You are on page 1of 467

Research Design and Methodology in Studies on L2 Tense and Aspect

Studies in Second and Foreign


Language Education 2

Editors
Anna Uhl Chamot
Wai Meng Chan

De Gruyter Mouton
Research Design
and Methodology
in Studies on L2
Tense and Aspect
edited by
M. Rafael Salaberry
Llorenc Comajoan

De Gruyter Mouton
ISBN 978-1-934078-14-3
e-ISBN 978-1-934078-16-7
ISSN 2192-0982

Library of Congress Cataloging-in-Publication Data


A CIP catalog record for this book has been applied for at the Library of Congress.

Bibliographic information published by the Deutsche Nationalbibliothek


The Deutsche Nationalbibliothek lists this publication in the Deutsche
Nationalbibliografie; detailed bibliographic data are available in the Internet
at http://dnb.dnb.de.

2013 Walter de Gruyter, Inc., Boston/Berlin


Cover image: Creatas/Thinkstock
Typesetting: RoyalStandard, Hong Kong
Printing: Hubert & Co. GmbH & Co. KG, Gttingen

Printed on acid-free paper


Printed in Germany
www.degruyter.com
Acknowledgements

The authors gratefully acknowledge all the authors who contributed to the
volume for their own contributions and for acting as reviewers of other
chapters. We also acknowledge the substantive and useful feedback of
the reviewers commissioned by DeGruyter to review the entire volume,
as well as the critical reviews of readers of individual chapters: Michel
Achard, Robert De Keyser, Elena de Miguel, Alex Housen, John Norris,
Ana Teresa Perez-Leroux, and Jacqueline Toribio. The support by the
editors at De Gruyter was greatly appreciated for their usefulness and
good work. We especially need to mention the work of the late Cathleen
Petree, who accompanied us in the rst stages of the editorial process of
our book. After countless email exchanges with her, we especially remember
some of the lines of the quote that she used in her email signature: All
about us is noise. All about us is noise and bramble, thorn and din, each
one of our ancestors on our tongues. . . (from the poem Praise Song for
the Day by Elizabeth Alexander). Finally, we would like to acknowledge
the support of our families and academic institutions (the University of
Texas at Austin, the University of Vic, and the University Center for
Sociolinguistics and Communication at the University of Barcelona) for
their continued support.

Rafael and Llorenc


Table of contents

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v

Introduction
Research design and methodology in L2 studies of tense and aspect . . 1
Llorenc Comajoan and M. Rafael Salaberry

Part I. Theoretical representations of tense and aspect in L2 studies

Chapter 1
A Cognitive Grammar perspective on tense and aspect . . . . . . . . . . 11
Susanne Niemeier
Chapter 2
The Spanish preterite and imperfect from a cognitive point of view . . 57
Aintzane Doiz
Chapter 3
Frequency-based grammar and the acquisition of tense and aspect in
L2 learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Nick Ellis
Chapter 4
Generative approaches to the L2 acquisition of temporal-aspectual-
mood systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Dalila Ayoun and Jason Rothman

Part II. Research design and methodology in L2 studies of tense


and aspect

Chapter 5
Research design: A two-way predicational system is better than a
four-way approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Paz Gonzalez
Chapter 6
Research design: Operationalizing and testing hypotheses . . . . . . . . 187
M. Rafael Salaberry
Chapter 7
Research design: From text to task . . . . . . . . . . . . . . . . . . . . . . . . . 219
Kathleen Bardovi-Harlig
viii Table of contents

Chapter 8
Dening and coding data: Lexical aspect in L2 studies . . . . . . . . . . 271
Yasuhiro Shirai
Chapter 9
Dening and coding data: Narrative discourse grounding in
L2 studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
Llorenc Comajoan
Chapter 10
Data analysis: Quantitative approaches . . . . . . . . . . . . . . . . . . . . . 357
Robert Bayley
Chapter 11
Data analysis: The qualitative analysis of actionality in learner
language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Anna Giacalone-Ramat and Stefano Rastelli
Chapter 12
Integrating the analyses of tense and aspect across research and
methodological frameworks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez
Author biographies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
Subject index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
Introduction
Research design and methodology in L2 studies of
tense and aspect

Llorenc Comajoan and M. Rafael Salaberry

Time and its linguistic expression is such a pervasive topic that it is no


wonder that it has occupied a central role in the study of linguistics in
general and second language acquisition (SLA) more specically.1 The
current volume focuses on dierent matters pertaining to research design
and methodology in studies on second language (L2) tense and aspect
dened from a broad perspective; that is, spanning from cognitive to
generative linguistics, from qualitative to qualitative methodologies. The
main goal of the volume is to bring to light the main issues regarding
research design by examining the role of theoretical representations of
aspect and methodological procedures. The volume is divided into two
main sections. The rst section provides a discussion of dierent theoretical
approaches to study tense and aspect in SLA. The second section focuses
specically on various factors regarding methodological conditions and
constraints that directly aect the collection of empirical evidence to sub-
stantiate theoretical hypotheses.
The rst section, focusing on theoretical issues, is comprised of four
chapters, with a predominant focus on cognitive approaches. The rst two
chapters discuss cognitive grammar, whereas the third chapter addresses
the role of frequency in the acquisition of L2 tense and aspect features.
The fourth chapter provides a counterpoint by examining the generative
(minimalist) perspective to L2 tense and aspect. More specically, in
Chapter 1, Susanne Niemeier provides a clear introduction to cognitive
linguistics and cognitive grammar in general through an analysis of the
English tense and aspect system. Taking Langackers approach and Mental
Space Theory as a point of departure, the author argues that prototypical as
well as nonprototypical uses of tense and aspect can be accounted for by
making reference to general categorical rules (as opposed to exceptions).

1. See Binnick (1991) for the history of how tense and aspect have been studied
from dierent perspectives; Bybee, Perkins, and Pagliuca (1994) on how dif-
ferent languages code time; and Bardovi-Harlig (2000); Slabakova (2001);
Salaberry and Shirai (2002); Ayoun and Salaberry (2005); Rocca (2007); and
Salaberry (2008) for SLA studies.
2 Llorenc Comajoan and M. Rafael Salaberry

The chapter is particularly useful because for every section of the discus-
sion it provides specic applications of cognitive grammar to the L2 class-
room to teach tense and aspect features. In the second chapter, Aintzane
Doiz discusses the various aspectual meanings of the Spanish Preterite
and Imperfect, also from the pespective of Cognitive Grammar. Thus, the
chapters by Niemeier and Doiz complement each other in the sense that
they apply the same theoretical apparatus to describe two tense-aspectual
systems, namely English and Spanish. In particular, both authors remark
the need to provide L2 learners with cognitive explanations that help them
relate how they view situations (cognitively) and the ways to express them
in the L2. Furthermore, both Niemeier and Doiz emphasize the power of
cognitive grammar in the way it handles exceptions to specic uses of
tense-aspect forms in English and Spanish. Finally, both authors agree on
the fact that further research still needs to fully investigate to what extent
cognitive grammar can help or hinder learners with the process of acquir-
ing a second language. In this view, the chapters by Niemeier and Doiz
will be of value to researchers who are also L2 teachers and may be inter-
ested in trying a cognitive approach to the teaching of tense and aspect.
In Chapter 3, Nick Ellis, provides a review of frequency-based grammar
(e.g., Construction Grammar) and examines the role of frequency in lan-
guage cognition and SLA. Ellis provides a review of how frequency and
learning are related by examining input frequency, form salience and per-
ception, and prototypicality and contingency of form-meaning mapping.
The second section of the article examines how such determinants of
learning were applied to the study of L2 tense and aspect in a study by
Wul et al. (2009). In the last section of the chapter, the author argues
for a dynamic model of usage that integrates all the factors that aect lan-
guage constructions; that is, he advocates for research that is not limited
to univariate analyses of data but that is rather multivariate and interac-
tive. The fourth chapter is devoted to a review of generative approaches
to the L2 acquisition of temporal-aspectual-mood systems. Dalila Ayoun
and Jason Rothman provide an in-depth introduction to the generative
theoretical approach in its current form and review current generative L2
literature by examining how dierent studies and methodologies in L1-L2
pairings have provided empirical evidence to support the dierent hypo-
theses advanced within this theoretical framework. Their critical review
highlights the rich body of empirical evidence accumulated by the eld of
generative linguistics in SLA. In their conclusion, Ayoun and Rothman
argue that, when taken as a whole, there is evidence for the position that
L2 adult learners are not impaired in their acquisition of functional cate-
Introduction 3

gories and their features (contrary to some of the hypotheses they have
reviewed in their chapter). Ayoun and Rothman contextualize their con-
clusion within the background of current and future work that deals with
the interface between syntax and other grammar components.
The second section of the volume comprises seven chapters, which can
be further subdivided into three subsections: (a) theoretical issues that
directly aect the research design of studies on tense-aspect (Chapters 5
and 6), (b) methodological factors that aect the analysis of tense-aspect
data (Chapters 7, 8, and 9), and (c) the use of qualititative and quantita-
tive types of analysis of data (Chapters 10 and 11). The rst two chapters
of this section (by Gonzalez and Salaberry, respectively) bridge the rst
and second section of the volume by focusing on the dierences in the
characterization of the theoretical construct of aspect for the development
of appropriate research hypotheses. In her chapter, Paz Gonzalez presents
a discussion of aspect in Spanish and argues that a two-way distinction of
predicational aspect (as opposed to the traditional, Vendlerian four-way
distinction) may be more accurate to describe the learners interlanguage.
More specically, Gonzalez argues that native speakers rely on the com-
positional aspect of verbal predicates (including arguments and adjuncts)
to the extent that they accept uses of Spanish preterite and imperfect that
isolated might be considered ungrammatical. Learners accept them because
they rely on creating a non-verbalized context. Using evidence from a
previous study, the author argues for the Predication-Eect Hypothesis,
whereby learners rely on two aspectual features (durativity and termina-
tivity) to make a dual distinction (preterite and imperfect). From this
perspective, the mapping of two features onto two morphological forms
may facilitate the learners understanding of the Spanish tempo-aspectual
system.
The chapter by M. Rafael Salaberry continues the discussion of the
compositionality of verbal predicates and its eect on the research design
of studies and their hypotheses. More specically, Salaberry focuses on an
aspectual distinction that has been little studied, namely the acquisition of
preterite and imperfect contrasts in Spanish to mark iterativity and
habituality. The chapter provides a detailed discussion of how such mean-
ings are dependent on the interaction between the use of perfective and
imperfective Spanish markers and the use of specic adverbs. Salaberry
argues that dierent authors interpretations of the role of adverbial adjuncts
have prompted dierent operationalizations of research hypotheses asso-
ciated with iterativity and habituality. The author further argues that such
distinct operationalizations are directly related to various methodological
conditions of the research design (e.g., in the design of language prompts)
4 Llorenc Comajoan and M. Rafael Salaberry

as well as dierences in analyses of the empirical data obtained within spe-


cic research designs. Ultimately, the research design and the results of
studies cannot be easily disentangled from theoretical assumptions used
to guide the research design of a study. The chapter ends with an argu-
ment in favor of including more rather than fewer layers of meaning asso-
ciated with aspectual concepts in the design of studies (adjuncts, context,
and so on).
The second subsection on methodology is composed of three chapters
dealing with aspects related to the collection of data (Bardovi-Harlig)
and the analysis of the collected data (Shirai, Comajoan). In Chapter 7,
Kathleen Bardovi-Harlig discusses some of the advantages and drawbacks
of open-ended tasks to collect data on L2 tense-aspect phenomena, and
she relates that analysis to the use of controlled tasks. The chapter presents
an exhaustive review of dierent types of open-ended tasks that have been
used in SLA tense-aspect research: conversations, narrative extended mono-
logic discourse (narratives and conversational narratives, elicited narratives,
personal narratives, impersonal narratives, and personalized narratives),
and nonnarrative monologic discourse (description, argument, and irrealis).
For each type of task, the author discusses specic data examples, elicita-
tion techniques, and practical considerations.
The following two chapters, by Yas Shirai and Llorenc Comajoan,
respectively, follow a structure similar to the one adopted by Bardovi-
Harlig in that both authors examine how lexical aspect categories and
grounding categories have been dened and how particular constraints
and conditions of such denitions bring about problematic issues in the
coding of tense-aspect data. Chapter 8, in particular, tackles two particular
questions: (a) How many lexical-aspectual categories do we need to code?
and (b) Are aspectual categories similar across languages? The bulk of
the chapter is dedicated to a detailed discussion of the most commonly
used classication procedure of lexical aspectual categories, namely, Shirai
(1991) and Shirai and Andersen (1995). Shirai discusses in detail the vari-
ous tests for lexical use, as well as problematic cases for the classication
of predicates (e.g., predicates that can be accomplishments or achievements;
and the classication of predicates with the verbs say and come/go). Shirai
argues for a data coding methodology that is detailed and rigorous enough
so that it becomes as replicable as possible.
Chapter 9 begins with a presentation of the Discourse Hypothesis and a
discussion of denitions of narratives according to dierent authors. Next,
several denitions of grounding are evaluated through an examination of
how dierent studies have operationalized the two main grounding con-
Introduction 5

cepts: foreground and background. For ease of presentation, Comajoan


divides the denitions into early denitions (based on the application
of denitions from functional linguistics) and critical denitions, which
examine the concepts critically and attempt to nd clearer criteria for their
denition. Comajoans discussion focuses primarily on the establishment
of the Discourse Hypothesis, how foreground and background were dened
within this hypothesis, and how the two concepts have evolved in current
studies. The nal section of the chapter addresses some of the problematic
cases for the coding of foreground and background, namely its relation-
ship with morphology and syntax (e.g., subordination), coding in dierent
types of texts, and interpretation of the learners intended meaning.
The nal subsection on methodological matters is comprised by two
chapters (10 and 11) that focus on the advantages and disadvantages of
quantitative and qualitative perspectives for the analysis of tense-aspect
data. In Chapter 10, Robert Bayley begins with some practical information
regarding matters of coding quantitative data, such as learners intended
meaning and how to interpret and code it, and the coding of all the
possible inuences in the L2 acquisition of tense-aspect. The bulk of the
chapter is devoted to a discussion of three methods that can deal with
multiple variables (linguistic and nonlinguistic) in the L2 acquisition of
tense and aspect; namely, multivariate analyses (focusing on logistic regres-
sion), testing alternative hypotheses, and implicational scaling. The chapter
will be of use to current and future researchers who are interested in
collecting dierent types of data on L2 tense-aspect and investigating
how dierent variables interact and contribute to the explanation of results
for dierent tasks. Finally, in Chapter 11, Anna Giacalone-Ramat and
Stefano Rastelli argue for the need to incorporate qualitative characteriza-
tions of tense-aspect data into research programs by examining some of
the advantages and drawbacks of qualitative analysis. From their perspec-
tive, the traditional type of quantitative analysis, with its emphasis on
classication (operational tests), misrepresents the important fact that
learners often may not have the same representations as those posited by
the researcher. Thus, Giacalone-Ramat and Rastelli propose a rationale
and procedure of qualitative research on L2 aspect based on three compo-
nents: tracking the same referent in dierent moments of time and exam-
ining within-subject variability, making a comparison of L1-L2 verb pairs
that can have dierent actional characteristics (e.g., see, watch; look for,
nd ), and comparing contexts in which aspectual pairs are produced in
learner data (i.e., scope widening). The authors apply the three principles
to past and current studies of Italian L2 by Chinese students and show
6 Llorenc Comajoan and M. Rafael Salaberry

that learners at low levels use actionally underspecied verbs (contra the
Lexical Aspectual Hypothesis 2 ) and use other cues to mark actionality.
They argue that their results are not caused by frequency in the input,
but rather they seem to be general and not restricted to learners with
dierent L1s. In their conclusion, Giacalone-Ramat and Rastelli advo-
cate for the use of qualitative research to investigate closely the learners
semantic representations and posit hypotheses that then can be examined
in a quantitative way.
The volume closes with a chapter by M. Rafael Salaberry, Llorenc
Comajoan, and Paz Gonzalez, in which the main themes of the volume
are discussed and related to matters of integrating research theories and
methodologies as well as language teaching. More specically, the authors
discuss four issues. First, they review the theoretical constructs discussed
in the volume and examine them from the perspective of the dependent
variable to be accounted for and the independent variables that contribute
to the acquisition of L2 tense and aspect. Second, they discuss the main
trends regarding methodology that have been addressed in the volume;
namely, the need for the use of multivariate and rigorous data analyses
that can allow for replication of studies. Third, the authors discuss the
advantages and drawbacks of adopting a multiple methods perspective.
And nally, a section on the eect of explicit instruction on the acquisition
of L2 tense and aspect is included, in which three guidelines for eective
instruction are provided. The article concludes with a call for further re-
search that creates links between theoretical approaches and methodologies
with the aim to establish stronger collaboration among researchers of L2
tense and aspect.

References

Ayoun, Dalila & M. Rafael Salaberry


2005 Tense and aspect in Romance languages. Amsterdam: John Ben-
jamins.
Bardovi-Harlig, Kathleen
2000 Tense and aspect in second language acquisition: Form, meaning,
and use. Oxford: Blackwell.

2. The terms Lexical Aspect Hypothesis (LAH) and Aspect Hypothesis (AH) are
used indistinctively by dierent authors in the chapters of the volume.
Introduction 7

Binnick, Robert
1991 Time and the verb: A guide to tense and aspect. Oxford: Oxford
University Press.
Bybee, Joan, Revere Perkins & William Pagliuca
1994 The evolution of grammar: Tense, aspect, and modality in the lan-
guages of the world. Chicago: University of Chicago Press.
Rocca, Sonia
2007 Child second language acquisition: A bi-directional study of English
and Italian tense-aspect morphology. Amsterdam: John Benjamins.
Salaberry, M. Rafael
2008 Marking past tense in second language acquisition. London:
Continuum.
Salaberry, M. Rafael & Yasuhiro Shirai
2002 The L2 acquisition of tense-aspect morphology. Amsterdam: John
Benjamins.
Shirai, Yasuhiro
1991 Primacy of aspect in language acquisition: Simplied input and pro-
totype. Los Angeles, CA: UCLA, Ph.D. dissertation.
Shirai, Yasuhiro & Roger Andersen
1995 The acquisition of tense-aspect morphology: A prototype account.
Language 71. 743762.
Slabakova, Roumyana
2001 Telicity in second language. Amsterdam: John Benjamins.
Wul, Stefanie, Nick Ellis, Ute Romer, Kathleen Bardovi-Harlig & Chelsea LeBlanc
2009 The acquisition of tense-aspect: Converging evidence from cor-
pora, cognition, and learner constructions. Modern Language
Journal 93. 354369.
Part I. Theoretical representations of tense and
aspect in L2 studies
Chapter 1
A Cognitive Grammar perspective on tense and aspect

Susanne Niemeier

1. Introduction

Cognitive Grammar (CG), belonging to the framework of Cognitive Lin-


guistics (CL), is founded mainly on earlier work by Langacker (1987a, 1991)
and is a relatively recent linguistic approach. Focusing on the English tense/
aspect (TA) system, this chapter will outline the theoretical basis for CGs
view on tense and aspect and also provide reasons for CGs didactic poten-
tial. Although CG is not a completely uniform approach, the theoretical
description of the English TA system takes a prominent place within all
of the various CL approaches to grammar (e.g., Fauconnier 1994, 1997;
Fauconnier and Turner 2002; Langacker 1991a, b, 2000, 2001, 2002,
2008a, b, c, 2009, 2011; Radden and Dirven 2007; Taylor 2002). All of
these frameworks have in common that they treat language as an integral
facet of cognition and regard grammatical phenomena such as TA as
meaningful (i.e., providing language users with clues for understanding).
The meanings of grammatical phenomena are of course more abstract
than those of lexical items for example, the meaning of the English plural
-s can be glossed as more than one but they are nevertheless helpful
for the correct interpretation of the utterance in question. Based on the
assumption that situations are generally not reected directly in linguistic
forms but by means of elaborate cognitive construals, the ways in which
people structure their experiences through language can be investigated.
Therefore, a crucial role in CG is assigned to the notion and analysis of
construals.
More recently, a range of publications has been dedicated to CG applica-
tions to the areas of second language acquisition and grammar instruction
(e.g., Achard and Niemeier 2004; De Knop and De Rycker 2008; Doiz-
Bienzobas 1995, 2002; Putz, Niemeier, and Dirven 2001a, b; Robinson and
Ellis 2008, Salaberry 2008; Tyler 2008; Tyler, Mueller, and Vu 2010) in
which researchers argue that due to its usage-based nature and its focus
on (conceptual) meaning, cognitive grammar may oer foreign language
learners a descriptively adequate and intuitively comprehensible account
of grammar.
12 Susanne Niemeier

Focusing mainly on Langackers foundational approach to tense and


aspect (but also integrating further approaches along the line, such as
Fauconnier 1994, 1997; Niemeier and Reif 2008; Radden and Dirven
2007 as well as Tyler and Evans 2001), this chapter intends to address
the following questions: How can the phenomenon of the interaction of
situation type (or lexical/inherent aspect) and grammatical/viewpoint aspect
be explained in CG terms, and how can it be explained to the foreign
language learner? How can apparent restrictions on the combination of
certain TA components and the role of the ground in the conceptualiza-
tion of situations expressed by tensed verbs be explained in CG terms
and also to the foreign language learner? Furthermore, how can we take
account of the diverse non-temporal uses of tense and present this network
of extended uses of the ed-morpheme in a foreign language classroom?
It will be demonstrated how a prototype account of TA categories (e.g.,
Shirai and Andersen 1995, Shirai 2002) might render extended or more
peripheral uses plausible to the language learner (which in traditional
TEFL (Teaching English as a Foreign Language) approaches tend to be
treated as exceptions). Concerning non-prototypical uses of aspect, the
focus is on the use of the progressive with iterative processes (as in She
kicked her little brother versus She was kicking her little brother), the use of
the progressive with involuntary sensory perception (as in I see a bird y
by versus I am seeing Tom tonight) and the use of the non-progressive
with performatives (as in I promise to be on time). With respect to tense,
the following non-temporal uses of the present and past tenses are dis-
cussed: the expression of reality and irreality (epistemic stance), the desig-
nation of salience (foregrounding/backgrounding), and the attenuation of
speech acts (politeness phenomena), which all seem to rely on the image
schema of proximity versus distance, be it in a temporal or in a non-
temporal way. Finally, I briey comment on the necessity of empirical
research in the foreign language classroom, outlining potential research
topics and methodological procedures.

2. Basic CL assumptions

As mentioned above, CL (and as such also CG) sees all facets of language,
including grammar, as meaningful, and posits in contrast to other ap-
proaches such as Transformation Grammar that meaning is the most
important issue in language. Meaning is not only present in lexis, where
it is easily discernible, but also in grammar, albeit in a more abstract
Cognitive perspective 13

way. Grammar is reducible to symbolic relationships, that is, to form-


meaning pairings. Consequently, CL does not see lexis and grammar as
two dierent structural principles of language, but as belonging together
and as being located on a continuum, in which lexis is situated at one
pole and grammar at the other pole, with various in-between positions.
In Langackers terms, lexicon, morphology and syntax form a continuum
of symbolic units serving to structure conceptual content for expressive
purposes (1987a: 35). As a CG analysis of tense and aspect relies on the
abstract meaning potential of the two phenomena in question, it may be
better equipped than other approaches to deal with tense and aspect in a
foreign language classroom setting, where meaningful explanations are the
basic ingredient for the learners understanding.
Furthermore, CL is based on the assumption that meaning is on the
one hand embodied (Gibbs 2006 or Ziemke, Zlatev, and Frank 2007)
as it derives from our general cognition, world-view and experiences (for
example, the expression a warm welcome metaphorically derives from our
experience that to feel human warmth is positive, which goes back to our
earliest experiences in life when as an infant we were held by a parent) and,
on the other hand, meaning is seen as socially embedded (cf. Tomasellos
2005 usage-based approach to language acquisition). CG and CL thus
attempt to explain facts about language in terms of other properties and
mechanisms of the human mind, body, and social environment. They
furthermore hold that language is a reection of human cognition and
conceptualization. Based on the premise that human perceptions of the
world are always ltered through our particular physical and neurological
architecture, CL argues in a rather constructivist way that humans do
not have direct access to an objective, external reality (see, for example,
Lako 1987: 266 or Evans and Green 2006: 47).
Rather, what people do have direct access to are their own subjective
conceptualizations. For example, language users can decide which perspec-
tive they want to take on a certain event what to foreground, what to
background, what to omit, and what to focus on. Thus, a simple event
such as two people shaking hands can linguistically be presented in very
dierent ways: linguistic representations of such an event may dier con-
cerning, for instance, who instigated the handshake, what the handshake
meant (greeting, congratulation, sealing of a deal, etc.), what the people
felt, etc. The linguistic representation of an event is rooted in the concep-
tualization that the event evokes in a language user or that the language
user decides to adopt. Conceptualizations arise from the complex interac-
14 Susanne Niemeier

tions of a persons rich cognitive abilities and our species-specic interac-


tions with the external spatial physical-social world.
Crucially, while claiming that human access to the external world is
indirect and ltered, CL also argues that our interactions with the physical-
social world are fundamental to how our cognition is shaped. Basic force
dynamics, such as our understanding of gravity or motion along a path,
provide foundational schemata that give structure to our understanding
of many other domains of experience. For example, if we are told that a
plane has landed at Frankfurt airport, due to our basic knowledge of the
source-path-goal schema we know that it must have started somewhere
(source), and that it must have own through the air (path). As we
metaphorically conceive of time as abstract motion on the time line (at
least in Western societies), we also know that if we are told that an event
lasted until a certain point of time (goal), it must have started before
(source) and must have had some duration (path). As we will see when
discussing the non-progressive versus the progressive aspect, dierent con-
ceptualizations of the same event are involved, where the non-progressive
foregrounds the goal and the progressive foregrounds the path.
Humans, who are fundamentally social in character, use language as a
tool to interact with others. Language allows us to externalize our internal
conceptualizations in order to make them accessible to other humans.
Depending on how exactly we want to present our concepts, we can use
dierent construals. For instance, if we want to focus on the fact that a
certain event is ongoing at speech time (What are you doing? Im reading
the newspaper), we may want to use the progressive aspect. If, however, we
want to present the same event in a holistic way, such as in an enumera-
tion of events, we may want to use the non-progressive aspect (Every
morning I have a cup of coee and read the newspaper).
The primary function of language is communication and we learn lan-
guage by using it in communicative and social contexts. This is especially
true for the rst language, but also for further and foreign languages.
Since language is understood to reect conceptualization, language is all
about meaning, as also emphasized by CLs focus on the symbolic aspects
of language. CL sees linguistic meaning not as referential and objective,
but as subjective, dynamic, exible, encyclopaedic and usage-based.
Not only words and expressions but also the grammar, or morphosyntax,
of a language reects conceptualization and is therefore meaningful. A
simple example for this is the iconic eect of the English plural morpheme
-s: more form is more meaning; that is, adding an extra morpheme to a
Cognitive perspective 15

noun adds the extra meaning of plurality.1 CGs focus on the motivated,
meaningful connections between forms that are often ignored by other
theories of language is one more reason why a CG approach may be
useful to second language pedagogy, because what is motivated and mean-
ingful can be explained.
This means furthermore that so-called exceptions are no exceptions
after all, but that there are explainable reasons why they behave in a dif-
ferent way than the prototypical forms. Examples for this claim can be
found in Section 3.2, dealing with non-prototypical uses of aspect that are
treated as exceptions in many textbooks of English as a foreign language.
CL applications thus invite a change of perspective in that they do not
posit a clear borderline between rules and exceptions but instead refer to
language phenomena as situated within a radial network of meaning with
more prototypical instances at the core and more marginal instances on
the fringes, all of them related and explainable (Radden 1992; Tyler and
Evans 2004).
Learners should not be expected anymore to learn by heart seemingly
idiosyncratic exceptions, which they may not understand and which are
therefore hard for them to memorize, but to reconstruct them via the con-
nections to the prototypes. For example, as will be shown later on, the use
of the non-progressive with verbs of involuntary sensory perception as in I
see a bird y by is not to be treated as an exception but can be attributed
to the fact that the event of seeing is too short to zoom into and therefore
has no duration which could be focused on. Such an approach is also be-
lieved to be helpful for teachers as they can use the motivated connections
as explanatory tools. The main aim then is to make learners aware of the
motivation behind linguistic phenomena and to help them understand
how language works, as understanding is seen as a precondition for learn-
ing. This seems to be possible via the inherent explanatory power of CL
approaches. Furthermore, grammar and lexis are not seen as separate
from each other (at least in German EFL textbooks there is always a
distinction between grammar/structure and vocabulary/content), but
as two poles of a continuum, thus structured by the same organizational
principles. In the following sections, the above assumptions will be out-
lined using the English TA system as an example.

1. This is similar in pidgin and creole languages which often rely on reduplication
in order to indicate plurality, as the languages may not yet have developed a
specialized plural morpheme.
16 Susanne Niemeier

3. Aspect

Although every single English verb is by denition always marked for


aspect as well as for tense (and also for modality) at the same time, aspect
and tense will be looked at separately in the following paragraphs. At
some important points, however, the interaction between the two subsys-
tems will be highlighted. Assuming that grammatical units, just like lexical
ones, are meaningful in the sense that they possess a phonological and
a conceptual pole (Langacker 1991b), the aspect system and the tense
system also carry meaning. In contrast to lexical units, however, gram-
matical units are used to express rather abstract meanings, such as, for
example, being relevant for a time before the communicative present in
the case of the past tense marker. The dierent meanings of the aspect
system and the tense system may interact and inuence each other. When
we look at tense, we take an external perspective on situations that species
the time of a situation as well as its reality status, whereas when we look
at aspect we adopt an internal perspective; thus, we are concerned with
the internal temporal structure of a situation (Comrie 1976: 3), which is
actually part of the situation itself.
Situations can be viewed from dierent perspectives, and the language
user can normally choose between dierent ways of presenting a situation.
Langacker calls a perspective a viewing arrangement, which he sees as
the overall relationship between the viewers and the situation being
viewed (Langacker 2008a: 73) and denes it as follows:

Inherent in every usage event is a presupposed viewing arrangement,


pertaining to the relationship between the conceptualizers and the situation
being viewed. The default arrangement nds the speaker and hearer
together in a xed location, from which they report on actual occurrences
in the world around them. There are however numerous kinds of departures
from this canonical circumstance. The departures help make it evident that
the default arrangement, so easily taken for granted, is nonetheless an essen-
tial part of the conceptual substrate supporting the interpretation of expres-
sions. Whether canonical or special, the viewing arrangement has a shaping
inuence on the conception entertained and consequently on the linguistic
structure used to code it (Langacker 2001: 16.).

In other words, our own perspective on the situation to be represented


linguistically is integrated into the upcoming representation itself, because
the neutral situation i.e., speaker and hearer conceptualizing the same
Cognitive perspective 17

situation from exactly the same perspective will be extremely rare. The
speaker encodes his/her subjectivity grammatically and the hearer decodes
it not necessarily in exactly the same way, which, at least from a con-
structivist point of view, is seen as an impossibility but the speaker has
at least provided the hearer with valid hints concerning his/her subjective
perspective, for example whether an action is seen as being in progress or
as completed. For instance, when a speaker says I am living in Paris, this
may indicate to the hearer that Paris is only the speakers temporary loca-
tion, as the use of the progressive with inherently unbounded verbs such as
to live indicates that the situation may change.

3.1. Interaction of situation type and grammatical aspect


In contrast to heavily tense-prominent languages (Bhat 1999), as for
instance German, Swedish, or Danish, in English every single verb has
to be marked for aspect.2 Although Bhat argues that English is a tense-
prominent language as well, it is so to a lesser extent (see also Dekel
2010), and we may have to assume a continuum here. Looking at the
other end of this continuum, English diers from aspect-prominent lan-
guages, such as for example Slavic languages, as it only has a clearly
marked progressive aspect, but no special marker for the contrasting ele-
ment in the pair.3 Although there is no widespread agreement about the
existence of a non-progressive aspect in English, I assume in this chapter
(following Radden and Dirven 2007: 177196) that there is a cross-wise
aspectual contrast in English, aecting both processes and states, which
are, according to Langacker, subtypes of situations.
According to Radden and Dirven (2007: 47), the term situation is to be
understood in the sense of events that happen or states that things are

2. It may be mainly due to this dierence that aspect errors are the most frequent
errors committed by German learners of English (Niemeier and Reif 2008).
German verbs do not need to be marked for aspect, as aspect is either indi-
cated in various optional lexical ways or omitted altogether. Therefore, many
German learners of English tacitly assume that aspect is not an obligatory
category in English. Most EFL textbooks used in Germany do not explain this
to them either.
3. In Slavic languages, aspect is an important formal category marking imper-
fective and perfective construals of situations (see Schmiedtova and Flecken
2008).
18 Susanne Niemeier

in, although the distinction between events (which I prefer to call processes,
following Langacker) and states is not always a clear-cut one. Processes
can further be subdivided into durative and punctual processes, and states
can be subdivided into permanent and transitory states.
Starting with prototypical scenarios, situations can be classied into
two categories according to their inherent temporal structure: they either
refer to inherently unbounded situations (which Langacker calls imperfec-
tive, 2008: 147) or to inherently bounded situations (which Langacker calls
perfective). However, categorisation is exible and subject to subtle con-
ceptual inuence from a variety of sources (Langacker 2008: 148).
Langackers opposition between perfectives and imperfectives is dierent
from, but partly compatible with Vendlers categories (Vendler 1967), as
Langacker states that the imperfective class is equivalent to what Vendler
1967 calls states, his other three categories (achievements, activities
and accomplishments) are subclasses of perfectives (1987b: 79). While
Langacker is aware that activities designate processes that are easily con-
strued as being internally homogeneous (2009: 189) and thus bear some
resemblance to imperfectives, he claims that they are nonetheless conceived
as occurring in bounded episodes, and bounding is the critical property for
the perfective/imperfective contrast.
Inherently unbounded situations are internally homogeneous and not
susceptible to change (Niemeier and Reif 2008; Williams 2002); thus,
they are not expected to come to an end. This is not the same as saying
that they will never end, and they must of course have had a beginning at
some point in time, it is just . . . that the verb itself excludes them [i.e., the
beginning and the end] from what it puts onstage for focused viewing
(Langacker 2008: 147). Such situations can either be permanent states,
such as <BE BRITISH>, which according to our commonsensical world
knowledge is normally not going to change throughout the lifetime of
a person, or they can be potentially transient states, such as <LIVE IN
LONDON>, for which a change cannot be ruled out, but is neither prob-
able nor predictable and is thus not part of our viewing frame.
By contrast, inherently bounded situations are internally heterogeneous
and susceptible to change, because they allow internal development and are
expected to come to an end at some point. Inherently bounded situations
can either have explicit boundaries, such as <BUILD A SNOWMAN>,
which is by denition over once the snowman is nished, or they can
have implicit boundaries, such as <WANDER ABOUT THE PARK>, a
process that, although it does not have a xed endpoint, is very unlikely
Cognitive perspective 19

to last forever.4 As already mentioned, our world knowledge provides us


with an intuitive understanding of the default readings of situations, due
to which the prototypical interpretations of situations like those outlined
above develop naturally, at least for native speakers. We may, however,
not want to focus on the prototypical perspective but indicate that we see
a given situation dierently or want the hearer to see it dierently. This is
where grammatical aspect comes into play.
Grammatical aspect (i.e., the use of either the non-progressive or the
progressive form) interacts with lexical aspect in that it oers the speaker
a means to construe an idealized situation in dierent ways.5 Depending
on the type of situation (whether it is inherently bounded or unbounded),
and depending on whether it is construed as a single situation or as a
repeated situation, grammatical aspect can have dierent conceptual eects.
If we compare the following two sentences, we see that they both refer
to the same idealized situation, that is, <BUILD A SNOWMAN>, an
inherently bounded situation with explicit boundaries. However, this situa-
tion is realized dierently in (1a) and (1b) as far as grammatical aspect is
concerned:
(1) a. Nick built a snowman.
b. Nick was building a snowman.
As we can see, (1a) is the default (i.e., the unmarked mode for the verb
to build ), as the situation as such is inherently bounded, thus involving
change through time. The situations beginning (initial boundary) as well
as its endpoint (nal boundary) are part of the mental representation. At
some point in time, Nick starts to roll the snowballs that are then stacked
up one upon another and decorated with a hat, a scarf, a broom and so on
in order to become the nished snowman in the end.
While the non-progressive aspect in (1a) expresses that the situation is
viewed in its entirety and that both its beginning and its endpoint fall
within the scope of predication, the use of the progressive aspect in (1b)

4. At this point the bounded situation <BUILD A SNOWMAN> would turn


into the inherently unbounded situation <BE A SNOWMAN>, a potentially
transient state because it is not expected to last forever, but only until the
weather gets warmer.
5. Idealized situation is to be understood in the sense of Smith (2009: 9), who
claims that when speakers talk about actual situations, they invoke abstract
representations, or idealized situations types. The idealized situation types are
abstractions that represent the properties characteristic of dierent situations.
20 Susanne Niemeier

has the eect of unbounding the situation.6 Another notable dierence


between (1a) and (1b) consists in the fact that while the situation in (1a) is
perceived as being complete in itself and is thus not susceptible to change
anymore, the situation in (1b) is construed as being in progress and is thus
(at least potentially) susceptible to change (Williams 2002: 88).
We nd the same unbounding eect if we look at inherently bounded
situations with implicit boundaries, such as in (2a,b):
(2) a. We wandered about the park at night.
b. We were wandering about the park at night.
The only dierence between the snowman example in (1) and the park exam-
ple in (2) is that the boundaries of the situation <BUILD A SNOWMAN>
are explicit, while the boundaries of <WANDER ABOUT THE PARK>
remain implicit. A walk in the park starts at some point in time and needs
to end at some point in time. It is conceptualized as some kind of
bounded episode, irrespective of whether a natural endpoint is discernible
(Langacker 2001: 13). There is no change in state or any internal struc-
turing involved (Klein 1994, 1995), so we are not dealing with a proto-
typical example of an inherently unbounded situation here. Whereas in
(2a) we speak about a completed instance of the situation <WANDER
ABOUT THE PARK>, in (2b) we zoom into the situation, defocus its
boundaries, and focus instead on its middle part (i.e., its ongoingness).
As pointed out above, the use of the progressive aspect unbounds inher-
ently bounded situations. If we turn to inherently unbounded situations, it
becomes obvious that the progressive needs to have a dierent eect on
them as in these cases there is no need to unbound an already unbounded
situation anymore. Let us rst consider a prototypical state verb such as to
live:
(3) a. My best friend lives in London.
b. My best friend is living in London.
Example (3) shows that inherently unbounded situations such as (3a) are
construed as lasting states that are not susceptible to change when used in

6. This process is called a defocusing of boundaries by Schmiedtova and


Flecken (2008): both the initial and the nal boundaries of the situation are
not gone, but excluded from the speakers focus.
Cognitive perspective 21

the default mode (i.e., with the non-progressive aspect).7 When used with
the progressive as in (3b), however, implicit boundaries are added to the
situation and it is no longer seen as a lasting, but instead as a temporary
state (i.e., construed as having implicit boundaries and as being susceptible
to change). Instead of defocusing the boundaries as it does with inher-
ently bounded situations as in (1b) and (2b) the progressive aspect with
inherently unbounded situations imposes boundaries.
Such a conversion from an indenitely lasting state to a temporary state
is often visible in the description of characters or of peoples behaviour:
(4) a. You are arrogant.
b. You are being arrogant.
When we utter a sentence like (4a), we are referring to a characteristic
quality of a person or to a persons general style of behaviour, whereas
when we use (4b), we are referring to the current, temporary behaviour of
a person, independent of their normal behaviour and character. They may
not be arrogant at all, but in this one special moment they are acting as if
they were.
In other words, we are dealing here with a cross-wise aspectual contrast
(cf. also Radden and Dirven 2007: Chapter 8). That is, the progressive
changes the default boundary situation, a view that diers slightly from
Langackers perspective. Langacker claims that the overall eect of a pro-
gressive is (. . .) to convert a perfective process into an imperfective one
(2008: 155), which coincides with the defocusing of boundaries for inher-
ently bounded situations. However, he does not mention the cross-wise
eect, namely that the progressive can impose boundaries on an inherently
unbounded situation but analyzes this latter phenomenon dierently: he

7. In textbook grammars and other learner grammars, learners are generally


confronted with a list of state verbs and are usually told that these verbs
cannot take the progressive aspect (for more details see Niemeier 2008). Such
a list is supposed to be learnt by heart, which is not exactly a useful learning
strategy, as the learners may face diculties because they do not know any
reason why the exceptions dier from the rule and may thus be unable to
construct or reconstruct their meaning. Although the role of frequency of a
linguistic phenomenon has largely been underestimated so far (see Chapter 3),
frequency does not seem likely to be a decisive factor for a rehabilitation of
the learning by heart strategy, because exceptions in grammar tend to be
infrequent and thus do not normally appear as salient features in the linguistic
input a learner receives instead, they are by denition non-prototypical.
22 Susanne Niemeier

claims that when an imperfective verb is construed as referring to a


bounded episode/temporary state, the progressive only admissible with
perfectives can be used because the normally imperfective verb in that
case relates to a (non-prototypical) perfective instance. Opposing these
two views may seem akin to a chicken-or-the-egg question (i.e., what
came rst: the concept that inuenced the form or the form that inuenced
the concept?). I would like to interpret this potential dilemma rather as
either speaker- or hearer-oriented: for the speaker, the concept comes rst
and needs to be expressed accordingly. That is, seeing a situation as
unbounded/imperfective may trigger the use of the progressive, whereas for
the hearer, the form used helps him/her to interpret the concept/meaning
that the speaker wants to get across.8 For teachers this would imply that
the concepts their learners want to express are the most important assets, as
these concepts inuence the grammatical form to be chosen. For example,
when dealing with aspect in a class of English as a foreign language, the
idea of zooming into a situation visualizable by a lens or a keyhole
stencil (Niemeier and Reif 2008) might be employed in order to have
the learners think about what they really want to express. In other words,
learners should rst be made aware of what they want to express so that
they can then focus on how to express this and employ the form that ts
the intended meaning.
To sum up, the non-progressive and the progressive aspect have an
eect on the construal of a situation, indicating how the internal constitu-
tion of a situation is viewed. With inherently bounded situations, the non-
progressive aspect construes the situation as complete in itself, including
its boundaries (i.e., its starting point and its endpoint), as in (1a) and
(2a). The progressive aspect, by contrast, has the eect of defocusing the
boundaries of the situation by zooming in on the situation (Langacker
2000: 228) and thus construing it as ongoing, as in (1b) and (2b). With
inherently unbounded situations, the non-progressive construes the situa-
tion as continuing indenitely, as in (3a) and (4a). The progressive, by
contrast, imposes implicit temporal boundaries on the situation, as in
(3b) and (4b), seeing it as potentially susceptible to external change and

8. For EFL purposes, working with the concept of the cross-wise aspectual con-
trast has proven to be a lot more fruitful as learners can relate to it far better
than they can to Langackers interpretation. The concept of boundaries
along with the idea that the progressive changes the default setting of the
boundaries of a verbs meaning is apparently easier to understand than the
concept of (im)perfectivity (personal experience).
Cognitive perspective 23

thus focusing on its temporariness and its possible transition to adjacent


states.
Making learners aware of the meaning behind the aspectual system of
English a major source of errors at least for German learners of English,
but presumably not only for these learners may help avoid some of their
errors, as they would not have to learn rules by heart and then simply
reproduce them (or forget them). Instead, they can start from scratch: by
understanding what aspect is about and how a grasp of the aspectual
system enables them to ne-tune their utterances, they can express their
own perspectives. This will potentially also change the learners notion of
grammar as an opaque end in itself. The boundaries (or lack of these)
are quite easily visualizable (Niemeier 2008; Niemeier and Reif 2008),
thus enabling double coding. As the learners are not inhibited by rules
with exceptions, they can handle the language more freely and more crea-
tively. They could, for example, create fantasy stories about what they
saw when they were clandestinely looking through a keyhole, witnessing
action in progress and trying to interpret it (for further examples, see
Niemeier 2008 and Niemeier and Reif 2008).

3.2. Non-prototypical uses of aspect


Accounts of aspect in textbook and other learner grammars usually present
aspect in terms of rules and exceptions to these rules.9 For example,
Ungerer et al. (1992) just state without any further explanation that
iterative verbs are used in the progressive form to express a series of
repeated actions, but that when a denite number of repetitions is stated,
the simple past has to be used.10 This is not helpful for learners as they
will have problems to understand why this is the case. In contrast to this,
CG argues that what are traditionally called exceptions are not excep-
tions at all, but rather non-prototypical uses of aspect which are meaning-
ful and therefore explainable. In this section, three apparent deviations

9. That is, if they present aspect at all, because it is generally integrated into
the chapters on tense, most frequently without even making use of the term
aspect. Although it is debatable whether learners need the technical terms,
they should at least be aware of the concepts behind the terms and know the
dierence between them.
10. Ungerer et al. (1992) is a book-size learner grammar. Although it is as of
this writing one of the better grammar books for German schools and
although it acknowledges some cognitive principles, it still works with rules
and (unexplained) exceptions.
24 Susanne Niemeier

will be discussed, namely the use of the progressive with iterative pro-
cesses, the use of the progressive with verbs of involuntary sensory percep-
tion, and the use of the non-progressive with performatives.

3.2.1. Iterative processes 11


Starting with iterative processes (such as kick, hop, knock, etc.), at rst sight
they seem to be prototypical members of the class of inherently bounded
situations, as the processes that they describe are clearly delimited in time.
Still, such verbs do not normally allow the progressive aspect. If we com-
pare the following two sentences, it is obvious that they describe dierent
scenarios:
(5) a. Daria kicked her little brother.
b. Daria was kicking her little brother.
In (5a), the protagonist kicked her brother only once, whereas in (5b), she
gave him several kicks. We cannot use the progressive to refer to a single
kick because the process is too short to zoom into: in other words, it does
not allow an internal perspective. With the inherently bounded situation
<KICK>, we have a starting point as well as an endpoint, but as these
two nearly coincide temporally, the process itself does not have any dura-
tion. Using the progressive with an inherently bounded situation means
defocusing the situations starting point as well as its endpoint, which in
the case of such punctual, short processes like <KICK> leaves actually
nothing to view. Still, the progressive can of course be found with such
verbs, as in (5b), only the interpretation then changes to that of an itera-
tive, repetitive process. When we use the progressive with such minimal-
duration verbs, we conceptually extend the situations duration in order
to be able to zoom into the middle phase between the starting point and
the end point, which then results in an interpretation of repetitive short
actions. We therefore interpret sentences such as (5b) as referring to a
succession of short processes such as kicking or hopping, which are,

11. I will use the term iterative here for simplicitys sake. If an event of no or
very short duration is meant, we should technically speak about a semelfac-
tive, whereas when this event is repeated, we speak of an iterative. This section
of the chapter will cover both concepts. Semelfactives have uniplex structure,
whereas iterative have multiplex structure (see Evans and Green 2006: 519 on
the concept of plexity). For a dierentiation between iteratives and habituals
from a Langackerian perspective, see Salaberry (2008).
Cognitive perspective 25

according to Langacker (2008a: 156), construed as constituting a single


overall event of bounded duration. We can then highlight and zoom
into the internal duration of this overall situation, defocusing the beginning
and the endpoint of the sisters tantrum. If such a series of short, punctual
processes were lmed and replayed in slow motion, also a single kick could
be referred to as She is kicking her brother because the duration of that
process would then be long enough to zoom into (Niemeier 2008: 317).
What we are dealing with here is therefore not an exception or a devia-
tion, but a semantic mismatch, (i.e., meaning inconsistencies) between verbs
denoting short processes and the concept of unboundedness. If there is no
middle phase between the boundaries to zoom into, unbounding is not
possible. It becomes possible again if we reconstruct the overall situation
as consisting of several repeated sub-processes. If our learners are already
familiar with the notion of a situations boundaries, it is quite easy for
them to discover why short, bounded, non-prototypical processes have to
be treated dierently from prototypical bounded situations. However, this
claim is still waiting for empirical validation.
As suggested in Niemeier (2008), teachers could introduce the notion of
boundaries via the count-noun/mass-noun distinction, which presents no
problem for German learners of English. Once the learners know to pay
attention to boundaries, the content within the boundaries and the
important role of their own perspective, they can be made aware of the
fact that there is no content in iterative processes and no possibility to
zoom into the situation. This then rules out a defocusing of the boundaries
and therefore the use of the progressive aspect, which would focus on the
(non-existent) content, is ruled out as well. A repetition of a short pro-
cess would then extend the situation, creating duration and thus content
between the boundaries, and at the same time allow for a zooming in
perspective.

3.2.2. Verbs of involuntary sensory perception


The second apparent deviation to be discussed is closely related to the rst
one. Verbs of involuntary sensory perception (e.g., see, hear, feel, smell,
taste) generally describe very brief processes and the majority of EFL text-
books and grammars lists them as exceptions because they do not take the
progressive aspect.12 Again, CG argues that these verbs are no exceptions

12. One example may serve as illustration: in the brand-new and widely distributed
textbook series Green Line for higher secondary schools in Germany (Gym-
nasien), we nd the following text in the grammar section of volume 5: Die
26 Susanne Niemeier

but that they do not necessarily invite an internal perspectivization due to


their brevity, just as described above for verbs like kick, hop, nod, and the
like. If we verbalize a perception such as <SEE A FLASH> or <HEAR A
GUNSHOT>, then these processes will be over before we will have had
the time to contemplate or describe them, (i.e., the processes lack duration
and their onsets and osets nearly coincide). In this respect, verbs of punc-
tual involuntary sensory perception behave in a similar way as the verbs
for short processes described in 3.2. In contrast to these, however, verbs
of involuntary sensory perception do not allow an iterative interpretation
as this would represent a semantic incompatibility if something is in-
voluntary, it cannot be repeated at will because by denition ones will
is not involved. If such processes are repeated, then we are dealing with
voluntary sensory perception (watch/observe/look at the reworks; listen
to gunshots during a battle, etc.) and the verbs for voluntary sensory per-
ceptions behave like any other inherently bounded verbs.13
However, verbs of involuntary sensory perception can nevertheless
be used with the progressive aspect in two respects. On the one hand, the
processes being involuntarily perceived may have some kind of extension
as for example in <FEEL PAIN> (i.e., may not be punctual such as
<SEE A FLASH> or <HEAR A GUNSHOT>), in which case they are

progressive form ist normalerweise nicht moglich bei Verben, die eine Sinnes-
wahrnehmung beschreiben wie to notice, to see, to hear, to sound, to smell,
to look (aussehen) etc.. (Horner et al. 2010: 118). Translation: the progres-
sive form is normally not possible with verbs describing sensory perception,
such as to notice, to see, to hear, to sound, to smell, to look (referring to out-
ward appearance), etc.. Not only does the book fail to distinguish between
voluntary and involuntary sensory perception, but it furthermore adds the
verbs to notice and to look to the category of sensory perception and most
lamentable of all it does not provide any reasons for the verbs deviating
behaviour nor does it explain what is meant by normally.
13. Such a lexical switch is not only to be found in English. Ibarretxe (1999) studies
the polysemy of perception verbs in English, Basque, and Spanish, basing her
research on the classication of perception verbs following Viberg (1984) and
Gisborne (1996), who claim that on the basis of the semantic role of their sub-
jects, perception verbs can be divided into three groups: experience, activity,
and percept. Not all languages ll all groups with dierent verbs, but when
a verb is used for more than one category, it is used with a polysemous
meaning. In German, we can nd a similar distribution of verbs of involun-
tary versus voluntary perception as in English, which is partly achieved via
prexation (sehen versus ansehen/betrachten; horen versus zuhoren; fuhlen
versus befuhlen, anfassen; riechen versus beschnuppern).
Cognitive perspective 27

treated like any other bounded situation that has a duration we can zoom
into for unbounding it. Although one might argue that such situations do
not possess a long duration, according to Langacker, stability and dura-
tion are relative and what matters is whether a situation is construed as
stable for the purpose at hand and whether this stability endures through
the stretch of time considered relevant (2008a: 149). If a sensation is pre-
sented as constant for the brief temporal interval in question, it can be
unbounded and used as an imperfective, as in I was feeling some pain in
my left knee this morning this is largely dependent on whether the object
noun accompanying the verb in question invites a punctual interpretation
(e.g., ash, gunshot) or not (e.g., light, music, pain).
On the other hand, we nd uses such as the one depicted in (6) below:
(6) She is seeing Peter tonight.
In this example, <SEE> metonymically stands for <MEET>, referring to
the complete scenario connected to meeting a person, furthermore hinting
at a touch of romance. <MEET> is a prototypical inherently bounded
verb, encompassing various subprocesses (e.g., going to a bar, seeing
Peter there, walking towards him, greeting him, having a drink with him,
etc.) and, therefore having a certain duration which can be zoomed into. If
verbs of involuntary perception are not being used in their basic, proto-
typical sense but in more marginal, mostly metonymically motivated
senses which no longer focus on the actual sensory perception in question
but rather on the complete scenario, they gain duration and can therefore
semantically accept an inner perspectivization, which in turn allows the
progressive aspect. The sensory perception then just serves as a metonymic
link.

3.2.3. Performatives
The third apparent deviation relates to the use of non-progressives with
performatives. Performatives explicitly name speech acts ( promise, admit,
swear, pronounce, sentence, apologize, etc.) and the speaker is always the
subject. They are always bounded and are always uttered in the present
tense as uttering them coincides with the notion of doing them, there-
fore we not only need to look at the (un)boundedness of such situations
but also at the notion of the present tense. When we use the present tense,
our speech time should temporally coincide with the situation time, but
with a vast majority of situations, this is not possible. A situation such as
<READ A BOOK> usually takes a lot longer than uttering the English
28 Susanne Niemeier

sentence *My son reads a book, and therefore the preconditions for the use
of the simple present are violated.14 Although <READ> is an inherently
bounded situation, it cannot be framed as bounded when used in the pres-
ent tense as it does not comprise both of the necessary boundaries the
starting point is included but the endpoint has not yet been reached. Rad-
den and Dirven (2007: 208) state that the conceptual boundaries of most
events do not neatly coincide with the temporal boundaries of uttering the
speech act describing the event, therefore the simple present cannot nor-
mally be used to describe bounded situations happening in the present
time.
If we use a performative speech act such as I promise to write the paper
next week, we avoid the above-mentioned dilemma, as this sentence does
not describe a situation but is the promise itself (i.e., by uttering the
sentence the speaker performs the act of promising). This means that there
is temporal identity between the utterance of the speech act on the one
hand and performing the act of promising on the other hand. Therefore,
situation time and speech time completely coincide. The verb to promise
is bounded and as the whole utterance contains the promise, the starting
point of the event as well as the endpoint fall within our viewing arrange-
ment. In Langackers terms (2001: 26), performatives not only tolerate
but actually require the present tense. The reason is that a performative
represents a special viewing arrangement in which the process put onstage
and proled is the speech event itself .
Performatives furthermore lack an epistemic problem that normally
arises with the use of the present tense, namely that of speaker knowledge.
If we want to speak about a situation in the present, we rst of all have
to identify it in order to be able to describe it. However, as such mental
processing may take some time, and as the utterance itself has to correlate
temporally with the situation when we use the simple present, the action is
frequently well underway before we can even start to talk about it. As
a consequence, we are dealing with a temporal incongruity relative to
the starting point of the situation.15 For example, when we see somebody
repairing their car, the act of repairing the car takes a lot longer than to
say *Peter repairs his car (the durational problem) and furthermore we

14. In other languages, however, such as in Spanish, French or German, this


utterance would be perfectly acceptable in the simple present.
15. In Langackers terms, the epistemic problem has to do with the fact that by
the time we observe an event to nd out what is happening, it is already too
late to initiate a description that precisely coincides with it (2009: 191).
Cognitive perspective 29

rst have to identify what Peter is actually doing, as he might also be


cleaning the car or inspecting it (the epistemic problem). None of these
two problems exists with performatives because, on the one hand, the
speech act is the situation, and on the other hand, the speaker performs
the action intentionally (which means that there is no need for him/her to
rst identify the action, as it is known to him/her anyway). Thus, the
explanatory potential of the CG view on aspect with respect to performa-
tives is able to cater for this non-prototypical aspectual use. If the basic
aspects (boundaries) are known to the learners, they can extend these
in order to explain more marginal aspectual uses.

4. Tense

Tense is our grammaticalized conceptualization of time. The description


of this grammatical phenomenon needs to begin with a reminder that
tense, aspect, and modality are essentially non-separable issues and are
only separated here for practical reasons. Our human experience only
allows us to experience the present time directly, as the past has already
happened and can only be accessed by recall, although it can have reality
status. The future has not yet happened and therefore can neither be expe-
rienced directly nor does it have reality status. Cognitive linguists are
divided into two camps concerning the idea of how many tenses English
has: Langacker strongly argues for only two tenses, the present and the
past, which according to him both have reality status and are morphol-
ogically marked on the verb, whereas the future for him belongs to the
modal system because it has no reality status and is non-inectional.16
Taylor (2002: 394) also states that English has only two tenses. Radden
and Dirven (2007: 224), on the other hand, maintain that English has
three tenses: the present, the past, and the future, although they admit the
following:

. . . our projection of events into the future always involves a certain amount
of uncertainty [. . .] Future situations are therefore very much subject to
peoples imagination. As a result, English has a number of future tense
forms expressing shades of (un)certainty about a future situation.

16. In his 1991a publication (332), Langacker still speaks about the future tense,
whereas from his 1991b publication onwards he seems to have changed his
views.
30 Susanne Niemeier

Expressions such as peoples imagination and (un)certainty, however,


can also be seen as arguments for shifting future-related verb constructions
to the realm of what Langacker (2008a: 306) calls conceived reality,
namely modality.17 It is also interesting to note that according to Radden
and Dirven (2007: 227), those future forms with the highest degree of
certainty are what they call the planned future (Im getting married
next month) and the scheduled future (My train leaves at six), both of
which use present tense morphology. Furthermore, on a conceptual level
it can be argued that only the present and the past tenses are used to signal
relevance time for situations that are construed as having reality status
(i.e., situations located in factual reality). This is why the present chapter
will follow Langackers assumption that there are only two tenses in
English and that future time is expressed by modality.18
With respect to foreign language teaching, the question whether to
teach that there are two tenses in English or instead to teach that there
are three tenses is hard to answer. The three-tenses option is compatible
with all traditional textbooks and theories, and the learners will probably
be familiar with this approach from lessons in their native language, where
from primary school onwards normally three tenses are identied. On the
other hand, the two-tenses approach makes more conceptual sense for the
reasons mentioned above (i.e., the connection to reality). This should be
easy to internalize for the learners and it also helps to explain certain uses
of the present tense, as for example the fact that using the present tense to
refer to the future lends more reality status to the future event in question.
Furthermore, in this way the future auxiliary will does not have to be
dierentiated from the modal auxiliary will, as they can both be seen
as referring to potentiality space. Therefore, although tradition sees this
dierently, I would advocate for the use of the two-tenses approach in
the foreign language classroom.

17. For a Cognitive Linguistic way of teaching English modal verbs to speakers of
other languages, see Tyler (2008) and Tyler, Mueller, and Ho (2010).
18. As the present chapter sees future as a modality, it will not refer to the various
types of future. Radden and Dirven (2007: 225) dierentiate the following
kinds of future: predicted future (Well have some sunshine), matter-of-course
future (Ill be seeing you), intentional future (Im going to get married ), con-
tingent future (Its going to rain), planned future (Im getting married next
month), scheduled future (My train leaves at six) and background future (If I
see him, Ill send him home).
Cognitive perspective 31

4.1. Speech time, relevance time, and situation time


Langacker as well as Taylor (2002) subsume tense under the keyword
grounding, as the term ground is used in CG to indicate the speech
event, its participants (speaker and hearer), their interaction, and the
immediate circumstances (notably, the time and place of speaking) (Lan-
gacker 2008a: 259). Tense situates the proled relationship of an utterance
with respect to the speakers current conception of reality.
In a CG analysis of tense, three dierent components interact (see
Figure 1).19 The rst component is the present moment of speech, or
speech time. We live in current reality, therefore the base of the tense
system is the present moment of speech, or speech time. In Radden
and Dirvens terms, speech time oers an anchor to locate the occurrence
of situations in time (2007: 202). It is the moment in time at which a
communicative instance is produced. If we speak about today, the present
moment of speech is part of the time region indicated by this lexical item.
Similarly, if we speak about yesterday or tomorrow, these lexical items are
still interpreted in relation to speech time (i.e., to the present moment of
speech), although they do not include the present moment of speech time
but refer to time intervals before or beyond speech time.
The second component is the time span for which the speaker sees the
proposition of the utterance as relevant (i.e., relevance time). The relevant
time span for a proposition such as We were snowed in yesterday is past
time, more precisely yesterday, as indicated by the grammatical past tense
marker were and the temporal adverb yesterday.
The third component in the relation between tense and time is the time
at which a situation (i.e., a process or a state) is instantiated. It is therefore
called situation time.20 Situation time can correspond exactly to relevance
time, as in (7):

19. This distinction is very similar to Kleins distinction between time of utterance,
time of situation, and topic time (Klein 1994). However, the two models
do not coincide completely. According to Klein, tense serves to relate topic
times and not situations to utterance time. Klein claims that the situa-
tions themselves are not linked directly to utterance time, but only to topic
time. This linking is then done by aspect (Klein 1994, 1995). Relevance
time, on the other hand, refers to the speakers viewpoint, i.e., to the time
span for which the speaker sees the proposition of the utterance as relevant.
20. Radden and Dirven (2007: 202) use event time but I have decided to follow
Niemeier and Reif (2008), who use situation time, because in this context a
situation can refer either to a process or to a state, and I believe it is con-
32 Susanne Niemeier

(7) We were snowed in yesterday.


On the other hand, relevance time can also comprise only a temporal sec-
tion of situation time, as in example (8) below; or it can be dierent from
situation time, as in the newspaper headline Snowstorm cuts o villages.
Although the situation took place a day before the headline appeared
(i.e., in the past), and was remedied since, the newspaper headline wants
to present it as being relevant to its readers present time in order to moti-
vate potential readers to buy the newspaper and read the article. In the
article itself, the past tense is then used, making relevance time equivalent
to situation time.
Tense does not locate the processes or states on the time axis, but it
rather allows the speaker to select a time span that is relevant for what
the speaker wants to say. This becomes clear when taking a closer look at
the already mentioned example (8):
(8) Emily and Charlotte came home to see her, but she was dead . . .
(British National Corpus: FNY 448)
Relevance time as well as situation time of the rst part of the sentence are
clearly in the past, but for the second part of the sentence, relevance time
is in the past, whereas situation time encompasses not only the past but
also the present, as the deceased remains dead also at the present moment
(i.e., the situation still applies at speech time). In other words, the past
tense in was dead does not primarily full the function of locating the
state of <BE DEAD> in past time, but instead it expresses that the rele-
vant time span for what the speaker is saying lies somewhere in past time
(relevance time), whereas the state of the persons being dead comprises
both past and present time (situation time). In this example then, rele-
vance time only includes a part of situation time, namely the relevant
time stretch which lies in the past.

4.2. Simple tenses


Radden and Dirven call the simple tenses (i.e., the present tense and the
past tense), deictic times as they relate to speech time, the only moment
that is available to us in our perception of time (2007: 204). They go on

ceptually somewhat dicult to subsume states under events, a term which


tends to have a more processual character. A similar problem occurs with
Langackers terminology, where process is used as a cover term for both
processes and states.
Cognitive perspective 33

Figure 1. Speech time, relevance time, and situation time

to explain that as speech time is always in the present but situation time
may be in the past the use of the present tense always locates a situation
at or around or including speech time, whereas the past tense locates a
situation at a time earlier than speech time. The simple tenses furthermore
give information about the reality status of a situation. Whereas the present
tense gives information about the immediate reality of a situation, the
past tense gives information about what Radden and Dirven call known
reality and what Langacker calls conceived reality (2008a: 301), argu-
ing that our knowledge about reality is partial and also not necessary
infallibly accurate, and therefore can only be a part of factual reality.
What we consider as known by us we frequently simply embrace (. . .) as
established knowledge. For a particular conceptualizer, C, this constitutes
conceived reality. It is what C accepts as being real. Langacker (1991:
245) sees the canonical temporal distinction between present tense and
past tense as a proximal/distal contrast in the epistemic sphere. In the
following two sections, the two tenses with reality status will be briey
characterized.
34 Susanne Niemeier

4.2.1. Present tense


Notwithstanding its name, the simple present is, according to Langacker
(2001), one of the most complex tenses of English. Langacker (1991: 242)
relates the present tense to immediate reality and argues that speech time
and reference time exactly co-occur in canonical uses of the present tense.21
What makes the present tense so complex is that such a co-occurrence is
not frequently to be found. Instead, there is a range of non-present uses
of the present tense, where what is being coded linguistically is not the
actual occurrence of events, but their virtual occurrence as part of a non-
canonical viewing arrangement (Langacker 2001: 30). In this section,
I discuss the various uses of the simple present for the true present,
for imperfective/bounded situations, for the narration of demonstrations,
for the scheduled future, for stage directions, for the historical present,
for eternal truths, and for timeless situations as well as for habituals.
The true present tense is not frequently used. As already described
above, bounded events happening in the present at the time of speaking
are generally too long to exactly coincide with the utterance itself (apart
from very few instances such as The balloon pops). We are not only deal-
ing with a durational problem here, but also with an epistemic one. Not
only does the utterance need to cover the whole event, but a speaker also
generally needs to rst observe a situation in order to identify it and to be
able to talk about it, which takes even more time away from the small
time frame provided by the utterance. This is why bounded situations in
the present generally take the progressive aspect in English, in this way
getting rid of the boundaries that would otherwise not t into the small
time frame. The zooming-in eect then proles only that portion of the
homogeneous state of aairs that is valid at speech time, such as in He

21. This is called epistemic immediacy by de Wit and Brisard (2009: 4), who
argue that the simple present entails a notion of epistemic necessity and the
present progressive a notion of epistemic contingency in the speakers concep-
tion of reality. In his 2009 publication, Langacker agrees that he paid too little
attention to the modal import of the present tense and also speaks about
epistemic immediacy (Chapter 7) and recently (2011), he has devoted a com-
plete paper to the commonalities and dierences of the temporal-coincidence
approach (treating the English present as tense) and the epistemic-immediacy
approach (treating the English present as modality), arguing that temporal
coincidence provides the basis for epistemic immediacy, the former one being
the category prototype and the latter one being the more general and sche-
matic account.
Cognitive perspective 35

is reading a book. On the other hand, many uses of the simple present
tense do not refer to speech time, but to either the future, or the past, or
to so-called timeless situations or eternal truths. As Langacker (2001)
argues, these latter ones are extended uses of the true present. For him,
the present tense indicates that a full instantiation of the proled process
occurs and precisely coincides with the time of speaking (2001: 22).22
Langackers denition of the present tense also accounts for imperfec-
tive/unbounded situations such as He resembles his grandfather. Such a
resemblance is valid without boundaries and is therefore imperfective. It is
true at any time, thus also for the moment of speaking, and in using the
simple present, that portion of the resemblance is highlighted that is in
focus at speech time: since an imperfective process is internally homoge-
neous and not characterized in terms of bounding, any subpart singled out
for proling will itself constitute a valid instance of the process type in
question (Langacker 2001: 23).
A further use of the simple present is to be observed in the narration of
demonstrations, such as in cooking programmes on television: I put a
tablespoon of butter in the pan. It melts quickly. Now I put the llet in. I
cook it at low temperature for ve minutes . . . (Langacker 2001: 28).
Although uttering these sentences requires a lot less time than the prepara-
tions and the ve minutes of cooking, Langacker claims that the simple
present is not used here for the description of actual situations but instead,
it is to be seen as reading o entries from a list or scenario (i.e., it refers to
the virtual occurrence of the situation which then again coincides with the
time of speaking).
A similar explanation can be given for the so-called scheduled future
use of the present tense (My train leaves at six). Here again, we are refer-
ring to the representation of this situation on a virtual schedule in our
minds and not to the actual situation. Such scheduled future uses generally
incorporate a precise time expression (at six) and do not work for situa-
tions that cannot be scheduled (*I fall ill next week). Although a virtual
schedule belongs to the future, it is stable and reliable also at the present
moment of speaking and can thus be regarded as a representation of
an anticipated actual event and as a virtual occurrence of that situa-
tion which coincides temporally with the moment of speaking (following

22. Langacker only refers to English, other languages such as German, which
has no grammaticalized progressive form express not only the true present
and timeless situations but also actions at speech time with the simple present.
36 Susanne Niemeier

Langacker 2001: 31). The same virtual reading applies to stage direc-
tions as well as to the use of the historical present, where past situations
are retold using the simple present. In this latter case, a past situation is
virtually replayed, and the use of the present tense underlines its salience,
its still being vividly recalled by the experiencer. In this context, Langacker
speaks of . . . event representations. Even when these correspond in some
fashion to actual events, the represented events are the ones directly coded
linguistically and proled by the present tense verb (Langacker 2001: 33).
A further related non-prototypical use of the simple present concerns
so-called eternal truths and timeless situations, such as The kangaroo
is a marsupial. This utterance does not refer to any specic kangaroo
but to a virtual instance of a kangaroo. This virtual situation belongs to
an open-ended set of actual instantiations, distributed throughout the
time span during which the generalization holds (Langacker 2001: 33).
The use of the present tense indicates that the speaker is referring to a
sub-part of this eternal truth at the moment of speaking. The same is true
for habituals (I drive to work every morning), where the utterance does
not refer to any actual instance of driving I may even be uttering this
sentence on a weekend when I am not driving to work but to a virtual
instance of driving.
Again, we are not dealing with exceptions in the case of the simple
present but with explainable meaning extensions. In an EFL classroom,
the basic meaning of the simple present as referring to situations occurring
in the here-and-now should be introduced at a very early point and will
then form a basis for the explanation of the extended uses as listed in this
section. The actual or virtual coincidence between speech time and reference
time is easy to visualize, for example by using a time axis and drawing
reference time-circles as well as situation time-circles in dierent colours
the reference time-circles will always be identical to speech time, whereas
the situation time-circles may be larger (they will, for example, cover the
complete time axis in the case of eternal truths) but will always contain
speech time as well as reference time. The learners can then verbalize the
illustrations and thus via the commonalities between the various proto-
typical and extended instances of use develop and advance their con-
cepts of the meaning of the present tense.23

23. One example for research on teaching tense in the English classroom via
Cognitive Grammar is Meuniers project Cognitive Grammar and EFL
at the University of Louvain, which started in the summer of 2010 and for
which no results have been published so far (see http://www.uclouvain.be/en-
323144.html, last date of access: October 17, 2012).
Cognitive perspective 37

4.2.2. Past tense


Whereas the present tense, as shown in the previous section, locates a
situation exactly at the time of speaking, the past tense morpheme im-
poses an immediate scope prior to the speech event (Langacker 2001:
22) and in contrast to the present tense is not limited concerning the length
of the proled process. It conveys distance from the speech event (Taylor
2002: 394) and refers to facts. A sequence of past tense verbs is often used
in narratives, and the individual events are interpreted as successive. That
is, the rst situation evoked by a past tense locates the narration in the
past and the following situations are interpreted to follow this rst situa-
tion in sequential order. As all of these situations are seen as completed,
they are interpreted as successive. If they co-occurred, the progressive
aspect needs to be used for the background situation.
Langacker (1991: 242) relates the past tense to non-immediate reality,
however, as will be pointed out below, this is only the prototypical, tem-
poral use of the past tense. The past tense is conventionally also used to
signal less commitment to the reality of a situation, that is, it is used to
signal epistemic distance as in if I were you (Evans and Green 2006: 396;
Gurevich 2010: 90). Furthermore, it is frequently used to indicate a lack of
intimacy as in My childs father was Italian, a lack of salience as in This
book is a masterpiece. It was published in London, (see Tyler and Evans
2001: 71), or a heightened degree of politeness as in I wanted to ask you
something (Radden and Dirven 2007: 211). In all of these cases the -ed
morpheme does not refer to non-immediate reality but to non-temporal
events. The dierent uses of the past tense morphology are outlined in
detail in Section 4.4.3.

4.3. Complex tenses


All other tense forms apart from the present tense and the past tense are
complex tenses (Radden and Dirven 2007: 204), which serve to locate
anterior or posterior situations relative to a reference time.24 Anterior
times are expressed as perfect tenses stating that something happened at an
earlier time than reference time; posterior times belonging to futurity
are expressed by prospective forms, generally consisting of grammaticalized
lexical items such as going to or be about to, stating that something may

24. They are called complex tenses because in contrast to the deictic tenses they
involve two temporal relations: a deictic temporal reference point and either a
backwards or a forwards look from this reference point.
38 Susanne Niemeier

happen at a later time than at reference time. Anterior times are the present
perfect and the past perfect.25 They have in common that a situation is seen
as located before relevance time. In complex tenses, the speakers viewpoint
is of importance because two times spheres are concerned: the rst one is
located in one of the deictic times the past or the present and the second
one consists of a look backwards (anterior times) or a look forwards
(posterior times, i.e., uses of the future). In this way, the speaker locates
anterior (or posterior) events relative to the deictic reference time.
In the case of the present perfect, a situation has occurred before the
present time but it may still continue until now (I have never seen him
before, which remains valid until I see the person) or may still have
an impact on the present (see example (9)). According to Radden and
Dirven (2007: 205), the time conguration described by the present
perfect is unique among the anterior times in that it involves only one
time sphere and a relation from event time to speech time (. . .) The unique
status of the present perfect is due to its present reference time: its imme-
diacy makes the present the more prominent time. This ts in nicely
with the fact that the present tense has the meaning of proximity (5).
However, although Radden and Dirven claim that we are only dealing
with one time sphere in the case of the present perfect, the temporal rela-
tion to a situation in the past is still obvious and, especially for didactic
reasons, I would recommend to highlight the fact that in the case of the
present perfect the time sphere of the past is involved as well, as otherwise
learners might have problems dierentiating between the deictic tenses and
the complex tenses.
(9) Ive broken my leg, so it is dicult for me to go shopping right now.
In example (9), speech time is right now (i.e., in the present), whereas
the breaking of the leg happened in the past. Using the present perfect
indicates that the past event of breaking the leg is seen from the perspec-
tive of speech time as presently being an obstacle concerning everyday life
routines. The present relevance of the anterior situation is highlighted,
which is why it can be seen as part of the overall situation and can there-
fore be said to have not only a temporal but also an aspectual meaning

25. If the future was treated as a tense, we would also have to mention the future
perfect here (I will have passed my driving license by the time I turn 18). If the
future is not considered to be a tense, then we are dealing with a combination
of modality and the auxiliary have -ed.
Cognitive perspective 39

(Radden and Dirven 2007: 206) insofar as the left boundary of the inher-
ently bounded situation <GO SHOPPING> is extended into the past so as
to also include the anterior situation. This is not possible for past perfects:
(10) I had broken my leg, so it was dicult for me to go shopping during
the weeks following the accident.
This sentence can only be understood in a purely temporal sense speech
time is in the present, relevance time is in the past, and situation time is
before relevance time. The use of the past perfect relates the situation
time to the relevance time.
Although perfect tenses are analyzed by Radden and Dirven (2007) in
their chapter on tense, this view is not unanimously shared. Langacker,
for example, treats the perfect not as a grounding element but as a
grounded structure, because in contrast to tense and modals, the perfect
(have -ed ) is optional and not obligatory in a verb and can only appear
in non-nite clauses from which tense and modals are excluded (2008a:
300). I tend to agree with Radden and Dirven; that is, I tend to see the
perfect as a tense because just as the other tenses it has reality status and
it is (partly) marked directly on the verb. However, it is not a prototypical
instance of tense but has an aspectual meaning as well, because on the
one hand, the speakers viewpoint is involved and, on the other hand, the
perfect is built in a similar way to the progressive aspect (i.e., it is directly
marked on the verb but also requires an auxiliary verb: have in the case
of the perfect, be in the case of the progressive aspect). The perfect can
furthermore also have a modal meaning because in its anterior uses it
refers to the future, relative to the reference point taken (reference point
in the past: I was going to ask her out yesterday when suddenly . . . ; refer-
ence point in the present: Im going to ask her out; reference point in the
future: I will have asked her out before the end of the week).
For learners, it is important to realize how many time spheres are
involved. In the case of deictic times represented by the simple tenses
only one time sphere is referred to, whereas in the case of complex times
represented by the complex tenses two times spheres are concerned: the
rst one being located in one of the deictic times and the second one con-
sisting of a look backwards (anterior times) or a look forwards (posterior
times, i.e., uses of the future). This can be illustrated by using a time axis,
indicating the deictic reference points as well as including arrows pointing
into the direction of the speakers shifted viewpoint (i.e., pointing to the
left in the case of anterior times). Complex tenses may also be visualized
by referring to Mental Space Theory (Section 4.4) as this helps to show the
40 Susanne Niemeier

interaction between the dierent time spheres quite clearly. Learners can
be trained to focus on the three elements of speech time, relevance time,
and situation time and with their help, decode temporal structures without
being confused.

4.4. Tense and Mental Space Theory


In the following sections, Langackers as well as Radden and Dirvens
understanding of tense will be combined with insights from Fauconniers
(1994, 1997) and Fauconnier and Turners (2002) Mental Space Theory
in order to generate a synthesized CG view on tense. The basic idea of
Fauconniers model is that when discourse participants interact, they
mentally construct small conceptual packets, called mental spaces (Fau-
connier 1998: 252). These mental spaces contain elements (i.e., conceptual
information on the things or people talked about), as well as information
on the reality status of the situation and its relevance time. This model
may be relevant for foreign language teachers as it allows visualization,
for example, for tracking the temporal shifts in a conversation, and thus
helps to clarify the thought processes involved.
The mental space that serves as the basis or starting point of an interac-
tion is called base space. Importantly, mental space theory makes a clear
distinction between base space and all other mental spaces. Base space is
the situation at speech time, the here and now, in which the speaker
is the deictic centre. It serves as an anchor for expressing both the reality
status and the temporal relevance of any situation that is being communi-
cated and it contains conceptual information on the things or people
talked about, the space and time of the interaction (speech time), as well
as information about the interaction context. All of this is normally taken
for granted and will therefore not necessarily be verbalized explicitly. The
status of things or people in the discourse is expressed by the determiner
system and is known as reference; the status of situations and their rele-
vance time is expressed by the tense, aspect, and modality systems.
A new mental space can be set up or an already established space can
be referred back to at any moment in the discourse. Most mental spaces
are opened implicitly, but this can also be done in a more explicit way.
Such explicit space builders are, for example, time adverbials, such as the
adverb yesterday in example (7) above (We were snowed in yesterday), or
temporal expressions such as in 1992, or the rst clause in the sequence
of two or more clauses in a narrative context such as in example (8) above
(Emily and Charlotte came home to see her, but she was dead . . .). All these
Cognitive perspective 41

space builders denote the temporal setting serving as a situations back-


ground and the hearer implicitly understands that base space has been
left and a new space has been opened up.
Other explicit space builders, referring to the reality status of a situation,
are expressions such as Possibly or Nick believes that . . . . Grammatical
markers for tense and aspect, to which modality can be added, are not
explicit space builders themselves, but provide the interaction partner with
clues concerning the space that is relevant at that point (Fauconnier
1994: 33). Fauconnier gives the nice example of In 1929, the lady with
white hair was blonde (1994: 29), where 1929 builds up a new mental time
space in the conversation, namely in the past. If this sentence is followed
by a past tense, we stay in the mental time space of 1929, whereas if the
sentence is followed by a present tense (Today she still wishes to be
blonde), we shift back to base space (now). Should this sentence be
followed by a conditional ( you would have enjoyed meeting her then), we
shift to a counterfactual space that provides us with information about
the reality status of a situation, which in that case would be located in
non-reality.
Grammatical tenses and aspects and their combinations serve to indi-
cate relative relations between spaces and, crucially, to keep track of the
discourse position of the participants which space is in focus, which one
serves as base and what shifts are taking place (Fauconnier 1994: xi,
emphasis in original). The reality status of a situation can be indicated
either by means of tense forms which are described in more detail in the
next section or by modal verbs. To sum up, situations set relevant for
base space are referred to by the use of the present tense and are always
located in reality. Mental spaces can be opened from base space for
situations whose relevance time is anterior to speech time and which are
therefore expressed by the past tense, or else whose relevance time is
posterior to speech time and which are therefore expressed by the will
form. Learners can be trained to identify base space via a diagram and
would then try to track a speakers train of thought to other spaces. A sim-
plied example deals with the book cover text of Harry Potter, 6 (Figure 2).
The text starts in base space, in the here-and-now, then the perspec-
tive shifts to the past. Such an illustration can then be extended in order to
include other spaces, for example hypothetical or counterfactual spaces.
Besides the temporal use of tense morphemes, we nd a number of non-
temporal and thus non-prototypical uses of tense morphemes, related to
the expression of reality versus irreality, to the notion of intimacy, to the
designation of salience, and to politeness (i.e., the attenuation of speech
42 Susanne Niemeier

Figure 2. Mental spaces in the (slightly adapted) blurb of Harry Potter and the
Half-Blood Prince

acts), which will all be described in the following sections. The temporal
use of tense morphemes does not primarily mean locating situations in
time, but rather indicates which time span is relevant for what we are
saying (i.e., the temporal use of tense and other morphemes is already
much more subtle than what has usually been understood by temporal
meaning). Furthermore, the same morphological forms can also be used
to encode non-temporal meanings. In the same way that, starting from
base space, we can open further reality-related spaces with past, present,
or future relevance time, we can also open up other mental spaces, as, for
instance, potentiality space. One kind of potentiality space is modality
space, others are hypothetical space or counterfactual space. Still further
possible spaces are, amongst others, interactive spaces such as narrative
space or politeness space.

4.4.1. Hypothetical/counterfactual space


In an assertive speech act, the reality status of a situation is taken for
granted and therefore not explicitly marked. However, if we are dealing
with a speech act based in irreality, a hypothetical space is opened up
which needs to be marked explicitly. In such a case, we are dealing with
either non-counterfactual hypotheticality or counterfactual hypotheticality,
both of which are marked by the use of past tense morphology.
A hypothetical space can be created by means of the space builder if
in combination with a set of tense forms that indicate epistemic as well
as temporal relevance.
Cognitive perspective 43

(11) a. If he is German, as you claim, lets talk to him right now.


b. If he shows up, lets ask him straight away.
c. If he showed up, we would ask him then.
d. If he had shown up, we could have asked him then.

The possibility that a hypothetical situation will become real is still given,
for example in the immediate future (11a) or in the more remote future
(11b). In terms of their reality status, the rst two situations in example
(11) are located either in the fuzzy peripheral area of base space (11a) or
close to the reality of base space, but already outside it (11b). If we are
dealing with an even higher degree of hypotheticality, such as in (11c),
where it is highly unlikely but not yet impossible that the person in
question will show up, the epistemic distance is greater. And if the episte-
mic distance to the reality of base space becomes still greater, we are deal-
ing with counterfactuality (11d), dened by Fauconnier and Turner as
forced incompatibility between spaces (2002: 230). A counterfactual
hypothetical statement indicates that it is impossible for the situation to
happen. The statements in (11c) and (11d) do not dier concerning the
events described or with respect to their times of occurrence, they only
dier with regards to their epistemic distance to base space. At the same
time, as the reality status of the situation changes, dierent relevance times
are involved, as illustrated by the use of the adverbs right now, straight
away, and then:
English past tense morphology clearly signals distance in two dierent
but related ways: it can either refer to temporal distance to base space (as in
One of the last times he saw the headmaster. . .) or it can refer to epistemic
distance to base space (as in example 11c) or it can refer to both temporal
and epistemic distance to base space (as in example 11d). These uses are
closely related, the second one may also be seen as a metaphorical exten-
sion of the rst one (epistemic distance is temporal distance). Both the
temporal and the epistemic uses can best be understood in their relation to
the double function of base space (i.e., its reality status function and its
relevance time function). Sentence (11d) illustrates how the interaction
between temporal and epistemic relevance is expressed grammatically:
since the situation is located in temporal as well as epistemic distance
from the base, a double backshift takes place grammatically, as can be
seen by the use of the past perfect in had shown up.
In example (11), the tense morphemes in both clauses coincide but this
need not be the case. In example (12), again accessing counterfactual
44 Susanne Niemeier

hypothetical space, the person talked about does not know, and in (12b)
the male protagonist has not told the truth. Although situation time in
(12a) is present time, the past tense is used to express epistemic distance
vis-a-vis the base. This example shows that past tense morphology cannot
only be used to indicate relevance time that is located anterior to (and thus
at a temporal distance from) the base, but also to express epistemic dis-
tance from the base. The same applies to (12b), where situation time is in
the past and where the situation itself is construed as a counterfactual one,
therefore the past perfect is used, indicating temporal as well as epistemic
distance from the base.
(12) a. If only he knew! (British National Corpus: AEB 3109)
b. If he had told her the truth, she would not have believed him.
(British National Corpus: HR8 858)

Every speaker has many grammatical options at hand to communicate


his/her thoughts, whereas from the hearers perspective, tense and aspect
marking allows us to reconstruct the reality and time spaces set up from
base space (Fauconnier 1997: 78) and the relations between them.

4.4.2. Narrative space


Another type of space that can be opened up from base space is narrative
space. Narrations deal with past events but frequently especially in very
lively oral narrations the present time can be used (see Section 4.2.1),
such as in the narrative discourse in example (13):
(13) a. Erm, Im just sitting in front of the car last night and erm . . .
(British National Corpus: KC2 3048)
Here, the narrative space allows a conict between situation time, which is
past time, and relevance time, which is presented as if it were present time.
However, no hearer will have a problem to understand that the situation is
in the past although the present tense is being used, because he will follow
the speaker from base space (i.e., from the here and now) into narrative
space that is per se located in the past. The fact that the present is being
used signals that the speaker is mentally reliving the events of last night
and is foregrounding them, which is why they still have an impact at
speech time.26 This impact is not only felt by the speaker, but also by the

26. See Tyler and Evans (2001: 72): past tense signals background and support-
ing status and present tense signals foreground status.
Cognitive perspective 45

hearer as they are both drawn into the situation as if they had been
present as onlookers. Instead of creating distance a function of the past
tense the use of the present tense creates closeness. If such closeness is
aimed at, the use of the past tense is not possible. The use of the present
tense in narrations (of movies, plays, or books, etc.) has been discussed by
other scholars, and alternative reasons for its use have been proposed.
For example, in Fleischmans (1990: 15) view, the use of the present is
due to its atemporality, as such narrations can be revisited on multiple
occasions. Although Fleischman (1990) as well as Klein (1994) point out
that personal narratives in contrast to movie or plot narratives are
rather told in the past tense, the present tense is also used especially for
the lively oral narrations mentioned above, as the speaker virtually re-
lives the events s/he is talking about.27 The speaker as well as the hearer
know that base space has been left and narrative space has been entered.
Therefore, the use of the present tense does not create any temporal and/
or epistemic conict. The CG perspective on the present tense, however,
goes beyond Fleischmans strictly temporal view of tense by claiming that
in addition to indicating atemporality the abstract meaning of the
present tense is to indicate the speakers proximal perspective vis-a-vis his
or her utterances.

4.4.3. Politeness space


Whereas the use of the present tense indicates that the situation described
is close to the speaker, the past tense indicates distance with respect to the
situation or the hearer. In example (14), the past tense is used although
situation time is the present time, as the question is asked at that precise
moment:
(14) I wanted to ask you something. (British National Corpus:
HTN 2787)

27. In his 2009 publication, Klein seems to see this slightly dierently: In the
narrative present, the whole action is in the past (. . .) but that at least some
of (sic) situations are presented as if they were present. There are two com-
mon interpretations of this use: the situations are felt to be present at the
time of utterance, or the speaker imagines himself to be present in the situa-
tion. Under the rst interpretation, the situations are somehow shifted in
time, and under the second interpretation, the deictic anchoring is shifted
(49). Klein does not further elaborate on this view.
46 Susanne Niemeier

According to Tyler and Evans (2001: 95), we conventionally understand


that the use of past tense does not place the desire to ask the question in
the past, but rather that it attenuates and so makes such requests less face-
threatening and hence more polite. That is, for reasons of politeness the
speaker presents the utterances relevance time as non-present time, which
has the eect that the request seems to be less urgent so that the hearer
may feel less imposed upon and thus less face-threatened. Using the past
tense in requests, suggestions, invitations, commands, and reprimands is
conventionally interpreted as polite, mitigating the amount of imposition on
the addressee. By moving from base space to politeness space, the speaker
signals a cognitive, non-temporal distance between him-/herself and the
hearer and by using past tense morphology, the speaker signals this dis-
tance also grammatically.
The perspective on the functions of tense morphology described above
ties in well with Tyler and Evans approach of pragmatic strengthening:

. . . due to the way in which we actually experience the notions of intimacy,


salience, actuality, and politeness, namely in terms of proximal-distal spatial
relations, and the fact that time-reference is experienced in terms of analo-
gous spatial relations, in certain situations tense morphemes which canoni-
cally signal time-reference can implicate a non-temporal relation. Through
usage-based conventionalization, i.e., pragmatic strengthening, a conventional
time-meaning can become associated with a particular tense morpheme
(2011: 65).

According to Tyler and Evans, non-temporal meaning extensions asso-


ciated with tense are explainable by the polysemy of tense morphology.
By entrenchment (i.e., pragmatic strengthening), tense morphemes have
become associated with non-temporal meanings. The concept of intimacy,
for example, in its basic meaning refers to physical intimacy, but in an
extended meaning we can also either feel close or feel distant to some-
body who is not necessarily spatially close or spatially distant to us. There-
fore, we may talk in the past tense about people we do not feel close to.
When we start telling a love story from our past, we may say things like
We met in Brisbane. He was an architect . . . may well still be alive and still
be an architect, but by using past tense morphology, we distance ourselves
from this specic episode of our life.
The concept of salience is explainable in quite a similar way, as what
is physically closer to us is more salient and vice versa, which is why we
can speak metaphorically of issues closer at hand or about distant
Cognitive perspective 47

rumours. Combining these insights with the use of tense morphology, we


can foreground important facts and make them more salient by using the
present tense and we can background less important facts by using the
past tense. This is in line with certain uses of the present tense for lively
narratives, the historical present and other non-canonical uses (see Section
4.2.1). As Tyler and Evans argue, the use of the present tense to make a
particular event more real would seem to be related to the use of the
present tense to denote greater salience and hence importance in terms of
information structure (2001: 93).
To sum up, tense morphology cannot only be used to symbolize temporal
relevance, but it can furthermore be used to express epistemic relevance,
salience, and attenuation in certain contexts. Both the temporal and the
non-temporal uses of tense forms can be ascribed to a common con-
ceptual basis, the proximal/distal (or immediacy/non-immediacy) schema
(Langacker 1991: 249).28 This means that while the present tense is always
used to express proximity/immediacy be it temporal, epistemic, narra-
tive, or formal proximity/immediacy , the past tense always indicates
distance/non-immediacy. In other words, in a very iconic way, more form
is more meaning, i.e., the addition of the past tense morpheme signals the
addition of the notion of distance, either with respect to relevance time,
reality status, backgrounding, or social commitment. The CG view thus
extends the mainly temporal view of tense that researchers such as Fleischman
(1990) or Klein (1994) have propagated by also incorporating pragmatic
aspects, especially concerning speaker perspective.

28. Tyler and Evans account for the fact that time is associated with the spatial
concepts of proximity and distance by referring to the experiential correlation
between the two concepts. When we speak, we are in the here-and-now, which
functions as our deictic center: We cannot help but experience the present
moment in terms of our immediate physical surroundings and our sensory
perceptions of them (. . .) Traversing a certain distance inevitably correlates
with the elapse of a certain amount of time. Thus, elements of the spatial
domain, such as movement from one location to another and distance, have
become strongly associated with the elapse of time (2001: 81). In other
words, the concepts of space and time are so intertwined in human cognition
that they are frequently co-activated, which is amongst other phenomena
visible in the non-temporal use of temporal morphology, when we speak about
the past tense as signaling distance (a spatial concept) in various ways.
48 Susanne Niemeier

5. Conclusion

Although this chapters aim was primordially an account of the CG per-


spective on tense and aspect, the teaching angle has been addressed
throughout the contribution whenever appropriate. Concerning the teach-
ing and learning of English as a foreign language, CG has not yet reached
regular textbooks, teaching materials, teacher training, or even teachers
awareness. Therefore, in order to be able to argue convincingly for the
benets of a CG-based approach to teaching tense and aspect, we need a
lot more research, especially empirically oriented, and we need materials
development based on such research.
For an empirical study to be fruitful and valid, however, we do not
only need a test group, but also a control group, which means that a
second class would have to be taught in a dierent, more traditional way
so that the results of these two groups could then be reliably compared at
certain points during the study, a process which may well cover several
years. Such ideal research scenarios are hard to nd, therefore it seems
much more realistic to use smaller test items than tense or aspect in their
totality and to restrict the time for the studies. The results found will
not be as positive as they might be under ideal test conditions, but if they
are indeed positive even under less-than-optimal circumstances, this may
already pave the way towards a rethinking of grammar learning and
teaching in the direction of CG-based instruction.
For Spanish as a foreign language the research situation seems to be
somewhat better than for English, as there are quite some publications
dealing with CG-inspired approaches, for example Marras and Cadierno
(2008) on constructions or Llopis Garca (2010) on mood selection. Espe-
cially interesting in this respect is Lopez Garca (2005), which outlines
hands-on ways for teaching Spanish grammar in a CG way for all gram-
matical areas, including tense and aspect. For English, there are among
others several contributions by Tyler and Evans on prepositions (2003,
2004) or by Alejo Gonzalez (2010), Dirven (2001) and Kurtyka (2001) on
phrasal verbs, but they all remain quite theoretical and do not report
on classroom experiments or give practical advice for teaching. Tyler,
Mueller, and Ho (2010) report on an actual teaching experiment concern-
ing modality, however, the group size was very small and the learners
were adult learners, therefore the results are not immediately replicable
in the regular foreign language classroom. So far, there is no published
empirical research on CG-inspired ways in which to teach English tense
and aspect at school.
Cognitive perspective 49

There is certainly no lack of research questions such as for example a


denition and description of the acquisition sequence of the tense-aspect
system. It would also be interesting to nd out whether CG-based instruc-
tion is equally well suited for every learning style, whether visualizations
of grammatical phenomena are perceived as helpful and if so by which
types of learner, which kind of visualization works best, how to introduce
topics and which terminology to use in the classroom at various levels, or
whether the learners error rate after CG-based instruction is lower and
if it is, in which areas (analysis of utterances, closed exercises, half-open
exercises, or open exercises such as free text production).29
The teaching of tense and aspect, two of the most error-prone areas for
ESL learners, could presumably prot very much from a CG approach
with its focus on the meaningfulness of grammar. Such an approach enables
teachers to begin with prototypical instances on the basis of which their
learners can understand the admittedly abstract meaning of specic
grammatical phenomena. In later lessons, teachers can then gradually
introduce non-prototypical instances of the phenomena in question. It
is important that learners are made aware of the fact that these more
marginal instances are not necessarily interrelated but nevertheless always
related to the basic meaning.
For example, the proximity versus distance meanings of the tenses are
to be found in the use of the -ed morpheme in temporal relations, which
acts as a prototype, but they are also to be found in other areas, for example
in referring to irreality, as outlined above. Connecting these various areas
in teaching makes it easier for learners to see that the abstract meaning
of in this case the ed-morpheme stays intact in all of them, and the
learners thus have the chance to get an integrated view on grammatical
areas that in traditional grammar teaching remain unconnected (as for

29. Boers (2004) claims that learners retain the meanings of metaphorical expres-
sions better if they can be made aware of their motivations, and that such
awareness raising works best for intermediate learners (beginners have too
little vocabulary at their disposal, advanced learners refrain too much from
taking risks) with an analytic and imager cognitive style. In Boers experi-
ments, analytic learners were clearly able to distinguish literal from gurative
usage, whereas learners with a more holistic style found it harder to identify
source domains of metaphors. Imagers were better than verbalisers because
they could associate novel gurative expressions more easily with mental
pictures or concrete scenes. There is no comparable research for grammar
teaching but as meaning stays the most important factor, one might assume
that also in grammar teaching, analytic and imager learners might benet
more than others.
50 Susanne Niemeier

example the past tense and conditionals; or aspect and the distinction
between mass nouns and count nouns). This will allow them to establish
meaningful neural connections and come up with grounded hypotheses
for the usage of these grammatical phenomena.
Combining Langackers approach to tense with Mental Space Theory
allows learners to trace the thought processes behind utterances as shown
before and to determine which mental space(s) the utterances refer to,
whether the meaning is temporal and/or epistemic and therefore arrive at
a correct interpretation, which should ideally also serve as a model for the
learners own future utterances in the foreign language. Learners should
start in the here-and-now, (i.e., in base space), and work their way up to
other mental spaces, while detecting the meanings of and the relations
between the various tense and aspect uses. Concerning the teaching of
aspect, the notion of boundaries may come in helpful, especially when
introduced via the teaching of mass and count nouns (see also Niemeier
2008).
Tense and aspect are present from the very beginning of ESL learning,
not necessarily in an explicit way but at least in the teachers language. As
soon as learners start producing their own utterances, they will use verbs
and therefore also tense and aspect. In other words, CG-oriented foreign
language instruction should ideally begin with the very rst lesson in the
foreign language because only then can it unfold its full potential.

References

Achard, Michel & Susanne Niemeier (Eds.)


2004 Cognitive linguistics, second language acquisition, and foreign lan-
guage teaching. Berlin/New York: Mouton de Gruyter.
Alejo-Gonzalez, Rafael
2010 Making sense of phrasal verbs: A cognitive linguistic account. In
Jeannette Littlemore & Constanze Juchem-Grundmann (Eds.),
AILA Review 23: Applied cognitive linguistics and second lan-
guage learning and teaching, 5071. Amsterdam: Benjamins.
Bhat, D. N. Shankara
1999 The prominence of tense, aspect and mood. Amsterdam: Ben-
jamins.
Boers, Frank
2004 Expanding learners vocabulary through metaphor awareness:
What expansion, what learners, what vocabulary? In Michel
Achard & Susanne Niemeier (Eds.), Cognitive linguistics, second
language acquisition, and foreign language teaching, 211232.
Berlin/New York: Mouton de Gruyter.
Cognitive perspective 51

Comrie, Bernard
1976 Aspect. An introduction to the study of verbal aspect and related
problems. Cambridge: Cambridge University Press.
Dekel, Nurit
2010 A matter of time: Tense, mood and aspect in spontaneous spoken
Israeli Hebrew. Ph.D. Thesis, University of Amsterdam. Avail-
able at: http://dare.uva.nl/document/174501. Oct. 17 (2 August,
2011).
De Knop, Sabine & Teun De Rycker (Eds.)
2008 Cognitive approaches to pedagogical grammar. Berlin/New York:
Mouton de Gruyter.
De Wit, Astrid & Frank Brisard
2009 Expressions of epistemic contingency in the use of the English
present progressive. Papers of the Linguistic Society of Belgium
4. Available at: http://webh01.ua.ac.be/linguist/SBKL/sbkl2009/
dew2009.pdf. Oct. 17 (13 August, 2011).
Dirven, Rene
2001 English phrasal verbs: Theory and didactic application. In Martin
Putz, Susanne Niemeier & Rene Dirven (Eds.), Applied Cogni-
tive Linguistics I: Theory and language acquisition, 328. Berlin/
New York: Mouton de Gruyter.
Doiz-Bienzobas, Aintzane
1995 The preterite and the imperfect in Spanish: Past situation versus
past viewpoint. San Diego, CA: University of California-San
Diego, Ph.D. dissertation.
Doiz-Bienzobas, Aintzane
2002 The preterite and the imperfect as grounding predications. In
Frank Brisard (Ed.), Grounding: The epistemic footing of deixis
and reference, 299347. Berlin: Mouton de Gruyter.
Evans, Vyvyan & Melanie Green
2006 Cognitive linguistics. An introduction. Edinburgh: Edinburgh Uni-
versity Press.
Fauconnier, Gilles
1994 Mental spaces: Aspects of meaning construction in natural lan-
guage. Cambridge: Cambridge University Press.
Fauconnier, Gilles
1997 Mappings in thought and language. Cambridge: Cambridge Uni-
versity Press.
Fauconnier, Gilles & Mark Turner
2002 The way we think. Conceptual blending and the minds hidden
complexities. New York: Basic Books.
Fleischman, Suzanne
1990 Tense and narrativity. London: Routledge.
52 Susanne Niemeier

Gibbs, Raymond W.
2006 Embodiment and cognitive science. New York: Cambridge Uni-
versity Press.
Gisborne, Nikolas
1996 English perception verbs. London: University College London,
Ph.D. dissertation.
Gurevich, Olga
2010 Conditional constructions in English and Russian. In Hans C.
Boas (Ed.), Contrastive studies in Construction Grammar, 87102.
Amsterdam: Benjamins.
Horner, Marion, Elizabeth Daymond, Jennifer Baer-Engel & Peter Lampater
2010 Green Line 16. Stuttgart: Klett.
Ibarretxe Antunano, Iraide
1999 Polysemy and metaphor in perception verbs: A cross-linguistic
study. Ph.D. Thesis, University of Edinburgh. Available at
http://www.unizar.es/linguisticageneral/articulos/Ibarretxe-
PhD-Thesis-99.pdf. Oct. 17 (3 August, 2011).
Klein, Wolfgang
1994 Time in language. London/New York: Routledge.
Klein, Wolfgang
1995 A simplest analysis of the English tense-aspect system. In Wolfgang
Riehle & Hugo Keiper (Eds.), Anglistentag 1994 Graz: Proceedings
of the Conference of the German Association of University Teachers
of English, 139152. Tubingen: Niemeyer.
Klein, Wolfgang
2009 How time is encoded. In Wolfgang Klein & Ping Li (Eds.), The
expression of time, 3981. Berlin/New York: Mouton de Gruyter.
Klein, Wolfgang & Ping Li (Eds.)
2009 The expression of time. Berlin/New York: Mouton de Gruyter.
Kurtyka, Andrzej
2001 Teaching English phrasal verbs: A cognitive approach. In Martin
Putz, Susanne Niemeier & Rene Dirven (Eds.), Applied Cognitive
Linguistics I: Theory and language acquisition, 87116. Berlin/
New York: Mouton de Gruyter.
Lako, George
1987 Women, re, and dangerous things. Chicago: The University of
Chicago Press.
Langacker, Ronald W.
1987a Foundations of cognitive grammar. Vol. 1: Theoretical prerequi-
sites. Stanford: Stanford University Press.
Langacker, Ronald W.
1987b Nouns and verbs. Language 63(1). 5394.
Langacker, Ronald W.
1991a Concept, image, and symbol: The cognitive basis of grammar.
Berlin/New York: Mouton de Gruyter.
Cognitive perspective 53

Langacker, Ronald W.
1991b Foundations of Cognitive Grammar. Vol. II: Descriptive applica-
tion. Stanford: Stanford University Press.
Langacker, Ronald W.
2000 Grammar and conceptualization. Berlin/New York: Mouton de
Gruyter.
Langacker, Ronald W.
2001 Cognitive linguistics, language pedagogy, and the English present
tense. In Martin Putz, Susanne Niemeier & Rene Dirven (Eds.),
Applied Cognitive Linguistics I: Theory and language acquisition,
339. Berlin/New York: Mouton de Gruyter.
Langacker, Ronald W.
2002 Langacker, Ronald W. (2002), Remarks on the English ground-
ing systems. In Frank Brisard (Ed.), Grounding: The epistemic
footing of deixis and reference, 2938. Berlin/New York: Mouton
de Gruyter.
Langacker, Ronald W.
2008a Cognitive Grammar: A basic introduction. Oxford: Oxford Uni-
versity Press.
Langacker, Ronald W.
2008b Cognitive Grammar as a basis for language instruction. In Peter
Robinson & Nick C. Ellis (Eds.), Handbook of cognitive linguis-
tics and second language acquisition, 6688. New York/London:
Routledge.
Langacker, Ronald W.
2008c The relevance of Cognitive Grammar for language pedagogy. In
Sabine de Knop & Teun de Rycker (Eds.), Cognitive approaches
to pedagogical grammar, 735. Berlin/New York: Mouton de
Gruyter.
Langacker, Ronald W.
2009 Investigations in Cognitive Grammar. Berlin/New York: Mouton
de Gruyter.
Langacker, Ronald W.
2011 The English present: Temporal coincidence versus epistemic
immediacy. In Adeline Patard & Frank Brisard (Eds.), Cognitive
approaches to tense, aspect and epistemic modality, 4586. Amster-
dam: Benjamins.
Llopis Garca, Reyes
2010 Why Cognitive Grammar works in the L2 classroom: A case
study of mood selection in Spanish. In Jeannette Littlemore &
Constanze Juchem-Grundmann (Eds.), AILA Review 23: Applied
cognitive linguistics and second language learning and teaching, 72
94. Amsterdam: Benjamins.
Lopez Garca, Angel
2005 Gramatica Cognitiva para profesores de espanol L2. Madrid: Arco.
54 Susanne Niemeier

Marras, Valentina & Teresa Cadierno


2008 Spanish gustar versus English like: A cognitive analysis of the con-
structions and its implication for SLA. In Andrea Tyler, Yiyoung
Kim & Mari Takada (Eds.), Language in the context of use:
Cognitive and usage-based approaches to language and language
learning, 233252. Berlin/New York: Mouton de Gruyter.
Niemeier, Susanne
2008 The notion of boundedness/unboundedness in the foreign language
classroom. In Frank Boers & Seth Lindstromberg (Eds.), Cognitive
Linguistic approaches to teaching vocabulary and phraseology,
309327. Berlin/New York: Mouton de Gruyter.
Niemeier, Susanne & Monika Reif
2008 Progress isnt simple teaching English tense and aspect to
German learners. In Sabine de Knop & Teun de Rycker (Eds.),
Cognitive approaches to pedagogical grammar, 325255. Berlin/
New York: Mouton de Gruyter.
Putz, Martin, Susanne Niemeier & Rene Dirven (Eds.)
2001 Applied Cognitive Linguistics I: Theory and language acquisition.
Berlin/New York: Mouton de Gruyter.
Radden, Gunter
1992 The cognitive approach to natural language. In Martin Putz
(Ed.), Thirty years of linguistic evolution, 513541. Amsterdam:
Benjamins.
Radden, Gunter & Rene Dirven
2007 Cognitive English grammar. Amsterdam/Philadelphia: John Ben-
jamins.
Robinson, Peter & Nick C. Ellis (Eds.)
2008 Handbook of Cognitive Linguistics and second language acquisi-
tion. New York/London: Routledge.
Salaberry, M. Rafael
2008 Marking past tense in second language acquisition: A theoretical
model. London: Continuum.
Schmiedtova, Barbara & Monique Flecken
2008 Aspectual concepts across languages: Some considerations for
second language learning. In Sabine de Knop & Teun de Rycker
(Eds.), Cognitive approaches to pedagogical grammar, 357384.
Berlin/New York: Mouton de Gruyter.
Shirai, Yasuhiro & Roger W. Andersen
1995 The acquisition of tense-aspect morphology. A prototype account.
Language. Vol. 71(4). 743762.
Shirai, Yasuhiro
2002 The prototype hypothesis of tense-aspect acquisition in second
language. In Rafael Maximo Salaberry (Ed.), The L2 acquisition
of tense-aspect morphology, 455493. Amsterdam: Benjamins.
Cognitive perspective 55

Smith, Carlota
2009 A speaker-based approach to aspect. In Helen Aristar-Dry,
Richard P. Meier & Emilie Destruel (Eds.), Text, time, and con-
text, 724. Dordrecht: Springer.
Taylor, John R.
2002 Cognitive Grammar. Oxford/New York: Oxford University Press.
Tomasello, Michael
2005 Constructing a language A usage-based theory of language
acquisition. Cambridge: Harvard University Press.
Tyler, Andrea & Vyvyan Evans
2001 The relation between experience, conceptual structure and mean-
ing: Non-temporal uses of tense and language teaching. In Martin
Putz, Susanne Niemeier & Rene Dirven (Eds.), Applied cognitive
linguistics I: Theory and language acquisition, 63105. Berlin/
New York: Mouton de Gruyter.
Tyler, Andrea & Vyvyan Evans
2003 The semantics of English prepositions: Spatial scenes, embodied
meaning and cognition. Cambridge: Cambridge University Press.
Tyler, Andrea & Vyvyan Evans
2004 Applying Cognitive Linguistics in pedagogical grammar: The
case of over. In Michel Achard & Susanne Niemeier (Eds.),
Cognitive linguistics, second language acquisition, and foreign
language teaching, 257280. Berlin/New York: Mouton de
Gruyter.
Tyler, Andrea
2008 Cognitive Linguistics and second language instruction. In Peter
Robinson & Nick C. Ellis (Eds.), Handbook of cognitive linguistics
and second language acquisition, 456488. New York/London:
Routledge.
Tyler, Andrea, Charles M. Mueller & Vu Ho
2010 Applying Cognitive Linguistics to instructed L2 learning: The
English modals. In Jeannette Littlemore & Constanze Juchem-
Grundmann (Eds.), AILA Review 23: Applied cognitive linguistics
and second language learning and teaching, 3049. Amsterdam:
Benjamins.
Ungerer, Friedrich, Peter Pasch, Peter Lampater & Rosemary Hellyer-Jones
1992 Learning English: A guide to grammar. Stuttgart: Klett.
Vendler, Zino
1967 Linguistics in Philosophy. Ithaca, NY: Cornell University Press.
Viberg, Ake
1984 The verbs of perception: A typological study. In Brian Butter-
worth, Bernard Comrie & Osten Dahl (Eds.), Explanations for
language universals, 123162. Berlin: Mouton de Gruyter.
56 Susanne Niemeier

Williams, Christopher
2002 Non-progressive and progressive aspect in English. Fasano: Schena
editore.
Ziemke, Tom, Jordan Zlatev & Roslyn M. Frank (Eds.)
2007 Embodiment. Berlin/New York: Mouton de Gruyter.

Acknowledgement

I wish to thank Fred Thompson (University Koblenz-Landau, Germany)


for proofreading the manuscript as well as my two reviewers for their
valuable comments and suggestions. All remaining aws are, as usual,
my own responsibility.
Chapter 2
The Spanish preterite and imperfect from a cognitive
point of view1

Aintzane Doiz

1. Introduction

The goal of this chapter is to provide a semantic characterization of the


imperfect (henceforth, IMP) and of the preterite (henceforth, PRET) in
Spanish, two past forms that have attracted a lot of attention in the litera-
ture. In particular, two parameters will be used to account for the forms:
the distinction between the actual occurrence reading and the property read-
ing proposed by Doiz-Bienzobas (1995, 2002), and the distinction between
real versus virtual events put forward by Langacker (2001b, 2008, 2009). It
will be shown that this analysis captures the subtle dierences between the
meanings of the two forms in context and accounts for the preference in the
choice or the acceptability of one of the two forms in discourse.
The paper is organized as follows. Section 2 summarizes some of the
main assumptions of Cognitive Grammar (especially Langacker 1987,
1991) and introduces the analytic tools that will be used throughout this
chapter. In addition, it provides an overview of the English present and
past tenses based on Langacker (1991, 2009) that will be relevant for the
purpose of this chapter. Section 3 proposes an analysis of the PRET and
the IMP in terms of the actual occurrence reading and the property reading
that will serve to account for the readings and the acceptability of the forms
in combination with certain predicates. Section 4 introduces Langackers
characterization of virtual versus non-virtual events (2001a, 2009). This char-
acterization will be applied in Section 5 to the understanding of the use of the
IMP to talk about scheduled future situations, generics, speakers expect-

1. The results presented in this paper are part of the research project IT31110
(Department of Education, University and Research of the Basque Govern-
ment) and the UFI11/06 (the University of the Basque Country UPV/EHU). I
would like to thank the editors of this volume and two anonymous reviewers
for their comments on an earlier draft of this chapter. Any errors or mistakes
are, however, entirely my responsibility.
58 Aintzane Doiz

ations, and irrealis. Section 6 summarizes the main points of the analysis
of the IMP and the PRET proposed in the chapter.

2. Prerequisites and a cognitive account of the English present and


past tense

In the rst part of this section, I introduce some basic notions that have
been developed within Cognitive semantics/Cognitive Grammar during
the last quarter century and which will be helpful for the understanding of
this chapter.2 In the second part, I provide a simplied description of the
characterization of the present and the past tense provided by Langacker
within Cognitive Grammar.

2.1. Prerequisites
Speakers may conceptualize and portray the same situation in dierent
ways; this ability is referred to as construal. Since every linguistic expres-
sion incorporates a particular way of construing the conceptual content it
evokes, construal is [. . .] part of the conventional semantic value of lexical
and grammatical elements (Langacker 2009: 14). Thus, from the Cogni-
tive semantics perspective, meaning is identied with conceptualization,
i.e., mental experience, which is ultimately to be explicated in terms of
cognitive processing (Langacker 1991: 5).
The notion of construal has four dimensions: specicity, perspective,
scope, and prominence. In this chapter we focus on two dimensions of
construal: scope an aspect of perspective and proling a kind of
prominence. Scope is the extent of the conceptual content an expression
invokes as the basis for its meaning (Langacker 2009: 14). For instance,
in order to conceptualize physical objects, the domain of space has to be
evoked since objects exist in space, occupy a space, and their size and
shape are also manifested in space. But only the spatial expanse sucient
for the object to manifest in it is required; we do not need to bring the
entire universe within our scope of conception. In the case of the notion
table, for example, the delimited spatial region invoked to support it is its
spatial scope or spatial domain as illustrated in Figure 1.

2. See Chapter 1 for a summary of the main assumptions within Cognitive


Linguistics.
Cognitive perspective on Spanish 59

Figure 1. Spatial scope and space for the notion table

Within a domain, we can often distinguish between the maximal scope


(MS) and the immediate scope (IS). The MS stands for the overall content
of an expression and the IS is the portion within the MS which is relevant
to identify the meaning of an expression. For instance, the IS of elbow is
an arm and its MS is the body as a whole. Both scopes are necessary
because it is impossible to characterize the notion of elbow without invoking
the notion of arm, which in turns invokes the notion of body. Finally, the
conceptual content of an expression is also characterized by the proling
a kind of prominence and the prole it invokes (i.e., the entity it desig-
nates or refers to). Following the example of elbow, we can say that elbow
proles a specic section of the arm, namely, the bend of the arm between
the forearm and the upper arm represented with the heavy lines in Figure 2.

Figure 2. Maximal scope (MS), immediate scope (IS), and prole in the notion
elbow
60 Aintzane Doiz

Nouns and verbs prole dierent kinds of entities. Nouns prole a kind
of a thing in an abstract sense. By contrast, verbs prole an event, a rela-
tionship that unfolds through time.3 Relationships may designate a bounded
event or process, as in the case of perfective predicates, or an unbounded
process, as in the case of imperfective predicates (Langacker 2008, 2009).
Imperfective predicates correspond to stative verbs (e.g., to be, to love)
within Vendlers categorization. Imperfectives lack an intrinsic beginning
or ending, they are internally homogenous and contractible (any subpart
of the process is an instance of the process), and they are non-replicable
(when two instances of an imperfective are combined, a larger instance
of the process is obtained). Perfective predicates subsume the categories
of achievement, accomplishment, and activity verbs proposed by Vendler
(1967). They designate dynamic processes with natural boundaries. They
are non-homogenous, i.e., they have dierent stages: in the case of the
perfective predicate to write a letter, for instance, there are usually four
dierent stages: writing the name of the addressee, starting the letter,
writing some paragraphs, and nishing the letter. The stages are not con-
tractible (all the stages that constitute the process need to be present in
order to have an instance of the process) and they are replicable, that is
to say, two successive acts of the process designated by a perfective predi-
cate constitute two instances of the process type rather than a bigger or
longer instance of the process (e.g., two instances of letter writing result
in two dierent letters.)4
Figures 3a and 3b contain the representation of an abstract perfective
and an abstract imperfective predicate, respectively (Langacker 2008, 2009:
19). In Figure 3a, the perfective process is represented by the rectangle with
the heavy lines within the IS. In Figure 3b, the imperfective process is
represented by the proled section of the rectangle; the beginning and the
ending of the process within the IS are not proled. The two kinds of pro-
cesses develop through time as indicated by the arrow, the time line t.

3. See Langacker (1987, 2009) for a more detailed account.


4. Activities such as to sing and to eat are perfective in that they naturally occur
in bounded episodes but they are internally homogeneous (Langacker 2009:
22).
Cognitive perspective on Spanish 61

Figure 3a. Perfective process

Figure 3b. Imperfective process

2.2. The English present tense and the past tense


Tense morphemes impose an immediate temporal scope and direct the
speakers attention to a particular span of time relative to a speech event
(Langacker 1991, 2009: 23). The value of the present tense is that of
temporal coincidence: the situation is coextensive with the speech event,
the here-and-now of the speaker. The function of the past tense is to
impose an immediate temporal scope that is prior to the time of speaking
62 Aintzane Doiz

(Langacker 2009: 22).5 Figure 4a contains the representation of a present


tense morpheme with a perfective process (e.g., *He writes a book) and
an imperfective process (e.g., He knows Italian). Figure 4b contains the
representation of a past tense morpheme with a perfective process (e.g.,
He wrote a book) and an imperfective process (e.g., He knew Italian). The
speech event is represented by the box with the squiggly lines over the time
line (t).

Figure 4a. The present tense

Figure 4b. The past tense

5. These are the characterizations of the prototypical temporal uses of the English
present tense and past tense (Langacker 1991, 2001b, 2009). Langacker (2009)
and Brisard (1999, 2001, 2002) argue that a more general characterization of
the two tenses is epistemic.
Cognitive perspective on Spanish 63

On the one hand, perfectives do not usually occur with the English
present tense (e.g., *He writes a book). Langacker argues that the ungram-
maticality is due to durational and epistemic reasons. Firstly, the dura-
tional problem refers to the fact that processes designated by perfectives
are, generally speaking, too long to coincide with the speech time: a
sentence like He writes a book may be uttered in a very short time, but the
actual duration of the process it designates is longer than the time it takes
to be uttered. Secondly, the epistemic motivation derives from the fact
that by the time the addressee has heard enough of the utterance to be
able to recognize the process designated, it is too late to make the beginning
of the event coincide with its description at the speech event.6 On the other
hand, imperfectives may occur with the present tense (He knows Italian)
because any proled subpart of the process is equivalent to the other and,
consequently, the part of the imperfective process that coincides with the
speech time is in itself an instance of the process type (Figure 4a). Both,
perfectives (He wrote a book) and imperfectives (He knew Italian) occur
with the past tense. In the case of perfectives with the past tense, the entire
situation is contained within the IS, which means that the situation may be
apprehended or viewed as a whole from the speech time by the speaker
and the hearer (Figure 4b). There are no durational or epistemic constraints.
In the case of imperfectives with the past tense, the situation is designated
without its beginning or ending.

3. The past tense in Spanish: Actual occurrence reading versus


property reading

Like the past tense in English, the Spanish PRET and IMP impose an
immediate temporal scope that is prior to the time of speaking. However,
the PRET and the IMP construe the situations they modify in a dierent
way. I propose the following characterizations:7 (I) Predicates with the
PRET evoke the actual occurrence reading; they state that something

6. However, perfectives may occur with the present tense in the case of performa-
tives (I order you to go now) or schedules (The train leaves at 6.00) because
there are no durational or epistemic incompatibilities in these two contexts.
7. These characterizations dier in some respects to the ones proposed in Doiz-
Bienzobas (1995, 2002). However, the notions of actual occurrence versus
property do not dier from the original proposal. Leonetti (2004) also refers
to the notion of property for the characterization of the IMP.
64 Aintzane Doiz

happened or that something has changed. Hence, the PRET evokes the
occurrence of an actual bounded event which started and nished at some
point in the past, as illustrated in (1a). It may also designate a bounded
state in combination with some imperfective predicates (1b).
(1) a. Juan condujo muy rapido ayer.
Juan drove-PRET very quickly yesterday.
Juan drove very quickly yesterday.
b. Juan estuvo aqu (ya no).
Juan was-PRET here (not anymore).
(II) Predicates with the IMP evoke the property reading. Therefore, unlike
the PRET, the IMP construes the situation it modies as an unbounded
state of aairs or an unbounded property which is generally stated of
the subject of the sentence. In (2a) Juan had the property of being a fast
driver; in (2b) Juan had the property of being Swedish.

(2) a. Juan conduca muy rapido.


Juan drove-IMP too quickly.
Juan was a fast driver.
b. Juan era sueco.
Juan was-IMP Swedish.
It is up to the speaker and hearer to decide which construal, the actual
occurrence reading which is evoked by the PRET, or the property reading,
associated with the IMP, best ts her communicative intention and the
situation being talked about.
In 3.1, I illustrate the actual occurrence reading imposed by the PRET
and the property reading associated with the IMP in combination with
imperfective and perfective predicates. In Section 3.2, I account for the
acceptability judgements of perfectives and imperfectives with the PRET
and the IMP.

3.1. The IMP and the PRET in contrast


3.1.1. Imperfectives PRET/IMP
Imperfective predicates may occur with the IMP and the PRET, but the
interpretation of the sentences they appear in is very dierent. Consider
(3a) and (3b), where the imperfective predicate to be interesting, an intrin-
Cognitive perspective on Spanish 65

sically unbounded homogeneous state, may occur with the PRET or with
the IMP.
(3) a. La pelcula fue interesante (?por lo menos en teora).
The movie was-PRET interesting (?at least in theory).
b. La pelcula era interesante (por lo menos en teora).
The movie was-IMP interesting (at least in theory).
When the PRET is used (3a), the actual occurrence reading surfaces. The
sentence evokes the occurrence of an event, namely, the showing of the
movie, as conrmed by the unacceptability of adding the phrase at least
in theory, which explicitly negates the occurrence of the event (Doiz-
Bienzobas 1995, 2002). When the IMP is used (3b), however, there need
not be a showing of the movie, and consequently, the phrase at least in
theory is ne. The state of being interesting is construed as an unbounded
property of the subject, the movie.

3.1.2. Perfectives PRET/IMP


As in the case of imperfectives, perfective predicates (i.e., non-stative predi-
cates) may be modied by the PRET and the IMP. However, the construals
evoked by the two predicates are very dierent, as illustrated by (4) and (5).
When the PRET is used, speaker A wants to know what happened on
Tuesday in terms of weather (actual occurrence reading): whether it rained
or not. Speaker B answers that it rained on Tuesday morning, but that the
weather was good in the afternoon. The perfective predicate, hacer bueno
to make good in combination with the PRET designates the occurrence
of an action (it made good).
(4) Speaker A:
? ?
Que tiempo hizo el martes? Llovio?
What made-PRET (was) the weather (like) on Tuesday? Did it
rain-PRET?
Speaker B:
S, por la manana llovio, pero a la tarde hizo bueno.
Yes, in the morning it rained-PRET, but in the afternoon it
made-PRET (was) good (the weather was good).
By contrast, when the perfective predicate to make good is modied by
the IMP (5), the predicate does not respond to the question What happened
66 Aintzane Doiz

weatherwise on Tuesday? In this case, the predicate designates a property


attributed to the topic. The sentence it made-IMP good describes an
unbounded property or a state of aairs similar to the one evoked by the
sentences it was rainy or it was cold. The property reading surfaces and it
serves as the background for the occurrence of the event We went to the
beach:8
(5) Speaker A:
?
Que hicisteis el martes?
What did you do-PRET on Tuesday?
Speaker B:
Como el martes haca bueno, fuimos a la playa.
Since the weather made-IMP (was) good on Tuesday, we went-
PRET to the beach.
The analysis of the use of the IMP in (5) is compatible with Ducrots anal-
ysis of the sentence provided in (6) (Ducrot 1979: 6):
(6) El ano pasado en Pars haca/hizo calor.
Last year it made-IMP/PRET (was) hot in Paris.
Ducrot (1979) proposed that predicates with the IMP designate a property
that is applicable to the topic as a whole. In the case of the sentence with
the IMP in (6), the speaker is saying that it was hot all year round in Paris.
On the other hand, when the PRET is used, the property designated by the
predicate may be assigned to a subpart of the topic: it may be the case that
it was hot just some days that year. The subpart reading is supported by
the fact that the PRET, and not the IMP, is normally used when there is
explicit reference to some subparts of the temporal frame, e.g., three times
within a week in (7) (sentence provided by Leonetti 2004: 497):
(7) La semana pasada, ?llova/llovio tres veces.
Last week, it rained-?IMP/PRET three times.

8. The occurrences of backgrounded situations in the IMP and of foregrounded


situations in the PRET have long been noted in the literature. Hopper and
Thompson (1980) established a correlation between backgrounding and atelicity
(unboundedness), and foregrounding and telicity (boundedness). Fleischman
(1989, 1990) also stated that the function of the French imparfait, similar to
the Spanish IMP, is to provide backgrounded information at the textual level
(see Chapter 9, for a discussion of foreground and background).
Cognitive perspective on Spanish 67

Leonetti (2004: 497) derived the global property reading associated with
the IMP from the imperfective nature of the IMP.9 He argued that, since
the IMP construes the predicate it modies as an unbounded and homo-
geneous property (i.e., as an imperfective), it is unnatural to assign the
property to a part. Consequently, the property is stated of the whole tem-
poral frame in which the situation holds.
The characterization of the IMP that has been proposed here also ac-
counts for the so-called imperfecto narrativo (IMP of narration) and
its two variants, the imperfecto biograco (IMP of biographies) and the
imperfecto de ruptura (IMP of breakage).10 These uses of the IMP are
generally attested at the beginning and at the end of narratives, where the
function of backgrounding or the framing of the narration has greater
prominence (Fernandez Ramrez 1986). Consider the sentences in (8a) and
(8b), which illustrate the imperfecto biograco and the imperfecto de ruptura,
respectively:
(8) a. A los tres anos, el general mora de una forma misteriosa.
Three years later, the general died-IMP in a mysterious way.
b. Media hora despues la herida paraba de sangrar.
Half an hour later the wound stopped-IMP bleeding.
The use of the IMP results in the understanding of the predicate as a state
of aairs or a property predicated of a temporal topic: the temporal topic
of three years later in (8a) is characterized by the generals death; in (8b)
the temporal topic of half an hour later is characterized by a property of
the wound. The use of the PRET in these sentences would result in the
actual occurrence reading whereby the predicates state that something
happened at a particular time in the past.

3.2. The unacceptability of the PRET and the IMP


While the IMP and the PRET may occur both with imperfectives and
perfectives (Section 3.1.), there are contexts in which only one of the two
forms is possible. As stated by Fernandez Ramrez (1986: 281): Parece

9. Leonetti proposed that IMP is the equivalent of individual-level predicates


(Carlson 1977; Kratzer 1989; Diesing 1992) in the verbal system. For reasons
of space, I will not discuss this point in detail.
10. Bertinetto (1986: 392 and beyond) and Vetters (1996) also discuss these uses
of the corresponding Italian and French IMP, respectively. See Fernandez
Ramrez (1986) and Gomez (2002) for more information on the narrative
uses of the IMP.
68 Aintzane Doiz

pues que existe cierta concurrencia del imperfecto y del preterito. Pero esta
libertad esta limitada en la mayora de los casos y determinada por una
serie de factores que sera necesario jar caso por caso.11
According to grammar books and course books for students of Spanish,
imperfective predicates (i.e., states) tend to occur with the IMP and perfec-
tive predicates (i.e., activities, accomplishments, and achievements) with
the PRET. In other words, intrinsically unbounded situations (i.e., imper-
fectives) occur with the IMP because it portrays the situations it modies
as unbounded; intrinsically bounded situations (i.e., perfectives) occur with
the PRET because it portrays the situations it modies as bounded. Hence,
the unacceptability of the IMP or the PRET in certain contexts is attributed
to the clash between the intrinsic boundedness of the situation designated
by the predicate and the boundedness of the situation evoked by the PRET
and the IMP. Under this account, the predicate to be of age is unaccept-
able with the PRET in (9) because the PRET imposes some boundaries to
the intrinsically unbounded situation. By contrast, the perfective predicate,
to write a letter, is unacceptable with the IMP because the IMP portrays
the intrinsically bounded situation as unbounded (10).

Imperfective ?PRET/IMP
(9) Como ?fue/era mayor de edad, condujo el coche.
Since he was-?PRET/IMP of age, he drove the car.
Perfective PRET/?IMP
(10) Ayer escrib/?escriba una carta.
Yesterday I wrote-PRET/?IMP a letter.
However, the analysis of the compatibility of the IMP/PRET in context
in terms of boundedness does not account for the unacceptability of the
IMP with some imperfective predicates, and for the unacceptability of the
PRET with some perfective predicates. I consider these two cases in the
remainder of this section.

Imperfectives *IMP
As we have already seen, imperfective predicates may normally occur with
the IMP (2b and 9), but this is not always the case as illustrated in (11),

11. Hence, there seems to be some overlap between the IMP and the PRET. But
the possibility of using either form is limited in the majority of the cases and is
conditioned by some factors that need to be determined case by case (trans-
lation provided by the author of the chapter).
Cognitive perspective on Spanish 69

where the predicate to be those three cannot be modied by the IMP;


and thus it has to appear in the PRET.12
(11) Speaker A:
Who did it?
Speaker B:
Fueron/?Eran esas tres.
It was-PRET/?IMP those three (girls).
Speaker A knows that somebody did something, but she does not know
who it was, so she asks speaker B (who did it?). Since the sentence evokes
an actual occurrence reading (i.e., something happened), speaker B codes
the missing information in the PRET (it was-PRET those three girls).
By contrast, the use of the IMP is unacceptable in this context because it
portrays the three women as the entities of whom a property is stated, a
construal which is not an appropriate reply to the question posed by
speaker A.
Similarly, (12) also illustrates the unacceptability of an imperfective
predicate with the IMP and its acceptability with the PRET with an im-
perfective verb when the actual occurrence reading is the only appropriate
construal of the situation in the context.
(12) Fue/?Era inteligente: no contesto.
She was-PRET/?IMP intelligent: she did not reply.
When the PRET is used, the rst proposition states that the subject did
something that made him seem intelligent. The proposition after the colon
species what that action was, namely, not replying. Since the two pro-
positions have a coherence relation, they result in an acceptable discourse.
However, when the IMP is used, the predicate modied by the IMP states
that the subject was characterized by the property of being intelligent. The
clause after the colon states that the subject did not say a word. Since
there is no logical coherence between the two propositions, the sentence
is unacceptable.13

12. A similar example is analyzed in Doiz-Bienzobas (1995, 2002).


13. As noted by Brisard (2010: 493), the idea that there should be a coherence
relation between two subsequent units in a discourse is a requirement which
can be formulated in terms of Gricean maxims or relevance conditions (Sperber
and Wilson 1995).
70 Aintzane Doiz

Perfectives *PRET
Counter to the predictions made by an analysis of the PRET in terms of
boundedness, sometimes perfective predicates may not be modied by the
PRET, as illustrated in (13):14
(13) La carta *dijo/deca lo mucho que le quera.
The letter said-*PRET/IMP how much he loved her.
When the PRET is used, the predicate said how much he loved her desig-
nates the occurrence of an event in which an agent did something: the letter
uttered some words in the past. Under normal circumstances, only people
may carry out this activity, so the unacceptability of the sentence in (13)
with the PRET, where the letter is responsible for the utterance, comes as
no surprise. When the IMP is used, however, the letter is not construed as
actually talking. The predicate designates a property, which is attributed
to the subject: the letter was characterized by having the message of how
much he loved her. This sentence portrays a pragmatically feasible scenario,
and hence, it is acceptable.15
Last but not least, the analysis proposed in this chapter accounts for the
general tendency for imperfectives to occur with the IMP and for perfectives
to occur with the PRET, as explained by the traditional analyses in terms
of boundedness. Consider (9) and (10), repeated here under (14) and (16).
(14) Como ?fue/era mayor de edad, condujo el coche.
Since he was-?PRET/IMP of age, he drove the car.
When the IMP is used, the predicate being of age is construed as a
property attributed to the subject, and this property explains his being
able to drive the car: since he was of age, he drove the car. The IMP is
acceptable. The PRET, on the other hand, construes the predicate being
of age as an actual occurrence (i.e., something happened: he became of
age). Therefore the reading of the sentence is: Since he became of age, he
drove the car. The lack of a coherence relation between the two proposi-

14. A similar example is analyzed in Doiz (1995, 2002) in more detail.


15. As one of the editors of this volume pointed out, predicates similar to to say
such as estipular to stipulate, detallar to specify, establecer to establish
may take an inanimate subject in the PRET. In these cases, the predicate
designates an activity that may be carried out by the inanimate subject. I
suggest that the selectional restrictions regarding the animacy of the subjects
of these verbs are dierent from the ones associated with the verb to say.
Cognitive perspective on Spanish 71

tions renders the PRET unacceptable. In order for the PRET to be accept-
able, a context where the predicate needs to be construed as an actual
occurrence is required, as in (15):
(15) Cuando fue mayor de edad, condujo el coche.
When he was-PRET of age, he drove the car.
When he became of age, he drove the car.
In (15), the temporal conjunction when marks the beginning of the situa-
tion becoming of age. The speaker states that the subject was able to drive
when something happened, namely, when the subject became of age (actual
occurrence reading). Consequently, in this case, the PRET is grammatical.
Finally, the so-called preference of perfectives to occur with the PRET
is also accounted for under the characterizations of the PRET and the
IMP provided here. In (16) the PRET is the preferred choice for the per-
fective predicate to write a letter: the perfective situation is perceived as
an actual occurrence. In order for the IMP be acceptable, the perfective
predicate has to be embedded in a discourse in which the property reading
is appropriate, such as in literary contexts where the narrative uses of the
IMP are found, as discussed in Section 3.1.
(16) Ayer escrib/?escriba una carta.
Yesterday I wrote-PRET/?IMP a letter.
As a summary, in this section I have discussed the dierences in the inter-
pretation of the sentences whose predicates take the PRET and the IMP,
and I have accounted for the acceptability judgements of the two forms
with various imperfective and perfective predicates. In both tasks, the
characterization of the PRET and the IMP in terms of the actual occurrence
reading versus property reading was crucial. In Section 4, I introduce the
distinction between virtual and non-virtual events proposed by Langacker
(2003, 2008, 2009), which will be shown to be relevant for the understand-
ing of the IMP and the PRET in Section 5.

4. Actual versus virtual events in English

Linguistic expressions may reect the direct description of an actual or


represented event, or they may code a virtual or representing event of
the actual event. The former reects the default apprehension or viewing
arrangement of a situation by the speaker and hearer (Figure 5a); the latter
portrays a special apprehension or viewing arrangement (Figure 5b) (Lan-
gacker 2001a, 2009):
72 Aintzane Doiz

Figure 5a. Default viewing arrangement of an event

Figure 5b. Special viewing arrangement of an event

On the one hand, the default viewing arrangement occurs when the speaker
and hearer observe and report on actual occurrences. On the other hand,
the special viewing arrangement is taken when what is linguistically coded
is not an actual or represented event but a representing or virtual event.
Scheduled future events, generics, and the historical present are all mental
constructions involving instances of virtual or representing events.
Generics, which include general truths, laws of nature, and established
social practices, such as the ones illustrated in (17) and (18) (sentences
from Langacker 2009: 28), describe how the world is supposed to work
(Goldsmith and Woisetschlaeger 1982; Langacker 2009). That is to say,
generics describe structural generalizations based on what is common to
actual occurrences; they do not describe the actual occurrences themselves.
Cognitive perspective on Spanish 73

Consequently, when speakers state a generalization, they are describing a


virtual or representing situation, which captures the common aspects of all
the corresponding actual situations. When the generalization is accessible at
the speech event, the present tense is used, as in (17). When the generaliza-
tion is accessible at a past point in time, the past tense is used (18).
(17) A man proposes to a woman. [Thats how its done.]
(18) In those days, a man proposed to a woman. [Now anything goes.]
Figures 6 and 7 contain the representations of present singular generics and
past singular generics, respectively. The boxes which include the circles
containing the letters m, w, and the arrow stand for the events (or the rela-
tionships) of a man proposing to a woman (m ! w). The various instances
of the events of a man proposing to a woman from which the generaliza-
tion is drawn are the actual situations. The generalizations are represented
by the proled boxes with the heavy lines within the IS. The empty boxes
to the left and to the right of the boxes with the heavy lines stand for other
(unspecied) generalizations that the speaker may have. It is important to
note that, whereas the actual situations are not located at specic points in
time, the generalizations (i.e., the virtual generic situations) are. In the case
of present generics, the generalization coincides with the time of speaking
(the box with the squiggly lines), as represented in Figure 6; in the case of
past singular generics, the generalization is in eect in the past, as repre-
sented in Figure 7.

Figure 6. Present singular generics (Langacker 2009: 28)


74 Aintzane Doiz

Figure 7. Past singular generics

Like generics, scheduled situations also involve virtual events. A schedule


comprises a series of virtual events, each the mental representation of an
anticipated actual event (Langacker 2009: 27). Consider (19) and (20) which
designate a present and a past schedule, respectively:
(19) We have to hurry. The plane leaves in ten minutes. (Langacker
1999: 94, 2001a: 31, 2001b: 268, 2003: 22, 2009: 26)
(20) She was rushing through the airport. The plane left in ten minutes.
(Langacker 2009: 26)
Scheduled situations evoke two events. The anticipated event, the planes
actual leaving, which may or may not take place, and the virtual or the
representing event, that is, the knowledge that the speaker has of the
anticipated event. It is the virtual event that is linguistically coded, not
the anticipated situation. When the time in which the speaker apprehends
the virtual event coincides with the time of speaking, the virtual situation is
coded in the present tense in English, as illustrated in (19). When the speaker
refers to a mental schedule that was in eect at an earlier time with respect
to the speech event, the past tense is used in English (Langacker 2009: 37),
as illustrated in (20). Hence, the scheduled future use of the present tense
and of the past tense are sketched in Figures 8 and 9, respectively.
Cognitive perspective on Spanish 75

Figure 8. Scheduled future situations in the present (Langacker 2009: 9)

Figure 9. Scheduled future situations in the past

The proled boxes (i.e., the boxes with the heavy lines) within the IS
represent the schedule, the knowledge or virtual situation the speaker has,
namely, the situation The plane leaves in ten minutes in the cases of (19)
and (20). This knowledge is accessible at the speech time when the present
tense is used, as in (19) (Figure 8), or at a time prior to the speech time
when the past tense is used, as in (20) (Figure 9). The boxes to the left
76 Aintzane Doiz

and to the right of the box in bold represent other potential virtual events,
which have not or were not accessed by the speaker. The boxes with the
dotted lines are the anticipated actual events which correspond to the
virtual events: in (19) and (20) they stand for the future situation, namely,
for the planes actual departure, which may or may not take place.16
It is important to bear in mind that genericity and scheduling are not
part of the meaning of the present tense or the past tense; they are not a
meaning of these tenses. Genericity and scheduling are mental constructs
found in a particular context that are coded in the present tense or in the
past tense in accordance to the prototypical temporal value of the tenses,
as discussed in Section 2.2. (Langacker 2009: 28).

5. Virtual events in Spanish and the IMP

In Spanish, habituals, generics, and schedules occur with the IMP. Simi-
larly, speakers expectations and the expression of irrealis are also coded
in the IMP. Drawing from the analysis of genericity and scheduling pro-
vided by Langacker (Section 4), I propose that the IMP is compatible
with virtuality.17

Past singular generics/past habituals


In Spanish, the reading of past genericity and habituality surfaces with the
IMP (21a), not with the PRET (21b):
(21) a. [Antes] un barbero sacaba muelas. [Ahora ya no].
[Before] a barber took-IMP out back teeth. [Not any more].
b. Un barbero saco muelas.
A barber took-PRET out back teeth.

16. The participants of the situations and the relationship linking them (i.e., the
plane leaving in ten minutes) have not been represented within the proled/
non-proled boxes in Figures 8 and 9.
17. See Brisard (2010: 505) for an analysis of the French imparfait as a marker of
a kind of virtual reality: The French imparfait either shifts the viewpoint to
the past, thereby virtualizing it (making the viewpoint and situation virtually
available at the time of speaking), or shifts it to another space than the actual
one, also virtual (which may then be interpreted in context as hypothetical
etc.).
Cognitive perspective on Spanish 77

On the one hand, when the IMP is used (21a), the speaker is referring to
a generalization (i.e., to a virtual event) that is extracted from the com-
monality among the actual events in the past; the speaker is not referring
to a particular instance of the situation. On the other hand, when the
PRET is used (21b), the situation designated by the sentence is the actual
occurrence of an event in the past: the speaker is referring to a specic
time when a barber took out some back teeth. Since the generic/habitual
reading surfaces with the IMP, it is concluded that the IMP is compatible
with the linguistic coding of virtual events.

Scheduled/planned future situations


Scheduled or planned future situations have been referred to as anticipated
(Doiz-Bienzobas 1995, 2002) or prospective situations/readings (Leonetti
and Escandell-Vidal 2003) in the literature. These situations are coded in
the IMP in Spanish, as illustrated by (22a) and (22b):
(22) a. Al ano siguiente haba estas. (Pero se cancelaron/Fueron
divertidas).
The following year there were-IMP some festivities.
The following year there were going to be some festivities.
(But they got cancelled/They were a lot of fun).
b. El tren sala a las 6. (Se estropeo/Salio en punto)
The train left-IMP at 6 oclock.
The train was leaving (was scheduled to leave) at 6 oclock.
(It broke down/It departed on time).
In (22a) and in (22b) the speaker is referring to a situation which has been
planned/scheduled to take place at a time in the future with respect to
a past reference time; the speaker does not state whether the situation
has taken place or not (They were a lot of fun/It departed on time; They
were cancelled/It broke down). Following Langackers characterization of
scheduled future situations, it is proposed that what is coded linguistically
in (22a) and (22b) is the virtual plan, the schedule regarding the future
occurrence of the situation, rather than the actual future situation itself.
In this view, the IMP is compatible with past virtual situations.
By contrast, when the PRET is used, the anticipated reading or scheduled
reading does not surface. The sentences in (23a) and (23b) designate a situa-
tion which actually occurred in the past as shown by the unacceptability of
adding But they were cancelled (23a) or It didnt leave (23b). There is no
reference to a scheduled or to an anticipated event:
78 Aintzane Doiz

(23) a. Al ano siguiente hubo estas. (?Pero se cancelaron/Fueron


divertidas).
The following year there were-PRET festivities.
The following year some festivities took place.
(?But they got cancelled/They were a lot of fun).
b. El tren salio a las 6. (Llegamos a tiempo/?No salio).
The train left-PRET at 6 oclock.
(We arrived in time to catch it/?It didnt leave).
The mental construction of a schedule also underlies the use of the so-
called imperfecto de error (the imperfect of mistakes) illustrated in (24).
This use is discussed in Grijelmo (2006: 254) and in some Spanish grammar
books.
?
(24) El conyuge sorprendido le dijo a su pareja: No venas manana
de Italia?
The surprised husband told his wife: Not come-IMP tomorrow
from Italy?
The surprised husband told his wife: Werent you coming back
tomorrow from Italy?
In (24), the husband expresses his surprise at the early return of his wife by
asking: Werent you coming back tomorrow from Italy? At the time in
which the question is asked, the speaker knows that his wife is not coming
home on the following day because she is already home. In this case, the
speaker is accessing his knowledge about the scheduled date of return of
his wife. The question is posed to get his wifes conrmation of the validity
of his schedule, of the virtual event. That is why the IMP is used. Since it
may be the case that the speaker is not necessarily wrong (his schedule
could have been right but the actual event did not follow the timing stated
in the schedule, as in the present example), the label the imperfect of
mistakes is not quite appropriate. The speaker is merely seeking to have
his schedule or piece of knowledge conrmed. As in the previous cases,
this reading is not part of the semantic contribution of the IMP.

The speakers expectations


The speakers expectations are the set of beliefs that the speaker has regard-
ing the way things should have been in the past or should be in the present
or in the future. Since expectations are not actual situations, but rather, they
Cognitive perspective on Spanish 79

are virtual events with corresponding potential actual situations, the IMP
is used. Compare (25a) and (25b):
(25) a. ?
Donde estuviste ayer, Juan?
Where were-PRET you yesterday, Juan?
b. ?
Donde estabas ayer, Juan?
Where were-IMP you yesterday, Juan?
When the PRET is used (25a), the speaker wants to know where the
addressee, Juan, was at a specic point in the past. The speaker is access-
ing an actual past situation. In this case, the speaker does not have any
pre-conceived idea of where Juan was or should/could have been, and
she expects a straight forward answer such as I was at the movies, I went
shopping, etc. No further explanation is required or expected from the
hearer. By contrast, when the IMP is used (25b), the question posed by
the speaker is not a neutral question. In this case the speaker is referring
to her expectation regarding the hearers location at some moment in the
past; the question refers to some knowledge or a virtual idea. By question-
ing her knowledge or expectation, the speaker implies that the expectation
is not met and, consequently, she is asking the hearer to explain to provide
a reason. The process is something like:

Given some piece of information/knowledge I had, I expected to see you


at the party ! You were not at the party ! I am asking you where were
you given the fact that you were not where you were expected? ! Why
werent you there?

The emergence of certain implications that are present when the IMP is
used is a desired eect that is exploited by speakers as illustrated by the
dialogue in (26a). Lets imagine that a crime has been committed. A police
ocer, speaker A, is conducting an investigation and is cross-examining a
suspect, speaker B. The choice of the IMP or the PRET reects a dierent
understanding of the situation, as reected by the police ocers attitude
towards speaker B. If speaker A uses the IMP, speaker B is the suspect of
some wrong-doing (26a); if she uses the PRET, speaker B is not a suspect
for the time being (26b). The answers provided by speaker B reect the
specic scenarios associated with the sentences with the PRET and the
IMP, respectively:
80 Aintzane Doiz

(26) a. ?
Speaker A: Que haca ayer a las 3 en el despacho?
( ! Por que estaba all?)
?
Speaker B: Tena que terminar un informe y me quede hasta tarde.
Speaker A: What did you do-IMP yesterday at 3 oclock at the
oce? ( ! Why were you there?)
Speaker B: I had-IMP to nish a report and I stayed late in the
oce.
b. ?
Speaker A: Que hizo ayer a las 3 en el despacho?
Speaker B: Termine un informe.
Speaker A: What did you do-PRET yesterday at 3 oclock in
the oce?
Speaker B: I nished-PRET a report.
When the PRET is used (26b), the ocer is enquiring about the specic
activities that the suspect carried out at his oce at a certain time. Speaker
B provides a straightforward reply. However, when the IMP is used (26a),
speaker A is accessing her expectations; she asks speaker B what he was
doing at the oce at a time when he should have been home (according
to her knowledge/expectations). The process is something like:

Yesterday you were at the oce at three in the night. I know that this is
unusual; people are usually at home at that time. So ! what were you doing
there that could not wait until the morning? ! Why were you there?

Since the reply given by speaker B in (26a) designates an actual event


(I stayed late in the oce) that does not coincide with the police ocers
expectations, speaker B feels the need to add an explanation, I had to
nish a report, a planned situation which is coded in the IMP.

Intentions/Wishes
An intention is a volition that is intended to be carried out; a wish is a
hope or a desire for something. Frequently intentions are based on wishes,
and wishes are translated into volitions. For our purposes, it is not impor-
tant to draw a line separating one kind of event from the other. What is
important, though, is to acknowledge the virtual nature of the wish or
the intention designated.
In Spanish, present wishes or intentions occur with the present tense
(27a); and past wishes/intentions with the IMP (27b):
Cognitive perspective on Spanish 81

(27) a. Espero ir a Hawai de vacaciones en julio.


I intend (hope) to go to Hawaii on holidays in July.
b. Yo esperaba ir a Hawai de vacaciones en julio.
I hoped-IMP (it was my intention) to go on holidays to
Hawaii in July.
In (27a), the speaker is accessing a virtual event, which describes her wish
at the time of speech; the corresponding actual event is located in the
future. When the IMP is used (27b), the sentence designates a past wish;
it is a virtual event with a corresponding actual occurrence, which may
have taken place or not.

Irrealis
The mental constructs that have been analyzed in the preceding sections
(e.g., schedules, generalizations, expectations, and wishes) involve virtual
events that are part of the speakers accepted reality (Langacker 2009: 31).
In the remainder of this section, I consider the use of the IMP for the
linguistic coding of some virtual events that are not part of the speakers
reality, namely, hypotheticals, dreams, and childrens play games, which
all belong in irrealis.
The relationship between the past tense and non-actuality or irrealis has
been widely acknowledged in the literature (see Chapter 1; Brisard 2010;
Doiz-Bienzobas 1995, 2002; Fleishman 1989; Hutchison 1985; Langacker
1987; and Steele 1975, among others):
Distance in the past frequently serves to remove a real event from actuality/
reality, relocating it in a hypothetical world. It should therefore come as
no surprise to nd languages using the PAST to refer to (present or time-
less) situations predicated as occurring in explicitly hypothetical or imagi-
nary worlds, e.g., those of dreams and of childrens make-believe games
(Fleischman 1989: 14).

As a result, it has been argued that the basic meaning of the past tense
should be understood as distance from reality. Langacker (1991: 245,
2009), for example, provides a comprehensive account of tense in English
where the present and past tenses designate the proximal/distal contrast in
the epistemic sphere.18
In the case of languages with two past forms like Spanish, the IMP, and
not the PRET, is normally used to express irrealis. It has been proposed

18. Brisard (2010) provides an account of the French imparfait along the same lines
lines too.
82 Aintzane Doiz

that the choice of the IMP in this context results from the unboundedness
or imperfectivity of the situations designated by the predicates with the
IMP (Leonetti 2004),19 whereas others have attributed this correlation to
the backgrounding function of the IMP as opposed to the foregrounding
role of the PRET (Hopper 1979). I propose that the use of the IMP to
convey irrealis is accounted for by an analysis that proposes the com-
patibility of the IMP with virtuality. I will consider a few cases in the
remainder of this section.

a) Dreams
Dreams are part of irrealis and are in eect at a time prior to the speech
time. They are virtual (i.e., non-actual) events that are told through recall.
In Castilian Spanish, dreams are normally coded in the IMP, as illustrated
in (28):
(28) Sone que ganaba/*gane la lotera y que me compraba/*compre
una bicicleta.
I dreamed that I won-IMP/*PRET the lottery and that I
bought-IMP/*PRET a bicycle.
It should be noted that the retelling of a past event through recall does not
license the use of the IMP, as illustrated by the ungrammaticality of the
IMP to designate a real situation in (29).
(29) Ayer me acorde de que Juan hizo/*haca la mili en Madrid.20
Yesterday I remembered that Juan did-PRET/*IMP the compulsory
military service in Madrid.
Yesterday I remembered that Juan performed on the compulsory
military service in Madrid.

19. Leonetti (2004) proposes that the anaphoric value of the IMP, which derives
from its imperfectivity, requires the IMP to be anchored to a (temporal) refer-
ence point that has already been introduced in the discourse. Accordingly, the
use of the IMP in dreams, counterfactuals, and sentences such as the ones
I am discussing in this section is a side-eect of the anaphoric character of
the IMP: the predicate modied by the IMP is interpreted relative to a
temporal or other kind of frame which has already been introduced. In the
present cases, it could be a dream, a movie, or a hypothetical space. By con-
trast, the PRET is a non-anaphorical form. It introduces its own temporal
reference in reality.
20. The use of the IMP in (29) is grammatical when the sentence with the IMP
takes the scheduled future reading, that is to say, when it involves a virtual
situation. The grammaticality of this reading is predicted by the analysis pro-
posed here.
Cognitive perspective on Spanish 83

In order for the IMP to be grammatical, the situation has to be virtual, as


in (28). When virtuality (e.g., dreams) and reality are mixed in the same
discourse, linguistic coding can keep them apart, as illustrated in (30a)
(Doiz-Bienzobas 1995, 2002).

(30) a. En el sueno, la senora que trajo el libro el otro da era mi ta.


In the dream, the lady who brought-PRET the book the other
day was my aunt.
b. En el sueno, la senora que traa el libro (*el otro da) era mi ta.
In the dream, the lady who brought-IMP the book (*the other
day) was my aunt.

In (30a) and (30b) the fact that the lady is part of the dream is reected by
the use of the IMP in the copula to be. In (30a) the PRET is used to
designate an event that is part of reality, namely, the lady who brought
the book. This is an event that took place at some point in the past as
specied by the time adverbial the other day. By contrast, the IMP is
used when the event the lady who brought the book is stated to be part of
the dream. Accordingly, the event cannot be modied by the temporal
adverbial the other day (30b).

b) Hypothetical sentences
The use of the past tense morphology to refer to epistemic distance in a
hypothetical space accounts for the occurrence of the past tense in counter-
factual sentences in English (see Chapter 1; Langacker 2009). In Spanish,
the past subjunctive, a past tense, is also used in the protasis of counter-
factual hypothetical sentences. The verb in the apodosis is modied by
the conditional and/or the IMP, as illustrated by the sentence in (31):

(31) Si tuviera dinero, me comprara/compraba una casa.


If I had-PAST SUBJ money, I buy-COND/IMP a house.
If I had money, I would buy a house.
Intuitively speaking, there is a sense in which the sentence is more assertive
with the IMP than with the conditional (Fernandez Ramrez 1986; Porcar
Miralles 1993). I propose that the dierence in the degree of assertiveness
between the two forms results from the fact that when the IMP is used, a
plan/schedule in a hypothetical space is designated. When the conditional
84 Aintzane Doiz

is used, however, a prediction is designated. Plans/schedules are less tenta-


tive than predictions because the former evoke situations that are arranged
beforehand and that are meant to be followed. The grammaticality judg-
ments of the IMP and the conditional in (32) support this idea:
(32) Si viajaras mas, no *estabas/estaras aburrida.
If you travelled more, you were-*IMP/COND not bored.
If you travelled more, you would not be bored.
When the IMP is used, the state of being bored is perceived as a scheduled/
planned situation. Since it is unlikely to schedule this kind of state, the
IMP is ruled out. But when the conditional is used, the situation of being
bored is perceived as a prediction that will come true if the condition stated
in the subordinate clause is met. Since it is possible to predict being bored
in a particular situation, the use of the conditional is feasible.
Finally, our account predicts that the PRET may not be used in hypo-
thetical sentences because it construes the situation it modies as an actual
past situation that is incompatible with the hypothetical space evoked by
the protasis as in (33):
(33) Si tuviera dinero, me *compre una casa.
If I had money, I bought-*PRET a house.
In order for the PRET to be grammatical, the condition expressed in the
protasis has to be met in reality, as it is the case in one of the readings
of (34).
(34) Si Juan estuvo en Pars, robo el banco.
If Juan was in Paris, he robbed-PRET the bank.
In (34) the speaker does not know whether Juan was in Paris or not. But
she knows that if the condition of being in Paris is met in reality, then the
fact that he robbed the bank is also true. That is to say, when the condi-
tion in the protasis of the hypothetical sentence is met in the reality space,
the apodosis is satised in the actual world too. In this case, there is no
irrealis and the PRET is used.

c) Childrens play games


The IMP is also used in the coding of the preparations undertaken by
children before they start role-playing games. This use has long been noted
in the literature (Doiz-Bienzobas 1995; Fernandez Ramrez 1986: 275;
Cognitive perspective on Spanish 85

Gomez 2002: 149; Hernandez 1996: 433) and has been referred to as
imperfecto ludico or the imperfecto de fantasa (35):
(35) ?Jugamos? Tu eras el polica y yo el ladron y entonces tu robabas un
banco. . .
Shall we play? You were-IMP the policeman and I the thief and
then you robbed-IMP a bank. . .
When children play pretend games, they do a lot of planning regarding
who plays which character and regarding the course of the events that
are going to be acted out. In other words, they have to provide a plan or
script that will have to be agreed on prior to their game. In fact, scripts
are very similar to schedules: a script is something like a schedule, one
dierence being that the represented events can occur at any time [. . .]
(Langacker 2009: 27). Scripts, like schedules, are also virtual situations,
and therefore, the IMP is used.

6. Conclusions

In this chapter, I have provided a characterization of the Spanish IMP and


PRET in terms of two sets of distinctions, the actual occurrence/property
reading and the virtuality/nonvirtuality distinction. The PRET is stated to
construe a situation as a bounded occurrence that took place at a moment
in the past. Therefore, situations with the PRET are anchored in reality.
The IMP portrays a situation as an unbounded property or a state of
aairs, which may be compatible with the expression of virtual mental
constructs.
These characterizations have allowed us to account for the following
facts: (a) the tendency for perfectives to occur with the PRET and some
imperfectives with the IMP; (b) the unacceptability of certain predicates
with the PRET or the IMP; (c) the dierences in the interpretations of
the sentences with the IMP and the PRET; and (d) the use of the IMP to
refer to dierent kinds of situations such as schedules, generics, expecta-
tions, or dreams.
The characterizations that I have provided in this chapter will hopefully
help students of Spanish to grasp the dierences in the semantics of the
two forms and to master their use. At a more abstract level, the usefulness
of cognitive semantic constructs has been supported.
86 Aintzane Doiz

References

Bertinetto, Pier
1986 Tempo, aspetto e azione nel verbo italiano. Florence: Accademia
della Crusca.
Brisard, Frank
1999 A critique of localism in and about tense theory. Antwerp: Uni-
versity of Antwerp, Ph.D. dissertation.
Brisard, Frank
2001 Be going to: An exercise in grounding. Journal of Linguistics 37.
251285.
Brisard, Frank
2002 The English Present. In Frank Brisard (Ed.), Grounding: The
epistemic footing of deixis and reference, 251298. Berlin/New
York: Mouton de Gruyter.
Brisard, Frank
2010 Aspects of virtuality in the meaning of the French imparfait. Lin-
guistics 48(2). 487524.
Carlson, Gregory
1977 A unied analysis of the English bare plural. Linguistics and
Philosophy 1. 413457.
Diesing, Molly
1992 Indenites. Cambridge, Mass.: MIT Press.
Doiz-Bienzobas, Aintzane
1995 The preterite and the imperfect in Spanish: Past situation versus
past viewpoint. San Diego, CA: University of California, San
Diego, Ph.D. dissertation.
Doiz-Bienzobas, Aintzane
2002 The preterite and the imperfect as grounding predications. In
Frank Brisard (Ed.), Grounding: The epistemic footing of deixis
and reference, 299348. Berlin/New York: Mouton de Gruyter.
Ducrot, Oswald
1979 Limparfait en francais. Linguistische Berichte 60. 123.
Fernandez Ramrez, Salvador
1986 Gramatica espanola. 4. El verbo y la oracion. Madrid: Grafur.
Fleischman, Suzanne
1989 Temporal distance: A basic linguistic metaphor. Studies in Lan-
guage 13. 150.
Fleischman, Suzanne
1990 Tense and narrativity, from medieval performance to modern c-
tion. Austin: University of Texas Press.
Goldsmith, John & Erich Woisetschlaeger
1982 The logic of the English progressive. Linguistic Inquiry 13. 79
89.
Cognitive perspective on Spanish 87

Gomez, Leonardo
2002 Gramatica didactica del espanol. 8th edition. Madrid: SM.
Grijelmo, Alex
2006 La gramatica descomplicada. Madrid: Taurus.
Hernandez, Cesar
1996 Gramatica funcional del espanol. 3rd edition. Madrid: Gredos.
Hopper, Paul
1979 Aspect and foregrounding in discourse. In Givon Talmy (Ed.),
Syntax and semantics. Discourse and syntax, 213242. London:
Academic Press.
Hopper, Paul J. & Sandra A. Thompson
1980 Transitivity in grammar and discourse. Language 56. 251299.
Hutchinson, Chris
1985 The where of the when: Some arguments for a non-temporal
interpretation of tense. Papers in Linguistics 21/22. 135.
Kratzer, Angelika
1989 Stage and individual level predicates: Papers on quantication,
NSF Grant Report, Department of Linguistics. Amherst: Univer-
sity of Massachusetts.
Langacker, Ronald W.
1987 Foundations of Cognitive Grammar. Volume I: Theoretical pre-
requisites. Stanford, CA: Stanford University Press.
Langacker, Ronald W.
1991 Foundations of Cognitive Grammar. Volume II: Descriptive appli-
cation. Stanford, CA: Stanford University Press.
Langacker, Ronald W.
1999 Virtual reality. Studies in the Linguistic Sciences 29. 77103.
Langacker, Ronald W.
2001a Cognitive Linguistics, language pedagogy and the English pres-
ent tense. In Martin Putz, Susanne Niemeier & Rene Dirven
(Eds.), Applied Cognitive Linguistics I: Theory and language acqui-
sition, 339. Berlin/New York: Mouton de Gruyter.
Langacker, Ronald W.
2001b The English present tense. English Language and Linguistics:
251272.
Langacker, Ronald W.
2003 Extreme subjectication: English tense and modals. In Hubert
Cuyckens, Thomas Berg, Rene Dirven & Klaus-Uwe Panther
(Eds.), Motivation in language: Studies in honor of Gunter Radden,
326. Amsterdam: John Benjamins.
Langacker, Ronald W.
2008 Cognitive Grammar. A basic introduction. Oxford: Oxford Uni-
versity Press.
88 Aintzane Doiz

Langacker, Ronald W.
2009 Investigations in Cognitive Grammar. Berlin/New York: Mouton
de Gruyter.
Langacker, Ronald, W.
2011 The English present: Temporal coincidence versus epistemic
immedicacy. In Frank Brisard & Adeline Patard (Eds.), Cogni-
tive approaches to tense, aspect and epistemic modality, 4586.
Amsterdam/Philadelphia: John Benjamins.
Leonetti, Manuel & Victoria Escandell
2003 On the quotative readings of Spanish imperfecto. Cuadernos de
Lingustica X. 135154.
Leonetti, Manuel
2004 Por que el imperfecto es anaforico. In Luis Garca Fernandez &
Bruno Camus Bergareche (Eds.), El preterito imperfecto, 481
507. Madrid: Gredos.
Porcar Miralles, Margarita
1993 La oracion condicional. La evolucion de los esquemas verbales con-
dicionales desde el Latn al Espanol actual. Castello: Universitat
Jaume I.
Sperber, Dan & Deirdre Wilson
1995 Relevance: Communication and cognition, 2nd edition. Oxford:
Blackwell.
Steele, Susan
1975 Past and irrealis: Just what does it all mean? International Journal
of American Linguistics 41. 200217.
Vendler, Zeno
1967 Linguistics in Philosophy. Ithaca, N.Y.: Cornell University Press.
Vetters, Cari
1996 Temps, aspect et narration. Amsterdam/Atlanta: Rodopi.
Chapter 3
Frequency-based grammar and the acquisition of
tense and aspect in L2 learning1

Nick Ellis

1. Frequency and language cognition

The last 50 years of psycholinguistic research has demonstrated language


processing to be exquisitely sensitive to usage frequency at all levels of
language representation: phonology and phonotactics, reading, spelling,
lexis, morphosyntax, formulaic language, language comprehension, gram-
maticality, sentence production, and syntax (Ellis 2002a). Language knowl-
edge involves statistical knowledge, so humans learn more easily and pro-
cess more uently high frequency forms and regular patterns which are
exemplied by many types and which have few competitors. Psycholinguistic
perspectives thus hold that language learning is the implicit associative learn-
ing of representations that reect the probabilities of occurrence of form-
function mappings. Frequency is a key determinant of acquisition because
rules of language, at all levels of analysis from phonology, through syn-
tax, to discourse, are structural regularities which emerge from learners
lifetime unconscious analysis of the distributional characteristics of the
language input.
It is these ideas which underpin the last 30 years of investigations of
language cognition using connectionist and statistical models (Christiansen
and Chater 2001; Elman et al. 1996; Rumelhart and McClelland 1986),
the competition model of language learning and processing (Bates and
MacWhinney 1987; MacWhinney 1987b, 1997), the investigation of how
frequency and repetition bring about form in language and how probabilis-
tic knowledge drives language comprehension and production (Bod, Hay,
and Jannedy 2003; Bybee and Hopper 2001; Ellis 2002a, 2002b; Jurafsky
2002; Jurafsky and Martin 2000), and the proper empirical investigations
of the structure of language by means of corpus analysis.

1. The author thanks Rafael Salaberry and Llorenc Comajoan for their construc-
tive editing of this chapter.
90 Nick Ellis

Frequency, learning, and language come together in usage-based ap-


proaches which hold that we learn linguistic constructions while engaging
in communication, the interpersonal communicative and cognitive pro-
cesses that everywhere and always shape language (Chapter 1; Slobin 1997).
Constructions are form-meaning mappings, conventionalized in the speech
community, and entrenched as language knowledge in the learners mind.
They are the symbolic units of language relating the dening properties of
their morphological, syntactic, and lexical form with particular semantic,
pragmatic, and discourse functions (Bates and MacWhinney 1987; Bybee
2008; Croft 2001; Croft and Cruise 2004; Goldberg 1995, 2003, 2006;
Lako 1987; Langacker 1987; Robinson and Ellis 2008; Tomasello 2003).
Goldbergs (2006) Construction Grammar argues that all grammatical
phenomena can be understood as learned pairings of form (from mor-
phemes, words, idioms, to partially lexically lled and fully general phrasal
patterns) and their associated semantic or discourse functions: the network
of constructions captures our grammatical knowledge in toto, i.e., its con-
structions all the way down (Goldberg 2006: 18). Such beliefs, increasingly
inuential in the study of child language acquisition, have turned upside
down generative assumptions of innate language acquisition devices, the
continuity hypothesis, and top-down, rule-governed, processing, bringing
back data-driven, emergent accounts of linguistic systematicities. Construc-
tionist theories of child language acquisition use dense longitudinal corpora
to chart the emergence of creative linguistic competence from childrens
analyses of the utterances in their usage history and from their abstraction
of regularities within them (Goldberg 1995, 2003, 2006; Tomasello 2003,
1998). Children typically begin with phrases whose verbs are only conserva-
tively extended to other structures. A common developmental sequence is
from formula to low-scope slot-and-frame pattern, to creative construction.

2. Frequency and concept learning

It is human categorization ability that provides the most persuasive testa-


ment to our incessant unconscious tallying of associations. We know that
natural categories are fuzzy rather than monothetic. Wittgensteins (1953)
consideration of the concept game showed that no set of features that we
can list covers all the things that we call games, ranging as the exemplars
variously do from soccer, through chess, bridge, and poker, to solitaire.
Instead, what organizes these exemplars into the game category is a set of
family resemblances among these members son may be like mother, and
Frequency-based grammar 91

mother like sister, but in a very dierent way. And we learn about these
families, like our own, from experience. Exemplars are similar if they
have many features in common and few distinctive attributes (features
belonging to one but not the other); the more similar are two objects on
these quantitative grounds, the faster are people at judging them to be
similar (Tversky 1977). Prototypes, exemplars which are most typical of
a category, are those which are similar to many members of that category
and not similar to members of other categories. Again, the operationalisa-
tion of this criterion predicts the speed of human categorization perfor-
mance people more quickly classify as birds sparrows (or other average
sized, average colored, average beaked, average featured specimens) than
they do birds with less common features or feature combinations like kiwis
or penguins (Rosch and Mervis 1975; Rosch et al. 1976).
Prototypes are judged faster and more accurately, even if they themselves
have never been seen before someone who has never seen a sparrow, yet
who has experienced the rest of the run of the avian mill, will still be fast
and accurate in judging it to be a bird (Posner and Keele 1970). Such eects
make it very clear that although people do not go around consciously
counting features, they nevertheless have very accurate knowledge of the
underlying frequency distributions and their central tendencies. Cognitive
theories of categorization and generalization show how schematic construc-
tions are abstracted over less schematic ones that are inferred inductively
by the learner in acquisition (Harnad 1987; Lako 1987; Taylor 1998).

3. Frequency and second language acquisition

Language learners, L1 and L2 both, share the goal of understanding lan-


guage and how it works. Since they achieve this based upon their experi-
ence of language usage, there are many commonalities between rst and
second language acquisition that can be understood from corpus analyses
of input and cognitive- and psycho-linguistic analyses of construction
acquisition following associative and cognitive principles of learning and
categorization. Therefore usage-based approaches, cognitive linguistics,
and corpus linguistics are increasingly inuential in L2 research too
(Collins and Ellis 2009; Ellis 1998, 2003; Ellis and Cadierno 2009; Robinson
and Ellis 2008), albeit with the twist that since they have previously devoted
considerable resources to the estimation of the characteristics of another
language the native tongue in which they have considerable uency
92 Nick Ellis

L2 learners computations and inductions are often aected by transfer,


with L1-tuned expectations and selective attention (Ellis 2006b) blinding
the acquisition system to aspects of the L2 sample, thus biasing their esti-
mation from naturalistic usage and producing the limited attainment that
is typical of adult L2 acquisition. In this view, L2 acquisition is dierent
from L1 acquisition in that it involves processes of construction and
reconstruction.

4. Construction learning as associative learning from usage

If constructions as form-function mappings are the units of language, then


language acquisition involves inducing these associations from experience
of language usage. Constructionist accounts of language acquisition thus
involve the distributional analysis of the language stream and the parallel
analysis of contingent perceptual activity, with abstract constructions being
learned from the conspiracy of concrete exemplars of usage following statis-
tical learning mechanisms (Christiansen and Chater 2001) relating input
and learner cognition. Psychological analyses of the learning of construc-
tions as form-meaning pairs is informed by the literature on the associa-
tive learning of cue-outcome contingencies, where the usual determinants
include: (1) factors relating to the form, such as frequency and salience; (2)
factors relating to the interpretation, such as signicance in the compre-
hension of the overall utterance, prototypicality, generality, and redun-
dancy; (3) factors relating to the contingency of form and function; and
(4) factors relating to learner attention, such as automaticity, transfer, over-
shadowing, and blocking (Ellis 2002a, 2003, 2006a, 2008b). These various
psycholinguistic factors conspire in the acquisition and use of any linguistic
construction.
These determinants of learning can be usefully categorized into factors
relating to (1) input frequency (type-token frequency, Zipan distribution,
recency), (2) form salience and perception, (3) prototypicality of meaning
and redundancy), and (4) contingency of form-function mapping.

4.1. Input frequency


4.1.1. Construction frequency
Frequency of exposure promotes learning. Ellis (2002) review illustrates
how frequency aects the processing of phonology and phonotactics, read-
ing, spelling, lexis, morphosyntax, formulaic language, language compre-
hension, grammaticality, sentence production, and syntax. That language
Frequency-based grammar 93

users are sensitive to the input frequencies of these patterns entails that
they must have registered their occurrence in processing. These frequency
eects are thus compelling evidence for usage-based models of language
acquisition that emphasize the role of input.

4.1.2. Type and token frequency


Token frequency counts how often a particular form appears in the input.
Type frequency, on the other hand, refers to the number of distinct lexical
items that can be substituted in a given slot in a construction, whether it is
a word-level construction for inection or a syntactic construction specify-
ing the relation among words. For example, the regular English past
tense -ed has a very high type frequency because it applies to thousands
of dierent types of verbs, whereas the vowel change exemplied in swam
and rang has much lower type frequency. The productivity of phonologi-
cal, morphological, and syntactic patterns is a function of type rather than
token frequency (Bybee and Hopper 2001). This is because: (a) the more
lexical items that are heard in a certain position in a construction, the less
likely it is that the construction is associated with a particular lexical item
and the more likely it is that a general category is formed over the items
that occur in that position; (b) the more items the category must cover, the
more general are its criterial features and the more likely it is to extend to
new items; and (c) high type frequency ensures that a construction is used
frequently, thus strengthening its representational schema and making it
more accessible for further use with new items (Bybee and Thompson
2000). In contrast, high token frequency promotes the entrenchment or
conservation of irregular forms and idioms; the irregular forms only survive
because they are high frequency. These ndings support languages place
at the center of cognitive research into human categorization, which also
emphasizes the importance of type frequency in classication.

4.1.3. Zipan distribution


In the early stages of learning categories from exemplars, acquisition is
optimized by the introduction of an initial, low-variance sample centered
upon prototypical exemplars (Elio and Anderson 1981, 1984). This low
variance sample allows learners to get a x on what will account for most
of the category members. The bounds of the category are dened later by
experience of the full breadth of exemplar types. Goldberg Casenhiser and
Sethuraman (2004) demonstrated that in samples of child language acqui-
sition, for a variety of verb-argument constructions (VACs), there is a
94 Nick Ellis

strong tendency for one single verb to occur with very high frequency in
comparison to other verbs used, a prole which closely mirrors that of
the mothers speech to these children.
In natural language, Zipf s law (Zipf 1935) describes how the highest
frequency words account for the most linguistic tokens: the constitutes
nearly 7% of the Brown Corpus of English usage, to more than 3%; while
about half the total vocabulary of about 50,000 words are hapax legomena:
words that occur only once in the corpus. If pf is the proportion of words
whose frequency in a given language sample is f, then pf P f b, with b Q 1.
Zipf (1949) showed this scaling relation holds across a wide variety of lan-
guage samples. Subsequent research has shown that many language events
(e.g., frequencies of phoneme and letter strings, of words, of grammatical
constructs, of formulaic phrases, etc.) across scales of analysis follow this
law (Ferrer i Cancho and Sole 2001, 2003). It has strong empirical support
as a linguistic universal and has important implications for language struc-
ture, use, and acquisition.
Goldberg et al. (2004) show that Zipf s law applies within VACs too,
and they argue that this promotes acquisition: tokens of one particular
verb account for the lions share of instances of each particular argument
frame; this pathbreaking verb also is the one with the prototypical mean-
ing from which the construction is derived (see also Ninio 1999, 2006). Ellis
and Ferreira-Junior (2009a, 2009b) investigate eects upon naturalistic
second language acquisition of type/token distributions in the islands
comprising the linguistic form of English verb-argument constructions
(VACs: VL verb locative, VOL verb object locative, VOO ditransitive) in
the ESF corpus (Perdue 1993). They show that in the naturalistic L2
acquisition of English, VAC verb type/token distribution in the input is
Zipan and learners rst acquire the most frequent, prototypical and
generic exemplar (e.g., put in VOL, give in VOO, etc.). Their work further
illustrates how acquisition is aected by the frequency and frequency distri-
bution of exemplars within each island of the construction (e.g., [Subj V
Obj Oblpath/loc]), by their prototypicality, and, using a variety of psycho-
logical (Shanks 1995) and corpus linguistic association metrics (Gries and
Stefanowitsch 2004; Stefanowitsch and Gries 2003), by their contingency of
form-function mapping. Ellis and Larsen-Freeman (2009a) describe con-
nectionist serial-recurrent network models of these various factors as they
play out in the emergence of constructions as generalized linguistic schema
from their frequency distributions in the input.
This fundamental claim that Zipan distributional properties of language
usage helps to make language learnable has thus begun to be explored
Frequency-based grammar 95

for these three verb argument constructions, at least. Ellis and ODonnell
(2012) are exploring its generality across a wide range of VACs in 100
million words of English.

4.1.4. Recency
Language processing also reects recency eects. This phenomenon, known
as priming, may be observed in phonology, conceptual representations,
lexical choice, and syntax (Pickering and Ferreira 2008). Syntactic priming
refers to the phenomenon of using a particular syntactic structure given
prior exposure to the same structure. This behavior has been observed
when speakers hear, speak, read, or write sentences (Bock 1986; Pickering
2006; Pickering and Garrod 2006). For L2 acquisition, Gries and Wul
(2005) provided evidence (a) that advanced L2 learners of English showed
syntactic priming for ditransitive (e.g., The racing driver showed the helpful
mechanic) and prepositional dative (e.g., The racing driver showed the torn
overall . . .) argument structure constructions in a sentence completion
task, (b) that their semantic knowledge of argument structure construc-
tions aected their grouping of sentences in a sorting task, and (c) that
their priming eects closely resembled those of native speakers of English
in that they were very highly correlated with native speakers verbal sub-
categorization preferences whilst completely uncorrelated with the sub-
categorization preferences of the German translation equivalents of these
verbs. There is now a growing body of research demonstrating such L2
syntactic priming eects (McDonough 2006; McDonough and Mackey
2006; McDonough and Tromovich 2008).

4.2. Form (salience and perception)


The general perceived strength of stimuli is commonly referred to as their
salience. Low salience cues tend to be less readily learned. Ellis (2006a,
2006b) summarized the associative learning research demonstrating that
selective attention, salience, expectation, and surprise are key elements in
the analysis of all learning, animal and human alike. As the Rescorla-
Wagner (1972) model encapsulates, the amount of learning induced from
an experience of a cue-outcome association depends crucially upon the
salience of the cue and the importance of the outcome.
Many grammatical meaning-form relationships, particularly those that
are notoriously dicult for second language learners like grammatical
particles and inections such as the third person singular -s of English,
are of low salience in the language stream. For example, some forms are
96 Nick Ellis

more salient: today is a stronger psychophysical form in the input than is


the morpheme -s marking third person singular present tense, thus while
both provide cues to present time, today is much more likely to be perceived,
and -s can thus become overshadowed and blocked, making it dicult for
L2 learners of English to acquire (Ellis 2006b, 2008a; Goldschneider and
DeKeyser 2001).

4.3. Prototypicality of meaning and redundancy


4.3.1. Prototypicality of meaning
Categories have graded structure, with some members being better exem-
plars than others. In the prototype theory of concepts (Rosch and Mervis
1975; Rosch et al. 1976), the prototype as an idealized central description
is the best example of the category, appropriately summarizing the most
representative attributes of a category. As the typical instance of a cate-
gory, it serves as the benchmark against which surrounding, less repre-
sentative instances are classied. The greater the token frequency of an
exemplar, the more it contributes to dening the category, and the greater
the likelihood it will be considered the prototype. The best way to teach
a concept is to show an example of it. So the best way to introduce a
category is to show a prototypical example. Ellis and Ferreira-Junior
(2009a) show that the verbs that L2 learners rst used in particular VACs
are prototypical and generic in function (go for VL, put for VOL, and
give for VOO). The same has been shown for child language acquisi-
tion, where a small group of semantically general verbs, often referred
to as light verbs (e.g., go, do, make, come) are learned early (Clark 1978;
Ninio 1999; Pinker 1989). Ninio argues that, because most of their seman-
tics consist of some schematic notion of transitivity with the addition of
a minimum specic element, they are semantically suitable, salient, and
frequent; hence, learners start transitive word combinations with these
generic verbs. Thereafter, as Clark describes, many uses of these verbs
are replaced, as children get older, by more specic terms. . . . General pur-
pose verbs, of course, continue to be used but become proportionately less
frequent as children acquire more words for specic categories of actions
(53). Notwithstanding the fact that prototypicality can help L2 learners
during the beginning stages of acquisition of complex, graded and fuzzy
concepts (such as tense-aspect meanings), the acquisition of the less proto-
typical exemplars of a complex concept remains an area of fertile research.
This is particularly relevant in the case of target items that can only be
concurrently dened at various levels of representation of language (e.g.,
Frequency-based grammar 97

lexical, morphosyntactic, discursive, and pragmatic at the same time). In


the sections below, we address this issue through the specic analysis of
specic tense-aspect meanings that could potentially be outside of the realm
of the basic concept.

4.3.2. Redundancy
The Rescorla-Wagner model (1972) also summarizes how redundant cues
tend not to be acquired. Not only are many grammatical meaning-form
relationships low in salience, but they can also be redundant in the under-
standing of the meaning of an utterance. For example, it is often unneces-
sary to interpret inections marking grammatical meanings such as tense
because they are usually accompanied by adverbs that indicate the tem-
poral reference. Second language learners reliance upon adverbial over
inectional cues to tense has been extensively documented in longitudinal
studies of naturalistic acquisition (Bardovi-Harlig 2000; Dietrich, Klein,
and Noyau 1995), training experiments (Ellis 2007; Ellis and Sagarra 2010),
and studies of L2 language processing (Van Patten 2006).

4.4. Contingency of form-function mapping


Psychological research into associative learning has long recognized that
while frequency of form is important, so too is contingency of mapping
(Shanks 1995). Consider how, in the learning of the category of birds,
while eyes and wings are equally frequently experienced features in the
exemplars, it is wings that are distinctive in dierentiating birds from
other animals. Wings are important features to learning the category of
birds, because they are reliably associated with class membership, eyes
are neither. Raw frequency of occurrence is less important than the con-
tingency between cue and interpretation. Distinctiveness or reliability of
form-function mapping is a driving force of all associative learning, to
the degree that the eld of its study has been known as contingency learn-
ing since Rescorla (1968) showed that for classical conditioning, if one
removed the contingency between the conditioned stimulus (CS) and the
unconditioned (US), preserving the temporal pairing between CS and US
but adding additional trials where the US appeared on its own, then animals
did not develop a conditioned response to the CS. This result was a mile-
stone in the development of learning theory because it implied that it was
contingency, not temporal pairing, that generated conditioned responding.
Contingency, and its associated aspects of predictive value, information
gain, and statistical association, have been at the core of learning theory
98 Nick Ellis

ever since. It is central in psycholinguistic theories of language acquisition


too (Ellis 2006a, 2006b, 2008b; Gries and Wul 2005; MacWhinney 1987b),
with the most developed account for L2 acquisition being that of the
Competition model (MacWhinney 1987a, 1997, 2001). Ellis and Ferreira-
Junior (2009b) use delta P and collostructional analysis measures (Gries
and Stefanowitsch 2004; Stefanowitsch and Gries 2003) to investigate
eects of form-function contingency upon L2 VAC acquisition. Boyd and
Goldberg (2009) use conditional probabilities to investigate contingency
eects in VAC acquisition. This is still an active area of inquiry, and more
research is required before we know which statistical measures of form-
function contingency are more predictive of acquisition and processing.

4.5. The many aspects of frequency and their research consequences


Interference with any of these aspects reduces learnability: constructions
of low salience of form are hard to learn, constructions where there is low
reliability or contingency between form and meaning are hard to learn, con-
structions with subtle construals yet to be discerned are hard to learn, and
constructions of low frequency of occurrence tend to be acquired later. Such
ndings suggest that the learning of linguistic constructions, like other con-
cepts, can be understood according to psychological principles of category
learning.

5. Applications of frequency-based grammar to the study of L2


tense and aspect

5.1. The Aspect Hypothesis


The study of tense and aspect (TA) has been a paradigm case in cognitive
and functional SLA theory because of the pioneering work of such scholars
as Roger Andersen, Yas Shirai, and Kathleen Bardovi-Harlig. Andersen
was the rst L2 researcher to pose the idea following similar studies in L1
acquisition that L2 language learners are initially inuenced by the inher-
ent semantic aspect of verbs in the acquisition of TA morphology axed to
these verbs. Andersen argued that L2 learners start out by using the perfec-
tive past morpheme with telic verbs (achievements and accomplishments,
with a clear endpoint) before they extend its use to atelic verbs (activity
and stative, with no inherent endpoint). After the perfective form is estab-
lished, learners start to mark states with the imperfective form and later
Frequency-based grammar 99

spread its use to dynamic verbs towards telic events. Conversely, progres-
sive marking is preferentially used with dynamic verbs (activities, accom-
plishments, and achievements) and it is rst used with activity verbs (atelic)
before it spreads to telic verbs. That is, progressivity is preferentially
marked rst with verbs that focus our attention on the process rather
than the end-state of the process.
This inuence of the inherent lexical semantics of verbal predicates on
the acquisition of morphosyntactic marking led to an important hypo-
thesis of TA acquisition in terms of cognitive psychological processes of
prototype formation (Andersen and Shirai 1994, 1996; Shirai and Andersen
1995). The Aspect Hypothesis (AH) (see Chapters 8 and 11; Andersen and
Shirai 1994) proposes that the abstract grammatical schema for perfective
past generalizes from more concrete beginnings close to the prototypic
centre in the clear exemplications of telic achievements and accomplish-
ments. Likewise abstract progressive morphology emerges from concrete
exemplars in the semantics of activities and states.
Andersens hypothesis was based on the analysis of L2 Spanish data
collected among adolescent learners in the natural (non-classroom based)
social environment of acquisition. Even though Andersen did not obtain
data to conrm all stages of acquisition of past tense morphology, he pro-
posed a sequence of acquisition of eight phases. The strong association of
the lexical semantics of verbal predicates was predicted to occur during
the initial four stages. Andersen argued further that the nal four stages
are necessary in the model to account for the fact that learners are even-
tually able to use both markers of past tense aspect with any lexical aspec-
tual class, thus breaking the categorical pairing of one lexical aspectual
class and one grammatical marker (e.g., states and Imperfect, achievements
and Preterite). The last four stages point to the fact that the appropriate
use of TA verbal endings brings about a level of discursive and semantic
complexity that accounts for the diculty L2 learners have in the process
of acquisition. That is, a comprehensive account of how L2 learners
approach the level of representation of TA meanings among native speakers
must eventually go beyond the level of lexical aspect (cf., input frequency
and prototypicality of meaning), incorporating in the process the variety
of cues that underpin the more complex representations underpinning
nativelike levels of grammatical aspect.
Aspect-before-tense phenomena also prevail in L2 acquisition (Ander-
sen and Shirai 1994; Bardovi-Harlig 2000; Indefrey and Gullberg 2008;
Li and Shirai 2000). Adult language learners too are sensitive to the lexical
aspects of verbs, initially using combinations of lexical and grammatical
100 Nick Ellis

aspect that are maximally compatible, with telicity being a particularly


salient feature.2
Bardovi-Harlig (2000), in particular, presents an extensive functional
analysis of the acquisition of L2 TA morphology in terms of cognitive
principles and semantic prototypes. Thus L2 learners from a wide variety
of L1/L2 combinations rst use perfective past marking on achievements
and accomplishments, and only later extend this to activities and state.
Similarly, in L2s that have progressive aspect, progressive marking begins
with activities and only extend slowly thereafter to accomplishments and
achievements.
Despite this support for the AH, the original argument about the eect
of the inherent lexical semantics of the verbal predicate on the morpho-
syntactic marking of TA was underspecied with regards the timing of
this eect: Does lexical aspect guide the process from the beginning stages
of acquisition and later subside as learners are able to use both gramma-
tical markers with every verb type as the AH suggests? Or, does the eect
of lexical aspect increase with experience in the L2? Previous researchers
have not been clear on this point. Robison (1990) argued that the eect
of lexical aspect occurs when L2 verb morphemes enter the interlanguage
of an adult language learner, but also that . . . verbal morphology corre-
lates with lexical aspect at least during some stage during the development
of an interlanguage (Robison 1990: 329330, italics added).
Wiberg (1996) and Salaberry (1999) argued for an expansion of the
claim made by Bergstrom (1995) showing that the perfective form was
used with all lexical aspectual classes (not just dynamic verbs) during the
very beginning stages of acquisition. Again, these results do not reject the
eect of a past tense prototypical marker; quite the opposite. Nevertheless,
the eect of straight lexical-grammatical pairings is weaker than expected
by the AH. Also, more recent studies have shown that the eect of lexical
aspect tends to increase with exposure to the L2. This is contrary to the
expectation that lexical semantics has maximum eect at rst until non-
prototypical pairings are eventually incorporated to the L2 system. In
fact, even early proponents of the AH have acknowledged the replication
of ndings that demonstrate the increasing rather than decreasing eect of

2. There is a substantial amount of empirical evidence oered in favor of the


Aspect Hypothesis in SLA (e.g., Bardovi-Harlig 1998, 2000; Bardovi-Harlig
and Reynolds 1995; Bergstrom 1995; Camps 2002, 2005; Collins 2002, 2004;
Comajoan 2001, 2006; Hasbun 1995; Salaberry 1998; Shirai and Kurono
1998, which examined cloze passages, and Bardovi-Harlig 1998, 2000, which
investigated oral production data obtained from narratives).
Frequency-based grammar 101

prototypical tense-aspect markings. Thus, Shirai (2004: 103) states that at


least in some contexts in cross-sectional studies involving production
data, the prototypical association becomes stronger as the learners pro-
ciency increases. These results are in agreement with the importance of
type and token frequency; that is, the productivity of a pattern is a func-
tion of type frequency; the more forms that exemplify a pattern, the more
productive that pattern becomes (see Section 4.1.2.)
Perhaps the strongest evidence in favor of the increasing association
of grammatical marking of tense-aspect and lexical aspectual classes as
learners acquire more experience in the L2 is provided by Salaberry (2011)
with a study that compared the claims of the AH and the Discourse
Hypothesis (DH). The ndings are important because this study used
a large number of participants, thus providing a more robust data set
than is normally used in TA studies. The results showed both that L2
learners increased their use of past tense markers in association with the
inherent lexical meanings of verb phrase, and, more importantly, that
native speakers had the highest association of prototypical pairings in
their use. Thus, L2 learners seem to be converging, in asymptotic terms,
towards the native speaker norm. That is, the main factor behind this
change seems to be the distributional bias present in native speakers
choices, which is clearly related to exposure and frequency of data. Obvi-
ously, as L2 learners gain more experience in the language and have more
exposure to language samples, they are able to converge more and more
towards the native speaker standard.
A frequency-based approach argues that frequency/prototypicality eects
are there from the very get-go, because they determine the sample of
language which a learner is likely to experience. Zipf s law entails that
particular exemplars are very high frequency these are the ones a learner
is going to experience rst, and these are the ones that therefore seed the
system. If, as is typical in language, the high frequency forms in a con-
struction are also prototypical in meaning, then these are the ones a
learner will sample (Sections 4.1.3 and 4.3.1). These results are in line with
other studies that have investigated the inuence of input frequency on TA
acquisition in L1 (Shirai, Slobin, and Weist 1998) and L2 (Andersen 1990).
More specically, Andersen (1990, the Distributional Bias Hypothesis) ob-
served that the input available to learners exhibits distributional patterns
similar to those observed in learners productions: Native speakers in inter-
action with other native speakers tend to use each verb morpheme with a
specic class of verbs, also following the aspect hypothesis (Andersen and
Shirai 1994: 137). Such input frequency biases should aid the statistical
learning of TA constructions.
102 Nick Ellis

5.2. The eect of frequency-based constructionist biases on the


acquisition of L2 aspect
Wul et al. (2009) analyzed the eect of the constructionist principles
outlined in Section 4 (input frequency, prototypicality of meaning, and
contingency of form-function mapping) for learning TA meanings using
corpus linguistic analyses of representative samples of language input and
of learner language. The study was designed to test frequency-based con-
structionist hypotheses for the acquisition of English L2 TA constructions
as cognitive categories. The particular hypotheses used in this study, and
the ndings relating to them, were as follows:

H1: Natural language data has a distributional bias whereby some verb
types occupy each TA construction much more frequently than others, the
distribution of the types constituting each construction being Zipan.

In order to examine frequency biases in the input, we retrieved verb form


frequencies for all verbs from two native speaker corpora taken to repre-
sent the type of language input adult L2 learners are exposed to: the 10
million word spoken section of the British National Corpus (BNCspoken)
and the 1.7 million word Michigan Corpus of Academic Spoken English
(MICASE, Simpson et al. 2002). All verb form frequencies were retrieved
from CLAWS-tagged versions of BNCspoken and MICASE, respectively.
When we analyzed the verbs tagged as simple past or progressive, their
frequency distributions across the dierent TA categories was Zipan: the
frequency with which verbs occur with a certain TA category is inversely
proportional to their rank in the frequency table, with the most frequent
verb types accounting for the lions share of all occurrences of any given
TA morpheme. Unlike for the VAC data in Ellis and Ferreira-Junior
(2009a, b) however, the top ten most frequent verbs within each category
were not typically distinctive of that category, because the very highest fre-
quency verbs in the language (like do, be, have, and get) naturally occupy
the top ranks across all TA categories.

H2: More-frequent verbs in each TA construction are distinctively associated


with that construction in the input.

In order to determine which verbs are particularly associated with the


progressive and the perfective more systematically, we computed form-
meaning contingencies (see Section 4.4), in this case using a multiple dis-
Frequency-based grammar 103

tinctive collexeme analysis (MDCA) for the BNCspoken and MICASE


data sets (Gries and Stefanowitsch 2004). The association-based distribu-
tions showed that a small number of verbs are extremely highly associated
with a particular TA category, and association strength drops exponen-
tially thereafter. Ranking the top ten most distinctively associated verbs
for each TA reected intuitions about verbs that typically occur with the
dierent TA categories: the past and perfect TA columns were occupied
by highly telic verbs such as die, crash, explode, lose, or nish; the progres-
sive preferred continuous action verbs like sit, play, walk, and run. These
distinctively-associated verbs, while not the highest frequency in the lan-
guage (H1), are frequently experienced in that construction.

H3: The verbs most distinctively associated with each TA construction in


the input are prototypical of the meaning of that construction.

In order to investigate the prototypicality of the verbs, we obtained native


speaker telicity ratings for a range of verbs selected from these analyses
from 20 native speakers of American English. A questionnaire presented
the verbs in isolation, without arguments, and in their base forms. Sub-
jects were instructed to evaluate each verb with regard to how strongly it
implies an endpoint expressed in values from 1 (if there is no endpoint im-
plied) to 7 (if an endpoint is strongly implied). Three examples were given:
smash as a highly telic verb, continue as an example of a verb that is
located at the opposite, atelic end of the continuum, and swim as an exam-
ple of a verb that falls somewhere in between.
The resulting Telicity Rating data demonstrated that those verbs distinc-
tively associated with past tense in the input received signicantly higher
telicity ratings than verbs associated with the progressive (MICASE data:
t 2.107; df 18; p .049; BNC spoken data: t 4.356; df 18;
p < .001).

H4: The rst-learned verbs in each TA construction are prototypical of


that constructions functional interpretation in terms of their telicity/lexical
aspect.

Wul et al. analyzed oral production data collected by Bardovi-Harlig


(2000), who had 37 English beginning L2 learners from 5 dierent L1
backgrounds watch an excerpt of Modern Times and then tell the story in
their own words. The resulting narratives produced an average of 51 verb
tokens. All verb forms were coded for TA morphology (that is, simple
104 Nick Ellis

past, past progressive, pluperfect, present, present progressive, progressive


without auxiliaries, present perfect, or uninterpretable). For the purpose
of their study, Wul et al. selected from this data set verbs that occurred
more than 10 times overall and which were distinctly associated with present,
simple past, or progressive as determined by a chi-square test. The 5 most
frequently occurring past tense verbs in the learner production data (say,
see, steal, take, tell ) and the 5 most frequently occurring progressive verbs
(begin, eat, run, think, walk) diered signicantly in their mean telicity
ratings (t 2.838; df 9; p < .01), with the past tense verbs being
judged more telic and the progressive verbs more atelic.
In sum, the results of Wul et al. (2009) suggested that the verbs rst
learned by adults in the progressive are also frequent in the progressive in
the input, distinctively associated with the progressive in the input, and
highly atelic (i.e., signicantly less telic than verbs frequent and associated
with past tense in the input). Likewise, the verbs rst learned in past tense
are frequent in past tense in the input, highly distinctive for past tense in the
input, and highly telic. These ndings provide some support for the hypo-
thesis that the learning of tense and aspect, like that of other linguistic
constructions, can be understood according to psychological principles of
category learning. In terms of the frequency-based associative, cognitive,
and functional properties of TA construction learning: (1) The rst-learned
verbs in each TA construction are those which appear frequently in that
construction in the input. (2) The rst-learned pathbreaking verbs for
each TA construction are distinctive of that construction the contin-
gency of forms and function is reliable. (3) The rst-learned verbs in each
TA construction are those which are prototypical of the constructions
functional interpretation in terms of telicity/lexical aspect. In sum, TA
construction learning is sensitive to input frequency, reliabilities of form-
function mapping, and prototypicality of lexical aspect in English.
Although the analyses of spoken language carried out by Wul et al.
(2009) involved quite extensive corpus analysis, it is a stretch to claim
that the language sampled therein was properly representative of that to
which the ESL learners had been exposed. Additionally, the learner data
was small, far from dense, and it covered only a very short period of initial
acquisition. Finally, the study focused on L2 English only as the target
language. We turn next to the analysis of more advanced levels of L2
Spanish, a language with a complex representation of tense-aspect markers
to investigate the eect of input frequency, prototypicality, and the map-
pings of form and meaning.
Frequency-based grammar 105

5.3. The eect of frequency-based constructionist biases on acquisition


of Spanish L2 aspect

The eects described for English above are, by and large, also relevant for
the analysis of L2 Spanish data. However, the analysis of more advanced
Spanish data to be discussed below (from Salaberry 2011) brings about a
challenge for any constructionist approach operating only at a lexical level:
the distinction between the aspectual concepts of iterativity and habituality
as shown in sentences (1a) and (1b).
(1a) Cuando era nino, Lucas jugaba al futbol. [habitual]
When [he] was a child, Lucas played/used to/would (IMP) play
soccer.
(1b) Por anos, Lucas jugo al futbol. [iterative]
For years, Lucas played (PRET) soccer.
The main challenge for learners is that the use of the perfective form to
make reference to extended events in the past is predicated on the facts
that iterativity (a) is not very frequent in the input, and (b) it does not
represent a prototypical marker of iteration (i.e., the imperfective form is
the prototypical marker, as documented in one of the most traditional and
used rules taught to Spanish learners). On the other hand, learners can
benet from the fact that the grammatical marking of iterated events pro-
vides a direct mapping of form and function (i.e., iterativity is always
marked with the Preterite, whereas habituality is marked with the Imper-
fect). That is, L2 learners need to go beyond the realm of prototypical
pairings of lexical aspect and grammatical markings to learn some specic
aspectual meanings that are clearly marked in Spanish through the choice
of perfective or imperfective marker. To do so, however, L2 learners must
take into account broader pieces of discourse than would be required
to make decisions about straightforward lexical-grammatical pairings (as
discussed in the analysis of English data above).
If the challenge is to process ever-longer pieces of discourse to make
judgments on the aspectual representation of eventualities, one of the rst,
most immediate elements that has to be considered to mark aspectual con-
trasts is the role of adverbial phrases. For instance, Menendez-Benito (2001)
shows how adverbial phrases can change the prototypical meaning of the
perfective marker in Spanish (i.e., episodic meanings) to represent the itera-
tion of eventualities (i.e., an aspectual concept reserved for the imperfective
marker). The diculty brought about by the broader discourse prompted
by the computation of adverbials (on top of the analysis of external and
106 Nick Ellis

internal arguments) for the marking of iterated eventualities is corroborated


by the few studies that have looked at this area of studies. Previous studies
(e.g., Perez-Leroux et al. 2007; Salaberry and Martins 2011; Slabakova and
Montrul 2007) show, categorically, that L2 learners even highly advanced
learners fail to recognize the aspectual meaning of iterativity (conveyed
through the use of the Preterite) as distinct from the meaning of habituality
(conveyed through the use of the Imperfect).
A constructionist explanation that the use of the Spanish Preterite to
express iterativity is dicult for L2 learners to acquire would rst point
to the facts that iterative meanings of the Preterite are neither frequent in
the input nor prototypical of the perfective form. A richer analysis of the
problem is, nevertheless, possible given that the focus of Construction
Grammar is as much about constructions above the word level (e.g.,
grounding information) as about lexical or morphological units (e.g.,
lexico-semantic information), thus we assign a prominent role to the con-
spiracy of cues in processing (see Section 4.1.3 and the acquisition of Verb-
argument constructions). In this respect, native speakers systematically use
cues provided by adverbial phrases to select the use of Preterite or Imper-
fect to mark either iterativity or habituality. The debate is whether we can
correlate the use of Preterite and Imperfect with generic and durational
adverbial phrases as proposed by Menendez-Benito (2001), or specic
adverbial phrase constructions as proposed by Salaberry and Martins
(2011), or some other alternative option. Further research needs to inves-
tigate how the lexical-level cues act in combination with adverbial phrases,
and how learners may be more sensitive to some cues (lexical or discourse-
building) in this conspiracy at dierent stages of language acquisition
(Salaberry 2008, 2011; Rosi 2010).

6. Conclusions and future research directions

The rst part of this chapter gathered a range of frequency-related factors


that inuence the acquisition of any linguistic constructions:
1. the frequency, the frequency distribution, and the salience of the form
types,
2. the frequency, the frequency distribution, the prototypicality and gen-
erality of the semantic types, their importance in interpreting the over-
all construction,
3. the reliabilities of the mapping between 1 and 2, and
4. the degree to which the dierent elements in the construction are
mutually informative and form predictable chunks.
Frequency-based grammar 107

The second part applied these factors to TA acquisition. Before learners


can recognize or use TA constructions productively, they have to analyze
them, to identify their linguistic form, and then map it to meaning. Each
construction has its own form, meaning, and corresponding mapping
pattern. Research shows that the input that learners get is biased so
that they experience past tense forms predominantly with verbs which are
distinctively associated with more telic construals, and progressive forms
predominantly with verbs which are distinctively associated with more
atelic construals. Language lines up with the world, or, better, with the
way we construe it. Our understanding of the world lines up with our
language. Our actions in the world, our categorization of the world, and
our talk about these actions and classications occur in broadly parallel
relative frequencies. Such parallels make constructions learnable.
There are many factors involved, and research to date has tended to
look at each hypothesis by hypothesis, variable by variable, one at a time.
But they interact. And what we really want is a model of usage and its
eects upon acquisition. We can measure these factors individually. But
such counts are vague indicators of how the demands of human interac-
tion aect the content and ongoing co-adaptation of discourse, how this
is perceived and interpreted, how usage episodes are assimilated into the
learners system, and how the system reacts accordingly (see Chapters 10,
11, and 12).
Usage is rich in latent linguistic structure, thus frequencies of usage
count in the emergence of linguistic constructions. Corpus Linguistics pro-
vides the proper empirical means whereby language input can be counted.
But this is not enough; we also require an understanding of the psychology
of cognition, learning, attention, and development. Sensation is not per-
ception, and the psychophysical relations mapping physical onto psycho-
logical scales are complex. The world of conscious experience is not the
world itself but a construal crucially determined by attentional limitations,
prior knowledge, embodiment, and context. Not every experience is equal
eects of practice are greatest at early stages but eventually reach asymp-
tote. The associative learning of constructions as form-meaning pairs is
aected by: factors relating to the form such as frequency and salience;
factors relating to the interpretation such as signicance in the comprehen-
sion of the overall utterance, prototypicality, generality, and redundancy;
factors relating to the contingency of form and function; and factors relat-
ing to learner attention, such as automaticity, transfer, and blocking.
108 Nick Ellis

Univariate counts are vague indicators of how the demands of human


interaction aect the content and ongoing co-adaptation of discourse, how
this is perceived and interpreted, how usage episodes are assimilated into
the learners system, and how the linguistic system reacts accordingly. We
need models of learning, language, meaning, usage, interaction, develop-
ment, and emergence that take all these factors into account dynamically.
Some progress on language and meaning comes from cognitive linguistics
(Robinson and Ellis 2008), though this is often non-quantitative research.
Some progress on language usage comes from corpus linguistics (Gries
and Divjak, in press), though all too often this is cognition-light. Some
progress on interaction comes from work on the Interaction Hypothesis
(Mackey and Gass 2006), though too often this is language-light. Some
progress on emergence is being made in emergentism and complexity
theory (Ellis 1998; Ellis and Larsen Freeman 2006a; Ellis and Larsen-
Freeman 2009b; Elman et al. 1996; Larsen-Freeman 1997; Larsen-Freeman
and Cameron 2008; MacWhinney 1999) which analyzes how complex
patterns emerge from the interactions of many agents, how each emergent
level cannot come into being except by involving the levels that lie below
it, and how at each higher level there are new and emergent kinds of related-
ness not found below. These approaches align well with dynamic system
theory, which considers how cognitive, social, and environmental factors are
in continuous interactions, where ux and individual variation abound, and
where cause-eect relationships are non-linear, multivariate, and interactive
in time (de Bot, Lowie, and Verspoor 2007; Ellis 2008a; Ellis and Larsen
Freeman 2006a, 2006b; Port and van Gelder 1995; Spencer, Thomas, and
McClelland 2009; Spivey 2006; van Geert 1991). But research in emergence
and dynamic system theory are often light in the details of the component
parts.
Recent developments in corpus linguistics, neuro-linguistic programming,
and computer simulation suggest that a tractable approach is to combine the
qualitative linguistic analyses of construction grammar and corpus linguistics
as applied to longitudinal corpora of learner language and large samples
of representative input. These can then to be brought together in quanti-
tative computer simulations of construction acquisition (Christiansen and
Chater 2001), either connectionist, agent-based, or exemplar-driven, illus-
trated, for example, in the initial explorations of MacWhinney and Leinbach
(1991), Ellis and Schmidt (1998), Li and Shirai (2000), and Ellis with Larsen-
Freeman (2009a). Even then, much will remain to be done in building into
such models more sophisticated representation of salience of form and its
perception, meaning and embodiment, and learner attention.
Frequency-based grammar 109

The analyses of the psychological representation of TA meanings, the


linguistic means by which these representations are explicitly conveyed in
usage, and their developmental sequences in interlanguage together pro-
vide a rich testing-ground for investigation of cognitive and linguistic
universals of TA and of the role of frequency-tuning in the usage-based
abstraction of constructions as categories.

References

Andersen, Roger
1990 Models, processes, principles and strategies: second language
acquisition inside and outside of the classroom. In B. Van Patten
& J. Lee (Eds.), Second language acquisition Foreign language
learning, 4568. Clevedon: Multilingual Matters.
Andersen, Roger & Yasuhiro Shirai
1994 Discourse motivations for some cognitive acquisition principles.
Studies in Second Language Acquisition 16. 133156.
Andersen, Roger & Yasuhiro Shirai
1996 The primacy of aspect in rst and second language acquisition:
The pidgin-creole connection. In W. C. Ritchie & T. K. Bhatia
(Eds.), Handbook of second language acquisition, 527570. San
Diego, CA: Academic Press.
Bardovi-Harlig, Kathleen
1998 Narrative structure and lexical aspect: Conspiring factors in sec-
ond language acquisition of tense-aspect morphology. Studies in
Second Language Acquisition 20. 471508.
Bardovi-Harlig, Kathleen
2000 Tense and aspect in second language acquisition: Form, meaning,
and use. Oxford: Blackwell.
Bardovi-Harlig, Kathleen & Dudley Reynolds
1995 The role of lexical aspect in the acquisition of tense and aspect.
TESOL Quarterly 29. 107131.
Bates, Elizabeth & Brian MacWhinney
1987 Competition, variation, and language learning. In B. MacWhinney
(Ed.), Mechanisms of language acquisition, 157193. Hillsdale,
NJ: Lawrence Erlbaum Associates.
Bock, J. Kathrin
1986 Syntactic persistence in language production. Cognitive Psychology
18. 355387.
Bod, Rens, Jennifer Hay & Stefanie Jannedy (Eds.)
2003 Probabilistic linguistics. Cambridge, MA: MIT Press.
Boyd, Jeremy & Adele Goldberg
2009 Input eects within a constructionist framework. Modern Lan-
guage Journal 93(2). 418429.
110 Nick Ellis

Bybee, Joan
2008 Usage-based grammar and second language acquisition. In P.
Robinson & N. C. Ellis (Eds.), Handbook of cognitive linguistics
and second language acquisition, 216236. London: Routledge.
Bybee, Joan & Paul Hopper (Eds.)
2001 Frequency and the emergence of linguistic structure. Amsterdam:
Benjamins.
Bybee, Joan & Sandra Thompson
2000 Three frequency eects in syntax. Berkeley Linguistic Society 23.
6585.
Christiansen, Morten H. & Nick Chater (Eds.)
2001 Connectionist psycholinguistics. Westport, CO: Ablex.
Clark, Eve
1978 Discovering what words can do. In D. Farkas, W. M. Jacobsen
& K. W. Todrys (Eds.), Papers from the parasession on the
lexicon, Chicago Linguistics Society April 1415 1978, 3457.
Chicago: Chicago Linguistics Society.
Collins, Laura
2002 The roles of L1 inuence and lexical aspect in the acquisition of
temporal morphology. Language Learning 52. 4394.
Collins, Laura & Nick Ellis
2009 Input and second language construction learning: frequency,
form, and function: Introduction to the special issue. Modern
Language Journal 93(2). 329335.
Croft, William
2001 Radical construction grammar: Syntactic theory in typological
perspective. Oxford: Oxford University Press.
Croft, William & Alan Cruise
2004 Cognitive linguistics. Cambridge: Cambridge University Press.
de Bot, Kees, Wandar Lowie & Marjoliyn Verspoor
2007 A dynamic systems theory to second language acquisition. Bilin-
gualism: Language and Cognition 10. 721.
Dietrich, Rainer, Wolfgang Klein & Colette Noyau (Eds.)
1995 The acquisition of temporality in a second language. Amsterdam:
John Benjamins.
Elio, Renee & John Anderson
1981 The eects of category generalizations and instance similarity on
schema abstraction. Journal of Experimental Psychology: Human
Learning and Memory 7(6). 397417.
Elio, Renee & John Anderson
1984 The eects of information order and learning mode on schema
abstraction. Memory and Cognition 12(1). 2030.
Ellis, Nick
1998 Emergentism, connectionism and language learning. Language
Learning 48(4). 631664.
Frequency-based grammar 111

Ellis, Nick
2002a Frequency eects in language processing: A review with implica-
tions for theories of implicit and explicit language acquisition.
Studies in Second Language Acquisition 24(2). 143188.
Ellis, Nick
2002b Reections on frequency eects in language processing. Studies
in Second Language Acquisition 24(2). 297339.
Ellis, Nick
2003 Constructions, chunking, and connectionism: The emergence
of second language structure. In C. Doughty & M. H. Long
(Eds.), Handbook of second language acquisition, 63103.
Oxford: Blackwell.
Ellis, Nick
2006a Language acquisition as rational contingency learning. Applied
Linguistics 27(1). 124.
Ellis, Nick
2006b Selective attention and transfer phenomena in SLA: Contingency,
cue competition, salience, interference, overshadowing, blocking,
and perceptual learning. Applied Linguistics 27(2). 131.
Ellis, Nick
2007 Blocking and learned attention in language acquisition. In D. S.
McNamara & J. G. Trafton (Eds.), Proceedings of the 29th
Annual Cognitive Science Society, 965970. Austin: Cognitive
Science Society.
Ellis, Nick
2008a The dynamics of language use, language change, and rst and
second language acquisition. Modern Language Journal 41(3).
232249.
Ellis, Nick
2008b Usage-based and form-focused language acquisition: The asso-
ciative learning of constructions, learned-attention, and the
limited L2 endstate. In P. Robinson & N. C. Ellis (Eds.), Hand-
book of cognitive linguistics and second language acquisition,
372405. London: Routledge.
Ellis, Nick & Teresa Cadierno
2009 Constructing a second language. Annual Review of Cognitive
Linguistics 7 (Special section).
Ellis, Nick & Fernando Ferreira-Junior
2009a Construction learning as a function of frequency, frequency dis-
tribution, and function. Modern Language Journal 93. 370386.
Ellis, Nick & Fernando Ferreira-Junior
2009b Constructions and their acquisition: Islands and the distinctive-
ness of their occupancy. Annual Review of Cognitive Linguistics,
111139.
112 Nick Ellis

Ellis, Nick & Diane Larsen Freeman (Eds.)


2006a Language emergence: Implications for applied linguistics. Applied
Linguistics 27(4).
Ellis, Nick & Diane Larsen Freeman
2006b Language emergence: Implications for applied linguistics (intro-
duction to the special issue). Applied Linguistics 27(4). 558589.
Ellis, Nick & Diane Larsen Freeman
2009a Constructing a second language: Analyses and computational
simulations of the emergence of linguistic constructions from
usage. Language Learning 59 (Supplement 1). 93128.
Ellis, Nick & Diane Larsen Freeman (Eds.)
2009b Language as a complex adaptive system (special issue). Lan-
guage Learning 59. Supplement 1.
Ellis, Nick & Matthew ODonnell
in press Statistical construction learning: Does a Zipan problem space
ensure robust language learning? In J. Rebuschatand J. Williams
(Eds.), Statistical learning and language acquisition, 265304.
Berlin: Mouton de Gruyter.
Ellis, Nick & Nuria Sagarra
2010 The bounds of adult language acquisition: Blocking and learned
attention. Studies in Second Language Acquisition 32(4). 553580.
Ellis, Nick & Richard Schmidt
1998 Rules or associations in the acquisition of morphology? The fre-
quency by regularity interaction in human and PDP learning of
morphosyntax. Language and Cognitive Processes, 13 (2 and 3).
307336.
Elman, Jerey, Elizabeth Bates, Mark Johnson, Annette Karmilo-Smith, Domenico
Parisi & Kim Plunkett
1996 Rethinking innateness: A connectionist perspective on development.
Cambridge, MA: MIT Press.
Ferrer i Cancho, Ramon & Richard Sole
2001 The small world of human language. Proceedings of the Royal
Society of London, B. 268. 22612265.
Ferrer i Cancho, Ramon & Richard Sole
2003 Least eort and the origins of scaling in human language. PNAS
100. 788791.
Goldberg, Adele
1995 Constructions: A construction grammar approach to argument
structure. Chicago: University of Chicago Press.
Goldberg, Adele
2003 Constructions: a new theoretical approach to language. Trends
in Cognitive Science 7. 219224.
Goldberg, Adele
2006 Constructions at work: The nature of generalization in language.
Oxford: Oxford University Press.
Frequency-based grammar 113

Goldberg, Adele, Devin Casenhiser & Nitya Sethuraman


2004 Learning argument structure generalizations. Cognitive Linguis-
tics 15. 289316.
Goldschneider, Jennifer & Robert DeKeyser
2001 Explaining the natural order of L2 morpheme acquisition in
English: A meta-analysis of multiple determinants. Language
Learning 51. 150.
Gries, Stefan & Dagmar Divjak (Eds.)
in press Frequency eects in cognitive linguistics (Vol. 1): Statistical eects
in learnability, processing and change. Berlin: Mouton de Gruyter.
Gries, Stefan & Anatol Stefanowitsch
2004 Extending collostructional analysis: a corpus-based perspective
on alternations. International Journal of Corpus Linguistics 9.
97129.
Gries, Stefan & Stefanie Wul
2005 Do foreign language learners also have constructions? Evidence
from priming, sorting, and corpora. Annual Review of Cognitive
Linguistics 3. 182200.
Harnad, Stevan (Ed.)
1987 Categorical perception: The groundwork of cognition. New York:
Cambridge University Press.
Indefrey, Peter & Marianne Gullberg
2008 Time to speak: Cognitive and neural prerequisites for time in
language. Language Learning 58.
Jurafsky, Dan
2002 Probabilistic modeling in psycholinguistics: Linguistic compre-
hension and production. In R. Bod, J. Hay & S. Jannedy (Eds.),
Probabilistic Linguistics, 3996. Harvard, MA: MIT Press.
Jurafsky, Daniel & James Martin
2000 Speech and language processing: An introduction to natural lan-
guage processing, speech recognition, and computational linguistics.
Englewood Clis, NJ: Prentice-Hall.
Lako, George
1987 Women, re, and dangerous things: What categories reveal about
the mind. Chicago: University of Chicago Press.
Langacker, Ronald
1987 Foundations of cognitive grammar: Vol. 1. Theoretical prerequisites.
Stanford, CA: Stanford University Press.
Larsen-Freeman, Diane
1997 Chaos/complexity science and second language acquisition.
Applied Linguistics 18. 141165.
Larsen-Freeman, Diane & Lynne Cameron
2008 Complex systems and applied linguistics. Oxford: Oxford Univer-
sity Press.
114 Nick Ellis

Li, Ping & Yasuhiro Shirai


2000 The acquisition of lexical and grammatical aspect. Amsterdam:
De Gruyter.
Lieberman, Erez, Jean-Baptiste Michel, Joe Jackson, Tina Tang & Martin Nowak
2007 Quantifying the evolutionary dynamics of language. Nature
449(7163). 713716.
Mackey, Allison & Susan Gass
2006 Pushing the methodological boundaries in interaction research:
An Introduction to the special issue. Studies in Second Language
Acquisition 28. 169178.
MacWhinney, Brian
1987a Applying the Competition Model to bilingualism. Applied Psycho-
linguistics 8(4). 315327.
MacWhinney, Brian
1987b The Competition Model. In B. MacWhinney (Ed.), Mechanisms
of language acquisition, 249308. Hillsdale, NJ: Erlbaum.
MacWhinney, Brian
1997 Second language acquisition and the Competition Model. In A.
M. B. De Groot & J. F. Kroll (Eds.), Tutorials in bilingualism:
Psycholinguistic perspectives, 113142. Mahwah, NJ: Lawrence
Erlbaum Associates.
MacWhinney, Brian
2001 The competition model: The input, the context, and the brain.
In P. Robinson (Ed.), Cognition and second language instruction,
6990. New York: Cambridge University Press.
MacWhinney, Brian
1999 The emergence of language. Hillsdale, NJ: Erlbaum.
MacWhinney, Brian & Jared Leinbach
1991 Implementations are not conceptualizations: Revising the verb
learning model. Cognition 40(12). 121157.
McDonough, Kim
2006 Interaction and syntactic priming: English L2 speakers produc-
tion of dative constructions. Studies in Second Language Acquisi-
tion 28. 179207.
McDonough, Kim & Allison Mackey
2006 Responses to recasts: Repetitions, primed production and lin-
guistic development. Language Learning 56. 693720.
McDonough, Kim & Pavel Tromovich
2008 Using priming methods in second language research. London:
Routledge.
Menendez-Benito, Paula
2002 Aspect and adverbial quantication in Spanish. Paper presented at
the Proceedings of the 32nd North Eastern Linguistics Society,
Amherst, MA.
Frequency-based grammar 115

Ninio, Anat
1999 Pathbreaking verbs in syntactic development and the question
of prototypical transitivity. Journal of Child Language 26. 619
653.
Ninio, Anat
2006 Language and the learning curve: A new theory of syntactic develop-
ment. Oxford: Oxford University Press.
Pagel, Mark, Quentin Atkinson & Andrew Meade
2007 Frequency of word-use predicts rates of lexical evolution through-
out Indo-European history. Nature 449(7163). 717720.
Perez-Leroux, Ana, Alejandro Cuza, Monika Majzlanova & Jeanette Sanchez
Naranjo
2007 Non-native recognition of the iterative and habitual meanings of
Spanish preterite and imperfect tenses. In J. Liceras, H. Zobl &
H. Goodluck (Eds.), Formal features in second language acquisi-
tion, 432451. Amsterdam: John Benjamins.
Pickering, Martin
2006 The dance of dialogue. The Psychologist 19. 734737.
Pickering, Martin & Victor Ferreira
2008 Structural priming: A critical review. Psychological Bulletin 134.
427459.
Pickering, Martin & Simon Garrod
2006 Alignment as the basis for successful communication. Research
on Language and Computation 4. 203228.
Pinker, Steven
1989 Learnability and cognition: The acquisition of argument structure.
Cambridge, MA: Bradford Books.
Port, Robert & Timothy Van Gelder
1995 Mind as motion: Explorations in the dynamics of cognition. Boston
MA: MIT Press.
Posner, Michael & Steven Keele
1970 Retention of abstract ideas. Journal of Experimental Psychology
83. 304308.
Rescorla, Robert
1968 Probability of shock in the presence and absence of CS in fear
conditioning. Journal of Comparative and Physiological Psychology
66. 15.
Rescorla, Robert & Allan Wagner
1972 A theory of Pavlovian conditioning: Variations in the eective-
ness of reinforcement and nonreinforcement. In A. H. Black &
W. F. Prokasy (Eds.), Classical conditioning II: Current theory
and research, 6499. New York: Appleton-Century-Crofts.
Robinson, Peter & Nick Ellis (Eds.)
2008 A handbook of cognitive linguistics and second language acquisi-
tion. London: Routledge.
116 Nick Ellis

Rosch, Eleanor & Carolyn Mervis


1975 Cognitive representations of semantic categories. Journal of
Experimental Psychology: General 104. 192233.
Rosch, Eleanor, Carolyn Mervis, Wayne Gray, David Johnson & Penny Boyes-
Braem
1976 Basic objects in natural categories. Cognitive Psychology 8. 382
439.
Rosi, Fabiana
2009 Learning aspect in Italian L2: Corpus annotation, acquisitional
patterns, and connectionist modelling. Milan, Italy: FrancoAngeli.
Salaberry, M. Rafael
2008 Marking past tense in second language acquisition. London: Con-
tinuum Press.
Salaberry, M. Rafael
2011 Assessing the eect of lexical aspect and grounding on the acqui-
sition of L2 Spanish past tense morphology among L1 English
speakers. Bilingualism: Language and Cognition 14(2). 184202.
Salaberry, M. Rafael & Custodio Martins
2011 Non-accidental generalizations of the Preterite and Imperfect in
Spanish. Paper presented at the CHRONOS 10: Tenth Interna-
tional Conference on Tense, Aspect, Evidentiality and Modality,
Aston University, Birmingham, England.
Shanks, David
1995 The psychology of associative learning. New York: Cambridge
University Press.
Shirai, Yasuhiro & Roger Andersen
1995 The acquisition of tense-aspect morphology. Language 71. 743
762.
Shirai, Yasuhiro, Dan Slobin & Richard Weist
1998 Introduction: the acquisition of tense-aspect morphology. First
Language 18. 245253.
Simpson, Rita, Sarah Briggs, Janine Ovens & John Swales
2002 The Michigan Corpus of Academic Spoken English. Retrieved
from http://micase.elicorpora.info.
Slabakova, Roumyana & Silvina Montrul
2007 L2 acquisition at the grammar-discourse interface: Aspectual
shifts in L2 Spanish. In J. Liceras, H. Zobl & H. Goodluck
(Eds.), Formal features in second language acquisition, 452483.
Mahwah, New Jersey: Lawrence Erlbaum.
Slobin, Dan
1997 The origins of grammaticizable notions: Beyond the individual
mind. In D. I. Slobin (Ed.), The crosslinguistic study of language
acquisition (Vol. 5), 265323. Mahwah, NJ: Erlbaum.
Spencer, John, Michael Thomas & James McClelland (Eds.)
2009 Toward a unied theory of development: Connectionism and
dynamic systems theory re-considered. New York: Oxford.
Frequency-based grammar 117

Spivey, Michael
2006 The continuity of mind. Oxford: Oxford University Press.
Stefanowitsch, Anatol & Stefan Gries
2003 Collostructions: Investigating the interaction between words and
constructions. International Journal of Corpus Linguistics 8. 209
243.
Taylor, John
1998 Syntactic constructions as prototype categories. In M. Tomasello
(Ed.), The new psychology of language: Cognitive and functional
approaches to language structure, 177202. Mahwah, NJ: Erl-
baum.
Tomasello, Michael
2003 Constructing a language. Boston, MA: Harvard University Press.
Tomasello, Michael
1998 The new psychology of language: Cognitive and functional ap-
proaches to language structure. Mahwah, NJ: Erlbaum.
Tversky, Amos
1977 Features of similarity. Psychological Review 84. 327352.
van Geert, Paul
1991 A dynamic systems model of cognitive and language growth.
Psychological Review 98. 353.
Van Patten, Bill
2006 Input processing. In B. Van Patten & J. Williams (Eds.), Theories
in second language acquisition: An introduction, 116. Mahwah,
NJ: Lawrence Erlbaum.
Wittgenstein, Ludwig
1953 Philosophical investigations (G.E.M. Anscombe, Trans.). Oxford:
Blackwell.
Wul, Stefanie, Nick Ellis, Ute Romer, Kathleen Bardovi-Harlig & Chelsea
LeBlanc
2009 The acquisition of tense-aspect: Converging evidence from cor-
pora, cognition, and learner constructions. Modern Language
Journal 93. 354369.
Zipf, George
1935 The psycho-biology of language: An introduction to dynamic phi-
lology. Cambridge, MA: The M.I.T. Press.
Chapter 4
Generative approaches to the L2 acquisition of
temporal-aspectual-mood systems

Dalila Ayoun and Jason Rothman

1. Introduction

The objective of this chapter is to present a concise overview of the current


literature on the adult second language (L2) acquisition of tense, aspect
and modality (TAM) from the generative perspective. In doing so, we
highlight how such studies approach the topic and how they contribute to
our understanding of the acquisition of L2 TAM systems and morpho-
syntax more generally. First, we present the theoretical underpinnings of
current generative morpho-syntactic theory and contextualize the assump-
tions proposed under this paradigm.
In reviewing the six most well-known hypotheses that have been tested
with empirical data, Salaberry and Ayoun (2005) present a non-exhaustive
overview of the factors relevant to various approaches to the investigation
of L2 TAM systems: (a) pragmatic factors; (b) semantic factors (i.e., the
Lexical Aspect Hypothesis; (c) contextual factors (i.e., the Discourse Hypo-
thesis); (d) input-based factors: (i.e., the Distributional Bias Hypothesis);
(e) cognitive processing factors (i.e., the Default Past Tense Hypothesis);
and (f ) syntactic factors (i.e., the UG-Minimalist Hypothesis). They correctly
point out that the UG-Minimalist Hypothesis relies on syntax to provide
an explanatory account of tense and (lexical and grammatical) aspectual
phenomena (e.g., early studies by DeMiguel 1992 and Giorgi and Pianesi
1997). However, empirical studies soon expanded the strictly syntactic line
of inquiry by concerning themselves with the interaction between morpho-
syntax and semantics (e.g., Schell 2000), going beyond the traditional
functional categories such as the Inner and Outer Aspect Phrases and
Mood Phrase that host aspect and mood/modality. For instance, such
studies have examined target-like use of inectional morphology based
on the productive use of perfective and non-perfective morphology in
personal narratives.
Most research conducted within a Universal Grammar (UG) perspec-
tive shows that no one theoretical approach can neatly account for TAM
120 Dalila Ayoun and Jason Rothman

systems in isolation because TAM is dependent on factors from many


linguistic domains at the same time syntax, morphology, semantics,
and discourse/pragmatics as well as the integration of information
(operationalized as features) across internal and external linguistic-cognitive
interfaces. By increasingly focusing on how these various areas are inter-
connected (e.g., Jackendo 2002; Reinhart 2006; Rothman and Slabakova
2011; Sorace 2011; White 2011), examining TAM properties as interface-
conditioned properties has become crucial both from a theoretical and an
applied linguistic perspective. In particular, doing so has placed L2 TAM
studies at the center of current generative second language acquisition
(SLA) research from which the Interface Hypothesis (Sorace 2011 inter
alia) can be tested (e.g., Gabriele and Canales 2011; Iverson, Kempchinsky,
and Rothman 2008).
The Interface Hypothesis claims that properties at interfaces overlap-
ping points in the mental representation of grammar where syntactic com-
putation involves the integration of properties between the syntactic and
at least one other linguistic-cognitive module are more formally complex
than narrow syntactic properties, resulting in the increased likelihood of
residual optionality even at the near-native L2 steady state. In fact, since
TAM properties are relevant to both internal interfaces (i.e., the interac-
tion between two or more grammar internal modules such as morphology-
syntax, lexicon-syntax and syntax-semantics) and external interfaces (i.e.,
the interaction between a grammar internal module and cognition more
generally such as syntax-discourse-pragmatics interface), examining various
TAM domains in adult L2 acquisition can also test the most current in-
stantiations of the Interface Hypothesis, claiming a dierence in formal
complexity and processing-based expected outcomes between internal and
external interfaces (e.g., Sorace and Serratrice 2009; Tsimpli and Sorace
2006; Sorace 2011).
Although we adopt the position that L2 grammars continue to be con-
strained by an accessible UG, thereby proposing that functional categories
and their features (see Section 3.3.) are in principle acquirable in adult-
hood, we are also careful to take into account cognitive factors which
may conspire to keep L2 learners from being fully successful in the map-
ping of abstract features to the appropriate L2 surface morphology, pre-
sumably due to processing diculties, L1 transfer eects in various domains
such as prosodic transfer, and/or the inherent diculty of remapping and
reassembling features (e.g., Goad and White 2008, Lardiere 2009; Slabakova
2008, but see Bley-Vroman 2009).
Generative perspectives 121

2. Generative Linguistics as a theory of representation and acquisition

UG is an innate language faculty equipped with abstract principles of


grammar and features that give rise to parametric dierences across lan-
guages. It is also a theory of the structure and acquisition of language
that nds its genesis in the broader break within the cognitive sciences
from behaviorism in the late 1950s and early 1960s (e.g., Chomsky 1959,
1965). UG theory is also known as the parameter-setting theory or as the
Principles and Parameters theory (P&P) (Chomsky 1981), details of which
are further developed below. Generative acquisition theory diers from
other cognitive-based linguistic theories such as connectionism (e.g., Bybee
2001; Ellis 2008) and emergentism (e.g., OGrady 2005) despite the fact
that all cognitive linguistic theories agree that language and its acquisition
are unique byproducts of human cognition and experience with primary
linguistic data (i.e., the input to which children and adults alike are ex-
posed). Only UG theory maintains that language structure and its acquisi-
tion are constrained by innate knowledge specic to language in the sense
of modularity; that is, the human mind is hypothesized to be highly spe-
cialized with inborn linguistic structure. As Valian (2009) describes it,
nativist views take the position that language acquisition starts with inborn
abstract entities, for example, innate knowledge of functional syntactic
categories such as the Determiner Phrase (DP), Verb Phrase (VP) and
Complementizer Phrase (CP). Details of how these functional categories
work in particular languages of the environment are lled in via experience
with primary linguistic data.
Alternatively, the converse is true of other cognitive theories: language
acquirers are said to use the details of the input whereby frequencies and
general cognitive learning mechanisms combine to derive abstract cate-
gories such as DP, VP, and CP (see Chapter 3). The question, thus, is
not one of linguistic description as most cognitive-based theories use
similar grammatically descriptive tools, nor is it a debate on whether or
not language acquisition entails mental representations specic for lan-
guage. Linguistic nativist researchers who approach language acquisition
from a cognitive science angle agree that Language is a complex, specialized
skill, which develops in the child spontaneously, without conscious eort
or formal instruction, is deployed without awareness of its underlying
logic, is qualitatively the same in every individual, and is distinct from
more general abilities to process information or behave intelligently
(Pinker 1994: 18). Citing the logical problem of language acquisition
(Chomsky 1965, 1981) for the case of children the acquisition of complex
122 Dalila Ayoun and Jason Rothman

linguistic knowledge that is argued to be lacking in available input UG


theory maintains that associative (domain-general) learning alone insu-
ciently accounts for the entirety of acquisition.1 In Chomskys (1965: 58)
original words, which hold true four decades later in the view of genera-
tive linguists:

A consideration of the character of the grammar that is acquired, the de-


generate quality and narrowly limited extent of available data, the striking
uniformity of the resulting grammars, and their independence of intelli-
gence, motivation and emotional state, over wide ranges of variation, leave
little hope that much of the structure of language can be learned by an
organism initially uniformed as to its general character.

Herein, we highlight what the generative perspective contributes to the


understanding of how adult acquisition of TAM occurs, but also how
examining these domains specically contributes to a more complete under-
standing of how non-native language is represented in the mind/brain of
individual learners. Some of the reviewed L2 TAM studies (e.g., Iverson,
Kempchinsky, and Rothman 2008; Slabakova and Montrul 2003) provide
evidence of a similar logical problem for L2 acquisition (e.g., Schwartz
1998), that is, evidence of L2 poverty of the stimulus knowledge related
to target TAM systems. Such evidence strongly suggests a continued role
for UG in adult language acquisition despite observable dierences in
developmental sequence and ultimate attainment that are also amply exem-
plied in the studies we review below.

3. Linguistic theory and language acquisition

Explanation in formal linguistic theory has at least two main goals. The
rst is to specify a model for a universal faculty of language (i.e., UG)
and which of its properties come into play in dening the form of individual
linguistic systems, that is, the particular languages that people speak. The
second goal is to elucidate the course of linguistic development, isolat and

1. See a special issue of the journal The Linguistic Review (2002, volumes 1 and 2)
entitled A review of the poverty of the stimulus argument for a collection of
recent articles that address the debate inherent to this topic, as well as Rothman
and Iverson (2008), who use the criteria set forth in Pullman and Scholz (2002)
to demonstrate that poverty of the stimulus, at least for complex semantic
entailments derived via the acquisition of syntactic derivation, robustly exists.
Generative perspectives 123

then explaining the steps through which individuals proceed from an


initial state of underspecied structures through a series of consecutive
states of more detailed specication. According to the UG approach,
when exposed to sucient primary linguistic data, the child is able to
identify in the input the triggers whose acquisition results in the sequential
transition from the universal initial state to the ultimate attainment of a
mature, arguably steady state particular language grammar (e.g., Guasti
2002; Snyder 2007).

3.1. Parameterization: explaining cross-linguistic variation


As mentioned above, Principles and Parameters (P&P) is the general frame-
work adopted by UG acquisition approaches (Chomsky 1981), of which
the Minimalist Program (Chomsky 1992, 1995, 2001) is the most recent
instantiation. It maintains basic assumptions regarding language acquisi-
tion such as: languages are based on simple principles that interact to
form often intricate structures (Chomsky 1993: 2); and UG continues to
[provide] a xed system of principles and a nite array of nitely valued
parameters. The language-particular rules reduce to choice of values for
these parameters (Chomsky 1993: 4). Universal principles are constraints
on grammatical well-formedness, which by denition, are logistically sim-
plistic (i.e., economic) and equally true of every natural human language.
For example, the fact that all languages minimally require a subject referent
for all verbal predicates is formally captured under the Extended Project
Principle. However, in order to account for cross-linguistic variations, it is
proposed that principles are associated with linguistic parameters assumed
to present at least two dierent options, or settings, yielding dierent pos-
sibilities that languages can variably adopt. Setting relevant parameters to
the appropriate value generates the core grammar of the language being
acquired (but see Ayoun 2003 for a critical review of parameter setting
theory).
Let us take the Telicity Parameter as an example of such cross-linguistic
dierences. A basic semantic feature along with dynamicity and durativity,
telicity (from Greek telos meaning limit, end, or goal) is an inter-
pretable feature (i.e., it has semantic content): [telic] events have an
inherent, natural endpoint in time, whereas [telic] events do not. Thus,
following the widely used Mourelatos-Vendlerian classication of predi-
cates, states and activities are atelic, whereas achievement and accomplish-
ment predicates are telic (e.g., Smith 1997).
124 Dalila Ayoun and Jason Rothman

Bulgarian, English, and Spanish instantiate dierent settings of the


Telicity Parameter as illustrated in the following examples:2
(1) a. Zlatko ate pizza *in fteen minutes / for fteen minutes.
b. Zlatko coma (la) pizza *en quince minutos/por quince minutos.
c. Zlatko jade pizza *za petnajset minuti/petnajset minuti
Zlatko ate/was eating (some) pizza for fteen minutes.
(2) a. Zlatko ate the pizza in ten minutes / for ten minutes.
b. Zlatko comio la pizza en diez minutos/*por diez minutos.
c. Zlatko iz-jade pizza-ta za deset minuti/*deset minuti.
Zlatko prex-ate pizza-DET in ten minutes/*for ten minutes3
Zlatko ate up (all) the pizza in ten minutes/*for ten minutes.
In examples (1) and (2), it can be seen that the three languages encode
telicity dierently. Germanic and Romance languages require the morpho-
syntax of the Direct Object-Determiner Phrase (DO-DP) to encode an
instance of the telicity feature; for example an article, a quantier, or accu-
sative Case. Conversely, Slavic languages require that telicity be assigned
to the functional head by the perfective/imperfective morphology of the
main lexical verb, which then binds the DO-DP in its specier. As a result,
perfective heads uniformly assign the [telic] feature (as in 2c), and imper-
fective ones [telic] in Slavic languages (as in 1c). This means, for example,
that stative predicates can only be marked as atelic and thus imperfective
(without a prex); if a perfective prex obtains, the meaning is inchoative
for the state such as to be in love as opposed to fall in love. Telicity/
perfectivity marking is only independent of tense in Slavic languages
whereby overt morphological marking (i.e., the completive prex) surfaces
more than in the past as in Romance. Whereas Romance languages have
dedicated perfective and imperfective morphology in the past, unlike English
for example, telicity cannot be predictably calculated from such morphology
reliably.
This non-exhaustive discussion of the Telicity Parameter briey exem-
plies how parameterization is hypothesized to work in general. To sum
up, as a universal semantic construct, telicity must be expressed in all

2. We thank Roumyana Slabakova for the Bulgarian examples and her helpful
explanations.
3. The prex is a completive prex on the main verb. The verb with the prex is
perfective in (2), the one without the prex is imperfective in (1).
Generative perspectives 125

languages. However, the way in which particular grammars encode telicity


is left open to a specic, yet limited number of possibilities, or parametric
values. Languages like Spanish, English, and Bulgarian, for example, dif-
fer on the surface in terms of how telicity is expressed because they have
dierent settings of this parameter.
Generative theory has always maintained that the task of converging
on a specic syntactic system crucially involves the learning of the lexicon.
Even early versions of the P&P framework argued that the locus of
syntactic parameterization was found in the specication of features of
individual lexical items (e.g., Borer 1984; Hyams 1986). The Minimalist
Program holds close to this idea in that syntactic operations are signi-
cantly, if not completely, dependent on the interaction of features of
lexical items. Under a feature-based approach as developed most recently
in the Minimalist Program (see Section 3.3.), parametric variation can be
dened in terms of the feature specication of (functional) lexical items
that determine syntactic structure and variation. For instance, a prediction
is that the telicity encoding contrast between Spanish and Bulgarian as
seen above is determined by variation in the feature specication of the
relevant inectional head.

3.2. Empirical support: generative studies of language acquisition


Generative studies in child language acquisition (see e.g., Hamann 2008
for a review) have shown over the past three decades how a P&P model
best explains not only how individuals acquire the properties of specic
grammars that distinguish them from others, but also how acquisition of
individual grammars is constrained by domain-specic linguistic knowl-
edge. Similar goals apply to adult L2 acquisition as well, although in this
context, the direct role UG plays is less clear in light of the observable
dierences in L1 and L2 acquisition for developmental route and ultimate
attainment. For close to three decades, two core questions taking various
forms have dominated the research program of generative L2 studies: (a)
exploring the role UG and previous linguistic knowledge play in the devel-
opment and ultimate attainment of non-native systems and (b) exploring
what other variables conspire to explain L1/L2 dierences if one accepts
that UG, in part or wholly, is accessible in adulthood (e.g., Ayoun 2005a, b;
Hawkins and Hattori 2006; Lardiere 2009; Rothman 2008; Sorace 2005,
2011).
Some researchers take L1/L2 disparity as evidence for inaccessibility
to UG in its entirety (e.g., Bley-Vroman 1990; Clahsen and Hong 1995;
126 Dalila Ayoun and Jason Rothman

Meisel 1997) or in part (Beck 1998; Hawkins and Chan 1997). However,
others maintain that children and adults have the same access to UG, but
that previous linguistic experience, reduced processing resources, mapping
problems, physiological dierences, and other extra-linguistic factors, to
name just a few, play a role in determining how adult language acquisition
proceeds dierently and results in accented speech, production errors, and
possibly fossilization (see e.g., Lardiere 2009; Sorace 2005). As Slabakova
(2009: 158) suggests, the absolute question of L1/L2 fundamental dier-
ences must shift to understanding a more elaborate picture of the SLA
process. Even so, we take the position that there must be access to UG,
since evidence from adult knowledge of poverty-of-the-stimulus properties
exists, as reported in a signicant number of studies across several L1/L2
language pairing combinations (see Slabakova 2008 and Rothman 2008
for reviews of this literature). Some of these studies, which we review in
greater detail in what follows, involve the acquisition of new temporal and
aspectual features (see Section 4). Taken together, these studies demonstrate
that L1 properties transfer to the initial state of L2 acquisition creating an
inevitably dierent state from the underspecied initial state for L1 acqui-
sition, which alters the comparative learning task of arriving at the target
grammar for child and adult acquirers. However, evidence from advanced
learner grammars strongly favors the position that acquisition of new syn-
tactic and semantic properties is nevertheless possible in adulthood (see
Slabakova 2008).

3.3. Feature-based syntax


As a general proviso, we nd it prudent to point out from the outset that
the construct of syntactic and semantic features, at least in the sense meant
here, is a theory-internal notion, not an intuitive one. Yet, understanding
what the label feature means is crucial to understanding the literature we
review for generative SLA, and so we describe and defend its raison detre
in this section. Within the syntactic framework of the Minimalist Program
(Chomsky 1995), it is argued that inectional features correspond to func-
tional heads that project phrases (in a traditional X-bar manner). A feature
is a syntactic or semantic (abstract, but not always) element that correlates
to its grammatical function. There are two types of formal or syntactic fea-
tures: features that make a semantic contribution (e.g., [present] or [telic] as
seen above) hence referred to as interpretable features and features with
a purely syntactic function (e.g., Case on Nouns, or Agreement marking on
verbs) hence referred to as uninterpretable features.
Generative perspectives 127

Following earlier suggestions by Borer (1984) and Hyams (1986),


Chomsky (1995) and Giorgi and Pianesi (1997) proposed a feature-matching
approach under which functional categories (such as Determiner Phrase and
Tense Phrase) are derived within a clause to meet the feature-matching
needs of lexical items within a given sentence (numeration). Functional
categories themselves, being derived out of feature-checking needs, are
economical transports for (bundles) of morpho-syntactic and semantic
features but not primitive formatives of the syntax itself. Features can be
bundled in dierent ways depending on the language; however, features
must abide by the linear checking order imposed on them. Cross-linguistic
parametric dierences, as discussed above, are understood as language-to-
language variation in the featural inventory of specic grammars. The
checking of features has syntactic, semantic, and morpho-phonological
reexes such that cross-linguistic dierences and similarities are largely
reduced to distinctions in the feature compositions of particular grammars.
After acquiring the features relevant for any given language, the universal
components responsible for the checking of these features give rise to the
semantics of natural language and how the same universal system can have
dierent outputs. Poverty-of-the stimulus knowledge is thus simply the
consequence of feature checking.
Tense, mood, lexical, and viewpoint aspect are understood within Mini-
malism as features (e.g., Borer 2005; Giorgi and Pianesi 1997; Kempchinsky
2009; Quer 1998; Smith 1997). Semantic and syntactic features such as
[past], [telic], [mood], and [perfective] are hypothesized to be encoded in the
lexicon of the worlds languages in dierent ways, yielding cross-linguistic
dierences in TAM systems. Although all languages have means to encode
TAM information, it is not the case that all do so syntactically or in the
same syntactic manner. For example, by most accounts (e.g., Lin 2003,
2005), Chinese does not encode tense overtly, but dierentiates between
present, past, and future temporality by relying on temporal adverbials
(such as today and yesterday) and discourse pragmatics as well as aspec-
tual markers (i.e., classiers). So, it might be the case that Chinese does
not project (or make use) of the functional category TP/IP since there is
no feature-checking motivation for it. However, AspP must be projected
given the system of lexical classiers. Whereas Romance languages encode
mood (e.g., indicative versus subjunctive) morpho-syntatically, it is not
clear that English does at all, and certainly not always in the same way de-
pending on what type of subjunctive is implicated, for example volitional or
mandative subjunctive versus polarity subjunctive (Iverson, Kempchinsky,
128 Dalila Ayoun and Jason Rothman

and Rothman 2008).4 Yet, both languages have means to express and derive
the same semantic computations subsumed under the subjunctive morphol-
ogy of Romance languages; for instance, Spanish is assumed to project the
functional category MoodP (e.g., Kempchinsky 2009). Two other impor-
tant functional categories to account for the relationship between temporal
arguments and lexical aspect are the Inner Aspect Phrase (Travis 1992),
which represents lexical aspect (i.e., lexical perfectivity or telicity) and Outer
Aspect Phrase (Zagona 1994), which represents syntactic aspect (i.e., per-
fective or imperfective).
Regardless of how tense, mood, or aspect may be encoded in a language,
empirical studies of L2 acquisition from a generative perspective ask
whether learners can acquire the functional features within inectional
morphology and the aspectual semantic properties associated with the
relevant functional categories. The current literature oers two diametrically
opposed views on the adult L2 acquisition of functional categories and their
features. At one end of the spectrum is the view that (some) functional
categories cannot be acquired because adult L2 acquisition is no longer
constrained by UG in the same way as in childhood and/or is subjected
to a critical period (see various impairment hypotheses proposed by
Hawkins and Chan 1997, Hawkins and Liszka 2003 and representational
decit hypotheses, e.g., Hawkins 2003, Tsimpli and Dimitrakopoulou
2007). At the other end of the spectrum are those who believe that adult
SLA is constrained by UG and who support the Full Access/Full Transfer
hypothesis (e.g., Schwartz and Sprouse 1994, 1996), which proposes that
the learners L1 grammar constitutes the L2 initial state (full transfer) but
that parameter resetting to the L2 values is possible (full access)). Others
believe that although adult SLA is constrained by UG, it is subjected to
external factors (e.g., Missing Inection Hypothesis, Haznedar and Schwartz
1997; Missing Surface Inection Hypothesis, Prevost and White 2000).

4. Iverson et al. (2008) tested whether or not American English speakers accept
sentences of the following types:
i) The doctor recommends that newborn children be fed every 3 hours.
ii) The doctor recommends that newborn children are fed every 3 hours.
They found that speakers in their late teenage years and early twenties either
preferred sentences like (ii) or only accepted such sentences, whereas older
individuals either highly preferred (i) or only accepted (i) as grammatical.
They used this to argue that younger American English speakers might not
have a MoodP instantiated at all in their grammar, at least in their non-
formal, colloquial dialect.
Generative perspectives 129

The view that we adopt here is that adult L2 acquisition is constrained by


UG, and thus features and associated functional categories are in principle
acquirable (e.g., Ayoun 2005a,b; Rothman 2008). Learners have appro-
priate syntactic representations, but fail to consistently produce target-
like inectional morphology due to diculties in mapping form to meaning
or performance issues (e.g., Missing Surface Inection Hypothesis; Prevost
and White 2000; Feature Reassembly, Lardiere 2009). In reviewing the
generative L2 TAM literature below, one ultimate goal is to assess whether
the current ndings tend to support one theoretical stance against the other.

4. Review of the generative L2 literature

The acquisition of TAM systems by L2 learners from a generative perspec-


tive has been a growing line of empirical research since the early 2000s.
The rst studies focused on the acquisition of L2 English by Bulgarian
native speakers (Slabakova 2000, 2003), or both Bulgarian and Spanish
native speakers (Slabakova 2001) to test the semantic-morphology inter-
face, the acquisition of L2 Spanish by native speakers of English (Montrul
2002; Montrul and Slabakova 2002, 2003; Slabakova and Montrul 2000,
2002, 2003), as well as the dierences between Standard Russian and
American Russian in Russian-English bilingual speakers (Pereltsvaig 2002).
Then, a variety of studies started to target other L2 languages, such as
French (Leung 2002; Ayoun 2005a,b), Spanish (Schell 2000) and Portu-
guese (Goodin-Mayeda and Rothman 2007; Rothman and Iverson 2008).
Covering a wide range of L1 backgrounds, these studies investigate the L2
(or L3) acquisition of lexical aspect (telicity), grammatical or viewpoint
aspect (perfectiveness and iterativity), and mood/modality (the indicative
and subjunctive). The studies are reviewed in chronological order and
by L2 target language (English, Spanish, French, Portuguese, as well as
Russian and Spanish as heritage languages).

4.1. Tense and aspect in L2 English


Slabakova (2000, 2001) tested the acquisition of the telicity marking
parameter by Bulgarian-speaking learners of English. The telicity marking
parameter subsumes three related constructions verb-particle, resultative
secondary predicates, and double objects and presents two main contrasts:
(a) the [telic] morpheme occupies a dierent functional category in English
(i.e., AspP) as compared to Slavic (PerfP); and (b) the [telic] morpheme
is null in English, and depends on the objects cardinality, whereas it is
130 Dalila Ayoun and Jason Rothman

overt in Slavic, it is lexicalized as a pre-verb, and the objects cardinality


is irrelevant in establishing telicity. In Slabakova (2001), three groups of
L2 English learners and two groups of native speaker controls performed
four elicitation tasks: an aspect task, a translation task, a grammaticality
judgment task, and a stories task. Group and individual results indicated
that the learners ability and consistency in judging and distinguishing
between telic and atelic sentences increased with their level of prociency.
In Slabakova (2000), the hypothesis of L1 transfer was supported in
that the results of an aspectual interpretation task showed that both the
English native speakers and the Spanish L2 learners recognized the contrast
between telic and atelic sentences, because they preferred atelic sentences in
combination with a habitual context to telic sentences in a habitual context,
contrary to the Bulgarian L2 learners. The Bulgarian-speaking learners
were signicantly less accurate in their judgments of telic sentences than
atelic sentences indicating that they initially transferred the L1 value of
the aspectual parameter, maybe because they were still at beginning and
intermediate levels of prociency. Slabakova (2003) investigated whether
Bulgarian-speaking L2 learners of English in instructional settings are
able to acquire the functional category AspP with its semantic implica-
tions, none of which are part of the Bulgarian TAM system: (a) eventive
English verbs are inherently perfective; (b) the present tense is associated
with a quanticational feature and a generic operator; (c) progressive mor-
phology has a continuous interpretation; and (d) statives may have a habitual
interpretation in the simple past. Participants in low, high, and advanced
levels of competence performed an elicited production task and a truth
value judgment task (Crain and Thornton 1998; Grimshaw and Rosen
1990). The results of the production task revealed that learners had acquired
the simple and progressive aspect inectional morphology, whereas the re-
sults of the truth value judgment task indicated that the learners improved
in their ability to accurately map inectional morphology to appropriate
semantic contexts with their prociency level. Learners at the advanced
prociency level showed that they had acquired AspP and its semantic
values.
Testing the hypothesis that L2 learners cannot acquire new morpho-
syntactic features that are not instantiated in the L1, and examining the
suppliance of past morphology in production, Hawkins and Liszka (2003)
found that Chinese native speakers (n 2) performed worse than Japanese
(n 5) and German (n 5) native speakers in a retelling task using a short
extract from a Charlie Chaplin lm and a personal recount of a happy
experience. The Japanese and German native speakers produced regular
simple past tense morphology in over 90% of the obligatory contexts,
Generative perspectives 131

whereas the Chinese native speakers supplied the regular simple past tense
in about 62% of the obligatory contexts; however, they performed much
better with irregular simple past morphology by correctly inecting verbs
for simple past in roughly 84% of the obligatory contexts. Hawkins and
Liszka concluded that these results support the Failed Functional Features
hypothesis (Hawkins and Chan 1997), since only Chinese does not morpho-
syntactically encode tense. However, it is very dicult to make any general-
izable claims from such a small data set.
Campos (2009) claimed that L1 prosodic transfer might provide a better
explanation for Hawkins and Liszkas results since Chinese, similarly to
Spanish, does not allow for complex codas. Moreover, and although it is
not framed in a generative perspective, Yang and Yuan Huangs (2004)
study provided contradictory evidence to Hawkins and Liszkas study,
showing that 10 year-old to 19 year-old Chinese ESL learners (n 453 at
5 dierent levels of prociency, a much larger, and hence reliable, sample
than in Hawkins and Liszkas study) as instructed learners without expo-
sure to English outside the classroom achieved very high success rates
in marking past verbal morphology in the obligatory contexts (90.6% to
99.5%) of personal narratives.
In Gabriele, Martohardjono, and McClure (2005) (based on Gabriele
2005), native speakers of Japanese (n 83) at dierent levels of prociency
in L2 English intermediate (n 38) or advanced (n 45) were admin-
istered an interpretation task (Klein, Martohardjono, and Valian 1999)
designed to investigate whether Japanese learners interpretation of the
English past progressive would interact with L1 lexical semantics. Japanese
te-iru allows both progressive and perfective interpretations, depending on
the predicate, whereas English be ing always denotes a progressive inter-
pretation. Findings showed that the past progressive was more dicult
than the simple past for both activity and change of state verbs, even for
the advanced group learners. They allowed a perfective reading for the
English past progressive even when the L1 Japanese interpretation is pro-
gressive, leading Gabriele, Martohardjono, and McClure (2005) to suggest
that the perfective is a default interpretation in the mental representation
of te-iru.
Gabriele and Maekawa (2008) compared three groups of L2 English
learners native speakers of Chinese (n 32), Korean (n 18), and
Japanese (n 55) at dierent levels of prociency. Contrary to Japanese
and Korean, Chinese does not encode tense morpho-syntactically and pre-
sumably does not have a Tense projection (Lin 2005). The participants
completed an interpretation task with short stories focusing on the present
132 Dalila Ayoun and Jason Rothman

and past progressive in English as well as regular and irregular past tense
forms as distractors. The Chinese learners in the high prociency group
correctly distinguished the present from the past progressive, but the inter-
mediate prociency group was not as successful. The L1 Japanese and L1
Korean learners performed well as expected. Gabriele and Maekawas
(2008) ndings provide evidence against Hawkins and Liszkas (2003) pro-
posal of a permanent syntactic decit in the L2 when a feature such as
[epast] is not instantiated in the L1.
In Gabriele (2009), Japanese speakers (n 101) L2 English learners
at low, intermediate, and high levels of prociency performed a story com-
patibility task which targeted telic predicated with simple past (as a control),
present progressive, and past progressive morphology. Then, English native
speakers (n 33) L2 Japanese learners at two levels of prociency (low
and high) performed the Japanese equivalent of the story interpretation
task to test for the acquisition of the simple past marker -ta and the pro-
gressive marker te-iru with telic predicates. Both L2 Japanese and L2
English learners performed well on accomplishments with the progressive
and on achievements with simple past, but not as well on achievements
with the progressive -ing or imperfective marker te-iru. Moreover, the L2
English learners performed poorly on the preemption (i.e., L1 interpre-
tations) context at all prociency levels by accepting sentences in which
the present progressive referred to complete events. Preemption was less
dicult for L2 Japanese learners although some learners in the low-
prociency group allowed the imperfective marker te-iru to refer to ongo-
ing events. Gabriele (2009) suggested that the complexity of the semantic
computation and the (lack of ) transparency of input cues may explain
why preemption appears to be so dicult.
Liszka (2006) investigated the implications that a representational decit
approach may have on pragmatic processes from a Relevance theoretic per-
spective (Sperber and Wilson 1986/95) by asking French-speaking learners
of L2 English (n 8) to perform one written task (a narrative to elicit
events and states) and two oral tasks (description of a video and picture
clip to elicit states and ongoing events and a contextualized dialogue).
The focus was on the L2 acquisition of the distributional and interpreta-
tional properties of the English present simple and present progressive;
the latter is not grammatically encoded in L1 French. Following Al-
Hamad et al. (2002: 55), Liszka assumed that the dierence in temporal
interpretation between habitual and event in progress is directly con-
nected to whether the verb raises overtly to T(ense) in the syntax. Strong
uninterpretable features in French lead to overt verb raising whereas only
auxiliaries raise to T in the present progressive English. Results show that
Generative perspectives 133

participants production of the simple present was native-like and they


were successful in assigning target-like meaning to present simple forms
(100%) and present progressive forms (93.7%) (92), but learners over-
generalized the present simple forms to present progressive contexts in
the oral tasks.
In Liszka (2009), the ndings from a larger number of participants
(n 16) showed very high accuracy scores on assigning target-like mean-
ings to present simple forms (100%) and present progressive forms (92.7%)
in the oral description task. In the video clip description tasks, whereas the
accuracy scores for the present simple forms remained high (94.4%), the
use of present progressive forms dropped to 52.8% of obligatory contexts
which may not simply be a task eect due the real-time processing pressure
because a similar pattern was obtained for the contextualized dialogue
task (94.2% for simple present and 54.8% for present progressive).

4.2. Tense and aspect in L2 Spanish


L2 Spanish has been targeted by a large number of studies mostly with
English-speaking learners presumably because the two languages instantiate
partially overlapping TAM systems. In a rare longitudinal study of English-
speaking university students (n 5) participating in a study abroad pro-
gram in Spain, Schell (2000) assumed the existence of Inner and Outer
Aspect Phrases to account for preterite and imperfect past tenses in Spanish
(see also Ayoun and Salaberry 2005).5 This study focused on the interaction
between morphosyntax and semantics by examining how L2 learners
may acquire the form-meaning relationship of Spanish tense and aspect.
It assumed the three stages of the constructionist approach (Herschensohn
2000): feature strengths are initially unset and learners end up mastering
the lexical aspect of most verbs by the advanced stages, after the feature
values at the Inner Aspect Phrase have been specied. High accuracy rates
(85% to 100%) on three worksheets and a total of 4 written samples were
collected (one every two months for nine months) indicate successful
acquisition of past tense morphology.

5. In a personal communication (March 2011), Salaberry points out that Schells


(2000) study may be taken as evidence against the strictly syntactic view in
which adjuncts are considered as pragmatic in nature, adding that Schell spe-
cically points out that In an isolated monoclausal sentence in which there
are no adjuncts or aspectual references, the grammatical aspect assigned will
automatically match the perfectivity (or imperfectivity) of the Inner Aspect
Phrase (3536).
134 Dalila Ayoun and Jason Rothman

In a series of studies based on the analysis of Giorgi and Pianesi (1997)


for English and Spanish, S. Montrul and R. Slabakova investigated the
acquisition of the interpretive properties of the preterite and the imperfect
tenses in Spanish by English-speaking learners. First, in Slabakova and
Montrul (2000), the results of a truth value judgment task indicated that
intermediate learners were signicantly more accurate with the imperfect
than with the preterite when using state verbs such as poder be able to,
whereas the advanced group showed greater accuracy with both tenses,
suggesting that instructed L2 learners can learn properties that are usually
not explicitly taught. In Slabakova and Montrul (2002), participants per-
formed a sentence conjunction judgment task to investigate the interpreta-
tions L2 learners assign to preterite and imperfect past morphology.
Group and individual results revealed a clear developmental trend with
all advanced learners and some intermediate learners, demonstrating the
ability to dierentiate between the semantic features associated with gram-
matical aspect.6
Montrul and Slabakova (2002) rst used a morphology test to deter-
mine whether L2 learners could distinguish the morphological endings of
the preterite and the imperfect in a narrative. They then administered a
sentence conjunction judgment task designed to test the semantic implica-
tions of these two past tense forms. The group results showed that the L2
learners who had obtained accuracy scores of at least 80% on the mor-
phology test seemed to have acquired the semantic interpretations of the
preterite and the imperfect, whereas those who scored below 80% were still
struggling with the semantic contrasts, particularly with state and achieve-
ment predicates. The individual results suggested that the acquisition of
morphology precedes the acquisition of semantics, and that both types
of acquisition are gradual developments (140), although Slabakova (2006)
characterized this claim as premature considering later data indicating
that acquisition of morphological forms may be preceded by acquisition
of semantics. It was concluded that the formal features associated with the
functional category AspP are acquirable and unimpaired in SLA (141).
Montrul and Slabakova (2003) compared the performance of English-
speaking advanced learners and English near-native speakers of Spanish
with native speakers of Spanish on two tasks. The sentence-conjunction
judgment task did not reveal any dierences between the near-native

6. Salaberry (2008: 165170) argues that by rejecting the eect of adjuncts, the
eects of inner and outer aspect are conated into one single construct that is
methodologically dicult to tease apart.
Generative perspectives 135

speakers and the native speakers, whereas on the truth-value judgment


task, only the near-natives performed like the native speakers in recogniz-
ing the generic restriction on the preterite and specic-generic ambiguity
of the imperfect, the latter being clearly more dicult than the former for
the advanced and superior learners. It was however concluded that near-
native competence is possible in the domain of aspectual interpretations.
Following a prociency test and an inectional morphology recogni-
tion test (adapted from Salaberry 1997), Slabakova and Montrul (2003)
asked an intermediate group (n 40) and an advanced group (n 29) of
English-speaking learners of Spanish to perform a truth-value judgment
task with 80 story-sentence combinations to test their knowledge of the
preterite and the imperfect as well as the possible generic interpretation
of pronominal subjects. In Spanish, the subject pronoun can be interpreted
as having a generic meaning (people in general) or a specic meaning
(we) only when the predicate is in the imperfect; in the preterite, the
subject pronoun cannot have generic interpretation. Since this distinction
is not explicitly taught in foreign language classrooms, it presents a case of
poverty of the stimulus. It was found that the semantic properties of the
Spanish preterite and imperfect, the universal semantic conditionals (a
habitual clause reading implies a generic pronominal subject), and the
negative constraint on the generic interpretation of the preterite were
acquired by the advanced learners whereas the intermediate learners
displayed a clear developmental trend toward the successful acquisition
of these properties.7
The studies carried out by Montrul and Slabakova have been critically re-
viewed by Salaberry (2008: 177184), who claims, for instance, that the
methodologies in these studies do not consider possible coercion eects
that arise with endpoint adjuncts, thus confounding inner and outer
aspect. However, it is not clear that Montrul and Slabakova reject a priori
the eects of adjuncts. In fact, some of their most robust ndings are those
with achievement predicates for which adjuncts are simply not relevant.
For example, to sell the house and to reach the summit do not have
adjuncts, as the objects are clearly arguments. In a later study, Slabakova
and Montrul (2007) indeed examined the pragmatics of aspect and were
able to replicate the ndings from previous studies with achievements.8

7. But see Salaberry (2008: 177184) for a critique of this studys methodology.
8. More generally speaking, in my view, lexical aspect can depend on anything
that is in the VP, I mean the type of verb, arguments and adjuncts as in to go to
the park versus to go toward the park (Slabakova, personal communication).
136 Dalila Ayoun and Jason Rothman

Borgovono and Prevost (2003) tested the acquisition of polarity subjunc-


tive triggered by the presupposition/lack of presupposition contrast by
French-speaking learners of L2 Spanish (n 25) in two prociency groups.
Polarity subjunctive refers to a subjunctive licensed by an operator such as
a negation or an interrogation (Stowell 1993). In Spanish, polarity sub-
junctive is licensed with epistemic, perception and communication verbs,
whereas in French only negated epistemics license polarity subjunctive
in embedded clauses (153). However, the interpretations of the choice
of mood are the same in the two languages. The participants performed a
pen-and-paper truth-value interpretation task with 48 short scenarios in
French followed by a test sentence in Spanish in either the subjunctive or
indicative, depending on the presuppositional status of the lower clause.
The learners were expected to correctly allow the subjunctive in Spanish
perception and communication verbs (contrary to L1 French), while re-
jecting it when the lower clause does not contain a presupposition. Their
performance depended on their prociency level and revealed a gradual
learning process. Although they were not always accurate in their mood
selection, they displayed near-native accuracy in their performance.
Rothman (2008) proposes the Competing Systems Hypothesis to explain
why tutored L2 learners continue to produce non-target-like morphological
forms in spite of their advanced prociency: it is hypothesized that peda-
gogical rules create a separate system of learned knowledge that competes
with generative competence. The hypothesis was tested by examining the
performance of three groups of participants native speakers of Spanish
(n 20), advanced tutored L2 Spanish learners (n 20), and advanced
naturalistic L2 Spanish learners (n 11) on a binary-choice cloze test
and a ll-in-the-blank production task targeting the [eperfective] contrast
between the preterite and the imperfect. Findings reveal that half of the
tutored learners and all the naturalistic learners performed within native
speaker range on the cloze test. Overall, 25% of the tutored learners
diered from the native speakers on the production task, while there was
no dierence between naturalistic learners and native speakers. It was con-
cluded that the predictions of the Competing Systems Hypothesis were
conrmed because whereas naturalistic learners consistently performed in
a native-like fashion, the performance of the tutored learners (except for 4
participants who were native-like) revealed patterns traceable to the over-
simplied pedagogical rules they had been taught, suggesting that natural
positive evidence is most conducive to achieving a native-like grammar
since it is devoid of oversimplied prescriptive explanation. To be fair, it
is obvious that natural and classroom contexts represent complex environ-
Generative perspectives 137

ments that include several factors that can aect the outcome in each con-
text. That is, there may be interaction eects with other variables at play
(other than just pedagogical explanations). In fact, qualitative and quanti-
tative dierences in the input that naturalistic and classroom learners are
likely to have received, are variables that could be just as, if not more
explanatory, for the dierences Rothman discusses. However, we should
keep in mind that both groups of L2 learners are of highly advanced pro-
ciency and although the classroom learners were instructed learners, they
had more than mere classroom exposure over the course of time it took
for them to achieve their high prociency level. The point to be made is
that the two experiments were designed to focus on uses where pedago-
gical descriptions do not conict with naturalistic input and one where
pedagogical simplication does. The results across the two groups show
that it is reasonable to conclude that pedagogical rules play a role in the
performance on the tasks. Thus, even if other co-occurring variables are
at play, Rothmans conclusions remain reasonable.
Following impairment hypotheses according to which uninterpretable
features not present in the L1 are not acquirable in the L2, Diaz, Bel, and
Bekiou (2007) assume that only L2 Spanish learners whose L1 encodes the
[eperfective] aspectual distinction morpho-syntactically are able to acquire
the Spanish preterite-imperfect contrast. Specic learnability predictions are
made for each L1 language background or type (Romance, Modern Greek,
Chinese, Japanese, and Slavic). Seventy participants at intermediate and
advanced levels of prociency (but few participants per language group
and prociency level, from 5 to 12) completed three written stories by pro-
viding the appropriate verb form in the past. Findings show that: (a) the
presence of the [eperfective] feature in L1 helps the Romance and Greek
L2 learners; (b) eects of morphological contrast with Spanish were found
mostly for the Slavic group but also the Asian group; and (c) eects of the
Aspect Hypothesis (Aspect before Tense) are conrmed only for states and
achievements. Finally, it is argued that diculties encountered across
language groups with activities and accomplishments, the two aspectual
classes that are sensitive to the cardinality of the DP object, appear to
spring from the interaction between interpretable and uninterpretable fea-
tures (507).
Hsien-jen Chin (2008) is another study investigating the acquisition of
the [eperfective] contrast between the Spanish preterite and imperfect by
Chinese- and English-speaking learners. Both English and Spanish overtly
mark tense morphology whereas Chinese uses aspectual markers for the per-
138 Dalila Ayoun and Jason Rothman

fective, experiencer, durative, and progressive. Three groups of participants


English-speaking (n 22) and Chinese-speaking (n 14) L2 Spanish
learners, and Spanish native speakers (n 11) as controls were adminis-
tered a morphology test and an acceptability test (following Montrul and
Slabakova 2002, 2004). The morphology test revealed signicant dierences
among the three groups (from 70.27% to 75.76% for the L2 learners)
following L1 inuences. The group results on the acceptability test show
that overall, the English L1 learners distinguished contrasts on telic predi-
cates but not on states, whereas the Chinese L1 learners did not appear to
be sensitive to any of the contrasts. The individual results indicate that
only a small number of English L1 learners recognized the contrasts, and
that there was no clear prociency eect. However, these ndings should
be considered with caution because of the low prociency level of the L2
learners, and also because the Chinese-speaking participants were actually
L3 Spanish learners.
Iverson, Kempchinsky, and Rothman (2008) focus on the syntax-
pragmatics interface by testing the acquisition of two types of subjunctive
complements in intermediate and advanced English learners of L2 Spanish:
complements to volitional verbs (e.g., querer to want) which exemplify a
purely syntactic relationship between the matrix verb and the subordinate
clause compared to complements with negated epistemic verbs (e.g., no
creer to not believe), which involve the syntax-discourse interface in the
mood selection (indicative versus subjunctive). In English, volitional com-
plements use the subjunctive only formal registers (the indicative and
for-to indicative clauses are preferred) while negated epistemic verbs are
only found in the indicative. The participants a native control group
(n 13) and two L2 groups, one advanced (n 18) and one intermediate
(n 25) performed a grammaticality judgment task testing indicative
and subjunctive complement clauses with volitional and negated epistemic
verbs whose results reveal signicant dierences between groups on all
sentence types. However, post hoc tests did not detect any signicant
dierences between the advanced learners and the native speakers in any
rating of categories, nor did the intermediate learners dier signicantly
from the native speakers except in negated epistemics with subjunctive
embedded clauses.
Bruhn de Garavito and Valenzuela (2008) administered a grammaticality
judgment task and a sentence selection task to an Anglophone group of
advanced L2 Spanish learners (n 21) and a group of Spanish native
speakers (n 21) to test the aspectual distinction between two types of
passives: eventive passives are marked with ser, which is aspectually neutral,
Generative perspectives 139

while stative passives are marked with estar, which expresses the perfective.
Both copulas mean to be. In English, the aspectual distinction between
eventives and statives is not marked in any specic way, hence the diculty
for Anglophone native speakers to make the appropriate selection between
the two Spanish copulas. The ndings are mixed in that the L2 learners
performance on both tasks showed that they distinguished between the two
copulas followed by adjectives, but on the sentence selection task, their
performance with regards to the interpretation of the subject was random.
It is likely that, as suggested by de Garavito and Valenzuela (334), the
problem is related to processing at the interface between semantics and
syntax (see also Sorace 2006).

4.3. Tense and aspect in L2 French and Portuguese


Leung (2002, 2005) compared L2/L3 French learners Vietnamese mono-
lingual and Cantonese-English bilingual learners at beginning and advanced
levels of prociency along with English and French controls to test the
Full Access/Full Transfer hypothesis and Failed Functional Features hypo-
thesis by administering four dierent elicitation tasks: two written production
tasks (a composition task to elicit past tense morphology, a sentence com-
pletion task to elicit verbal agreement and adverbs) and two preference
tasks (one to test for Case, niteness, and agreement and the second to
test for adverb placement). The ndings, which reveal near-native-like per-
formance on agreement and adverbs as well as niteness and agreement,
show L1 transfer at the initial stages, but also that features which are not
present in the L1 are acquirable in the L2 contra the Failed Functional
Features hypothesis, which claims the oppositive (i.e., L2 features that
are not present in the L1 will not be fully acquired).
Ayoun (2005a) investigated the acquisition of past temporality by English-
speaking learners of French (n 37) as a foreign language in an instructed
setting at three dierent prociency levels (a group of French native speakers
(n 14) served as controls) and Ayoun (2005b) investigated the acquisition
of future temporality with the same participants. In Ayoun (2005a), the
analyses of the personal narratives revealed that the interlanguage of L2
learners displayed well-formed sentences with correctly inected verbs for
tense, person, and number as well as appropriate negation and adverb
placement. These ndings indicate that the interlanguage of these L2
learners projects the appropriate functional categories associated with
strong features. They also seem to have acquired semantic contrasts, but
with varying degrees of mastery across the dierent lexical classes. The
140 Dalila Ayoun and Jason Rothman

results of the cloze test, however, revealed signicant dierences between


groups across all aspectual classes. Thus, the prediction that L2 learners
of French would be successful in acquiring the semantic properties of the
functional category AspP was only supported by the results of the personal
narratives, which was explained as a task eect. The second and third pre-
dictions (that learners would be successful in acquiring target-like use of
inectional morphology, and that they would be successful in acquiring
the appropriate feature strength of AspP) appear to be supported, as
evidenced by their productive use of perfective and non-perfective mor-
phology in the personal narratives. Moreover, it was also found that the
L2 learners did not seem to have diculties in the mapping of surface
morphology: almost all past participles and lexical verbs were properly
inected for tense, number, and gender, with rare gender agreement errors.
The tense/aspect system of these L2 learners is not yet target-like, but their
current system shows contrasts and systematicity, signalling the appropriate
feature strength of AspP and its semantic properties. In Ayoun (2005b),
the two elication tasks produced mixed ndings: the L2 learners performed
better on the personal narrative task than on the cloze task: the latter re-
vealed signicant variations depending on the morphological forms and
the predicates. There was also a clear prociency eect from the lowest
to the highest levels of prociency. Nevertheless, all three learner groups
used a variety of morphological forms to express futurity in their personal
narratives, indicative of the acquisition of temporal and modal values
associated with the future.
Iverson and Rothman (2008) asked advanced English learners of adult
L2 Portuguese (n 17) to perform two sentence conjunction judgment
tasks to test their knowledge of [eaccidental] interpretative nuances with
preterite and imperfect adverbially quantied sentences. Results show a
native-like performance of the advanced L2 learners in 10 out of 11 contexts,
providing evidence that morphosyntactic features and phrasal semantics
can be acquired by adult learners contra impairment hypotheses.9 In a

9. We should acknowledge that apparently some theoretical authors and even


native speaker judgments shed some doubt on the extent to which the seman-
tic distinctions being tested here are truly generalizable and/or accurate. Of
course, we believe that the semantic nuances at play here are indeed general-
izable, albeit extremely subtle. As it relates to the methodology in Iverson and
Rothman (2008) and Rothman and Iverson (2008), there perhaps is a con-
found based on the people who were tested in that both groups represent
highly educated speakers and it might be that they are able to get at these
interpretations more easily given the inherent complexity of the tasks them-
selves.
Generative perspectives 141

related study, Rothman and Iverson (2008) tested 31 learners of L2 Portu-


guese focusing on other semantic entailments related to the aspectual dis-
tinction of adverbial quantication whereby only such perfective sentences
lose the otherwise available kind-denoting reading of the subject DP,
retaining only the group-denoting reading (e.g., Menendez-Benito 2002).
Rothman and Iverson (2008) showed that the same learners from Iverson
and Rothman (2008), who demonstrated the target [eaccidental] interpre-
tative nuances, were also successful in interpreting restrictions on DP inter-
pretations, related to the same aspectual distinction (but see Chapter 6).

4.4. Tense and aspect and heritage learners


Investigating the dierences between Standard Russian and American
Russian (the variety of Russian spoken by second-generation Russian-
speaking immigrants in the U.S.) led Pereltsvaig (2002) to claim that
American Russian encodes lexical aspect, rather than grammatical aspect.
She followed Travis (2000) in assuming a syntactic structure with two func-
tional projections: InnerAspP to encode lexical aspect located between
the two VP-shells and OuterAspP, to encode grammatical aspect, located
outside the VP-shells in the I-I-domain. Pereltsvaig argues that in Standard
Russian, aspectual morphology is associated with OuterAspo, whereas in
American Russian it is associated with InnerAspo. Past/nonpast tense
morphology is in To in both Standard and American Russian. She further-
more assumes that vP encodes agentivity/transitivity instead of lexical
aspect (contra e.g., Pinon 1995; Verkuyl 1999; Slabakova 2001). The anal-
ysis is based on 150 verb forms extracted from a corpus representing the
Russian spoken by English L2 dominant speakers for whom Russian is
the rst but clearly a secondary language that they use rarely, and in which
they are illiterate. It appears that the American Russian aspectual system
does not mark a clear preference for either the perfective or the imperfec-
tive aspect, and most predicates are always used in the same aspectual
form, leading Pereltsvaig to conclude that in American Russian, aspec-
tual marking encodes a lexical semantic property of the verb (7). She
further observes that American Russian should be characterized as lack-
ing syntactic agreement altogether; in other words, American Russian
speakers cannot do either subject-verb or DP-internal agreement through
feature-checking or feature-matching (15). Thus, with respect to func-
tional projections, lexical aspect is marked by aspectual morphology,
but grammatical aspect is not marked at all (i.e., OuterAspP is missing)
whereas tense is marked correctly, but subject-verb agreement is not.
142 Dalila Ayoun and Jason Rothman

Pereltsvaig rejects the possibility of L2 transfer because it makes two pre-


dictions which are not borne out: imperfective morphology is not used
similarly to English -ing, and perfective morphology does not mirror
English perfect aspect. The loss of uninterpretable features is accounted
for by L1 attrition, dened as incomplete L1 acquisition, because these
speakers become English-dominant before puberty, and do not benet from
exposure to Standard Russian.
Montrul (2002) compared bilingual speakers of English and Spanish as
a heritage language with monolingual Spanish speakers to investigate
whether the age of onset of bilingualism inuences ultimate attainment.
Assuming AspP to be a functional category where the [eperfective] fea-
tures are checked, Montrul (2002) addresses the question of whether both
the morphological features and the semantic features are aected by lan-
guage loss, or whether bilingual speakers display patterns of incomplete-
ness in the Spanish preterite/imperfect contrast. The results of the inter-
pretation and production tasks revealed signicant dierences between
bilingual and monolingual speakers on achievement predicates in the
imperfect, and stative verbs in both the preterite and the imperfect,
whereas the early child bilingual speakers diered from the monolingual
speakers on stative verbs in the preterite. In other words, the simultaneous
bilinguals displayed incomplete acquisition, whereas the early child bilinguals
showed some evidence of attrition, mostly for stative verbs in the preterite.
The wide range of prociency shown by the bilingual speakers also con-
tributed to the conclusion that, contra Toribio (2001), for many partici-
pants, the loss of morphological features is accompanied by the loss of the
formal features, which provides support for the critical period hypothesis.
Montrul (2007) examines the status of the subjunctive mood in heritage
bilingual Spanish grammars. Adopting the basic tenets of Soraces (2005)
Interface Hypothesis to make predictions on where heritage bilingual
competence is more likely to match and diverge from monolingual compe-
tence, this study demonstrates that heritage speakers grammars clearly
control the underlying syntax of the subjunctive mood, yet dier from
native monolinguals in interesting ways. Heritage bilinguals productively
use dedicated subjunctive morphology in appropriate contexts: they per-
form like monolinguals when the subjunctive is subcategorized by the
matrix verb as complements of volitional predicates. However, heritage
speakers dier from monolinguals signicantly with respect to the distri-
bution of the subjunctive in environments where both the indicative and
subjunctive are possible. However, where the pragmatics favors the use of
one or the other to signal a subtle shift in semantic interpretation, heritage
Generative perspectives 143

speakers dier from monolinguals signicantly. For example, after phrases


such as mientras que while or cuando when, after which both indicative
or subjunctive use are delimited for monolinguals depending on the dis-
course, heritage speakers default to using the indicative. We agree with
Montrul that such evidence supports the Interface Hypothesis.

5. Discussion and conclusion

Space limitations being what they are, we are unable to cover in absolute
depth or even exhaustively all the available studies examining the L2
acquisition of TAM from a generative perspective. However, what we
have been able to show in the selection of studies we have reviewed are
the wide variety of language pairings and the depth of the questions that
comprise the research of L2 TAM acquisition within the UG paradigm.
As is true of all L2 studies from a UG perspective, in addition to their
earnest attempt at documenting and explaining the developmental route
of TAM acquisition in a way that meets descriptive and explanatory ade-
quacy, these studies also contribute to theory internal questions of a larger
remit as well as provide important data sets towards larger questions of
great relevance to SLA research more generally. These studies combine
to inform questions pertaining to the age-of-acquisition and critical period
for L2 acquisition debates. They provide crucial evidence towards whether
or not and the extent to which the same mechanisms implicated in
child acquisition remain available to adult learners in the course of L2
acquisition. Because UG based studies on L2 TAM acquisition adopt
formal theoretical proposals as their point of reference, they are also able
to provide a unique testing ground for theoretical proposals of formal
syntax and semantics.
As it relates to the whole of the data that we have reviewed, there is a
noticeable developmental trend showing that learners in foreign language
instructed settings show gradual improvement from the beginning to more
advanced prociency levels to eventually successfully acquiring the tense/
aspect systems of their respective L2s. The studies that tested near-native
speakers also arrived at the conclusion that ultimate attainment was possible,
while stressing that TAM appears to be a sensitive domain of L2 language
acquisition that may be subjected to some critical period and/or learnability
eects, be it for bilingual speakers of heritage languages (Montrul 2002,
2007), or for English-American Russian bilingual speakers who become
dominant in English before puberty, and fail to be exposed to Standard
Russian (Pereltsvaig 2002). Even studies which set out to test permanent
144 Dalila Ayoun and Jason Rothman

syntactic decit proposals in the L2 when an L2 feature is not instantiated


in the L1 did not nd strong evidence that adult learners are inevitably
impaired in their acquisition of functional categories and their features.
Quite to the contrary, learners appear to arrive at a TAM system with
well established contrasts and systematicity. If adult L2 acquisition were
indeed permanently impaired for new feature acquisition, learners perfor-
mance would not display such high accuracy rates in obligatory contexts
and across a variety of elicitation tasks, as was seen in the majority of the
research we reviewed.
Although we are able to highlight the apparent success of some L2
learners in the domains of L2 TAM acquisition, there are obvious diculties
inherent to TAM acquisition in adulthood. Such diculties seem to reside
in the complexities of TAM systems and arise in dierent ways for various
learners depending on the L1 and L2 pairing. Subcomponents of tense,
aspect, and modality can be more or less formally complex and this is
seen not only in the developmental sequencing of TAM in child L1 acqui-
sition, but also echoed in the trends that obtain in L2 TAM studies as
well. For example, studies on the acquisition of subjunctive modality in
Spanish as a heritage language and as an L2 (e.g., Montrul 2010; Iverson
et al. 2008) show dierential success in ultimate attainment and timing of
convergence of sub-properties in interlangue developmental route depend-
ing on the interfaces implicated. Unlike other TAM properties, such as
TAM-related morphological paradigms and their canonical distributional
use, for example when subjunctive modality is strictly subcategorized by
volitional predicates, interface-conditioned properties at the syntax-discourse
interface (so-called polarity subjunctive with epistemic predicates) are more
subject to vulnerability. Interestingly, these same properties are most often
not systematically taught in the classroom. However, we do not take lack
of instruction to be the source of L2 dierences because acquisition, as
opposed to learning, is largely believed within the UG paradigm to be
independent of instruction (see White 1991; Schwartz and Gubala-Ryzak
1992). As outlined in the empirical literature, it has been decidedly shown
that many related properties that are not explicitly taught, come to be
acquired and are part of the mental representation of the L2. In this
regard, we highlight the studies by Slabakova and Montrul (2003) and
Iverson et al. (2008), both of which exemplify L2 learners at advanced
stages of Spanish prociency who demonstrate native-like knowledge of
untaught complex semantic entailments (properties at the syntax-semantics
interface and even syntax-discourse pragmatics) related to L2 aspectual and
modality features that are morphosyntactically dierent in the learners
Generative perspectives 145

L1. In both these studies, as well as many others we reviewed, it did


not seem to be the case that interface properties inevitably lead to non-
convergence (i.e., the learnerss L2 output in not target-like) or to residual
optionality (i.e., the learners interlanguage incorrectly displays two forms
as optional).
In recent years, Sorace and colleagues (see Sorace 2011 for a review) have
modied the Interface Hypothesis to claim that some interfaces are more
complex and formidable for L2 acquisition than others. The most current
instantiation of this hypothesis maintains that internal interfaces should be
less problematic than external ones for L2 convergence, due to dierences
in the attentional (processing) resource assignment needed (Sorace and
Serratrice 2009; Sorace 2011). Dividing interfaces into internal and external
interfaces still leaves the hypothesis amenable to the comparative exami-
nation of L2 TAM studies. An immediate consequence of the predictive
dierences between L2 internal and external interfaces for TAM explains
the general success that advanced and near-native L2 learners demonstrate
in syntax-semantics and lexicon-semantics-syntax related to L2 TAM systems:
these are internal interfaces (e.g., Slabakova 2001; Slabakova and Montrul
2003; Rothman and Iverson 2008) and the seemingly greater diculty
(residual optionality) for TAM properties tend to appear at the syntax-
pragmatics interface, an external interface (e.g., Iverson et al. 2008). How-
ever, whereas external interfaces might mean more protracted convergence
in general, evidence from L2 TAM acquisition, namely some of the indi-
vidual data from Iverson et al. (2008) do not seem to support the conten-
tion that residual optionality is an inescapable outcome of L2 acquisition.
Future research that examines more TAM properties relevant to both
internal and external interfaces will be in a privileged position to oer evi-
dence in support of this general line of research or add empirical evidence
towards its more precise renement. Since the most recent instantiation
of the Interface Hypothesis envisions the diculty in external interfaces
to stem from limitations in processing capacity for learners who have
more than one mental grammar in the same mind, more work using online
methodologies (eye-tracking, fMRI, ERP, and other technologies) are needed
and are to date largely non-existent within the UG tradition of L2 TAM
acquisition studies. The dearth in empirical work is promising since it oers
the opportunity for new research that can go beyond what has already
been shown in the well-studied area of generative SLA.
Although there has been signicant work examining L2 TAM systems
under the generative paradigm over the past two decades, there is much
room for more studies. Particularly welcomed will be those that expand
146 Dalila Ayoun and Jason Rothman

the L1/L2 pairings to include less commonly studied languages whose


complexity will help us to take the L2 study of TAM to a dierent level,
improving the methodologies used, the questions we ask, and the corpora
of evidence that would allow us to draw more complete conclusions about
SLA in general.

References

Al-Hamad, Majed, Eidah Al-Malki, Gabriela Casillas, Florencia Franceschina,


Roger Hawkins, James Hawthorne, Daniela Karadzovska, Kazue
Kato, Sarah Liszka, Cristobal Lozano, Shiro Ojima, Natsumi
Okuwaki & Emma Thomas
2002 Interpretation of English tense morphophonology by advanced
L2 speakers. EUROSLA Yearbook 2. 4969.
Ayoun, Dalila
2003 Parameter setting in language acquisition. London: Continuum
Press.
Ayoun, Dalila
2005a The acquisition of tense and aspect in L2 French from a Universal
Grammar perspective. In D. Ayoun & R. Salaberry (Eds.), Tense
and aspect in Romance Languages: Theoretical and applied per-
spectives, 79127. Amsterdam: John Benjamins.
Ayoun, Dalila
2005b The development of the future in L2 French. Paper presented
at the Second Language Research Forum, Teachers College,
Columbia University.
Ayoun, Dalila & M. Rafael Salaberry (Eds.)
2005 Tense and aspect in Romance Languages: Theoretical and applied
perspectives. Amsterdam: John Benjamins.
Beck, Maria
1998 L2 acquisition and obligatory head movement: English-speaking
learners of German and the local impairment hypothesis. Studies
in Second Language Acquisition 20. 311348.
Bley-Vroman, Robert
2009 The evolving context of the fundamental dierence hypothesis.
Studies in Second Language Acquisition 31. 175198.
Bley-Vroman, Robert
1990 The logical problem of foreign language learning. Linguistic Anal-
ysis 20. 349.
Borer, Hagit
1984 Parametric syntax. Dordrecht: Foris.
Borer, Hagit
2005 The normal course of events. Structuring tense. Volume 2. Oxford:
Oxford University Press.
Generative perspectives 147

Borgonovo, Claudia & Phillipe Prevost


2003 Knowledge of polarity subjunctive in L2 Spanish. In B. Beachley,
A. Brown & F. Conlin (Eds.), Proceedings of BUCLD 27, 97108.
Sommerville, MA: Cascadilla Press.
Bybee, Joan
2001 Phonology and language use. Cambridge: Cambridge University
Press.
Campos, Gonzalo
2009 L2 production of English past morphology in advanced Spanish
natives: syntactic decits or phonotactic transfer? In M. Bowles,
T. Ionin, S. Montrul & A. Tremblay (Eds.), Proceedings of the
10th Generative Approaches to Second Language Acquisition Con-
ference (GASLA 2009), 210219. Somerville, MA: Cascadilla
Proceedings Project.
Chomsky, Noam
1959 A review of B. F. Skinners Verbal Behavior. Language 35.
2658.
Chomsky, Noam
1965 Aspects of the theory of syntax. Cambridge: M.I.T. Press.
Chomsky, Noam
1981 Lectures on government and binding. Dordrecht: Foris.
Chomsky, Noam
1992 A minimalist program for linguistic theory. MIT Occasional
Papers in Linguistics 1. 167.
Chomsky, Noam
1993 A minimalist program for linguistic theory. In K. Hale & S. J.
Keyser (Eds.), The view from Building 20: Essays in linguistics in
honor of Sylvain Bromberger, 152. Cambridge, MA: MIT Press.
Chomsky, Noam
1995 The minimalist program. Cambridge, MA: MIT Press.
Chomsky, Noam
2001 Derivation by phase. In M. Kenstowicz (Ed.), Ken Hale: A life
in language, 152. Cambridge, MA: MIT Press.
Clahsen, Harald & Upyong Hong
1995 Agreement and null subjects in German L2 development: New
evidence from reaction-time experiments. Second Language Research
11. 5787.
Crain, Stephen & Rosalind Thornton
1998 Investigations in Universal Grammar. Cambridge, MA: MIT Press.
Daz, Lourdes, Aurora Bel & Konstantina Bekiou
2007 The role of morphological features in the acquisition of Spanish
aspectual dierences. In J. Liceras, H. Zobl & H. Goodluck
(Eds.), The role of formal features in second language acquisition,
485511. Mahwah, NJ: Lawrence Erlbaum.
148 Dalila Ayoun and Jason Rothman

DeMiguel, Elena
1992 El Aspecto en la sintaxis del espanol: Perfectividad e imperfectivi-
dad [Aspect in Spanish syntax: Perfectivity and imperfectivity].
Madrid: Ediciones de la Universidad Autonoma de Madrid.
Ellis, Nick
2008 The dynamics of second language emergence: cycles of language
use, language change, and language acquisition. The Modern
Language Journal 92(2). 232249.
Gabriele, Alison
2005 The acquisition of aspect in a second language: A bidirectional
study of learners of English and Japanese. New York, NY:
CUNY Graduate Center, Ph.D. dissertation.
Gabriele, Alison
2008 Calculating telicity in native and non-native English. In R.
Slabakova, J. Rothman, P. Kempchinsky & E. Gavruseva (Eds.),
Proceedings of the 9th Generative Approaches to Second Language
Acquisition Conference (GASLA 2007), 3746. Somerville, MA:
Cascadilla Proceedings Project.
Gabriele, Alison
2009 Transfer and transition in the SLA of aspect. Studies in Second
Language Acquisition 31. 371402.
Gabriele, Alison & Alonso Canales
2011 No time like the present: Examining transfer at the interfaces in
second language acquisition. Lingua 121. 670687.
Gabriele, Alison & Junko Maekawa
2008 Interpreting tense in a second language. EUROSLA Yearbook 8:
79106.
Gabriele, Alison, Gita Martohardjono & William McClure
2005 Evaluating the role of the L1 in the L2 acquisition of aspect. In
J. Cohen, K. T. McAlister, K. Rolstad & J. MacSwan (Eds.),
Proceedings of the 4th International Symposium on Bilingualism,
808826. Somerville, MA: Cascadilla Press.
Giorgi, Alessandra & Fabio Pianesi
1997 Tense and aspect: From semantics to morphosyntax. Oxford:
Oxford University Press.
Goad, Heather & Lydia White
2008 Prosodic structure and the representation of functional morphol-
ogy: a nativist approach. Lingua 118. 577594.
Goodin-Mayeda, Elizabeth & Jason Rothman
2007 The acquisition of aspect in Portuguese Spanish: Exploring native/
non-native performance dierences. In S. Baauw, F. Dirjkoningen
& M. Pinto (Eds.), Romance languages and linguistic theory 2005,
131148. Amsterdam: John Benjamins.
Grimshaw, Jane & Sara Rosen
1990 Knowledge and obedience: The developmental status of the
Binding theory. Linguistic Inquiry 21. 188221.
Generative perspectives 149

Guasti, Mara Teresa


2002 Language acquisition: The growth of grammar. Cambridge, MA:
MIT Press.
Hamann, Cornelia
2008 The rst language acquisition of French from a generative per-
spective. In D. Ayoun (Ed.), Studies in French Applied Linguis-
tics, 135181. Amsterdam: John Benjamins.
Hawkins, Roger
2001 Second language syntax: A generative introduction. Malden, MA:
Blackwell.
Hawkins, Roger
2003 Representational decit theories of adult SLA: Evidence, coun-
terevidence and implications. Plenary talk given at EUROSLA,
Edinburgh, UK, September 2003.
Hawkins, Roger, Gabriela Casillas, Hajime Hattori, James Hawthorne, Ritta
Husted, Cristobal Lozano, Aya Okamoto, Emma Thomas &
Kazumi Yamada
2008 The semantic eects of verb raising and its consequences in
second language grammars. In J. M. Liceras, H. Zobl & H.
Goodluck (Eds.), The role of formal features in second language
acquisition, 328351. New York: Lawrence Erlbaum.
Hawkins, Roger & Cecilia Yuet-hung Chan
1997 Partial availability of Universal Grammar in second language
acquisition: The failed formal features hypothesis. Second Lan-
guage Research 13. 187226.
Hawkins, Roger & Hajime Hattori
2006 Interpretation of English multiple wh-questions by Japanese
speakers: A missing uninterpretable feature account. Second Lan-
guage Research 22(3). 269301.
Hawkins, Roger & Sarah Liszka
2003 Locating the source of defective past tense marking in advanced
L2 English speakers. In R. van Hout, A. Hulk, F. Kuiken & R.
Towell (Eds.), The lexicon-syntax interface in second language
acquisition, 2144. Amsterdam: John Benjamins.
Herschensohn, Julia
2000 The second time around. Minimalism and L2 acquisition. Amster-
dam: John Benjamins.
Hsien-je Chin, Diana
2008 A cross-linguistic investigation on the acquisition of Spanish
aspect. In J. Bruhn de Garavito & E. Valenzuela (Eds.), Selected
proceedings of the 10th Hispanic Linguistics Symposium, 3650.
Somerville, MA: Cascadilla Proceedings Project.
Hyams, Nina
1986 Language acquisition and the theory of parameters. Dordrecht:
Reidel.
150 Dalila Ayoun and Jason Rothman

Iverson, Michael, Paula Kempchinsky & Jason Rothman


2008 Interface vulnerability and knowledge of the subjunctive/indica-
tive distinction with negated epistemic predicates in L2 Spanish.
EUROSLA Yearbook 8. 135163.
Iverson, Michael & Jason Rothman
2008 Adverbial quantication and perfective/imperfective interpretive
nuances in L2 Portuguese. In R. Slabakova, J. Rothman, P.
Kempchinsky & E. Gavruseva, E. (Eds.), Proceedings of the 9th
Generative Approaches to Second Language Acquisition Con-
ference (GASLA 2007), 7080. Somerville, MA: Cascadilla Pro-
ceedings Project.
Jackendo, Ray
2002 Foundations of language: Brain, meaning, grammar, evolution.
Oxford: Oxford University Press.
Kempchinsky, Paula
2009 What can the subjunctive disjoint reference eect tell us about
the subjunctive? Lingua 119. 17881810.
Klein, Elaine, Gita Martohardjono & Virginia Valian
1999 Learning and using a rst and second language. CUNY Collabo-
rative Grant.
Lardiere, Donna
2009 Some thoughts on the contrastive analysis of features in second
language acquisition. Second Language Research 25. 173227.
Lardiere, Donna
2003 Second language knowledge of [ePast] versus [eFinite]. In J.
Liceras, H. Zobl & H. Goodluck (Eds.), Proceedings of the 6th
Generative Approaches to Second Language Acquisition: L2 Links,
176189. Somerville, MA: Cascadilla Proceedings Project.
Leung, Yan-kit Ingrid
2002 Functional categories in second and third language acquisition:
a cross-linguistic study of the acquisition of English and French
by Vietnamese and Chinese speakers. Montreal: Canada, McGill
University, Ph.D. dissertation.
Leung, Yan-kit Ingrid
2005 L2 versus L3 initial state: a comparative study of the acquisition of
French DPs by Vietnamese monolinguals and Cantonese-English
bilinguals. Bilingualism: Language and Cognition 8. 3961.
Lin, Jo-wang
2003 Temporal reference in Mandarin Chinese. Journal of East Asian
Linguistics 12. 259311.
Lin, Jo-wang
2005 Time in a language without tense: The case of Chinese. Journal
of Semantics 23. 153.
Liszka, Sarah Ann
2006 Advanced grammars and pragmatic processes. Exploring the
interface. EUROSLA Yearbook 6. 7999.
Generative perspectives 151

Liszka, Sarah Ann


2009 Associating meaning to form in advanced L2 speakers. An inves-
tigation into the acquisition of the English present simple and
present progressive. In N. Snape, Y. K. Leung & M. Sharwood
Smith (Eds.), Representational decits in SLA: Studies in honor
of Roger Hawkins, 229246. Amsterdam: John Benjamins.
Meisel, Jurgen
1997 The acquisition of the syntax of negation in French and German:
contrasting rst and second language development. Second Lan-
guage Research 13. 227263.
Menendez-Benito, Paula
2002 Aspect and adverbial quantication in Spanish. In M. Hirotani
(Ed.), Proceedings of the 32nd North Eastern Linguistics Society,
Graduate Linguistic Student Association, 365382. University of
Massachusetts, Amherst.
Montrul, Silvina
2002 Incomplete acquisition and attrition of Spanish tense/aspect dis-
tinctions in adult bilinguals. Bilingualism: Language and Cogni-
tion 5. 3968.
Montrul, Silvina
2005 Second language acquisition and rst language loss in adult early
bilinguals: Exploring some dierences and similarities. Second
Language Research 21. 199249.
Montrul, Silvina
2007 Interpreting mood distinctions in Spanish as a heritage language.
In K. Potowski & R. Cameron (Eds.), Spanish in contact. Policy,
social and linguistic inquiries, 2340. Amsterdam: John Benjamins.
Montrul, Silvina & Roumyana Slabakova
2002 Acquiring morphosyntactic and semantic properties of preterite
and imperfect tenses in L2 Spanish. In A. T. Perez-Leroux &
J. Liceras (Eds.), The acquisition of Spanish morphosyntax: The
L1-L2 connection, 113149. Dordrecht: Kluwer.
Montrul, Silvina & Roumyana Slabakova
2003 Competence similarities between native and near-native speakers:
an investigation of the preterite/imperfect contrast in Spanish.
Studies in Second Language Acquisition 25. 351398.
OGrady, William
2005 Syntactic carpentry: An emergentist approach to syntax. Mahwah,
NJ: Lawrence Erlbaum.
Pereltsvaig, Asya
2002 Aspect lost, aspect regained: Restructuring of aspectual marking
in American Russian. Unpublished manuscript. University of
Troms. Available at http://www.hum.uit.no/a/pereltsvaig/pro-
fessional.htm.
152 Dalila Ayoun and Jason Rothman

Perez-Leroux, Ana, Alejandro Cuza, Monika Majzlanova & Jeanette Sanchez


Naranjo
2007 Non-native recognition of the iterative and habitual meanings of
Spanish preterite and imperfect tenses. In J. Liceras, H. Zobl &
H. Goodluck (Eds.), Formal features in second language acquisi-
tion, 432451. Amsterdam: John Benjamins.
Pinker, Steven
1994 The language instinct: How the mind creates language. New York:
Harper Collins Publishers.
Pinon, Christopher
1995 An ontology for event semantics. Stanford, CA: Stanford Univer-
sity, Ph.D. dissertation.
Pollock, Jean-Yves
1989 Verb movement, Universal Grammar, and the structure of IP.
Linguistic Inquiry 20. 365424.
Pollock, Jean-Yves
1997 Langage et cognition. Introduction au programme minimaliste de
la grammaire generative. Paris: Presses Universitaires de France.
Prevost, Phillipe & Lydia White
2000 Missing surface inection or impairment in second language
acquisition? Evidence from tense and agreement. Second Lan-
guage Research 16. 103134.
Pullum, Georey & Barbara Scholz
2002 Empirical assessment of stimulus poverty arguments. The Linguis-
tic Review 19 (special issue, Nos. 12: A Review of The Poverty
of Stimulus Argument, N. Ritter, Ed.): 950.
Quer, Josep
1998 Mood at the interfaces. Utrecht, Amsterdam: Utrecht Institute of
Linguistics, Ph.D. dissertation.
Reinhart, Tanya
2006 Interface strategies. Optimal and costly computations. Cambridge,
MA: MIT Press.
Rothman, Jason
2008 Aspectual morphology use in adult L2 Spanish and the com-
peting systems hypothesis: when pedagogical and linguistic rules
conict. Languages in Contrast 8(1). 74106.
Rothman, Jason & Michael Iverson
2008 Poverty-of-the-stimulus and L2 epistemology: Considering L2
knowledge of aspectual phrasal semantics. Language Acquisition:
A Journal of Developmental Linguistics 15(4). 270314.
Rothman, Jason & Roumyana Slabakova
2011 The mind-context divide: on acquisition at the linguistic inter-
faces. Lingua 121(4). 568577.
Generative perspectives 153

Salaberry, M. Rafael
1997 The acquisition of aspect in past tense French among L2 class-
room learners. In A. Melby (Ed.), The 1996 Twenty-Third LACUS
Forum, 521532. Chapel Hill, NC: Linguistic Association of
Canada and the United States.
Salaberry, M. Rafael & Dalila Ayoun
2005 The development of L2 tense-aspect in the Romance languages.
In D. Ayoun & R. Salaberry (Eds.), Tense and aspect in romance
languages: Theoretical and applied perspectives, 134. Amster-
dam: John Benjamins.
Sanz, Montserrat
1999 Aktionsart and transitive phrases. In E. Trevino & J. Lema
(Eds.), Semantic issues in romance syntax, 247261. Amsterdam:
John Benjamins.
Sanz, Montserrat
2000 Events and predication. A new approach to syntactic processing in
English and Spanish. Amsterdam: John Benjamins.
Schell, Karyn
2000 Functional categories and the acquisition of aspect in L2 Spanish:
A longitudinal study. Seattle, WA: University of Washington,
Ph.D. dissertation.
Schwartz, Bonnie & Rex Sprouse
1994 Word order and nominative case in nonnative language acquisi-
tion: A longitudinal study of (L1 Turkish) German interlanguage.
In T. Hoekstra & B. D. Schwartz (Eds.), Language acquisi-
tion studies in generative grammar, 317368. Amsterdam: John
Benjamins.
Schwartz, Bonnie & Rex Sprouse
1996 L2 cognitive states and the full transfer/full access model. Second
Language Research 12. 4072.
Schwartz, Bonnie
1998 The second language instinct. Lingua 106. 133160.
Schwartz, Bonnie & Magda Gubala-Ryzak
1992 Learnability and grammar re-organization in L2A: Against nega-
tive evidence causing the unlearning of verb movement. Second
Language Research 8(1). 138.
Slabakova, Roumyana
2000 L1 transfer revisited: The L2 acquisition of telicity marking in
English by Spanish and Bulgarian native speakers. Linguistics
38. 739770.
Slabakova, Roumyana
2001 Telicity in the second language. Amsterdam: John Benjamins.
Slabakova, Roumyana
2002 Review article. Recent research on the acquisition of aspect: An
embarrassment of riches? Second Language Research 18. 172188.
154 Dalila Ayoun and Jason Rothman

Slabakova, Roumyana
2003 Semantic evidence for functional categories in interlanguage
grammars. Second Language Research 19. 4275.
Slabakova, Roumyana
2006 Is there a critical period for the acquisition of semantics. Second
Language Research 22(3). 302338.
Slabakova, Roumyana
2008 Meaning in the second language. Berlin: Mouton de Gruyter.
Slabakova, Roumyana
2009 L2 fundamentals. Studies in Second Language Acquisition 31(2).
119.
Slabakova, Roumyana & Silvina Montrul
2000 Acquiring semantic-properties of preterite and imperfect tenses
in L2 Spanish. In S. C. Howell, S. A. Fish & T. Keith-Lucas
(Eds.), Proceedings of the 24th Boston University Conference on
Language Development, 534545. Somerville, MA: Cascadilla
Press.
Slabakova, Roumyana & Silvina Montrul
2002 On viewpoint aspect interpretation and its L2 acquisition: A UG
perspective. In R. Salaberry & Y. Shirai (Eds.), The L2 acquisi-
tion of tense-aspect morphology, 363395. Amsterdam: John
Benjamins.
Slabakova, Roumyana & Silvina Montrul
2003 Genericity and aspect in L2 acquisition. Language Acquisition
11. 165196.
Slabakova, Roumyana & Silvina Montrul
2008 Aspectual shifts: Grammatical and pragmatic knowledge in L2
acquisition. In J. M. Liceras, H. Zobl & H. Goodluck, H. (Eds.),
The role of formal features in second language acquisition, 455
483. New York: Lawrence Erlbaum.
Smith, Carlota
1997 The parameter of aspect (2nd ed.). Boston: Kluwer.
Snyder, William
2007 Child language. The parametric approach. Oxford: Oxford Uni-
versity Press.
Solt, Stephanie, Yana Pugach, Elaine Klein, Kent Adams, Iglika Stoyneshka &
Tamara Rose
2004 L2 perception and production of the English regular past: evi-
dence of phonological eects. In L. Brugos, L. Micciulla & C.
E. Smith (Eds.), Proceedings of the 28th annual Boston University
Conference on Language Development, 553564. Somerville, MA:
Cascadilla Press.
Sorace, Antonella
2005 Syntactic optionality at interfaces. In L. Cornips & K. Corrigan
(Eds.), Syntax and variation: Reconciling the biological and the
social, 46111. Amsterdam: John Benjamins.
Generative perspectives 155

Sorace, Antonella
2006 Possible manifestations of shallow processing in advanced second
language speakers. Applied Psycholinguistics 27. 8891.
Sorace, Antonella
2011 Pinning down the concept of interface in bilingualism. Linguis-
tic Approaches to Bilingualism 1(1). 133.
Sorace, Antonella & Ludovica Serratrice
2009 Internal and external interfaces in bilingual language develop-
ment: Beyond structural overlap. International Journal of Bilin-
gualism 13. 116.
Sperber, Dan & Deirdre Wilson
1995 Relevance: Communication and cognition. Oxford: Blackwell.
Stowell, Tim
1993 Syntax of tense. Los Angeles, CA: UCLA, Ph.D. dissertation.
Tenny, Carol
1992 The aspectual interface hypothesis. In I. Sag & A. Szabolcsi
(Eds.), Lexical matters, CSLI lecture Notes, 127. Stanford,
CA: CSLI Publications.
Toribio, A. Jacqueline
2001 On Spanish language decline. In A. H.-J. Do, L. Domnguez &
A. Johansen (Eds.), Proceedings of the 25th Annual Boston Univer-
sity Conference on Language Development, 768779. Somerville,
MA: Cascadilla Press.
Travis, Lisa
2000 Event structure in syntax. In C. Tenny & J. Pustejovsky (Eds.),
Events as grammatical objects: The converging perspectives of
lexical semantics and syntax, 145185. Stanford, CA: CSLI
Publications.
Tsimpli, Ianthi Maria & Maria Dimitrakopoulou
2007 The interpretability hypothesis: Evidence from wh-interrogatives
in second language acquisition. Second Language Research 23.
215242.
Tsimpli, Ianthi Maria & Antonella Sorace
2006 Dierentiating interfaces: L2 performance in syntax-semantics
and syntax-discourse phenomena. In D. Bamman, T. Magnitskaia
& C. Zaller (Eds.), Proceedings of the 30th Annual Boston Univer-
sity Conference on Language Development, BUCLD 30, 653664.
Somerville, MA: Cascadilla Press.
Valian, Virginia
2009 Innate syntax still the best hypothesis. Plenary address given
at the Boston University Conference on Language Development
34.
Verkuyl, Hans
1999 Aspectual issues: Studies on time and quantity (Vol. 12). Stanford,
CA: CSLI Publications.
156 Dalila Ayoun and Jason Rothman

White, Lydia
1991 Adverb placement in second language acquisition: Some eects
of positive and negative evidence in the classroom. Second Lan-
guage Research 7. 133161.
White, Lydia
2011 Second language acquisition at the interfaces. Lingua 121(4).
577591.
Yang, Suying & Yue Yuan Huang
2004 The impact of the absence of grammatical tense in L1 on the
acquisition of the tense-aspect system in L2. International Review
of Applied Linguistics 42. 4970.
Part II. Research design and methodology in L2
studies of tense and aspect
Chapter 5
Research design: A two-way predicational system is
better than a four-way approach1

Paz Gonzalez

1. Introduction

This chapter presents the Predication-eect Hypothesis, an approach to


the issue of interpreting the particular distribution of the past aspectual
morphology in interlanguage. More specically, it argues that a two-way
predicational system is more appropriate for the learning of Spanish as
a second language (L2) than a four-way approach. On the theoretical
side, a particular description of the Spanish aspectual system is given in
Section 2. Following Verkuyl (1997) and Krifka (1998), a distinction is
made between predicational and grammatical aspect. Section 3 discusses
the intuitions of native speakers about the preferences of verb forms accord-
ing to varying aspectual sentence information. It is shown that there is no
correlation between predicational aspect and grammatical aspect according
to native speakers. Section 4 analyzes the distribution of the two Spanish
past tense forms in the learners interlanguage, the Aspect Hypothesis
(AH) and the Default Past Tense Hypothesis (DPTH) are described, and
an alternative approach to the L2 acquisition of grammatical aspect is
presented; namely, the Predication-eect Hypothesis, which takes as its
theoretical starting point the cross-linguistic analysis of the rst section.
The chapter provides further evidence of what a number of previous studies
on the matter propose: L2 research in the acquisition of grammatical aspect
conrms the juxtaposition of grammatical aspect and inherent aspect. More
specically, evidence shows that the distinction terminative-durative of the

1. I am very thankful to Henk Verkuyl for his invaluable help with the clarica-
tion of the used terminology, and to Huub van den Berg for his help with the
analysis of the ndings of the two empirical studies presented here. I would
also like to thank Rafael Salaberry and Llorenc Comajoan and the external
reviewer for their comments on earlier versions of this chapter. However,
I am the only one responsible for any inaccuracies or shortcomings of this
chapter.
160 Paz Gonzalez

Predication-eect Hypothesis is eective because it maps not only into the


perfective-imperfective contrast in number, but also in meaning, as will be
demonstrated in Section 4.3.

2. Aspect

2.1. Two types of aspect


The notion of aspect has been used to distinguish aspectual information
from temporal information. This distinction, however, only captures part
of the complexity, as there are two aspectual phenomena to account for:
lexical and grammatical aspect. On the one hand, following well-known
proposals (Verkuyl 1997; Borik 2002; and Krifka 1998, among others),
a distinction will be made between predicational aspect (also known as
inherent aspect) and grammatical aspect (also known as viewpoint aspect).
On the other hand, although grammatical aspect and predicational aspect
are two distinct linguistic categories, it is almost impossible to entirely
exclude one from the discussion of the other.

2.2. Predicational aspect


Predicational aspect is essentially the modern successor to the notion of
Aktionsart as developed in the study of Slavic aspect. Aktionsart is a
property of verbal predicates, and it concerns the internal temporal con-
stituency of a (type of ) situation denoted by a given predicate (Bache
1985). The problem with the notion of Aktionsart is that it has been mostly
used to describe a lexical property of the verb, instead of a property of the
verb and its arguments. Nowadays, it has become clear that it is the verb
and its arguments that provide the aspectual information about the aspec-
tual nature of a predication (Verkuyl 1997, Krifka 1998), but the old
distinction between Aktionsart and grammatical aspect is still under dis-
cussion as the distinction between predicational aspect and grammatical
aspect.
In order to understand the principle behind the Aspect Hypothesis
(AH, Andersen 1989, 1991), a short description of Vendlers well-known
classication of verb classes is necessary. The verb classes in Vendler
(1957) have become a well-established classication of verb meanings
according to their inherent aspectual information. In this view, verbs can
be states, activities, accomplishments, or achievements. State verbs do no
encode either phases or endpoints (e.g., to want); activity verbs encode
situations consisting of phases but no endpoints (e.g., to walk); accom-
plishment verbs have phases and an endpoint (e.g., to walk a mile); and
Two-way predicational system 161

achievements have an endpoint which is instantaneous, and therefore


without phases (e.g., to discover). Predicational aspect is compositionally
formed from the interaction between the verb and its arguments (as shown
below), whereas grammatical aspect is said to express the perspective on
the eventuality described, and it is encoded in the verbal morphology of
many languages (e.g., in Spanish).
By distinguishing grammatical aspect and predicational aspect, one can
posit a tense-aspect triangle: tense, grammatical aspect, and predicational
aspect. As shown below, each of the three concepts makes its own contri-
bution to the expression of temporality within the sentence. For instance,
the representation of (1a) given in (1b) expresses that the basic, rst-level
aspectual information comes from the tenseless predication; that is, [Noa
drink a lemonade].
(1) a. Noa drank a lemonade.
b. past [Noa drink a lemonade]
In this sense, predicational aspect diers crucially from the temporality
contributed by tense. The notion of completion, which is intuitively asso-
ciated with (1a), is already there before tense is taken into account,
because the tenseless sentence expresses termination due to the choice of
the verb and the arguments. The tenseless predication in (1b) is termina-
tive (cf. Verkuyl 1972; Krifka 1989, who uses the term telic in this connec-
tion, among many others), because the information expressed by com-
bining a lemonade and drink into a verb phrase such as drink a
lemonade, and by combining this verb phrase and Noa into Noa
drink a lemonade expresses something that presents itself as a unit, as
something that can be discerned as complete when compared to the rest
of the domain of discourse.
The terms terminative/durative are preferred to the terms telic/atelic,
because the origin of telic is the Greek word telos, which means target,
and may be too closely associated with lexical meaning (Verkuyl 1972). A
verb phrase like write a letter may contain something that can be called
target, but this notion does not really work for phrases such as spilling
three glasses, begin a walk, or pass away. In these examples, it is far-
fetched, impossible, or forced to speak of a target. To underline that the
level of aspectual representation, corresponding to the tenseless part of a
sentence, is in fact a predication (the verb and its arguments), the aspectual
information collected at that level is named predicational aspect (Vet 1994).2

2. Some authors use the term eventuality description to refer to the same notion
(De Swart 1998).
162 Paz Gonzalez

Predicational aspect crucially concerns the information about the rela-


tionship between the verb and its arguments. If the speaker of example (1)
had used as the direct object the mass noun lemonade, the subject
nobody or the verb want, the aspectual value of the predication would
have changed (as shown in sentences (2c)(2e), which are all durative (or
atelic)).
(2) a. Noa drank lemonade.
b. past [Noa drink lemonade]
c. Nobody drank lemonade.
d. Noa wanted a lemonade.
e. Nobody wanted lemonade.
In separating the tense information past from the tenseless predication
as in (2b), it is assumed that the aspectual value of a predication deter-
mined at the tenseless level of representation remains intact. Therefore,
its value is taken as independent of any specic tense information. The
tense information has an eect on the predication, but it will not change
its aspectual value. This picture seems to hold for Germanic languages
like Dutch and English (Verkuyl 1993), but in other languages the con-
nection between tense and aspect is more intimate than in Germanic lan-
guages. For instance, Spanish (like other Romance languages) has two
inectional forms for the past tense sentence (1a); that is, drank can be
translated as bebio or beba (examples (3) and (4)).

(3) Jari bebio una limonada.


past perfective [Jari drink a lemonade]
Jari drank a lemonade.
(4) Jari beba una limonada.
past imperfective [Jari drink a lemonade]
Jari drank/was drinking a lemonade.

The dierence between the two past forms can only be understood by
assuming that certain tense forms express aspectual information. This
aspectual information is called grammatical aspect and is encoded between
the tenseless level where predicational aspect is determined and the
past temporal level. In particular, grammatical aspect characterizes the
domain in which the eventuality of the predication takes place either
as perfective as in (3), or imperfective as in (4). The notion of domain
Two-way predicational system 163

is semantic in the sense that sentences are interpreted in a domain of


interpretation. A temporal domain concerns the temporal structure and
temporal entities such as events. For instance, for French, there are several
proposals in which the formula in (5) has been used to account for the
dierent roles in the tense-aspect triangle (Vet 1994, de Swart 1998).
(5) tense [aspect [predication]]
The general idea connected with this formula is that the lowest level of the
eventuality description is taken as expressing predicational aspect, that the
next step is provided by aspectual operators expressing perspective (the
perfective-imperfective distinction), and that the nal step is the applica-
tion of the tense operator.

2.3. The Plus Principle


Using an event semantic framework, Verkuyl (1972, 1993) showed that in
Germanic languages what has been called lexical aspect is in fact composi-
tionally formed with the help of two aspectual features: [add-to] and [sqa].
To account for the non-stative nature of verbs like write, eat, hit,
etc., as opposed to stative verbs like hate, love, and want, a verbal
semantic feature called [add-to] is assumed as the contribution of the verb
to the aspectuality at a tenseless level. This feature expresses the dynamicity
of the verb. For dynamic verbs, such as write, eat, etc., the value of
the feature is positive; whereas for stative verbs, such as hate, love
etc., the value of the feature is negative. The contribution of the NP inter-
nal argument to aspectual information can be accounted for in terms of
an NP-feature [sqa], where sqa stands for Specied Quantity of A, A
being the denotation of the Noun of the NP. This feature expresses the
quantication and delimitation of the arguments. Thus, delimited NPs like
an apple, three beers, a piece of bread, etc. are labelled [sqa],
whereas NPs like apples, beer, bread, etc. are [sqa].
Predicational aspect thus involves a compositional amalgamation of
lexical semantic information given by the verb and delimiting information
given by the arguments. Only a combination of a [add-to]-verb with
[sqa]-arguments yields compositionally well-formed terminative aspect.
All other cases are durative. This combination describes a bipartition of
predicational values: terminative versus durative. This is what Verkuyl
(1999) called the Plus Principle, since it refers to the requirement that all
aspectual features ([addto] and [sqa]) involved in a terminative predication
are plus values. As seen in the tenseless examples in (6), terminative predi-
164 Paz Gonzalez

cations are the marked case, as opposed to the durative predications, which
are unmarked and thus more common in language production.
Aspectual value
(6) a. Teo write a letter
[sqa] [addto] [sqa] ! terminative
b. Teo write letters
[sqa] [addto] [sqa] ! durative
c. Children write a letter
[sqa] [addto] [sqa] ! durative3
d. Teo expect a letter 4
[sqa] [addto] [sqa] ! durative
Gonzalez and Verkuyl (2003) showed that the Plus Principle is also a valid
tool to characterize Spanish predicational aspect and that both operators
receive the same values regardless of the fact a sentence is produced in
Spanish or in Dutch.
In sum, the Plus Principle turns out to be a useful tool for drawing
attention to the contribution of the predication as a whole to aspectual
information, because it expresses the requirement that all atomic entities
involved in the aspectual composition at this level have to have a positive
value in order to derive a terminative predication.

2.4. Grammatical aspect


Grammatical aspect concerns the representation of the completion of the
temporal domain in which an eventuality is hosted in the past. If the
domain is characterized as bounded, the perfective aspect results, whereas
if the domain can be characterized as not necessarily complete or bounded,
then the imperfective is called for (Gonzalez 2003). Grammatical aspect is
not a strict temporal notion, as it is not deictic. It merely modies in some

3. In a sentence such as (6c), the denotation of the external argument is [-sqa],


which also yields a compositionally formed durative aspect. Examples involv-
ing [-sqa] external arguments are not analysed here. This is a result or con-
sequence of the fact that there is an asymmetry between the internal and the
external argument; the verb and its internal arguments have closer ties as far
as aspectual structure is concerned.
4. In a sentence such as in 6d, the denotation of the verb is [-add to], and there-
fore it yields a compositionally formed durative aspect.
Two-way predicational system 165

way a domain in the past in which the eventuality is located as to its being
bounded or not, so that information can be obtained about the way which
the eventuality is presented.5
Grammatical aspect in Spanish is morphologically marked in a systematic
way by means of specic morphemes: it is visible in the alternation of aspects
in the past, since the morphological encoding of grammatical aspect is
only obligatory in the past tense (imperfective-perfective). The perfective-
imperfective distinction in Spanish is inectional, which means that every
verb has both a perfective and an imperfective past form. Spanish has two
simple past tense forms, both of them introducing aspectual as well as
temporal information together with the verb form itself. Table 1 provides
examples of terminative and durative predications and shows how they
interact with the two Spanish simple past tenses.
The terminology used in Table 1 deserves some clarication. In 1931,
the Real Academia Espanola (ocial institution responsible for regulating
the Spanish Language) reected on the terminological problem of consider-
ing the perfective forms as perfect forms (see Rojo 1988). Spanish distin-
guishes the perfect verb forms from the non-perfect ones, and it has two
parallel complete series of tenses to express them: the perfect and the
imperfect. The correspondence could not be more exact: every simple
tense or imperfect form corresponds in a one-to-one way to a compound
tense or perfect form. To treat the past perfective as a past perfect form
would completely break the system, because it would be a simple form
but yet perfect in meaning. Because of this, the Academia resorted to
considering the past perfective a preterito indenido basing this nomen-
clature on the fact that it expresses sometimes the eventuality as incipient
and sometimes as nished. However, to call it preterite was confusing,
as it just means past. To solve the disputed point of whether the perfective
forms should be treated as perfect or not, the following terminological
proposal is suggested: the preterito indenido or preterito perfecto
simple is here called preterito perfectivo; and the preterito imperfecto
is in the same manner here called preterito imperfectivo.

5. However, aspect is also been dened in deictic terms by Klein (1994) and
Doiz-Bienzobas (1995). According to Klein, any given utterance is composed
of two components, one nite and one non-nite. The non-nite component is
what carries lexical content, the nite component is the Topic Time, which is
the time for which the claim about the situation has been made.
166 Paz Gonzalez

Table 1. Spanish simple pasts


Preterito Imperfectivo Preterito Perfectivo
(Past Imperfective) (Past Perfective)

Durative (7a) Luca cantaba. (7b) Luca canto.


Luca sang imperfective Luca sang perfective
Terminative (7c) Luca cantaba una cancion. (7d) Luca canto una cancion.
Luca sang imperfective a song Luca sang perfective a song

Put very simply, the imperfective forms in examples (7a) and (7c) in
Table 1 imply that the period of time or domain hosting the eventuality
described by the predication is of an incomplete nature (it does not have
a right boundary), whereas the perfective forms in (7b) and (7d) imply
that the domain hosting the eventuality described by the predication is
closed o. This causes the process of Luca singing in example (7b) to be
presented as bounded. In this sense, there are two sorts of completion in
(7d). One is expressed by the terminative nature of the predication, whereas
the other is the completion of the domain in which the eventuality is hosted.
The dierence between imperfective and perfective taken in this sense is not
stated in terms of the temporal location of the eventuality itself, which is, in
both cases, the past tense, nor in terms of the aspectual properties of the
eventuality. What matters here is the representation of the temporal domain
where the eventuality takes place.
The question of whether the dierences between perfective and imper-
fective forms are temporal or aspectual has attracted much discussion in
Spanish linguistics. The idea that the dierence is temporal was rst pro-
posed by Bello (1847), who dened the imperfective form as a co-preterite,
treating it as a present in the past. Therefore, the imperfective form, in
Bellos theory, had its own temporal characteristics. However, the aspec-
tual description of the two simple past tenses in Spanish can be better
described as, in accordance with Garca Fernandez (1998), a past imper-
fective (combination of past tense and imperfective aspect) and a past
perfective (combination of past tense an perfective aspect) (Table 1).

2.5. Aspectual incongruence: Incompatibilities?


In general terms, both aspectual levels (grammatical and predicational)
interact without inuencing each other. The following sentences illustrate
that it is possible to obtain a sentence with perfective marking and a termi-
Two-way predicational system 167

native predication (8), a sentence with perfective marking and a durative


predication (9), a sentence with imperfective marking and a terminative
predication (10), and a sentence with imperfective marking and a durative
predication (11).
(8) Ayer le dos artculos en media hora.
Yesterday read 1stp past perfective two articles in half hour
Yesterday I read two articles in half an hour.
(9) Ayer com porqueras todo el da.
Yesterday eat 1stp past perfective junk food the whole day
Yesterday I ate junk food the whole day.
(10) Cada manana Diego compraba el periodico.
Every morning Diego buy 3rdp past imperfective the
newspaper
Every morning Diego bought the newspaper.
(11) Cada manana Pablo compraba bizcochos.
Every morning Pablo buy 3rdp past imperfective cakes
Every morning Pablo bought cakes.

All the sentences in (8)(11) are grammatical, which indicates that the two
levels of aspectual representation interact without transforming internal
values in those languages where the distinction perfective-imperfective
is formally present. Hence, aspect needs to deal simultaneously with the
characterization of the predicational aspect and with the question of
whether the period of time in which the predication is situated is left as
open or closed o. This task of dealing with two aspectual levels may
lead to aspectual combinations within sentences, which may look like
incompatibilities, as is the case in sentences (9) and (10), in which the
senses of completion of the sentence and the predication it contains are
reversed.
The possible incompatibilities with imperfective verbal forms are reviewed
rst (see Chapter 1). The combination of imperfective marking and durative
predication, as the more natural combination of two incomplete levels,
does not present any complications. That is, a sentence that features an
imperfective form, implying that the temporal domain hosting the described
eventuality is not complete, does not conict with the fact that the eventu-
ality is durative, because both aspectual levels are dened as non-complete
168 Paz Gonzalez

(sentence (11)). However, sentences with an imperfective form, but with


compositionally formed terminative aspect can create a problem, as in
sentence (10).
A quick overview of the dierent readings of the imperfective is needed
to understand possible incompatibilities. Consider the following sentences:
(12) Elin cantaba una cancion cada domingo.
Elin sang a song every Sunday.
Habitual reading
(13) Ayer Elin cantaba una cancion porque estaba contenta.
Yesterday Elin sang a song because she was happy.
Episodic reading
(14) Elin cantaba una cancion cuando me la encontre.
Elin was singing a song when I found her.
Progressive reading
A range of completion values at the predicational level can be traced
between the three imperfective readings. In sentence (12), only the tense-
less predication in Elin cantaba una cancion, with habitual reading,6 can
be understood as terminative, since the eventuality of singing a song
occurs several times but each time the eventuality is complete, it becomes,
so to speak, an accumulation of terminative eventualities. Sentence (13) is
not incompatible with whether in reality Elin nished singing that one song.
However, this information is not given by the sentence itself. Finally,
sentence (14) cannot express a terminative reading due to the presence of
the when-clause and hence the sentence becomes durative: the imperfective
form of the verb overrules and modies the tenseless terminative meaning
given by the internal structure of the verb and its arguments by preventing
the whole predication from being actualised in real time. To be able to still
consider this sentence with a progressive meaning as containing a termina-
tive predication, the progressive could be understood as making a commit-
ment to the process that is part of the event. In this way, the predication
keeps its tenseless terminative value; but the focus is on its progress in real
time, not on its termination or completion.

6. There is another reading of the imperfective in this sentence: if the verb is


punctual, it is possible that it expresses that the event was about to happen
but never happened. This is the imperfecto de conato (interrupted) (Fernandez
and de Miguel 1999).
Two-way predicational system 169

Imperfectivity, whether it is traditionally a property of a tense system


or an aspect system, can be given a uniform analysis based on the notion
of an incomplete temporal domain. The range of available interpretations
is the same in the case of terminative and durative sentences in the imper-
fective. Imperfectivity is stronger in the sense that it is able to override
the terminative-durative distinction. A terminative predication together
with a progressive reading is the only case in which the two aspectual
levels may not work independently (example 14). However, this problem
is solved if the reading of progressivity is understood as focussing on the
progress of the event rather than on its being presented as complete.
The present approach suggests that an adequate analysis of the imper-
fective value of grammatical aspect cannot be given in terms of an even-
tuality description, whether it concerns the part versus whole relation (as in
the progressive reading of the imperfective) or its durative versus termina-
tive character. Such an analysis should be adopted in terms of a character-
ization of the temporal domain, hosting an eventuality. In other words, if
the temporal domain in which an eventuality is hosted in the past is charac-
terized as not necessarily complete or bounded, then the imperfective is
called for.
The perfective form allows for the other type of aspectual incompatibil-
ities. The combination of perfective and terminative aspectual values, as
the more natural combination of two complete levels, does not create
a problem (as shown in sentence (8)). The other combination with the
perfective, that is, a perfective verbal form and durative predication, in
sentence (9), repeated here for convenience as (15), may be considered
problematic for the idea of independency of aspectual levels.

(15) Ayer com porqueras todo el da.


Yesterday eat 1stp past perfective junk food whole the day
Yesterday I ate junk food the whole day.
One could say that intuition would direct the predication [I eat junk food]
towards expressing an endpoint in this sentence, because of the perfective
marker (in com ), although from an inherent aspect point of view, it
should be characterized as a durative predication, as junk food is [-sqa].
However, the eating of junk food by someone is known to end, not
because it becomes a terminative predication, but because the period of
time when the durative predication takes place is bounded (requirement
of the perfective form). Sentence (15) does, therefore, not create a problem
for the idea of independence of aspectual levels either.
170 Paz Gonzalez

Another possible incompatibility may be found in those sentences with


stative verbs (included in durative predications):
(16) Olivia estuvo enferma.
Olivia be 3rdp past perfective sick
Olivia was sick.
It can be armed in (16) that [estar enferma] (be sick) is a durative
predication and that the domain accommodating the state of being sick is
complete and bounded, as in cell (7b) of Table 1. There is, again, no real
incompatibility of aspectual meanings. Thus, it can be concluded that
none of the perfective durative combinations is a problem for a two-level
aspectual description analysis.7
Nevertheless, there are ungrammatical cases with some types of predi-
cations and the perfective form, which turn out to be pragmatic incom-
patibilities.8 Permanent predications such as that in (17a) express qualities
that cannot be subjected to variation (cf. Garca Fernandez 1999).9 In
general, they cannot be modied by adverbial complements (17b), nor by
temporal subordinated sentences as in (17c), nor can they, in languages
with a perfective-imperfective distinction, appear with the perfective (17d).
However, they are grammatical with the imperfective (17e).
(17) a. [Julia to be from Barcelona]
b. *Julia was from Barcelona since a couple of months.
c. *Julia was from Barcelona since she was born.
d. *Julia fue de Barcelona.
Julia be 3rdp past perfective from Barcelona
Julia was from Barcelona.
e. Julia era de Barcelona.
Julia be 3rdp past imperfective from Barcelona
Julia was from Barcelona.

7. For another theoretical standpoint on the interaction of the two aspectual levels,
see de Swart (1998). De Swart describes what she calls aspectual shifts and
coercion, proposing the idea that perfectives only describe events, whereas
imperfectives only describe states and processes.
8. In this context, a connection can be made again to Kleins Topic Time (1986).
According to Klein, the relationship between Topic Time and Time of the
Situation (lexical aspect) represents imperfective aspect.
9. Examples (17a), (17b), and (17c) are in English to show that the quality of
these predications is not language specic.
Two-way predicational system 171

This semantic incompatibility emerges also with the present perfect, as it is


shown in (17f ).
(17) f. *Julia ha sido de Barcelona.
Julia has been from Barcelona.
Sentence (17f ) exemplies that the incompatibility is not aspectual in
nature, as the predication is not only incompatible with the perfective
aspectual marker, but also with the perfect temporal marker.
In sum, the aspectual description of the Spanish language given in this
section assumes a separation of two levels of aspectual information, each
of them contributing its own particular interpretation of the semantic
notion completion: a) an interpretation at the predicational aspect level
(terminative versus durative predications), and b) an interpretation at the
grammatical aspect level (perfective versus imperfective forms)

3. Use of grammatical aspect in L1

The purpose of this section is to contribute to a better understanding of


grammatical aspect as used by Spanish speakers. As argued in the previous
section, there is a separation between predicational aspect and grammatical
aspect, and each of the two levels can be characterized as expressing infor-
mation about discrete or continuous temporal units: at the predicational
level in the form of the bounded-unbounded opposition, and at the gram-
matical level in the form of the complete-incomplete opposition. The inde-
pendency of the levels may be hard to recognize because the same type of
features seems to characterize both. If the expectation of non-interference
between levels is fullled, then speakers of Spanish will claim that any past
sentence, no matter how the predicational level is being characterized (as
bounded or unbounded), can accept both grammatical aspects.
Gonzalez (2003) empirically showed that, despite the inuence of the
lower lexical level, the nal choice of grammatical aspect by native speakers
of Spanish is made according to the contextual aspectual information,
provided by adjuncts or sentential information. In other words, the predi-
cational aspect that the sentence contains is not the only one inuencing
the grammatical aspect choice made by the native speakers. In the study,
the informants were confronted with pairs of sentences whose only dier-
ence was the form of the verb. A questionnaire for native speakers called
for acceptance or rejection of 15 pairs of sentences with perfective and
imperfective markings, and the results showed that native speakers intuitions
172 Paz Gonzalez

varied, even though there were some patterns that stood out. Sentences
including extra information pointing to some sort of repetitive, habitual,
or progressive meaning were logically accepted with the imperfective form.
For instance, sentences (18) and (19) only dier aspectually in the form of
the verb.
(18) Los barcos llegaban cada media hora.
The ships arrive past imperfective every half hour
The ships arrived every half an hour.
(19) Los trenes llegaron cada media hora.
The trains arrive past perfective every half hour
The trains arrived every half an hour.
These sentences have extra aspectual information that points to repetition
(every half an hour). Thus, sentence (18) with the imperfective form
was judged grammatical by the majority of the informants. On the other
hand, nearly half of the informants found (19) also grammatical. The
native informants seemed to have created a situation where the sentence
is also acceptable with the perfective form, and this is achieved by (im-
plicitly) adding another layer of aspectual information, in this case some-
thing like (20).
(20) (Ayer) los trenes llegaron cada media hora.
Yesterday the trains arrive past perfective every half hour
(Yesterday) the trains arrived every half an hour.
Yesterday limits the time, and it provides a beginning and an end to the
event, making the temporal domain complete. Therefore, in this case, the
perfective is also an acceptable form. This may lead us to think that aspec-
tual choice is not only a predicational phenomenon, but also a sentential,
even contextual, discourse phenomenon (Chapter 6; Fleischmann 1985;
Silva Corvalan 1983).
Sentences including extra information pointing to some sort of limita-
tion of time were logically mostly accepted with the perfective form. Sen-
tences (21) and (22), again, only dier in the form of the verb.
(21) Ayer por la manana Ana compro el periodico.
Yesterday in the morning Ana buy past perfective the
newspaper
Yesterday morning, Ana bought the newspaper.
Two-way predicational system 173

(22) Ayer por la manana Ana compraba una revista.


Yesterday in the morning Ana buy.past.imperfective a magazine
Yesterday morning, Ana bought a magazine.
These sentences have extra information regarding a specic point in time,
with its own beginning and end (yesterday morning). Therefore, the
sentence with the perfective form was judged grammatical by nearly all
of the informants. However, a small percentage of the informants also
considered sentence (22) acceptable. This is probably due to the fact that
the imperfective simple past in Spanish may convey an ongoing or repeti-
tive sense to the meaning of the sentence. That is, on the one hand, the
imperfective form expressing ongoingness has the unusual property of
always having to be in relation to another tense for the sentence where it
occurs to fully function. On the other hand, the habitual/repetitive mean-
ing of the imperfective form can function on its own, as long as the time
domain is clearly specied, as in (22).
Thus, sentence (22) can never express habituality, since it has been
already xed to only one time in the past. There is only one time domain,
which means that a repeated temporal domain is out of the question.10
Therefore, the only way this sentence can be used with an imperfective
marker is by understanding it as representing an episodic or progressive
situation. And this can only be fullled if some context is added (in this
case, if another tense is added for the sentence to function with imper-
fective marking). The informants who accepted (22) probably invented a
whole extra situation such as in (23), where the imperfective can function
adequately.
(23) Ayer por la manana Tom compraba una revista cuando se encontro
con Luca.
Yesterday during the morning Tom buy past imperfective a
magazine when him nd past perfective with Luca
Yesterday morning Tom was buying a magazine when he met Luca.
Doiz-Bienzobas (1995) has another explanation for the use of the im-
perfective with telic events, which can also explain why the informants

10. The repetition is not of the eventuality, but of the hosting domain. This is why
Ayer por la manana Ramon compraba tres revistas (Yesterday morning Ramon
bought imperfective three magazines) is ruled out without the help of
extra contextual information.
174 Paz Gonzalez

accepted (22). In her view, the imperfective in this context is used to desig-
nate the concluding event of a series of events, to express emotions and
make judgements, and to create a temporal setting.
For sentences with no additional aspectual information, both options
(perfective and imperfective) are equally acceptable. An example of this
type of pair of sentences is given in (24) and (25).
(24) Por la noche contabamos cuentos.
During the night tell 1stp pl past imperfective stories
At night time we told stories.
(25) Por la noche contamos cuentos.
During the night tell 1stp pl past perfective stories
At night time we told stories.
If aspect were an intra-sentential phenomenon, it would be expected that
one of both sentences would be more acceptable than the other (according
to the predicational aspect of the sentence). This is not the case. Both
sentences were equally accepted because the predication in itself does not
give enough information as to which aspectual form the verbs need to
take. Thus, it was easy for the informants to imagine the two required
contexts, one for the perfective form and another one for the imperfective
form.11
In sum, from this section, two main conclusions can be reached. First,
the predicational level oers no aspectual information that may interfere
with the choice of past tense for L1 speakers, and, therefore, predicational
aspect and grammatical aspect do not interfere with each other. Second,
sentential and contextual aspectual information is crucial to properly under-
stand how grammatical aspect is used by native speakers (see Chapter 6).

4. The acquisition of grammatical aspect in L2 Spanish

Several studies have investigated the challenging issue of interpreting the


particular distribution of past aspectual morphology in interlanguage

11. A reviewer pointed out that por la noche is ambiguous, as it can mean
every night or during that period of time and therefore both interpreta-
tions are possible.
Two-way predicational system 175

(e.g., Li and Shirai 2000; Bardovi-Harlig 2000; and Salaberry 2008). More
specically, Spanish has aspectual morphological verbal markers, which
do not have a counterpart in languages from the Germanic family. This
fact may have contributed to the interest in the study of the acquisition
of tense and aspect by L2 learners of Spanish who have a Germanic lan-
guages as their L1, as the latter face one of the most signicant challenges:
the notorious distinction between the two simple past tense forms and the
mismatch in aspectual representations between their L1 and L2 (Salaberry
2008).
The goal of the present section is to characterize the initial stage(s) of
the acquisition of the L2 Spanish aspectual system. Two theoretical views
are discussed. First, the aspectual division on verb classes (Vendler 1957,
described in Section 2.2) is applied to a description of the distribution of
interlanguage past tense morphology of beginning learners. Second, a
two-way predicational-aspectual analysis, that is, a division into termina-
tive and durative predications, is considered as a possible tool for an
optimal description of what happens in the interlanguage of Dutch L2
learners of Spanish. Moreover, other pragmatic-discourse factors will also
be taken into account as they may add relevant clues to the characterization
of the distribution of morphological markers in L2 Spanish. Following
Salaberry (2008), several characterizations of the initial state of the acqui-
sition of temporal systems in a L2 are discussed. Although these character-
izations do not follow the same theoretical framework, reviewing them
jointly can be important in building an adequate model of tense-aspect
acquisition. The theoretical perspectives to be summarized are the follow-
ing: Lexical Aspect Hypothesis (Andersen 1991), the Default Past Tense
Hypothesis (Salaberry 2000), and the Predication-eect Hypothesis (see
Section 2).

4.1. The Aspect Hypothesis (AH)


A number of acquisition studies have tackled the inuence of aspectual
classes (as dened by Vendler 1957) on the interlanguage tense-aspect mor-
phology of L2 learners of Spanish. The main contribution to this line of
research is the Aspect Hypothesis (or Lexical Aspect Hypothesis), which
states that in beginning stages of language acquisition, only inherent
aspectual distinctions, neither tense nor grammatical aspect, are encoded
by verbal morphology. The hypothesis was rst posited by Andersen (1986),
who studied the acquisition of Spanish as a second language by two English-
speaking children using Vendler (1957)s four-way division: states, activities,
176 Paz Gonzalez

accomplishments, and achievements. He presented a developmental sequence


for encoding tense and aspect with past inections containing 9 stages
(Table 2).

Table 2. Stages in the acquisition of perfective/imperfective forms in L2 Spanish.


Stage States Activities Accomplishments Achievements
1
2/3 p
4 i p
5 i i p p
6 i i ip p
7 i ip ip p
8 i ip ip ip
9 ip ip ip ip

Note: A hyphen indicates that no past morphology is used, that is, present or base
forms are used; i: imperfective forms; p: perfective forms; ip: imperfective and perfec-
tive forms.

The AH makes predictions about two acquisition features: rstly, the


distribution of verbal morphology, and secondly, the development of
aspect morphology. Both predictions behave according to inherent aspec-
tual meanings. According to the AH,12 the perfective forms are used rst
with achievements, then with accomplishments, spreading later to activities
and, nally, states. Imperfective forms appear later than perfective forms,
and they appear rst with states, spreading later to activities, accomplish-
ments and, nally, achievements (Table 2). According to Andersen (1986),
in the rst stages of acquisition of verbal past morphology, the perfective
form is only found with achievements, and the imperfective form only
with states. Later on, the perfective is also found with those verbs referring
to accomplishments and the imperfective with activities. That is, telic
verbs (achievements and accomplishments) are used only with perfective
forms and atelic verbs (states and activities) only with imperfective forms.

12. The Aspect Hypothesis makes two more claims that are not considered rele-
vant for the purposes of this description (in languages with progressive aspect,
it emerges in activities, then accomplishments and achievements; progressive
markings are not incorrectly overextended to statives).
Two-way predicational system 177

The rst atelic verbs that are found with perfective are activities; the rst
telic verbs found with the imperfective are accomplishments. Finally, both
forms are found with the four types of predicates (Table 2). When verbal
morphology emerges in the learners interlanguage, it appears as if it is in
complementary distribution to the aspectual class of the verb.
A number of studies have provided evidence in favor of the AH
(Hasbun 1995, Bergstrom 1995, Cadierno 2000; for a review of studies
see Bardovi-Harlig 2000 and Salaberry 2008). However, some important
weaknesses qualify its claims. As Salaberry (2008: 104) pointed out, the
LAH does not oer a theoretical account for what happens after lexical
aspect alone can no longer determine the choice of perfective or imperfec-
tive. Moreover, it also contains methodological limitations that prevent
adequate generalizations (see Chapters 8 and 11). However, despite these
weaknesses, as Shirai and Kurono (1998: 284) argued, studies on the
acquisition of verb morphology that systematically investigated the rela-
tionship between verb morphology and inherent aspect have consistently
indicated that (. . .) adults acquiring an L2 are strongly inuenced by the
inherent aspect of the verb to which the morphology is attached.

4.2. The Default Past Tense Hypothesis (DPTH)


Salaberry (1999, 2000, 2008) described the particular distribution of the
past aspectual morphology at various stages of development. Salaberry
(1999) studied the distribution and development of past tense markers in
three sets of data: movie narratives, cloze and editing tasks, and speak-
aloud protocols. The languages involved in his study were English (L1)
and Spanish (L2). At the very beginning (stage 1, 2nd semester students
following Spanish academic instruction), the learners in Salaberrys study
relied on a single marker of past tense verbal morphology: the perfective
(see also Wiberg, 1996 for similar ndings with L2 Italian data with ado-
lescents). This nding was explained as being an under-application of the
rule of past tense formation in Spanish. By not using both forms, which are
necessary to have a complete system in Spanish, according to Salaberry,
the learners were under-applying the rules of past tense use in Spanish. In
sum, Salaberry argued that the learners of his study used the perfective as
a default marker of past tense. In stage 2, students may realize that they
need to use two dierent markers of past tense, and they use them accord-
ing to the inherent lexical semantics associated with verb types (lexical
aspect). In stage 3, students are more procient, and they unexpectedly
178 Paz Gonzalez

overgeneralize the use of past tense markers associated with specic lexical
aspectual classes (the prototypical cases). Salaberrys explanation for this
unexpected overgeneralization was that the degree of association between
lexical aspectual classes and inectional morphology increases with time
to the point that it overgeneralizes to a level that surpasses the target
language distribution. Finally, in stage 4, students were able to focus on
language-specic discursive-pragmatic factors that provided them with
information about when to use both markings of verbal morphology.
There are two main dierences between Salaberrys stages of acquisi-
tion of L2 tense-aspect morphology and those presented by Andersens
original AH. First, Salaberrys rst stage, which presents the perfective as
the default past tense in beginning stages of interlanguage, was not pre-
dicted by the AH. Second, studies supporting the AH have not found data
where the more procient students over-generalized the prototypical asso-
ciations of verbal morphology and aspectual category (Salaberrys over-
generalization stage). Finally, a few studies have disputed the claims of
the DPTH arguing that it is not clear that beginners use perfective mor-
phology across all aspectual categories (Comajoan 2009) and that beginners
do not only use perfective morphology (Bergstrom 1995; Camps 2002; and
Comajoan 2001).

4.3. The Predication-eect Hypothesis


A two-way predicational aspect division, based on the information depicted
by the verbal predication, portrays the use of the interlanguage past tense
morphology from a dierent perspective than a four-way inherent verbal
semantic distinction. As argued in Section 2.2, predicational aspect crucially
concerns the information about the relationship between the verb and
its arguments. Durativity and terminativity are the two aspectual values
characterizing predicational aspect and recognized at the tenseless level of
the bare eventuality. The four-way distinction (states, activities, accom-
plishments, and achievements) is based on allegedly verbal meanings,
whereas the two-way distinction (the two-way predicational aspect division)
is based on predicational meanings. The distribution of aspectual classes in
Andersen (1986) was primarily based on verb meanings, and Salaberry
(1999) also based his second stage on lexical aspect according to Vendlers
classication. However, Salaberry considered three categories, as achieve-
ments and accomplishments were combined as telic events, and the contri-
bution of internal arguments and external arguments was also considered.
Two-way predicational system 179

In contrast, the Predication-eect Hypothesis postulates that it may be


sucient to distinguish two opposite types of predicational aspect: as in
Salaberrys study, both achievements and accomplishments can be dened
as pertaining to discrete units, therefore terminative; and activities and
states can be dened as predications that are left open, therefore durative.
The theoretical analysis described in Section 2 has argued for the exis-
tence of two levels of aspectual information within the sentence: pre-
dicational aspect and grammatical aspect. Each of the two levels can be
characterized as expressing information about its completion: at the pre-
dicational level, it is in the form of the opposition terminative-durative;
whereas at the grammatical level, it is in the form of the opposition perfec-
tive-imperfective. In other words, both oppositions reect a distinction in
completion, on the one hand terminative-durative and on the other hand
perfective-imperfective. In this view, the challenge for L2 learners (no
matter how you divide verbal aspectual classes) seems to arise when the
sense of completion of both levels does not match. If this situation arises,
the predicational level may take over, inuencing the choice of verb form;
that is, when both verbal forms emerge, the completion value of the
predication may be responsible for the choices in interlanguage, and
not the meaning of the verb on its own. The independency of the levels
(predicational and grammatical) may be hard to identify because the
same type of feature seems to characterize them both: the completion
feature. Nevertheless, the learner has to become aware that the two levels
can interact. Only then can learners broaden their interlanguage aspectual
system to accommodate those options that at rst seem unorthodox. In
this sense, it is to be expected that learners will rst use the grammatical
aspectual completion marker (perfective) with those predications that are
also marked as complete, that is, with terminative predications; and they
will use the grammatical aspectual non-completion marker (imperfective)
with durative predications. The less expected combinations are perfectives
with durative predications and imperfectives with terminative predications,
and as argued by Salaberry (2008), it is the task of the learner to discover
that those options are as valid and as common as those where both com-
pletion levels match.
Having a two-way distinction at the predicational level allows for a
simpler and more straightforward comparison with the two opposites at
the grammatical level: terminative-durative versus perfective-imperfective.
It may very well be that this juxtaposition of aspectual levels (predica-
tional and grammatical) allows for some complementary distribution in
interlanguage. That is, during beginning stages of acquisition, perfective
180 Paz Gonzalez

morphology will be clustered with terminative predications, whereas im-


perfective morphology will occur more often with durative predications
(Gonzalez 2003). However, there may be other factors that can interfere
with this proposed complementarity of aspectual meanings. First of all,
narrating a story in Spanish typically calls for the perfective (for a review
of studies defending a discursive approach to grammatical aspect, see
Salaberry 2000), and therefore, the perfective is the most prominent past
tense marker. Secondly, the durative value is the unmarked predication
in language production, and as unmarked, it appears more often. Most
sentences in L1 and L2 language production contain durative predica-
tions.13 These two factors together (namely, perfective as the most used
past tense form and durative as the most common predication type) pose
a problem to the hypothesis as presented so far, because it predicts perfec-
tive markers to appear with verbs in terminative predications and imper-
fectives with verbs in durative predications. Therefore, the Predication-
eect Hypothesis is reformulated as follows: 1) Perfective will appear rst
and with all types of predications; and 2) imperfectives will appear later,
and, when present, will occur more often with durative predications.
Even though the perfective is expected to appear only with terminative
predications, this will not be the case. This is because the perfective is the
semantically prominent past tense marker and will appear everywhere,
irrespective of the type of predication (claim 1 above). On the other hand,
the inherent aspect of the predication also plays a role in interlanguage.
This entails that when the imperfective emerges, it will be more often than
not with durative predications (claim 2 above). The dierence with the
DPTH is found in the second claim, as the second stage in the DPTH
relies on the inherent aspectual meaning of the verb, and the second claim
of the Predication-eect Hypothesis relies on the compositional aspect of
the predication.
Further research needs to substantiate the claims of the Predication-
eect Hypothesis, but evidence from a study of 17 Dutch classroom L2
learners of Spanish following a beginners course shows that a two-way
distinction can prove fruitful for the study of the L2 acquisition of Spanish
tense and aspect (Gonzalez 2003). The participants in the study wrote a
number of compositions, and the analysis of data showed that it was not
the lexical semantics of the verb what partly inuenced the verb choice but

13. This is where the Discourse Hypothesis or the interaction of the Aspect and
Discourse hypotheses come into play (see Chapter 9; Comajoan 2000; Salaberry
2008; and Rosi 2009, among others).
Two-way predicational system 181

rather the predicational aspect. The two-way distinction of the Predication-


eect Hypothesis was more clarifying than the four-way distinction used for
the Aspect Hypothesis, because it mapped not only into the perfective-
imperfective opposition in number but also in meaning (both presenting
their own understanding of completion (see Section 2.4.)).
In sum, the Predication-eect Hypothesis states that the durative-
terminative distinction for the description of the development of aspectual
systems of L2 learners proves more relevant than Vendlers four-way clas-
sication or the intrinsic meaning of the verb, because the imperfective-
perfective choice that learners make at the beginning stages of acquisition
is related to the predicational aspect of the eventuality they aim to describe.
In other words, the four-way distinction is not enlightening to learners
because a pattern in the use of the Spanish past tenses according to the
four-verb types of Vendler may not exist.

5. Conclusion

This chapter has covered a range of research on various issues related to


grammatical aspect. On the theoretical side, a particular description of
the Spanish aspectual system has been provided in Section 2 according to
which Spanish can encode completion at the tense level and at the predica-
tional level. In the tenseless aspectual analysis, every predication receives
its tenseless aspectual meaning from the combination of the lexical mean-
ing of the verb and the delimiting information of the arguments. In this
sense, Spanish grammatical aspect is a sort of lubricant between tense
and the tenseless predication, between the tense level and the predicational
level; as it provides aspectual information but uses tense forms as the
carriers of such information. The analysis presented in Section 2 allows
for predictions at theoretical and empirical levels. Section 3 has discussed
the intuitions of native speakers about the preferences of verb forms accord-
ing to varying aspectual sentence information and how their intuitions on
the acceptance of the grammatical aspect forms depends on other aspec-
tual information contained in a simple sentence. The results have showed
that there is no correlation between predicational aspect and grammatical
aspect in the eyes of native speakers, which means that when choosing
between the perfective-imperfective verb forms, natives are not inuenced
by the lower aspectual level of completion (predicational). In other words,
in Spanish, the terminative-durative distinction does not play a role on the
natives choice of simple past tense verb form, and native speakers choose
182 Paz Gonzalez

between the two inected aspectual forms according to the contextual dis-
course information, not according to the atemporal predicational informa-
tion. Section 4 has analyzed the distribution of the two past tense forms in
Spanish in the learners interlanguage. The AH and the DPTH have been
described and a dierent approach to the L2 acquisition of grammatical
aspect has been proposed, namely, the Predication-eect Hypothesis.
Second language acquisition research involving the acquisition of
grammatical aspect conrms the juxtaposition of grammatical aspect and
inherent aspect. According to most studies on the matter, L2 verbal mor-
phology is closely related to lexical aspect, as when L2 verbal morphemes
emerge in interlanguage, they are not uniformly distributed across all
verbs, but rather are distributed according to the lexical aspectual cate-
gories of verbal predicates (AH). Nevertheless, although the line of thought
started with the AH has denitely hinted in the right direction, it could be
more accurate. Considering the lexical meaning of the verb as the carrier
of all inherent aspectual information is a rst step, but it does not cover
the relevance of the semantics of internal arguments for the overall mean-
ing of the bare aspectual level. Thus, it is not the lexical semantics of the
verbs what partly inuences the verb choice but predicational aspect. The
two-way distinction of the Predication-eect Hypothesis is eective because
it maps not only into the perfective-imperfective division in number but
also in meaning (both presenting their own understanding of completion).
To sum up, essentially, it makes more sense having two types of aspect,
at a predicational level and at a grammatical level, which can have
conicting meanings in the minds of the learners if each level has the
same division type, namely a two-way partition. If an aspectual system is
described with two levels (one predicational, one grammatical) and in
each level only one opposition, it seems to provide a more powerful expla-
nation to imagine the two levels conicting than when one of the levels
contains four values and the other one only two.

References

Andersen, Roger
1989 La adquisicion de la morfologa verbal. Lingustica 1. 90142.
Andersen, Roger
1991 Developmental sequences: the emergence of aspect marking in
second language acquisition. In T. Hueber & C. A. Ferguson
(Eds.), Crosscurrents in SLA and linguistic theories, 305324.
Amsterdam/Philadelphia: John Benjamins.
Two-way predicational system 183

Bardovi-Harlig, Kathleen
1999 From morpheme studies to temporal semantics. Studies in Second
Language Acquisition 21. 341382.
Bardovi-Harlig, Kathleen
2000 Tense and aspect in second language acquisition: form, meaning
and use. Malden, MA: Blackwell.
Bergstrom, Anna
1995 The expression of past temporal reference by English-speaking
learners of French. State College, PA: Pennsylvania State Uni-
versity, Ph.D. dissertation.
Bello, Andres
1988 Gramatica de la lengua castellana destinada al uso de los ameri-
canos. Revised edition of Bello (1947). Madrid: Arco.
Borik, Olga
2002 Aspect and reference time. Utrecht: The Netherlands: University
of Utrecht, Ph.D. dissertation.
Cadierno, Teresa
2000 The acquisition of Spanish grammatical aspect by Danish ad-
vanced language learners. Spanish Applied Linguistics 4. 153.
Camps, Joaquim
2002 Aspectual distinctions in Spanish as a foreign language: The
early stages of oral production. IRAL 40. 179210.
Comajoan, Llorenc
2009 The early L2 acquisition of past morphology: Perfective mor-
phology as an aspectual marker or default tense marker? In D.
Eddington (Ed.), Selected proceedings of the 6th Conference on
the acquisition of Spanish and Portuguese as rst and second lan-
guages, 3143. Somerville, MA: Cascadilla Proceedings Project.
Dietrich, Rainer, Wolfgang Klein & Colette Noyau (Eds.)
1995 The acquisition of temporality in a second language. Amsterdam:
Benjamins.
Doiz-Bienzobas, Aintzane
1995 The preterite and the Imperfect in Spanish: Past situation versus
past view point. San Diego, CA: University of California-San
Diego, Ph.D. dissertation.
Doiz-Bienzobas, Aintzane
2002 The preterite and the imperfect as grounding predications. In F.
Brisard (Ed.), Grounding: The epistemic footing of deixis and
reference, 299347. Berlin: Mounton de Gruyter.
Fernandez Lagunilla, Marina & Elena de Miguel
1999 Relaciones entre el lexico y la sintaxis: adverbios de foco y de-
limitadores aspectuales. Verba 26. 97128.
Fleischman, Suzanne
1985 Discourse functions of tense-aspect oppositions in narrative:
Toward a theory of grounding. Linguistics 23(6). 851882.
184 Paz Gonzalez

Garca Fernandez, Luis


1998 El aspecto gramatical en espanol. Cuadernos de lengua espanola.
Madrid: Arco.
Gonzalez, Paz & Hans Verkuyl
2003 Eects of inherent aspect on the use of Spanish grammatical
aspect. Manuscript.
Gonzalez, Paz & Hans Verkuyl
2003 Aspects on aspect. Theory and applications of grammatical aspect
in Spanish. Utrecht, The Netherlands: LOT.
Hasbun, Leila
1995 The role of lexical aspect in the acquisition of the tense/aspect sys-
tem in L2 Spanish. Bloomington, IN: Indiana University, Ph.D.
dissertation.
Klein, Wolfgang
1994 Time in language. London: Routledge.
Krifka, Manfred
1998 The origins of telicity. In R. Rothstein (Ed.), Events and grammar,
197235. Dordrecht: Kluwer Academic Publishers.
Krifka, Manfred
1989 Nominal reference, temporal constitution and quantication in
event semantics. In R. Bartsch, J. van Benthemand & P. van
Emde Boas (Eds.), Semantics and contextual expression, 75115.
Dordrecht: Floris Publications.
Li, Ping & Yasuhiro Shirai
2000 The acquisition of lexical and grammatical aspect. Berlin: Mouton
de Gruyter.
Rojo, Guillermo
1988 Temporalidad y aspecto en el verbo espanol. LEA. Lingustica
Espanola Actual 10. 195216.
Rosi, Fabiana
2009 Learning aspect in Italian L2: Corpus annotation, acquisitional
patterns, and connectionist modelling. Milano: FrancoAngeli.
Salaberry, M. Rafael
1999 The development of past tense verbal morphology in classroom
L2 Spanish. Applied Linguistics 20(2). 151178.
Salaberry, M. Rafael
2000 The development of past tense morphology in L2 Spanish. Am-
sterdam/Philadelphia: John Benjamins.
Salaberry, M. Rafael
2008 Marking past tense in second language acquisition. A theoretical
model. London: Continuum.
Shirai, Yasuhiro & Atsuko Kurono
1998 The acquisition of tense-aspect marking in Japanese as a second
language. Language Learning 48. 245279.
Two-way predicational system 185

Silva-Corvalan, Carmen
1983 Tense and aspect in oral Spanish narrative: Context and mean-
ing. Language 59. 760780.
von Stutterheim, Christiane & Wolfgang Klein
1987 A concept-oriented approach to second language studies. In C.
W. Pfa (Ed.), First and second language acquisition processes,
191205. Cambridge, MA: Newbury House.
de Swart, Henriette
1998 Aspect shift and coercion. Natural Language and Linguistic
Theory 16. 347385.
Vendler, Zeno
1957 Verb and times. Philosophical Review 66. 143160.
Verkuyl, Hans
1972 On the compositional nature of the aspects. FLSS 15. Dordrecht:
D. Reidel Publishing Company.
Verkuyl, Hans
1993 A theory of aspectuality. The interaction between temporal and
atemporal structure. Cambridge: Cambridge University Press.
Verkuyl, Hans
1999 Aspectual issues. Studies on time and quantity. California: CLSI
Publications, Stanford University.
Vet, Co
1994 Petite grammaire de lAktionsart et de laspect. Cahiers de Gram-
marie 19. 117.
Weist, Richard
1986 Tense and aspect. In P. Fletcher & M. Garman (Eds.), Language
acquisition: Studies in rst language development, 356374. Cam-
bridge: Cambridge University Press.
Wiberg, Eva
1996 Reference to past events in bilingual Italian-Swedish children of
school age. Linguistics 34. 1087114.
Chapter 6
Research design: Operationalizing and testing
hypotheses

M. Rafael Salaberry

1. Introduction

To investigate aspectual knowledge among both native speakers and second


language (L2) learners, it is important to dene and circumscribe: (a) the
specic theoretical construct that denes aspect, and (b) the methodological
procedure to gain access to the selected theoretical representation of aspect.
More specically, the wide range of denitions of tense-aspect knowledge,
and the variety of procedures to assess such knowledge are directly reected
on distinct ways to operationalize and assess research hypotheses. In the
present chapter, I will analyze how both theoretical descriptions and meth-
odological procedures in the study of tense-aspect knowledge may lead to
divergent outcomes (under similar experimental conditions) or, alterna-
tively, divergent outcomes even when sharing theoretical descriptions.
For reasons of space, I will restrict my analysis to one particular com-
ponent of aspectual knowledge (iterativity) as realized in the Romance
languages (i.e., Portuguese and Spanish). More specically, I will analyze
various empirical studies which seem to provide contradictory ndings to
the extent that (a) studies using decontextualized prompts match the nd-
ings from other studies using sentence-level prompts, while at the same time
(b) a study with contextualized prompts is not replicated with another one
using discursive prompts.

2. Aspectual representation of iterativity

2.1. Iterativity as a distinct aspectual concept


Aspect refers to the temporal structure of situations (Smith 1997: 3). The
most commonly discussed aspectual contrast brought about by the use
of perfective and imperfective past tense morphology is the notion of
boundedness (e.g., Depraetre 1995; de Swart 1998; Klein 1986; Smith 1997).
188 M. Rafael Salaberry

There are, however, other aspectual notions that have an impact on the
native-like selection of past tense morphology. For instance, Filip (1999:
119, emphasis added) specically states that (p)luralities of eventualities,
which are expressed by iterative and habitual sentences, belong to a dimen-
sion of conceptualization that is orthogonal to the classication of verbal
predicates and sentences into states, processes and events.1 Bertinetto
(1994) makes a similar argument stating that habituality belongs to the
subdomain of aspect, whereas lexical aspectual classes belong to the sub-
domain of actionality.
Overall, situations that are repeated in time (cf. iterated eventualities)
are represented as either iteratives or habituals.2 Comrie (1976: 27) justies
the theoretical contrast between habituality and iterativity, because the
repetition of an event is not enough to make it a habitual.3 That is, all of
the repeated instances of the event can be viewed as a single situation,
albeit with internal structure, and referred to by a perfective form. Binnick
(1991: 155) proposes further that the imperfective may convey habituality
(repetitive episodes somewhat distantly spaced in time are viewed as a
unit) whereas the perfective conveys iterativity (repetitive episodes rather
closely spaced in time and viewed as a unit). For instance, in Spanish and
Portuguese, both single and iterated telic events can be conveyed with the
use of Preterite (PRET) in (1a), whereas habituality is typically expressed
with the use of the Imperfect (IMP) in (1b). On the other hand, the con-
cept of iterativity is expressed with the Preterite in (1c).

1. In Filips nomenclature, processes correspond to activities, whereas events


correspond to telic eventualities.
2. Note that the term iterative may refer to the overall concept of iterated even-
tualities in general, or it may refer to the more specic concept that stands
in contrast with the correlated concept of habituality (e.g., Langacker 2000;
Lenci and Bertinetto 2000).
3. One reviewer points out that the distinction between iterative and habitual
from Shirai (1991) should be considered for the present analysis. However,
Shirais contrast between iterative and habitual is predicated on the contrast
of repeated actions on a single occasion (iterative) versus repeated actions on
dierent occasions (habitual). A similar position is advanced by Klein (1994).
As is clear from the examples to be reviewed, such denition is not necessarily
the one represented in the the uses of the Spanish Preterite and Imperfect.
Operationalizing hypotheses 189

(1) a. Ayer el cartero llego tarde.


Yesterday the mailman arrived (PRET) late.
b. Cuando era nino, el cartero llegaba tarde.
When I was a child, the mailman arrived (IMP) late.
c. Durante muchos meses, el cartero llego tarde.
For months the mailman arrived (PRET) late.
English native speakers are very familiar with the contrast of single events
versus habitual events (1a versus 1b), both through high frequency repre-
sentations of such contrast in L2 input data, along with explicit instruction
on such contrast. Less well known among these learners is the contrast
between single and iterated events (1a versus 1c).

2.2. Interpretations of ndings on the L2 acquisition of the concept


of iterativity
Previous studies carried out with Spanish L2 data (e.g., Perez-Leroux et al.
2007; Salaberry 2012; Salaberry and Scholes to appear; Slabakova and
Montrul 2007) consistently show that whereas both English native speakers
learning Spanish and Spanish native speakers consistently accept the use
of the Preterite with sentences depicting a single event (1a), only Spanish
L2 learners tend to reject the use of the Preterite depicting an iterated telic
event (1c). Despite the analysis of the apparently consistent ndings about
the rejection of the Preterite as an iterativity marker in Spanish L2 data,
there is at least one recent study based on L2 Portuguese data that reveals
that non-native speakers are arguably capable of correctly assigning the
use of the perfective form to iterated events (i.e., Rothman and Iverson
2008). This study brings up the distinct possibility that the selection of L2
(i.e., Spanish or Portuguese) makes a dierence in the acquisition of the
concept of iterativity among English speakers. However, as I will discuss
below, it is also possible that both the theoretical framework and the type
of language input will have an eect on the outcome of the assessment.
It is important to note that, despite the fact that the studies with Spanish
L2 data cited above have provided similar results with regards to the reac-
tions of learners to the linguistic prompts (cf. Spanish L2 learners rejection
of the Preterite to mark iterated eventualities), the researchers interpreta-
tions of the ndings are not the same. For instance, Slabakova and Montrul
(2007) argued that L1 English speakers are not able to transfer knowledge
of how iterated events are marked in L1 English to L2 Spanish. Slabakova
190 M. Rafael Salaberry

and Montrul note, however, that this shortcoming is not to be regarded as a


limitation on the acquisition of the L2 grammar per se. That is, Slabakova
and Montrul propose that the type of aspectual information provided by
adverbials is outside of the scope of grammatical representation, and more
properly, part of the realm of pragmatics knowledge.4 In contrast, Perez-
Leroux et al. propose that knowledge about iterativity remains within the
realm of the L2 grammar. That is, whereas Montrul and Slabakova depict
acquisition as feature activation, for Perez-Leroux et al., acquisition is
theoretically represented as lexical development, where learners must boot-
strap the selectional features of each functional head independently
(4434). Thus, L2 learners learn the semantic selectional restrictions asso-
ciated with the perfective rst (the Imperfect is acquired later). As Perez-
Leroux et al. point out (443), within Slabakova and Montruls position,
variations from the target depend solely on performance factors,
whereas within the framework adopted by Perez-Leroux et al. students
responses leaners could in principle determine the s(emantic)-selectional
features of each head independently.5
Finally, Salaberry (2012) argues that the development of a representa-
tion of aspectual meanings among adult L2 learners is determined primarily
through L1 transfer of aspectual representations. For instance, through
transfer from their native language, L1 English speakers may initially assume
that the category Simple Past corresponds to the Preterite and that uses of
the periphrastic progressive are largely equivalent to the Imperfect. Learners
would of course use these generalizations as rules of thumb, quickly realizing
there are exceptions to them. The use of these basic generalizations is
eventually complexied with the use of additional mostly lexically-based
patches (cf. lexical aspectual categories or specic verbs) that help learners
reconcile their overly simplied system with the available data from the L2.

3. Theoretical descriptions: Operationalizing hypotheses

In this section, I review three distinct theoretical perspectives on the repre-


sentation of the iterative-habitual contrast in the most recent L2 acquisition

4. This is a valid methodological decision that can help us circumscribe the


theoretical analysis of data within a limited contextual space (i.e., adverbial
information is left out of the picture so to speak). Being a methodological
decision, however, it cannot circumvent the fact that native speakers intui-
tions about aspectual interpretations are contextualized.
5. The position advocated by Perez-Leroux is compatible with the argument
advanced by the Default Past Tense Hypothesis (Salaberry 1999, 2003, 2008).
Operationalizing hypotheses 191

literature: Smiths basic versus derived-level interpretations, Menendez-


Benitos accidental versus non-accidental generalizations, and Langackers
actual versus structural plane. Each position to be described will progres-
sively expand the range of discourse contexts available for the interpreta-
tion of aspectual meanings.
I argue that aspect cannot be properly understood without reference
to a level of contextualization above the verb phrase. In essence, it is
not possible to dene or assess the construct of aspect without adopting a
discursively-distributed denition of aspectual meanings (cf. Salaberry
2008). That is, aspect, by its very nature, can neither be understood nor
explained if its theoretical construct is dened out of context. Within this
perspective, the dichotomy of core-peripheral grammar or syntactic-pragmatic
representations of aspect can be reconceptualized as invariant and contex-
tualized meanings (cf. Binnick 1991; Doiz-Bienzobas 1995, 2002).

3.1. Basic-level versus derived level of interpretation


The eect of contextual information beyond the eect of internal and exter-
nal arguments has generated a multitude of theoretical accounts. Some
researchers have approached this problem by way of expanding on the
basic aspectual distinctions they have proposed. For instance, Smith
makes a distinction between basic-level verb constellations and derived-
level verb constellations (1997: 5456). The former are determined by
compositional rules that include the eect of the main verb and its argu-
ments (both internal and external), whereas the latter includes not only
the basic-level verb constellation, but also the eect of adverbials (e.g.,
for an hour), superlexical morphemes (e.g., to begin, to nish) and verbal
morphology (e.g., progressive). For the purpose of this chapter, I will
focus on the eect of adverbials only.
Smith (1997: 4) argues that the following examples (originally from
Verkuyl) show very clearly that situation type meaning is compositional:
it is built up with the verb, arguments and adverbs of a sentence.6 Smith
points out that the combined meaning of the external argument (the subject
of the sentence), and the adverbial ( for years in 2a) entails that sentence
(2a) is atelic and (2b) is telic.

6. Examples (2a) and (2b) are numbered (39a) and (39b) respectively in Smith
(1997).
192 M. Rafael Salaberry

(2) a. [Famous movie stars] discovered that little spa for years.
[atelic]
b. [A famous movie star] discovered that little spa.
[telic]
Note that if we added the same adverbial phrase used in (2a) to sentence
(2b), we would be shifting the telic nature of that sentence (without chang-
ing the external argument) to an atelic one.
(2) c. A famous movie star discovered that little spa [for years].
[atelic]
That is, in (2c) one single movie star visited the same spa over several
years (metaphorically discovering it each time).
Smith does not make a principled distinction among dierent types of
iteration of eventualities (for a discussion of the problem of subsuming
perfectivity within habituality, see Lenci and Bertinetto 2000).7 Essentially,
Smith proposed that in the context of a frequency adverbial, almost all
verb constellations can be taken as habitual (35, emphasis added). And,
in the case of sentences without an explicit frequency adverbial, Smith
argued that our world knowledge leads us to interpret sentences as habitual.
For instance, Smith states that [g]etting up and reading the newspaper are
ordinary and likely to be taken as habitual.8
(3) a. Eva got up at noon last summer.
(4) a. Mary read the newspaper last summer.
Interestingly, however, the Spanish translations of the sentences above
are more naturally expressed with the imperfective and perfective forms
respectively.
(3) b. Eva ?se levanto/se levantaba al medioda el verano pasado.
Eva got up (PRET/IMP) at noon last summer.
(4) b. Mara leyo/?lea el periodico el verano pasado.
Mary read (PRET/IMP) the newspaper last summer.

7. Lenci and Bertinetto (2000: 251) surmise that it might be straightforward to


include the perfective sentence in the domain of habituality. However, the
consequence of this solution is that we would lose the correlation between
habituality and imperfectivity.
8. These examples are presented in Smith (1991: 41), but not in the (1997) edition.
The overall argument, however, is the same in both editions.
Operationalizing hypotheses 193

The almost complementary distribution of the perfective and imperfec-


tive markers in the above sentences in Spanish (and other Romance lan-
guages for that matter) brings up two important theoretical issues. First,
the use of contrasting morphological markers shows that Spanish speakers
are aware of meaningful aspectual distinctions that are, for that reason,
signaled through morphosyntactic means. That is to say, Smiths conclu-
sion that the notion of repetition or iteration belongs to the realm of prag-
matics (cf. world knowledge) based on examples from English appears
to be unsatisfactory, or at a minimum, incomplete given the above examples
from Spanish.
Second, the possible complementary distribution shown above is a
tendency that can be easily reversed if we added some relevant context
(in the form of additional adverbials) as in (3c) and (4c) below. Thus, the
strong tendency of the Preterite to convey punctuality (default meaning) in
association with a telic event (prototypical match) cannot be easily over-
come with the addition of the durative adverbial (last year) in sentence
(3b).9 If, however, another adverbial were to be added, in particular one
that signals the notion of repetition more clearly (todos los das in 3c),
then the Preterite shifts meaning from punctuality to iterativity and the
sentence is acceptable.
(3) c. El verano pasado, Eva se levanto al medioda [todos los das].
Last summer, Eva got up (PRET) at noon [every day].
(4) c. Mara lea el periodico [con regularidad] el verano pasado.
Mary [regularly] read (IMP) the newspaper last summer.
Similarly, the preference for the selection of the Preterite in (4b) above
may possibly be associated with the fact that reading the newspaper (leer
el periodico) is an atelic event. That is, the strong tendency of the Imper-
fect to convey background information (default meaning) in association
with an atelic event (prototypical match) cannot be easily overcome with
the addition of the durative adverbial (last year) in (3b). If, however, another
adverbial that further specied the notion of repetition (con regularidad in
4c) were to be added, then the Imperfect would shift meaning from signal-
ing background information to conveying habituality. As a consequence,
sentence (4c) becomes acceptable. In sum, despite Smiths assertion, the

9. An equivalent association is the one between lexical aspect and grammatical


aspect, which Michaelis (2004) describes as isomorphic characterization.
194 M. Rafael Salaberry

iterativity-habituality contrast appears to belong to the realm of gram-


matical conceptualization, and, more importantly, such contrast seems to
be dependent on a broad level of contextualization.

3.2. Accidental versus non-accidental


A more encompassing perspective of the role of adjuncts within an overall
composition of aspectual meanings is provided by Menendez-Benito (2002:
365), who argues that [i]t is clear that to ultimately be considered success-
ful, an account of the perfective/imperfective contrast should lead to an
explanation of how these aspects interact with adverbs . . . Menendez-
Benito underlines that, at a basic level, the Imperfect focuses on the fact
that eventualities are generalizable (5a), whereas the Preterite serves to
focus on the episodic nature of particular eventualities (5b).10
(5) a. Cuando vena a mi casa, Juan fumaba.
When he used to/came (IMP) to visit, Juan used to/smoked
(IMP).
b. Cuando vino a mi casa, Juan fumo.
When he came (PRET) to visit, Juan smoked (PRET).
She notes, however, that when our descriptions of eventualities set in
past tense contexts include adverbial quantiers (e.g., siempre), both the
Preterite and Imperfect convey the notion ofgeneralizations of particular
episodes (366) as shown in sentences (6a) and (6b) below.11
(6) a. Siempre que vena a mi casa, Juan fumaba.
Whenever he used to/came to visit (IMP), Juan used to/smoked
(IMP).
b. Siempre que vino a mi casa, Juan fumo.
Whenever he came to visit (PRET), Juan smoked (PRET).
That is, both sentences above make reference to the iteration of the even-
tuality of smoking. Thus, in principle we should predict imperfective and
perfective forms to be interchangeable when the sentence contains an
adverbial quantier (369).

10. The following examples (514) are from Menendez-Benito.


11. Bonami (1997) points out that aspectual morphology is associated with a
default quantier, which can be overriden by an overt adverb of quantication.
Operationalizing hypotheses 195

Nevertheless, Menendez-Benito adds that this prediction is not borne


out: even when an adverbial quantier is present, imperfective and perfec-
tive forms do not have the same distribution . . . (372). That is, Spanish
perfective and imperfective behave dierently with respect to generic adverbs
(e.g., normalmente normally) and durational phrases (e.g., durante dos anos
for two years). In essence, the imperfective can combine with generic
adverbs (7a), but the perfective does not render a felicitous sentence (7b).
In contrast, Menendez-Benito argues, the perfective is acceptable with
durational phrases (8a), but the imperfective is not (8b).12
(7) a. El ano pasado Juan normalmente iba al cine.
Last year, Juan normally went (IMP) to the movies.
b. *El ano pasado Juan normalmente fue al cine.
Last year, Juan normally went (PRET) to the movies.
(8) a. Durante dos anos Juan fue al cine.
During two years, Juan went (PRET) to the movies.
b. *Durante dos anos Juan iba al cine.
During two years, Juan went (IMP) to the movies.
To solve the dilemma about the distinct distribution of past tense forms,
Menendez-Benito argues that Goodmans (1947) contrast between non-
accidental (9) and accidental (10) statements posits an interesting correla-
tion with iterated eventualities.
(9) All dimes are silver.
(10) All the coins in my pocket are silver.
More specically, Menendez-Benito argues that [w]hile the generaliza-
tions made by the imperfective sentences pattern with sentences like (9),
the generalizations made by perfective sentences pattern with sentences

12. Menendez-Benito mentions a third category: exceptive temporal phrases that


denote a particular time interval. For instance, in the sentence Siempre que le
hice/haca una observacion, se lo tomo/tomaba bien excepto el lunes dos de abril
de 1998, [Whenever I made/used to make an observation, he accepted/used to
accept it, it except on Monday April 2, 1998] Menendez-Benito argues that the
use of the perfective form is probably more acceptable than the imperfective.
This judgment of unacceptability is, however, questionable. In fact, Menendez-
Benito recognizes that the use of the imperfective is not necessarily ungram-
matical given that she does not categorize the use of the imperfective form as
ungrammatical, but rather unlikely (She marks it with ?? Instead of *).
196 M. Rafael Salaberry

like (10) (373, emphasis added). That is, sentences with the imperfective
(11) convey the notion of non-accidental generalizations (law-like) and
sentences with the perfective (12) convey the notion of accidental general-
izations. Menendez-Benitos argument leads us to conclude that accidental
generalizations are exceptional in the sense that they are restricted to the
specic time frame in which they happened: they are equivalent to the
exceptional circumstance that all coins in my pocket happen to be made
of silver.
(11) Siempre que vena a mi casa, Juan fumaba.
Whenever he used to/came to visit (IMP), Juan used to/smoked
(IMP).
(12) Siempre que vino a mi casa, Juan fumo.
Whenever he came to visit (PRET), Juan smoked (PRET).
To substantiate the proposed correlation with Goodmans proposal,
Menendez-Benito argues that accidental generalizations make reference
to stage-level predicates, whereas non-accidental generalizations make
reference to kind-denoting subjects.13 For instance, she states that (13a)
conveys the meaning that the price of the book was $40, whereas (13b)
conveys the meaning that, for instance, someone bought the book and it
cost $40.
(13) a. El libro costaba $40.
The book cost (IMP) $40.
b. El libro costo $40.
The book cost (PRET) $40.
This argument is problematic given that Menendez-Benito concedes that
there are contexts in which a perfective verb may get an individual-level
interpretation as in (14).

13. Menendez-Benito specically states that Non-accidental generalizations are


law-like statements, and make reference to kind-denoting subjects, thus they
support the truth of counterfactuals (372374). For reasons of space, I will
not discuss this argument in detail. Nevertheless, I believe this assertion is
tautological, thus it is impossible to disambiguate the referential meaning of
the counterfactual statement. For instance, Menendez-Benito provides no
independent evidence to demonstrate that the counterfactual makes reference
to the imperfective sentence, but not the perfective one (cf. En la Republica,
siempre que alguien era/fue acusado de un crimen, tena/tuvo un juicio justo,
In the Republic, when someone was accused of a crime, he always had (IMP/
PRET) a fair trial).
Operationalizing hypotheses 197

(14) Juan supo frances hasta que se mudo a los EEUU.14


Juan knew (PRET) French until he moved to the US.
Even though Menendez-Benito disregards such example as exceptional
(I believe that the tendency noted above is strong enough to still count as
evidence for my claim, 376377), there are many examples that uphold the
notion that the perfective form can be used with individual level predicates.
For instance, Guell (1998: 99) points out that the perfective may be used
with the same stative verb saber (to know) in contexts that are even more
puzzling to non-native speakers than the previous one, as shown in (15).
(15) Lo supo/*saba durante mucho tiempo.
(S/he) knew (PRET/IMP) it for a long time.
Indeed, we can easily nd contextual support to convey an individual-level
interpretation of the main example used by Menendez-Benito and conclude
that the verb costar (to cost) in the perfective should be given an individual-
level interpretation as well:
(16) El libro siempre costo $40, hasta que Amazon lo puso a la venta.
The book always cost (PRET) $40, until Amazon started selling it.
In sum, the proposal to characterize the types of iterativity brought
about by the use of perfective and imperfective Spanish past tense markers
in terms of accidental versus non-accidental sentences (Goodman 1947)
has notable shortcomings. Nevertheless, Menendez-Benitos argument serves
to make a compelling case to consider the information provided by adverbial
phrases (adjuncts) as part of grammatical knowledge, and more importantly,
to advance the argument that such (contextually-based) knowedge has to
be incorporated to any model of grammatical competence.

14. One reviewer questioned the grammaticality of sentence (14). I believe that
Menendez-Benito is right to consider it grammatical/acceptable in the context
of a non-prototypical interpretation. For instance, one possible context that
would license this sentence is one in which the time-frame that serves as back-
ground information is one in which Juan spoke French as a second or third
language, and then, after a long period of absence from the French-speaking
scene, he is no longer able to keep up with a language that he no longer uses
regularly.
198 M. Rafael Salaberry

3.3. Actual versus structural plane


A third theoretical perspective on the contrastive meaning between iteratives
and habituals is the one advanced by Langacker (2000, 2009). Langacker
conceptualizes the information provided by adverbials as part of gram-
matical knowledge. More specically, he draws a distinction between the
actual plane comprising event instances that are conceived as actually
occurring, and the structural plane in which event instances [have] no
status in actuality (2000: 251). Langacker argues further that repetitive
sentences (iterative) are part of the actual plane, whereas habitual sentences
belong to the structural plane. For instance, the repetitive in (17) shows the
component events of individual instances of cats and birds anchored to
particular points in time (expressing actual, episodic, events). In contrast,
in a habitual sentence as in (18), the component events are not anchored
to any particular points in time (thus, conveying a habitual meaning).
(17) Repetitive: My cat repeatedly stalked that bird.
(18) Habitual: My cat stalks that bird every morning.
In essence, iterated-event predications express actual events like other kinds
of episodic predications. In contrast, habitual sentences express structural
events similar to the meanings expressed by generic sentences.15
Langackers proposal is not unique given that other researchers have
made similar arguments. For instance, Bhatt (1999: 53) contended that
the habitual diers from iterative and frequentative crucially by the fact
that the former is inductive whereas the latter are deductive. Inductive
refers to the fact that an event is habitual upon the observation of only
one occurrence of the event. In contrast, for iteratives, we need to observe
several occurrences of the event. For instance, frequency adverbials such
as once, twice, ten times are relevant for iteratives but not for habituals.
That is, the anchoring/deductive nature of iterated events is felicitous in
the context of durational phrases (e.g., durante dos anos for two years),
an argument advanced by Menendez-Benito as well.

15. Interestingly, the contrast proposed by Langacker is also made by several


other researchers working within very dierent theoretical models. For instance,
Lenci and Bertinetto (2000: 25) express that . . . language represents series of
events in two dierent modes: as normative habits or simply as accidental
pluralities of occurrences . . . Perfective sentences can represent series of events
or the frequency of events, but always remaining the vehicles of episodic propo-
sitions.
Operationalizing hypotheses 199

Doiz-Bienzobas (1995: 107) extended Langackers conceptualization of


iterative and habitual sentences to analyze the meaning contrasts of Spanish
Preterite and Imperfect. Doiz-Bienzobas and she concluded that habituals
are represented with the Imperfect (19a) whereas iterated meanings are
represented with the Preterite (19b):
(19) a. El ano pasado iba a nadar todos los das.
Last year I used to go (IMP) swimming every day.
b. El ano pasado fui a nadar todos los das.
Last year I went (PRET) swimming every day.
Doiz-Bienzobas argues that the dierent instances of the event of swim-
ming depicted in (19a) are not anchored to particular points in time (struc-
tural plane); thus, any one of them may fail to be present without neces-
sarily aecting the interpretation of the habitual meaning of the sentence.
In contrast, all of the instances of the event of swimming depicted in (19b)
are anchored to particular points in time. Thus, sentence (19b) implies that
the speaker went to swim every day last year.16
In sum, Smiths contrast between basic and derived level interpretations
leaves out of the representation of aspectual meanings the inherently
essential component of adverbials. Thus, Smiths proposal leads to an
incomplete and decontextualized representation of aspect that fails to
account for the apparent contrasts between iterative and habitual sentences
in Spanish. Menendez-Benitos proposal solves this problem by explicitly
recognizing the eect of adjuncts on aspectual interpretations. However,
Menendez-Benitos claim introduces certain restrictions: the perfective
form is associated with exceptional, accidental cases only. Some of the
data we reviewed above seem to raise doubts about the generalizability of
her contention. Finally, Langackers proposal (actual versus structural
plane) is the most open to the encompassing eect of context on the repre-
sentation of aspectual contrasts; thus it seems to capture the fact that both
iterative and habitual sentences may be used to refer to the same eventual-
ity in reality. In other words, Langackers denition captures the fact that

16. Comrie (1976: 29) argues that habituals no longer hold only as a result of an
implicature, which is weaker than an implication: if denied, the situation no
longer holds, but if not denied, or suspended by an explicit remark from the
speaker to the eect that he does not know whether or not it holds, then it will
be taken to hold.
200 M. Rafael Salaberry

there is an aspectual meaning contrast (i.e., grammatical) introduced by


the use of the perfective or imperfective form. The notion of the anchoring
of eventualities in association with the use of specic adverbial phrases
(cf. Menendez-Benitos generic adverbials versus durative phrases) and
the selection of either Preterite or Imperfect is more apt to capture the
iterative-habitual contrast in meaning.

4. Methodological procedures: Testing hypotheses

Evaluating speakers knowledge about aspectual distinctions within the


scope of broad contextualization of the construct is complex because there
are multiple variables to consider for the selection of the actual linguistic
prompts to study aspect. This is even more of a challenge in cases when
linguistic prompts are longer than a single sentence, a necessary incon-
venience of methodological design given the discussion on the theoretical
representation of aspect presented above.

4.1. Language prompts


Assessing speakers knowledge about theoretical constructs that do not
have a high frequency of use in most speech tasks used in L2 research
presents an important challenge for the research design of any study. In
particular, the representation of iterative and habitual meanings will be
distinctly represented through various levels of contextualization language
of prompts that need to be carefully analyzed given the conclusions
gathered in the previous section with respect to the contextual scope of
the denition of iterative-habitual. In this section, I rst present a brief
review of the basic experimental set-up of relevant previous studies. Then,
I oer a more detailed discussion of one of the studies that has used a
more encompassing and necessary discursive frame of analysis and yet, it
has oered contradictory ndings with reference to previous studies.
Slabakova and Montrul (2007) analysed data from 60 English native
speakers and 27 native Spanish speakers. The English native speakers
were divided into two groups according to their prociency in Spanish:
27 advanced learners and 33 intermediate learners. The test instrument
was a grammaticality judgement test that consisted of 49 pairs of sentences
testing six conditions of shifted aspectual interpretations and one set of
seven distractor sentences. The sentences were presented in pairs to test
Operationalizing hypotheses 201

the meaning contrast of verbs with a basic and a shifted interpretation


brought about by grammatical or pragmatic means. The grammatically
based shifted interpretations of our interest were the ones in which achieve-
ments were shifted into habituals with the addition of an adverbial (sentences
20a and 20b)
(20) a. Durante muchos meses, el tren del medioda llego tarde.
For months, the 12 oclock train arrived (PRET) late.
b. Ayer, el tren del medioda llego tarde.
Yesterday, the 12 oclock train arrived (PRET) late.
Informants judged all sentence pairs on a ve-point scale from 1 (unaccept-
able, I would never say this) to 5 (perfect, I would say this). The results
of the study revealed that learners failed to distinguish the dierence in
meaning represented by telic events that had been iterated (20b). Further
substantiation of these results comes from the study carried out by Salaberry
and Scholes (to appear), who replicated the ndings from Slabakova and
Montrul with the inclusion of participants who had even more advanced
knowledge of the L2 than the ones used in the original study of Slabakova
and Montrul.
Perez-Leroux et al. (2007) expanded on the previous results by focusing
on the eects of explicit versus implicit markers of iterativity. Thus, they
analysed data from 41 L2 Spanish students and 10 native Spanish speakers
on the eects of unique, habitual, and iterated situations with the use of a
50-item grammaticality judgement task and an 18-item translation task.
The learners were divided into two levels of prociency based on course
enrollments in second or third year. There were two dierent conditions
that triggered an iterative interpretation: explicit with the use of what
they call iteration adverbial (e.g., repetidamente, repeatedly) as shown
on (21a,b) and implicit with the use of a duration adverbial (e.g., por das,
for days) as depicted in (22a,b). The unique situation was acceptable with
the Preterite, the habitual with the Imperfect, and both iterated situations
were acceptable with the Preterite only.
(21) a. El terremoto sacudio la ciudad por das.
The earthquake shook (PRET) the city for days
b. El terremoto sacuda la ciudad por das.
The earthquake shook (IMP) the city for days.
202 M. Rafael Salaberry

(22) a. El terremoto sacudio la ciudad repetidamente.


The earthquake shook (PRET) the city repeatedly.
b. El terremoto sacuda la ciudad repetidamente.
The earthquake shook (IMP) the city repeatedly.

The results revealed that the L2 learners accepted the prototypical com-
binations of the Preterite with punctual eventualities and the Imperfect
with habitual eventualities. However, learners failed to reject the (ungram-
matical) use of the Imperfect with both implicit and explicit iterated events
(21b and 22b). Perez-Leroux et al. attribute the outcome of their study to
the eect of instruction. They argue that [g]rammatical instruction does
not distinguish between these senses, it merely states that repetition in
the past is expressed in the Imperfect form (434).
Finally, Salaberry (2012) expanded the analysis of previous studies based
on sentence-based prompts with the use of more extended discourse-based
prompts that contextualized the aspectual choices, thus maximizing the
probability that participants would focus their attention on the target
aspectual construct. His ndings replicated the results of previous studies
pointing to a clear dissociation in the interpretation of iterative and habitual
choices. That is, advanced L2 speakers (cf. teaching assistants) paralled
the choices of native speakers with regards to the preferred aspectual
marker to signal habituality, but failed to mark the contrast signalled by
iterative sentences as distinct from habitual choices.

4.2. Analysis of contradicting data


Despite the apparent consistency in the outcome of studies that investigated
knowledge about iterativity and habituality among L2 learners, there is one
study that has failed to replicate such ndings. More importantly, such an
outcome has been used to substantiate a strong argument for the existence
of ultimate attainment in aspect (see Chapter 4). In this section, I analyze
the research design of the study carried out by Rothman and Iverson
(2008). This is justied for three reasons. First, their study is framed
within Menendez-Benitos (2002) theoretical proposal about accidental
and non-accidental generalizations, a proposal that explicitly incorporates
the eect of adverbials to the composition of aspectual information. Second,
their study uses linguistic prompts longer than single sentences, and thus it
provides for a more complex and accurate assessment of (discourse-based)
aspectual knowledge. Third, the ndings from Rothman and Iverson dier
from the outcome revealed by the other studies focused on the assessment
of iterativity with Spanish L2 data as described in previous sections.
Operationalizing hypotheses 203

4.2.1. Research design and outcomes of the study


The study carried out by Rothman and Iverson (2008) used language
prompts that provided participants with a broad discursive context to
help them decide their choices of aspectual meanings.17 The participants
were 31 non-natives and 19 native speakers of Portuguese from Brazil. The
non-natives were recruited among undergraduate and graduate American
university students enrolled in a study-abroad program in Salvador, Brazil.
They were divided into two groups of prociency: intermediate (14 learners)
and advanced (17 learners).18 Rothman and Iverson used three tests: two
sentence-conjunction tests and one context-sentence matching test. One of
the two sentence-conjunction tests and the context-sentence matching test
primarily focused on the assessment of the distinction between accidental
versus. non-accidental meanings conveyed by the use of the imperfective
and perfective in Portuguese (see examples below). The remaining sentence-
conjunction test was focused primarily on the contrast between kind-
denoting and group-denoting readings.19 All tests were contextualized
with a short paragraph. The main dierence between each type of test
was that the sentence-conjunction tests had one sentence after the short
paragraph that had to be rated on a 5-point scale ranging from semanti-
cally odd (2) to perfectly acceptable (2). In contrast, in the context-
sentence matching test, participants read a short paragraph that served as
context for two follow-up sentences essentially identical except for the fact
that one of the two sentences contained the imperfective form and the
other one the perfective form. Participants in the study were asked to
choose which sentence was better according to the given context provided
in the paragraph.

17. Iverson and Rothman (2008) is a smaller study that precedes and is subsumed
by the one published in Rothman and Iverson (2008).
18. Of the 17 advanced learners, three were not part of the main pool of partici-
pants (i.e., American university students studying in a summer language pro-
gram), but rather natural learners who had acquired Portuguese while living
in Brazil. Rothman and Iverson do not justify why they added these learners
to their pool of subjects, especially considering that the prociency tests had
already divided the participants into an equal number of intermediate and
advanced learners (i.e., 14 and 14).
19. Menendez-Benito (2002) proposes that the imperfective allows both types of sub-
ject DP readings, whereas the perfective allows only group denoting readings.
204 M. Rafael Salaberry

The statistical analysis of the ndings of the study conducted by Rothman


and Iverson revealed that the advanced group performed like the control
group of native speakers, but that the intermediate learners failed to
produce native-like results.20 Nevertheless, both groups showed a much
stronger preference for preterite in both episodic and accidental general-
ization sentence types (302). Rothman and Iverson ultimately conclude
that these results along with similar ndings from the other tests used in
their study constitute evidence in favor of adult UG-continuity theories
(307). As stated before, however, these ndings with Portuguese data
stand in contrast with the results from the studies carried out with L2
Spanish speakers. How can we reconcile the distinct outcome? In the
following section, I analyze the language prompts used in this study to
assess the possibility that the responses provided by native speakers may
have been unreliable due to methodological factors.

4.2.2. Language prompts

I will restrict my discussion to the analysis of the outcome of the context-


sentence matching test.21 Rothman and Iverson describe the rst context
below as an example of an accidental generalization. Arguably, an acci-
dental generalization would trigger the use of the sentence with the perfec-
tive form (option b, in bold is the expected response).

1. Tem uma garota na minha classe que conversa muito. O nome dela e
Jimena. O problema com ela e que, como fala tanto, ela nunca me deixa pre-
star atencao. Entao, trato de ir a biblioteca em horarios nao muito comuns
para nao encontrar ela. Mesmo assim, as vezes nao dou sorte e na semana
passada . . .

20. This was true across all three tests, with the only exception of contexts show-
ing non-accidental generalizations presented in the rst sentence-conjunction
test.
21. Not only are all three types of tests focused on assessing the same type of
knowledge (cf. the grammatical marking of iterated eventualities), but they
also use variations of the same linguistic triggers. In particular, most contexts
are represented with variations of the use of the adverbial sempre (always)
with a limited number of time frames (last year, last week, etc.).
Operationalizing hypotheses 205

Theres a girl in my class who talks a lot. Her name is Jimena. The problem
with her is, as she talks so much, she never lets me pay attention. Therefore, I
try to go to the library during unusual times so I dont see her. Even so,
sometimes I dont have luck and last week . . .
a. empre que eu ia a biblioteca, eu encontrava com ela.
every time I went-IMP to the library, I saw-IMP her.
b. sempre que eu fui a biblioteca, eu encontrei com ela.
every time I went-PRET to the library, I ended up seeing-PRET her.

In contrast, the context of the second sentence below is described as an


example of a non-accidental generalization, and as such it is expected to
trigger the use of the sentence with the imperfective form (option a, in bold
is he expected response).

2. Lorenzo ama produtos lacteos. De fato, ele os cosome todos os dias. Ele
gosta desde o queijo camembert ate o iogurte sem sabor. No ano passado ele
cou na minha casa para uma visita e . . .
Lorenzo loves dairy products. In fact, he has them every day. He likes
everything from camembert cheese to unavored yogurt. Last year he stayed
in my house while visiting and . . .
a. sempre que eu servia pao, ele pedia queijo.
every time I served-IMP bread, he asked for-IMP cheese.
b. sempre que eu servi pao, ele pediu queijo.
every time I served-PRET bread, he asked-PRET for cheese.

The analysis of the selected language prompts brings up important ques-


tions. First, it is open to debate whether answer (a) for context 1 would not
be regarded as acceptable. Indeed, in another study, Salaberry and Martins
(2012) showed that native speakers consider both options acceptable. In fact,
this problem is acknowledged also by Rothman and Iverson in a footnote
(297), in which they point out that a reviewer claimed that s/he was not
convinced that answer (a) in the imperfect should be regarded as ungram-
matical. Rothman and Iverson concur with the reviewers concern sating
that [they] had some reservations as well. Nevertheless, what is puzzling
is that the native speakers in their study categorically favored answer (b)
over (a).22 How can we reconcile this contradiction? The only apparent
answer would be the potential eect of the distinct linguistic context that
accompanies each option.

22. Rothman and Iverson (2008: 297) underline that this is exactly what Menendez-
Benito claims. However, as discussed above, these data can also be reanalyzed
from a dierent theoretical perspective, in which case the data are not easily
accounted for as stated above.
206 M. Rafael Salaberry

Indeed, the context that precedes the target sentences provides linguistic
cues that may generate unintended artifacts. That is, options (a) and (b) in
the text prompts (1) and (2) above are dierentiated according to factors
other than the intended independent and dependent variables. In particu-
lar, I note two linguistic contrasts that may lead to inconsistent results: (a)
the type of adverbial, and (b) the recency of the time period. First, the use
of the adverbial sempre (always) is not as conducive to triggering the
notion of iterativity, as the use of the adverbial cada vez (each time)
would be. In fact, Menendez-Benito (2002: 374) explicitly makes this point
when she argues that the use of cada vez in sentence (23b) below sounds
more natural than sentence (23a) in which sempre replaces cada vez.
Menendez-Benito argues further that the adverbial cada vez in sentence
(23b) puts emphasis on the accidental character of perfective sentences.
(23) a. En la clase de ayer, siempre que un chico hizo una pregunta,
una chica le dio la respuesta.
In yesterdays class, whenever a boy asked (PRET) a question,
a girl gave (PRET) him the answer.
b. En la clase de ayer, cada vez que un chico hizo una pregunta,
una chica le dio la respuesta.
In yesterdays class, every time a boy asked (PRET) a question,
a girl gave (PRET) him the answer.
Second, the contexts that accompany the sentences that are expected to
be marked with the imperfect (non-accidental) consistently use adverbial
phrases rmly placed in a more remote past tense than is the case for the
ones that are expected to be marked with the preterite (accidental): no ano
passado (last year) versus na semana pasada (last week) respectively.
For instance, notice that in the following pairs of sentences, the selection
of the more remote past tense period in (24) favors the use of the Imperfect
in (24a). In contrast, the use of a more recent time period in (25) favors the
use of the preterite (25b).
(24) a. Cuando Julian era estudiante universitario, cada vez que le
preguntaban saba la respuesta.
When Julian was a university student, each time they asked
(IMP) him a question, he knew (IMP) the response.
b. ?*Cuando Julian era estudiante universitario, cada vez que le
preguntaron supo la respuesta.
When Julian was a university student, each time they asked
(PRET) him a question, he knew (PRET) the response.
Operationalizing hypotheses 207

(25) a. ?En la clase de ayer, cada vez que le preguntaban saba la


respuesta.
In yesterdays class, each time they asked (IMP) him a question,
he knew (IMP) the response.
b. En la clase de ayer, cada vez que le preguntaron supo la respuesta.
In yesterdays class, each time they asked (PRET) him a
question, he knew (PRET) the response.
Given that the study by Rothman and Iverson was framed within
Menendez-Benitos theoretical proposal, the inconsistent use of types of
adverbials raises important questions. That is, Menendez-Benitos argu-
ment is prompted by the very fact that, even when both the Preterite and
the Imperfect convey the notion of iterated eventualities, their distribution
varies in association with dierent types of adverbials (such as durative
and generic). A possible solution to this dilemma is to assess the eect of
adverbials on the acceptability of various uses of past tense markers not as
generic or durative, but rather as constructions (Michaelis 2004) embedded
in specic contextual frames, above and beyond the eect of morphological
markers and adverbials on their own.
Alternatively, it is possible to analyze the previous prompt sentences
from the perspective of Doiz-Bienzobas account (i.e., the anchoring of
events on the timeline). Thus, in context (25) above, answer (b) seems to
indicate that the specic times when the speaker bumped into Jimena
can be linked to non-generic specic times. In contrast, answer (a) in the
context of (25) signals that the request for cheese is not anchored to any
specic time.

5. Discussion

Thus far, I have described various denitions about the representation of


iterated eventualities. Furthermore, I have summarized and critiqued the
implementation of one particular proposal (Menendez-Benitos) as repre-
sented in the linguistic prompts used to assess judgments of grammaticality
of iterativity and habituality in one particular L2 empirical study (Rothman
and Iverson). The previous analysis shows the importance of assessing the
eect of adjuncts (e.g., adverbial phrases) and the ever-broadening dis-
cursive contexts (grounding, text type, world knowledge, etc.) on the over-
all composition of aspectual meanings. In this last section, I bring together
208 M. Rafael Salaberry

the analysis of the role of theoretical frameworks and the empirical imple-
mentation of the tenets of such theoretical models through research design
(e.g., the design of linguistic prompts) for the purpose of understanding
the process of theorization about the acquisition of aspectual knowledge.
The focus of this section is restricted to the analysis of three theoretical
conclusions that one may, arguably, draw from the previous analysis of se-
lected theoretical and empirical studies. First, the representation of aspec-
tual knowledge is distributed across several layers of information spanning
a broad discursive continuum. Second, among the most immediate layers
of meaning that have an eect on aspectual composition, the eect of
some frequently used adverbials (e.g., durante, during/for, siempre, always)
seems to be among the strongest and most obvious ones. Third, there
seems to be some default selections of morphological markings in associa-
tion with adverbial phrases to mark iterativity and habituality.

5.1. Layering of meaning


As early as 1973, Verkuyl explicitly stated that aspect is not a phenomenon
that can be restricted to the analysis of the lexical predicate alone, but one
that requires the computation of at least internal and external arguments,
and possibly more layers of information. For instance, how can we assess
the grammaticality/acceptability of the following pair of sentences?

(26) a. Lucas siempre supo la respuesta.


Lucas always knew (PRET) the answer.
b. Lucas siempre saba la respuesta.
Lucas always knew (IMP) the answer.

Clearly, as discussed in detail in previous sections, to make a decision we


need more contextual support than the one provided in the bare sentences
above. On the other hand, it would not be impossible for both native and
non-native speakers to make a decision if they were required to make one
(e.g., as required by an experimental study). That does not mean, never-
theless, that context would be left out of the assessment even if it is not
explicitly detailed in the sentences discussed. In such experimental condi-
tions, whatever response participants provide will be aected by the con-
texts they conjure up to make sense of such decontextualized sentences
(Coppieters 1987). The problem with this type of analysis is that we intro-
Operationalizing hypotheses 209

duce a variable (i.e., expanded context) that we cannot assess or control


because it is hidden and implicit. This is a potential problem that was
present in some of the studies reviewed above that used decontextualized
sentences as prompts (Perez-Leroux et al. 2007; Salaberry and Scholes to
appear; Slabakova and Montrul 2007).
Despite the fact that we cannot ascertain what contextual information
speakers bring to the task of assessing the grammaticality of decontextual-
ized utterances, the available evidence shows that native speakers seem
to be more attuned to the complex meanings of successive layers of con-
textual information than non-native speakers are (e.g., Coppieters 1987;
Garca and van Putte 1988; Guell 1998). Moreover, whereas in some cases
this sensitivity to broader pieces of discourse leads native speakers to
make categorical and homogeneous judgments (e.g., Coppieters 1987,
Garca and van Putte 1988), in other cases, it leads to uncertainty. For
instance, let us analyze one sentence (27) from Guell (1998) in which the
main predicate (atelic pasear, to take a stroll) has been iterated with an
adverbial (cada da, every day), which has been further circumscribed
within a specic time period (la semana pasada, last week).

(27) La semana pasada Antonio paseo/paseaba por el parque cada da.


Last week Antonio strolled (PRET/IMP) in the park every day.

Given the two choices of Preterite and Imperfect, non-native speakers


categorically selected the Imperfect. That is, they preferred the morpho-
logical marker that correlates most distinctly with the atelicity of the
verbal predicate (thus, conrming a narrow reading of aspectual interpre-
tation: they conrmed/maintained the meaning of the verbal predicate).
Native speakers, on the other hand, displayed great variation in their
responses splitting in half their preferences between the Preterite and
Imperfect. In other words, the consideration of three layers of aspectual
information triggered some uncertainty among native speakers about which
component(s) of the overall composition of the aspectual meaning of the
sentence to favor. That is, native speakers were aected in their responses
by the addition of a higher level of contextualization of the basic sentence.
How far down the discursive continuum can we go to assess the aspec-
tual value of a proposition? At what point have we broadened the context
to the point that world knowledge becomes the critical factor that deter-
mines whether a sentence is grammatical or not? Brisard (2010: 489), for
one, argues that aspectual contrasts may be more likely a pragmatics issue
rather than simply morphosyntax:
210 M. Rafael Salaberry

. . . much work in interpreting the concrete (temporal or modal) values of


this tense depends on pragmatic inferences on the basis of contextually pro-
vided information and is, as such, not to be attributed to the semantics of
the imparfait proper.

Against this theoretical claim, however, the empirical data show that native
speakers, as a group, seem to have distinct responses that dier from the
ones that non-native speakers select as a group (or various groups depend-
ing on prociency). Clearly, there is some type of knowledge that con-
sistently underlies the decisions that each population of speakers makes.
Ultimately, this knowledge is systematic, and that systematicity requires
some type of representation in the overall grammatical system of both
native and non-native speakers.

5.2. Constructions
As mentioned above, among the layers of meaning that have an eect on
aspectual composition, some frequently used adverbials seem to be strong
predictors of aspectual selection. For instance, in the following variation
of the sentence reviewed above, the durational adverbial (durante dos
anos, during two years) triggers the use of the perfective.

(28) Durante dos anos, fui a nadar todos los das.


During two years, I went (PRET) swimming every day.

Michaelis (2004: 7) argues that the adverbial in (28) coerces the basic
meaning of one time event associated with the use of the Preterite, shift-
ing it to several, many instantiations of the event. Thus, Michaelis states
that [c]oercion eects, rather than representing a special form of com-
position are by-products of the ordinary signications of constructions.
As a consequence, constructions can alter what words (and their syntac-
tic projections) designate (30). That is, the specic meanings of lexical
items (from the verbal predicate to the adverbial adjunct) cannot over-
come the combination of meanings of all lexical units above and beyond
each one of those meanings separately (cf. the construction). More spe-
cically, Michaelis proposes the Override Principle: If a lexical item is
semantically incompatible with its syntactic context, the meaning of the
lexical item conforms to the meaning of the structure in which it is
Operationalizing hypotheses 211

embedded (25). Thus, aspectual encoding is, by its very nature, an ad hoc
categorization of aspectual meanings.23
It is important to note that, within Michaelis model of constructions,
the eect of adjuncts (adverbials) on the aspectual interpretation of even-
tualities is not to be regarded as pragmatic, but rather as part of the gram-
matical representation of the language. That is, coerced meanings will
never contain anything that the rule of morphosyntactic combination does
not (30). In line with Michaelis proposal about constructions, we have
seen that Menendez-Benito (2002) shows that Spanish perfective and imper-
fective behave dierently with respect to generic adverbs (e.g., normalmente
normally, tpicamente typically) and durational phrases (e.g., durante dos
anos for two years). For instance, as argued by Menendez-Benito, generic
adverbs are more likely to combine with the imperfective, but not with the
perfective:
(29) a. ?El ano pasado, Lucas normalmente fue al cine.24
Last year, Lucas normally went (PRET) to the movie theater.
b. El ano pasado, Lucas normalmente iba al cine.
Last year, Lucas normally went (IMP) to the movie theater.
Note, however, that given the right contextualization, generic adverbs are
actually quite felicitous with the use of the perfective form in Spanish, as
shown in (29c).
(29) c. El ano pasado, Lucas normalmente fue al cine de la calle
Montevideo.
Last year, Lucas normally went (PRET) to the movie theater
of Montevideo Street.
A second problem with the argument about the incompatibility of the use
of generic adverbials with the Preterite is that not all generic adverbials
are the same. Some of these adverbials, such as normalmente normally,

23. For reasons of space, I will not discuss any specic type of constructions.
Michaelis discusses in detail constructions with frame adverbials and fre-
quency adverbials.
24. Although Menendez-Benito qualies sentences like (29a) as ungrammatical
with an asterisk, it is probably more accurate to use a question mark to signal
the apparent preference for the use of the Imperfect. That is, we cannot neces-
sarily reject the possible use of the version with the Preterite.
212 M. Rafael Salaberry

generalmente generally, usualmente usually, a veces sometimes, and de


vez en cuando from time to time, convey the meaning that some, but not
all, episodes actually happened. In contrast, other types of generic adver-
bials, such as repetidamente repeatedly, periodicamente periodically, and
constantemente constantly, convey the meaning that all episodes actually
happened. Sentence (29d) shows that the Preterite is indeed perfectly
acceptable with such adverbials.

(29) d. El ano pasado, Lucas fue repetidamente al cine.


Last year, Lucas repeatedly went (PRET) to the movie theater.

In this regard, according to Langackers denition, all episodes of the


described eventuality (i.e., going to the movie theater) are anchored in
time. Thus, the description of the eventuality is part of the actual plane. In
these cases, the use of the Preterite is likely to be acceptable, as in (29d), and
also by the following example borrowed from Perez Leroux et al.s (2007)
test:

(30) Los ninos se cambiaron de asiento repetidamente.


The children exchanged seats repeatedly.

Finally, and, in consonance with the argument made in the previous


section regarding the multilayering of aspectual representation, if the
second layer of adverbial phrases is conducive to the use of the perfective
form, the adverbial normalmente (normally) will not block the use of the
Preterite.

(31) Durante muchos meses, el cartero normalmente llego tarde.


For months, the mailman normally arrived (PRET) late.

The above discussion about the problems with the variability of aspec-
tual meanings brought about by the use of adverbials that are supposed to
convey a consistent interpretation (i.e., generic adverbials are incom-
patible with the Preterite) leads us to the analysis of correlated problems
with the design of linguistic prompts in acquisition studies. One such prob-
lem is the possible inuence of the selection of one particular adverbial as
opposed to another one. For instance, Menendez-Benito notes that the
adverbial sempre (always) is likely to trigger the use of the imperfective
(non-accidental or habitual depending on the chosen theoretical frame-
work of reference), whereas the use of the adverbial cada vez que (each
Operationalizing hypotheses 213

time) would trigger the use of the perfective (accidental or iterative).25


That is to say, we are compelled to understand what specic type of
knowledge is exhibited by native speakers to make the selections accord-
ing to the outcome correctly specied by Menendez-Benito.

5.3. Default meanings and constructions


The concept of constructions (Michaelis 2004) is not incompatible with the
notion that the morphological markers of tense-aspect in Spanish and
Portuguese have some basic, default meanings that become more nuanced
in the presence of various (additional) layers of contextual information
(i.e., the constructions). Let us review in more detail one example that
was discussed in previous sections herein reproduced as (32a).
(32) a. El verano pasado, Mara leyo/?lea el periodico.
Last summer, Mary read (PRET/IMP) the newspaper.
As previously described, in sentence (32a) reading the newspaper (leer el
periodico) is an atelic event. Thus, the association of the Imperfect (a
marker of background information) with an atelic event generates a proto-
typical match that highlights the function of backgrounding of the Imper-
fect. A background scene, however, requires the specication of another
event to be proled against that background. The absence of the other
event to be proled in the foreground makes the sentence unacceptable.
Indeed, through an analysis of past tense French, Brisard (2010: 507) pro-
poses that a scene-setting use of the imparfait creates certain expecta-
tions . . . because the notion of backgrounding that is involved only make
sense interactionally relative to a situation occurring in the foreground.
The following example provided by Brisard demonstrates the apparent
incomplete meaning of sentence (33a) given the lack of a foregrounded
element:
(33) a. Lannee derniere, je passais mes vacances en Suisse.
Last year, I spent my holidays in Switzerland.
b. Et quest-ce qui sest passe alors?
And what (has) happened then?

25. I should note, however, that these are tendencies, and thus, it is possible that
these selections are somewhat interchangeable.
214 M. Rafael Salaberry

Let us now analyze the eect of additional adverbial information on the


aspectual interpretation of the overall composition of the original utterance
in (32a).
(32) b. El verano pasado, Mara ?leyo/lea el periodico por la manana.
Last summer, Mary read (PRET/IMP) the newspaper in the
morning.
c. El verano pasado, Mara ?leyo/lea el periodico durante dos
horas por la manana.
Last summer, Mary read (PRET/IMP) the newspaper for two
hours in the morning.
Sentences (32b) and (32c) provide a more proper visualization of back-
ground information, thus making the use of the Imperfect more felicitous.26
In the case of (32b), the adverbial phrase further species the atelic nature
of the verbal predicate (homogeneous eventuality) making it less likely that
the Preterite could be used to make reference to anchored eventualities in
the past. Sentence (32c) shows that even the addition of the durational
adverbial is not enough to counteract the eect of the main adverbial that
determines that the eventuality is a homogeneous one, thus more properly
described (in the context of the overall construction) as a habitual.
Similarly, the unacceptability of the Preterite in (33a) is possibly related
to the fact that getting up (levantarse) is a telic event. Thus, the strong
tendency of the Preterite to convey punctuality (cf. default meaning) in
association with a telic event (prototypical match) cannot be easily over-
come with the addition of the durative adverbial (i.e., last year).
(33) a. Eva ?se levanto/se levantaba al medioda el verano pasado.
Eva got up (PRET/IMP) at noon last summer.
Once again, however, if we were to add an adverbial that signals the
notion of repetition more clearly (todos los das), then the Preterite is
placed within the bounds of a construction that shifts the predicates

26. The analysis of the (mis)use of the Preterite can be done along similar lines:
the Preterite, as a basic tense marker that conveys that specic instances of
reading happened in the past (cf. Doiz-Bienzobas 1995, 2002) is now incom-
patible with a mostly homogeneous, non-quantized event (cf. Michaelis 2004).
Operationalizing hypotheses 215

meaning from its default interpretaion of punctuality to the meaning of


iterativity; thus the sentence becomes more acceptable.

(33) b. Eva se levanto [todos los das] al medioda el verano pasado.


Last summer, Eva got up (PRET) at noon [every day].

In sum, we can argue that there are some invariant aspectual mean-
ings (Salaberry 2008) associated with the Spanish Preterite and Imperfect
that may trigger incompatibilities with constructions that would other-
wise convey consistent meanings. The Imperfect has a strong and invariant
meaning associated with the discursive function of backgrounding narrative
information. Thus, it may clash with the use of constructions that do not
allow to convey the notion of background information. Similarly, the
Preterite has a strong and invariant meaning associated with the function
of conveying episodic narrative information. Thus, its default meaning
may clash with the use of constructions that trigger habitual meanings as
opposed to iterative meanings.

6. Conclusion

The study of the acquisition and development of tense-aspect knowledge


among second language learners is notoriously complex. One of the most
important reasons for this complexity is the fact that tense-aspect knowl-
edge spans a range of domains including syntax, semantics, discourse, and
pragmatics. Not surprisingly, previous research on tense-aspect knowledge
provides contradictory ndings on central theoretical constructs (e.g., ulti-
mate attainment in tense-aspect).
Given the above claim, it is not surprising that even minor rearrange-
ments of information may trigger fundamental changes in the intepreta-
tion of aspectual meanings. The operationalization of the theoretical con-
cept of aspect and the methodological procedures to be used to assess
the broad meanings conveyed by aspectual markers presents a challenge.
However, despite the challenges posed by the multilayered nature of aspec-
tual knowledge, or perhaps precisely because of them, one can argue that
research on the acquisition and development of tense-aspect represents
one of the promising areas of L2 research to assess the validity of various
models and hypotheses of second language acquisition in general.
216 M. Rafael Salaberry

References

Bertinetto, Pier
1994 Stative, progressives, and habituals: Analogies and dierences.
Linguistics 32. 391423.
Bhat, D. N. Shankara
1999 The prominence of tense, aspect and mood. Amsterdam/Philadel-
phia: John Benjamins.
Binnick, Robert
1991 Time and the verb. Oxford: Blackwell.
Bonomi, Andrea
1997 Aspect, quantication and when-clauses in Italian. Linguistics
and Philosophy 17. 57108.
Brisard, Franc
2010 Aspects of virtuality in the meaning of the French imparfait.
Linguistics 48(2). 487524.
Comrie, Bernard
1976 Aspect. Cambridge: Cambridge University Press.
Coppieters, Rene
1987 Competence dierences between native and near-native speakers.
Language 63. 544573.
de Swart, Henriette
1998 Aspect shift and coercion. Natural Language and Linguistic
Theory 16(2). 347385.
Depraetre, Ilse
1995 On the necessity of distinguishing between (un)boundedness and
(a)telicity. Linguistics and Philosophy 18. 119.
Doiz-Bienzobas, Aintzane
1995 The preterite and the imperfect in Spanish: Past situation versus
past viewpoint. San Diego, CA: University of California-San
Diego, Ph.D. dissertation.
Doiz-Bienzobas, Aintzane
2002 The preterite and the imperfect as grounding predications. In
F. Brisard (Ed.), Grounding: The epistemic footing of deixis and
reference, 299347. Berlin: Mouton de Gruyter.
Filip, Hannah
1999 Aspect, eventuality types and nominal reference. New York:
Garland Publishing.
Garca, Elvira & Florimon van Putte
1988 The value of contrast: Contrasting the value of strategies. IRAL
26. 263281.
Goodman, Nelson
1947 The problem of counterfactual conditionals. Journal of Philosophy
12. 607653.
Operationalizing hypotheses 217

Guell, Lourdes
1998 La adquisicion del tiempo verbal en el aprendizaje del espanol
como lengua extranjera. Barcelona, Spain: Universitat de Barce-
lona, Ph.D. dissertation.
Iverson, Michael & Jason Rothman
2008 Adverbial quantication and perfective/imperfective interpretive
nuances in L2 Portuguese. In R. Slabakova, J. Rothman, P.
Kempchinsky & E. Gavruseva (Eds.), Proceedings of the 9th
Generative Approaches to Second Language Acquisition Con-
ference (GASLA 2007), 7080. Somerville, MA: Cascadilla Pro-
ceedings Project.
Langacker, Ronald
1987 Foundations of cognitive grammar, Vol. 1: Theoretical perspectives.
Stanford, California: Standford University Press.
Langacker, Ronald
2000 Grammar and conceptualization. Berlin: Mouton de Gruyter.
Langacker, Ronald
2009 Investigations in Cognitive Grammar. Berlin/New York: Mouton
de Gruyter.
Lenci, Alessandro & Pier Marco Bertinetto
2000 Aspects, adverbs and events: Habituality versus perfectivity. In
J. Higginbotham, F. Pianesi & A. Varzi (Eds.), Speaking of
events, 245287. New York: Oxford University Press.
Menendez-Benito, Paula
2002 Aspect and adverbial quantication in Spanish. Paper presented at
the Proceedings of the 32nd North Eastern Linguistics Society,
Amherst, MA.
Michaelis, Laura
2004 Type shifting in Construction Grammar: An integrated approach
to aspectual coercion. Cognitive Linguistics 15. 167.
Perez-Leroux, Ana, Alejandro Cuza, Monika Majzlanova & Jeanette Sanchez
Naranjo
2007 Non-native recognition of the iterative and habitual meanings of
Spanish preterite and imperfect tenses. In J. Liceras, H. Zobl &
H. Goodluck (Eds.), Formal features in second language acquisi-
tion, 432451. Amsterdam: John Benjamins.
Rothman, Jason & Michael Iverson
2008 Beyond morphological use: What semantic knowledge tells us
about aspect in L2 Portuguese. Language Acquisition 15(4). 270
314.
Salaberry, M. Rafael
1999 The development of past tense verbal morphology in classroom
L2 Spanish. Applied Linguistics 20(2). 151178.
Salaberry, M. Rafael
2003 Tense aspect in verbal morphology. Hispania 3(86). 559573.
218 M. Rafael Salaberry

Salaberry, M. Rafael
2008 Marking past tense in second language acquisition: A theoretical
model. London: Continuum Press.
Salaberry, M. Rafael
2012 Contrasting preterite and imperfect use among advanced L2
learners: Judgments of iterated eventualities in Spanish. Interna-
tional Review of Applied Linguistics.
Salaberry, M. Rafael & Custodio Martins
2012 Dierential distribution of Spanish Preterite and Imperfect with
iterated eventualities among native speakers. In E. Labeau &
Qiaochao Zhang (Eds.), Cahiers Chronos. Amsterdam/New York:
Rodopi.
Salaberry, M. Rafael & Joe Scholes
to appear The eects of instruction on the development of knowledege
about iterativity in L2 Spanish.
Slabakova, Roumyana & Silvina Montrul
2007 L2 acquisition at the grammar-discourse interface: Aspectual
shifts in L2 Spanish. In J. Liceras, H. Zobl & H. Goodluck
(Eds.), Formal features in second language acquisition, 452483.
Mahwah, New Jersey: Lawrence Erlbaum.
Smith, Carlota
1997 The parameter of aspect. Dordrecht: Kluwer Academic Press.
Chapter 7
Research design: From text to task

Kathleen Bardovi-Harlig

1. Introduction

In the study of second language (L2) tense and aspect as in other areas
of research the best task is the task that best addresses the research ques-
tion. In the absence of a specic research question, it is dicult to weigh
the costs and benets of particular approaches because the costs that a
researcher is willing to incur depend heavily on what benets are to be
gained in the context of a given problem. Nevertheless, in the spirit of
addressing the most general question of how we study development in L2
tense-aspect systems, this chapter discusses the use of open-ended tasks.
Open-ended tasks are designed to encourage learners and native speakers
to produce the kind of language that they produce naturally, that is, to tell
stories and recount events, to describe and to argue, to predict and hypoth-
esize, and to have conversations. Compared to other elicitation tasks, re-
searchers exert only a light touch on language production, guiding learners
subtly to the construction of texts. Because dierent types of texts have
dierent tense-aspect proles, understanding the dierence among text types
and what they can contribute to tense-aspect research is a necessary rst
step to determining method. To that end, this chapter is organized by
types of spoken and written discourse.
Open-ended tasks can be contrasted with controlled production tasks
such as cloze passages and other tasks (see Bardovi-Harlig 2000, for a
discussion of the range of tasks used in tense-aspect research). As in other
areas of investigation in second language acquisition (SLA) research, the
study of communicative language may need to be supplemented by planned
elicitation to assure the balance and contrast of verb types, tenses, gram-
matical aspect, lexical aspect, and adverbials (to name a few linguistic
devices used in the expression of temporality; see Bardovi-Harlig 2000,
2005 for a discussion of the distribution of aspectual categories in dierent
types of texts and tasks). After reviewing general considerations in SLA
research that favor the selection of communicative production (Section
220 Kathleen Bardovi-Harlig

1.1), task characteristics (1.2), and talk and text (1.3), this chapter considers
conversations as dyadic discourse (Section 2) and extended monologic dis-
course. Narratives are discussed in Section 3 and nonnarratives in Section
4. Nonnarrative texts are further divided into description, argument, and
hypothetical or future-oriented texts. Each section ends with specic elici-
tation tasks that have been used successfully in the literature.

1.1. Communicative production tasks in SLA


The decision to employ open-ended tasks in the study of the development
of L2 temporal expression stems from the interest in observing inter-
language at work in the act of communication. Communication gives
learners an opportunity to build their contributions in speech or writing
from the linguistic devices that they have available. The use of world
knowledge, text structure, lexicon, lexical temporal expressions, and verbal
morphology come together in rich and varied discourse that reects the
voices of individual learners and their current level of development. Com-
municative texts provide fertile grounds for both hypothesis-testing and
functional approaches and for form-oriented and meaning-oriented studies
(Bardovi-Harlig 2000, 2007). In addition, temporal reference is pervasive
(unlike some other linguistic constructs) and in some languages obligatorily
marked in every nite clause making it a particularly good candidate for
investigation by relatively open-ended production. However, not every
form-meaning association is found in every text, and that is where a famil-
iarity with text type proves invaluable to the researcher.
In addition, research in SLA more broadly suggests that the greater the
focus on communication, the less likely the learner is to employ explicit
knowledge (Dulay, Burt, and Krashen 1982; Ellis 2008). Whereas not all
facets of language are explicitly taught, tense-aspect morphology is generally
a high priority of grammar-based instruction, and most classroom learners
are likely to have had exposure to explicit instruction. Avoiding learned
knowledge is a relevant factor in L2 task construction, and one which the
use of communicative tasks directly addresses. Moreover, open-ended tasks
are also well-suited to longitudinal studies over a given period of time.
Because learners construct their own text, such tasks avoid the problem
of learners learning the test as they might with various cloze passages or
other item-focused tasks which have target responses. In addition, topics
which allow learner creativity and avoid repetition increase learners interest
in participating in a study.
Text to task 221

1.2. Task characteristics


The conditions under which L2 production data are collected are not
unique to the study of temporality, but they are worth reviewing here.
Language samples may be oral or written; they are also increasingly
computer-mediated. Oral production can be either monologic (produced
by one speaker) or dyadic or multi-partied (produced by two or more
speakers). Written production is generally monologic, although computer-
mediated communication may result in more cases of written dyadic
communication. Oral communication is the most likely to be spontaneous,
but is not exclusively unplanned. Whereas turns in conversation can be
completely spontaneous, personal narratives, for example, can be repeated
across conversations (see, for example, the concept of discourse domains
which learners perfect through multiple tellings; Douglas and Selinker
1994; Selinker and Douglas 1988). The degree of comparability is a result
of research design. The more similar the task completed by all learners,
the more comparable the resulting language samples. Beyond spontaneous
conversations, tasks may range from an interviewer asking every learner
the same question to showing a lm for the purposes of eliciting a lm
retell, but since text construction is up to the informant, comparability in
open-ended tasks is always relative and will never achieve the level of
comparability that results from highly controlled tasks.

1.3. Talk and text


This chapter divides dyadic and multi-partied discourse, or conversation,
from extended monologic discourse (Berman and Verhoeven 2002), which
includes both narrative and nonnarrative discourse.1 Relevant characteristics
of texts include time frame, person, personal viewpoints, and whether they
occur in a real world or not. Of primary concern in the study of temporality
is the time frame of the text. For example, narratives are past-oriented
whereas conversations and nonnarrative texts may be past-oriented, but
may just as likely be present- or future-oriented. Person determines use of
verbal morphology, and among the texts considered here, conversation
is the only likely source of second person usage (in addition to rst and
third), whereas narratives and nonnarrative are not. In contrast, the variable

1. Although Smith (2003) explicitly excludes conversation from her discourse


modes, conversation (whether spontaneous or elicited through interviews) has
been a very important source of unmonitored expression both in SLA and
sociolinguistics and is included here.
222 Kathleen Bardovi-Harlig

personal refers to the speakers relationship to the text. A personal text


draws on a learners experience (and may be in rst or third person),
whereas as impersonal texts do not, and instead respond to an external
stimulus from the researcher. A personalized text is a hybrid, and is dis-
cussed in Section 3.3. Realis/irrealis refers to whether the text reports
events that are real in their discourse world. Irrealis refers to unrealized
events, including hypothetical, future, and imagined events in both actual
and ctional worlds. What links inquiry into dierent discourse types, and
studies of L2 tense and aspect, is the observation that the dominant tem-
poral relation is determined by the discourse type (Caenepeel 1995; Smith
2003, 2005; Ragnarsdottir et al. 2002, for developmental L1 research; von
Stutterheim 1991, for L2).
The following sections explore conversation and extended monologic
texts (Berman and Verhoeven 2002). Within extended monologic texts,
narrative and nonnarrative texts are discussed. Nonnarrative texts are
further divided into description, argument, and irrealis texts. Before em-
barking on a discussion of these categories, it is important to note that
any type of text may contain another type of text. Text, like other linguistic
structures, is recursive and exhibits textual embedding. A narrative may
appear within a narrative, a narrative within an argument or a descrip-
tion, a description within a narrative, or a narrative in a conversation.
For example, the narrative in (1) opens with a description (Bardovi-Harlig
1992a); the narrative begins with the rst event Punias mom said labeled
[1]. (Oral features of language such as lexical searches and repetitions have
been deleted for illustration purposes).
(1) In Kohala there were ten sharks. And then nobody can go to the
beach. So they just eat potato and some food. But, the lobsters were
in the water and nobody can reach it.
[1] One day, Punias mom said I wish that I have a lobster with
my potato
[2] And when Punia heard that, he decided to go to the beach
[3] He went there
[4] He throw a rock in the sea. . . .
Texts have main structures and side structures (or foreground and
background) and these too may be recursive. Main structures attend to
the main function of the text, and side structures allow for elaboration
which results in a richer language sample. Thus, rather than considering
a text to be exclusively of one type or another we will consider a text to
Text to task 223

be predominately one type. The presence of passages within a text that


contrast with the main function of the text provides additional contexts
for the study of how tense-aspect interacts with text structure. The follow-
ing sections are organized by major category of discourse. A general deni-
tion is provided, variations are considered, and specic means of elicitation
are discussed; these are followed by practical considerations that might be
useful for researchers planning to elicit the type of text under consideration.

2. Conversation and conversational interviews

Conversation constitutes the most familiar form of authentic unplanned


discourse. Conversation is oral, interactive, and either dyadic or multi-
partied. In conversation, speakers take turns, introduce and maintain
topics, negotiate meaning, and attend to face concerns. Participants gener-
ally have a social connection that leads them to talk in the rst place and
those relationships are further enhanced by talk. Conversations exhibit a
variety of local temporal frames in a single encounter and constitute a
rich source of data. Because they are unique, spontaneous conversations
are generally low on comparability for specic linguistic contexts or ideas,
although they are somewhat more comparable in turn-taking and other
discourse features not of immediate concern to temporal reference.
In tense-aspect research, conversational data generally derive from
conversational interviews (e.g., Bardovi-Harlig 2000; Dietrich, Noyau, and
Klein 1995; Giacalone Ramat 2002; Kihlstedt 2002; Salsbury 2000). When
learners and interviewers get to know each other well, as in the case of a
longitudinal study, the interviews can be increasingly conversational with
almost equal exchanges of turns where either interlocutor may initiate a
topic. However, the expressed purpose of many conversational interviews
is not to produce equal turns, but rather to elicit extended language
samples from the learner-participant. Conversational interviews generally
give learners fewer opportunities to negotiate topics than conversations.
Nevertheless, interlocutors also often encourage learners to say more
than they would on their own, thus creating more contexts for temporal
reference.
Conversations can exhibit a full range of temporal frames at the level of
individual topics. Although corpus studies characterize conversations as
exhibiting nonpast verb forms (e.g., Biber 1989), that is a general charac-
terization that does not preclude the use of past (as will become obvious in
the case of conversational narratives discussed in the following section). A
range of temporal frames occur spontaneously in conversation, especially
224 Kathleen Bardovi-Harlig

when initiated by a learner. However, even in conversational interviews


there are both planned and unplanned questions. Because no single time
frame is associated with conversation it should be a good source of adver-
bials and other explicit time framing devices. These shifts also provide
contexts for a range of tense-aspect morphology (see also Wiberg 2002),
and as a result the literature reects the use of conversational interviews
to study a range of tense-aspect forms including future (Bardovi-Harlig
2004a; Howard 2012; Wiberg 2002), pluperfect (Bardovi-Harlig 1994;
Howard 2005), past (Bardovi-Harlig 2000; Bayley 1994; Howard 2009;
Kihlstedt 2002; inter alia), and present perfect (Bardovi-Harlig 1997a, 1997b).
The exibility of conversations in allowing for a variety of temporal
frames is illustrated in the following excerpts in which learners initiate a
new temporal frame. In Example (2) the interviewer follows the learners
use of tense-aspect. In (3), when the interviewer supplied a frame which
did not match what the learner wanted to say, he supplied his own, last
weekend as an example.2 (In the transcripts of oral production, slashes
show overlapping speech).
(2) Present perfect (Bardovi-Harlig 1997b; L1 Arabic)
I (Interviewer): . . .Thank you for doing that.
L (Learner): Youre welcome. So, where have you been, Kathleen?
I havent see \\ \\ you.
I: \\Well, since Ive. . .\\ seen you, I went to,
California.
(3) Past reference (Bardovi-Harlig 1992b, L1 Arabic)
I: Do you guys, do you n your friends go someplace on the weekends?
MA: Yeah, last weekend I went to Ohio, King Island.
One concern of assessing underlying competence from dyadic interactions
is the contribution of the interlocutor through modeling or scaolding.
Wiberg (2002) briey discussed this concern when describing how she
prompted learners in order to elicit additional future-oriented responses
when their contributions either strayed from the established time frame

2. Unattributed examples come from my longitudinal data, consisting of 1,576


written texts and 175 oral texts, from 16 learners of English as a second lan-
guage. Examples that were previously published or that come from other
sources are cited.
Text to task 225

Table 1. Usage patterns across speaker turns and likely knowledge sources
Usage pattern across speaker turns Interpretation
The learner uses the target tense-aspect The learner demonstrates the tense-
form independently (no use by the aspect form-meaning association
interviewer, no prompting)
The learner uses the target tense-aspect The learner knows the tense-aspect
form after the interviewer or a prompt form and usage well enough to accom-
modate, but does not demonstrate
independent use.
The learner does not use the target Learner demonstrates comprehension
tense-aspect form in an environment and in some cases co-occurrence
which the interviewer creates, but restrictions
instead uses the appropriate adverbials
in a short answer
The learner does not use the target Two interpretations are possible: the
tense-aspect form in an environment learner under-uses the tense-aspect
which the interviewer creates, but uses form or the interlocutor has set up a
an alternative verb form time-frame that the learner overrides
The learner does not use the targeted The learner demonstrates lack of
tense-aspect form in an environment comprehension of the form-meaning
which the interviewer creates and correspondence
furthermore indicates noncomprehen-
sion of the form.

or when their responses were very short. She also noted that this standard
interview procedure was also used with native speakers. Bardovi-Harlig
(1997b) explored the relation of interlocutor turns to learner production
in conversational interviews. Learner production of a tense-aspect form
can be evaluated vis-a-vis interlocutor turns in at least ve ways (Table 1):
independent production by the learner (Examples 23), repetition (Exam-
ples 45), use of relevant adverbials (Example 6), use of an alternative
form (Example 7), and nonuse with noncomprehension (Example 8). Ex-
amples (46) and (8) are from Bardovi-Harlig (1997b).
(4) Repetition [SA, L1 Arabic]
L: Its uh, I think small city.
I: From what I hear. // // I mean I dont, Ive never been.
L: //Yeah.//
L: Me too. Ive never been there.
226 Kathleen Bardovi-Harlig

(5) Apparent Repetition [TZ, L1 Arabic]


I: Have you uh seen magic shows, is that, is that uh common?
L: In Saudi Arabia? no, but Ive seen others. Other countries,
like Egypt.
(6) Relevant adverbials [WS, L1 Korean]
I: Mm-hm. Thats good. How long have you been here in
Bloomington?
L: Oh, maybe since March.
I: Thats right. Cuz you started in level 2, right?
(7) Alternative form [LU, L1 Spanish, Bardovi-Harlig 2004b]
I: What are you going to do for Christmas Break?
L: I will go back to my country on December 8.
(8) Nonuse [TO, L1 Japanese]
I: Have you traveled anywhere else in the United States?
L: Travel?
I: Uh-huh. Have you been to any other cities in the United States?
L: Uh, have you been? Uh. . .
I: Have you visited other places in the United States?
L: No.
I: No? Only Bloomington?
L: Ah no no no no. St. Louis.
Of the ve outcomes, the most challenging for analysts to interpret is
imitation, but one thing to keep in mind is that repetition of morphology
and syntax requires a certain level of grammatical development. Learners
cannot repeat what they have not already produced or are on the verge of
being able to produce (for L1 acquisition, see Dale 1975; for a review of
elicited imitation in L2 acquisition, see Bley-Vroman and Chaudron 1994).
Moreover, the learners language sample can be checked for independent
use in other contexts or, if a longitudinal study, in earlier or later texts.
Such a check revealed that the learner in (4) had used only one present
perfect successfully in the oral sample and three widely spaced written
tokens up to that point, suggesting the likelihood of a repetition. In con-
trast, the learner in (5) had shown many previous correct and indepen-
dent uses of present perfect in both oral and written samples prior to the
Text to task 227

conversation from which the example was excerpted, making repetition


without understanding the form-meaning association unlikely. Further-
more, the use of the will-future in response to the going to future used by
the interviewer in Example (7) shows that comprehension of the framing is
not sucient for repetition; the go-future had not been used by the learner
at the time of the example.
Even though interlocutors may suggest temporal frames, learners may
also override the context set up by the interviewer. In Example (9), the
interviewer asked her question as though the learners musical career in
his country had ended (using the simple past). The learners reply (with
the present perfect) indicates that he intends to play in an orchestra in his
home country after graduation (and this accords with what he reported
elsewhere in his longitudinal data).
(9) Present perfect override (LU, L1 Spanish, Bardovi-Harlig 1997a)
L: . . . . And also I like to play recital in marimba, like a solo . . .
player.
I: Did you uh, give any concerts . . . in Puerto Rico?
L: \\yeah: I,\\ I, . . . I have played abou-:: six concerts in Puerto Rico
in dierent part of my country. And the last: abou- two weeks
ago, I play at one concert
I: mm-hmm
L: at university too
The widespread use of conversational interviews for both learners and
native speakers speaks to its value as a source for temporal expression.
Means of collecting conversation and implementing conversational inter-
views for research on temporal reference are discussed next.

2.1. Eliciting conversation


Conversation may be elicited in various settings. Learners could record
their target language encounters resulting in truly spontaneous conversa-
tions. Classroom activities can be recorded, and depending on the time
frames or tense-aspect forms being investigated may include pair and
group work, discussions, problem-solving activities, and event planning
such as class picnics, skits, service activities, or other class events. Conver-
sations during co-curricular activities such as language tables, conversation-
partner meetings, or other target language opportunities may also be
recorded.
228 Kathleen Bardovi-Harlig

The recording of truly spontaneous authentic conversation is always


somewhat of a challenge, which is why researchers often resort to conver-
sational interviews. Conversational interviews are the most common type
of elicited conversations in L2 tense-aspect research, although they fre-
quently lack the turn taking characteristics of unplanned conversations,
since the goal of the interviewer is to elicit uninterrupted language samples
from the informant. A number of things can be done to increase conversa-
tional features in conversational interview data. Researchers can get to
know their participants, suggest topics of interest to learners in order to
promote learner participation, and let learners take the lead on topic nom-
ination when possible. Conversational interviews with two learners and a
single interviewer (Bardovi-Harlig and Salsbury 2004; Bayley 1994) maxi-
mize learner participation and allow the interviewer to play a lesser role.
Conversational interviews generally explore a number of topics and last
for about an hour (but some are shorter, see for example, Kihlstedt 2002).
Howards (2005, 2012) interviews of Irish learners of French included
topics such as hobbies and pastimes, holidays, university studies, employ-
ment, Franco-Irish relations, as well as Labovs danger of death and
premonitions modules. Shirais (2002) interviews of Chinese learners of
Japanese covered daily activities, past experience in China and Japan,
and future plans. Clementss (2003) interview of a Chinese learner of
Spanish asked the informant about her birth place, education and profes-
sional life in China, her family, reasons for emigrating, likes and dislikes
of Spanish society, food, professional life in Spain, and her future in
Spain. The choice of topics was controlled by the interviewers in order to
include topics about the present, past, and future.

2.2. Practical considerations


Conversations and conversational interviews allow the greatest exibility
in tailoring a task to an individual. Interlocutors or interviewers can follow
up on answers by learners and ask for clarication or elaboration. In
conversations, negotiation of meaning also allows analysts to observe
understanding in action. Moreover, a variety of topics can probe a range
of time frames. Conversations and conversational interviews can be used
with speakers at all literacy levels. Training interviewers (including our-
selves) to allow longer silences than they would normally encourages
more talk from learners; asking wh-questions rather than Yes/No ques-
tions increases turn-length.
Text to task 229

3. Extended monologic discourse: narrative

In contrast to conversation which is characterized by turn-taking and thus


co-construction of texts by two or more speakers, extended monologic
texts (Berman and Verhoeven 2002) are constructed by a single speaker or
writer. This section follows the binary division into narrative and nonnarra-
tive discourse (Georgakopoulou and Goutsos 2000). Within nonnarrative
discourse, description, argument, and irrealis texts are considered in terms
of their contribution to understanding the development and use of tense-
aspect by L2 learners.
As noted by Georgakopoulou and Goutsos (2000), the study of narra-
tive and nonnarrative discourse has often been separated, with the study
of nonnarrative discourse lagging behind. In contrast, they observe, Narra-
tive analysis, by contrast, is one of the best and most extensively researched
areas of multi-disciplinary study of discourse (117). The study of narra-
tive whether viewed methodologically as a vehicle for tense-aspect or as
structured by tense-aspect benets from rich traditions of analysis (see
also Chapters 9 and 11). In contrast, as Georgakopoulou and Goutsos
(2000) point out, there is no uniformly accepted denition of nonnarrative
(except in contrast to narrative, as its name suggests) nor is there a generally
accepted further division of nonnarrative into text types (see Georgako-
poulou and Goutsos 2000, for a history of analysis of nonnarrative, and
Smith 2003, for discourse mode, which is beyond the scope of this chapter).
In his article A Typology of English Texts, Biber (1998) uses ve sets of
binary features to distinguish texts: involved versus informational produc-
tion; narrative versus, nonnarrative concerns; explicit versus. situational
dependent references; overt expressions of persuasion; and abstract versus.
nonabstract. In her book Modes of Discourse, Smith (2003) distinguishes
ve discourse modes that commonly appear in written texts: narrative,
description, report, information, and argument.3
In addition to the division of narrative and nonnarrative on the basis of
discourse analysis (see, for example, Biber 1998; Georgakopoulou and
Goutsos 2000), there are also acquisitional distinctions. Narrative struc-
ture is acquired relatively early by children (Berman and Slobin 1994;
Berman and Verhoeven 2002) and requires no specic formal education.
In contrast, some nonnarrative types, such as argument and expository
texts, are mastered much later and may be inuenced by formal education

3. The ve domains are not exhaustive; conversation and procedural texts are
excluded (Smith 2005: 223237).
230 Kathleen Bardovi-Harlig

as suggested by the multi-national, cross-linguistic, and cross-sectional


study of the production of text reported in two thematic volumes of Written
Language and Literacy 55(12). Berman and Verhoeven (2002: 6) reported
the following:

even the youngest children, the 910 year old grade-schoolers, make a clear
distinction between the two types of texts [narrative and expository] in con-
tent and linguistic means of expressions. . . . But the mastery of expository
text production in the sense of a coherent set of core propositions, express-
ing key ideas elaborated by relevant illustrative and motivational material
is not manifested until much later, usually only in high school.4

These considerations contribute to the fact that narratives are the more fre-
quently studied type of extended monologic text in tense-aspect research.

3.1. Narrative discourse


A narrative is a text in which the speaker relates a series of real or ctive
events in the order in which they took place (Dahl 1984: 116). Narrative
has also been dened as one method of recapitulating past experience
by matching a verbal sequence of clauses to the sequence of events which
actually occurred (Labov and Waletzky 1967: 20). Narratives are past ori-
ented by denition.5 Smith (2003: 19) characterizes narrative as dynamic
and located in time. Narratives are typically monologic, they can be oral
or written, may be personal, impersonal, or personalized, rst or third
person, and can occur spontaneously in conversation or be elicited. They
exhibit main and side sequences. Narratives can be highly individual and
have low comparability or they can relate the same or similar events for
greater comparability. In addition, producing a narrative is something that
every speaker can do; it is not culturally specic or learned in school; and it
can be done by very low level learners as well as advanced learners, with
skilled narrators in both groups.
The chronological order of events in a narrative provides a temporal
anchor against which learners use of temporal expression can be evaluated.
Narratives have internal structure, often referred to as foreground and back-
ground. The foreground relates events belonging to the skeletal structure
of the discourse (Hopper 1979) and consists of clauses which move time

4. For a crosslinguistic developmental study that includes even younger children,


see Berman and Slobin (1994).
5. Other texts such as recipes, instructions, and play-by-play sports commentaries
are also organized by chronological order, but do not constitute narratives
according to Dahls denition.
Text to task 231

forward (Dry 1981, 1983). The temporal point of reference of any one event
in the foreground is understood as following that of the event preceding
it. So important is the concept of sequentiality that foreground clauses may
be dened by the interpretation of their order: if a change in the order of
the two clauses results in a change in the interpretation of what actually
happened, then those two clauses are narrative [i.e., foreground] clauses
(Schirin 1981: 47; see also Labov 1972; Labov and Waletzky 1967; Fleisch-
man 1985; Hopper and Thompson 1980; Reinhart 1984; von Stutterheim
1991) Narrative foreground will carry perfective past.6 Narrative excerpts
(10) and (11) show a sequence of events reported as foreground in narra-
tives by learners of English as a second language (Bardovi-Harlig 1995).
The numbers in the square brackets represent the order of the foregrounded
events. (Learner spelling in written narratives has been preserved.)7
(10) Oral Modern Times narrative excerpt (L1 Japanese)

Foreground Background
[1] Then she, stole the bread.
[2] And the- she ran away
[3] and she . . . hit the Chaplin . . .
I: Mm-hm. S: . . . and uh, dierent woman
saw the, she stole the, uh,
[4] . . . and uh, and cried, bread . . .
[5] and chased her, I: Mm-hm.
[6] and the employer caught her,
[7] and, but Chaplin said,
I did it, you know, I . . . stole . . . the breads.

(11) Written Modern Times narrative excerpt (L1 Arabic)

Foreground Background
[1] she stol abread
[2] and they cutch her
[3] she met charlie the nice man who was trying to helpe her
[4] when he said
she didnt sleel the bread I did that

6. For a discussion of tense switching in narrative see Schirin (1981).


7. For a detailed discussion of grounding, see Chapter 9 and Bardovi-Harlig (1995,
1998, 2000).
232 Kathleen Bardovi-Harlig

All narratives have a foreground. Depending on the purpose of the


narrative, the skill and level of engagement of the speaker (or writer), and
the context in which the narrative is told, many narratives also have a
background. In narrative excerpts (10) and (11) there is one background
clause each. The background of a narrative supports its foreground. In
contrast to the single function of the foreground to move time forward,
the background may support the foreground in a variety of ways. Crucial
to research on temporality, the background has no set time frame. For
example, a background clause may reveal a prior event (located before
the narrated event on the time line), make a prediction about the outcome
of an event (located after the event on the time line), or refer to a simulta-
neous event (located at the same point or interval on the time line), for
example, where the stealing of the bread and the viewing of it coincide
as in Excerpt 10. The background also sets the scene (called orientation)
as in (12) and (13) which in turn often includes description, and provides
evaluation as in (12), or explanation/identication as in (13) (Bardovi-
Harlig 1995). Scene setting promotes the use of stative predicates and
imperfect.

(12) Oral, L1 Spanish

Foreground Background
Orientation And then, in another situation, in
another place, there is a old lady that
is really hungry is starving- She is
starving. She is starving and so a- a
business of bread, loaf of bread, [. . .]
[. . .] to support it.
Foreground [1] And she decided to
hope to take out the
the loaf of bread.
Evaluation And although this was a wrong thing.
Text to task 233

(13) Written, L1 Arabic

Foreground Background
Orientation This movie talk about
the man and the women.
One day the women was
too hungre,
Action [1] so she stole a big piece of bread
[2] but when she was ranning
[3] she hit the man
Identication who is Charlie Chaplin.
Although the man
Identication who is a Baker
[4] he said to the policeman that
there is women who stole his bread,
[5] so the policman go after the women.

From a storytelling perspective, the background enhances the narrative.


From the perspective of research on temporal expression, the background
provides contexts in which learners use a variety of temporal devices includ-
ing adverbials, lexical devices, and a range of tense-aspect morphology
making it the ideal context for imperfective verb forms, including progres-
sive (Bardovi-Harlig 2005). This is further enhanced because departures
from chronological order are generally indicated explicitly (Bardovi-Harlig
1992b, 1994; Klein 1986).
On the one hand, all narratives share the same basic structure, and on
the other, there are characteristics of narratives that are amplied in spe-
cic contexts of production. Because these characteristics are relevant to
the study of L2 tense-aspect, the following sections consider conversational
and elicited narratives and within elicited narratives, personal, impersonal,
and personalized narratives.
3.2. Conversational narratives
Narratives occur naturally in conversation. They may be the main focus of
a conversation as when a speaker says, Guess what happened at school
today! or the result of a topic like getting speeding tickets in which
friends all contribute a story from their own experience. Just as likely, a
narrative may occur in conversation to illustrate a point. Investigating
situated narratives reveals how they are used in spontaneous communica-
tion. Like all narratives, a conversational narrative has a single primary
speaker. Conversation accommodates narratives by suspending the typical
turn-taking dynamics, and allowing the narrator to have an extended turn.
234 Kathleen Bardovi-Harlig

As an illustration, consider the narrative in (14) told by a male learner


to a male interviewer with whom he had developed a rapport. Fahad, the
learner-narrator of this narrative, uses the narrative to illustrate his claim
that not having a car would allow him to study, but that owning a car
leads to distraction, and when he sees it he thinks Where are we going?
Before considering the narrative, rst consider a point of geography: At
the time of the narrative, the learner lived in Bloomington, Indiana, 218
miles to the southeast of Bloomington, Illinois. The two are often confused
even by American travelers, and faculty members who are scheduled to give
a talk at Indiana University have been known to call from the Bloomington,
Illinois airport asking for directions to Bloomington, Indiana.
The excerpt opens with the conversational exchange that frames the
narrative which begins on line 20.
(14) Lost in America, Conversational narrative (Excerpt from Fahad, L1
Arabic)
1 S: Yeah, but I, I bought the car for [ol] reason because
2 I, my apartments far \\ \\ from here.
3 I: \\uh\\
4 I: Mmm hmm
5 S: But, I dont like to buy a car and I dont like to have
6 a car, cause the car will, will make me like I dont like to study
7 and I want go just around. ((laugh))
8 I: Mmm hmm, where all, where have you gone in
9 your car? You went to Cincinnati and Indianapolis, where
10 else?
11 S: yeah, I ((laugh)), Uh, just [xxx], I mean, now I
12 cant I cant go a freeway, if I have
13 I: mm hm
14 S: car, but if if I dont have car just I can study and I
15 dont care about anything. Yeah but but now, if you can just I
16 think Where are we going,
17 I: \\oh you have\\
18 S: Cincinnati, San Louis or Chicago, [a]-on freeway,
19 and like for three weeks ago,
20 [1] I went with my friend to s-uh to San Louis,
21 [2] and, we uh we slept there, and when we uh
22 when we back from there,
23 [3] I said to him Do you want us go to Chicago
Text to task 235

24 [4] he said uh OK,


25 [5] I said OK, and
26 [6] then that way we found uh Bloomington. [I:
27 ((mued laugh))]
28 Uh, I mean, I dont know I-uh, we-we-we dont
29 know, we didnt know [I: mmm hmm]
30 this is for I dont know,
31 [7] we-we said this is Bloomington.
32 [8] And he said, Do you want us go to change our
33 clothes, and uh continue the way?
34 [9] I said him OK.
35 [10] And we entered the, we entered the,
36 Bloomington, Illinois,
37 [11] and there, we found everything is changed!
38 ((laugh)) [I: ((laugh))] One, ((laugh)),
39 [12] I said uh One one day, the one day changed
40 everything. ((laugh)) [I: ((laugh))]
41 [13] Yeah and I I ask-ed the someone there,
42 the gas station, I asked I said to him Excuse me, do
43 you know where the nineteenth street is?
44 [14] he said I dont have a nineteenth street.
45 [15] I said Oh, when did they take that ((laugh))
46 nineteenth street?
47 [16] Yeah uh, and we just we enter from here
48 and we, uh we [I: ((laugh))] exit I mean we uh go from
49 uh I mean uh back,
50 and we we lost, we [ ] lost in the in America
51 and just go around we dont know where are going.
52 [17] And I ask uh someone about the way
53 [18] he said Take uh fty-ve south.
54 [19] And we took uh fty s-uh fty-ve south,
55 [20] and the fty-uh fty-ve uh south took
56 me to Indianapolis. And from Indianapolis to
57 Bloomingtons,
58 we we [sta] we, from I mean in the, I I
59 start to mm from San Louis three oclock,
60 I I came here to Bloomington one
61 oclock in that night. [coda]
62 S: Yeah too much, just driving
236 Kathleen Bardovi-Harlig

63 I: whew
64 S: and ((laugh)) too much too much just . . .
65 I: ((laugh)). Yeah you do have to be careful in, you
66 cant spend all your time in your
67 S: yeah ((laugh))
68 I: car and then, no time in your books.

The speaker depends heavily on reported speech to convey the story


line in his narrative; 10 of the 20 foreground clauses use reported speech
[35, 79, 12, 14, 15, and 19]. The narrative is also largely unelaborated in
that it consists essentially of the main story line. The few background clauses
provide an interpretation of the foreground event we found Bloomington in
line 26 [6], observing we didnt know in line 29 (namely, we didnt know
that we were in the wrong Bloomington) and the resultant state is sum-
marized in line 50 We lost . . . in America. In line 51 (and just go around
we dont know where are going) the background provides orienting
action; orienting action sets the scene but does not move narrative time
forward (Labov 1982; Bardovi-Harlig 1992a). The narrative ends tradi-
tionally in a coda; a coda brings a narrative back to the present (Labov
1982).8 Fahad oers a coda which brings the narrative back to the present
place, I start from San Louis three oclock, I came here to Bloomington
one oclock in that night and sums up with too much driving. The
minimal backgrounding in this narrative may be due to its function of
illustrating a point rather than the telling of the story itself.
The interviewers rst turn following the narrative responds to the main
point made by the speaker, namely that if he has a car, he cannot study. In
fact, the introductory sentence by the narrator and the response by the
interviewer could appear contiguously in sequential turns as in (15).
(15) (modied sequence)
F: But, I dont like to buy a car and I dont like to have a car, cause
the car will, will make me like I dont like to study and I want go
just around.
I: Yeah you do have to be careful in, you cant spend all your time
in your car and then, no time in your books.

8. See also Bardovi-Harlig (1992a) for analysis of L2 narratives in the Labov


(1982) framework.
Text to task 237

The learner is obviously comfortable speaking, and the learner and inter-
viewer have developed a rapport during his time in the English program.
The learner nominates topics and requires very little prompting from his
interlocutor. The conversation, and the narratives that occur in it, is a rea-
sonable length to provide a usable language sample.

3.2.1. Collecting conversational narratives


Conversational narratives occur in conversations and are not elicited per
se, but rather collected. Any means discussed for eliciting conversation
may also lead to conversational narratives (see Section 2.1).
There is no direct written correlate for conversation. However, there is
a source for spontaneous written production of narratives in the form of
journals. Like conversations, journals contain a variety of text types, includ-
ing personal narratives. Bardovi-Harlig (1992b, 1994, 1997a, 2000) made
extensive use of ungraded free-production daily journals which learners
kept throughout the longitudinal study. Many entries at the lower levels
received reciprocal entries from teachers in the form of dialog journals.
Some students made six or seven entries a week while others made only
two or three entries. The benet of ungraded, unpressured written format
manifests in longer more elaborate text than conversation which can be
scaolded. This is illustrated in the comparison of (16) from conversa-
tional interviews and (17) from journal entries by the same learners writ-
ing about the same trips.
(16) Conversational interviews
a. I: So we came to class last week and you were gone. Your
parents came?
YJ: We went to trip.
b. I: So tell me what you did this weekend.
HK: Oh, I went to Chicago.
(17) Journal entries
a. YJ: I left Bloomington to Boston on May 19 because I have to
visit my cousin. (YJ, T5)
b. HK: I got up at 5:00 am and we drove to Chicago this morning.
(HK, T5.5)
When the same learners write, they must set the scene since a reader
cannot contribute to the construction of the text in the same way as an
238 Kathleen Bardovi-Harlig

interlocutor. Thus as writers, learners must fully specify the scene, includ-
ing temporal reference; this is seem in the way in which YJ and HK set the
scene when telling about their respective trips in their journals.

3.2.2. Practical considerations


Like conversation and conversational interviews more generally, narra-
tives found in conversation can be highly individual and tuned to specic
learners. In conversation, speakers are often motivated to relate a narra-
tive in order to make a point. For learners who are never at a loss for
words (and some of the learners in my longitudinal study fell into that
category) conversational interviews lead to extended narratives. For others
who are less inclined to talk, however, such open-ended tasks provide little
structure. In addition, even among elaborated personal narratives there
is little comparability. For example, the Lost in America narrative in
Example (14) is one of a kind. No directly comparable narratives were
produced by other learners in the study (Bardovi-Harlig 1992b, 1994,
1997, 2000). Because SLA research values language samples that can be
compared, elicited narratives are considered next as a means of increasing
comparability.

3.3. Elicited narratives


To address the issue of comparability, narrative research often elicits sets
of narratives from multiple speakers or writers. Native-speaker and learner
narratives alike have been elicited for study of discourse structure and
temporal expression. Elicited narratives may be personal narratives with
content determined by the speaker in response to a prompt or question;
they may also be impersonal with topic and potential content determined
by the researcher, and they may be oral or written. Whatever the specic
conditions of production, the speaker ultimately controls the content and
structure of the narrative. The following sections consider and compare
three types of elicited narratives: personal, impersonal, and personalized
narratives.

3.4. Personal narratives


Although the most genuine personal narratives arise spontaneously in
conversation, personal narratives can also be elicited by asking speakers
to respond to a question or prompt. Personal narratives, whether elicited
Text to task 239

or spontaneous, draw on a speakers experiences and may be told in rst


or third person, and they may be oral or written. Some of the best-known
sets of personal narratives include the danger of death stories collected by
Labov (1972; Labov and Waletzky 1967) and harassment stories collected
by Tannen (1983). In L2 tense-aspect research, Bayley (1994) collected
accounts of the uprising in Tiananmen Square in June 1989 and experiences
during the Bay Area earthquake of October 1989; other topics included
experiences in host and home countries (Shirai 2002), memories of the rst
few months in the host country (Lopez-Ortega 2000), and travel narratives
(Du and Li 2002). Liskin-Gasparro (1996) recorded a single learners
account of being robbed in a taxi in Spain and harassed by a group of
young men later that same evening; the learner told the story twice, once
in each of two oral prociency interviews. Excerpt (18) presents the narra-
tive of the robbery from the advanced-level interview.

(18) Taxi robbery excerpt (Advanced level, Liskin-Gasparro 1996: 275)

Line Foreground Background Translation


1 [1] Regrese a mi casa I went home alone, but in
sola, pero en un taxi a taxi
2 alguien me haba someone had told me
dicho
3 que en un taxi es that in a cab youre really
bastante seguro safe
((laugh))
4 [2] fui I left
5 [3] cog un taxi I caught a cab
6 [4] y al llegar a mi and when I got home,
casa el me pido he asked me for more
mucho mas dinero money
7 que haba puesto en than what was on the
la maquina machine
8 ?como se llama? how do you say it?
9 que cuenta el dinero that counts the money
10 [5] y yo pregunte por and I asked why
que
11 no debera haber I shouldnt have done
hecho ((laugh)) esto ((laugh)) that
240 Kathleen Bardovi-Harlig

12 pero s but yes


13 [6] y el se enfado and he got mad
14 [7] y agarro mi bolsa and he grabbed my purse
15 [8] y el tomo todo el and he took all the money
dinero all in it
16 [9] y empezo a conducir and he started to drive
otra vez again
17 [10] y yo todava estaba and I was still
[I: all] [I: there]
18 [11] entonces yo agarre so then I grabbed my purse
mi bolsa
19 [12] salte del coche I jumped out of the car
20 que estaba en marcha that was moving
21 y bueno, me asusto and well, he really scared
bastante me [evaluation]

The example illustrates the features of a personal narrative including


background clauses that interpret the situation in lines 23, provide an
evaluation of the wisdom of the action reported in lines 1112, and a
description of the moving car followed by an evaluation in lines 2021.
The background clauses provide a context for the past perfect, haba dicho
had told and the modal perfect no debera haber hecho I should not
have done that which are tense-aspect forms that do not appear in the
foreground. Personal narratives are recognized for their rich background
as seen in Example (18) and discussed more fully in Section 3.3.3.

3.4.1. Eliciting personal narratives


Personal narratives have typically been elicited by asking a general ques-
tion, although many of the studies cited earlier do not give the specic
question used to elicit the narratives. To elicit the danger of death stories
Labov (1972) asked, Were you ever in a situation where you thought you
were in serious danger of getting killed? Tannen (1983) reported asking
whether New Yorkers had had any experiences in the subway.
Elicited narratives are generally the focus of the interview in which they
occur and an extended turn is guaranteed. (Even though conversation
naturally suspends normal turn-taking during a narrative the speaker
Text to task 241

may have a sense of not taking too long to tell a story.) On the other
hand, conversational motivation for telling a story is removed, and a new
one has to be invented and conveyed in the course of the data elicitation.
Sucient rapport has to be established for the potential narrator to want
to tell the narrative. In some cases, L2 data collection benets from learners
wanting to say as much as possible in the second language and being able
to do so without fear of losing their turn at talk.
Although most of the personal narratives collected in L2 tense-aspect
research have been oral, written narratives are valuable as well. Written
narratives are more likely to be planned than oral narratives, but time
constraints on the writing tasks can limit the amount of explicit knowledge
brought to bear on narrative production. Writing may also reduce anxiety
concerning oral skills and mediate nontargetlike pronunciation. Written
and oral narratives may often be elicited by the same techniques.
Questions asked of learners should be formulated with some sensitivity
to what they might be willing to talk about. Not everyone wants to tell a
stranger about their most embarrassing moment, for example. If it was
embarrassing in the rst place, it will be embarrassing in the retelling.
Topics should be chosen carefully with an eye toward what students are
likely to have experienced. For example, Jourdenais (1998) asked students
about their initial experiences during their rst job, during their rst few
days of college, and as they began to learn Spanish. Choosing topics care-
fully is even more important when learners and researchers have no rela-
tionship beyond the single elicitation event.
One way around the issue of determining the topic of a personal narra-
tive (so much of a persons life may not be the business of a researcher,
after all) is to avoid a specic selection, and instead to let the learner
choose from a short list of topics. The use of emotion cards (rst used by
Collier, Kuiken, and Enzle 1982, and Davitz 1969) were introduced to
L2 research by Rintell (1989) to elicit language expressing emotion and
adopted for tense-mood-aspect elicitation by Salsbury (2000) and Comajoan
(2001). Following Rintell, learners are presented with a set of emotion cards,
with one emotion written on each card (e.g., happy, sad, angry) and asked
to select an emotion that they are willing to talk about. They are then
prompted to tell about a time that you felt sad. The advantage is that
the stimulus is very minimal and that the emotion is selected by the
learner. This lessens the comparability somewhat in that learners do not
all respond to the same prompt, but they may ultimately produce longer
narratives because they chose the topic. Salsbury used longer prompts as
242 Kathleen Bardovi-Harlig

the learners in his longitudinal study gained in prociency, but maintained


learner topic-selection by presenting the questions on individual cards.
Another means of enhancing elicitation of personal narratives is to pro-
vide an example of an event or situation in the form of pictures or lm and
then ask learners whether they have had a similar experience (Bardovi-
Harlig, longitudinal corpus 1994, 1997a, 2000; Berman and Verhoeven
2002). Bardovi-Harlig used picture sequences which included going to a
circus and trying out the tricks at home, playing pranks at school, and
being helped by a stranger to get speakers to start talking and to introduce
topics prior to asking them for their own similar experiences with a
prompt such as Did you ever do that. . . . Tell me about it. Berman and
Verhoeven showed a lm depicting dierent types of conict before their
elicitation of oral and written texts on the same topic.
It is interesting to note that not all questions work as prompts. Berman
and Verhoeven (2002) reported that students from some countries would
not discuss interpersonal conict and instead were given a substitute ques-
tion about a time when someone helped them (34, footnote 10.) Tannens
(1983) harassment stories began with the goal of collecting narratives
about experiences on the New York subway, but ended up as narratives
about molestation as the theme recurred across narratives. When she
collected a comparable set of narratives from Greek women she asked
directly if they had had any experiences being molested.

3.4.2. Practical considerations


Some of the practical considerations were mentioned above. In addition
to being sensitive about topics, interviewers should also be prepared with
alternative topics (as were Berman and Verhoeven 2002). If interviewers
are working in a team, it helps if they can confer between interviews and
make similar adjustments to keep the interviews aligned. Individual inter-
viewers can keep notes to increase similarities in their approaches to dier-
ent speakers. Shared cultural events, weather-related experiences, or air-
port travel are topics that are often fruitful. Even with elicited narratives
on the same general theme, there is still an element of the unknown for the
researcher. For example, personal narratives may be comparable only
thematically, but not in detail, and analysts do not have an external map
of the order of events. Both of these issues are addressed through the use
of impersonal narratives.
Text to task 243

3.5. Impersonal narratives


Among the open-ended tasks, those eliciting impersonal narratives dominate
L2 tense-aspect research. They are favored for their high comparability
and the degree of researcher control over the stimulus. The impersonal
narratives used in L2 research are ctional (but do not have to be) and
always third person. Examples (10)(13) are impersonal narratives from
Modern Times retell tasks.

3.5.1. Eliciting impersonal narratives


Impersonal narratives can be elicited easily by means of lms or picture
retell tasks. The lm or picture retell tasks get their name from the process
by which it is elicited. Participants are presented with a story on lm or in
a series of stationary pictures and are asked to tell the story. Sequences of
events are presented as well as visual details (varying with the media used),
but the participant is free to add to or reduce the detail of the story.
Silent lms or excerpts have been used because the lms are intended
to provide a stimulus, but not a verbal model. Films that have been used
to elicit narratives include the Pear Stories produced for the purpose of
narrative elicitation (Chafe 1980), Modern Times (Chaplin 1936; dierent
segments used by Bardovi-Harlig 1995, 1998; Dietrich, Klein, and Noyau
1995; also Bergstrom 1995, 1997; Comajoan 2001, 2005a,b, 2006; Hasbun
1995; Salaberry 1998), Tin Toy (Lasseter 1995; used by Bardovi-Harlig 1999),
The Sorcerers Apprentice (Algar, Pearce, and Fallberg 1940; used by Laord
1996), Quest (Montgomery 1996; used by Carroll, Rossdeutscher, Lambert,
and von Stutterheim 2008) and others. Schmiedtovas (2004) use of eleven
short television commercials (1955 seconds long) were carefully selected
for their portrayal of simultaneous actions and ordered events in a study
of the L2 acquisition of Czech. Leclercq (2009) used ve of Schmiedtovas
commercials to study simultaneity in English.
In addition, stationary pictures such as those found in comic strips
(Comajoan 2006) and wordless picture books such as Frog, Where are
you? (Mercer 1969) have been used by Bamberg and Marchman (1990),
Berman and Slobin (1994) and Housen, Kemps, and Pierrerd (2009). Word-
less picture books can be modied to emphasize actions (such as modifying
an action of ying up to ying up, up, up; Burghardt 2010, forthcoming)
and can be projected via computer.
Narratives can also be retold from a live, oral performance of story or
folktale (Bardovi-Harlig 1992a used Punia and the King of the Sharks, a
Hawaiian trickster tale, from MacDonald 1986). Some learners nd a
244 Kathleen Bardovi-Harlig

task based exclusively on listening in the L2 to be too demanding (only 16


of 22 learners were able to complete the task used by Bardovi-Harlig
1992a). In addition, an oral presentation provides a model for temporal
reference, which adds another variable. Interestingly, even with a verbal
model, some learners elaborated on the text in innovative ways.
Procedures dier on whether the learners tell the narrative to a re-
searcher or associate who has seen the lm (Bardovi-Harlig 1995), a fellow
student who has not seen the lm (Salaberry 1998) or self-recording with
no interlocutor (Labeau 2009). Written impersonal narratives do not have
an obvious reader, but Labeau had learners email her their written narra-
tives within an hour of completing the oral picture retell task, thus estab-
lishing a known reader. Practice also varies on whether the learners view
the lm once or twice. Viewing the lm twice is thought to lessen anxiety
and reduce memory load.
As with personal narratives, most studies collect oral retells. Some
studies have collected both oral and written narratives (Bardovi-Harlig
1995; Labeau 2009); only one study (Clachar 2005) seems to have collected
only written narratives.
The emphasis on retell tasks comes from the freshness of the details and
the action. It has been suggested that lm retells are superior to the use of
classic stories or culturally specic folktales or myths which tend to lack
details in accounts by learners who do not know them well or have not
heard them for a while. However, folktales and myths can be used as con-
tent for retell tasks as Comajoan (2001, 2005a,b) did with an illustrated
book of the tale of Saint George.

3.5.2. Practical considerations


Because stimuli used to elicit impersonal narratives are selected by the
researcher, the order of the events is known in advance, an advantage
especially when working with language samples from lower level learners.
Researchers have the opportunity to select stimuli that elicits features that
they are interested in. Pictures and even lm can be relatively easily modi-
ed, for example, to increase the number of simultaneous events, or augment
an action (from going up to going up, up, up, Burghardt 2010). Narrative
retell tasks have the potential of equalizing the length of narratives by
giving speakers a certain amount of content to work with. This was em-
phasized by the narratives collected by Du and Li (2002) who found
that the elicited Pear Story narratives were longer than personal travel
narratives told by the same learners of Mandarin (whereas the personal
narratives by native speakers were longer than their Pear Stories).
Text to task 245

The use of still pictures to elicit narrative has to be done with some care
to have the learners rst read the pictures, after which the researcher
removes the pictures and the learners tell the story. This increases the
focus on the action. Learners have a tendency to describe a picture that is
in front of them. This can be turned into an advantage for the elicitation
of descriptions discussed later.

3.6. Comparing personal and impersonal narratives


Noyau (1984, 1990) argued for the superiority of personal over impersonal
narratives citing structure and motivation as two main benets. She claimed
that lm retell tasks provide an inherent structure whereas personal narra-
tives oer greater potential for observing how learners manage temporal
reference on their own. More importantly, Noyau (1984) claims that the
motivation of the speaker for sharing his own experience gives maximal
expression of his repertoire (115) leading to greater elaboration of the
background. In support of this argument, native speaker personal narratives
show greater backgrounding than foregrounding (Schirin 1981), as do
personal narratives by learners (Comajoan 2001, 2005b; Liskin-Gasparro
2000; Trevise 1987). Trevise (1987) noted that one of her narrators ex-
planations in the narrative about how he escaped from Argentina is so
important to both the narrator and narrative that it seems hardly less
signicant than the action.
In addition to inuencing the depth of the background, whether a
narrative is personal or impersonal, elicited or conversational may also
inuence the rate of suppliance for both perfective and imperfective mor-
phology. Personal and conversational narratives show higher rates of use
of verbal morphology than impersonal narratives (Comajoan 2001, 2005b
for Catalan; Giacalone Ramat and Ban 1990 for Italian; Liskin-Gasparro
2000 for Spanish).
Although lm and story retell tasks can elicit rich background (see
Bardovi-Harlig 1992a, 1995), especially with visually rich animation (Laord
1996), learners can comply with a researchers request to tell what hap-
pened by relating the foreground alone. In any retell task, however, there
may be pressure that leads learners to sacrice background for foreground
(Tomlin 1984).

3.7. Personalized narratives


The use of personalized narratives attempts to capitalize on the benets of
personal narratives (greater detail, more background, and increased use
246 Kathleen Bardovi-Harlig

of verbal morphology) while maintaining control over content and com-


parability. In the retell task participants take the role of a protagonist in
a lm or story and tell the story from the rst person perspective. Learners
show dierent use of tense-aspect morphology in personalized narratives
and greater elaboration of background compared to impersonal narratives.
Moreover, personalized narratives look ahead, interpret, evaluate, and
give reasons, features that Noyau (1984, 1990) claims to be present only
in personal narratives. In addition, some learners also show a range of
tense-aspect forms not seen in the impersonal narratives. Bardovi-Harlig
(1999) used a computer-animated lm, Tin Toy (Lasseter 1989) to explore
personalized narratives. In Tin Toy, a baby terrorizes his toys by picking
them up, chewing on them, drooling on them, and then smashing them
down on the oor. Only the tin toy, a mechanical one-man band, dares
to oer himself to the baby to play with. After learners viewed the lm,
they were asked rst to retell the story. These instructions produced im-
personal narratives. Next, the learners were asked to take the part of one
of the protagonists, the heroic mechanical toy or the toy-smashing baby.
The personalized narratives provided background information which
was absent relative to the same reported foreground action in the im-
personal narration, including adding motivation of the character and emo-
tions; in addition, the narrators oered interpretations of the foreground
events. In Excerpt (19) line 7 the learner narrating the impersonal account
reports that the toys are hiding, and in the personalized narrative in (20)
line 7, he expands on why the toys are hiding, and in line 9 what the toys
are thinking (they are surprised) following foreground predicate he started
to cry.

(19) Impersonal narrative (L1 Arabic)


[5] the musician man run away from the baby.
[6] And, uh . . . the, the baby
was . . . uh, running after the musician man,
[7] till the . . . the musician man hide uh, under the couch or
something. And theres all of the toys who were /hide/, hiding
from the baby
[8] And after that the baby fell down
[9] and he started to cry.
/Nnn/ . . .
[10] After that the musician . . .
Text to task 247

(20) Personalized narrative: narrator as the baby


[5] The musician man just run away from me
[6] and I follow [him?].
And I was so happy. And I was eager to catch it.
[7] Uh, then he just uh, he just hide from me with the . . . other
toys which they were, they, they were afraid of me.
[8] And uh, I fall [fell?] down, eh,
[9] then I started to cry. I was crying, and all of the toys
[surprised?]: Whats happening?
[10] There is a musician man get out . . .
In addition, in (21) we hear the voice of the Tin Toy who feels sorry for
the baby (compare this to predicate [9] in (19) and (20)).
(21) Narrator as the tin toy
Foreground Background
[17] and he, he was crying, started to cry.
Uh, then uh, his voice was very loud, and I just felt sorry
for him

3.7.1. Eliciting personalized narratives


To elicit personalized narratives, learners are told to imagine that they are
one of the characters from a story. For example, in the case of the Tin Toy
narratives excerpted above, learners were told imagine that you are this
toy . . . and tell me the story . . . and tell me what you remember about
what happened to you. or imagine that youre the baby. Okay? And
tell the story as if you were the baby. Okay? One interviewer added,
What happened to you today, Mr. Toy? (Bardovi-Harlig, longitudinal
corpus 2000). Note that the instructions were followed by several com-
prehension checks (Okay?) by the interviewers.

3.7.2. Practical considerations


Eliciting personalized narratives requires a stimulus with a sympathetic
protagonist. Most of the narratives used in L2 elicitations could be used
in this way. This turned out to be a conceptually dicult task which
some participants declined to do citing the fact that babies cannot talk or
think. Because of this, it seems to be important to make sure that the pro-
tagonists are higher up on the animacy hierarchy (Hill 1988; Silverstein
1976) than babies or toys. (This also suggests that animals might be prob-
248 Kathleen Bardovi-Harlig

lematic main characters for personalized narratives. See Carroll, Ross-


deutscher, Lambert, and von Stutterheim 2008, for a discussion of charac-
teristics of protagonists.) Learners who completed the task seemed to have
fun with it and, more importantly, they showed noticeable increases in back-
grounding in the rst person narratives over the third person narratives.

3.8. Summary: working with narratives


Regardless of mode, whether elicited or spontaneous, personal or impersonal,
narratives allow researchers to document learners use of tense-aspect in a
past-time chronologically ordered framework. The sequence of chronolog-
ically ordered events in the foreground establishes a timeline against which
the background is understood: The foreground will exhibit perfective past
(simple past in English). In contrast, anterior, simultaneous, and future
events and situations in the background all provide contexts for the use
of diverse tense-aspect morphology. These factors combine to create a
rich environment for the study of temporal expression. Oral narratives
can be told by speakers of any age or education level; moreover, well-
developed narratives are not the exclusive domain of the procient speaker.
By denition, narrative discourse privileges chronological sequencing as it
answers the question what happened next? The following section con-
siders texts with other organization and goals.

4. Extended monologic discourse: nonnarrative texts

As previously noted, nonnarrative discourse often has been dened in


opposition to narrative in terms of what it is not, rather than what it is.
Following Georgakopoulou and Goutsos (2000: 124), nonnarrative dis-
course (a) establishes a generic truth through analysis, synthesis, or argu-
mentation; (b) contains detailed information about structure or process;
(c) establishes a position that needs to be supported; and (d) realizes a
variety of speech acts.
In the same way that we can use narrative to study how time moves
forward and the means that speakers use to convey it, we can use other
types of text to study how time stands still and how learners convey that.
Nonnarratives exhibit use of imperfective past, future, and modals (Bardovi-
Harlig 2005). No single type of nonnarrative text has received as much
attention linguistically as the narrative, especially in the tense-aspect litera-
ture. This section considers description, argument, and irrealis as three
examples of nonnarrative text which have been investigated in L2 tense-
Text to task 249

aspect research. (Although not part of classic studies of discourse, I have


added irrealis here because eliciting irrealis texts require special elicitation
tasks.)
In the construction of the narrative, chronological action is emphasized
against all else. In contrast, in descriptions, states are highlighted and action
is relegated to the explanatory (von Stutterheim and Klein 1989; von
Stutterheim 1991). Von Stutterheim and Klein argue that narrative struc-
ture, grounding, and description can be understood in a more general
approach to main structures and side structures. In a narrative, the fore-
ground answers the question What happened next? from which unbounded
states, habituals, and generics are excluded. In contrast, a description answers
questions such as, What did/does X look like? (von Stutterheim and
Klein 1989) or What was it like there? (Carroll and von Stutterheim
2003). In descriptions, according to von Stutterheim (1991: 391), specic
temporal reference is normally excluded, and temporal location on the
time axis leads to side structures, exactly the reverse picture from the
narrative.9 In a narrative, temporal reference is moved forward by bounded
events; in a description, unbounded states imply the maintenance of the
temporal frame.
Comparisons among text types have been made by Bronckart and
Bourdin (1993) for young students ages 10, 12, and 14, by Ragnarsdottir
et al. (2002) for students at dierent academic levels, ages 910, 1213,
1617, and university graduate students, and Smith (2003) for published
texts. Bronckart and Bourdin (1993) compared use of morphology in a
narrative, a report, a letter and an explanation of a process by students
at age 10, 12, and 14. These comparisons show that not only do texts
dier from each other, but tense-aspect distribution may also dier across
languages (see also Carroll and von Stutterheim 2003, on this issue.)10 It
should also be noted, however, that tasks can be manipulated to encourage
one tense over another, as discussed in Section 4.1.1. A comparison of
temporal and modal features in narrative, descriptions, and arguments in
written L2 English showed that learner descriptions and arguments were
characterized by nonpast time and greater use of modals than narratives,
making these text types ideal means of investigating nonpast and modality
(Ewert 2006). In addition, descriptions showed more stative verbs than

9. Smiths (2003) analysis is a bit dierent. For Smith, descriptions are located in
time, but static. Movement of time would trigger a side sequence.
10. An examination of learner texts crosslinguistically is beyond the scope of this
chapter.
250 Kathleen Bardovi-Harlig

narratives or arguments making them potentially good contexts for imper-


fective morphology.

4.1. Description
Smith (2003: 20) characterizes descriptions as being located in time (which
is static) focusing on specic objects, people, mental states, and ongoing
events. Descriptions may be oral or written and may occur as part of the
background of a narrative as in (1) and (12) or an independent text as in
(22) and (23).
Independent descriptive texts have the potential of being longer than
descriptions found in narrative background (in learner texts) and may be
more fully developed (the dierence in oral-written modes may be a con-
found here). In Excerpt (22) from an essay describing the qualities of a
good leader (Ewert 2006: 198), the use of present and modal can maintain
the time frame. Original learner spelling is maintained.
(22) Description essay (Ewert 2006: 198) (In the excerpts that follow,
means paragraph.)
I think
karizma [charisma] is one of the most important quality in being a
leader.
Secondly, I think
that the good leader can persuade other people who against their
bill or idea.
Without any violence, the leader can change the opposive sides
thinking.
By speech, Lincon or Washinton can move the public.
This ability is the necessity of good leader.
which can persuade other people
(23) Description essay with narrative side structure (Ewert 2008: 20)
First, they have the constant will or passion.
In their life time, they were attacked so many times,
but no one can defeat them,
Because they attracted the people,
so the people was alway their side.
Nelson Mandela was arrested
Text to task 251

and put into jail for eighteen years.


His children was killed.
However, he never gave up.
He had the dream to overcome the human discrimination and
establish the freedom.
So was Martin Luther King.
After eighteen year ghting, he got the freedom
and became a president of South Africa Republic.
I was much moved by their life.
The true leaders aect us not only at that times but also even now.

Excerpt (23) shows a description with a narrative side structure (indi-


cated by italics) in which the statives, modals, and habituals (indicated by
adverbials) in the description contrast with the narrative.
There are two main ways to elicit descriptions: to enhance background
in narratives and to elicit independent descriptions. One means of enhanc-
ing description is to employ the personal narrative for reasons discussed
earlier. (See, for example, Liskin-Gasparro 1996.) Another means is to
manipulate a stimulus to promote description. Visually interesting lms
such as the Sorcerers Apprentice used by Laord (1996) enhance descrip-
tion as would documentary lms such as Babies (Billot et al. 2010) whose
settings are perhaps the main point of the lm. Following von Sutterheim
and Klein (1989) and Carroll and von Stutterheim (2003), tasks for elicit-
ing description should compel participants to answer questions such as
What is/was X like? and What was it like there?
Increasing description by increasing narrative background is a natural
outcome of using static pictures. Static pictures can be problematic for
eliciting narratives (because they promote description) and this makes
them ideal when the goal is to elicit description. Labeaus (2009) use of
nine static pictures in a wordless cartoon set in prehistoric times promoted
the use of background description in a study of the French imparfait
(imperfect). Comajoan (2005a) employed a richly illustrated picture book
about the legend of Saint George and found greater background with the
static pictures than with a lm retell task. Kim (2012) also used visually
detailed pictures to elicit descriptions. Using a blind between the researcher
and the learner so that only the learner saw the picture, Kim encouraged
learners to produce longer oral samples by asking, I would like to know
more about the lady. Please tell me about her clothes and appearance.
252 Kathleen Bardovi-Harlig

Independent descriptive texts may be elicited in a variety of ways. The


prompt used with emotion cards (Rintell 1989) can be varied from elicit-
ing narratives to eliciting description. After the learner identies the emo-
tion he or she would like to talk about from a stack of cards, the researchers
uses the prompt tell me about things that make you sad/happy/anxious
(Salsbury 2000). The focus away from events to things shifts the focus of
the text. One of the challenges of working with descriptions is nding topics
that generate descriptions of sucient length to constitute a reasonable
sample, and allowing learners to choose their topics (as discussed earlier)
helps with this.
Another means of eliciting description is to make it the explicit focus
of the prompt by naming the text type as in DescriptionQualities of a
Good Leader: Describe the qualities (characteristics) of a good leader.
Use specic details and examples in your description (Ewert 2006). Such
prompts are common for eliciting written texts in academic language pro-
grams. If a description is used to investigate the emergence of past imper-
fective forms, then there is methodological challenge of maintaining the
focus on past time reference. Descriptive prompts which have a past time
reference were employed by Jourdenais (1998) in which learners were
asked to describe their initial impressions during their rst jobs, during
their rst few days of college, and as they began to learn Spanish.
An additional way to elicit description may be to have learners emulate
newspaper reports, a text characterized by the central role of description
(Caenepeel 1995). Because the immediate situational context of the reader
and the reported events do not coincide, the writer must convey facts
about the setting and circumstances in which the newsworthy events have
occurred, resulting in a descriptive focus. Extensive clips of news footage
could be shown to learners who then write their own newspaper reports.
This task would work best with participants who have had experience with
newspapers or on-line news reporting. An oral version might ask learners
to simulate a live broadcast for radio or television.

4.2. Argument
Arguments put forward a claim, comment, or position and support it in
some way. Smith (2003) describes arguments as atemporal. Arguments
are a typical academic text type and unlike the other text types considered
in this chapter are primarily written, although the arguing of positions can
also be found in conversation, and formal oral debate. Ewert (2006, 2008)
elicited written texts and compared arguments to description and narra-
Text to task 253

tive. So far, the use of arguments has only been minimally exploited in L2
tense-aspect research.
Example (24) shows an excerpt from an argument essay that takes the
stand that technology enhances the quality of education.

(24) Excerpt from argument essay (Ewert 2006: 98)


Also, making students better to live current world is important.
It means
that students must be used to using technology such as computers.
Because, no one can live without such kinds of technology in
practical world.
For example, almost all companies use and organize their works by
computers.
If someone cant use computers,
he/she will not achieve.
So, using technology helps also the students technology experience.
In conclusion, teaching must change with pracical world.
(25) Argument with side structure (Ewert 2008: 16)
Todays technology is higher than past years.
People have many technology like calculators, computers,
television, and so on
They are support to education.
Quality of education is increased by technology.
Korea army used the brackboard of paper chapter
when they teaching about batter [battle] skills.
Three years ago, they changed the teaching way from brackboard and
paper chapters to CBT.
That was revolution of education.
Many practice soldier was understood the educations more easier.
At that time, dropping rate was decrease.
Engineers need calculators and computers
In (25) the learner provides a short narrative to illustrate his point (see
italicized text). The narrative also contains its own background clauses.
254 Kathleen Bardovi-Harlig

The use of argument in L2 tense-aspect research has not yet been fully
exploited. The use of argument prompts are well developed in the teaching
of composition and oral debate, and can be utilized in L2 tense-aspect
research with additional piloting and adjustment for specic purposes.
Ewert (2006) used a prompt related to technology and education, a topic
of relevance to her subject population: ArgumentTechnology and the
Quality of Education: Discuss your opinion on how technology (calcu-
lators, computers, television, video, etc.) aects the quality of education.
Use specic details and examples to support your ideas.
In a crosslinguistic study of the development of text production of L1
speakers from 4th grade (910 years old) to University graduate students
(2030 years old), Berman and Verhoeven (2002) collected oral and written
texts about problems between people. Subjects viewed a three-minute lm
depicting short unrelated scenes about social, moral, and physical conict
at school. No scene presented a resolution and no scene was related to the
other to make sure that the lm did not predispose the viewers to narra-
tive. The subjects were asked to give a talk in class and to write a com-
position about problems between people. They were further instructed
not to tell a story.

4.3. Irrealis: hypothetical and future texts


Realis refers to situations that have actually taken place or are actually
taking place in a real or ctional world, whereas irrealis is used for hypo-
thetical situations, including inductive generalizations and future (Comrie
1985; Dry 1983). Irrealis texts are also nonnarrative. The irrealis texts that
are most likely to be considered in tense-aspect studies are future-oriented
texts.11
Temporality is only one concept that makes up the future. Cross-
linguistic research on tense-aspect systems agrees that, unlike the purely
temporal relationships of the past and the present, the future also encom-
passes modality (Dahl 1985; see also Bybee, Perkins, and Pagliuca 1994).
Future reference may thus include modal readings of possibility, probability,
intention, and desire or volition, of which intentionality is the most common
(Bybee 1985; Bybee, Perkins, and Pagliuca 1994; Dahl 1985; Fleischman

11. Once future is introduced in any discussion of tense-aspect, the gate to modality
is open. Due to length restrictions, this chapter does not review data collection
in studies of L2 modality. The reader is referred to Giacalone Ramat and
Crocco Galeas (1995), Salsbury (2000), and von Stutterheim (1993).
Text to task 255

1982). In fact, Dahls crosslinguistic survey shows that the most typical
uses of future involve actions that are planned by the agent of the sentence
(1985: 105), whereas cases of pure prediction are less frequent. The prac-
tical outcome of this observation is that most future texts will be rst
person. The tendency to rst person in future-oriented texts or passages
can be seen in Example (26) produced in response to a writing prompt
asking learners to consider life ve years from the time of writing (which
was in 1990, hence the references to 1995).
(26) Five Years from Now
a. Five years from now, I will be twenty eight years old at 1995.
It is very dicult for me to guess what I will be doing. . . .
(HK, Written, L1 Japanese)
b. I hope in 1995 I will graguet [graduate] from university in
seecced [and succeed] if I did it I will work in the caverman
[government] in my cantery [country] and I want to make a
privet pesness [private business] . . . (MA, Written, L1 Arabic)
Future and hypothetical texts are not only written. In Example (27) the
learner was asked by the interviewer to describe a fantasy (dened by the
interviewer as something that could probably never happen; Salsbury
2000).
(27) As you know, the computer will be like, like ten years the future,
will be most things in your life . . . so many the computer will solve
will solve a lot of problems, like maybe the computer will gonna,
like help them how to see the, like farthest place in the universe,
they found life or something, maybe he can like, imagine, like,
maybe they can bring something from other planet computer can,
like translate it or imagine it.
(28) Irrealis passage from Modern Times
[16] and imagined a future of them
Chaplin is the womans husbund. It is deal. There are many food in
their home. They look like very happy.
[17] but the polise man came
An irrealis passage may also appear in other text types including narra-
tives. Modern Times contains a day-dream sequence introduced by the
title Can you imagine us in a little home like that? which encourages
narrators in the construction of irrealis text. Learners use bounding state-
256 Kathleen Bardovi-Harlig

ments as in (28), Imagined the future of them and the polise man
came and in between set up the irrealis passage morphologically by using
present, thus distinguishing it from the foreground of the narrative which
occurred in past (Bardovi-Harlig 1995).

4.3.1. Eliciting hypothetical and future texts


One of the most common means of eliciting future texts is to ask questions
about future plans for a certain time period (Moses 2002) or a specic
event or situation such as career plans (Howard 2012). In his longitudinal
study of future expression in French, Moses asked one question in each
of ve conversational interviews, What are your plans for Thanksgiving
holiday/next semesters schedule/weekend activities/summer vacation plans?
The questions were enhanced by story squares, a series of two or three
pictures with one time label for present, now, and another one or two
for the relevant future, for example Thanksgiving, or next semester.
The story squares had two rows. The rst row was labeled moi (me) in
which the pictures depicted the researchers activities. The now squares
depicted the researcher conducting interviews, and the next semester
square showed him writing his dissertation. The next row was blank, but
labeled toi (you) for the learners. Following the interviews, learners
completed written texts on the same topic designed as letters to a friend
in order to maintain the informal register of the interview. The reciprocal
information exchange, where the interviewer also divulged his own plans
added to the conversational feel and helped build rapport being the inter-
viewer and the participants.
A longitudinal study of future expression in English (Bardovi-Harlig
2004a, b, 2005) drew on an oral and written learner corpus which included
essays assigned by the intensive English program with prompts such as
ve years from now or What do you expect life to be like in 10 years/
in 2040? (Excerpt 26 is a response to the rst prompt.) In written mode,
learner journals also provided a good source of future as learners wrote
about their plans and aspirations. The prompts planned by Howard and
Moses for their interviews, and the essay prompts given by the instructors
in Bardovi-Harligs study are designed to increase the number of future
uses in one text. In conversation, Wiberg (2002) notes, future does not
generally occur in long sequences; rather, it occurs in contrast to other
temporal reference. Nevertheless, it does occur rather frequently. As an
illustration, consider that although no tokens of the future were intention-
ally elicited in the corpus which sought to investigate past-time reference,
Text to task 257

over 1,100 tokens of future occurred in 175 conversations which were


introduced entirely by learners and their interlocutors in the unplanned
portions of friendly small talk (Bardovi-Harlig 2004a).
To present hypothetical situations Salsbury (2000) set up interpretive
sessions which utilized black and white candid photographs taken by pho-
tographers for Life magazine. Learners were shown a picture and asked to
tell a story about it; then they were asked specic questions about what
they thought the subjects in the photographs were thinking or feeling
and how they might react if they were in a similar situation. As students
became more procient in the course of the longitudinal study, prompts
became increasingly verbal and included questions or statements that
learners could address involving past regrets and future hopes and dreams.
An important principle in Salsburys elicitation method is that learners
were able to choose the topics that they addressed from among a four to
six topics.
In a study of topic-related cognitive demands on syntactic accuracy,
uency, and complexity, Tapia (1993) developed two future-oriented essay
prompts, one an experiential topic, In your opinion, what is one of the
biggest problems in your country? What will need to be done to solve this
problem? and the other a hypothetical topic, What do you think life will
be like in 2041 (50 years from now)? Describe what it will be like to live
then. What changes will there be by then? The experiential topic starts
o with a current issue and builds to the future (hence the label experien-
tial) whereas the second topic is entirely set in the rather distant future
for the college-age respondents, but still in their lifetimes. The hypotheti-
cal topic is an elaborated version of topics used frequently in eliciting the
future. In the elicitation of future there is a tension between conversation
in which future appears not infrequently but in short bursts, and the ex-
tended monologic interview and writing tasks in which the text is poten-
tially entirely in future. Both are promising sources.

4.3.2. Practical considerations


The greatest challenge in eliciting nonnarrative texts is avoiding narrative.
Speakers will often switch to narrative when possible. This is so widely
recognized that Berman and Verhoeven (2002) explicitly told their partici-
pants in the expository condition to not tell a story. Still pictures, which
are risky for narrative production if viewed while speaking, promote
description, and are thus ideal when description is the goal. Some types
of nonnarrative texts are almost exclusively written, and this excludes
258 Kathleen Bardovi-Harlig

either very young learners or adults with low literacy skills; on the other
hand, these same texts are common in academic writing and can be easily
elicited in instructional settings.

4.4. Summary: working with nonnarrative texts


Nonnarrative texts have not been as widely investigated in L2 tense-aspect
research as narrative has been. As noted at the outset, discourse analysts
do not always agree on divisions among nonnarrative texts. Nevertheless,
the inclusion of a variety of nonnarrative texts enriches L2 tense-aspect
research by providing environments (both at the textual and clausal levels)
for the expression of temporal relations other than forward movement of
time. The use of independent texts (rather than background clauses) allows
speakers and writers to highlight static, atemporal, or irrealis expression in
texts ordered by principles other than chronological order. Testing the
aspect hypothesis or the discourse hypothesis on texts other than narra-
tives would allow researchers to avoid unrecognized narrative confounds.

5. Concluding remarks

The use of conversation, conversational interviews, and open-ended tasks


that result in extended monologic discourse have played a considerable
role in the study of L2 temporal expression and tense-aspect development.
Each type of language sample has its own benets and its own costs. What
they share is the opportunity for learners to make meaning and to use
their emergent tense-aspect systems (as well as other means of temporal
expression) in the act of communication. The communicative focus directs
learners attention away from explicit knowledge which is likely to play a
role in tense-aspect use under monitored conditions. Conversation oers
the greatest range of tense-aspect morphology given that a conversation
can have any possible temporal frame, depending on the topic. This, it
is likely to display a range of tense-aspect forms. Speakers may change
temporal reference, sometimes quite rapidly (for example, imagine moving
from reporting an event earlier in the day to weekend plans). Depending
on the overall oral-aural competence of a learner and his underlying
tense-aspect system, a learner may display agility in following the lead of
his interlocutor (or not), or if inclined, may lead the conversation. The
turn structure is both a cost and a benet in that turn taking supports
lower-level speakers (a benet) and a cost in that relatively balanced turns
preclude extended turns for an individual speaker, thus limiting a sample.
Text to task 259

Conversational narratives are the one exception to regular turn taking


because turns are often suspended to allow an extended narrative to occur.
Conversational narratives, like other extended monologic texts, allow
speakers to create an entire text on their own. Conversational narratives
are typically framed by a claim which they illustrate. Speakers are invested
in making their point and add rich support in the form of background to
help them do so. The benet of personal narratives, whether spontaneous
or elicited, is speaker involvement which leads to more elaboration (and
every clause provides more evidence for the system which underpins it);
however, the greatest cost of personal narratives is the lack of comparability.
Because of its disciplinary investment in language samples that can be
compared, SLA studies in tense-aspect research have favored elicited
narratives. Several means have been developed to elicit impersonal narra-
tives which were detailed in the preceding sections. The comparability of
texts has paid o by providing relatively large corpora of narratives for
analysis. The cost is in the less elaborated background which is thought
to stem from the impersonal perspective. A hybrid elicitation task yielding
personalized narratives (a personal perspective of third-party events) has
shown promise in increasing background information relative to foreground
events. Narratives have been used successfully in small case studies and
large scale studies, with learners of all ages, levels of prociency, and
levels of literacy and education. Narratives are excellent sources for the
study of perfective past with foreground providing a contrast to the diverse
time frames and tense-aspect morphology found in the background.
Finally, this chapter considered nonnarrative texts and the role they
can play in providing data for the study of statives and imperfectives.
Although there has recently been a move to add nonnarrative texts to the
study of tense-aspect development, overall, nonnarrative texts have not
been exploited fully. The production of some types of nonnarrative texts
such as descriptions and future-oriented texts (including things that are
imagined, and hopes and plans for the future) can be elicited from a range
of informants. Other nonnarrative texts such as expository and argument
texts are often written and their structure is typically learned during
schooling; they thus seem to be relatively more limited as to the range of
learners who can produce them. No doubt there will be an increased use
of such texts as research moves from the focus on perfective to imperfec-
tive forms and functions in L2 acquisition research.
In summary, this chapter has investigated both talk and text types, con-
sidered the characteristics of each in terms of tense and aspect distribution
and function, and shown how tasks which promote extended discourse can
260 Kathleen Bardovi-Harlig

be conducted across languages and with populations at various levels of


prociency and most at any level of L1 or L2 literacy. The data which
extended discourse provides can be used to test a wide range of hypotheses
and to understand form-to-meaning and meaning-to-form associations in
developing L2 tense-aspect systems.

References

Algar, James (Director), Perce Pearce & Carl Fallberg (Writers)


1940 Sorcerers apprentice. (In Fantasia). [Motion picture]. United
States: Walt Disney Productions.
Bamberg, Michael & Virginia Marchman
1990 What holds a narrative together? The linguistic encoding of
episode boundaries. Papers in Pragmatics 4. 58121.
Bardovi-Harlig, Kathleen
1992a The telling of a tale: Discourse structure and tense use in learners
narratives. In L. F. Bouton & Y. Kachru (Eds.), Pragmatics and
language learning, Monograph 3, 144161. University of Illinois,
Urbana-Champaign: Division of English as an International
Language.
Bardovi-Harlig, Kathleen
1992b The use of adverbials and natural order in the development of
temporal expression. IRAL 30. 299320.
Bardovi-Harlig, Kathleen
1994 Reverse-order reports and the acquisition of tense: Beyond the
principle of chronological order. Language Learning 44. 243
282.
Bardovi-Harlig, Kathleen
1995 A narrative perspective on the development of the tense/aspect
system in second language acquisition. Studies in Second Lan-
guage Acquisition 17. 263291.
Bardovi-Harlig, Kathleen
1997a Another piece of the puzzle: The emergence of the present perfect.
Language Learning 47. 375422.
Bardovi-Harlig, Kathleen
1997b Assessing grammatical development in interactional contexts.
TESOL Quarterly 31. 797806.
Bardovi-Harlig, Kathleen
1998 Narrative structure and lexical aspect: Conspiring factors in
second language acquisition of tense-aspect morphology. Studies
in Second Language Acquisition 20. 471508.
Bardovi-Harlig, Kathleen
1999 Examining the role of text type in L2 tense-aspect research:
Broadening our horizons. In P. Robinson (Ed.), Proceedings
Text to task 261

of the Third Pacic Second Language Research Forum (Vol. 1,


129138. Tokyo: The Pacic Second Language Research
Forum.
Bardovi-Harlig, Kathleen
2000 Tense and aspect in second language acquisition: Form, meaning,
and use. Oxford: Blackwell.
Bardovi-Harlig, Kathleen
2004a The emergence of grammaticalized future expression in longitu-
dinal production data. In M. Overstreet, S. Rott, B. VanPatten
& J. Williams (Eds.), Form and meaning in second language
acquisition, 115137. Mahwah, NJ: Erlbaum.
Bardovi-Harlig, Kathleen
2004b Monopolizing the future: How the go-future breaks into will s
territory and what it tells us about SLA. EuroSLA Yearbook 4.
177201. Amsterdam: Benjamins.
Bardovi-Harlig, Kathleen
2005 Tracking the elusive imperfect in adult second language acquisi-
tion: Rening the hunt. In P. Kempchinsky & R. Slabakova
(Eds.), Aspectual inquiries, 397419. Dordrecht: Kluwer.
Bardovi-Harlig, Kathleen
2007 One functional approach to SLA: The concept-oriented approach.
In B. VanPatten & J. Williams (Eds.), Theories in second lan-
guage acquisition, 5775. Mahwah, NJ: Erlbaum.
Bardovi-Harlig, Kathleen & Tom Salsbury
2004 The organization of turns in the disagreements of L2 learners: A
longitudinal perspective. In D. Boxer & A. D. Cohen (Eds.),
Studying speaking to inform second language learning, 199227.
Clevedon: Multilingual Matters.
Bayley, Robert
1994 Interlanguage variation and the quantitative paradigm: Past
tense marking in Chinese-English. In S. Gass, A. Cohen & E.
Tarone (Eds.), Research methodology in second language acquisi-
tion, 157181. Hillsdale, NJ: Erlbaum.
Bergstrom, Anna
1995 The expression of past temporal reference by English-speaking
learners of French. State College, PA: Pennsylvania State Uni-
versity, Ph.D. dissertation.
Bergstrom, Anna
1997 Linuence des distinctions aspectuelles sur lacquisition des
temps en francais langue etrangere. [The inuence of aspectual
distinctions on the acquisition of tense in French as a foreign
language.] Acquisition et interaction en langue etrangere 9. 52
82.
Berman, Ruth A. & Dan Slobin
1994 Relating events in narrative: A crosslinguistic developmental study.
Hillsdale, NJ: Erlbaum.
262 Kathleen Bardovi-Harlig

Berman, Ruth A. & Ludo Verhoeven


2002 Cross-linguistic perspectives on the development of text-produc-
tion abilities. Written Language and Literacy 5. 143.
Biber, Douglas
1989 A typology of English texts. Linguistics 27. 343.
Billot, Amandine, Alain Chabat, Christine Rouxel (Producers) & Thomas Balmes
(Director)
2010 Babies [Motion picture]. United States: Focus Features.
Bley-Vroman, Robert & Craig Chaudron
1994 Elicited imitation as a measure of second-language competence.
In E. Tarone, S. M. Gass & A. D. Cohen (Eds.), Research meth-
odology in second language acquisition, 227243. Hillsdale, NJ:
Lawrence Erlbaum.
Bronckart, Jean-Paul & Beatrice Bourdin
1993 Lacquisition des valeurs des temps des verbes: Etude compara-
tive de lallemand, du basque, du catalan, du francais et de lita-
lien. [The acquisition of tenses: a comparative study of German,
Basque, Catalan, French, and Italian.] Langue Francaise 97.
102124.
Burghardt, Beatrix
2010 The acquisition of grammatical aspect: The use of prexes in L2
Hungarian narrative. Indiana University, Ph.D. qualifying paper.
forthcoming The acquisition of grammatical aspect through directional expres-
sions. Bloomington, IN: Indiana University, Ph.D. dissertation.
Bybee, Joan
1985 Morphology: A study of the relation between meaning and form.
Amsterdam: Benjamins.
Bybee, Joan, Revere Perkins & William Pagliuca
1994 The evolution of grammar: Tense, aspect, and modality in the lan-
guages of the world. Chicago: University of Chicago Press.
Caenepeel, Mimo
1995 Aspect and text structure. Linguistics 33. 213253.
Carroll, Mary, Antje Rossdeutscher, Monique Lambert & Christiane von
Stutterheim
2008 Subordination in narratives and macrostructural planning: A
comparative point of view. In C. Fabricius-Hansen & W.
Ramm (Eds.), Subordination versus coordination in sentence
and text, 161184. Amsterdam: Benjamins.
Carroll, Mary & Christiane von Stutterheim
2003 Typology and information organisation: Perspective taking and
language-specic eects in the construal of events. In A. Giacalone
Ramat (Ed.), Typology and second language acquisition, 365
402. Berlin: de Gruyter.
Chafe, William (Ed.)
1980 The Pear stories. Norwood, NJ: Ablex.
Text to task 263

Chaplin, Charlie (Director)


1936 Modern times [Motion picture]. United States: United Artists.
Clachar, Arlene
2005 Creole English speakers treatment of tense-aspect morphology
in English interlanguage written discourse. Language Learning
55. 275334.
Clements, J. Clancy
2003 The tense-aspect system in pidgins and naturalistically learned
L2. Studies in Second Language Acquisition 25. 245281.
Collier, Gary, Don Kuiken & Michael Enzle
1982 The role of grammatical qualication in the expression and
perception of emotion. Journal of Psycholinguistic Research 11.
631650.
Comajoan, Llorenc
2001 The acquisition of past temporality and past morphology in Catalan
as a foreign language. Bloomington, IN: Indiana University, Ph.D.
dissertation.
Comajoan, Llorenc
2005a The acquisition of perfective and imperfective morphology and
the marking of discourse grounding in Catalan. In D. Ayoun &
M. R. Salaberry (Eds.), Tense and aspect in Romance languages,
3578. Amsterdam: Benjamins.
Comajoan, Llorenc
2005b The emergence of L2 imperfective morphology in narrative re-
tellings and conversations. Catalan Review 19. 79108.
Comajoan, Llorenc
2006 The Aspect Hypothesis: Development of morphology and appro-
priateness of use. Language Learning 56. 201268.
Comrie, Bernard
1985 Tense. New York: Cambridge University Press.
Dahl, Osten
1984 Temporal distance: Remoteness distinctions in tense-aspect systems.
In B. Butterworth, B., Comrie & O. Dahl (Eds.), Explanations
for language universals, 105122. Berlin: Mouton.
Dahl, Osten
1985 Tense and aspect systems. Oxford: Basil Blackwell.
Dale, Philip
1975 Language development: Structure and function. New York: Holt
Rinehart and Winston.
Davitz, Joel R.
1969 The language of emotions. New York: McGraw Hill.
Dietrich, Rainer, Wolfgang Klein & Colette Noyau (Eds.)
1995 The acquisition of temporality in a second language. Amsterdam:
John Benjamins.
264 Kathleen Bardovi-Harlig

Douglas, Dan & Larry Selinker


1994 Research methodology in context-based second language research.
In E. E. Tarone, S. M. Gass & A. D. Cohen (Eds.), Research
methodology in second language acquisition, 119131. Hillsdale,
NJ: Erlbaum.
Dry, Helen
1981 Sentence aspect and the movement of narrative time. Text 1.
23340.
Dry, Helen
1983 The movement of narrative time. Journal of Literary Semantics
12. 1953.
Du, Patricia & Duan-Duan Li
2002 The acquisition and use of perfective aspect in Mandarin. In
M. R. Salaberry & Y. Shirai (Eds.), Tense-aspect morphology in
L2 acquisition, 417453. Amsterdam: John Benjamins.
Dulay, Heidi, Marina Burt & Stephen Krashen
1982 Language Two. Oxford: Oxford University Press.
Ellis, Rod
2008 The study of second language acquisition (2nd ed.). Oxford:
Oxford University Press.
Ewert, Doreen E.
2006 The expression of temporality in the written discourse of L2
learners of English: Distinguishing text-types and text passages.
Bloomington, IN: Indiana University, Ph.D. dissertation.
Ewert, Doreen E.
2008 Non-narrative writing for pre-academic L2 adult learners.
INTESOL Journal 5. 922.
Fleischman, Suzanne
1982 The future in thought and language. Cambridge: Cambridge Uni-
versity Press.
Fleischman, Suzanne
1985 Discourse functions of tense-aspect oppositions in narrative:
Toward a theory of grounding. Linguistics 23. 851882.
Georgakopoulou, Alexandra & Dionysis Goutsos
2000 Mapping the world of discourse: The narrative versus non-
narrative distinction. Semiotica 131. 113141.
Giacalone Ramat, Anna
2002 How do learners acquire the three classic categories of temporal-
ity? Evidence from L2 Italian. In M. R. Salaberry & Y. Shirai
(Eds.), Tense-aspect morphology in L2 acquisition, 221448. Am-
sterdam: John Benjamins.
Giacalone Ramat, Anna & Emanuele Ban
1990 The acquisition of temporality: A second language perspective.
Folia Linguistica XXIV/34. 405428.
Text to task 265

Giacalone Ramat, Anna & Crocco Galeas, G. (Eds.)


1995 From pragmatics to syntax: Modality in second language acquisi-
tion. Tubingen: Gunter Narr.
Hasbun, Leyla
1995 The role of lexical aspect in the acquisition of tense and gram-
matical aspect in Spanish as a foreign language. Bloomington,
IN: Indiana University, Ph.D. dissertation.
Hill, Jane H.
1988 Language, culture, and world view. In F. J. Newmeyer (Ed.), Lin-
guistics: The Cambridge survey (Vol. 4). Language: The socio-
cultural context, 1436. Cambridge: Cambridge University Press.
Hopper, Paul
1979 Aspect and foregrounding in discourse. In T. Givon (Ed.), Syn-
tax and semantics: Discourse and syntax, 213241. New York:
Academic Press.
Hopper, Paul & Sandra Thompson
1980 Transitivity in grammar and discourse. Language 56. 251299.
Housen, Alex, Nancy Kemps & Michel Pierrard
2009 The use of verb morphology of advanced L2 learners and native
speakers of French. In E. Labeau & F. Myles (Eds.), The ad-
vanced learner variety: The case of French, 4161. Oxford: Peter
Lang.
Howard, Martin
2005 The emergence and use of the plus-que-parfait in advanced
French interlanguage. In J.-M. Dewaele (Ed.), Focus on French
as a foreign language: Multidisciplinary perspectives, 6387.
Clevedon, England: Multilingual Matters.
Howard, Martin
2009 Short-versus long-term eects of naturalistic exposure on the
advanced instructed learners L2 development: A case-study. In
E. Labeau & F. Myles (Eds.), The advanced learner variety: The
case of French, 93123. Oxford: Peter Lang.
Howard, Martin
2012 From tense and aspect to modality the acquisition of future,
conditional and subjunctive morphology in L2 French: A pre-
liminary study. In E. Labeau (Ed.), Development of tense, aspect
and mood in L1 and L2, 201223. Amsterdam: Rodopi/Cahiers
Chronos.
Jourdenais, Renee
1998 The eects of textual enhancement on the acquisition of the
Spanish preterite and imperfect. Georgetown, DC: Georgetown
University, Ph.D. dissertation.
Kihlstedt, Maria
2002 Reference to past events in dialogue. In M. R. Salaberry & Y.
Shirai (Eds.), Tense-aspect morphology in L2 acquisition, 323
335. Amsterdam: John Benjamins.
266 Kathleen Bardovi-Harlig

Kim, Hyun-Jin
2012 A case study of tense-aspect marking by L2 learners of Korean.
In E. Labeau (Ed.), Development of tense, aspect and mood in L1
and L2, 153177. Amsterdam: Rodopi/Cahiers Chronos.
Klein, Wolfgang
1986 Second language acquisition. Cambridge: Cambridge University
Press.
Labeau, Emmanuelle
2009 An imperfect mastery: The acquisition of the functions of impar-
fait by Anglophone learners. In E. Labeau & F. Myles (Eds.),
The advanced learner variety: The case of French, 6392. Oxford:
Peter Lang.
Labov, William
1972 Language in the inner city. Philadelphia: University of Pennsyl-
vania Press.
Labov, William & Joshua Waletzky
1967 Narrative analysis: Oral versions of personal experience. In J.
Helm (Ed.), Essays on the verbal and visual arts, 1244. Seattle:
University of Washington Press.
Laord, Barbara
1996 The development of tense/aspect relations in L2 Spanish narra-
tives: Evidence to test competing theories. Paper presented at the
Second Language Research Forum, Tucson, AZ.
Lasseter, John (Writer, Director)
1988 Tin toy [Short lm]. In Pixar short lms collection: Volume 1.
United States: Walt Disney Home Entertainment.
Leclerq, Pascale
2009 The inuence of L2 French on near native French learners of
English. The case of simultaneity. In E. Labeau & F. Myles
(Eds.), The advanced learner variety: The case of French, 269
289. Oxford: Peter Lang.
Liskin-Gasparro, Judith
1996 Narrative strategies: A case study of developing storytelling
skills by a learner of Spanish. Modern Language Journal 80.
271286.
Liskin-Gasparro, Judith
2000 The acquisition of temporal expression in Spanish oral narra-
tives: Exploring learners perceptions. Hispania 83. 830844.
Lopez-Ortega, Nuria
2000 Tense, aspect, and narrative structure in Spanish as a second lan-
guage. Hispania 83. 488502.
MacDonald, Mary R.
1986 Twenty tellable tales: Audience participation folktales for the begin-
ning storyteller. New York: Wilson.
Text to task 267

Mercer, Marcel
1969 Frog, where are you? New York: Dial Press. Print.
Montgomery, Tyron (Director) & Thomas Stellmach (Writer)
1996 Quest [Animated short]. Germany: Montgomery Film/Thomas
Stellmach Animation.
Moses, John G.
2002 The expression of futurity by English-speaking learners of French.
Bloomington, IN: Indiana University, Ph.D. dissertation.
Noyau, Colette
1984 The development of means for temporality in French by adult
Spanish-speakers: Linguistic devices and communicative capacities.
In G. Extra & M. Mittner (Eds.), Studies in second language acqui-
sition by adult immigrants: Proceedings of the ESF/AILA sympo-
sium held on the 9th of August 1984 in Brussels, 113137. Tilburg,
Netherlands: Tilburg University, Department of Language and
Literature.
Noyau, Colette
1990 The development of means for temporality in the unguided
acquisition of L2: Crosslinguistic perspectives. In H. W. Dechert
(Ed.), Current trends in European second language acquisition
research, 143170. Clevedon, UK: Multilingual Matters.
Ragnarsdottir, Hrafnhildur, Melina Aparici, Dalia Cahana-Amiaty, Janet van
Hell & Anne Viguie-Simon
2002 Verbal structure and content in written discourse: Expository
and narrative texts. Written Language and Literacy 5. 95125.
Reinhart, Tanya
1984 Principles of gestalt perception in the temporal organization of
narrative texts. Linguistics 22. 779809.
Rintell, Ellen M.
1989 That reminds me of a story: The use of language to express
emotion by second-language learners and native speakers. In
M. R. Eisenstein (Ed.), The dynamic interlanguage, 237257.
New York: Plenum.
Salaberry, M. Rafael
1998 The development of aspectual distinctions in academic L2
French. Canadian Modern Language Review 54. 508542.
Salsbury, Tom
2000 The acquisitional grammaticalization of unreal conditionals and
modality in L2 English: A longitudinal perspective. Bloomington,
IN: Indiana University, Ph.D. dissertation.
Schirin, Deborah
1981 Tense variation in narrative. Language 57. 4562.
Schmiedtova, Barbara
2004 At the same time . . . The expression of simultaneity in learner
varieties. Berlin: Mouton de Gruyter.
268 Kathleen Bardovi-Harlig

Selinker, Larry & Dan Douglas


1988 Using discourse domains in creating interlanguage: Context,
theory, and research methodology. In J. Klegraf & D. Nehls,
(Eds.), Essays on the English language and applied linguistics
on the occasion of Gerhard Nickels 60th birthday, 357379.
Heidelberg: Julius Groos.
Shirai, Yasuhiro
2002 The proto-type hypothesis of tense-aspect in second language. In
R. Salaberry & Y. Shirai (Eds.), Tense-aspect morphology in L2
acquisition, 455478. Amsterdam: John Benjamins.
Silverstein, Michael
1976 Hierarchy of features and ergativity. In R. M. W. Dixon (Ed.),
Grammatical categories in Australian languages, 112177. Can-
berra: Australia Institute of Aboriginal Studies.
Smith, Carlota S.
2003 Modes of discourse. Cambridge: Cambridge University Press.
Smith, Carlota S.
2005 Aspectual entities and tense in discourse. In P. Kempchinsky
& R. Slabakova (Eds.), Aspectual inquiries, 223237. Dordrecht:
Kluwer.
Tannen, Deborah
1983 I take out the rockDok!: How Greek women tell about being
molested (and create involvement). Anthropological Linguistics
25. 359374.
Tapia, Elena
1993 Cognitive demand as a factor in interlanguage syntax: A study
in topics and texts. Bloomington, IN: Indiana University, Ph.D.
dissertation.
Tomlin, Russell
1984 The treatment of foreground-background in the on-line descrip-
tive discourse of second language learners. Studies in Second
Language Acquisition 6. 115142.
Trevise, Anne
1987 Toward an analysis of the (inter)language activity of referring
to time in narratives. In C. W. Pfa (Ed.), First and second
language acquisition processes, 225251. Cambridge, MA: New-
bury House.
von Stutterheim, Christiane
1991 Narrative and description: Temporal reference in second language
acquisition. In T. Huebner & C. A. Ferguson (Eds.), Cross-
currents in second language acquisition and linguistic theories,
385403. Amsterdam: Benjamins.
von Stutterheim, Christiane
1993 Modality: Function and form in discourse. In N. Dittmar &
A. Reich (Eds.), Modality in language acquisition, 326. Berlin:
de Gruyter.
Text to task 269

von Stutterheim, Christiane & Wolfgang Klein


1989 Referential movement in descriptive and narrative discourse. In
R. Dietrich & C. F. Graumann (Eds.), Language processing in
social context, 3976. Amsterdam: Elsevier.
Wiberg, Eva
2002 Information structure in dialogic future plans. In M. R. Salaberry
& Y. Shirai (Eds.), Tense-aspect morphology in L2 acquisition,
285321. Amsterdam: John Benjamins.
Chapter 8
Dening and coding data: Lexical aspect in L2 studies

Yasuhiro Shirai

1. Introduction

This chapter discusses issues surrounding the use of lexical aspect categories
in research on tense-aspect markers, mainly in second language acquisition
(SLA). I will rst discuss the issue concerning the optimal number of lexical
aspect categories to be distinguished, and the issues surrounding the dis-
crepancies of lexical aspect categories across languages. Then, I will present
the details of a classication procedure to provide a better sense of how
verb classications are conducted and the issues involved in classifying
problematic cases.
Since the 1970s, research on the relationship between tense-aspect markers
and characteristics of verb semantics in rst language acquisition has un-
covered an important correlation now known as the Aspect Hypothesis
(e.g., Andersen and Shirai 1994). Early studies such as the ones conducted
by Antinucci and Miller (1976), Stephany (1981), Aksu-Koc (1988), and
Bloom, Lifter and Haz (1980) used dierent methods to determine the
relevant verb semantic categories, and thus there were issues of replicability.
Weist et al. (1984) was the rst study to systematically use operational
tests based on Vendlers (1957) verb classication. Since then, many re-
searchers investigating the relationship between tense-aspect marking and
temporal semantics of verbs both in L1 and L2 acquisition have used
linguistic tests systematically so that studies can be comparable and repli-
cable. Yet, there are still discrepancies that need to be addressed to reach
even higher levels of systematicity.

2. How many categories?

Vendlers (1957) model of lexical aspect, which has been the most inuen-
tial both in linguistics and acquisition research, consists of four classes:
States, activities, accomplishments, and achievements. States (e.g., love)
describe a situation that is viewed as continuing to exist unless some out-
side situation makes it change. Activities (e.g., run) describe a dynamic
272 Yasuhiro Shirai

and durative situation that has an arbitrary endpoint, i.e., it can be termi-
nated at any time. In contrast, accomplishments (e.g., make a chair)
describe a situation that is dynamic and durative, but has a natural end-
point after which the particular action cannot continue (i.e., they are telic).
Finally, achievements describe an instantaneous and punctual situation;
that is, it can be reduced to a point on a time axis (e.g., die). States are
[dynamic], [telic], [punctual]; activities are [dynamic], [telic],
[punctual]; accomplishments are [dynamic], [telic], [punctual];
achievements are [dynamic], [telic], [punctual]. The following is a
graphic representation of these four categories.
State ______________ love, know, belong, think that . . .
Activity PPPPPPPPPPP run, walk, swim, think about . . .
Accomplishment PPPPPPPPPPP x paint a picture, make a chair,
walk to school
Achievement x fall, drop, win the race, die
Although Vendlers model is often taken as the standard, there are
other variations of lexical aspect categories in linguistic analysis. For
instance, Smith (1991, see also 1997) proposed a 5-way classication add-
ing another category: semelfactive, which is punctual but not telic (e.g.,
jump, knock). Verkuyl (1989), in turn, argued that the distinction between
achievements and accomplishments is not necessary. Furthermore, even if
dierent researchers use the same four-way distinction, they dier with
respect to how the four categories are operationalized; that is, they dier
in terms of the identicatory criteria used. For example, some researchers
take punctuality as a dening feature to distinguish accomplishments from
achievements (Smith 1991), whereas others (Van Valin 1993) do not (see
Shirai 1996).1
Why are there so many dierent proposals? We should remember that
verbal semantics (or language itself for that matter) is not such an entity
that can be classied into one, two, or three discrete categories by a small

1. Shirai (1996) pointed out that there are two approaches to lexical aspect, one
relying on temporal features (especially duration) which he called the temporal
approach, represented by Smith (1991), and one relying on logical structure
(e.g., VanValin 1993), which he called the decompositional approach. They
dier greatly in how they classify achievements and accomplishments. Essen-
tially, the latter position treats causative verbs as accomplishments regardless
of duration, and therefore walk the dog (activity in the temporal approach)
and throw the hat on the table (achievement in the temporal approach) are
accomplishments.
Lexical aspect 273

set of linguistic tests. It is well known that linguistic tests which are sup-
posed to give the same distinction (such as stative versus dynamic), often
give dierent results when applied to the same predicate (Parsons 1989).
For example, the linguistic test often used to determine whether a verb is
stative or dyanamic is incompatibility with progressive marking. If a
verb cannot be progressivized, it is stative. This test yields that [live] is a
dynamic verb since one can say I am living in Tokyo. However, another
linguistic test for stativity, such as no habitual reading in simple present
tense, yields that it is a stative verb. If a verb is dynamic, simple present
tense yields habitual reading. (I run three miles means that this person runs
three miles habitually, say every day). Since one can say I live in Tokyo
without any repetition involved, this tests yields [live] is stative, which con-
tradicts the progressive test mentioned above.
To make matters more complex, we cannot assume one classication
system is optimal for all languages, although lexical aspect categories are
often considered to be universal categories rooted in basic ontological
conceptual distinctions. For example, Smith (1991) proposed a universal
system of aspectual phenomena applicable to her 5-way classication to
English, Russian, French, Chinese, and Navajo. She did, however, modify
her system of lexical aspect in accounting for the aspectual system of
Navajo. Given these observations, it is not easy to evaluate which classi-
cation system is best for linguistic analysis, and it is beyond the scope of
this chapter to do such an evaluation. In what follows, I will focus on the
evaluation of dierent categories/classications used for second language
acquisition research.
Roger Andersen was the rst to investigate the eect of Vendlerean
lexical aspect in second language acquisition of tense-aspect morphology
(e.g., Andersen 1986, 1989). Although he did not publish his linguistic
tests, he was using them in analyzing data, and discussed such tests exten-
sively in his seminars at UCLA, which inuenced his students works, in-
cluding Shirai (1991). Robison (1990), who was Andersens student, was the
rst to publish a study that used a classication system based on linguistic
tests in SLA. His study on naturalist acquisition of English as a second
language by an immigrant, which was inuenced by Bickertons (1981)
Language Bioprogram Hypothesis, was therefore slightly dierent from
the standard Vendlerean four-way classication. He focused on the state-
process and the punctual-nonpunctual distinctions proposed by Bickerton,
and therefore, he did not distinguish between activities and accomplish-
ment, that is, the telic-atelic distinction. Robison (1993, 1995) added punc-
tual activity (i.e., semelfactive) and punctual state (which generally consists
274 Yasuhiro Shirai

of punctual change of state verbs of perception/cognition such as notice)


to Vendlers four categories.
Bardovi-Harlig and Bergstrom (1996) was a cross-linguistic study in
which both tests of English and French were used. This was the st pub-
lished study using the Vendleran four-way classication with operational
tests in SLA. Bardovi-Harlig (1998) also used the same tests for her study
on L2 English with the four-way Vendlerean classication. Salaberrys
studies (e.g., Salaberry 1999) on L2 Spanish usually employ a three-way
classication collapsing achievements and accomplishments as events. As
an extension, his work on English as a second language also employs 3-
way classications (Ayoun and Salaberry 2008).2 Researchers working on
other languages, such as Japanese (Shirai 1995; Shirai and Kurono 1998),
Korean (Lee and Kim 2007), and Catalan (Comajoan 2006) also used the
Vendlerean 4-way classications. Table 1 summarizes the published studies
that used linguistic tests to classify lexical aspect in second language acqui-
sition, although they are not meant to be exhaustive.3
In analyzing acquisition data, we cannot come to a denite conclusion
regarding which particular classication system universally works best for
all languages and for all purposes (see Chapter 11). There are various
advantages and disadvantages in particular systems of classication. Gen-
erally speaking, a ner distinction is better at the level of the initial classi-
cation. For example, a 6-way classication is better than a 4-way one.
This is because we can always choose to collapse two categories together
as one afterwards if necessary. However, there is always a trade-o when
using a large number of categories: If one has more rather than fewer cate-
gories, the overall process of categorizing verbs becomes time-consuming.
If, for example, the distinction between accomplishment versus achievement
is not always easy, and if we do not need to make this distinction, the
classication process will be much more straightforward than otherwise.
However, by abandoning this distinction, one might lose important infor-
mation about learners sensitivity to lexical aspect. One interesting nding

2. See Salaberry (2000: 3235) for his argument for collapsing two telic categories.
3. There are other studies that use the categories from Vendler, but not all such
studies report how linguistic tests are used. Wiberg (1996) and Collins et al.
(2009) used a 3-way, and Montrul (2002) used a 4-way classication but did
not report which tests were used to classify verbs. Rohde (1996) reported only
the tests for telic versus atelic distinction. Lopez-Ortega (2000) used a 4-way
categorization but she only stated that Robisons (1995) classication test was
used, which in fact was a 6-way classication, and therefore it is not very clear
how the verbs were classied into 4-way categories.
Lexical aspect 275

Table 1. Representative studies that used Vendlerean categories of lexical aspect


L2 / Study Vendlerean categories

English
Robison (1990) state versus process; punctual versus non-punctual
Robison (1995) 6-way (4-way punctual activity, punctual state)
Bardovi-Harlig & Bergstrom (1996) 4-way
Bardovi-Harlig (1998) 4-way
Gavruseva (2002) 4-way
Ayoun & Salaberry (2008) 3-way (state, activity, telic)
French
Bardovi-Harlig & Bergstrom (1996) 4-way
Japanese
Shirai (1995) 4-way
Shirai & Kurono (1995)
Spanish
Salaberry (1999) 3-way (state, activity, telic)
Camps (2002) 4-way
Korean
Lee & Kim (2007) 4-way
Catalan
Comajoan (2006) 4-way

from Bardovi-Harlig (1998) was that there was a statistically signicant


dierence between accomplishments and achievements in oral lm retell
tasks, while there was no dierence in written narratives. This interest-
ing task variation would have been missed had she collapsed the two telic
categories.
Another example of the danger of abandoning the achievement-accom-
plishment distinction can be illustrated by Collins et al.s (2009) study.
This study used three categories, and found that 72% of past tense forms
is used with telic verbs while 59% of progressive marking is used with
activity verbs in the ESL classroom input. Based on this nding, they argued
that past tense use is more restricted to prototypical verbs than progressive
marking, which makes it more dicult to acquire. However, previous
research (e.g., Shirai and Andersen 1995) showed that both progressive
and past markings are attached to their prototypes about 60% of the time
in native English speakers speech. This discrepancy is possibly related to
the use of a four-way versus a three-way classication. Shirai and Andersen
276 Yasuhiro Shirai

(1995) used a four-way classication, and they reported that about 10% of
both progressive and past markings were used with accomplishment verbs.
Therefore, it is likely that in Collins et al.s study too, both progressive
and past markers are attached to their prototype (i.e., activity verbs and
achievement verbs, respectively) around 60% of the time had they used
the four-way classication. In other words, the choice of three versus four
categories determined the outcome.
The choice of collapsing two telic categories (i.e., achievements and
accomplishments) is also inuenced by the target structure under investi-
gation. When the past or perfective category is the target, achievements
and accomplishments behave largely identically in terms of the meaning
they denote (change of state), whereas when they are used with imperfec-
tive aspect there is a substantial dierence in meaning between achievements
and accomplishments. For example, progressive aspect, which is dynamic
imperfective, generally yields action in progress meaning when it is used
with accomplishment verbs (e.g., He is making a chair). When the progres-
sive is attached to achievement verbs, however, it denotes the meaning
of preliminary stages of an event (Smith 1991), such as He is leaving or
He is reaching the summit. Thus, overall one would predict dierential
behavior between accomplishments and achievements. In the same vein,
when Japanese imperfective -teiru is attached to achievements it denotes
resultative state as in (1), whereas when it is used with accomplishments,
it denotes action in progress, as in (2).

Resultative: (1) Ken-wa sin-de i-ru.


Ken-Top die-Asp-Nonpast
Ken is dead.
Progressive: (2) Ken-wa hasit-te i-ru.
Ken-Top run-Asp-Nonpast
Ken is running.

That is, we obtain totally dierent meanings depending on whether imper-


fective aspect is attached to accomplishments or achievements. Thus, in
such cases it is more appropriate to make ner distinctions within telic
verbs without lumping together achievements and accomplishment.
This brings us to another question of cross-linguistic validity of lexical
aspect categories, which may aect the reliability of classication. When
I tried to translate the classication procedure for English (Shirai 1991)
into Japanese (Shirai 1993, 1998a), I found it very dicult to distinguish
Lexical aspect 277

activities from achievements in Japanese. This problem did not arise in


English, and so I added additional tests to distinguish between them.4
Likewise, when I tried to distinguish achievements from accomplishments
in Spanish by asking native Spanish speakers for acceptability judgments, it
was dicult to make this distinction. It is possible that dierent languages
may bring about specic diculties associated with particular distinctions.
These cross-linguistic dierences in aspectual organization may come into
play in making decisions in how many categories should be used in acquisi-
tion research.
In sum, we should not assume that there is one optimal number for
categories. This applies not only to acquisition research, but also in lin-
guistic analysis. The optimal number of categories should dier depending
on various factors, such as target language, target structures, and the target
of analysis (see Chapters 5 and 12).

3. Crosslinguistic discrepancy in lexical aspect

When we talk about a Vendlerean category, it is often the case that its
universality is emphasized. This is understandable because the attraction
of Vendlerean categories is the observation that in almost all languages
being studied these lexical aspect categories show very similar properties
and that their interaction with grammatical aspect can predict the aspec-
tual meaning of sentences (Smith 1991; Shirai 2000). These categories are
thus considered covert grammatical categories, and sometimes even con-
sidered innate (Smith 1991). Given the importance ascribed to the universal
nature of lexical aspect categories, it is not surprising that crosslinguistic
variations have not gured prominently either in linguistic analysis or
acquisition. However, it has been acknowledged that dierent languages
lexicalize similar notions dierently (e.g., Rothstein 2004; Tatevosov 2002).

4. The diculty associated with making the activity-achievement distinction,


which is more dicult than the achievement-accomplishment distinction in
Japanese, is probably related to the fact that Japanese has an imperfective
marking -teiru (see Shirai 2000 for further details), which clearly distinguishes
achievements and accomplishments as shown in the contrast between examples
(1) and (2). Japanese -teiru, which is obligatory in marking ongoing situation
in progress as in English, can refer to ongoing progressive as well as resultant
states. In other words, Japanese speakers always have to be sensitive to
whether a verb has duration or not.
278 Yasuhiro Shirai

For instance, the verb to know in English is stative, whereas its counter-
part in Italian is an activity (Giacalone Ramat 1997). Japanese does not
have a verb that corresponds to the meaning of the verb to know in these
languages, but the equivalent notion of knowing something or someone is
denoted by combining achievement verb siru come to know with the
imperfective aspect marker -teiru (Li and Shirai 2000).
This crosslinguistic variation in lexical aspect may have an important
implication for the study of tense and aspect acquisition. Although the
study concerning the Aspect Hypothesis has generally emphasized the uni-
versality observed in tense-aspect acquisition, eects of learners L1 have
been observed. Although an increasing number of studies have begun to
look at the eect of learners L1, they are only concerned with the L1
eect at the level of grammatical tense-aspect, and how it interacts with
lexical aspect (e.g., Rocca 2002; Collins 2002, 2004; Sugaya and Shirai
2007; Izquierdo and Collins 2008; Gabriele 2009; Gabriele and McClure
2011) without regard to the discrepancy at the level of lexical aspect.
Shirai and Nishi (2002) pointed out the possibility of the discrepancy at
the level of L1 and L2 lexical aspect having an inuence on the learners
form-meaning mapping. Nishi and Shirai (2007) tested this possibility by
investigating L1 English learners of Japanese, and found that when the
lexical aspect diers in the learners L1 and the target language, L2 learners
have diculty at the lower prociency level, indicating that initial form-
meaning association in aspect acquisition is constrained by L1 transfer.
More specically, students had signicantly lower scores in acceptability
judgment tests on the sentences involving imperfective -teiru with verbs
that are stative in English but achievement in Japanese (e.g., to know), com-
pared with verbs for which both languages have the same lexical aspect
(e.g., to fall ).
The L1-L2 discrepancy in lexical aspect has been discussed from meth-
odological perspectives as well. Lardiere (2003) pointed out that learners
L2 production of a particular verb may not be what researchers believe it
to be. When the learners L1 has a dierent lexical aspect than the target
language, even if the researcher classied a particular verb produced
according to linguistic tests, there is no guarantee that the learner is using
it with the same lexical aspectual value, in particular when the L1 and the
L2 have dierent lexical aspectual representations.5

5. Shirai (2007) replied to such methodological concerns stating that we should


balance the danger of overinterpretation and methodological rigor while treat-
ing use of linguistic test as a measure of enhancing replicability.
Lexical aspect 279

Given these concerns about L1-L2 dierences at the level of lexical


aspect, Nishi (2008) used written judgment tasks and oral picture descrip-
tion tasks to investigate whether the L1-L2 discrepancy inuences the
learning of L2 aspectual form. Nishi analyzed the use of Japanese imper-
fective marker -teiru by English, Chinese, and Korean L1 learners. This is
the only systematic study thus far which investigated the eect of cross-
linguistic variation in lexical aspect in the acquisition of tense-aspect
marking. By systematically manipulating L1-L2 discrepancies in the three
source languages and the target language (Japanese), Nishi found, among
other things, that L2 learners have diculty when there is a discrepancy at
the level of surface form-meaning correspondence, and that they nd
it more dicult to correctly judge or produce sentences involving L1-L2
discrepancies in lexical aspect. These studies suggest that L1 inuence
may be much stronger than originally thought. Based on these studies,
Shirai (2009: 184) suggested that the acquisition pattern represented by the
Aspect Hypothesis might at least be partially determined by L1 inuence:

Most of the L1s have corresponding past or perfective marking, for which
the prototype is telic and punctual (cf. Dahl 1985), which may transfer to
L2 learning of past or perfective marking. This may facilitate early acquisi-
tion of past tense with telic verbs. Likewise, acquisition of progressive mark-
ing with activities may be facilitated because in most languages, progressive
marking marks action in progress that is obtained when imperfective aspect
marking is combined with activity verbs, but other meanings (such as habitual,
futurate) are not always present (Bybee, Perkins, and Pagliuca 1994).

In other words, L1 inuence may be working as a constraint on prototype


formation. Learners create initial prototypes for each grammatical category
based on input frequencies characterized by the association of the past/
perfective marker with telic verbs, the imperfective marker with atelic
verbs, and the progressive marker with activity verbs. These associations
may be constrained by the learners L1 when corresponding L1 forms are
available. The strong eect of the L1 may account for the discrepancy in
foreign language learners and naturalistic learners that have been observed
in Spanish SLA. Whereas the ndings from Andersen (1991) and Lopez-
Ortega (2000) based on naturalistic learners were consistent with the
Aspect Hypothesis, Salaberrys (1999, 2011) ndings, based on classroom
learners, were not. This is probably because learners in the foreign lan-
guage environment are more inuenced by the L1, as Salaberry notes,
280 Yasuhiro Shirai

and thus they treat Spanish Preterite as if it were an English past tense
marker. In naturalistic settings, in turn, this past tense marker can be
more freely applied to atelic verbs and given that learners are more expe-
rienced with the target language, they begin to create semantic proto-
types.6 A recent study by Martelle (2010, 2011) on Russian as a foreign
language shows that learners treat the imperfective past tense form as
a marker of default past. This may plausibly be due to the fact that in
Russian the imperfective is morphologically unmarked and it is also
taught rst, thus learners may associate it with English simple past tense
marking.
The eect of the L1 is not always apparent because when L1 semantics
is a subset of L2 semantics, it does not result in errors. It may, however,
covertly inuence the learners form-meaning mapping. Only through
careful comparison of L1-L2 mappings in lexical aspect, grammatical
aspect (and tense), and the interaction between the two can we notice subtle
L1 eects. In sum, the crosslinguistic discrepancy of lexical aspect (Shirai
and Nishi 2002; Lardiere 2003) is an area that calls for more attention
and further research.

4. An example of classication procedure

Although many studies (see Table 1 above) report operational tests for
classifying verb tokens into lexical aspect classes, it is rare that they discuss
the detailed procedures to generate those classications. However, as those
who have tried actual coding of verbs know well, even if one uses linguistic
tests, there are many cases in which it is not easy to make coding decisions.
This is not surprising because linguistic semantics cannot be neatly classi-
ed into distinct categories given that linguistic categories have prototype
structures (e.g., Lako 1986), and because it is not easy to determine the
exact reference that speakers, especially L2 learners, have in mind.

6. Not just learning environment, but task type may also inuence the adherence
to the predictions of the Aspect Hypothesis (AH). The use of naturalistic tasks
(such as open ended conversation) give results that are more consistent with
the AH, whereas data prompted with the use of elicited narrative are less con-
sistent with the AH (Bonilla, in press).
Lexical aspect 281

In what follows, I will discuss details of how the verb classication


system used in Shirai and Andersen (1995, based on Shirai 1991) was per-
formed to illustrate the actual process of lexical aspect classication and
diculties involved. This classication system has been used by other
researchers both functionalists (e.g., Weist, Pawlak, and Carapella 2004;
Bayley, Alvarez-Calderon and Schecter 1998) and generativists (e.g.,
Gavruseva 2002, 2003). It has also been used, with minimal adaptation,
in the analysis of other languages, including Japanese (Shirai 1993, 1998a),
Turkish (Aksu-Koc 1998), Spanish (Salaberry 1999), Chinese (Jin and
Hendriks 2003; Chen and Shirai 2010), and Korean (Lee and Kim 2007).
Thus, we can treat Shirai and Andersens test as a good system that set the
standard in this area, and thus it is worth examining it in detail. First,
I will outline the procedure for coding and its motivation.7 I will then
discuss the choice of particular tests in Shirai (1991), and elaborate on
how aspectual systems as a linguistic choice is viewed in this classication
system. Finally, I will discuss some problematic cases that were encoun-
tered in coding, and how they were resolved.

4.1. Procedures to classify lexical aspect


One of the weaknesses in many previous studies in the past (and some
current ones) has been the lack of precise descriptions of the procedures
for determining lexical aspect. Prior to Shirai (1991), there had been
only two published acquisition studies that used operational tests (Weist,
Pawlak, and Carapella 1984; Robison 1990). The problem with this lack of
published procedures is that it aects the replicability and comparability of
the studies because dierent studies may be classifying the same verbs into
dierent categories.8 This problem is more common in the classication of
borderline cases, such as posture verbs, e.g., to sit, which can be classied
as a state or an activity (to be discussed below).

7. No study in linguistics that I know of has coded discourse in terms of lexical


aspect and discussed the problems involved except for Xiao and McEnery
(2004). It should be noted that R. Andersen and his students (including Robison
1990), at least since 1986, have been coding lexical aspect using operational
tests (e.g., Cushing 1987; Gonzales 1989, 1990).
8. Only a handful of studies reported the reliability of the Vendlerean classica-
tions (Cziko and Koda 1987, Shirai 1994, and Shirai and Andersen 1995, who
reported intra-rater reliability).
282 Yasuhiro Shirai

Figure 1. Schematic representation of the classication procedure

Shirai (1991) used the procedure represented in Figure 1 to classify verb


tokens into the Vendler-type lexical aspect categories. There are three
stages for coding lexical aspect. First, we need to read a small subset of
discourse until we are reasonably sure about the interpretation of the dis-
course segment. Second, we determine if repetition is involved, and code
such instances as either unitary, iterative, habitual, or iterative-habitual
following the criteria below:

Criteria for iterativity


The criterion for iterativity is simply whether the situation referred to is a
single unitary state/event/process, or not. For example, He coughed once
at noon yesterday, He sang for a few minutes yesterday, and He loved
Mary are coded as unitary, whereas he coughed for a few minutes is coded
as non-unitary (i.e., repeated) because there is repetition of the action of
coughing. There are three subcategories for repeated:
Unitary He coughed once at noon yesterday.
Repeated (a) Iterative He coughed for a few minutes.
(b) Habitual He walked to school for a month.
(c) Iterative-habitual He coughed for a few minutes every day.

As dened in Brinton (1988: 54), iterative portrays actions repeated on


the same occasion, (e.g., Hes kicking a ball outside right now) whereas
habitual portrays actions repeated on dierent occasions (e.g., Hes
kicking a ball for an hour every night these days).9 The category iterative-
habitual is to represent actions that are repeated not only on multiple occa-
sions but also on each of these occasions (to be discussed in more detail

9. I am aware that the distinction between the two can be murky. See Salaberry
(2008: 8185) for further discussion.
Lexical aspect 283

below). Finally, the third stage of the classication process requires that
we remove grammatical aspect/tense from the sentence, and then apply
the following operational tests (outlined in Steps 1 to 3) to determine its
lexical aspect.10
Operational tests for lexical aspect (Each test is used only in the clauses
remaining after the preceding test):
Step 1: State or non-state? Does it have a habitual interpretation in simple
present tense?
If no ! State (e.g., I love you.)
If yes ! Non-state (e.g., I eat bread.) ! Go to Step 2
Step 2: Activity or non-activity? Does X is Ving entail X has Ved
without an iterative/habitual meaning? In other words, if you stop in the
middle of Ving, have you done the act of V?
If yes ! Activity (e.g., run)
If no ! Non-activity (e.g., run a mile) ! Go to Step 3
Step 3: Accomplishment or achievement? If test (a) does not work, apply
test (b), and possibly (c).
a) If X Ved in (Y time; e.g., 10 minutes), then X was Ving during
that time.
If yes ! Accomplishment (e.g., He painted a picture)
If no ! Achievement (e.g., He noticed the picture)
It needs to be noted that it is possible to say X was Ving even right
after X began the action that led to the goal.

10. What needs to be coded is lexical aspect, not lexical aspect combined with
grammatical aspect. That is, morphology needs to be removed. Let us take
an obvious example from Japanese discussed above. In example (1), sin-deiru
be dead refers to a state of someone being dead, but that is not what needs to
be coded. What needs to be coded is sin(u) to die, i.e., the verb stem without
grammatical aspect markers, which is an achievement. Thus, in the context of
English, if the learner says hes walking to school we apply the tests (Steps
1,2, and 3) to the verb constellation [he walk to school].
284 Yasuhiro Shirai

b) Is there ambiguity with almost?


If yes ! Accomplishment (e.g., He almost painted a picture has two
readings; i.e., He almost started to paint a picture and He almost
nished painting a picture)
If no ! Achievement (e.g., He almost noticed the picture has only
one reading.)
c) X will VP in (Y time; e.g., 10 minutes) X will VP after (Y
time).
If no ! Accomplishment (e.g., He will paint a picture in an hour is
dierent from He will paint a picture after an hour, because the former
can mean He will spend an hour painting a picture, but the latter does
not.)
If yes ! Achievement (e.g., He will start singing in two minutes can
have only one reading, which is the same as in He will start singing
after two minutes, with no other reading possible.)

4.2. Tests for lexical aspect


The linguistic tests to operationalize lexical aspect categories rely on a
number of studies on the Vendlerean classication: Brinton (1988), Comrie
(1976), Dowty (1979), Lys and Mommer (1986), Mourelatos (1981), Robison
(1990), Sag (1973), and Vendler (1957). Dowty (1979) was especially in-
formative. One of the important observations that Shirai (1991) made in
using various tests to classify tokens in actual data is that acceptability
judgment tests in general are not very eective. In discussions of lexical
aspect in linguistics, the examples used are very clear cases, due to their
invented nature. In the actual classication of discourse data, however,
there are many unclear, borderline cases for which acceptability judgments
are dicult. This is, in a sense, inevitable since it has been observed that
acceptability judgment tests often create disagreements among raters (Labov
1975).
Let us take an example of one of the stativity tests often used: disjunc-
tion of progressive with stative verbs. According to this test, a verb is to
be coded stative if it is ungrammatical when the progressive marker is
applied. However, this is not always as clear-cut as the test seems to
indicate. Brown (1973: 323) heard the following sentence on an airplane:
(3) Are you wanting your suitcase down?
Lexical aspect 285

Sentence (3) is indeed a problematic case, and grammaticality judgments


may dier from coder to coder. Most of the tests Shirai chose do not rely
on grammaticality judgments but on the interpretation of sentences. Inter-
pretation of sentences should not be as fuzzy as grammaticality judgments.
For example, [want your suitcase down] is hard to classify if we use an
acceptability judgment test, but if we use an interpretation test (Does it
refer to habitual interpretation in simple present tense?), it is relatively
easy to determine that it is a state because [want your suitcase down]
does not have to involve habitual interpretation. Given enough informa-
tion for the interpretation of the clauses in question, its semantic interpre-
tation can be applied in a categorical fashion more easily than giving
acceptability judgments of borderline cases.11
Another criterion for choosing a test is the exclusion of agentivity. It is
often the case that agentivity is mistakenly regarded as an aspectual value
as pointed out by Brinton (1988) and Verkuyl (1989). This is inevitable
because there is a very strong correlation between high stativity and
low agentivity (Sag 1973). In Shirai (1991) a stativity test which renders
temporal aspectuality, not agentivity, was chosen: Does it have a habitual
meaning in simple present tense? As discussed in the previous paragraph,
[John run] involves a habitual interpretation, and therefore it is a non-
state (i.e., dynamic) whereas [John live (in Tokyo)] does not involve repe-
tition, and thus it can be classied as stative.

4.2.1. Situation, lexical aspect, and viewpoint aspect


In the classication scheme just described, it is assumed that it is possible
to separate grammatical (viewpoint) aspect from a sentence when coding
the lexical aspect. Let us take an example to illustrate this point. When
speakers want to describe a real-world situation, such as [Mary was run-
ning at 10 a.m. on January 5th 2009, with a view to running a mile]
(speech time: January 6th 2009), they have a variety of choices. They might
say one of the following, for example:
(4) Mary ran (yesterday).
(5) Mary was running.
(6) Mary was running a mile.

11. This may be due to the fact that in our everyday life we do not make a gram-
maticality judgment, whereas we always have to interpret sentences.
286 Yasuhiro Shirai

If the speakers goal is just to assert the fact that Mary ran (yesterday), (4)
will suce. If he wants to add more dynamicity/vividness to the assertion,
he can choose to impose a viewpoint aspect (in this case, progressive), as
in (5).12 If he wants to additionally assert that Mary was in the process of
running a mile, he can choose (6). The speaker makes these choices instan-
taneously. The point here is that verbalization, if the utterance contains a
verb phrase, involves (at least) two aspectual choices: which words to use
(lexical aspect), and which verbal morphemes to use (viewpoint aspect). In
terms of lexical aspectual choice, (4) and (5) are the same (activity), while
in terms of viewpoint aspectual choice, (5) and (6) are the same (progres-
sive). The hearer, then, uses the utterance and contextual information
(both linguistic and extralinguistic information) to interpret the utterance.
This process, leading to the formulation of the utterance and its interpre-
tation, is schematically represented in Figure 2.

Figure 2. Structure of the speakers utterance and the hearers interpretation

Given the above structure of the speakers aspectual choices and the
hearers interpretation, how are we to code sentences (4)(6)? What we
need to code is the lexical aspect that describes the real world situation.
Therefore, it is irrelevant that the situation being described is the same
for (4) to (6); what matters is the kind of aspectual choices made to
describe the situation. The choice made by the speaker as to what lexical
items should be used to describe the situation is important, and this is
what we need to code. We can use [Mary run] for (4) and (5) and [Mary
run a mile] for (6) as default values without grammatical aspect marking.
Then operational tests can be applied to these default values.

12. It is also possible to say that in (4) the speaker is choosing a viewpoint aspect,
which is the default (ran: simple non-progressive past).
Lexical aspect 287

Why do we need to rst read and interpret a segment of the transcrip-


tion before coding for lexical aspect? This is because the appropriate inter-
pretation is very dicult without information provided by the morphology,
the discourse context, and the comments on extralinguistic contexts pro-
vided by researchers who gathered the data. Consideration of both linguistic
and extralinguistic information is essential for classication because some
tokens with the same linguistic form can be interpreted in more than one
way. For example, [open the box] can be an achievement or an accom-
plishment depending on the kind of box being opened. If the box is the
kind that can be opened instantaneously, it is an achievement; if it is care-
fully wrapped and takes time to open, then it is an accomplishment (Shirai
and Andersen 1995). The information which makes it possible to reach
this level of understanding of [open the box] is necessary to obtain a
reliable classication. On the other hand, this does not mean that what is
coded are real-world situations. The reason we cannot code [open the
box] by itself is that in this case its lexical aspect is ambiguous without
contextual information. Here, situational information is used to disam-
biguate lexical aspect.
Let us take another example from Verkuyl (1989). He suggests that
walk a mile and walk a foot, which are usually considered accomplishment
and achievement predicates, respectively, can be quite dierent if the actor
is a giant or a dwarf. Walk a mile could be an achievement for a giant;
walk a foot an accomplishment for a dwarf. This example illustrates why
it is necessary to know exactly what the situation is that is being described
by the lexical aspect. As pointed out earlier, what we are coding is lexical
aspect that describes the real world situation, not lexical aspect without
regard to the situation.13 Robison (1990), among others, removed verbal
morphology from the transcripts when he coded lexical aspect to ensure
that morphology did not inuence coding of lexical aspect. For example,
progressive morphology tends to give the coder the impression that the
item being coded is an activity because of the strong association of both
activity and progressive with dynamic duration. I believe it is best not to

13. Verkuyl (1989) used these examples to argue that the achievement versus accom-
plishment distinction is not useful. This is too static a view of lexical aspect.
He considers that it is not necessary to make a linguistic distinction between
walk a mile and walk a foot in view of the fact that each can be an achieve-
ment or an accomplishment depending on the actor (dwarf versus giant). In
my scheme, walk a mile changes its lexical aspectual value depending on the
actor because what is linguistically important is lexical aspect that describes a
situation.
288 Yasuhiro Shirai

remove verbal morphology from the transcripts, considering that it is very


dicult to arrive at an appropriate interpretation of the situation being
described without verbal morphology, and that by applying operational
tests in a principled manner, the bias can be minimized. In fact, there are
advantages and disadvantages to deleting morphology. I prefer to take
morphological markers into account to ensure accuracy of interpretation.14
Another important point regarding the reliability of coding is the treat-
ment of time adverbials. How should the following sentences taken from
Chung and Timberlake (1985: 217) be coded?
(5) John painted until the sun went down.
(6) John painted from morning till night.
(7) John painted seventeen houses within three days.
If we apply the operational test If you stop in the middle of V-ing, have you
Ved? to these sentences, they are not activities but accomplishments. How-
ever, these types of sentences were coded as activities in Shirai (1991), not
as accomplishments. Since time adverbials often function as a perspective-
taker (Talmy 1988), and therefore function like viewpoint aspect, they
were not included in the verb constellation to be coded for lexical aspect.15
Exclusion of time adverbials was also practically motivated because one of
the tests in step 3 has in X time and during X time; if the test is
applied to [He break it in 3 minutes], the test will become He broke it in
3 minutes in 5 hours, which is absurd.16
It should be noted that in coding mother-child interaction, it was rarely
necessary to make this kind of decision, because time adverbials were
extremely rare. Probably in simple discourse such as mother-child interac-
tion, it may not be necessary to set the event frame (Chung and Timber-
lake 1985), using time adverbials since event frames may be implicit in the

14. One can also remove morphology from transcripts and just look at morphology
when it is necessary to disambiguate (Chapter 9; Comajoan 2001). I think this
is a reasonble approach as well. In any case, it is not easy to determine which
is a more serious threat to the study the danger of misinterpretation (by
removing morphology) or the bias in coding of lexical aspect (by keeping
morphology).
15. Talmy (1988: 183) points out that the time adverbial at exactly midday
(e.g., She climbed a ladder at exactly midday) involves a cognitive operation
of reduction whereas as we watched (e.g., She climbed a ladder as we
watched ) involves magnifying.
16. Of course, time adverbials were included in the interpretation of discourse. It
is just that they are not included in the verb constellations to be coded.
Lexical aspect 289

discourse. This, however, must be quite dierent for adult second lan-
guage acquisition, because L2 learners are known to establish temporal
reference using lexical information (Bardovi-Harlig 2000).

4.2.2. Unitary or repeated situation


The distinction between unitary situations and repeated situations may
have some relevance in terms of the acquisition of verbal morphology.17
To briey summarize, unitary situations usually exhibit an unmarked
association between lexical aspect and grammatical aspect/tense; i.e., (1)
achievement/accomplishment with perfective or simple past, (2) stative/
activity with imperfective, and (3) activity with progressive (Andersen
and Shirai 1994). On the other hand, repeated situations do not necessarily
show these associations. For example, for languages with a perfective/
imperfective distinction, if a situation is habitual, it is more typically ex-
pressed with an imperfective marker, even if accomplishment/achievement
verbs are used. In terms of the Distributional Bias Hypothesis (Andersen
1993; Andersen and Shirai 1994), for the hypothesis to be valid, the un-
marked association (i.e., perfective/past morphology with accomplishment/
achievement verbs, and progressive morphology with activity verbs) must
be kept strong in the input.
There is some confusion regarding the dierence between iterative and
habitual (Comrie 1976), which I clarify here because the distinction is
important for the analysis of lexical aspect in discourse. For Comrie,
iterative is simply a cover term for repeated actions, whereas habitual de-
scribes a situation which is characteristic of an extended period of time
(Comrie 1976: 2728). Therefore, He kicked a ball for an hour yesterday
(in describing a soccer players practice) is iterative, but not habitual.
He walked to the hospital for a month, on the other hand, is both iterative
and habitual for Comrie. Chung and Timberlake (1985) also use iterative
as a cover term for repeated situations, without using the term habitual.
Bybee (1985) and Brinton (1988), however, restrict iterative to repetition
on a single occasion. Brinton (1988: 54) states that Habitual may be dis-
tinguished from iterative, because the habitual portrays actions repeated
on dierent occasions, while the iterative portrays actions repeated on
the same occasion. Therefore, He kicked a ball for an hour yesterday is

17. See also Talmys (1988: 176) notion of plexity. Uniplex corresponds to uni-
tary, and multiplex to repeated, though his notion encompasses nominal
categories too.
290 Yasuhiro Shirai

iterative, whereas He walked to the hospital for a month is non-iterative


and habitual. I follow Bybees and Brintons denition because it is neces-
sary to make a distinction between repeated events that occur on one
occasion and those that are spread over many occasions. He is kicking
the ball, for example, should be qualitatively dierent from He is going to
school by bus (these days), although both of them involve repetition. In
sum, iteratives are taken to mean repeated actions on a single occasion,
whereas habitual refers to repeated actions on dierent occasions, which
(the speaker believes) persists for an extended period of time (see Huang
1999 and Shirai 2002 for L2 studies on the marking of habituality).18
It should be noted that most of the previous studies on lexical aspect
(e.g., Dowty 1979; Mourelatos 1981) treat [Verb indenite plural noun]
(or, to be more precise, nouns whose quantity is not specied, see Brinton
1988; Verkuyl 1989) as an activity:
(8) He ran for/*in three hours. (activity)
(9) He built houses for/*in three months. (activity)
However, in Shirai (1991), build houses in sentence (9) was treated as a
repeated situation (in this case, habitual) of an accomplishment at the
lexical aspectual level, not as an activity. Because the distinction is made
between unitary and repeated situations, it is possible to make a ner
aspectual distinction, e.g., habitual accomplishment, habitual achievement,
etc., which in other studies may be treated as activities.19 This distinction
between pure activity (e.g., play the guitar), which is unitary, and an activity
consisting of repeated actions (e.g., build houses) may be important for
L1 and L2 learners acquiring verbal morphology. For example, it may be
easier to acquire progressive morphology for a unitary activity [play the
guitar] than for a repeated habitual activity [build houses], and if we code
them both as activities, we cannot dierentiate the two.

18. Comrie (1976: 27) stated that a situation can be referred to by a habitual
form without there being any iterativity [i.e., repeatedness in my terminology]
at all, citing examples such as Simon used to believe in ghosts. It should be
noted that all the examples he gave to support his point involve stative verbs.
This suggests that statives can be habitual without repetition. In my classica-
tion, however, I chose to regard states that last over an extended period of
time as unitary, because my focus is not on extendedness of a situation, but
on whether there is repetition or not.
19. Camps (2002), for example, classied multiple event activities as activities.
Lexical aspect 291

4.3. Problematic cases in classication


Even with a sucient understanding of the speakers intent, there are
many diculties in applying operational tests for lexical aspect. In this
section, I discuss some of the problematic cases, and I provide some sug-
gestions on how to code them. These choices are made so that the classi-
cation can be of use in addressing pertinent issues in the area of language
acquisition (see Chapter 11 for an alternative approach).

4.3.1. Accomplishment or achievement


The distinction between accomplishments and achievements is especially
problematic. Shirai (1991) chose an entailment test from Dowty (1979)
for this distinction (i.e., If X Ved in Y time, then X was Ving during
that time period; see Step 3, test a). For example, He painted a picture in
two hours entails He was painting a picture during those two hours; there-
fore, [paint a picture] is an accomplishment. On the other hand, He noticed
the picture in two minutes does not entail He was noticing the picture during
two minutes; therefore [notice the picture] is an achievement. These are
clear-cut cases and not problematic. However, things are not always so
easy.
Let us take an example, reach the summit, which is often cited as an
example of an achievement. He reached the summit in two hours appears
to entail he was reaching the summit during those two hours. But in this
case, the meaning of he was reaching the summit is dierent in that it de-
notes the process leading up to the endpoint meaning (or preliminary
stages meaning, Smith 1991), that is, he was about to reach, on the verge
of reaching, or getting closer to the goal of reaching the summit (he was
not actually reaching the summit throughout those two hours). X was
Ving during that time in the test implies that X was Ving all the time
during that time. In other words, if reach the summit is an accomplish-
ment, then it should be possible to say He is reaching the summit at any
point during those two hours, even right after he started climbing the
mountain, which is possible for more prototypical accomplishments such
as [make a chair]. With reach the summit being an achievement, it is odd
to say He is reaching the summit if he has just started climbing the moun-
tain. On the other hand, it is perfectly all right to say He is running a mile
right after he has started running. Thus, if test (a) is hard to apply, one
additional criterion can be considered: Can you say X is V-ing right after
X started doing Y? If yes, the predicate is an accomplishment.
292 Yasuhiro Shirai

Another diculty in making an accomplishment versus achievement


distinction is that no single test can handle all the sentences to be coded,
which is the reason why three tests are necessary for this step, and possibly
more. First, test (a) cannot handle motion verbs that have both a starting
point and an endpoint and duration between them in real world situations
(e.g., come/go to school, bring something to a location). For example, He
went to the beach is not felicitous if test (a) is applied: ?He went to the
beach in two hours. This arises perhaps because the sentence, if acceptable,
can mean two things: (1) He left for the beach two hours after some refer-
ence time in the past, or (2) it took him two hours to get to the beach.
That is, it can refer to (1) a starting point focus versus (2) an endpoint
focus. These kinds of cases necessitate the application of test (b), ambiguity
with almost . There is no ambiguity for He almost went to the beach
because another interpretation (He almost got to the beach) is impossible.
Then, why not use test (b) only? Test (b) also has its own problem,
because if the sentence under analysis does not involve an action that the
subject can control (i.e., something the subject can do if he decides to do
it), the test does not work. For example, John, who is physically challenged,
almost walked 20 feet can have only one reading despite the fact that it is an
accomplishment, because walking 20 feet is not something John can do if he
decides to do it. Therefore, ambiguity with almost is absent, not because of
the temporal contour of the action, but because of some other semantic
elements (in this case perhaps self-controllability). Test (c) was used only
when neither test (a) nor (b) worked (as in the cases discussed above). In
other words, test (c) X will VP in (Y time; e.g., 10 minutes) X will
VP after (Y time) was used as a last resort when neither test (a) or (b)
works.
I discuss below how I coded say and come/go, which are particularly
problematic in terms of an accomplishment/achievement distinction.
These are high-frequency verbs in most languages and therefore how they
are coded has strong bearing on quantitative results, especially when a
token count is used.
Say X. Say often appeared in the data, as in She said Im hungry or
She said she ate breakfast. If the level of situation is being coded, say X
should probably be regarded as an accomplishment in that there is
duration that leads up to the point of nishing saying the sentence Im
hungry. However, what is coded here is lexical aspect. As Smith (1983)
claims, lexical aspect is the idealization of situations, so it is not crucial
whether it takes time before you nish saying something. In other words,
Lexical aspect 293

lexical aspect is how each predicate is idealized in our linguistic system.


The notion of saying X can be lexicalized as an achievement in one lan-
guage (e.g., English), but may be lexicalized as an accomplishment in
another language (e.g., French, Dunn 1998). I suggest that say X is not
an accomplishment but an achievement, because when we use say X
we rarely refer to its duration. Rather, we reduce it to a single point, as I
will argue below. Let us apply Step 3 (the accomplishment/achievement
test) and discuss the problems involved.
Test (a): If she said Im hungry in two seconds, then she was saying
Im hungry during that two seconds. This appears to hold, which suggests
that say X is an accomplishment. But do we talk about the duration of our
utterances? We do not normally refer to the duration (1 second, 2 seconds,
etc.) of someones utterance. If these cases are taken to be accomplish-
ments, because in some sense they last, then we would also have to code
She said Sorry or She said No! as accomplishments, even though
they last very little (We can say She said Sorry in 0,8 seconds). However,
sentences like Say Sorry are generally coded as an achievement.
Test (b): She almost said Im hungry. There appear to be two inter-
pretations. One interpretation is She almost said Im hungry but did not
say it. The other interpretation is She did start to say Im hungry, but
stopped in the middle and ended up saying Im hung . . .. However, this
second interpretation is very unnatural for the sentence She almost said
Im hungry.
Test (c): She will say Im hungry in two seconds is not necessarily the
same as She will say Im hungry after two seconds, because it is possible
that the former can also mean that she will complete her utterance within
two seconds, which argues for say X being an accomplishment. But we
normally do not talk about the duration of utterances. This becomes
apparent if we extend the time-frame from two seconds to two minutes.
She will say Im hungry in two minutes can only mean She will say
Im hungry after two minutes, because Im hungry would normally be
uttered in a matter of seconds, which suggests that say X is an achieve-
ment rather than an accomplishment.
In sum, if we refer to the duration of the utterance Im hungry, which
should take no more than a second or two, then it can be taken as an
accomplishment. In other words, if we treat the utterance Im hungry
as having duration, and therefore internal structure, then it can be an
accomplishment. On the other hand, if we do not focus on its duration but
take the utterance as being reduced to a point, then it should be treated as
an achievement. Then, what is our conceptualization of say X? Do we see
294 Yasuhiro Shirai

its internal structure, or do we see it as totality reduced to a point? As


noted above, Smith (1983, 1991) claimed that lexical aspect is an idealiza-
tion of a real world situation. Do we idealize say X as an achievement, or
as an accomplishment, i.e., having duration? When we quote someone by
using the verb phrase say X, we do not often quote such a long utterance
that we focus on its duration. This probably creates the idealization of say
X, whose lexical aspect is best characterized as an achievement.
Finally, the interaction of lexical aspect and grammatical aspect also
supports my classication. A punctual event, if combined with progres-
sive, gives either an iterative (He is punching a bag now) or a preliminary
stage sense (He is reaching the summit) (see also Quirk et al. 1985: 208
209). On the other hand, accomplishments, which refer to durative events,
with the progressive yield a meaning of action in progress. When we hear
a sentence in the progressive as She is saying shes hungry, what inter-
pretation do we have? Certainly, we do not sense an action in progress
(i.e., She is in the middle of saying shes hungry), because it is practically
impossible, at least in our daily life, to say She is saying shes hungry in the
middle of her utterance Im hungry. We usually interpret this as
iterative: she must be repeatedly saying Im hungry probably asking for
food. Alternatively, it is also possible that this sentence conveys that the
persons state of being hungry still persists at the moment of this utter-
ance.20 This eect comes from the imposition of duration by viewpoint
aspect. In any event, the action-in-progress sense is not normally possible.
Therefore, say X (X content of the utterance), should be coded as an
achievement, not as an accomplishment.
The above discussion does not mean that say is always an achievement.
For example, say a prayer is an accomplishment, as it typically has dura-
tion. When he was saying a prayer can have action in progress meaning,
and He said his prayer in 2 minutes entails He was saying his prayer during
the period of two minutes.
Based on the above considerations, I coded say X as punctual. This,
however, does not mean that I am claiming with absolute condence that
say X is categorically an achievement. This is an operational denition
taken for the purpose of acquisition research. In the end, the achieve-
ment/accomplishment distinction is not clear-cut when compared to other
distinctions.

20. This is what is observed in the analysis of Japanese conversational discourse


(Shirai and Nishi 2005).
Lexical aspect 295

Come/go. Another major problem concerns the coding of the deictic verbs
come and go. If what is coded is the situation, come/go to X, it might as
well be coded as an accomplishment because there is duration between
leaving the starting point and reaching the goal of coming and going. But
again, what is being coded is lexical aspect; thus, it is necessary to go
through the accomplishment/achievement tests. Because tests (a) and (c)
do not work (He went to the beach in 10 minutes, as discussed above, is
not acceptable for many native speakers of English), we can only rely on
test (b).
Test (b): He almost went to the beach. This has only one reading: He
almost decided to go to the beach, but he did not after all. He almost went
to the beach cannot mean that he almost got to the beach. It appears that go/
come to the beach reduces the action of going to the beach to a single point
in time. Therefore, test b shows that go to the beach is an achievement.
I tried three additional tests often used as tests for durativity: stop
Ving, nish Ving, and studiously/attentively/carefully Ved, which are
not compatible with punctual verbs because these tests requires some
duration (Dowty 1979). For example, one cannot stop doing something
unless there is some duration in this particular action. All these tests,
therefore, should result in anomalies for an achievement, but not for an
accomplishment. In fact, come/go to the beach results in an anomaly in
all three of them. Therefore, they should be taken as achievements, at least
in terms of classication.21

4.3.2. Habitual or stative


Another problematic distinction is that of stative versus habitual; namely,
the distinction between state and the cases of achievement, accomplish-
ment, and activity treated as habitual. Here, it should be made clear that
habitual is not lexical aspect. Therefore, in my classication scheme, this
distinction has to be made at the rst step (coding for repetition: habitual,
iterative, iterative-habitual, or unitary).
Many scholars observed similarities between states and habituals (e.g.,
Smith 1991, who calls habituals as derived states), but they are just similar
and not the same thing, which is also acknowledged in the literature. My

21. These three additional tests were used only when a classication was dicult
after applying the three regular tests, and they were not used very frequently.
These tests were not included in the regular tests, because they are accept-
ability judgment tests. As additional means for classication, however, they
are useful.
296 Yasuhiro Shirai

point in this section is not to address these similarities or dierences, but


to discuss how to deal with borderline cases between the two: Real stative
predicates (e.g., I love you) and derived stative predicates that are in fact
repeated dynamic situation (e.g., I watch TV all the time, which is habitual
repetition of an activity). Having said that, this is probably the case of
forcing a clear-cut distinction even though there is not such a clear line to
be drawn. Let us take an example:
(10) I call him Jack, even though his name is John.
Should I call him Jack be categorized as a state or a habitual of an achieve-
ment? Each occurrence of call him Jack is an achievement, but does (10)
refer to each act of calling him, or does it only refer to a general state of
my referring to him as Jack. What if I call him, say, Condor (his nick-
name)? Would it be a state, or a habitual of achievement? The distinction,
as suggested above, would not be clear-cut. In our mind, there would
probably be a continuum. At one end, we have a concrete instance of
actually calling somebody (as in Hey, Jack!), and at the other end, we
have a highly abstract conceptualization of the state of people referring
to him as Jack.
Assuming that there is this continuity, how should we code call in (10)?
In my classication, I treat this type of call as habitual because actual call-
ing is involved.22 This type of call should be quite dierent from unitary
states such as think, have, want, etc., which learners may encounter at
early stages, and which cannot have a habitual interpretation when Step
1 of the tests of lexical aspect is applied. By classifying this class of predi-
cates as habitual, we can avoid collapsing it with other state terms.

4.3.3. Activity or achievement


There is a class of verbs that should be coded as activities according to the
operational tests, but which proved somewhat problematic and, therefore,
needs special consideration. I will call this class resultative activity. In
fact, the verb by itself signies a punctual action, but if used with the pro-
gressive, the verb describes the resulting activity. This class of verbs is
analogous to some stative verbs showing entry into state (e.g., Now I

22. Therefore, say used in such a sentences as What does it say? when refer-
ring to the instructions on a pamphlet for a toy, etc, is not treated as habitual,
as it does not actually say anything verbally. It is interpreted as indicate,
which is state.
Lexical aspect 297

know it!) except that it shows entry into an activity.23 Let us review the
following pairs of examples:
(11) a. He looked at the table.
b. He is looking at the table.
(12) a. He touched the wall.
b. He is touching the wall.
(13) a. He kissed the girl.
b. He is kissing the girl.
In all these pairs, the past form describes a punctual event, whereas the
progressive form expresses the activity or state that has arisen as a result
of the punctual event.24 It should also be pointed out that these pairs can
be conceptualized as iteration of punctual events (especially 12.b and
13.b), because if somebody is kissing somebody, it could be interpreted as
an iterative event (multiple kisses). This argument depends entirely on the
real world situation. For example:
(14) *MOT: Nomi-is touching the recorder.
(Naomi, 2;1)
This utterance, by the mother of Naomi in Sachs (1983) data, is coded as
an iterative of an achievement verb, not as an activity, because Nomi
(Naomis nickname) must have touched the recorder several times, in
dierent ways, out of curiosity. If Nomis hand is static on the tape
recorder, it should be coded as an activity in that it passes the test of an
activity (If you stop in the middle of touching, have you touched, without
iteration?). In the same vein, he touched the wall should be coded as activity,
unless there is no indication that he touched the wall time and again.25
Another problem for this class was pointed out by Talmy (1985) with
the following sentences (Talmys 39.a and 39.b):
(15) a. He hid in the attic for an hour.
b. He hid in the attic when the sheri arrived.

23. For a more theoretical discussion of this class of verbs, see Shirai (1998b) and
Onozuka (2008).
24. Actually, sentence (a) of each pair can be an activity, for example, if for X
time is attached. But by default, the interpretation should be that of a punc-
tual event of entry into the resulting activity.
25. The verbs that belong to this class include: hang, look (at), sit (down), stand
(up), touch, hold, lie (down), hurt, hug, point to, stand up, and kiss.
298 Yasuhiro Shirai

This example shows that a particular term [hide in the attic] can be used
eectively to describe situations with dierent aspectual values: an activity
in (15a) and an achievement in (15b). However, since time adverbials are
excluded from coding, and only he hid in the attic is applied to the opera-
tional test, the above examples are coded in the present schema both as
activities.
This appears to create a problem if we are coding real world situations
for the purpose of investigating how learners encode the mapping between
real world situations and grammatical tense/aspect morphology. How-
ever, what we are coding is lexical aspect. To examine the consequence of
the decision to code lexical aspect only at the level of language, it is neces-
sary to separate two closely related, but dierent research questions in
relation to the Aspect Hypothesis. They are:
(16) Do learners use past marking more often to describe punctual verbs?
(17) Do learners use past marking more often to describe punctual
situations?
Sentence (16) investigates the relationship between verb morphology and
lexical aspect, whereas (17) investigates the relationship between verb mor-
phology and real world situations.26 Intuitively, the results should be very
similar, because punctual events are most probably described by inherently
punctual terms. The problem arises when we have a discrepancy between
real world situations and lexical aspect. For example, he looked at me
when I screamed is punctual (i.e., instantaneous shift of look) in terms of
real-world situation, whereas its lexical aspect is an activity. The frequency
of such discrepancy cases determines the degree of condence to which we
can extend the results of the study which concerns (16) (the relationship
between verb morphology and lexical aspect) to the research question in
(17) (the relationship between verb morphology and situational character-
istics). If such discrepancy cases are not frequent, we can safely extend the
answer for (16) to (17); if not, we need to be cautious.27
To see the eect of discrepancy between the two, it is necessary to code
cases when such discrepancies are found. Shirai (1991) used the following
denitions to classify the temporal characteristics of the situation:

26. It appears that the disagreements among researchers (Bickerton 1989; Cziko
1989; Weist 1989) stem from the lack of a distinction between the two, at least
in part.
27. Such discrepancies may be more frequent in other languages. For example, in
Japanese speakers often refer to stative situations by attaching imperfective
-teiru to achievement verbs (Shirai 2000), as noted above.
Lexical aspect 299

state: situations that last without any input of energy


activity: situations that have duration, but do not have an inherent end-
point
accomplishment: situations that have both duration and an inherent end-
point
achievement: situations that occur instantaneously

With this additional coding at the level of situations, the discrepancy


between lexical aspect and situational characteristics can be investigated.
Incidentally, Shirai (1991) found that there was not much dierence
between the two coding systems for L1 acquisition in English. This of
course does not guarantee that they yield similar results in L2 acquisition
or in the acquisition of other languages.28
Another diculty in making distinctions between achievements and
activities, which is very dierent in kind from that discussed above, in-
volves verbs such as wave, gallop, and shake, which appear to be iteratives
of an achievement verb, in the sense that these verbs describe an action
which consists of a series of punctual events. In fact, Quirk et al. (1985)
categorize the branches were shaking and the tops of the trees were wav-
ing together with hit and blink (achievement, or in the ve-way classica-
tion of Smith 1991, semelfactive). These were, however, coded as activities
in Shirai (1991). The dierence is that in the case of a punctual verb
(achievement or semelfactive), it is possible to have one jump, one blink,
etc., since they are by denition punctual and can occur instantaneously,
whereas it is impossible for a tree to shake once and instantaneously, or
at least we do not conceive shake in that way. This distinction is not
always easy, since it is not clear what constitutes one instance of waving
or shaking, which might suggest that there is a fuzzy boundary between
achievements and activities in these cases.29

28. Most SLA studies are concerned with the rst question: the relationship
between morphology and lexical aspect, not situational characteristics.
29. In fact, many scholars point out the fact that activities involve repetition of
small units of component activities (e.g., McClure 1993). The issue is how
each language lexicalizes these units as unitary. Walking is conceptualized as
unitary activity, and one cannot normally say He walked if one makes one
step. One might argue if we are talking about a man who lost his ability to
walk and one day suddenly moved his feet, one step can be conceived of as
walking, hence it is possible to say He walked! However, this is a stretch, or
exibility of language being used in context, and normally one cannot say He
walked if he just made one step. Thus, asking questions such as Can you
wave once? or If someone moves his hand left to right in a waving fashion,
can you say he waved? is pertinent to test whether a particular verb is an
activity or an achievement.
300 Yasuhiro Shirai

It should be noted again here that these classications are operational


denitions for the purpose of acquisition studies, and that the decisions
on many borderline cases are still debatable. However, the number of
such borderline cases are not so large as to constitute a problem for studies.
What we should strive for is to maximize methodological rigor by making
the classication process as transparent and replicable as possible, while
assuming some level of indeterminacy, stemming from (1) the non-discrete
nature of lexical aspect categories, and (2) the diculty of interpreting the
speakers intention (Shirai 2007).

5. Conclusion

This chapter discussed methodological issues involved in coding for verbs


into lexical aspect. Although many studies report linguistic tests used to
classify verbs, not many studies report the details of how these tests were
actually applied, which is not an ideal situation in terms of replicability
and comparability of dierent studies. After discussing two issues involved
in the classication of lexical aspect (i.e., the number of categories and
crosslinguistic variation), I presented a detailed description of the classi-
cation system used in Shirai and Andersen (1995), to show (1) what proce-
dures should be followed to enhance methodological rigor, and (2) the
diculty involved in the classication procedure. By presenting such details,
we can improve the methodological sophistication of studies in SLA and
further our understanding of tense-aspect acquisition.

References

Aksu-Koc, Ayhan
1988 The acquisition of aspect and modality: The case of past reference
in Turkish. Cambridge: Cambridge University Press.
Aksu-Koc, Ayhan
1998 The role of input vs universal predispositions in the emergence of
tense-aspect morphology: Evidence from Turkish. First Language
18. 255280.
Andersen, Roger
1986 The need for native language comparison data in interpreting sec-
ond language data. Unpublished manuscript. Forum lecture 1986
TESOL Summer Institute, University of Hawaii.
Lexical aspect 301

Andersen, Roger
1989 La adquisicion de la morfologa verbal. Lingustica 1. 89141.
Andersen, Roger
1991 Developmental sequences: The emergence of aspect marking in
second language acquisition. In T. Huebner & C. A. Ferguson
(Eds.), Crosscurrents in second language acquisition and linguistic
theories, 305324. Amsterdam: John Benjamins.
Andersen, Roger
1993 Four operating principles and input distribution as explanations
for underdeveloped and mature morphological systems. In K.
Hyltenstam & A. Viborg (Eds.), Progression and regression in lan-
guage, 309339. Cambridge: Cambridge University Press.
Andersen, Roger & Yasuhiro Shirai
1994 Discourse motivations for some cognitive acquisition principles.
Studies in Second Language Acquisition 16. 133156.
Antinucci, Francesco & Ruth Miller
1976 How children talk about what happened. Journal of Child Lan-
guage 3. 169189.
Ayoun, Dalila & M. Rafael Salaberry
2008 Acquisition of English tense-aspect morphology by advanced
French instructed learners. Language Learning 58. 555595.
Bayley, Robert, Alicia Alvarez-Calderon & Sandra Schecter
1998 Tense and aspect in Mexican-origin childrens Spanish: A proto-
type account of language shift. In E. Clark (Ed.), Proceedings of
the Twenty-ninth Annual Child Language Research Forum, 221
230. Stanford, CA: CSLI.
Bardovi-Harlig, Kathleen & Anna Bergstrom
1996 Acquisition of tense and aspect in second language and foreign
language learning: Learner narratives in ESL and EFL. The
Canadian Modern Language Review 52. 308330.
Bardovi-Harlig, Kathleen
1998 Narrative structure and lexical aspect: Conspiring factors in sec-
ond language acquisition of tense-aspect morphology. Studies in
Second Language Acquisition 20. 471508.
Bardovi-Harlig, Kathleen
2000 Tense and aspect in second language acquisition: Form, meaning,
and use. Malden, MA: Blackwell.
Bickerton, Derek
1981 Roots of language. Ann Arbor, MI: Karoma Publishers.
Bickerton, Derek
1989 The child, the bioprogram and the input data: A commentary on
Cziko. First Language 9. 3337.
Bloom, Lois, Karin Lifter & Jeremie Hatz
1980 Semantics of verbs and the development of verb inection in
child language. Language 56. 386412.
302 Yasuhiro Shirai

Bonilla, Carrie
in-press Tense or aspect? Initial past tense marking for beginning class-
room learners of Spanish. Hispania.
Brinton, Laurel
1988 The development of English aspectual systems: Aspectualizers and
post-verbal particles. Cambridge: Cambridge University Press.
Brown, Roger
1973 A rst language: The early stages. Cambridge, MA: Harvard
University Press.
Bybee, Joan
1985 Morphology: A study of the relation between meaning and form.
Amsterdam/Philadelphia: John Benjamins.
Bybee, Joan, Revere Perkins & William Pagliuca
1994 The evolution of grammar: Tense, aspect, and modality in the lan-
guages of the world. Chicago, IL: University of Chicago Press.
Camps, Joaquim
2002 Aspectual distinctions in Spanish as a foreign language: The
early stages of oral production. IRAL 40. 179210.
Chen, Jidong & Yasuhiro Shirai
2010 The development of aspectual marking in Mandarin Chinese.
Applied Psycholinguistics 31. 128.
Chung, Sandra & Alan Timberlake
1985 Tense, aspect, and mood. In T. Shopen (Ed.), Language typol-
ogy and syntactic description III: Grammatical categories and the
lexicon, 202258. Cambridge: Cambridge University Press.
Collins, Laura
2002 The roles of L1 inuence and lexcial aspect in the acquisition of
temporal morphology. Language Learning 52. 4394.
Collins, Laura
2004 The particulars on universals: a comparison of the acquisition of
tense-aspect morphology among Japanese- and French-speaking
learners of English. The Canadian Modern Language Review 61.
251274.
Collins, Laura, Pavel Tromovich, Joanna White, Walcir Cardoso & Marlise
Horst
2009 Some input on the easy/dicult grammar question. The Modern
Language Journal 93. 336353.
Comajoan, Llorenc
2001 The acquisition of Catalan L2 past morphology: Evidence for the
aspect and discourse hypotheses. Bloomington, IN: Indiana Uni-
versity, Ph.D. dissertation.
Comajoan, Llorenc
2006 The Aspect Hypothesis: Development of morphology and appro-
priateness of use. Language Learning 56. 201268.
Lexical aspect 303

Comrie, Bernard
1976 Aspect. Cambridge: Cambridge University Press.
Cushing, Sara T.
1987 Use of verb morphology in the English interlanguage of a native
speaker of Serahuli. Unpublished term paper, Applied Linguis-
tics, University of California, Los Angeles.
Cziko, Gary
1989 A review of the state-process and punctual-nonpunctual distinc-
tions in childrens acquisition of verbs. First Language 9. 131.
Cziko, Gary & Keiko Koda
1987 A Japanese childs use of stative and punctual verbs. Journal of
Child Language 14. 99111.
Dahl, Osten
1985 Tense and aspect systems. Oxford: Basil Blackwell.
Dowty, David
1979 Word meaning and Montague grammar: The semantics of verbs
and times in generative semantics and in Montagues PTQ.
Dordrecht: D. Reidel.
Dunn, William
1998 The Eect of Inherent Lexical Aspect and Grammatical Person
on Tense-switching in Oral Narratives. Paper presented at the
Annual Meeting of the American Association for Applied Lin-
guistics. Seattle, WA.
Gabriele, Allison
2009 Transfer and transition in the SLA of aspect: A bidirectional
study of learners of English and Japanese. Studies in Second
Language Acquisition 31. 371402.
Gabriele, Allison & William McClure
2011 Why only some imperfectives are learned imperfectly: A study of
Chinese learners of Japanese. Language Acquisition 18. 3983.
Gavruseva, Elena
2002 Is there primacy of aspect in child L2 English? Bilingualism:
Language and Cognition 5. 109130.
Gavruseva, Elena
2003 Aktionsart, aspect, and the acquisition of niteness in early child
grammar. Linguistics 41. 723755.
Giacalone Ramat, Anna
1997 Progressive periphrases, markedness, and second language data.
In In S. Eliasson & E. H. Jahr (Eds.), Language and its ecology:
Essays in memory of Einar Haugen, 261285. Berlin: Mouton de
Gruyter.
Gonzales, Patrick
1989 The emergence of verbal morphology in early child language: A
reanalysis of Eisenberg 1982. Los Angeles, CA: UCLA, Ph.D.
dissertation.
304 Yasuhiro Shirai

Gonzales, Patrick
1990 The imperfect/past progressive distinction in Spanish discourse: An
aspectual analysis. Los Angeles, CA: UCLA, Ph.D. dissertation.
Huang, Chiung-chih
1999 Tense-aspect marking by L2 learners of English and native
English speakers: Inherent lexical aspect and unitary versus
repeated situation types. Issues in Applied Linguistics 10(2).
113129.
Izquierdo, Jesus & Laura Collins
2008 The facilitative role of L1 inuence in tense-aspect marking: a
comparison of Hispanophone and Anglophone Learners of
French. The Modern Language Journal 92. 350368.
Jin, Limin & Henriette Hendriks
2003 The development of aspect marking in L1 and L2 Chinese.
Working Papers in English and Applied Linguistics 9. 6999.
Labov, William
1975 What is a linguistic fact? Lisse: Peter de Ridder Press.
Lako, George
1986 Women, re and dangerous things. Chicago: University of Chi-
cago Press.
Lardiere, Donna
2003 The comparative fallacy revisited: A reply to Lakshmanan and
Selinker (2001). Second Language Research 19. 129143.
Lee, EunHee & Hae-Young Kim
2007 On crosslinguistic variations in imperfective aspect: The case of
L2 Korean. Language Learning 57. 651685.
Li, Ping & Yasuhiro Shirai
2000 The acquisition of lexical and grammatical aspect. Berlin/New
York: Mouton de Gruyter.
Lopez-Ortega, Nuria
2000 Tense, aspect, and narrative structure in Spanish as a second lan-
guage. Hispania 83. 488502.
Lys, Franziska & Kerry Mommer
1986 The problem of aspectual verb classication: A two-level approach.
In A. M. Farley, P. T. Farley & K-E. McCullough (Eds.), CLS 22:
Papers from the General Session at the Twenty-Second Regional
Meeting, Chicago Linguistic Society, 216230.
Martelle, Wendy
2010 Aspectual production in L2 Russian written narratives: Evidence
for the Default Past Tense Hypothesis. Paper presented at the
Slavic Cognitive Linguistics Conference (SCLC 2010), Brown
University, Providence, RI.
Martelle, Wendy
2011 Testing the Aspect Hypothesis in L2 Russian. University of
Pittsburgh, Ph.D. dissertation.
Lexical aspect 305

McClure, William
1993 A semantic parameter: The progressive in Japanese and English.
In S. Choi (Ed.), Japanese/Korean linguistics 3, 254270. Stan-
ford, CA: CSLI.
Montrul, Silvina
2002 Incomplete acquisition and attrition of Spanish tense/aspect dis-
tinctions in adult bilinguals. Bilingualism: Language and Cogni-
tion 5. 3968.
Mourelatos, Alexander
1981 Events, processes, and states. In P. J. Tedeschi & A. Zaenen
(Eds.), Syntax and semantics, vol. 14: Tense and aspect, 191
212. New York: Academic Press.
Nishi, Yumiko
2008 Verb learning and the acquisition of aspect: Rethinking the
universality of lexical aspect and the signicance of L1 transfer.
Ithaca, NY: Cornell University, Ph.D. dissertation.
Nishi, Yumiko & Yasuhiro Shirai
2007 Where L1 semantic transfer occurs: The signicance of cross-
linguistic variation in lexical aspect in the L2 acquisition of
aspect. In Y. Matsumoto, D. Oshima, O. Robinson & P. Sells
(Eds.), Diversity in language: Perspectives and implications, 219
241. Stanford, CA: CSLI Publications.
Onozuka, Hiromi
2008 On the resultative reading of the imperfective aspect in English.
Australian Journal of Linguistics 28. 116.
Parsons, Terence
1989 The progressive in English: Events, states and processes. Linguis-
tics and Philosophy 12. 21341.
Quirk, Randolph, Sidney Greenbaum, Georey Leech & Jan Svartvik
1985 A comprehensive grammar of the English language. London:
Longman.
Robison, Richard
1990 The primacy of aspect: Aspectual marking in English interlan-
guage. Studies in Second Language Acquisition 12. 315330.
Robison, Richard
1993 Aspectual marking in English interlanguage: A cross-sectional
study. Los Angeles: University of California, Los Angeles,
Ph.D. dissertation.
Robison, Richard
1995 The aspect hypothesis revisited: A cross-sectional study of tense and
aspect marking in interlanguage. Applied Linguistics 16. 344370.
Rocca, Sonia
2002 Lexical aspect in child second language acquisition of temporal
morphology: A bidirectional study. In M. R. Salaberry & Y.
Shirai (Eds.), The L2 acquisition of tense-aspect morphology,
249284. Amsterdam/Philadelphia: John Benjamins.
306 Yasuhiro Shirai

Rohde, Andreas
1996 The aspect hypothesis and the emergence of tense distinctions in
naturalistic L2 acquisition. Linguistics 34. 11151137.
Rothstein, Susan
2004 Structuring events. In S. Rothstein (Ed.), Structuring events: a
study in the semantics of lexical aspect, 135. Malden, MA and
Oxford: Blackwell.
Sachs, Jacqueline
1983 Talking about the there and then: The emergence of displaced
reference in parent-child discourse. In K. E. Nelson (Ed.), Child-
rens language, Vol. 4, 128. Hillsdale, NJ: Lawrence Erlbaum.
Sag, Ivan
1973 On the state of progress on progressives and statives. In C. J. N.
Bailey & R. Shuy (Eds.), New ways of analyzing variation in
English, 8395. Washington, D.C.: Georgetown University Press.
Salaberry, M. Rafael
1999 The development of past tense verbal morphology in classroom
L2 Spanish. Applied Linguistics 20. 151178.
Salaberry, M. Rafael
2000 The development of past tense morphology in L2 Spanish. Am-
sterdam: John Benjamins.
Salaberry, M. Rafael
2008 Marking past tense in second language acquisition: A theoretical
model. London: Continuum.
Salaberry, M. Rafael
2011 Assessing the eect of lexical aspect and grounding on the acqui-
sition of L2 Spanish past tense morphology among L1 English
speakers. Bilingualism: Language and Cognition 14(2). 184202.
Shirai, Yasuhiro
1991 Primacy of aspect in language acquisition: Simplied input and
prototype. Los Angeles, CA: UCLA, Ph.D. dissertation.
Shirai, Yasuhiro
1993 Inherent aspect and the acquisition of tense/aspect morphology
in Japanese. In H. Nakajima & Y. Otsu (Eds.), Argument struc-
ture: Its syntax and acquisition, 185211. Tokyo: Kaitakusha.
Shirai, Yasuhiro
1994 On the overgeneralization of progressive marking on stative
verbs: Bioprogram or input? First Language 14. 6782.
Shirai, Yasuhiro
1995 Tense-aspect marking by L2 learners of Japanese. In D.
MacLaughlin & S. McEwen (Eds.), Proceedings of the 19th
Annual Boston University Conference on Language Development
2, 575586. Somerville, MA: Cascadilla Press.
Shirai, Yasuhiro
1996 Does accomplishment always have duration? The Dowty-Van
Valin system versus the Smith system of inherent aspect. In L. M.
Lexical aspect 307

Dobrin, K. Singer & L. McNair (Eds.), CLS 32: Papers from the
main session, 335345. Chicago: Chicago Linguistic Society.
Shirai, Yasuhiro
1998a The emergence of tense-aspect morphology in Japanese: Univer-
sal predisposition? First Language 18. 281309.
Shirai, Yasuhiro
1998b Where the progressive and the resultative meet: Imperfective
aspect in Japanese, Korean, Chinese and English. Studies in Lan-
guage 22. 661692.
Shirai, Yasuhiro
2000 The semantics of the Japanese imperfective -teiru: An integrative
approach. Journal of Pragmatics 32. 327361.
Shirai, Yasuhiro
2002 The prototype hypothesis of tense-aspect acquisition in second
language In M. Rafael Salaberry & Yasuhiro Shirai (Eds.), The
L2 acquisition of tense-aspect morphology, 451474. Amsterdam:
Benjamins.
Shirai, Yasuhiro
2007 The aspect hypothesis, the comparative fallacy, and the validity
of obligatory context analysis: A reply to Lardiere (2003). Second
Language Research 23. 5164.
Shirai, Yasuhiro
2009 Temporality in rst and second language acquisition. In W. Klein
& P. Li (Eds.), The expression of time, 167193. Berlin/New
York: Mouton de Gruyter.
Shirai, Yasuhiro & Roger Andersen
1995 The acquisition of tense-aspect morphology: A prototype account.
Language 71. 743762.
Shirai, Yasuhiro & Atsuko Kurono
1998 The acquisition of tense-aspect marking in Japanese as a second
language. Language Learning 48. 245279.
Shirai, Yasuhiro & Yumiko Nishi
2002 Lexicalization of aspectual structures in English and Japanese.
In A. Giacalone Ramat (Ed.), Typology and second language
acquisition, 267290. Berlin: Mouton de Gruyter.
Shirai, Yasuhiro & Yumiko Nishi
2005 How what we mean impacts how we talk: The Japanese imper-
fective aspect marker -teiru in conversation. In J. Frodesen &
C. Holten (Eds.), The power of context in language learning and
teaching, 3948. Boston, MA: Thomson Heinle.
Smith, Carlota
1983 A theory of aspectual choice. Language 59. 479501.
Smith, Carlota
1991 The parameter of aspect. Dordrecht: Kluwer.
Smith, Carlota
1997 The parameter of aspect. (2nd ed.) Dordrecht: Kluwer.
308 Yasuhiro Shirai

Stephany, Ursula
1981 Verbal grammar in modern Greek early child language. In P. S.
Dale & D. Ingram (Eds.), Child language: An international per-
spective, 4557. Baltimore, MD: University Park Press.
Sugaya, Natsue & Yasuhiro Shirai
2007 The acquisition of progressive and resultative meanings of the
imperfective aspect marker by L2 Learners of Japanese: Univer-
sals, transfer, or multiple factors? Studies in Second Language
Acquisition 29. 138.
Talmy, Leonard
1985 Lexicalization patterns: Semantic structure in lexical forms. In
T. Shopen (Ed.), Language typology and syntactic description
III: Grammatical categories and the lexicon, 57149. Cambridge:
Cambridge University Press.
Talmy, Leonard
1988 The relation of grammar to cognition. In Rudzka-Ostyn (Ed.),
Topics in cognitive linguistics, 165205. Amsterdam: John Ben-
jamins.
Tatevosov, Sergei
2002 The parameter of actionality. Linguistic Typology 6. 317401.
Van Valin, Robert D., Jr.
1993 A synopsis of Role and Reference Grammar. In Robert Van
Valin Jr. (Ed.), Advances in role and reference grammar, 1164.
Amsterdam: Benjamins.
Vendler, Zeno
1957 Verbs and times. Philosophical Review 66. 143160.
Verkuyl, Hans
1989 Aspectual classes and aspectual composition. Linguistics and
Philosophy 12. 3994.
Weist, Richard
1989 Aspects of the roots of language: Commentary on Cziko. First
Language 9. 4549.
Weist, Richard, Hanna Wysock, Katarzyna Witkowska-Stadnik, Ewa Buczowska
& Emilia Konieczna
1984 The defective tense hypothesis: On the emergence of tense and
aspect in child Polish. Journal of Child Language 11. 347374.
Weist, Richard M., Aleksandra Pawlak & Jenell Carapella
2004 Syntactic-semantic interface in the acquisition of verb morphol-
ogy. Journal of Child Language 31. 3160.
Wiberg, Eva
1996 Reference to past events in bilingual Italian-Swedish children of
school age. Linguistics 34. 10871114.
Xiao, Zhonghua & Anthony McEnery
2004 A corpus-based two-level model of situation aspect. Journal of
Linguistics 40. 32563.
Chapter 9
Dening and coding data: Narrative discourse
grounding in L2 studies

Llorenc Comajoan

1. Introduction

Time is one of the central concepts of human experience and as such it is


often expressed in language. A variety of linguistic devices can be used
to mark time; namely, tense, aspect, lexical aspect, temporal adverbials,
temporal particles, and discourse principles (Klein 2009: 3940). This
chapter deals with the last of the devices: discourse grounding principles.
More specically, it addresses how the concepts of foreground (F) and
background (B) have developed in linguistics and second language acqui-
sition (SLA) research, particularly in the investigation of the acquisition
and the use of tense-aspect morphology in a second language (L2). The
line of L2 research that has focused on this issue has been conducted in
the frame of what is known as the Discourse Hypothesis (DH).1
The term discourse hypothesis was rst used by Bardovi-Harlig (1994a),
and it is currently used in studies of the L2 acquisition of tense-aspect
morphology and its relationship to discourse. Briey, following research
in functional linguistics (Hopper 1979) and discourse analysis (Dry 1983),
Bardovi-Harlig (1994a: 13) formulated the hypothesis as follows: The
discourse hypothesis for interlanguage development states that learners
use emerging tense-aspect verbal morphology to distinguish foreground
from background in narratives. Other terms used for the same hypothesis
are discourse-pragmatic hypothesis (Housen 1997), discourse-functional hypo-
thesis (Housen 1998), gure-ground distinction (Andersen 1993), and discourse
motivation (Andersen and Shirai 1994). Even though the name of the hypo-
thesis makes reference to discourse, in reality the DH has mostly been
studied from the perspective of narratives. Thus, a more appropriate term
might have been the narrative discourse hypothesis.

1. This chapter does not address the results of studies focusing on the DH, but
rather its methodology of research. For a discussion of results within the DH,
see Bardovi-Harlig (2000), Comajoan (2005), Salaberry (2008, 2011), and the
studies summarized in Appendix 3.
310 Llorenc Comajoan

This chapter discusses how the two grounding constructs (F and B)


have been dened in dierent linguistic approaches and how they have
been used in L2 studies. Even though the focus will be on methodological
issues related to the DH in narrative discourse and the L2 acquisition and
use of tense-aspect morphology (i.e., appropriate use, not use of target-like
forms), references will be made to other types of discourse and methodol-
ogies when necessary (see also Chapters 7 and 11).
The chapter is divided into four sections. The rst one presents the
main object of study of the DH: narratives in L2 acquisition. The second
section discusses how F and B have been dened from dierent linguistic
perspectives. The next section deals with discourse grounding and SLA
and examines the dierent approaches to the study of L2 text production
and discusses how the F and B have been operationalized and coded in
previous research. Finally, the conclusion provides some guidelines regard-
ing methodology of tense-aspect research for future studies on discourse
grounding.

2. Narrative discourse

The linguistic material produced by language users can be classied into


dierent types (or genres) of discourse.2 This section focuses on what De
Fina and Geargakopoulou (2012) dened as the approach that views
narrative as a type of text as opposed to narrative as a mode of thought
that is part of human cognition, because early L2 studies mostly investi-
gated the linguistic features of narrative texts rather than their cognitive
epistemological features.
Classications of types of discourse go back to Aristotle, and they were
mostly developed for pedagogical purposes and lacked a clear formulation
(Georgakopoulou and Goutsos 1997). The traditional distinction between
narration, description, and argumentation has been reclassied multiple
times, and the current consensus is that there are no pure textual types
but rather dierent features of each type can be present in a single text.
Thus, Adam (1997) classied texts into ve prototypes: narrative, descrip-
tion, argumentation, explanation, and dialogue (see Chapter 7).

2. For all intents and purposes, the terms text and discourse are used synony-
mously and interchangeably in this chapter.
Coding Grounding 311

A more basic distinction that cuts across various text types is that between
narrative and nonnarrative discourse. Georgakopoulou and Goutsos (1997:
52) provided the characteristics of narrative and nonnarrative discourse
displayed in Table 1.

Table 1. Characteristics of narrative and nonnarrative discourse (Georgakopoulou


and Goutsos 1997)
Narrative discourse Nonnarrative discourse
Ordering Temporal sequencing Multiple sequencing (logical,
temporal, etc.)
Particularity Particular events Generic truths
Normativeness Disruption and re-establish- Stating (arguing, etc.) what
ment of equilibrium the norm is
Reference Reconstructed events Veriable events
Perspective Personal Impersonal
Context Under negotiation Permanent across contexts

The main characteristic of narrative discourse is that it is ordered, that


is, events are told in temporal sequencing. This is precisely one of the
distinctive features provided in what became a seminal study of narrative
discourse in discourse studies and, later, SLA, namely Labov and Waletzky
(1967: 13).3 They provided the following denition of narrative:

Narrative will be considered as one verbal technique for recapitulating expe-


rience, in particular, a technique of constructing narrative units which match
the temporal sequence of that experience. [. . .] Normally, narrative serves an
additional function of personal interest determined by a stimulus in the
social context in which the narrative occurs. We therefore distinguish two
functions of narrative: (1) referential and (2) evaluative.

The two distinctive features of their denition were temporal order and
the double function of narratives. Two clauses are in temporal order when
their textual/discourse order of mentioning follows the temporal sequence

3. See De Fina and Georgakopoulou (2012) for an appraisal of the Labovian


model.
312 Llorenc Comajoan

in which the events took place. In contrast, two clauses are not in tem-
poral order when they can be moved around without altering the inference
of the order of the series of non-linguistic, real-world events that they
encode. The sequentiality of clauses is exemplied in Appendix 1, a sample
narrative from a conversation between Barbara (B), an L1 English learner
of L2 Catalan, and the researcher (L) (Comajoan 2001). The narrative in
Appendix 1 is presented in Table 2 without the interaction of the inter-
locutor in order to distinguish the narrative, temporally ordered clauses
(marked as N) from the non-narrative, non-temporally ordered clauses.

Table 2. Temporally ordered clauses in a narrative


Line Narrative status Clause
1. When I was 10 years old,
2. I was more reserved and shy
3. and never talked in class.
4. N The teacher wrote a comment on the report card.
5. N My parents read the comment.
6. N My parents explained to me
7. that I needed to ask more questions.
8. N Next day we talked about a eld trip.
9. N I had a question.
10. N I asked the question.
11. I think
12. it was a stupid question.
13. Maybe the teacher had already talked about that thing
14. and I had not understood.
15. N (Repeat) I asked a question about this thing
16. that the teacher had already explained.
17. N (Repeat) I asked the question.
18. N The teacher answered in a sarcastic manner.

When applying Labovs (1972) denition of temporally ordered clauses to


the narrative in Appendix 1 and Table 2, it can be seen, for instance, that
clauses 13 in Table 2 could be moved around without altering the inter-
pretation of the order in which the actions took place (Table 3):
Coding Grounding 313

Table 3. Change or order of clauses in a narrative I (without change in


interpretation)
Line Narrative status Clause
4. N The teacher wrote a comment on the report card.
5. N My parents read the comment.
6. N My parents explained to me
7. that I needed to ask more questions.
1. When I was 10 years old,
2. I was more reserved and shy
3. and never talked in class.

However, altering the order of lines 56 from Table 2 does trigger a


change in the interpretation of how the actions occurred, since the parents
read the comment before they explained to their daughter that she needed
to talk more (Table 4):

Table 4. Change or order of clauses in a narrative II (with change in


interpretation)
Line Narrative status Clause
1. When I was 10 years old,
2. I was more reserved and shy
3. and never talked in class.
4. N The teacher wrote a comment on the report card.
6. N My parents explained to me
7. that I needed to ask more questions.
5. N My parents read the comment.

Narratives are not always told in the strict chronological order of the
real-world events that they refer to. For instance, in lines 3839 and line
41 (Appendix 1) (1314 and 16 in Table 2) the speaker refers to situations
that are not in order with the previous clause. In these cases, pluperfect
morphology is used to mark deviance from chronological order. However,
nonchronological order is not always explicitly marked (through linguistic
devices), and languages dier in the amount and manner of such marking.
314 Llorenc Comajoan

For instance, the narrative status of clauses 1 and 2 in Table 3 in English


is not morphologically marked, but they would be morphologically marked
in Spanish or Catalan through the use of imperfect morphology.
Labov and Waletkzky (1967) argued that narratives that are purely
referential are pointless in the sense that they do not fulll the evaluative
requirement. In their model, evaluation is dened as that part of the
narrative which reveals the attitude of the narrator towards the narrative
by emphasizing the relative importance of some narrative units as com-
pared to others (37). Evaluative sections are crucial in their model,
because evaluation helps dene the structure of the narrative; that is, to
make a distinction between complication and result.
Narratives are not just a string of narrative clauses one after the other,
but rather narratives have structure (Labov 1972; Labov and Waletzky
1967; see Adam 1997, 1994; Adam and Lorda 1999 for other features;
Polanyi 1989 for a similar approach) and texture (Fleischman 1990).4
The structure of full narratives is presented in Figure 1 (adapted from
Labov and Waletzky 1967).

Figure 1. Parts and structure of a narrative (Labov and Waletzky 1967)

4. Fleischman (1990) used the term texture to refer to the fact that some events
in a narrative have more relevance or are more salient than others from the
perspective of the speakers communicative goals, and thus the speaker may
mark them dierently in terms of grounding.
Coding Grounding 315

Each of the six parts of a narrative has an associated underlying question


regarding specic information (Figure 1). The abstract summarizes the
narrative. The orientation sets the narrative frame by providing informa-
tion regarding the characters, location, and their activity. The complicat-
ing action is the heart of the narrative and is composed of chronologically
ordered narrative clauses that recreate an experience. The evaluation
section allows the narrator to express the reason for telling the narrative.
As pointed out by Labov (1972), good narrators will use evaluation in
order to justify their narrative and avoid the underlying question So
what? from the interlocutor. Evaluation is not localized in a specic part
of the narrative, but it permeates the complicating action and resolution
parts of the narrative (indicated by the wavy lines in Figure 1). The reso-
lution part marks the end of the series of narrative events in the complicat-
ing action. Finally, the coda is a transition between the narrative and the
moment of speech (present). These six sections constitute full-formed nar-
ratives, but they are not necessarily found in every narrative. For instance,
Silva-Corvalan (1983) analyzed oral Spanish narratives and found that
very few had abstracts or codas.

3. Narrative discourse grounding: foreground and background

This section provides a review of the dierent denitions of F and B. The


approach is chronological, and the denitions are grouped in two periods
for ease of discussion: early denitions and critical denitions.

3.1. Early denitions


The meaning of grounding when applied to language refers to the dierent
ways in which events in the real world are given prominence in and
through language. The study of grounding from the perspective of how
events are perceived and represented has been mostly carried out within
the framework of cognitive linguistics. In fact, the development of cogni-
tive linguistics as a eld has spurred a number of terms and concepts that
are related to grounding, such as construal and perspectivization
(Verhagen 2007: 48), salience (Schmid 2007: 117), and prominence
(Langacker 2007: 434). Although all these terms have specic character-
istics in current perspectives in cognitive linguistics, L2 studies of the acqui-
sition of tense-aspect morphology mostly deal with concepts discussed in
earlier work in the functional-cognitive framework, namely F and B (e.g.,
Givon 1987; Hopper 1979; Talmy 1978).
316 Llorenc Comajoan

Three general comments can be made with respect to the early deni-
tions of F and B. First, the most oft-cited and inuential paper for their
denition is Hopper (1979), who characterized the F and B constructs as
follows:
1. Discourse grounding is a universal of narrative discourse: The fore-
ground is the actual story line, and the background is the supportive
material.
2. The dening property of the foreground is sequentiality, that is, icon-
icity between the order in which events took place in the real, non-
linguistic world and the way they are narrated. Other properties of
foreground can be derived from sequentiality (e.g., focus structure,
punctual verbs, contingency, and narrativity).
3. Discourse not the linguistic forms or categories of language is
primary, that is, linguistic categories emerge within discourse.
4. Grounding can be accomplished through dierent linguistic devices,
including tense-aspect morphology, word order, and voice.
Hopper (1979) provided some evidence for all characteristics, but in most
cases it was based on very few data and languages, and thus most of the
statements were open to further investigation (e.g., whether discourse
grounding is a universal feature of narrative discourse or not, or the dierent
linguistic devices that can be used to mark grounding).
Second, earlier studies already pointed out that the F and B were two
extremes of one continuum rather than two distinct categories. For instance,
Jones and Jones (1979) used the metaphor of a string of beads of dierent
sizes (prominence) to refer to the dierent degrees of prominence events
are given in language. Thus, they rejected a bipartite structure (F versus
B) and posited what they called the multiple-level hypothesis, whereby
discourse includes multiple levels (they mentioned ve) of signicant infor-
mation, which may be marked by specic grammatical devices in various
languages.
Finally, already from the beginning, the denition of F and B has been
a challenge for researchers in the study of discourse, because both con-
cepts are intricately related to other linguistic notions. For instance, Givon
(1984: 288) provided the linguistic correlates of foreground and back-
ground displayed in Table 5. The dierent features that characterize F
and B point out that verbal morphology is just one of the devices that
can be used to mark grounding. Other linguistic devices that contribute
to the marking of F and B are those reminiscent of the expression of
temporality (cf. Dietrich et al. 1995; Klein 1986, 1994b; Noyau 1990,
Coding Grounding 317

2002): adverbials (e.g., suddenly, then, no, lexical expressions providing


adverbial information, etc.), conjunctions (e.g., and, before, after, etc.),
clause structure (main versus subordinate, see Section 4.3.), and mood
(indiciative versus subjunctive in those languages where the distinction is
productive in verbal morphology, e.g., in the Romance languages).

Table 5. Characteristics of foreground and background (Givon 1984: 288)


Feature Foreground Background
Tense past present, future, habitual
Sequentiality in-sequence out-of-sequence, anterior, perfect
Durativity compact/punctual durative/continuous
Perfectivity perfective/completive imperfective/incompletive
Modality realis irrealis
(activeness) (action/event) (state)
(syntax) (main clauses) (subordinate clauses)

The multiple ways in which prominence and grounding could be marked


in language, reected in the F and B distinction, was more thoroughly and
more critically investigated in later studies, which are discussed in the
following section.

3.2. Critical denitions


Even though the dierences between F and B may be clear intuitively (i.e.,
F backbone of a narrative, B supporting material), there is no com-
monly accepted, comprehensive denition, as illustrated by the variety of
denitions listed below:
(1) That part of a discourse which does not immediately and crucially
contribute to the speakers goal, but which merely assists, amplies,
or comments on it, is referred to as background. By contrast, the
material which supplies the main points of the discourse is known as
foreground. (Hopper and Thompson 1980: 280)
(2) The foreground is composed of sentences which refer to sequenced
points on a timeline. The background is composed of those events
that either do not refer to a single point (imperfectives, habituals,
iteratives), or refer to a point that is not presented in fabula sequence
(e.g., sentences with pluperfect tense). (Dry 1983: 46)
318 Llorenc Comajoan

(3) The foreground, or the narrative skeleton is a (report of a)


sequence of events ordered on a time axis. This chain of events, in
and of itself, is meaningless. . . . Its interpretation or its signicance
can be determined only if we know the physical conditions of its
events, their motivations, the preceding circumstances or events that
led to them, the mental state of their agents, etc. In this sense, the
background enables us to perceive or understand the foreground
events. (Reinhart 1984: 789)

All three denitions refer to the same concepts but emphasize dierent
issues. The denition in (1) is the most general and focuses on the function
of the F-B distinction: The F expresses the main point according to the
speakers goal, whereas the B comments on it. The denition in (2) focuses
on the composition of the two parts: The F is made up of single-point
events that are sequentally ordered, whereas the opposite is true of the B.
The denition in (3) retakes the notion of sequentiality from the denition
in (2), but it emphasizes the notion of contrast and interpretation, indicat-
ing that the B is no less important than the F, because without B there
could be no contrast, and consequently the F could not exist.
After the early denitions of F and B, several papers critically examined
both constructs, attempting to nd clearer criteria for their denition and
examining problematic cases. For instance, Givon (1987) titled his work
Beyond foreground and background and Dry (1992) titled hers Fore-
grounding: An assessment. Two basic conclusions can be drawn from
these studies: (a) dening F and B is problematic, because too many factors
intervene in their denition; and (b) consequently, no single dening charac-
teristic can be found for the characterization of F and B, and thus a proto-
typical approach may be more appropriate than a categorical approach.
Givon (1987) attempted to overcome the F and B dichotomy by tack-
ling four main problems of their denition. Specically, he showed that
discourse grounding is dynamic rather than static in the sense that it pre-
supposes an online interactive model of text production in which what is
produced by a speaker is F because it is asserted or is pivotal, and it be-
comes B when it is a shared presupposition by the speaker. In other words,
online production of linguistic material does not allow for a straight-
forward classication; only after-the-fact analyses allow for this. The
second problem was that the correlation between B, presupposition, and
old information was not clear. This problem became a major focus of
investigation in other studies, which showed that subordinate clauses can
Coding Grounding 319

be used to advance time and contain sentences that are chronologically


ordered with those of the main clause, and thus can be part of the F (see
Sections 4.2 and 4.3). Third, Givon questioned the correlation between F,
sequentiality and main-line or gist. The fourth problem concerned the
binary or scalar nature of the F and B distinction, and it was argued that
the distinction was not discrete, but rather scalar as was the case for many
other morphosyntactic devices used to mark discourse grounding.
Dry (1992: 335) assessed the concept of foreground and stated that
the term foregrounding has become so diuse in its application that it
requires redenition, and that such redenition would be furthered by
clarifying the assumptions that dierent disciplines bring to the investiga-
tion. She reviewed most of the previous denitions and showed that
linguistic, literary, and psycholinguistic methodologies diered in their
assumptions. Some of the major arguments regarding linguistic methodol-
ogies in Dry (1992) were the following three: First, the term foregrounding
is used for two dierent concepts, which contributed to the terms am-
biguity: it was used to refer to the cognitive process and to the textual phe-
nomena that activated the process. Second, foregrounding is often related
to prominence, which itself can be interpreted as importance or salience
(see also Binnick 1991: 380). At the same time, both importance and
salience can be of dierent types, as shown in Figure 2 (from Dry 1992).

Figure 2. Characteristics of foreground (Dry 1992)

Finally, some of the problems with dening F are due to the application of
one discipline to another (e.g., from linguistics to literary analysis), and
for this reason, clear denitions and assumptions for each discipline are
needed.
From Givons and Drys studies reviewed here and from others that
examined F and B (e.g., Couper-Kuhlen 1987; Depraetere 1996; Dry 1992;
320 Llorenc Comajoan

Fleischman 1990; Givon 1987; Reinhart 1984), the general consensus


seemed to be that there was no single dening feature for the two con-
cepts: The foreground-background contrast is better viewed as a spectrum
or continuum in which saliency is a matter of degree, the dierent degrees
being expressed through an interplay of the semantic and grammatical
oppositions available in the language. (Fleischman 1990: 184)
However, even in those denitions that took into account several factors,
there was one feature that stood out in the denition of F: sequentiality
of events. Thus, Reinhart (1984: 801) argued for the following temporal
criteria to dene grounding5:
Narrativity: Only narrative units, i.e., textual units whose order matches
the order of the real-world events they report, can serve as foreground.
Punctuality: Units reporting punctual events can serve more easily as
foreground than units reporting durative, repetitive, or habitual events.
Completeness: A report of a completed event can serve more easily as
foreground than a report of an ongoing event.
It is worth noticing that the rst criterion is categorical (only ordered
events can be foreground), but the other two are gradable (can serve
more easily as foreground).
Depraetere (1996: 715) dened F from the perspective of prototypicality
and and also gave special status to the chronological order of sentences
(Table 6).

Table 6. Denition of foreground according to Depraetere (1996)


Foreground: Sequence of chronologically ordered
bounded situations whose sequence is
reected in the linear order in which
they are reported.
Prototypical Prototypical
Narrative present, past, future tense Pluperfect
Bounded telic Bounded atelic
Main clause
Subclause with loose bonds with main Subclause with tight bonds with the
clause main clause

5. In her denition of grounding, in addition to temporal criteria, Reinhart (1984)


also provided functional dependency and culture-dependent criteria.
Coding Grounding 321

The use of a prototypical scale, such as that in Table 6, stresses, for


instance, that even though it is the norm to mark the F with present,
past, or future tense, it may be possible to use pluperfect for a succession
of past events reported in a previous past moment. The denition in Table
6 also includes aspectual criteria regarding telicity and its relationship to
grounding.
Finally, Khalil (2000, 2003) reviewed all major denitions of F and B
and identied four main inadequacies (2000: 15). First, previous research
did not identify explicit and independent criteria for the denition of F
and B. Second, it was not clear whether F and B refer to the concept-level
(i.e., the cognitive status of information formulated by the speaker) or to
the linguistic level (i.e., the production of language per se). Third, there was
a trend to conate grounding with other discourse notions (e.g., coherence)
and with its surface structure expression (e.g., how grounding is coded lin-
guistically in main or subordinate clauses). Finally, most research focused
on narrative discourse and linguistic markers of such discourse. As an
alternative, Khalil (2006: 4) provided a framework in which grounding
is a language independent, universal principle of discourse organization
that is part of a continuum. Khalil argued that the assignment of ground-
ing values to propositions has its cognitive basis, namely the ways infor-
mation is structured and distributed in mental models of events and situa-
tions (5), which are independent of discourse.
In sum, these assessments are a witness to the vitality of research in
discourse grounding and to the evolution of the concepts in the last thirty
years. The reviews have mainly shown an evolution toward treating F and
B as part of a continuum that in terms of language is a reection of mental
models and that as such may vary depending on the speakers communi-
cative purpose. The next section investigates to what extent the same evo-
lution can be found in SLA studies.

4. Discourse grounding and L2 studies

4.1. Form-oriented and meaning-oriented approaches


The acquisition of L2 temporality has been examined from several perspec-
tives (Ayoun and Salaberry 2005; Bardovi-Harlig 2000; Montrul 2004;
Montrul and Salaberry 2003; Salaberry 2008; Salaberry and Shirai 2002),
and the role of how texts are produced by L2 learners and how discourse is
structured has mainly been studied from two methodological approaches,
which Bardovi-Harlig (2000) called the meaning-oriented and the form-
322 Llorenc Comajoan

oriented approaches to the study of temporality. Whereas the former


approach answers the question Which temporal concepts and relations
do learners express and how? the latter answers What is the distribution
of the emergent verbal morphology? or How is a particular linguistic
form or category (e.g., verbal morphology, adverbial expressions) used to
encode a particular temporal construct (e.g., past time, imperfectivity)?
The two approaches dier in the sense that the rst studies a number of
devices used by learners to mark temporality, whereas the second one
tends to focus on verbal morphology and its distribution.
The dierences between the meaning-oriented and the form-oriented
approaches have been mostly discussed in light of the acquisition of tense
and aspect and its relationship to order of acquisition and the role of
lexico-aspectual categories (What comes rst, tense marking or aspect
marking? cf. the Aspect Hypothesis (AH), Chapters 5, 8, and 11; Andersen
and Shirai 1996; Bardovi-Harlig 2000). In general, fewer studies have
examined the DH, and thus there has been little discussion on how the
dierences in the approaches could account for the production of texts by
L2 learners.
Even though meaning-oriented approach studies have focused on dis-
course, they have not taken the concepts of F and B as central. Rather,
early studies in the meaning-oriented approach discussed the pragmatic,
lexical, and morphological means to mark temporality (Perdue 1993;
Dietrich, Klein, and Noyau 1995). Later studies continued to study dis-
course, but, instead of examining F and B, they investigated three main
issues (Hendriks 2005): (a) how learners expressed and integrated informa-
tion from dierent semantic domains, (b) how their interlanguage developed
over time, and (c) what causal factors (cognitive, universal, language spe-
cic, etc.) could explain the changes in interlanguage. On the other hand,
studies in the form-oriented approach rst focused on the AH, but little by
little the emphasis shifted to include the DH and the interrelationship of
the AH, the DH, and other tempo-aspectual dimensions (Andersen and
Shirai 1994; Andersen 2002; Salaberry 2008, 2011; Comajoan 2001; Bardovi-
Harlig 2000; Rosi 2009; Appendix 3 at the end of this chapter).6

6. The studies included in Appendix 3 are within the form-oriented approach.


For a review of studies dealing with discourse grounding and the acquisition
of temporality in the meaning-oriented tradition, see Bardovi-Harlig (2000);
Comajoan (2005); Dietrich, Klein and Noyau (1995); and Watorek, Benazzo,
and Hickmann (2012).
Coding Grounding 323

There are currently two strands of research that investigate how learners
structure discourse, which can be seen as a continuation of meaning-
oriented and form-oriented studies, respectively. On the one hand, meaning-
oriented approaches, represented by studies such as Carroll et al. (2000),
Carroll and von Stutterheim (2002), von Stutterheim and Lambert (2005),
Hendriks (2005), and von Stutterheim, Carroll, and Klein (2009), study
learner varieties at large and how texts are produced. These studies have
mostly focused on narrative, descriptive, and conversation texts. The main
question addressed by such studies is the extent to which language-specic
preferences in information structure are driven by grammaticized means.
More specically, it investigates the extent to which . . . L2 learners apply
the principles of information structure of their target language, as con-
trasted with their native language (L1), in producing stretches of connected
discourse as telling a story, giving route directions, or describing the layout
of their home (Carroll and Lambert 2006: 54).
The meaning-oriented studies to L2 text production have the following
theoretical bases. First, information structure is to some extent language-
dependent, which means that events are structured temporally and dis-
coursively in dierent ways in dierent languages. That is, narratives or
descriptions are structured dierently in, for instance, German, Spanish,
and Arabic (e.g., more or less focus on endpoints of actions, more or less
emphasis on location of objects, etc.). Second, and as a consequence of
the previous statement, the acquisition of a L2 can be a challenge for
adult and young learners, because they may not be aware of the subtle dif-
ferences in text production in dierent languages. Thus, meaning-oriented
studies have tended to investigate the process whereby learners may (or
may not) attain native-speaker competence in text production and at
what stage of text production the dierences between learners and native
speakers can be explained (conceptual level, selection of information, or
selection of linguistic forms). Finally, this research approach is multi-
variable. That is, dierences in L2 text production cannot be explained
by a single feature, but rather by the conuence of several variables (e.g.,
temporal concepts, the role of the syntactic subject, and word order con-
straints), which at the same time have dierent roles in the learners L1.
On the other hand, current form-oriented studies have focused on the
interrelationship between the use of a selected linguistic device to encode
tense and aspect (most frequently, verbal morphology, but also other
lexical devices, such as temporal adverbs) and discourse grounding (emer-
324 Llorenc Comajoan

gence of F and B and distribution of forms in F and B).7 The main


theoretical premises of this approach are three (Bardovi-Harlig 2000;
Salaberry 2008). First, discourse is a central inuence on the distribution
of tense-aspect morphology, that is, discourse structure (F and B) and
how learners build texture in narratives plays an important role in how
tense-aspect morphology emerges (e.g., perfective rst in the F, imper-
fective rst in the B) and develops. In its formulation, the DH is more
descriptive than explanatory; that is, as a hypothesis it allows for a
description of specic observations (associations of morphology and other
linguistic devices with F and B) that otherwise might go unnoticed, and
it allows researchers to test its predictions. However, the DH does not
explicitly explain why learners follow the patterns found in their inter-
language; that is, what makes the F emerge before the B and why specic
combinations of lexical aspect, morphological aspect, and grounding are
so commonly found. Previous research, such as Andersen and Shirai
(1996), suggested that strong cognitive or linguistic universals can account
for the explanation of the DH, but studies have not yet fully addressed the
explanatory level of the DH. Some studies have attempted to explain its
predictions as a conuence of dierent variables (e.g., Salaberry 2008),
with some evidence that discourse grounding may be more powerful as a
variable to explain L2 development than lexical aspect (cf. Comajoan and
Perez Saldanya 2005; Rosi 2009; Salaberry 2011), but further research still
needs to investigate the psycholinguistic and cognitive validity of the DH
as an acquisitional principle in L2 studies. Second, in form-oriented studies,
the level of prociency and, related to this, the developmental stage of the
learners needs to be taken into account, because it is a likely factor in
determining the distribution of tense-aspect morphology relative to ground-
ing. Finally, studies within the form-oriented approach as was the case
for studies within the meaning-oriented one also adopt a multivariable
perspective in the sense that they investigate the interrelationship of lexical
aspect and discourse grounding in the emergence and development of L2
verbal morphology (e.g., perfective morphology emerges in telic predicates
in the F, whereas imperfective morphology emerges in atelic predicates in
the B) and tries to account for the associations of developing morphology

7. Form-oriented studies have also begun to study the relationship between the
DH and the AH and have searched for an answer to the following question:
Do the two hypotheses make the same predictions, and if so, which one
best describes L2 development at dierent stages of acquisition? For a dis-
cussion of the relationship of the two hypotheses, see Bardovi-Harlig (2000);
Comajoan and Perez Saldanya (2005); Rosi (2009); and Salaberry (2011).
Coding Grounding 325

and discourse structure (focusing on the constructs of F and B instead of


information structure as was the case in meaning-oriented studies).
The two approaches need not be seen as competing, but rather as com-
plementary, because they share a multivariable perspective with the acqui-
sition of temporality and they investigate one central question related
to temporality: How do learners answer the question What happened
next? (i.e., the quaestio model, Carroll and von Stutterheim 2003). The
main dierence is that meaning-oriented studies adopt a more global per-
spective than form-oriented studies, because in addition to examining
tense-aspect morphology they take into account a variety of linguistic
devices and ways to structure information. Whereas current meaning-
oriented studies often use native-speaker data to compare learner data to
target data, form-oriented studies tend to focus more on the development
of interlanguage, and thus they do not emphasize target-like competence
as much. This dierence is also related to the level of prociency of learners
that are under study in the two types of studies. Whereas most of the
current meaning-oriented studies focus on very advanced learners, form-
oriented studies tend to focus on beginner-intermediate learners. Thus,
meaning-oriented studies have focused more on the end-state, whereas
form-oriented studies have centered on the development of interlanguage
in the beginning and intermediate stages of acquisition. Finally, from a
methodological standpoint, meaning-oriented studies have not exclusively
focused on F and B and grounding, but also on other principles of infor-
mation (e.g., information selection, perspective taking, management of
events, etc.), whereas form-oriented studies (at least those focusing on the
DH) have investigated how the L2 acquisition of tense-aspect morphology
is distributed in the F and B.

4.2. Data coding in studies investigating the DH

This section reviews the denition of F and B in form-oriented studies in


SLA and their coding in the period from the rst studies that investigated
how verbal morphology was used in discourse to the formulation of the
DH and current discussions.8 The discussion is based on a review of all

8. De Fina and Georgakopoulou (2012) discussed a number of issues related to


the analysis of narrative from dierent perspectives (discourse and socio-
linguistics) but do not include SLA studies or the denition and coding of F
and B in their discussion.
326 Llorenc Comajoan

major published studies that directly address the DH (see Appendix 3),
and, for ease of presentation, they have been divided into three periods.9
In the early period, exemplied by studies such as Kumpf (1984),
Flashner (1989), Bardovi-Harlig (1992), and Housen (1994), the main
focus of research was to study the development of the learners interlan-
guage as an autonomous system (Andersen 1984) and, more specically,
how tense-aspect morphology developed in the F and the B. The main
characteristic of these early studies, as far as the denition of F and B
and their coding is concerned, was that the denitions and methodology
were based on works of functional linguistics, namely Hopper (1979) and
Givon (1982). Briey, the F sections were those parts of a narrative that
pushed events/the story line forward, and B those that did not.
Early studies deserve special attention for two main reasons. First, some
of the studies provided microanalyses of narratives and the way learners
used their resources for the construction of narratives. For instance,
Kumpf (1984) conducted a detailed analysis of one learners interlanguage
in conversations and narratives, coded the data for seven variables (dis-
course type, aspect, verb type, verb form, clause type, time reference, and
other), and she showed that tense-aspect morphology was not distributed
equally in the F and B. Bardovi-Harlig (1992) observed that lower-level
learners already use the few resources they have to mark the story line,
suggesting that level of acquisition is an important variable to take into
account. Second, already at this early period, researchers, such as Housen
(1994), noted the diculty of coding discourse grounding mainly due to
two reasons that further research will continue to mention: rst, the
researchers subjectivity in coding for F and B; and second, the use of a
priori formal and functional categories for data coding, which does not
provide an analysis of the development of interlanguage, but rather a
purely linguistic analysis (Box 1).
The second period of research in discourse grounding in L2 is charac-
terized by the establishment of the DH per se, mostly in studies by Bardovi-
Harlig (1998, 1999, 2000) that reviewed a number of studies in the acquisi-
tion of temporality and posited that the AH and DH may conspire in the
emergence and development of tense-aspect morphology. In terms of
methodology and the denition of F and B in L2 data analysis, this period

9. The periods are chronological but focus on the methodological characteristics


of L2 studies. The chronology of L2 studies does not necessarily match the
chronology of linguistic and discourse studies that are often taken as the basis
of L2 studies.
Coding Grounding 327

Box 1. Early period of study of the DH. Example: Housen (1994)

Denition of F and B:
e Based on Hopper (1979), Hopper and Thompson (1980), and Givon (1982).
e F: primary narrative information, pushes forward the story line.
e B: elements that do not directly contribute to the speakers primary
communicative goal but which merely comment, modify, or assist
(270).
Methodology:
e No specic coding methodology in the method section, but coding was
done independently of verbal morphology.10
e The last section of the paper, titled Methodological considerations,
mentioned the diculty of assigning categories to learners production
because of the researchers subjectivity. The author suggested asking the
learner what he or she meant in retrospect, but this may be too dependent
on the learners metalinguistic skills. The second diculty is using a priori
categories to code data.

added newer studies that contributed to the denition of the two con-
structs, namely Reinhart (1984) and Dry (1992), which would be the basis
for denition of F and B from then on (Box 2).

Box 2. Establishment of the DH. Example: Bardovi-Harlig (2000)

Denition of F and B:
e Based on Hopper (1979), Dry (1981, 1983, 1992), Fleischman (1985), and
Reinhart (1984).
e F: from Reinhart (1984): narrativity, punctuality, and completeness.
e B: multifunctional, supporting the F. Importance of tense-switching as a
device to produce narratives with discourse ground.
Methodology:
e F: clauses that moved the narrative time forward.
e Coding was done independently of tense-aspect morphology.
e Direct speech was excluded, but the verbs that introduced it were coded as
F if they were sequenced with other verbs.
e Interrater coding for reliable analysis (with a second experienced coder).

10. A specic section on coding methodology was cut by the editors of Housen
(1994) (Alex Housen, personal communication, 10 April 2010). In contrast
with Housen (1994), Housen (1995) provided a detailed discussion of coding
grounding (260266) and limitations of the coding (234235).
328 Llorenc Comajoan

The third period is one of consolidation of the study of the DH, and is
characterized by two factors: studies take Bardovi-Harlig (2000) as the
basis for the denition of grounding and apply it to other learners and
languages, and data coding becomes more sophisticated with the establish-
ment of learner corpora (Rosi 2009) and the use of statistical analyses
(Salaberry 2011). In this period, research more openly discusses method-
ological problems, such as the interpretation of data in coding F and B
(e.g., Salaberry 2008; Rosi 2009; see next section) (Box 3).

Box 3. Consolidation of the study of the DH. Example: Rosi (2009)

Denition of F and B:
e Based on Hopper (1979) and Dry (1992) with emphasis on syntax (main versus
subordinate clauses).
e Denition of the DH and F and B from Bardovi-Harlig (2000, 2002).

Methodology:
e Establishment of an annotated corpus of learner data.
e Coding for specic features, among them discourse grounding (others that
may be relevant to grounding were lexical aspect, grammatical aspect,
syntax, target and nontarget like forms).
e The following grounding features were coded and analyzed: F, B, predicates
of completive clauses depending on the F main clause, predicates of
completive clauses depending on the B main clause, predicates of incidental
clauses, metanarrative predicates (outside the story) (e.g., the scene begins
with . . .), and syntactic roles of the clause (main versus subordinate).
e The author acknowledged that coding the discourse functions of predicates
can be subjective (e.g., examples with the verb vedere to see and subordinate
clauses, 7475).

In sum, in the three periods of study of the DH, there has been an evo-
lution from dependence on references from studies and theories of func-
tional linguistics to an increasing discussion of how such theories apply to
research in SLA. This is of course related to the development of SLA as
a eld itself and to the development of research programs led by specic
research teams. With regard to the denition of F and B, there is still a
reliance on early references (e.g., Hopper, Dry, and so on), but current
research has advanced by critically reviewing previous works and rening
some denitions as they apply to L2 studies. In addition, some important
issues such as subjectivity in coding have emerged as central in regard
Coding Grounding 329

to the coding of the learners interlanguage. This issue is not specic to the
DH, as it also aects meaning-oriented studies. Two possible solutions can
be adopted. First, learner data can be compared to native-speaker data (as
in most current meaning-oriented studies), or they can be interpreted by the
researcher and investigated on their own, with references to interlanguage-
internal dynamics and mechanisms (without comparing them to native-
speaker data), following Andersens (1984) call for the autonomous analysis
of interlanguage data. The two approaches are complementary even though
they are often related to dierences in theoretical perspectives.
Finally, studies in the acquisition of L2 tense and aspect (and the DH)
have also advanced in their adoption of innovative methodologies to
tackle some of the complex issues in analyzing L2 text production. For
instance, Rosi (2009), Li and Shirai (2000), and Li and Zhao (2009) used
computational models of aspect to examine whether such models resemble
child and adult language acquisition; that is, whether the networks could
learn the oft-cited associations found in learner data (e.g., telic predicates
with perfective morphology in the F). The results are promising because
they show that simple but biologically plausible computational principles
in self-organizing neural networks can account for empirically observed
patterns in childrens acquisition of lexical aspect and grammatical mor-
phology, without a priori stipulations about the structure of meaning or
concept (Li and Zhao 2009: 266). These studies have shown that the
networks do learn the prototypical associations of verbal morphology,
lexical aspect, and discourse grounding, even though the L1 has an impor-
tant eect (i.e., associations are stronger in networks trained with Spanish-
Italian data than in German-Italian) (Rosi 2009) (see also Chapter 3). A
second type of innovative methodology is based on the use of eye tracking
in the acquisition of L2 tense and aspect. For instance, Stutterheim, Carroll
and Klein (2009) studied how learners attend to dierent types of events
(as on-going or not on-going) while watching a videoclip and by the use
of eye tracking data they found that there was a direct relationship
between having grammaticalized means to express an endpoint in the
speakers L1 and how the event was tracked (e.g., German speakers attended
more to on-going events than English speakers because German does not
grammaticalize ongoingness as overtly as English). Finally, a third type
of innovation is the design of experimental studies to test the AH and
the DH that allow for the use of data that can be analyzed statistically
(Salaberry 2011).
330 Llorenc Comajoan

4.3. Coding F and B in L2 data: problematic cases

This section addresses some of the most common problems in coding L2


data for the investigation of how the F and B are marked (see Chapter 8
for a similar approach to the coding of lexical aspect). Tomlin (1985) pro-
vided a good summary of some of the most important issues when inves-
tigating F and B, and even though he referred to the correlation between
subordination and discourse grounding specically, his recommendations
are applicable to research in the DH in general. He mentioned that four
issues needed special attention (88). First, the denitions of F and B must
be made explicit (without introspection) and be syntax-independent. This
means that not all main clauses must be coded automatically as F and all
subordinate clauses as F, since the relationship between grounding and
subordination is strong, but not categorical (see below).
Second, it is important to establish clear criteria for coding F and B
and their possible correlation with other linguistic variables. More speci-
cally, four types of relationships (and possible correlations) can be estab-
lished: syntactic coding, pragmatic coding, afunctional correlation, and no
association. Syntactic and pragmatic coding are relations between form
and function. In the case of syntactic coding, according to Tomlin, there
is a causal connection between form and function, because the occurrence
of one syntactic device necessarily marks a semantic or pragmatic function
independent of speaker and text type (e.g., -s marks plural; independent
clauses correlate with F information). Pragmatic coding is also indepen-
dent of speaker and text type, but the causal connection between form
and function is not as strong. For instance, the use of and in English
(e.g., John wrote a message and sent it) indicates sequentiality even though
there is nothing inherent in the conjunction that indicates sequentiality. It
may also be the case that there is some association between form and
function (due to their frequent co-occurrrence), but it is statistically weak
or nonsignicant. Finally, it may be be the case that there is no association
at all.
Third, studies should not be based on a few speakers who produce the
same types of data, but rather on a considerable number of participants in
order to avoid individual variation. It needs to be noted that Tomlin
(1985) adopted a quantitative approach that certainly requires a consider-
able number of participants, but this does not mean that case studies or
other types of qualitative research are deemed inappropriate.
Coding Grounding 331

Finally, researchers need to collect dierent types of data and test the
eect of discourse type in the use or nonuse of specic linguistic devices.
For instance, to examine the relation between grounding and syntactic
subordination, Tomlin (1985) used online descriptions of narratives that
participants watched on a television, oral-delayed narratives of the cartoon
they had watched, written-delayed narratives, and written-edited narratives
(the participants watched the cartoon, were allowed to take notes and
were allowed to edit their narratives in a period of four days).
Tomlin (1985) described what could be considered the ideal type of
coding, and it seems warranted that L2 studies should attempt to fulll
all four recommendations. However, some of them especially the rst
one may not always be possible, as discussed in the following section.
The second and third recommendations will also depend on the theoretical
perspective adopted by the researcher (somewhere on the quantitative-
qualitative continuum, see Chapter 11). Regarding the fourth recommen-
dation, since use of verbal morphology as well as other linguistic devices
is well known to depend on type of text, it is important to give it special
attention (see Chapter 7). Taking Tomlins (1985) recommendations as a
point of departure, four issues are discussed below regarding diculties
in coding F and B in L2 data, namely morphology-independent coding,
syntax-independent coding (subordination), coding in dierent types of texts,
and interpretation of the learners intended meaning (for lexical aspect
coding, see Chapters 8 and 11).
Ideally, one would want to apply the denitions of F and B without
being biased by morphology. This problem has been mentioned by several
researchers (e.g., Bardovi-Harlig 2000; Housen 1995), and in some re-
spects it may well be impossible to solve in a completely satisfactory way,
because both verb morphology and syntax contribute to the marking of F
and B. In other words, F and B are discourse-pragmatic phenomena, not
real-world phenomena, and they do not exist independently of their mark-
ing in language (including their morphosyntactic marking). Thus, one pos-
sible solution may be to carry out two codings, the rst one just looking at
the propositional content (verbs without morphology) of predicates, and a
second one with morphology. In this case, one avoids the bias of morphol-
ogy but not that of syntax, which is completely unavoidable. For instance,
Comajoan (2001) carried out both types of coding. When the two types of
coding are done, it is possible that dierences arise (e.g., one clause with
propositional content without verbal morphology is coded as F, and the
332 Llorenc Comajoan

same clause with verbal morphology is coded as B). In these cases, an


additional coder may be necessary to reach consensus.11
Regarding syntax-independent coding and its relationship to the coding
of F and B, temporal and relative subordinate clauses are the most prob-
lematic because they occur early and frequently in learner data and beg
the following question: Can subordinate clauses produced by L2 learners
be coded as F? Preposed temporal clauses with when, as argued by Ramsay
(1987), have the characteristics of F even though they are subordinate
clauses (see Dry 1983; Bardovi-Harlig 1995; Thompson 1987 for a discus-
sion of grounding and subordination). Thus, one possible solution to the
problem of coding when-clauses is to establish that preposed when-clauses
are considered F only if they move the story forward, answer the ques-
tion What happened? and are chronologically ordered. Preposed when-
clauses that contain old information, that is, information that has already
been mentioned or new information that does not move the story forward
can be coded as B. On a few occasions, preposed when-clauses may move
the story forward and contain new information. For instance, the material
in line 13 in (4) can be coded as B because the information in the when-
clause (sasusta gets scared) is old (it had already been mentioned in line
4) and not chronologically ordered (examples from narrative retellings and
conversations from Comajoan 2001).
(4) 1. D: #ah# despres #ahm# #ahm# s # la: la dona de la// del rei
#ahm# s, passa #ahm# po// por el campo [sp]
D: ah, then, ahm, ahm, yes, the woman of the king, ahm, yes,
goes by the eld
2. D: i viu #ah# e:l el drac
D: and sees ah the dragon
3. L: #mhm#
L: mhm
4. D: i #ahm# ella #m# sasusta
D: and ahm she gets scared
5. D: i #ahm# sierra [sp]
D: and ahm closes

11. In cases where coding of grounding diered in the clauses with verbal mor-
phology and those without, Comajoan (2001) adopted the coding obtained
from the clauses with verbal morphology (thus, favoring to some extent what
the learner had intended to produce).
Coding Grounding 333

6. D: sierra [ct pronunciation] #ahm# las #ahm# orelles?


[ct pronunciation]
D: closes her ears?
7. L: ulls
L: eyes
8. D: ulls/
D: eyes
9. L: orelles es aixo [pointing at ears, laughing]
L: these are ears
10. D: ah! dacord
D: oh, ok
11. D: # #ah# #ahm# quand #ahm# la princes #ahm# la dona o
#ah# potser la princesa de la del rei
D: ahm, when, the princess, ahm, the woman, or, maybe the
princess of the king
12. L: #mhm#
L: mhm
13. D: #ahm# sasusta
D: ahm gets scared
14. D: i potser grita
D: and maybe yells
The preposed when-clause in line 3 in (5) contains new information, but
it does not move the story forward (rather it answers the question Under
what circumstances did something happen?), and therefore it can be
coded as B.
(5) 1. D: i la #m# el policia no es conscient
D: and the policeman is not conscious
2. L: #mhm#
L: mhm
3. D: i <pero cuando trata de> [sp] #ahm# levantar-se [ct]
D: and but when he tries to stand up
4. D: #ahm# Charles Chaplin agafa #ahm# el #ahm# <la pelota>
[sp] #ahm#
D: Charles Chaplin takes the stick
334 Llorenc Comajoan

5. L: em sembla que es diu la porra, em sembla/, o el pal


L: I think it is called a porra, I think, or the stick
6. D: la po// la p// la poda #ahm# del policia
D: the policemans stick
7. D: i #ahm# # #ahm# tira la pola <a la cabeza> [sp]
[laughing] del policia
D: D: and hits the policemans head.
In line 3 in (6), the when-clause moves the story forward and is chronolog-
ically ordered, and thus it can be coded as F.
(6) 1. D: i despres #ahm# #ah# el policia #ah# no est:// no esta
conscient
D: and then the policeman is not conscious
2. L: #mhm#
L: mhm
3. D: i i ca// can #ahm# va #ah# despertar-se [ct]
D: and when he woke up,
4. D: #ahm# el Charlie Chaplin #ah# li va li va agafar el,
la barra?/ de la policia
D: Charlie Chaplin took the stick from him
5. L: #mhm#
L: mhm
6. D: i #ahm# li va trucar el el cabell?
D: and hit his hair
7. L: el cap
L: his head
8. D: cap, el cap, el cap
D: head, his head, his head
The same principle of chronological order can be used for complex and
relative clauses, which can be coded as F if they are in chronological order
with the previous material in the narrative. For instance, in the main
clause in line 8 in (7), the action of saying that he took the bread is
chronologically ordered, and thus it is considered F. However, the sub-
ordinate clause in line 9 in (7) is not chronologically ordered (it had been
mentioned in line 2), and thus it is considered B.
Coding Grounding 335

(7) 1. D: hi ha una dona


D: there is a woman
2. D: i a dona #ah# agafa #ahm# el pa en la pastiteria
D: and the woman takes the bread in the bakery
3. L: #mhm#
L: mhm
4. D: i ella #ahm# no paga
D: and she does not pay
5. L: #mhm#
L: mhm
6. D: i #ahm# el gerent #ahm# de #ahm# de la patisseria #ahm#
grita
D: and the manager of the bakery yells
7. D: i i llama una policia
D: and calls the police
8. D: i # el [spanish e] diu que
D: and he says that
9. D: #ahm# la dona agafa la el pa s// sin pagar [sp]
D: the woman takes the bread without paying
The relative clause in line 4 in (8) (people who saw. . .) is not chronologi-
cally ordered and, consequently, it can be coded as B.
(8) 1. D: i la policia// el el policia #ahm# #ahm# <trata de>
[sp] #ahm# ir [sp] #ahm# poner [sp] #ah# a
lhom #ahm# en la carcel [sp]
D: and the policeman tries to go and put the man in jail
2. L: #mhm#
L: mhm
3. D: mais [fr pron] pero hi ha altres persones #ahm# qui
D: but there are other people who
4. D: #ah# viuen que
D: see that
5. D: la la dona #ahm# ha sacat la #ahm# el pan
D: the woman has taken the bread.
336 Llorenc Comajoan

Coding F and B in narratives produced during retell tasks is relatively


straightforward, because such narratives are usually easy to identify and
there is minimal intervention from the interlocutor. However, in other
types of data (e.g., conversational data), narratives are embedded in or
mixed with other language material and interlocutors often interrupt. For
instance, conversation data in Comajoan (2001) often started with a ques-
tion, e.g., What did you do for Christmas? Should the answer to this
question be counted as F or B? The following trend was observed in the
data in Comajoan (2001). The beginnings of narratives (e.g., an answer
to a question) usually began with an abstract (Labov 1972) that sum-
marized the narrative. For example, the answer to the question about
Christmas might be I visited my family. Abstracts tended to have action
words that summarized the most salient events of the narrative, which
makes them similar to the F. By contrast, the ends or codas (Labov
1972) of the narratives tended to have evaluative comments, such as
It was fun, which are often related to B functions. In opposition to
Bardovi-Harligs (1992b) practice of coding abstracts as B because they
anticipate and summarize a following narrative, abstracts may be con-
sidered F although they do not technically move the story forward, and
codas can be coded as B. For instance, in line 5 in (9) vaig anar I went
and anem menjar we ate (line 10) were considered F, and havia muches
regales there were many gifts (line 25) was considered B. Coding the
form vaig anar in line 5 in (9) as F may seem contrary to the denition of
F as that part of the narrative that moves the story line forward, but it is
often dicult to discern whether sentences like this one are abstracts (and
outside the story line, and thus, B) or they are already part of the beginning
of a narrative (beginning the story line, and thus F). In sum, the main
recommendation is that the denition of F and B is well established and
that counterexamples or exceptions be made explicit (Tomlin 1985) and,
if necessary, be analyzed separately.
(9) 1. L: molt be [name] et volia demanar #ahm# com que fa poc
que va passar el Nadal et volia demanar: com va anar el
Nadal? mes o menys que vas fer? et va arribar la meva postal?
L: very good [name] I wanted to ask you, since Christmas
just nished, I wanted to ask you: How did Christmas go?
more or less, what did you do? did you receive my postcard?
2. D: #ah# no, no todava [sp]
D: no, not yet
Coding Grounding 337

3. L: no?
L: no?
4. D: no
D: no
5. D: per Nadal #ahm# #ah# he, s, s vaig anar a #ahm# la
la casa a la meva casa
D: For Christmas yes yes I went home
6. L: #mhm#
L: mhm
7. D: amb [or #ahm#] la meva mare i el #ahm# avui i unes
#ah# a// avies/ no unes #ahm# ties tambe
D: with my mother, grandfather and grandmothers, no,
I mean aunts also
8. L: #mhm#
L: mhm
9. L: #mhm#
L: mhm
10. D: i # i con [hard to understand] s, anem # co// s,
anem menjar?
D: and yes and we ate?
11. L: #mhm#
L: mhm
12. D: #ahm# el pernil dolc, the honey baked ham [en]
D: honey baked ham
13. L: [laughing] es bo, no?
L: it is good, right?
14. D: una vegada mes, s, i les patates/ #ahm# com se diu
#ahm# amb miel [sp]
D: once more, yes, and the potatoes with how do you say
honey?
15. L: #mhm#
L: mhm
16. L: amb mel
L: amb mel
338 Llorenc Comajoan

17. D: amb amb mel


D: with honey
18. L: wow, es bo, no?
L: wow, it is good, right?
19. D: sweet sweet potatoes
D: sweet potatoes
20. L: #mhm#
L: mhm
21. D: i #ahm# unes caseroles/ i com se diu #ah# shrimp?
D: and casseroles with how do you say shrimp?
22. L: #mhm# shrimp es gambes, ah! jo tambe
L: shrimp is gambes, yeah me too
23. D: gambes gambes pequenas [sp]
D: shrimp, small shrimp
24. L: #mhm#
L: mhm
25. D: i i despres #ahm# havia #m# muches [ct] regales/
que abrir
D: and then there were many presents to open.
When there is a narrative within a narrative, the two narratives can be
coded separately. For instance, in Comajoan (2001) one learner told in a
conversation that he went to the recording of a talk show and there was a
woman who told a joke. In this conversation, the joke becomes a narrative
within the narrative, and it was coded independently for grounding. After
telling the joke, the learner continued to narrate what happened at the
recording, and the coding of the macronarrative continued after the em-
bedded mininarrative.
A nal matter that deserves special attention regarding the coding of
F and B is the interpretation of the learners intended meaning as opposed
to the researchers inferred meaning (see also Chapters 8 and 11). As
argued by Housen (1995), it is practically impossible to discern exactly
what a learner may have meant when producing a narrative (or any other
type of uncontrolled task) and code it accordingly for grounding. In
other words, how does the researcher know that the learner considered
an event as part of the skeleton or main story line of a narrative (F) or,
Coding Grounding 339

otherwise, it was considered part of what amplies, describes, or explains


the main story line (B). Dealing with such a problem is not straight-
forward as it is not possible to nd out what learners have meant to
convey as they produce narratives or other types of text spontaneously.
It is possible that the researcher will listen to the interaction after the
recording of data and interpret a posteriori what the learner intended
to say. In this sense, it may be argued that the researcher-interviewer is
not truly communicating, but rather the researcher elicits data for the sake
of the study and the learner produces language for the sake of practicing
the language and following instructions. Thus, as argued by Schumann
(1984: 58), the communication is between the recording/transcription
and the researcher, not between the interviewer and the subject. One
way to deal with the problem of the learners intended meaning is to use
rich interpretation (Housen 1995); that is, resort to the multiple sources of
information about the learner and the text being produced to interpret the
intended meaning (e.g., use contextual knowledge, performance of similar
learners doing the same task, topic of narrative or conversation, and so
on). However, no matter how rich the interpretation is, the coding of
grounding in particular is likely to be subjective, as evidenced by the fact
that agreement between one coder and four more in Housen (1995) was in
84% of the cases for tense meaning, in 76% of cases for lexical aspect, and
in 63% of cases for grounding. An alternative method to eschew the problem
of interpretation of learner data is to compare two types of coding, one
based on operational tests (often based on the prototypical meaning of
verbal predicates) and the other based on the learners intended meaning.
Ahmadi (2010) compared the two types of coding for lexical aspect and
found that the number of verb morphology tokens distributed in the four
lexical aspect categories diered in the two types of coding (e.g., decrease
of states and accomplishments and increase of activities and achievements
in the coding based on prototypical meaning of verbal predicates), although
no statistical dierences were found between the two types of coding.
Applying a similar methodology to coding of grounding may prove even
more dicult than applying to lexical aspect, because there are well estab-
lished tests for lexical aspect and the prototypical meaning of verbs can
often be established reliably (but see Chapters 8 and 11), whereas tests
for grounding are not well established and prototypicality cannot be
applied to coding for grounding because in principle it is completely the
speakers choice to construct an event as F or B (i.e., it is relatively easy to
establish the prototypical lexicoaspectual meaning of a predicate like close
340 Llorenc Comajoan

the door, but it is practically impossible to reconstruct whether close the


door is F or B from a prototypical perspective).
In sum, coding for grounding is particularly thorny because it aects
coding for a construct that goes beyond the sentence level and thus it
incorporates many of the intervening linguistic and nonlinguistic variables
in discourse (e.g., subordination and speakers intentions). This problem
notwithstanding, research in the study of grounding and L2 learning con-
tinues to develop and nd new methods to circumvent some of the prob-
lems. The nal section of this chapter reviews some of the questions that
researchers carrying out studies that incorporate coding of F and B need
to consider in order to minimize the eect of coding procedures in their
studies.

5. Conclusion

The study of the concepts of F and B has gone a long way from the early
straightforward application of functional linguistics to current discussions
of the denition and coding of F and B. Researchers interested in embark-
ing upon research in the DH or studies related to (narrative) discourse
grounding are encouraged to consider the following questions to guide
their research, some of which are similar to any kind of SLA study and
some of which are specic to research related to the study of F and B
and become particularly relevant when investigating the DH.
Regarding the theoretical approach to be adopted, a set of questions
needs to be considered: What is the objective and the general research
question under investigation? Which analytic framework or approach is
the most appropriate for pursuing this objective? Is the overall aim to
investigate how the learners use and acquisition of verbal morphology
may be inuenced by discourse grounding (form-oriented approach) or,
rather, how learners build and structure (narrative) discourse (using a
variety of pragmatic, lexical and morphosyntactic devices)? The role of
the learners L1 also plays an important part in choosing what framework
is more appropriate to explain the learners interlanguage. Thus, whereas
current studies in the meaning-oriented approach (e.g., in studies by
Carroll and von Stutterheim) give a prominent role to the L1 and attempt
to describe and explain how the learners interlanguage functions and
develops, other studies (e.g., those investigating the AH and DH) do not
focus on the L1 as much.
Coding Grounding 341

Regarding methodology, three main questions need to be addressed.


First, what kind of L2 tasks will be used to study the development of
interlanguage and its relation to grounding? Normally the type of task
chosen to investigate any issue in SLA must be in accordance with the
research objectives and research questions, but in the case of the DH
this is particularly important since it has been found that narratives
elicited in dierent types of tasks may display dierences in the devel-
opment of L2 tense-aspect morphology (see Chapter 7). Second, what
variables will be considered in the analysis of data? As mentioned previ-
ously, early research tended to focus on the eect of one variable on the
acquisition of temporality (e.g., lexical aspect, discourse grounding), but
current studies tend to examine how several variables interact. Thus, it
may be useful to design a database that includes coding of several variables
in order to examine their possible interaction.12 One main issue related to
the design of the database is to consider how the researcher will code the
problematic cases of data, such as those described in the previous section
of this chapter. Thus, it is important to acknowledge that a certain degree
of subjectivity is part of the analysis of discourse and that some possible
ways to deal with such subjectivity is to operationalize denitions (of F
and B), acknowledge the problematic cases, discuss how they have been
handled in previous research, and specify clearly how some cases were
not included in the analysis or coded dierently (using tags such as
ambiguous). Third, is a native-speaker benchmark needed? The answer
to this question is complex, since it may seem obvious that any well-
designed study in SLA should include a control group that allows for a
comparison of learners and native speakers. If the study focuses on how
one language may predetermine how texts are produced (e.g., in studies
by Carroll and von Stutterheim), it is compulsory to include native speaker
data. However, if the study focuses on the development of the learners
interlanguage, a native speaker comparison baseline may not be required,
especially when the study focuses on the early stages of acquisition. In
these cases, data from dierent groups of learners with dierent L1s or
data from the same learners performing the narrative task in their L1
may be more appropriate than having a native speaker control group.
Finding appropriate answers to the previous questions will contribute
to the advancement of research in discourse grounding in SLA be it

12. Rosi (2009) may be a good example of a database that allows the researcher
to study how dierent variables interact in the acquisition of L2 tense-aspect
morphology.
342 Llorenc Comajoan

within the DH or other research frameworks and set a research agenda


for further studies. More specically, future studies could focus on the
following three issues. First, they should provide clear presentations of
the methodological points discussed in the current chapter so that other
researchers can replicate results and rene the hypotheses and results. For
instance, taking into account the importance of dierences in results
according to dierent types of text, as discussed in this chapter and
Bardovi-Harlig (see Chapter 7), future studies should investigate the eect
of text types in spontaneous conversations as well as experimental settings
and rene the DH. Second, further studies should study the interaction of
the marking of F and B with other variables that typically interact with
grounding, such as lexical aspect and morphological aspect. As mentioned
in the current chapter, previous and current studies have been studying the
interaction, but there is not yet a body of research that has provided for an
explanation of the interaction within one theoretical framework. Andersen
and Shirai (1994) and Bardovi-Harlig (1998, 2000) pioneered the study of
the interaction of grounding and aspect, and works such as Rosi (2009),
Salaberry (2011) and Vraciu (2012) have continued it within the frame-
work of the AH and DH, but as of today we do not yet have a compre-
hensive explanatory theory of how the hypotheses interact or may be sub-
sumed one within the other. A nal issue, which is related to the previous
one, has to do with the explanatory power of the hypotheses that are
related to the marking of F and B in SLA. Calls for a search for the
explanatory power of the DH are not new (cf. Andersen 2002), and further
research should tackle the always dicult mission of crossing research
boundaries and doing interdisciplinary work within dierent theoretical
frameworks in order to reach an answer to the question regarding expla-
nation of the DH (and AH) (see also Chapter 12). In this view, a future
research agenda should be able to bring together several frameworks and
methodologies to focus on the explanation of how L2 learners express
temporality.
Finally, although it may not be fashionable anymore to refer to peda-
gogical applications of research, as anyone who has taught in an L2 class-
room knows, teaching features related to tense and aspect (e.g., how to use
the dierent past tenses in narratives and conversations) occupies a con-
siderable amount of time and takes remarkable eort from the students.
Thus, it would be expected that after years of study of the expression of
L2 temporality and improvements in the design of studies, researchers
would be in a position to exchange with practitioners their perspective on
Coding Grounding 343

the teaching of tense and aspect features and how they relate to the pro-
duction and understanding of narratives (see also Chapter 12). No doubt
this would highly contribute to a better understanding of how classroom
learners develop in their interlanguage and to the development of SLA
research in general.

References

Adam, Jean-Michel
1994 Le texte narratif. Paris: Nathan.
Adam, Jean-Michel
1997 Les textes: Types et prototypes. Paris: Nathan.
Adam, Jean-Michel & Clara-Ubaldina Lorda
1999 Lingustica de los textos narrativos. Barcelona, Spain: Ariel.
Ahmadi, Newsha
2010 Instructed second language learning of English tense and aspect by
Persian speakers. Saarbrucken: Lambert Academic Publishing.
Andersen, Roger
1993 Four operating principles and input distribution as explanations
for underdeveloped and mature morphological systems. In K.
Hyltenstam & A. Viborg (Eds.), Progression and regression in
language, 309339. New York: Cambridge University Press.
Andersen, Roger
2002 The dimensions of pastness. In M. R. Salaberry & Y. Shirai
(Eds.), The L2 acquisition of tense-aspect morphology, 79106.
Amsterdam: Benjamins.
Andersen, Roger & Yasuhiro Shirai
1994 Discourse motivations for some cognitive acquisition principles.
Studies in Second Language Acquisition 16. 135156.
Ayoun, Dalila & M. Rafael Salaberry
2005 Tense and aspect in Romance languages: Theoretical and applied
perspectives. Amsterdam: Benjamins.
Bardovi-Harlig, Kathleen
1992 The telling of a tale: Discourse structure and tense use in learners
narratives. In L. Bouton & Y. Kachru (Eds.), Pragmatics and lan-
guage learning, Monograph (Vol. 3., 144161). Urbana-Champaign,
IL: Division of English as an International Language.
Bardovi-Harlig, Kathleen
1994 Anecdote or evidence? Evaluating support for hypotheses con-
cerning the development of tense and aspect. In S. Gass, A.
Cohen & E. Tarone (Eds.), Research methodology in second lan-
guage acquisition, 4160. Hillsdale, NJ: Erlbaum.
344 Llorenc Comajoan

Bardovi-Harlig, Kathleen
1995 A narrative perspective on the development of the tense/aspect
system in second language acquisition. Studies in Second Lan-
guage Acquisition 17. 263291.
Bardovi-Harlig, Kathleen
1998 Narrative structure and lexical aspect: Conspiring factors in
second language acquisition of tense-aspect morphology. Studies
in Second Language Acquisition 20. 471508.
Bardovi-Harlig, Kathleen
1999a Examining the role of text type in L2 tense-aspect research:
Broadening our horizons. In P. Robinson (Ed.), Representation
and process: Proceedings of the 3rd Pacic Second Language
Research Forum (Vol. 1), 129138. Tokyo: The Pacic Second
Language Research Forum.
Bardovi-Harlig, Kathleen
1999b From morpheme studies to temporal semantics. Studies in
Second Language Acquisition 21. 341382.
Bardovi-Harlig, Kathleen
2000 Telling of the tale. Studies in Second Language Acquisition 1(45).
456560.
Bardovi-Harlig, Kathleen
2000 Tense and aspect in second language acquisition: Form, meaning,
and use. Malden, MA: Blackwell.
Bardovi-Harlig, Kathleen
2002 Analyzing aspect. In M. R. Salaberry & Y. Shirai (Eds.), Tense-
aspect morphology in L2 acquisition, 129154. Amsterdam: John
Benjamins.
Binnick, Robert
1991 Time and the verb: A guide to tense and aspect. New York:
Oxford University Press.
Bronckart, Jean-Paul & Beatrice Bourdin
1993 Lacquisition des valeurs des temps des verbes. Langue Francaise
93. 102124.
Carroll, Mary & Monique Lambert
2006 Reorganizing principles of information structure in advanced
L2s: A study of French and German learners of English. In H.
Byrnes, H. Weger-Guntharp & K. Sprang (Eds.), Educating for
advanced foreign language capacities, 5473. Georgetown: George-
town University Press.
Carroll, Mary, Jorge Murcia-Serra, Marzena Watorek & Alessandra Bendiscioli
2000 The relevance of information organisation to second language
acquisition studies. Studies in Second Language Acquisition 22.
441466.
Coding Grounding 345

Carroll, Mary & Christiane von Stutterheim


2003 Typology and information organisation: perspective taking and
language-specic eects in the construal of events. In A. Giacalone
Ramat (Ed.), Typology and second language acquisition, 365
402. Berlin: Mouton de Gruyter.
Clachar, Arlene
2005 Creole English speakers treatment of tense-aspect morphology
in English interlanguage written discourse. Language Learning
55. 275334.
Comajoan, Llorenc
2001 The acquisition of Catalan L2 past morphology: Evidence for the
aspect and discourse hypotheses. Bloomington, IN: Indiana Uni-
versity, Ph.D. dissertation.
Comajoan, Llorenc
2005 The acquisition of perfective and imperfective morphology and
the marking of discourse grounding in Catalan. In D. Ayoun &
M. R. Salaberry (Eds.), Tense and aspect in Romance languages,
3578. Amsterdam: Benjamins.
Comajoan, Llorenc & Manuel Perez Saldanya
2005 Grammaticalization and language acquisition: Interaction of
lexical aspect and discourse. In D. Eddington (Ed), Selected pro-
ceedings of the 6th Conference on the Acquisition of Spanish and
Portuguese as First and Second Languages, 4455. Somerville,
MA: Cascadilla Proceedings Project.
Couper-Kuhlen, Elizabeth
1987 Temporal relations and reference time in narrative discourse. In
A. Schopf (Ed.), Essays on tensing (Vol. 1), 725. Tubingen,
Germany: Max Niemeyer.
De Fina, Anna & Alexandra Geargapoulou
2012 Analyzing narrative. Discourse and sociolinguistic perspectives.
New York: Cambridge University Press.
Depraetere, Ilse
1996 Foregrounding in English relative clauses. Linguistics 34. 699
731.
Dietrich, Rainer, Wolfgang Klein & Colette Noyau
1995 The acquisition of temporality in a second language. Philadelphia:
Benjamins.
Dry, Helen
1981 Sentence aspect and the movement of narrative time. Text 1.
233240.
Dry, Helen
1983 The movement of narrative time. Journal of Literary Semantics
12. 1953.
346 Llorenc Comajoan

Dry, Helen
1992 Foregrounding: An assessment. In S. J. Hwang & W. R.
Merrield (Eds.), Language in context: Essays for Robert E.
Longacre, 435450. Arlington, TX: The Summer Institute of
Linguistics and the University of Texas at Arlington.
Flashner, Vanessa
1989 Transfer of aspect in the English oral narratives of native Russian
speakers. In H. Dechert & M. Raupach (Eds.), Transfer in
language production, 7197. Norwood, NJ: Ablex.
Fleischman, Suzanne
1985 Discourse functions of tense-aspect oppositions in narrative:
Toward a theory of grounding. Linguistics 23. 851882.
Fleischman, Suzanne
1990 Tense and narrativity: From medieval performance to modern
ction. Austin, TX: University of Texas Press.
Georgakopoulou, Alexandra & Dionysis Goutsos
1997 Discourse analysis: An introduction. Edinburgh, United Kingdom:
Edinburgh University Press.
Givon, Tomas
1984 Syntax: A functional-typological introduction (Vol. 1). Philadel-
phia: Benjamins.
Givon, Tomas
1987 Beyond foreground and background. In R. S. Tomlin (Ed.),
Coherence and grounding in discourse, 175188. Philadelphia:
Benjamins.
Guell, Lourdes
1998 La adquisicion del tiempo verbal en el aprendizaje del espanol como
lengua extranjera. Barcelona, Spain, Universitat Autonoma de
Barcelona, Ph.D. dissertation.
Hendriks, Henriette
2005 The structure of learner varieties. Berlin: Mouton de Gruyter.
Hopper, Paul
1979 Aspect and foregrounding in discourse. In T. Givon (Ed.), Syn-
tax and semantics: Discourse and syntax, 213241. New York:
Academic Press.
Hopper, Paul & Sandra Thompson
1980 Transitivity in grammar and discourse. Language 56. 251299.
Housen, Alex
1994 Tense and aspect in second language acquisition: The Dutch
interlanguage of a native speaker of English. In C. Vet & C.
Vetters (Eds.), Tense and aspect in discourse, 257291. New
York: Mouton de Gruyter.
Housen, Alex
1995 Its about time. The acquisition of temporality in English as a
second language in a multilingual educational context. Brussels,
Belgium: University of Brussels, Ph.D. dissertation.
Coding Grounding 347

Housen, Alex
1997 Inherent semantics versus discourse-pragmatics in the L2-develop-
ment of tense-aspect. In L. Daz & C. Perez (Eds.), Views on the
acquisition and use of a second language, EUROSLA7 Proceed-
ings, 299312. Barcelona, Spain: Universitat Pompeu Fabra.
Housen, Alex
1998 Facteurs semantico-conceptuels et discursivo-fonctionnels dans
le developpement des systemes temporo-aspectuels: Apercu de
lacquisition de langlais comme langue etrangere. In S. Vogeleer,
A. Borillo, C. Vetters & M. Vuillaume (Eds.), Temps et discours,
257279. Leuven, Belgium: Peeters.
Howard, Michael
2004 On the interactional eect of linguistic factors in interlanguage
variation. The case of past time marking. International Review
of Applied Linguistics 42. 319334.
Jones, Larry & Linda Jones
1979 Multiple levels of information in discourse. In L. Jones (Ed.), Dis-
course studies in Mesoamerican languages, Vol. 1, 327. Dallas:
Summer Institute of Linguistics and the University of Texas at
Arlington.
Khalil, Esam N.
2000 Grounding in English and Arabic news discourse. Amsterdam:
Benjamins.
Khalil, Esam N.
2001 Grounding and its signalling: Evidence from short news texts.
Discourse Studies 3. 97.
Klein, Wolgang
2009 How time is encoded. In W. Klein & P. Li (Eds.), The expression
of time, 3982. Berlin: Mouton de Gruyter.
Kumpf, Lorraine
1984 Temporal systems and universality in interlanguage: A case
study. In F. Eckman, L. Bell & D. Nelson (Eds.), Universals
of second language acquisition, 132143. Rowley, MA: Newbury
House.
Labov, William
1972 Language in the inner city: Studies in the Black English Vernacular.
Philadelphia: University of Pennsylvania Press.
Labov, William & Joshua Waletzky
1967 Narrative analysis: Oral versions of personal experience. In J.
Helm (Ed.), Essays on the verbal and visual arts, 1244. Seattle,
WA: University of Washington Press.
Laord, Barbara
1996 The development of tense/aspect relations in L2 Spanish narra-
tives: Evidence to test competing hypotheses. Paper presented at
the 16th Annual Second Language Research Forum, Arizona
State University, Tempe.
348 Llorenc Comajoan

Langacker, Ronald
2007 Cognitive grammar. In D. Geeraerts & H. Cuyckens (Eds.),
The Oxford handbook of cognitive linguistics, 421462. Oxford:
Oxford University Press.
Li, Ping & Yasuhiro Shirai
2000 The acquisition of lexical and grammatical aspect. Berlin: Mouton
de Gruyter.
Li, Ping & Xiaowei Zhao
2009 Computational modeling of the expression of time. In W. Klein
and P. Li (Eds.), The expression of time, 241272. Berlin: Mouton
de Gruyter.
Liskin-Gasparro, Judy
1996 Narrative strategies: A case study of developing storytelling
skills by a learner of Spanish. The Modern Language Journal
80. 271286.
Lopez-Ortega, Nuria
2000 Tense, aspect, and narrative structure in Spanish as a second lan-
guage. Hispania 83. 488502.
MacWhinney, Brian
1995 The CHILDES project: Tools for analyzing talk. Mahwah:
Erlbaum.
Montrul, Silvina
2004 The acquisition of Spanish: Morphosyntactic development in
monolingual and bilingual L1 acquisition and adult L2 acquisition.
Amsterdam: Benjamins.
Montrul, Silvina & M. Rafael Salaberry
2003 The development of tense/aspect morphology in Spanish as a
second language. In B. Laord & M. R. Salaberry (Eds.), Spanish
second language acquisition, 4773. Georgetown: Georgetown
University Press.
Perdue, Clive
1993 Adult language acquisition: cross-linguistic perspectives. Cambridge:
Cambridge University Press.
Polanyi, Livia
1989 Telling the American story: A structural and cultural analysis of
conversational storytelling. Cambridge, MA: MIT Press.
Ramsay, Violeta
1987 The functional distribution of preposed and postposed if and
when clauses in written discourse. In R. S. Tomlin (Ed.),
Coherence and grounding in discourse, 383408. Philadelphia:
Benjamins.
Reinhart, Tanya
1984 Principles of gestalt perception in the temporal organization of
narrative texts. Linguistics 22. 779809.
Coding Grounding 349

Rosi, Fabiana
2009 Learning aspect in Italian L2: Corpus annotation, acquisitional
patterns, and connectionist modelling. Milan, Italy: FrancoAngeli.
Salaberry, M. Rafael
2008 Marking past tense in second language acquisition. London: Con-
tinuum Press.
Salaberry, M. Rafael
2011 Assessing the eect of lexical aspect and grounding on the acqui-
sition of L2 Spanish past tense morphology among L1 English
speakers. Bilingualism: Language and Cognition 14(2). 184202.
Salaberry, M. Rafael & Yasuhiro Shirai
2002 The L2 acquisition of tense-aspect morphology. Amsterdam:
Benjamins.
Schmid, Hans-Jorg
2007 Entrenchment, salience, and basic levels. In D. Geeraerts & H.
Cuyckens (Eds.), The Oxford handbook of cognitive linguistics,
117138. Oxford: Oxford University Press.
Silva-Corvalan, Carmen
1983 Tense and aspect in oral Spanish narrative: Context and mean-
ing. Language 59. 760780.
Talmy, Leonard
1978 Figure and Ground in Complex Sentences. In Joseph H. et al.
Greenberg (Eds.), Syntax, 625649. Stanford: Stanford University
Press.
Thompson, Sandra
1987 Subordination and narrative event structure. In. R. S. Tomlin
(Ed.) Coherence and grounding in discourse, 435454. Amsterdam:
Benjamins.
Tomlin, Russell S.
1985 Foreground-background information and the syntax of subordi-
nation. Text 5. 85122.
Verhagen, Arie
2007 Construal and perspectivization. In D. Geeraerts & H. Cuyckens
(Eds.), The Oxford handbook of cognitive linguistics, 4881.
Oxford: Oxford University Press.
von Stutterheim, Christiane, Mary Carroll & Wolfgang Klein
2009 New perspectives in analyzing aspectual distinctions across lan-
guages. In W. Klein & P. Li (Eds.), The expression of time,
195216. Berlin: Mouton de Gruyter.
von Stutterheim, Christiane & Wolfgang Klein
1987 A concept-oriented approach to second language studies. In C.
W. Pfa (Ed.), First and second language acquisition processes,
191205. Cambridge, MA: Newbury House.
von Stutterheim, Christiane & Monique Lambert
2005 Cross-linguistic analyses of temporal perspectives in text produc-
tion. In H. Hendriks (Ed.), The structure of learner varieties,
203262. Berlin: Mouton de Gruyter.
350 Llorenc Comajoan

Vraciu, Eleonora Alexandra


2012 Verbal morphology in advanced varieties of English L2: Aspect
or Discourse Hypothesis? In M. Watorek, S. Benazzo & M.
Hickmann (Eds.), Comparative perspectives on language acquisi-
tion. A tribute to Clive Perdue, 153169. Bristol: Multilingual
Matters.
Watorek, Marzena, Sandra Benazzo & Maya Hickmann
2012 Comparative perspectives on language acquisition. A tribute to
Clive Perdue. Bristol: Multilingual Matters.
Coding Grounding 351

Appendix 1. Sample narrative between Barbara (B, English L1, Catalan L2) and
the researcher (L). For transcription codes, see Appendix 2.

Line Catalan English translation

1 B: [laughing] #ahm# quan tenia # quants B: when I was, how old?


anys?
2 B: tenia com dot// dotze?, no, deu, deu anys B: I was twelve, no, ten, ten years old
3 L: #mhm# L: mhm
4 B: #ahm# era era mes reservada i mes tmida B: I was was more reserved and more shy
5 L: #mhm# L: mhm
6 B: i mai parlava en la en la classe B: I never talked in class
7 B: i per aixo el #ahm# el mestre va va escriure B: and for this reason the teacher wrote in my
en el report card? [en] report card?
8 L: #ah# a la, al butllet de notes, es diu L: report card, thats how you say it
9 B: butllet/ B: report card
10 L: butllet de notes L: report card
11 B: butllet de notes que no parlava bastant B: report card, that I did not talk enough
12 B: i que havia de de fer mes preguntes en la B: and that I had to ask more questions in class
classe
13 B: i #ah# per aixo # el els meus pares van B: and for this reason my parents read this
llegir aixo
14 B: i/ van explicar-me que B: and they explained to me that
15 B: havia de parlar mes i fer mes preguntes, B: I had to talk more and ask more questions
16 L: #mhm# L: mhm
17 B: que quan tenia una pregunta B: that when I had a question
18 B: havia de de dir/ al:/ alguna cosa B: I had to say something
19 L: #aha# L: mhm
20 B: i per aixo el # el dia el dia proper? B: and for this reason on the next day?
21 L: el seg// el proper o el seguent dia L: next day or following day
22 B: el seguent dia #ahm# vam parlar dun #ah# B: the next day we talked about a trip, a school
viatge, un #ahm# <school trip eld trip> trip
[en]
23 L: un viat// #ah# una excursio L: a school trip
24 B: un excursio i, no se, B: a trip, and, I dont know
25 B: tenia una pregunta B: I had a question
26 B: i vaig fer [laughing]/ B: and I asked it
27 B: vaig fer-la i, B: I asked it
28 B: no se, B: I dont know
29 B: crec que B: I think that
30 B: era una una pregun// pregunta una mica B: it was a question, a little bit stupid
estupida [laughing]
31 L: [laughing] L: [laughing]
32 L: [laughing] pero quina pregunta era? no L: [laughing] but what was the question? dont
ten recordes? you remember?
33 B: nooo/ #ahm# pot ser que:/ he de: B: no! #ahm# maybe
34 L: ah! L: oh
35 L: on anem? L: where are we going?
36 B: [laughing] B: [laughing]
37 L: [laughing] L: [laughing]
38 B: ja havia #ahm# parlat/ daquesta cosa/ B: he had already talked about this thing
39 B: i que no va entendre molt be B: and I did not understand very well
40 B: i #ah# vaig #ahm# #ah# vaig fer una B: and I asked the question about this
pregunta/ sobre aquesta cosa/
41 B: que #ah# ja havia explicat/ el mestre B: that the teacher had already explained
42 B: i vaig preguntar i, B: and I asked and
43 B: no se, [laughing]/ B: I dont know
44 B: el mestre va #ah# respondre duna mena B: the teacher replied in a very sarcastic
molt sarcastica manner.
352 Llorenc Comajoan

Appendix 2. Transcription codes


Most of the transcription codes were adapted from the CHAT transcription in
CHILDES (MacWhinney 1995).
Each participant is identied by a name (pseudonym) followed by a colon. The
researchers is identied by L:.
#ahm# and #ah# are hesitation marks.
Square brackets with an equal sign [] provide information about the immedi-
ately previous sentence. When the information refers to more than the previous
word, the aected words are inside < >. The information can be about pronun-
ciation, language (en: English, fr: French, sp: Spanish, ct: Catalan), emotion, and
so on. In the following example, the information within brackets indicates that the
word tness was used in English.
D: es un fanatic de tness [en]
D: he is a tness fanatic
Long utterances are indicated by :, and the sign @ indicates laughter.
The signs // indicate that the word is unnished and will be reformulated.
For instance:
D: pero pero el el el hom// el home deia
D: but but the the the ma the man said
The signs / indicate interruption by the listener. indicates the listeners
intervention. Thus, material marked with / and overlaps.
D: els acudits #ahm# tenen que ser limpias [sp]/
D: the jokes had to be clean
D: pero no: no era muy [sp]
D: but no, they were very
L: ah net!
L: oh! clean!
L: no ho eren? [laughing]
L: they werent?
Coding Grounding 353

Appendix 3. Studies that investigate the emergence and development


of verbal morphology in the foreground and background (Discourse
Hypothesis).

Study Denition of foreground (F) and background (B) and


coding methodology
Kumpf (1984) e F: clauses that push the event line forward; B: clauses
that do not
e Analysis of grounding in conversations and narra-
tives: in conversations, tense-aspect morphology is
less structured than in narratives.
e It examines how discourse structure (conversation
versus narrative; foreground versus background) is
related to aspect, verb types (e.g., transitive versus
intransitive), verb forms, clause types, and time
reference.
Flashner (1989) e Application of discourse analysis to interlanguage
data
e Analysis: clauses are coded according to the semantic
function within the text (foreground, background,
and subdivisions).
Bardovi-Harlig (1992) e Little information on how F and B were coded.
e Learners put to use the linguistic resources they have
to produce narratives: low-level learners mark the
story line.
Housen (1994) e Extraction of narrative episodes from conversations
and coded for F and B
e F: It provides primary narrative data and carries the
action forward; B: all other narrative clauses
e Attempts to code independently of grammatical form
to avoid circularity.
e Methodological concerns: (a) researchers subjectivity
in coding F and B, and (b) use of a priori formal and
functional categories
354 Llorenc Comajoan

Bardovi-Harlig (1995) e Specic research questions related to F and B and


coding of morphology regarding emergence of
morphology, presence of verbal forms in F and B,
and development of interlanguage
e Denition of F: Reinhart (1984): narrativity,
punctuality, and completeness; Dry (1983): newness
e Coding of F: Sequencing: move the narrative time
forward
e Coding of grounding independent of verbal
morphology
e Quoted speech was not counted as F, but the verb
that introduces quoted speech was counted as F if
sequenced.
e Interater coding
Liskin-Gasparro (1996) e Microanalysis of narratives: storyword (narrative)
versus nonstoryworld (evaluative)
e Diculties in coding: researchers interpretation of
what is evaluative and what is not
Laord (1996) e Little reference to how coding was done

Housen (1997) e F: situations in sequence and push the narrative plot


forward, answer What happened?,
B: remaining situations
e Reference to more rigorous procedures to code F and
B independent of morphology and to subjectivity of
coding
Housen (1998) e Similar to Housen (1997)

Bardovi-Harlig (1998) e Main reference for the establishment of the AH and


DH
e Methodology: same as Bardovi-Harlig (1995)

Guell (1998) e No denition of F and B per se. Reference to Pollak


(1960): base, framework, introduction and nucleus.
Nucleus is F, the rest is B.
e Coding of F and B made a priori by researcher and
what native speakers do.
Bardovi-Harlig (1999b) e Prociency is an important variable in the study of
how learners build texts.
e Reference to studies of narrative rather than
discourse
Lopez-Ortega (2000) e Denition of F and B according to Hopper and
Reinhart.
e No reference to how the data were coded.
Coding Grounding 355

Bardovi-Harlig (2000) e It reviews syntactic and semantic criteria to dene F


and B and adopts a semantic approach arguing that
it is more apt for L2 data for early learners.
e The F has a single function, whereas the B has
several ones.
Howard (2004) e Narrative discourse: sequentiality, quaestio model
e Diculties in coding: dependent and independent
relative clauses; advancing time clauses in
subordinate clauses were coded as B.
Comajoan (2005) e F: It moves the story forward, answers the question
What happened? and is chronologically ordered.
Easiest to verify: chronological order
e Issues in coding: priority of chronological order and
inuence of verbal morphology and syntax: relative
clauses and when-clauses.
Clachar (2005) e DH: Narrative discourse hypothesis
e F and B: Dry (1983): F: What happened next?
e No clear criteria on coding F and B

Salaberry (2008) e Three main strands of discourse-based approaches:


grounding, text structure, and sentence sequencing
e Coding: diculties and issues of interpretation
of data
Rosi (2009) e F: new information, not given, push the story
forward and ordered. B: descriptive functions
e It examines subjectivity of coding

Salaberry (2011) e F: chronological order, punctuality, and completeness


e Use of 3 additional raters for reliability of analysis
e Tests of narrative grounding: sequentiality
Chapter 10
Data Analysis: Quantitative approaches

Robert Bayley

1. Introduction

The quantitative analysis of the acquisition of tense and aspect in a second


language (L2) presents a number of challenges, particularly when examin-
ing natural speech of the type discussed by Bardovi-Harlig (Chapter 7).
The diculty arises for several reasons. First, although we have a great
deal of evidence that shows that aspectual class inuences the extent to
which L2 learners mark tense, aspect is not the only inuence involved
(Chapter 11; Bardovi-Harlig 1998, 2000; Bayley 1994; Wul et al. 2009).
That is, like other interlanguage features, tense and aspect are subject to
what Young and Bayley refer to as the principle of multiple causes,
which states that it is unlikely that any single contextual factor can
explain the variability in the [interlanguage] data (1996: 253). Several
studies have shown, for example, that saliency, operationalized as the
extent of the dierence between the present or base form of the verb and
the past tense form, signicantly aects the degree to which learners make
tense (e.g., Bayley 1994; Hakuta 1976; Wolfram 1985). In addition, in the
case of English regular verbs, the past tense marker may be absent as a
result of a phonological process, -t, -d deletion, that exists in all dialects
that have been studied to date (Bayley 1996; Wolfram 1985). As well,
Goad, White, and Steele (2003) have argued that cases of missing inec-
tions may be the result of prosodic transfer from the L1. In other cases,
the position of the aspectual marker in the sentence can have a signicant
eect, as in the case of the Chinese perfective aspect marker -le (Jia and
Bayley 2008). In addition, recent research suggests that frequency and
prototypicality also drive acquisition (Langman and Bayley 2002; Wul
et al. 2009). Social factors also aect an L2 speakers use of bare or
inected forms. Bayley (1994), for example, found that Chinese L1 learners
of English were more likely to mark verbs for past tense when they were
conversing with another Chinese learner than when they were conversing
with a native speaker of English.
358 Robert Bayley

Given the multiple factors that inuence whether a learner marks a


verb for tense, it is clear that either we must control for a large number
of potential inuences or employ a form of multivariate analysis that
allows us to examine the eects of multiple independent variables simulta-
neously if we wish to analyze natural speech where forms are unevenly dis-
tributed among aspectual categories as well as among other relevant
potentially inuencing factors. The need to control for many potential
inuences or to use multivariate analysis results not only from the possible
inuences enumerated in the previous paragraph, but also from the distri-
butional imbalance in the input that is one of the dening features of the
Aspect Hypothesis. Prototypical forms, such as preterite punctual verbs in
Spanish partirse (to break) (e.g., se partio), are far more common than
non-prototypical forms such as imperfect punctual verbs (e.g., se parta)
(Anderson 1991; Anderson and Shirai 1994). Hence, if we wish to analyze
learner production and to follow the principle of accountability, i.e., to
account for all of the possible forms in our data (Sanko 1990: 296), con-
ventional statistical procedures, which require the kind of balanced distribu-
tions characteristic of experimental data, are inadequate for our analysis.
Rather, to analyze data produced in natural conversation, the analytical
methods that have long been used in sociolinguistics may prove most useful.
Varbrul (Rousseau and Sanko 1978), a specialized application of
logistic regression, has long been sociolinguists preferred program for the
analysis of naturally occurring language data, and the current imple-
mentations, GoldVarb X and GoldVarb Lion (Sanko, Tagliamonte, and
Smith 2012), are still widely used. Although Varbrul has a number of
advantages, particularly ease of recoding and output that is relatively
easy to understand, it does have limitations and other programs, including
the logistic regression modules in most commercially available software
programs such as SPSS, produce equivalent results and have the advan-
tages of easier handling of interactions and being more familiar to many
researchers in second language acquisition (Morrison 2005). In addition, a
version of Varbrul, Rbrul, has recently been developed for R, the open
source statistics package (Johnson 2009; Gorman and Johnson, 2013).
Unlike GoldVarb, which is limited to nominal variables, Rbrul is able to
handle continuous variables, such as age or time in the target language
environment, variables that are obviously important for acquisition research.
Rbrul also enables the researcher to model the individual speaker as a
Data Analysis: Quantitative approaches 359

random variable, an important consideration for second language research


where it is not always clear that learners all follow the same path.1
In the following sections, I briey consider a number of issues that arise
prior to any statistical analysis of tense and aspect. These include coding
and how verbs should be grouped into categories for further analysis.
I then consider appropriate analysis for dierent kinds of data used to
examine the acquisition of tense and aspect. Finally, I illustrate ways of
analyzing natural language data with examples from studies of the acqui-
sition of tense and aspect in a variety of languages.

2. Preparing for quantitative analysis

Although we are concerned primarily with the appropriate means of quan-


titative analysis, a number of questions must be considered prior to any
analysis. Here, I consider issues that arise in determining the meaning of
learner utterances, the need for reliable coding of aspectual categories to
allow comparisons to be made across studies, and what features need to
be accounted for (or controlled) in addition to tense and aspect.

2.1. Coding
For any analysis of natural language data, coding is among the most
important and at times the most tedious of steps, particularly when large
quantities of data are involved. In the case of L2 speech, additional com-
plications may arise. For example, when we study fully procient native
speakers, we can assume that the surface meaning is the intended mean-
ing. Thus, we are safe in assuming that a present tense form refers to the
present. However, such an assumption can break down when we examine
interlanguage production, particularly the speech of less procient learners,
as I found when conducting a study of interlanguage tense marking (Bayley
1991). In an interview with an intermediate speaker of English, I asked
about the participants family. The speaker oered a long account about
his grandfather, a government ocial in China. All of the verbs in the
account were unmarked for tense, that is, they appeared to be present

1. Goldvarb X and Goldvarb Lion may be downloaded from /individual.


utoronto.ca/tagliamonte/goldvarb.htm. The most recent version of Rbrul and
several helpful papers and exercises may be downloaded from
www.danielezrajohnson.com/rbrul.html.
360 Robert Bayley

tense. I then asked: How old is your grandfather? The speaker responded:
Oh, he die in 1969. It was fortunate that I thought to ask about the
grandfathers age. Otherwise, the unmarked past tense forms would simply
have been coded as present-tense forms, with 3rd person singular -s missing.
Geeslin and Gudmestad (2010) provide a further example of the diculty
in interpreting non-native speaker utterances. In Spanish, the verbs ser to
be and estar to be both occur with adjectives and alternate depending
on the intended meaning:

(1) a. Mario esta muy guapo.


Mario looks very handsome (today).
b. Mario es muy guapo.
Mario is very handsome.

In analyzing a fully procient speaker we would have no diculty in inter-


preting the intended meaning. In the case of an L2 learner, however,
clearly we need additional evidence to determine whether the learner has
acquired the contrastive meaning of ser and estar.2
Of course, potential misunderstandings about the time of an event are
much less likely to arise in the case of narratives that deal with the inter-
viewees own experience, where events are necessarily in the past (see
Chapters 7, 8, 9, and 11 for further discussion of interpreting what the
speaker intends). Moreover, impersonal narratives where the speaker is
asked to retell the story of a brief lm or wordless picture book allow the
researcher to interpret both the time of the events described and provide
guidance as to the meaning of the verbs. Nevertheless, problems can still
arise even when the time of the event is clear. For example, Sebastian and
Slobin (1994) reported that the present was the dominant tense used in
narratives of Mercer Mayers (1969) wordless picture book, Frog where

2. The issue is actually somewhat more complicated than acquiring an obliga-


tory target language distinction. As shown by Silva-Corvalan (1994) for U.S.
Spanish and Geeslin and Guijarro-Fuentes (2008) for several contact varieties
of Spanish, estar is expanding at the expense of ser in a number of dierent
Spanish varieties. Hence, to interpret an L2 speakers use of the two copular
verbs, we need to know the patterns that exist in the dialect to which the
learner is exposed.
Data Analysis: Quantitative approaches 361

are you?, by Spanish 9 year-olds and adults.3 The past was the dominant
tense only in the narratives produced by the Spanish 4 and 5 year-olds.
However, a study of Frog narratives produced by Mexican American
and Mexican immigrant children in Texas, Bayley (1999) found that the
past was the dominant tense in the narratives produced by children from
6 to 12 years old, while the 4 and 5 year-olds were nearly evenly divided in
their use of the past and mixed tenses. Table 1 compares the results from
the two studies.

Table 1. Dominant tense in Mexican-origin and Spanish narratives (Source: Bayley,


1999: 11; source of Spanish data: Sabastian and Slobin, 1994: 244)
Speakers Present Past Mixed
Mexican-origin 1012 year-olds (n 27) 0 23 4
Mexican-origin 68 year-olds (n 11) 0 9 2
Mexican-origin 45 year-olds (n 11) 0 6 5
Spanish adults (n 12) 10 2 0
Spanish 9 year-olds (n 12) 8 4 0
Spanish 5 year-olds (n 12) 3 7 2
Spanish 4 year-olds (n 12) 2 8 2

There are a number of plausible reasons for the dierences between the
two studies, including dierences in elicitation technique (Bayley 1999) as
well as the possible inuence of English narrative style on the Mexican
background children.4 That issue is beyond the scope of this chapter.
However, as Berman and Slobin observe, although the past is the unmarked
tense for telling a narrative in English, speakers may choose to depart
from this tense by adopting the more marked narrative or historic use of
present tense for the main thread of their narrations (1994: 131). Given
the variability in native speech, how can we know that a language learner
did not intend to use the narrative present? That is, how do we know that
what appears at rst glance to be a missing past tense form is not an
attempt by a learner to use a rhetorical device that is found in native
speech? (Wolfson 1982).

3. Sebastian and Slobin used a criterion of 80% to determine the dominant or


anchor tense of a narrative.
4. In a related study of English speakers in the same volume, Berman and Slobin
(1994) found that the past was the dominant tense for 9 year-olds and half of
the 5 year-olds, but not for the adults or 3 year-olds.
362 Robert Bayley

Two answers are possible. First, we might simply posit that we are
examining learners use of their verbal morphology to express temporal
states. Hence, if a certain number of verbs with past temporal reference
are either unmarked or marked for present tense, we simply report that
fact, presumably along with aspectual categories. However, as suggested
in the latter part of this chapter, multivariate analysis provides a more
satisfactory answer at the level of the individual speaker if not at the level
of the individual token. If we code not only for temporal reference and
aspectual class, but also for potential inuences, such as the saliency of
the dierence between the past and present tense forms, and in the case of
English regular verbs, for the phonetic features of the following segment,
we can determine whether speakers are indeed using the narrative present,
as Bayley (1996) did in a study of past tense marking and consonant cluster
reduction by Chinese learners of English. In that study, speakers marked
more past reference verbs as past in elicited narratives than in relatively
informal interviews. For example, speakers marked 53% of regular verbs
in the elicited narratives, but only 35% in the conversational sections of
the interview. Had the learners been using the narrative present, we would
expect the opposite result.
To summarize, although interlanguage production data are inuenced
by a wide range of factors and we cannot always be certain what a learner
intended in any particular instance, an examination of overall patterns
across a range of styles, produced in response to a range of tasks, can
provide reasonable assurance about the speakers internal system. In L2
English, for example, the patterning of regular verbs that are subject to
consonant cluster reduction (e.g., [mist] can be reduced to [mis] before a
consonant), provide an especially useful diagnostic because the phonological
processes involved are independent of temporal or aspectual categories.
We turn now to issues involved in coding for aspect.
2.2. Coding for aspect
Most studies of the Aspect Hypothesis have used Vendlers (1967) cate-
gories and many, although certainly by no means all, have focused on the
relationship between lexical aspect and past tense marking (e.g., Anderson
1991; Bardovi-Harlig 1994, 1998, 2000; Bardovi-Harlig and Reynolds 1995;
Bayley 1999; Collins 2002; Salaberry 2000). Table 2 summarizes the fea-
tures of Vendlers categories, which are also illustrated in examples (2) to
(7), taken from the Spanish data used in Bayley (1999). Note that the data
come from working-class Mexican American children and do not always
follow the prescriptive norms of Spanish grammar.
Data Analysis: Quantitative approaches 363

Table 2. Features of the four verb classes (Andersen 1991: 113)


State Activity Accomplishment Achievement
Punctual
Telic
Dynamic

As illustrated in Table 2, stative verbs extend over an indenite period of


time, have no clear end point, and do not require a continuing input of
energy to continue, as in (2) and (3).
(2) Un da haba un nino y un perrito y una rana.
One day there was a boy and a dog and a frog.
(3) y crea que estaba muerta.
and he believed that it was dead.
Activity verbs, like statives, are [punctual] and [telic]. However, they
are [dynamic]. That is to say, they describe an action that takes place
over a period of time and that has no inherent endpoint as in (4):
(4) y el estaba pescando.
and he was shing.
Telic, or accomplishment, verbs are [punctual] and [dynamic]. They are
also characterized by an inherent endpoint, as in (5):
(5) se la llevo pa su casa.
he brought it [the turtle] home.
Finally, punctual, or achievement, verbs are [durative] (or [punctual]),
[telic], and [dynamic]. They are distinguished from accomplishment
verbs by the fact that the actions they describe have no perceptible dura-
tion, as in (6) and (7):
(6) se cayo el nino al agua.
the boy fell into the water.
(7) y luego la rana se desperto.
and then the frog woke up.
According to Andersen (1991), in languages like Spanish that grammatic-
alize both the perfective and the imperfective, children acquiring an L1
and older learners acquiring an L2 will use perfective morphology rst to
364 Robert Bayley

mark punctuals.5 Somewhat later, learners will use imperfect morphology


to mark states. However, as Salaberry and Shirai (2002: 6) point out,
researchers do not always agree on the relevant aspectual categories even
for a single language, much less cross-linguistically. At the very least, then,
it is necessary to clearly indicate the categories used and to provide su-
cient examples so that our studies may be compared within and across
languages (see also Chapter 8).

2.3. Additional factors


Table 3 illustrates some of the factors that have been hypothesized to
inuence the L2 acquisition of tense and aspect in a variety of languages.
When we consider the extensive (and still incomplete) list of possible
inuences, all of which have been shown to have an eect in one or
more studies, we can begin to appreciate some of the diculties in model-
ing variation. Moreover, not only do we nd a large number of possible
inuences, but also a number of possible inuences may well interact. For
example, in narrative, foreground clauses tend to be perfective and usually
contain accomplishments or achievements, whereas background clauses tend
to be imperfective and the verbs are most frequently stative or activity
verbs. Other possible combinations of factors present additional diculties.
For example, as noted earlier, accomplishments and achievements are
prototypically perfective. Thus, in a study of past tense marking we would
expect to nd most perfectives to be either telic or punctual. In contrast, as
we have seen, states and activities have the characteristics of ongoingness.
Hence, we would expect most to be imperfective. Results from Tajikas
(1999) study of the acquisition of English past tense marking by inter-
mediate-advanced Japanese women college students in Japan illustrate
the problem.

5. In Spanish, the preterite is most frequently used to mark telic events, i.e.,
events that have an inherent endpoint, and punctual events, i.e., events that
take no perceptible time. The imperfect is more often used to mark past states
or activities that have no inherent end point. Many Spanish imperfect tense
verbs may be translated into English using the progressive, e.g., estaba pescando,
he was shing. However, verbs of all lexical classes may occur with both
the preterite and the imperfect tenses, even if non-prototypical examples are
comparatively rare.
Data Analysis: Quantitative approaches 365

Table 3. Possible linguistic inuences on L2 tense and aspect marking


Possible linguistic inuence Target Language(s) Studies

Lexical aspect Spanish, English, Andersen, 1991; Ayoun &


French, Japanese Salaberry, 2008; Bardovi-Harlig,
1994, 1998, 2000; Bardovi-Harlig
& Reynolds, 1995; Bayley, 1999;
Collins, 2002; Salaberry, 2005;
Shirai & Kurono, 1998
Grammatical aspect English Bayley, 1991, 1994; Tajika, 1999
Foregrounding/ Catalan, English, Bardovi-Harlig, 1998; Comajoan,
backgrounding Spanish 2005; Kumpf, 1984; Salaberry,
2011; Tajika, 1999
Saliency English Bayley, 1991, 1994; Collins et al.,
2009; Hakuta, 1976; Wolfram,
1985
Frequency English, Hungarian Collins et al., 2009; Langman &
Bayley, 2002
Position of aspectual particle in Mandarin Jia & Bayley, 2008
the sentence
Transfer from L1 to L2 or L2 English, Mandarin, Bayley, 1991, 1994; Cai, 2010;
to L3 Portuguese, Russian Flashner, 1989; Salaberry, 2005
Discourse genre, task English, Mandarin Bayley, 1994; Jia & Bayley, 2008;
Tajika, 1999
Presence of temporal adverbs English Teutsch-Dwyer, 1994
Optionality (i.e., whether a Mandarin Jia & Bayley, 2008
grammatical marker is optional
or obligatory in the target
language)
Phonetic features of the preced- English Bayley, 1996; Wolfram, 1985
ing and following segments
Semantic universals Spanish Slabakova & Montrul, 2003

Note: The listed studies are intended as examples only, not as a comprehensive
list. Other studies have treated some of the same inuences mentioned here and
other inuences have also been proposed.

Tajika (1999) used Varbrul to analyze her data. The independent variables
examined included the foreground-background distinction, lexical aspect,
and perfectivity-imperfectivity. Of these three independent variables, she
reported that only grounding achieved statistical signicance, with verbs
366 Robert Bayley

in foreground clauses more likely to be marked for past tense than verbs
in background clauses. Overall the analysis appears sound, but there is
a potential problem. Foreground clauses are more likely to be perfective
and background clauses imperfective. Similarly, foreground clauses are
more likely to contain achievement or accomplishment verbs, i.e., they
are more likely to be events, while background clauses are more likely
to contain stative or activity verbs. That is, the three independent variables
are likely to interact. Tajika did not give measures of goodness-of-t in her
tables, that is, she did not provide information about how well the statistical
model t the data, nor did she provide cross-tabs that would allow us to
judge whether in fact there was interaction among these variables. It
would be surprising, though, if there were no interactions. Clearly, then
we need to consider a range of alternative analyses, and include or control
for possible sources of interaction. As shown in Table 3, numerous other
factors have been found to constrain L2 learners use of tense and aspect
morphology. Some, such as the position of the perfective aspect marker -le
in Mandarin, may be language specic, while others, such as frequency, if
indeed they apply, should have a similar eect across languages.
In analyzing interlanguage data for tense and aspect, not only are we
faced with numerous possible linguistic inuences, we also need to account
or control for numerous social factors and characteristics of the learner.
These include the rst language and any other languages spoken, pro-
ciency level, the conditions under which the language was or is being
acquired (formal classroom or untutored acquisition), age at which acqui-
sition began, and the learners social network(s) (predominantly or exclu-
sively native-language speakers or speakers of the target language). Finally,
we need to consider the discourse context and task type, both of which have
been shown to aect L2 speakers use of target-language forms (Bayley
1994, 1996; Tarone and Liu 1995; Tajika 1999). In the following section,
I illustrate some of the ways that the numerous potential inuences on
learners use of tense and aspectual forms may be analyzed, using methods
drawn from sociolinguistics.

3. Multivariate analysis

As Young and Bayley (1996) note, a considerable number of early studies


that examined variation in L2 performance investigated only a single variable
that the researcher hypothesized would inuence the learner. Ellis (1987), for
example, attributed variation in the use of the past tense by intermediate
Data Analysis: Quantitative approaches 367

learners of English from diverse backgrounds to the amount of planning


time available, while Selinker and Douglas (1985) found that variation in
discourse organization by a Spanish-speaking learner of English could be
attributed to the discourse topic. However, as the previous section indi-
cates, statistical models that consider only a single possible inuence on
variation are inadequate for understanding the issues involved. This is
because, in cases where many factors are known to inuence variation,
our question is not does X or Y inuence Z, but rather, which of the
multiple possible inuences signicantly inuences a learners choice of a
particular form and what is the relative strength of those inuences?
This problem is one that sociolinguists, who study natural language
produced in the speech community, have long dealt with. As noted above,
for many years, Varbrul (Cedergren and Sanko 1974; Rousseau and
Sanko 1978) has been the sociolinguists preferred tool. Varbrul, which
is designed to handle the type of unevenly distributed data found in natural
language, enables the researcher to model the many factors that can
potentially inuence a speakers choice of one or another variant, for
instance whether a past reference verb is marked for tense or not, to deter-
mine which factors signicantly aect the choice of a variant, and to assess
the relative strength of the various inuences. The details of how to con-
duct a Varbrul analysis are beyond the scope of this chapter, and they
are covered in detail elsewhere (Bayley 2002; Paolillo 2002; Tagliamonte
2006; Young and Bayley 1996).
In this section, I illustrate the use of Varbrul with data from several
studies. The rst example comes from a recent study of the acquisition
(and reacquisition) of the perfective aspectual particle -le by students in a
Chinese heritage language program in the southwestern United States (Jia
and Bayley 2008). Participants included 36 children and adolescents, 18
born in China and 18 in the U.S., ranging in age from 5 to 15 years old.
Children performed three tasks. The rst consisted of a narrative retelling
of Mayers (1969) Frog, where are you? Next, 24 students who were in their
second year at the school or beyond, divided evenly by birthplace and age
group, were given a multiple choice cloze test consisting of 15 sentences
and 20 verbs and asked to ll in the blanks with -le, , or another appro-
priate aspect marker. Sentences were presented in Chinese characters, with
pinyin supplied for dicult words that students might not know. Finally,
the 24 students who participated in Task 2 were presented with 10 pictures
and 10 sentences describing the pictures and asked to complete the sentences
with -le or . As in Task 2, sentences were presented in Chinese characters,
368 Robert Bayley

with pinyin supplied when necessary.6 Thus, we had both relatively free data,
with tokens distributed unevenly, as well as the type of controlled data
typical of experimental studies. We will focus on the procedures used for
the analysis of the narrative data.

3.1. Coding for -le


The status of the perfective aspect marker -le has been frequently studied
in Chinese linguistics, including studies of both rst and second language
acquisition (see e.g., Du and D. Li 2002; Erbaugh 1983; Kwan-Terry
1979; P. Li and Bowerman 1998; Shi 1989; Thompson 1968; Wang 1965;
Wen 1995, 1997). Grammatically, -le indicates a completed action, but the
completion of such action has no relationship to the time of speaking, so it
is possible to use -le to refer to actions in the present or future, as well as
to completed actions in the past. For language learners, acquisition of
aspectual marker -le is complicated by the existence of the homophonous
sentence nal particle le, which, among other functions, may indicate a
currently relevant state (CRS) rather than perfectivity (Li and Thompson
1981):
(8)
ranhou zhe ge mifeng jiu chu lai le
then this CL bee soon exit come CRS
Then the bee came out right away.7
The situation is further complicated because both particle le and verbal
sux -le may be sentence nal.
(9)
haizi ta xing -le
child 3sg awake PFV
The child woke up.
(10)
haizi yijing xing le
child already wake CRS
The child is already awake.

6. The 12 lowest level students were excluded from Tasks 2 and 3 because they
had not yet acquired sucient skill in reading Chinese characters.
7. The following abbreviations are used in the mopheme-by-morpheme glosses
in the Chinese examples: CL, classier; CRS, Currently Relevant State (le);
NOM, nominalizer (de); PFV, perfective aspect (-le); 1pl, rst person plural
pronoun; and 3sg, third person singular pronoun.
Data Analysis: Quantitative approaches 369

To examine childrens use of -le, we coded the frog narratives for a wide
range of factors, or independent variables, all of which had been shown to
have some inuence in one or more studies:

Dependent variable:
-le used grammatically, -le used ungrammatically, -le absent where expected
or allowed. Two Mandarin native-speaking graduate students with strong
backgrounds in linguistics conducted a reliability check and achieved an
inter-rater reliability rate of over 95%.

Independent variables (factors):


(1) Position of -le in the sentence: sentence nal; post verbal within
sentence.
Sentence nal:

haizi ta xing -le
child 3sg awake PFV
The child is awake.
Post verbal within the sentence:

xiao gou kanjian -le mifeng wo
little dog see PFV bee hive
The little dog saw the beehive.
(2) Optionality: -le required in standard Mandarin; -le optional in stan-
dard Mandarin.
Required:

nanhai ting (jian) le yi ge shengyin
boy hear PFV one CL noise
The boy heard a noise.
Optional:
() ,
xiao nanhai pa shang (-le) shitou zhuazhu -le ge shuzhi
little boy climb up PFV rock hold PFV CL branch.
The little boy climbed up the rock and took hold of a branch.
370 Robert Bayley

The optional -le occurs after the verb climb in the above example, which
is grammatically acceptable when the action took hold follows as part
of the series of actions (for a similar example, see notes in Du and D. Li
2002).
(3) Lexical aspect: punctual; telic-stative; semelfactive; activity; stative.8
(4) Grounding: foreground; background:

na nanhai shuizhao -le xiao qingwa pao -le
that boy sleep PFV little frog run PFV
The boy fell asleep while his frog ran away.
Here, the verb fell asleep is background, while the verb run away is
foregrounded.
(5) Birthplace/age of arrival of student: U.S.; China, arrived in U.S.
at age 5 or younger (that is, while they were still acquiring major
features of their L1 grammatical system); China, arrived in U.S.
at age 6 or older.
(6) Age of student: 57; 810; 1113; 1415.
(7) Number of years at the Chinese Heritage Language School: 1, 2, 3
or more.
(8) Language use at home (student): Mandarin primarily or exclusively;
Mandarin and English; English primarily or exclusively.
(9) Language use at home (parents): Mandarin primarily or exclusively;
Mandarin and English.
In distinguishing between perfective aspect marker -le and CRS sentence-
nal particle le, we followed Li and Thompson (1981), who distinguish
between the two homophonous forms on the basis of use.
All recent implementations of Varbrul enable the researcher to test
the signicance of factor groups (or independent variables) through the
step-up step-down procedure, analogous to step-wise regression in com-
mercial implementations of logistic regression. During the step-up proce-
dure one factor group at a time is added to see if the model improves

8. There is a considerable literature on Chinese aspectual categories, which is


briey summarized in Du and Li (2002) and Jia and Bayley (2008). For a
fuller explanation, see Li and Thompson (1981).
Data Analysis: Quantitative approaches 371

signicantly (with .05 as the alpha level). The reverse procedure is followed
in the step-down procedure. The program then generates output indicating
the best model from each procedure. Normally these are the same.
Varbrul assumes that factor groups are independent of one another.
In most cases, this is not a problem with well-dened linguistic factors,
particularly in phonology (although see Howard 2004 and Sigley 2003).
However, non-linguistic factors often interact. For example, we might
well expect childrens patterns of language use to change as they spend
more time in the target language community. In fact, we found that age
and birthplace interacted in our data. Therefore, following the procedure
outlined in Young and Bayley (1996: 279280), we recoded the data to
combine the factors from these two groups. That is, we combined the two
factor groups for age and birthplace into a single factor group consisting
of the following factors: age 57, born in China; age 57, born in the
U.S., etc. In addition, to produce the most parsimonious model that still
accounted for the observed variation, we combined factors within groups
that did not dier signicantly from one another.
Table 4 shows the results of the nal analysis. The results may be inter-
preted as follows. A weight, sometimes loosely called a probability, of
between .5 and 1.0 indicates that the factor favors the application value,
in this case the appropriate use of -le. A weight between 0 and .5 indicates

Table 4. Use of -le by Chinese heritage language learners


Factor group Factor N % -le Weight
Position of -le Sentence nal 200 83 .779
Postverbal in sentence 349 43 .327
Optionality Obligatory 484 63 .556
Optional 65 15 .156
Birthplace X age China 510 130 80 .709
China 1115 143 60 .521
United States 276 46 .386
Home language Mandarin 186 74 .613
Mandarin/English, English 363 49 .442
Total Input 549 58 .603
Notes: Application value: suppliance of -le. Factors within groups that did
not dierent signicantly from one another have been combined. Chi-square/
cell 0.7368; log likelihood 282.623; all factor groups signicant at p < .05.
372 Robert Bayley

that the factor disfavors the application value. The results, however, must
be interpreted in light of the input value, or corrected mean. In the study
under discussion, aspectual marker -le was supplied in 58% of the cases
where it could be used, resulting in an input value of .603. When -le is
not sentence nal, it has a weight of only .327, although -le is used appro-
priately in 43% of such cases. This is because the rate of -le use in such
cases is considerably lower than the overall rate of use, with a corrected
mean of .603.
Note that only four factor groups achieved statistical signicance at the
.05 level: the position of -le in the sentence, optionality, age by birthplace,
and the childs home language use. In contrast to our expectations and to
the results reported in Wen (1997), lexical aspect did not reach signicance.
This lack of signicance may stem from the fact that the study examined a
relatively small number of tokens, only 549. However, it does serve to point
out that studies of L2 aspect and tense need to consider the role of other
potential inuences in their design.
Our analysis for the other two tasks was much more straightforward.
With a controlled number of observations, we were able to use one-way
ANOVAs. Because ANOVA is covered extensively elsewhere (see e.g.,
Reitveld and van Hout 2005; Larson-Hall 2010), those results will not be
illustrated here. Interested readers may refer to Jia and Bayley (2008) for
details.9

3.2. The question of the individual


Typically sociolinguistic studies in the variationist tradition have concen-
trated on the social group rather than the individual as the unit of analysis.
Studies of variation in native languages, particularly in phonology, have
shown that, given sucient data, individual and group patterns match in
ne detail. Guys (1980) study of coronal stop deletion, e.g., just may be
pronounced as jus, in the English of New Yorkers and Philadelphians
is perhaps the best-known example. As Bayley and Langman (2004) note,
however, when it comes to L2 learners, we are on much less rm ground.
L2 learners, particularly immigrant learners, are exposed to varying types
of input and have varying degrees of access to the target language. Although
several studies have shown that L2 learners follow similar paths of acquisi-
tion (Bayley and Langman 2004; Regan 2004), those studies are fairly small.
Hence the question remains open.

9. For an interesting use of repeated measures ANOVA to study the relative


eects of lexical aspect and grounding in L2 Spanish acquisition, see Salaberry
(2011).
Data Analysis: Quantitative approaches 373

Here, I illustrate a case of multivariate analysis that considers the role


of individual speakers Langman and Bayleys (2002) study of the acqui-
sition of Hungarian verbal morphology by adult Chinese residents of
Budapest. The study, based on sociolinguistic interviews collected by
Langman in the early 1990s, presents a number of interesting issues. First,
it is among the earliest studies to examine the acquisition of an agglutina-
tive language (Hungarian) by speakers of an analytic language (Chinese).
Second, only one of the speakers examined had studied Hungarian formally.
Hence, the study deals primarily with informal acquisition. Third, Hungarian
has a number of typologically unusual properties that pose particular
challenges to a learner. Perhaps most unusual is the Janus-like double
agreement morpheme that marks person and number of the subject and
deniteness of the object (see Langman and Bayley 2002: 5961 for a
summary of the Hungarian verbal system), e.g., tanul-ok, 1 sg present
indicative indenite study or learn versus tanul-om, 1 sg present indica-
tive denite; tanul-sz, 2 sg present indicative indenite; tanul-od, 2 sg present
indicative denite, etc. The morphemes -ok, -om, -sz, and -od indicate the
person and number of the subject and the deniteness or indeniteness of
the object.
As is typical in studies of variation, whether in rst or second languages,
we performed several dierent analyses. Initially we coded four variants of
the dependent variable: 1) corresponds to native speaker usage; 2) inected
form, but does not correspond to native speaker usage; 3) base form, but
lacks inection where inection is required; 4) interlanguage form that
is not part of native speech. Forms that corresponded to native speaker
usage included third person singular indenite verbs, which correspond to
the base form that does not require inection. Because we wished to focus
on what speakers had acquired, forms that do not require inection were
excluded from the nal analysis.
In addition to the dependent variable, we coded for the following inde-
pendent variables:
1. Deniteness of the verb: denite; indenite
2. Tense: present; non-present (overwhelmingly past)
3. Vowel harmony
4. Verb class: regular verbs; -ik verbs, where the 3 sg ends in -ik as
opposed to ; irregular verbs, which consist of jonni come and menni
go; two highly frequent verbs, tudni know and forms of the verb
to be (i.e., vanni for present and past and lenni for the future).
5. Speaker: 9 speakers participated in the study, including 4 women and
5 men in their 20s and 30s.
374 Robert Bayley

Overall, speakers inected 43% of the verbs coded according to native


speaker norms; however, accuracy varied widely, ranging from 29% to
89%. Results of multivariate analysis with GoldVarb showed that all of
the linguistic factor groups coded proved to be statistically signicant. Of
perhaps most general interest, highly frequent verbs and irregular verbs
favored native-like usage, while regular verbs, which comprised nearly
half of the tokens, disfavored native-like usage. The results for individuals,
however, are our primary concern here.

Table 5. Verbal marking by Chinese acquirers of Hungarian,


with verbs that do not require inection excluded
(application value correctly marked)
Speaker (pseudonym) Weight
Xiao Wang .921
Zhou .649
Chang .590
Xiao Hong .432
Ai Hua .342
Liu .338
Pan .219
Mei .215
Lan .203
Input .429

Table 5 shows the factor weights for the 9 speakers who participated in
the study. Using the log-likelihood test summarized above, we combined
data from speakers whose accuracy level did not dier signicantly from
one another. This procedure resulted in four groups of two plus one highly
advanced speaker. We then ran separate analyses for each group. While
running separate analyses did result in rather small samples and in the
loss of some tokens that were either categorically marked or unmarked
for native-speaker usage, as well as some statistical uctuation due to small
sample size, it had the advantage of allowing us to determine which factors
had the strongest inuence for speakers of dierent prociency levels. We
were also able to investigate whether factors had the same eect on all
learners, regardless of prociency level, or whether learners at dierent
prociency levels were constrained by dierent types of inuences. The
results of these analyses are shown in Table 6.
Data Analysis: Quantitative approaches 375

Table 6. Analysis of Chinese-Hungarian verbal inectional marking by accuracy


level (Langman and Bayley 2002: 73)
Group Speaker(s) N % inected Input Signicant
correctly factors

1 Xiao Wang 124 89 .923 Verb type (other, .729; regular, .238)
2 Zhou, Chang 108 56 .631 Verb type (frequent, .789; irregular
and -ik, .677; regular, .208);
Tense (present, .629, non-present, .137);
Deniteness (denite, .725; indenite,
.252)
3 Ai Hua, Liu 216 36 .299 Verb type (vanni, .938; tudni, .698;
irregular and -ik, .518; regular, .211);
Person/number (1 sg, .609; 3 sg, .391;
other, .365);
Tense (present, .611, non-present,
.293); Deniteness (denite, .656;
indenite, .362)
4 Lan, Xiao Hong 277 31 .277 Verb type (vanni, .938; irregular and
-ik, .568; tudni, .525; regular, .287);
Person/number (1 sg, .615; 3 sg, .351;
other, .293);
Deniteness (denite, .658; indenite,
.392)
5 Pan, Mei 113 25 .189 Verb type (tudni, .855; vanni, .554; all
other verbs, .348);
Person/number (3 sg, .751; 1 sg, .558;
other, .129)

Notes: Group 1, X2/cell .000, log likelihood 38.328; Group 2, X2/cell


.3360, log likelihood 45.579; Group 3, X2/cell 1.6133, log likelihood
144.693; Group 4, X2/cell 1.1893, log likelihood 144.730; Group 5,
X2/cell .0908, log likelihood 50.534.

The results in Table 6 illustrate several general tendencies. First, and


perhaps most importantly for the current discussion, verb marking by
all speakers except Xiao Wang, who achieved an accuracy rate of 89%,
is constrained by more than one linguistic inuence. For the lowest pro-
ciency speakers, Pan and Mei, the frequent verbs tudni and vanni as
well as 3 sg and 1 sg favor native-like use (although only 25% of the verbs
analyzed were used according to native-speaker norms). For speakers at
the next highest level, Lan and Xiao Hong, denite verbs signicantly
favor native-speaker usage, while the eects of frequent verbs and 1 sg
376 Robert Bayley

are similar to their eects for speakers at the lowest level. Continuing up
the prociency scale, tense begins to have an eect with Ai Hua and Liu,
with the present tense favoring native-like usage. At the next highest level,
tense continues to be signicant, as are verb type and deniteness. Person/
number is no longer signicant, but that may be an artifact of the smaller
number of tokens for Chang and Zhou compared to Ai Hua and Liu.
Finally, when we reach the most procient speaker, Xiao Wang, whose
corrected mean, .923, has reached the usual criterion for acquisition, we nd
that only the distinction between regular and irregular verbs is signicant.
These results suggest that, given sucient data, we can use Varbrul and
other implementations of logistic regression to trace individual patterns of
acquisition.10 Indeed, as Berdan (1996) shows in his reanalysis of Schumanns
(1978) data on negation by Alberto, such analysis can even enable us to
correct long-held misconceptions. Although he does not deal with tense and
aspect, Berdan does show clearly that Alberto, although he certainly was not
a rapid acquirer, was making some progress in acquiring the English system
of negation during the time Schumann recorded him. Finally, although I
have used GoldVarb to illustrate the advantages of multivariate analysis, as
noted above, commercial implementations of logistic regression produce
equivalent results, as Morrison (2005) demonstrates.11

4. Testing alternative hypotheses

As we have seen, multiple inuences constrain L2 tense marking and


sometimes the distribution of forms in learner production makes it dicult
to disentangle various possible inuences. In this section, I briey consider
one method of disentangling the roles of narrative structure and lexical
aspect, both of which have ample theoretical and empirical support.
In one of the more important of an extensive series of studies of L2
tense and aspect, Bardovi-Harlig (1998) illustrated a solution to what she

10. Paolillo (2002) estimates that we need at least 100 tokens for a Varbrul analysis,
although more tokens are obviously better. Guy (1980), using English native-
speaker data to examine coronal stop deletion, suggests that individual and
group patterns converge once we reach 20 tokens per cell for each speaker.
11. For an example of how to convert GoldVarb output into the regression coe-
cients produced by the logistic regression module in SPSS, see Morrison (2005).
See Rietveld and van Hout (1993) for a clear discussion of the interpretation
of the output of the logistic regression module in SPSS.
Data Analysis: Quantitative approaches 377

referred to as the problem of conspiring factors in L2 tense-aspect


marking. Bardovi-Harlig elicited both oral and written narratives based
on a lm excerpt from 37 learners of English representing a wide range of
prociency levels. The total corpus contained 2,779 verb tokens, 1,318 in
the written narratives and 1,461 in the oral narratives. Table 7 shows the
distribution by modality, grounding, and aspectual class.

Table 7. Distribution of English verbs by grounding, modality, and aspectual


class (Bardovi-Harlig 1998: 483)
# Verbs Stative Activity Accomplishment Achievement
Written
Total 1,318 226 223 206 663
Background 402 214 73 29 86
Foreground 916 12 130 177 577
Oral
Total 1,461 246 235 314 666
Background 482 223 108 47 104
Foreground 979 23 127 267 562

Overall, the results show a clear eect for grounding, particularly in the
written narratives. The more advanced the learners, the greater the use of
the past tense and, in both written and oral narratives, learners used more
past tense forms in foreground than in background clauses (see Bardovi-
Harlig 1998: 489). However, as we see in Table 7, tokens were very un-
evenly distributed. For example only 1.85% of the verbs in foreground
clauses were statives and only 21.5% of the verbs in background clauses
were achievements.
To deal with this case of interaction, or conspiring factors, Bardovi-
Harlig (1998) presented separate analyses by level, grounding, and aspec-
tual class. Although there is some uctuation, particularly in the oral
narratives, the results show clearly that both grounding and aspectual
class constrain tense marking. Table 8 summarizes the results for all levels
by grounding, modality, and aspectual class (with statives excluded).
On the basis of results such as those presented in Table 8, which are
presented in much more detail in the original article, Bardovi-Harlig con-
cludes that achievements are the most likely verbs to be inected for
378 Robert Bayley

Table 8. Percentage of use of simple past by grounding,


modality, and aspectual class (Bardovi-Harlig 1998)
Activity Accomplishment Achievement
Written
Foreground 52 70 74
Background 10 35 70
Oral
Foreground 20 47 64
Background 15 30 69

simple past regardless of grounding. Accomplishments are next most likely


to be inected for simple past, but they are also aected by grounding,
with foregrounded accomplishments more likely to be inected than back-
grounded accomplishments. Activity verbs are the least likely of dynamic
verbs to be inected for simple past and again, foregrounded activities
are more likely to be marked than backgrounded activities (1998: 498).12
Perhaps most importantly, Bardovi-Harlig clearly shows the importance
of attending to multiple possible explanations of the patterns observed
in interlanguage production and she does so in a way that is highly trans-
parent.

5. An alternative method: Implicational scaling

Implicational scaling, introduced to linguistics by David De Camp (1971)


and subsequently used extensively in the study of creole languages, oers
a number of advantages, including transparency and a clear display of
individual speaker patterns. This section outlines those advantages, with
particular reference to testing the Aspect Hypothesis.
Table 9, adapted from De Camps work on the Jamaican post-creole
continuum, provides a convenient example of implicational scaling. This
particular example shows seven speakers ranged on various points from

12. Bardovi-Harligs (1998) result concerning the importance of lexical aspect is


contradicted, at least for L2 Spanish, in a recent study by Salaberry (2011),
who found that grounding is the construct that most clearly distinguishes
learners from native speakers (184).
Data Analysis: Quantitative approaches 379

basilectal creole to relatively standard English and six variables: nyam/eat,


nanny/granny, no ben/didnt, pikni/child, /t/ varying with /y/, and /d/ vary-
ing with //. A plus sign on the right implies that a speaker will use the
more standard forms of all variables to the left, while a minus sign implies
that a speaker will use the creole form of all variables to the right. Speaker
4, for example, who uses didnt rather than no ben will also use granny
rather than nanny and eat rather than nyam. Speaker 4, however, uses the
creole term pikni rather than the more standard child. Thus, this speaker
does not vary between /t/ and /y/ or between /d/ and //. Finally, a plus
implies that all of the speakers higher on the scale will also use the more
standard variant of that form, while a minus implies that all speakers
lower on the scale will use the creole form. Thus, speakers 1, 2, and 3,
like speaker 4, all use the more standard didnt rather than no ben, whereas
speakers 5, 6, and 7, again like speaker 4, all use the creole pikni rather
than the more standard child.

Table 9. Implicational scale for the Jamaican creole continuum (Adapted from
DeCamp 1971: 356 by Rickford 1991)
Variables
Speakers A B C D E F
1
2
3
4
5
6
7
Key: Minus (): nyam nanny no ben pikni /t/ /d/
Plus (): eat granny didnt child /tPy/ /dP/

When we consider the Aspect Hypothesis as proposed by Andersen


(1991), it is fairly obvious that we have implicational patterns for past mark-
ing. Table 10, for example, shows an implicational model of the acquisition
of the Spanish preterite. Speaker 1 has yet to acquire any past tense mor-
phology and hence all aspectual categories are marked with a minus sign.
380 Robert Bayley

Speaker 3 has acquired some inectional morphology and marks telics


and punctuals with preterite morphology, but not activities or states.
Speaker 5 has acquired the full system of preterite marking and uses it
with verbs of all aspectual classes. Note that for the imperfect, we would
simply reverse the aspectual classes.

Table 10. An implicational model of the acquisition of the Spanish preterite


Speaker Punctual Telic Activities States
1
2
3
4
5

Several years ago, I used implicational scaling to document the loss of


tense-aspect distinctions among Mexican-background children in northern
California and south Texas (Bayley 1999). Children ranged in age from
4 to 12 and had various levels of prociency in Spanish and English
(see Schecter and Bayley 2002 for an account of the families involved in
the study). Because we were interested in childrens prociency in both
Spanish and English we used two wordless picture books to elicit narra-
tives: Mayers (1969) Frog, where are you? and Mayer and Mayers (1971)
A boy, a dog, a frog, and a friend. Half of the children were asked to retell
Frog, where are you? in Spanish and half were asked to retell the alternate
story. We followed the same procedure in the English narrative elicitation
task. As the results in Table 11 suggest, both stories provided opportunities
for children to use verbs of all aspectual classes.
The basic question underlying the study was whether inectional mor-
phology is lost (or not fully acquired) by speakers in a community under-
going language shift in the reverse order that the Aspect Hypothesis pre-
dicts that it will be acquired. Hence, in contrast to the example in Table
10, I placed the most procient speakers, those who used appropriate
inectional morphology with verbs of all aspectual classes on the top of
the scale. Table 11 shows the results for use of the preterite by aspec-
tual class. Note that 96.4% of the cells follow the predicted implicational
pattern.
Data Analysis: Quantitative approaches 381

Table 11. Preterite tense use by aspectual class in Mexican-origin childrens


Spanish narratives (Bayley 1999)
Subject Narrative Achievement Accomplishment Activity State

SF03 Friend
SF05 Friend
SF12 Frog
SF13 Friend
SF15 Friend
SA15.2 Friend
SAN2 Friend
SAN11 Frog
PK08 Frog
PK11 Friend ()
SF07 Friend ()
SF08 Frog ()
SF09 Frog ()
SAN1 Friend ()
SA03 Friend ()
PK13 Frog ()
SF02 Frog
SF17 Friend
SF06 Frog
PK07 Frog
SAN3 Friend
SAN8 Friend
SAN9 Frog
SAN12 Friend
SF10 Frog
SF01 Friend
SF16 Frog
SF18 Frog
SA01 Frog
SA04 Friend
SA09 Friend
SA15.1 Frog
SA21 Friend
PK02 Frog
PK09 Friend
SAN5 Friend
SAN7 Friend
SAN10 Friend
SF11 Friend
SF19 Friend
SF20 Frog
SAN4 Friend
SA16 Friend
PK01 Friend
PK05 Friend
PK06 Friend
PK10 Friend
PK12 Frog
SA02 Frog
Notes: IR .964; cells that do not scale are in parentheses; SA - San Antonio 1012 year
olds, SF - San Francisco 1012 year olds, SAN San Antonio 67 year olds, PK San Antonio
45 year olds.
382 Robert Bayley

Analysis of imperfect marking by aspectual class produced similar re-


sults. It was thus possible to conclude that verbal inectional morphology
did in fact seem to be lost in the reverse order to which the Aspect Hypo-
thesis predicts that it will be acquired. However, the study does contain a
major limitation. Children were scored with a plus if they produced any
examples of a particular form for each aspectual class. This was necessary
because the elicited narratives were generally quite brief and because non-
prototypical forms such as preterite statives or imperfect achievements are
relatively rare. Hence, the results show that children who produced pre-
terite and imperfect forms of verbs of all aspectual classes had those forms
in their repertoire. The results do not, however, show the extent to which
they had full control of verbal morphology.
In summary, implicational scales, although widely used in creole studies
(Rickford, 2002), have been somewhat neglected in studies of L2 acquisi-
tion (see, e.g., Gatbonton 1978; Tromovich, Gatbonton, and Segalowitz
2007 for implicational studies of L2 phonology). However, given the fact
that the Aspect Hypothesis predicts an implicational pattern, such scales
appear highly suitable for the study of tense and aspect, particularly since
they enable us to clearly display results for individuals.

6. Conclusion

This chapter has provided examples of three ways of analyzing tense and
aspect in learner production: logistic regression with Varbrul, analysis by
separate factors, and implicational scaling. These types of analysis are use-
ful primarily for studies that rely on production data, particularly data
collected in interviews and in narrative elicitation tasks. For controlled
experiments, including cloze tests, where the distribution of verbs of dier-
ent aspectual classes can be controlled to deal with the distributional
imbalance that characterizes natural language, more conventional statisti-
cal methods such as ANOVA are preferred. Finally, to paraphrase Allan
Bell (1984), we have the issue of statistics as audience design. Although
Varbrul has been used in a variety of implementations for more than
three decades and a number of recent articles in the major SLA journals
have used the program to analyze their data (e.g., X. Li 2010; Rau, Chang,
and Tarone 2009), it is unfamiliar to many outside of sociolinguistics.
Therefore, we might be advised to use the logistic regression modules in
commercial packages or in R when we address audiences who are likely
to be more familiar with the type of output those programs produce.
Data Analysis: Quantitative approaches 383

References

Andersen, Roger
1991 Developmental sequences: The emergence of aspect marking in
second language acquisition. In T. G. Huebner & C. A. Ferguson
(Eds.), Crosscurrents in second language acquisition and learning
theories, 305324. Amsterdam: John Benjamins.
Andersen, Roger & Yasuhiro Shirai
1994 The primacy of aspect in rst and second language acquisition.
In William Ritchie & Tej J. Bhatia (Eds.), Handbook of second
language acquisition, 527570. San Diego: Academic.
Bardovi-Harlig, Kathleen
1994 Anecdote or evidence? Evaluating support for hypotheses con-
cerning the development of tense and aspect. In Elaine Tarone,
Susan M. Gass & Andrew D. Cohen (Eds.), Research methodology
in second language acquisition, 4160. Hillsdale, NJ: Lawrence Erl-
baum.
Bardovi-Harlig, Kathleen
1998 Narrative structure and lexical aspect: Conspiring factors in
second language acquisition of tense-aspect morphology. Studies
in Second Language Acquisition 20. 471508.
Bardovi-Harlig, Kathleen
2000 Tense and aspect in second language acquisition: Form, meaning,
and use. Oxford: Blackwell.
Bardovi-Harlig, Kathleen & Dudley Reynolds
1995 The role of lexical aspect in the acquisition of tense and aspect.
TESOL Quarterly 29. 107131.
Bayley, Robert
1991 Variation theory and second language learning: Linguistic and
social constraints on interlanguage tense marking. Stanford, CA:
Stanford University, Ph.D. dissertation.
Bayley, Robert
1994 Interlanguage variation and the quantitative paradigm: Past-tense
marking in Chinese-English. In E. Tarone, S. M. Gass & A. D.
Cohen (Eds.), Research methodology in second language acquisi-
tion, 157181. Hillsdale, NJ: Lawrence Erlbaum.
Bayley, Robert
1996 Competing constraints on variation in the speech of adult Chinese
learners of English. In R. Bayley & D. R. Preston (Eds.), Second
language acquisition and linguistic variation, 97120. Amster-
dam: John Benjamins.
Bayley, Robert
1999 The primacy of aspect hypothesis revisited: Evidence from lan-
guage shift. Southwest Journal of Linguistics 18(2). 122.
384 Robert Bayley

Bayley, Robert
2002 The quantitative paradigm. In J. K. Chambers, P. Trudgill &
N. Schilling-Estes (Eds.), The handbook of language variation
and change, 117141. Oxford: Blackwell.
Bayley, Robert & Juliet Langman
2004 Variation in the group and the individual: Evidence from second
language acquisition. IRAL 42. 303319.
Bell, Alan
1984 Language style as audience design. Language in Society 13: 145
204.
Berdan, Robert
1996 Disentangling language acquisition from language variation. In
Robert Bayley & Dennis R. Preston (Eds.), Second language
acquisition and linguistic variation, 203244. Amsterdam: John
Benjamins.
Berman, Ruth A. & Dan Slobin
1994 Development of linguistic forms: English. In Ruth A. Berman &
Dan I. Slobin (Eds.), Relating events in narrative: A cross-
linguistic developmental study, 127188. Hillsdale, NJ: Lawrence
Erlbaum.
Cai, Jinting
2010 The inuence of Chinese aspect marker le on the simple past
marking in English interlanguage: An empirical study of univer-
sity students in China. Chinese Journal of Applied Linguistics 33
(2). 3547.
Cedergren, Henrietta & David Sanko
1974 Performance as a statistical reection of competence. Language
50. 333355.
Collins, Laura
2002 The roles of L1 inuence and lexical aspect in the acquisition of
temporal morphology. Language Learning 52. 4394.
Collins, Laura, Pavel Trommovich, Joanna White, Waclir Cardoso & Marlise
Horst
2009 Some input on the easy/dicult grammar question: An empirical
study. Modern Language Journal 93. 336353.
Comajoan, Llorenc
2005 The acquisition of perfective and imperfective morphology and
the marking of discourse in Calalan. In Dalia Ayoun & M.
Rafael Salaberry (Eds.), Tense and aspect in Romance languages,
3378. Amsterdam: John Benjamins.
De Camp, David
1971 Toward a generative analysis of a post-creole continuum. In Dell
Hymes (Ed.), Pidginization and creolization of languages, 349
370. Cambridge: Cambridge University Press.
Data Analysis: Quantitative approaches 385

Du, Patricia & Duan Duan Li


2002 The acquisition and use of perfective aspect in Mandarin. In
M. R. Salaberry & Y. Shirai (Eds.), The L2 acquisition of tense-
aspect morphology, 417453. Amsterdam: John Benjamins.
Dunn-Rankin, Peter
1983 Scaling methods. Hillsdale, NJ: Lawrence Erlbaum.
Ellis, Rod
1987 Interlanguage variability in narrative discourse: Style shifting in
the use of the past tense. Studies in Second Language Acquisition
9. 120.
Erbaugh, Mary S.
1992 The acquisition of Mandarin. In Dan I. Slobin (Ed.), The crosslin-
guistic study of language acquisition, vol. 3, 373455. Hillsdale,
NJ: Lawrence Erlbaum.
Flashner, Vanessa E.
1989 Transfer of aspect in the English oral narratives of native Russian
speakers. In Hans W. Dechert & Manfred Raupach (Eds.),
Transfer in language production, 7297. Norwood, NJ: Ablex.
Gatbonton, Elizabeth
1978 Patterned phonetic variability in second-language speech: A
gradual diusion model. Canadian Modern Language Review
34. 335347.
Geeslin, Kimberly & Aarnes Gudmestad
2010 An exploration of the range and frequency of occurrence of forms
in potentially variable structures in second-language Spanish.
Studies in Second Language Acquisition 32. 433463.
Geeslin, Kimberly & Pedro Guijarro-Fuentes
2008 Variation in contemporary Spanish: Linguistic predictors of
estar in four cases of language contact. Bilingualism: Language
and Cognition 11. 365380.
Goad, Heather, Lydia White & Jerey Steele
2003 Missing inection in L2 acquisition: Defective syntax or L1-
constrained prosodic representation. Canadian Journal of Lin-
guistics 48: 243263.
Guy, Gregory R.
1980 Variation in the group and the individual: The case of nal stop
deletion. In William Labov (Ed.), Locating language in time and
space, 136. New York: Academic Press.
Gorman, Kyle & Daniel Ezra Johnson
2013 Quantitative analysis. In Robert Bayley, Richard Cameron &
Ceil Lucas (Eds.), The Oxford handbook of sociolinguistics, 214
240. Oxford: Oxford University Press.
Hakuta, Kenji
1976 A case study of a Japanese child learning English as a second
language. Language Learning 26. 321351.
386 Robert Bayley

Howard, Martin
2004 On the interactional eect of linguistic constraints on interlan-
guage variation: The case of past time marking. IRAL 42. 319
334.
Jia, Li & Robert Bayley
2008 The (re)acquisition of perfective aspect marking by Chinese
heritage language learners. In Agnes Weiyun He & Yun Xiao
(Eds.), Chinese as a heritage language: Fostering rooted world
citizenry, 205222. Honolulu: University of Hawaii, National
Foreign Language Resource Center.
Johnson, Daniel Ezra
2009 Getting o the GoldVarb standard: Introducing Rbrul for mixed-
eect variable rule analysis. Language and Linguistics Compass
3. 359383.
Kumpf, Lorraine
1984 Temporal systems and universality in interlanguage: A case
study. In Fred Eckman, Lawrence H. Bell & Diane Nelson
(Eds.), Universals of second language acquisition, 132143. Rowley,
MA: Newbury House.
Kwan-Terry, Anna
1979 Two progressive aspect markers in Chinese. In D. L. Nguyen
(Ed.), South-East Asian Linguistic Studies, Vol. 4 (Pacic Lin-
guistics, Series C, No. 44), 213232. Canberra: the Australian
National University.
Langman, Juliet & Robert Bayley
2002 The acquisition of verbal morphology by Chinese learners of
Hungarian. Language Variation and Change 14. 5577.
Larson-Hall, Jennifer
2010 A guide to doing statistics in second language research using
SPSS. New York: Routledge.
Li, Charles & Sandra A. Thompson
1981 Mandarin Chinese: A functional reference grammar. Berkeley:
University of California Press.
Li, Ping & Melissa Bowerman
1998 The acquisition of lexical and grammatical aspect in Chinese.
First Language 18. 311350.
Li, Xiaoshi
2010 Sociolinguistic variation in the speech of learners of Chinese as a
second language. Language Learning 60. 366408.
Mayer, Mercer
1969 Frog, where are you? New York: Dial Press.
Mayer, Mariana & Mercer Mayer
1971 A boy, a dog, a frog, and a friend. New York: Dial Press.
Data Analysis: Quantitative approaches 387

Morrison, Georey
2005 Dat is what the PM said: A quantitative analysis of Prime Minis-
ter Chretiens pronunciation of English voiced dental fricatives.
Cahiers linguistiques dOttawa 33. 121.
Paolillo, John
2002 Analyzing linguistic variation. Stanford, CA: Center for the
Study of Language and Information.
Rau, D. Victoria, Hui-Huan Ann Chang & Elaine Tarone
2009 Think or sink: Chinese learners acquisition of the English voice-
less fricative. Language Learning 59. 581621.
Regan, Vera
2004 The relationship between the group and the individual and the
acquisition of native speaker variation patterns: A preliminary
study. IRAL 42. 335347.
Reitveld, Toni & Roeland van Hout
1993 Statistical techniques for the study of language and language be-
haviour. Berlin: Mouton de Gruyter.
Reitveld, Toni & Roeland van Hout
2005 Statistics in language research: Analysis of variance. Berlin:
Mouton de Gruyter.
Rickford, John R.
2002 Implicational scales. In J. K. Chambers, Peter Trudgill &
Natalie Schilling-Estes (Eds.), The handbook of language varia-
tion and change, 142167. Oxford: Blackwell.
Rousseau, Pascal & David Sanko
1978 Advances in variable rule methodology. In David Sanko (Ed.),
Linguistic variation: Models and methods, 5769. New York:
Academic.
Salaberry, M. Rafael
2000 The development of past-tense morphology in L2 Spanish. Amster-
dam: John Benjamins.
Salaberry, M. Rafael
2005 Evidence for transfer of knowledge of aspect from L2 Spanish to
L3 Portuguese. In Dalia Ayoun & M. Rafael Salaberry (Eds.),
Tense and aspect in Romance languages, 179210. Amsterdam:
John Benjamins.
Salaberry, M. Rafael
2011 Assessing the eect of lexical aspect and grounding of L2 Spanish
past tense morphology among English speakers. Bilingualism:
Language and Cognition 14. 184202.
Sanko, David, Sali A. Tagliamonte & Eric Smith
2005 GoldVarb X. A multivariate analysis application. [Computer pro-
gram]. Toronto and Ottawa: Department of Linguistics, Univer-
sity of Toronto and Department of Mathematics, University of
Ottawa.
388 Robert Bayley

Sanko, David, Sali A. Tagliamonte & Eric Smith


2012 GoldVarb Lion. A multivariate analysis application for Macintosh.
[Computer program]. Toronto and Ottawa: Department of Lin-
guistics, University of Toronto and Department of Mathematics,
University of Ottawa.
Sanko, Gillian
1990 The grammaticalization of tense and aspect in Tok Pisin and
Sranan. Language Variation and Change 2. 293312.
Schumann, John
1978 The pidginization process: A model for second language acquisition.
Rowley, MA: Newbury House.
Sebastian, Eugenia & Dan I. Slobin
1994 Development of linguistic forms: Spanish. In Ruth A. Berman
& Dan I. Slobin (Eds.), Relating events in narrative: A cross-
linguistic developmental study, 239284. Hillsdale, NJ: Lawrence
Erlbaum.
Schecter, Sandra R. & Robert Bayley
2002 Language as cultural practice: Mexicanos en el norte. Mahwah,
NJ: Lawrence Erlbaum.
Selinker, Larry & Dan Douglas
1985 Wrestling with context in second language acquisition theory.
Applied Linguistics 6. 6792.
Shi, Ziqiang
1989 The grammaticalization of the particle le in Mandarin Chinese.
Language Variation and Change 1. 99114.
Shirai, Yasuhiro & Roger Andersen
1995 The acquisition of tense-aspect morphology: A prototype account.
Language 71. 743762.
Shirai, Yasuhiro & Atsuko Kurono
1998 The acquisition of tense-aspect marking in Japanese as a second
language. Language Learning 48. 245279.
Sigley, Robert
2003 The importance of interaction eects. Language Variation and
Change 15. 227253.
Silva-Corvalan, Carmen
1994 Language contact and change: Spanish in Los Angeles. Oxford:
Oxford University Press.
Slabakova, Roumyana & Silvina Montrul
2003 Genericity and aspect in L2 acquisition. Language Acquisition
11. 165196.
Tagliamonte, Sali A.
2006 Analysing sociolinguistic variation. Cambridge: Cambridge Uni-
versity Press.
Tajika, Hiroko
1999 Variable patterns of tense/aspect marking in interlanguage.
Minneapolis, MN: University of Minnesota, Ph.D. dissertation.
Data Analysis: Quantitative approaches 389

Tarone, Elaine & Liu, Guo-Qiang


1995 Situational context, variation and SLA theory. In Guy Cook
& Barbara Seidlhofer (Eds.), Principle and practice in applied
linguistics: Studies in honour of H. G. Widdowson, 107124.
Oxford: Oxford University Press.
Teutsch-Dwyer, Marya W.
1994 Means of expressing temporality in untutored adult second lan-
guage acquisition learner variety: A longitudinal analysis. Stan-
ford, CA: Stanford University, Ph.D. dissertation.
Thompson, J. Charles
1968 Aspects of the Chinese verb. Linguistics 38. 7076.
Tromovich, Pavel, Elizabeth Gatbonton & Norman Segalowitz
2007 A dynamic look at L2 phonological learning: Seeking processing
explanations for implicational phenomena. Studies in Second
Language Acquisition 29. 407448.
Vendler, Zeno
1967 Linguistics and philosophy. Ithaca, NY: Cornell University Press.
Wang, William S.-Y
1965 Two aspect markers in Mandarin. Language 41. 457470.
Wen, Xiaohong
1995 Second language acquisition of the Chinese particle le. Interna-
tional Journal of Applied Linguistics 5. 4562.
Wen, Xiaohong
1997 Acquisition of Chinese aspect: An analysis of the interlanguage
of learners of Chinese as a foreign language. ITL: Review of
Applied Linguistics 117/118. 126.
Wolfram, Walt
1985 Variation in interlanguage tense marking: A case for the obvious.
Language Learning 35. 229253.
Wolfson, Nessa
1982 On tense alternation and the need for analysis of native speaker
usage in second language acquisition. Language Learning 35.
229253.
Wul, Stefanie, Nick Ellis, Ute Romer, Kathleen Bardovi-Harlig & Chelsea
LeBlanc
2009 The acquisition of tense-aspect: Converging evidence from corpora,
cognition, and learner constructions. Modern Language Journal 93.
354369.
Young, Richard & Robert Bayley
1996 Varbrul analysis for second language acquisition research. In
R. Bayley & D. R. Preston (Eds.), Second language acquisition
and linguistic variation, 253306. Amsterdam: John Benjamins.
Chapter 11
Data analysis: The qualitative analysis of actionality
in learner language

Anna Giacalone-Ramat and Stefano Rastelli

1. The qualitative analysis of actionality in learner language

Qualitative research in second language (L2) acquisition is based on descrip-


tive data that do not make use of statistical procedures (Mackey and Gass
2005: 163). According to Mackey and Gass, qualitative research is an
open-ended process: research questions are inductive and rather than
being dened before data collection they emerge and are ne-tuned
as data are examined. By qualitative analysis in this study we mean a
corpus-driven research that attempts to describe the actional content1 of
the verbs used by L2 learners in the light of the independent nature of the
interlanguage (IL) and of its development. The analysis presented in this
article is qualitative because its conclusions (a) are inductive, (b) are based
on the observation of a few subjects, and (c) are not falsiable (i.e., the
conditions of learners productions are not controlled and replicable).

1.1. Actionality in a second language and the comparative fallacy


Despite its evident aws, a qualitative analysis of L2 actionality is not
only necessary, but also it is a pre-requisite for quantitative analysis in
order to avoid the comparative fallacy. Our point stems from a question
formulated in the title of Lakshmanan and Selinker (2001): How do we
know what learners know? This question gave rise to a lively point-counter-
point debate in the following years (see rst Lardiere 2003 and then Shirai
2007). In their article, Lakshmanan and Selinker provided some reasons
why the suppliance rate of a given form in an expected (mandatory) con-
text alone should not be considered a valid criterion for evaluating L2
acquisition. They pointed out that the relationship between the emergence

1. By actional content of a predicate we mean the lexical aspect features matrix


of that predicate; that is, its specication for being [edurative] and [etelic] in
learners representations.
392 Anna Giacalone-Ramat and Stefano Rastelli

and the use of an item (e.g., the inectional morphology) on one side and
the development of the representation of the corresponding abstract cate-
gory (e.g., Tense or Aspect) on the other side is not a direct one. In other
words, what a learner produces does not necessarily coincide with what
that learner meant to say and the recoverability of the learners intention
may be quite a dicult task to accomplish (401). Lakshmanan and
Selinker suggested that we need more rigorous criteria if we want to
avoid the comparative fallacy, which consists of super-imposing the target-
like categories on learner data (Bley-Vroman 1983; Klein and Perdue
1992). This section deals with the importance of discussing these criteria
(see also Chapters 8 and 9). Reviewing the ndings of an earlier work by
Lardiere (1998), Lakshmanan and Selinker suggested that verb semantics
(Actionality) and discourse grounding distinctions as proposed in the Aspect
Hypothesis (AH, Andersen 2002; Bardovi-Harlig 2000) studies are valid
criteria to discriminate genuine (non target-like oriented) obligatory con-
texts for past tense markings because telic verbs and foreground events
attract past perfective markings independently of the standpoint of the
target language.
In her reply to Lakshmanan and Selinker, Lardiere (2003) pointed out
that the matter of investigating a learners past tense marking in relation
to inherent lexical aspect (according to the Aspect Hypothesis) and dis-
course foregrounding is in fact at high risk of committing the comparative
fallacy. Lardiere stated that the Aspect Hypothesis studies appear to
assume native speakers intuitions about the meaning of verb stems in
assigning coding categories such as activity, achievement, etc. to the data,
and in applying diagnostic tests for those categories . . . these assump-
tions may indeed obscure our understanding of the L2 idiolect (136).
Lardiere added that in SLA literature she could not nd any discussion
about whether the linguistic tests normally utilized to code data for L1
aspectual categories (operational, compatibility tests) were applicable to
learner data (138). Actually, this point is taken up by Giacalone Ramat
and Rastelli (2008) and Rastelli (2008, 2009), who argue against the pre-
dictive power of actional tests designed for mature languages when used
for IL data (see Section 4.2). For instance, our data show that unlike in
L1 Italian in L2 Italian for-x-time expressions seem to be compatible
also with achievements, and in-x-time expressions seem to be compatible
with activities to the extent that it is dicult to conclude whether the L2
learners semantic representations coincide or dier from those of native
speakers. Shirai (2007), in his reply to Lardiere (2003), acknowledged
that AH studies might have assumed more semantic representations on
Data analysis: The qualitative analysis of actionality in learner language 393

the part of the learners than warranted and concluded that: . . . in the anal-
ysis of production data, one should attempt maximum rigour in classica-
tion without reading in too much about learners semantic representation
(60). Shirai attempted to reduce the risk taking into account the cross lin-
guistic variations of actional value (see Chapter 8, Sections 1, 2 and 3.1.1).
In this and in the following sections, we argue that the problems do not
arise from actional misclassication, but from taking for granted that L2
learners have clear and steady actional representations of the verbs they
are using. A previous and rigorous qualitative analysis is necessary not
only to validate production data as Shirai seems to suggest but also
for both preparing the eld and for narrowing the scope (thus increasing
the reliability) of quantitative analysis. Going beyond Lardiere (2003), we
ask: How do we know for example that a verb that is classied telic in
the target-language is also telic in the interlanguage? Shirai (Chapter 8)
suggests that L1 transfer and L1-L2 discrepancies should be taken into
account in order to answer this question. Unlike Shirai, we would like to
suggest that especially in initial IL the actional content of predicates
remains largely underspecied and that L2 learners are more concerned
about the general meaning of the verbs they use, rather than about the
actional content of those verbs. The question about learners representations
is the kind of a crucial and unavoidable question that quantitative research
alone cannot answer directly, but that it can benet from it indirectly.
Quantitative research can tell us, for instance, if the percentage of verbs
conjugated in the perfective past is signicantly higher for telic verbs than
for atelic verbs or whether perfective past is overextended more frequently
with telic verbs rather than with atelic ones. Instead, qualitative research
could be directed to understand whether learners know the peculiar prop-
erties of telicity of the second language. Investigating the properties of
telicity means answering three fundamental questions at least (which will
be possibly answered in Section 3): (1) Are prospective telic verbs in the
IL used also in virtually incompatible temporal frames (such as for x-time
expressions)? (2) Do L2 learners overuse a small number of basic verbs
regardless of their actional content? (3) Can qualitative longitudinal studies
on learner corpora reveal the existence of a bias for compositional or for
lexical expressions of actionality over time?
We are aware that qualitative analysis based on a limited number of
data (typically, on a convenience sample as Dornyei 2007 puts it) is
not sucient to draw conclusions about the signicance of facts (1)(3).
Nevertheless, qualitative analysis might be useful to cast doubts on the
394 Anna Giacalone-Ramat and Stefano Rastelli

fact that especially to L2 initial learners learning and knowing the


general meaning of a L2 verb entails knowing also its actional content.
To sum up, quantitative research alone runs the risk of being biased by
the assumption that L2 learners know the actional content of L2 verbs
the way native speakers do. In some important sense, the initial, not proved
assumptions made by qualitative research through the analysis of a few,
not representative data are the necessary, bad medicine for quantitative,
statistical analysis in order to avoid the comparative fallacy.
This paper is articulated as follows. In Section 1.2, three criteria of qual-
itative research are described in detail. In Section 1.3, we try to explain why
our proposal is alternative to the Distributional Bias Hypothesis. Section 2
is dedicated to a critical evaluation of the main ndings on the acquisi-
tion of aspect in L2 Italian. In Section 3, our data are presented, and in
Section 4 they are discussed in the light of the hypothesis that L2 learners
independently of their L1s may leave the actional content of predicates
underspecied also for a long time after they have learned the general
meaning of a verb. Finally, Section 5 underlines the advantages of quali-
tative investigation for future experimental research.

1.2. Rationale and procedures of qualitative research on L2 aspect


Dornyei (2007: 37) stated some important methodological points about
the characters of qualitative research that apply also to the procedure of
investigating qualitatively L2 actionality. Dornyei argued that the advan-
tage of qualitative research is its exploratory nature; that is, its capacity to
deal with new details or openings that may emerge during the process
of investigations. In the case of learners actionality, when approaching
data, one should not think of a verb being telic or atelic in virtue of its
lexical properties before having looked carefully at the context where it
occurs. Learners can use on target-like phrasal features (e.g., prepositions,
adverbs) to convey a telic interpretation for what we consider as being
atelic verbs. These phrasal features are often unpredictable, inconsistent,
and elusive. As such, they cannot be used as independent variables in a
correlation analysis, but they can only be deduced on a posteriori basis.
Dornyei also stated that qualitative research is more useful than quantita-
tive research for the purpose of longitudinal examination of dynamic phe-
nomena (Dornyei 2007: 40). If we advance the hypothesis that verb lexical
aspect too is learned gradually (it is a dynamic phenomenon) and that
learners do not begin by projecting the actional properties of verbs, then
we must have a considerable amount of longitudinal data at our disposal
Data analysis: The qualitative analysis of actionality in learner language 395

to support such a hypothesis. Thus, qualitative analysis relies on longitu-


dinal data.
Among the weaknesses of qualitative research quoted by Dornyei (2007),
sample size and generalizability are the most evident. Actually, the replica-
tion of ndings is a problem that aects all empirical studies (both quali-
tative and quantitative) in the eld of SLA research: the inconsistent and
unstable nature of second language knowledge also makes exact replica-
tion quite impossible (Gass and Selinker 2008: 72). In Section 3, we will
try to make even clearer that the goal of qualitative research at least as
we conceptualize it is to provide insights for quantitative research simply
by widening the scope of investigation to all phrasal features and by cast-
ing doubts about the initial assumption that learners aspectual representa-
tions resemble or even coincide to ours. In this respect, we do not think
that qualitative data should be counted or to put it dierently that
the analysis of qualitative data can be quantitative, for instance, by means
of percentage comparison (Ellis and Barkhuizen 2005: 253). In Section 3,
we thus suggest a sense in which qualitative research can be used as a
starting point for quantitative research in the eld of L2 actionality. The
procedure of qualitative analysis we are going to illustrate in the following
paragraphs is set out in three points:

(a) Referent check and within-subject variability. The rst point is trying to
establish to what degree we are in control of the referent, which is the state
of aairs described by a learner when recounting a scene. Let us assume
that the elective eld of application for qualitative research is re-telling
task data (see Chapter 7). The main goal of researchers should be to reach
an acceptable degree of control over the intended meaning; that is, over
what is being told by learners about a picture or a lm scene. There is an
additional source of confusion that complicates the task: L2 learners are
likely to use non target-like words and expressions that may mislead or
bias the interpreters towards one or another reading. One possible solution
is to look closely at within-subject variation; that is, at instances where
subjects serve as their own controls in repeated identical elicitation tasks at
dierent times. For instance, to elicit our data, the same subjects are asked
to re-tell the Frog story (see Section 3.1) twice or three times, at an interval
of 3 to 5 and 11 months (See Section 3). Within-subject variations are more
likely to be attributed to a modication in a learners competence rather
than to a dierence in the researchers interpretation (even though the
latter can never be neutralized).
396 Anna Giacalone-Ramat and Stefano Rastelli

(b) L1-L2 comparison against verb pairs. The second point is choosing
pairs of verbs in learner data which are both found to be very frequent in
the input and whose basic, general meaning seems to be already clear to
learners. These pairs are of two kinds: actional pairs and phasal pairs. In
native Italian, verbs of the former type share the basic meaning and dier
for one or two actional traits, in this case for the trait [etelic]. For example,
guardare to look, to watch is atelic, while vedere to see in its basic
meaning of perceiving that something entered the visual eld is telic
(see Section 3). Instead, verbs that form phasal pairs can be looked at as
being dierent phases of the same event (for the meaning of phase in the
event structure, see Moens 1987). For example, cercare to search may
serve as the preparatory phase of the event, while trovare to nd can
serve as its culmination point. Comparable to phasal verbs (to some extent
at least) are the reversive verbs (Cruse 1997) such as to teach / to learn,
to give / to receive.
Verkuyl (2005: 27) specied that the notion of culmination is ontological,
not a linguistic one. According to Verkuyl, this notion is rooted in our
tendency to give the nal limit of any event a prominent place given the
time course, but there is no detectable linguistic material at the VP-level
that can be said to express specically a culmination point. We too assume
that the notion of culmination is pre-linguistic. That is, learners might
want or need to express it long before they know which verbs are telic
and which are not in the lexicon of the target language. In this case, the
structural encoding of telicity may precede the lexical encoding (Rastelli
2008). Our hypothesis (Section 4) is that verbs belonging to actional and
phasal pairs are frequently exchanged by initial learners, regardless of
their actional content and that frequency in the input is not a relevant
factor to the choice of either verb. It is important, in order for the verbs
to be valid candidates for entering one or another pair, that learners are
aware of their general meaning, that is, of what is common between verbs
such as see and look (perception with the eyes), walking, going, coming
(displacement), listening and hearing (perception with the ear), etc. This
condition is not in contradiction with our research hypothesis: in fact the
general meaning of guardare and vedere is the physiological act of perceiv-
ing something with the eyes, whereas the general meaning of cercare and
trovare at least in the situation that we will analyze below is the over-
all event that encompasses both. Through the analysis of the sentences
where learners exchange such verbs in Section 3.2, we conclude that
general meanings and phasal meanings have primacy in acquisition over
aspectual features (namely over the [etelic] distinction) and more in
Data analysis: The qualitative analysis of actionality in learner language 397

general that our (native) semantic representations should not be com-


pared to those of learners if we want to avoid the risk of committing the
comparative fallacy.
Finally, all verbs entering both actional and phasal pairs should be
checked carefully against the corresponding learners L1 forms (see Section
3). We believe, in fact, that the actional templates of the rst language
might also aect the path and the rate of acquisition of the actional dis-
tinctions in the second language (see Chapter 8). This is true in two broad
senses at least. First, much more input is needed to gear a learners repre-
sentation from L1 to L2 when verbs which seem similar in their general
meaning dier in their actional content. Second, more generally, one-to-one
correspondences between verb meaning and actional traits do not con-
tribute to overload a learners working memory and are likely to be incor-
porated more quickly in the L2 system. In Section 3.3, we argue that the
phenomena pointed out by our qualitative research on the acquisition of
lexical aspect in L2 Italian are not limited to a particular group of learners
but they spread independently from the learners L1s.
(c) Scope widening. The third point is observing and comparing contexts
where verbs that form aspectual and event pairs occur in learner data.
Particular attention should be given to contexts where verbs occur in a
precise temporal framework, with expressions of time, or with phrasal
items that are used to dene temporal relations. Other important features
to be held in consideration are the number and the semantics of both
internal and external verb arguments, as it is widely reported in literature
about aspect. Fortunately, when the same elicitation task is repeated over
time (as it is in our research protocol), this very fact ensures that verb
arguments can be treated as independent variables without running the
risk of being taken as factors motivating changes occurring in the learners
lexical choices.

1.3. Qualitative analysis and the Distributional Bias Hypothesis


In this section, we discuss in what sense the target-language input may
aect directly a L2 learners aspectual competence. The question is whether
a learners knowledge of the actional content of L2 verbs is reinforced by
their exposure to the biased native input (for a critical review of the main
ndings of the Distributional Bias Hypothesis in L2 Italian, see Section 2).
We argue that qualitative analysis can possibly show that there exist
features or actional properties that initial L2 learners may not be aware
of, despite the fact that the verbs are very frequent in the input both in
absolute and in distributional terms.
398 Anna Giacalone-Ramat and Stefano Rastelli

The Distributional Bias Hypothesis states that native speakers in inter-


actions tend to use each verb morpheme with a specic class of verbs,
also following the aspect hypothesis (Andersen and Shirai 1994: 137).
According to this hypothesis, frequency of verbs in native input, strength
and exclusivity of form-function associations and paucity of non-prototypical
exemplars are predictors for primacy in acquisition of form-function
mappings. Wul et al. (2009) discussed how these dierent features in the
native input (among which frequency and prototipicality of lexical aspect)
aect the acquisition of the tense-aspect system. The authors reported
ndings that showed that in the native corpus the frequency with
which verbs occur with a certain tense-aspect category is inversely propor-
tional to its rank in the frequency table. The authors concluded that L2
learners are sensitive to the systematic correlation among three factors:
item frequency, distinctiveness (of the form-function mapping), and lexical
aspect (366). Ellis (Chapter 3) discusses this issue at length.
Our qualitative analysis of a longitudinal learner corpus shows things
from a dierent point of view. Namely, we want to address cases in which
L2 learners use pairs of verbs whose meanings seem exchangeable, the
preference for one or another being inconsistent over a long period of
time, mutable and not directly attributable to any factor reported in Wul
et al. (2009). Giacalone Ramat and Rastelli (2008) commented upon
narrative data elicited from MK, a Tigrinya (semitic language spoken in
Ethiopia) untutored adult learner of Italian who had been recorded for
seven months while reporting orally about his life in Italy. At the time of
the rst recording, MK had been living in Italy for one month. The study
focused on how MK gradually developed his awareness of the dierence
between the verbs imparare to learn, studiare to study or to attend
class and insegnare to teach throughout twelve recording sessions spaced
over twelve months. The verb imparare to learn in Italian is a gradual
completion verb (for a discussion on this category, see Bertinetto-Squartini
1995). Semantically, gradual completion verbs (such as sorgere to rise,
maturare to ripe), though being telic, express events made of a sequence
of successive partial attainments rather than only one nal achievement.
In contrast, studiare to study and insegnare to teach are atelic, activity
verbs in native Italian. The rst occurrences of imparare were found only
in the ninth recording, where MK used imparare in place of insegnare and
studiare, as sentences (1), and (2) respectively show:
(1) Cane se io imparato bene.
dog if I learned-PAST well
If I taught the dog well.
Data analysis: The qualitative analysis of actionality in learner language 399

(2) Fino a mercoled ho imparato al classe.


until to Wednesday (I) have-AUX learned-PASTP at the classroom
Until Wednesday I have attended class.
In Italian, the verb imparare in the perfective passato prossimo (ho imparato)
is not compatible with durative-culminative adverbials of the kind used by
MK ( no a mercoled, until Wednesday), unless it is completely deteli-
cized and is used as synonymous of the activity verb studiare to study or
to attend class. Again, in the eleventh recording, we found an attempt to
t the verb imparare into a bounded time-span frame, which was ex-
pressed by the adverbial expression tre quattro mesi (three four months):
(3) Quando ho imparato la lingua tre
when (I) have-AUX learned-PAST the language three
quattro mesi.
four months
When I had been learning the language (for) three four months.
In sentence (3), MK seems to be aware of the perfective and past values
associated with the Italian passato prossimo, but he still does not seem to
recognize any telicity (neither gradual nor punctual or semelfactive), in
the verb imparare, since he is not aware of the incompatibility with the
expression of duration. Our claim is that, to MK, the actional content of
the verb imparare might be roughly the equivalent of the Italian verb
studiare, even though a target-like alternation between imparare and
studiare is displayed in the same recording and even in the same sentence,
as in (4):
(4) Io non studio solo l elettricista . . . io sto imparando
I not study only the electrician I AUX learning-GER
la teoria.
the theory
Im not studying only electricity, Im also learning theory.
Only in the twelfth and last recording, MK seems to contrast sto imparando
Im learning with non ho imparato bene I did not learn well. By combin-
ing properly the actional content of imparare alternatively with imperfective
and perfective aspectual morphemes, MK nally seems to fully acknowl-
edge the peculiar two-fold [telic] and [activity] features of this verb:
400 Anna Giacalone-Ramat and Stefano Rastelli

(5) Pero ancora non sto parlando bene la lingua . . . non


but still not AUX speaking well the language not
ho imparato bene.
(I) have-AUX learned well
But Im not speaking the language well yet, I have not learned it
well yet.
The analysis of sentence (5) uttered only a few minutes after (4)
suggests possibly that at that point MK knows what distinguishes imparare
from studiare from the eventive standpoint (imparare can be the resulting
phase of studiare), even though he does not distinguish the two verbs from
the aspectual point of view. We hypothesize that the verbs imparare and
studiare form a phasal pair of the kind exemplied in Section 1.2, and
that the telic-atelic distinction is not yet operative or it is outranked in
MKs competence until the last recording at least. We hypothesize that
MK might know the general meaning of both studiare and imparare without
(yet) knowing their actional content (that is, what the belonging to one or
another actional class entails in terms of (in)compatibility with temporal
expressions). We do not know to what extent frequencies of lemma and
of form-function associations in native input have played a role in MKs
lexical choices. We do not think that counting the occurrences of studiare
and imparare in written and spoken Italian would be of any signicance
because we cannot factorize the kind of input MK was exposed to. We
have good reasons only to assume that especially MKs teachers and also
his interviewer utilized many times studiare, imparare, and insegnare, and
that they distinguished these verbs and used them in a target-like manner.
The source of the non target-like uses of MK is not in native input, and
thus it must be searched in some developmental factors, among which
perhaps the tendency of treating on a par verbs with similar meaning and
of grouping together verbs belonging to the same event irrespectively of
their actional endowments.

2. Some results from the acquisition of tense and aspect in L2 Italian

In this section, we oer a review of longitudinal studies on Italian L2


and critically evaluate the implications of the Aspect Hypothesis, which
has represented a major issue in the literature on rst and second language
acquisition (Andersen 2002). Analyses of tense and aspect marking in
learner Italian were rst carried out paying attention to the acquisition of
Data analysis: The qualitative analysis of actionality in learner language 401

morphological distinctions. As it is known, in early learners languages,


morphology is absent, irrespective of whether the target language and/or
the source language are rich in morphology (Giacalone Ramat 2009:
261). Early analyses took into account both the form-to-function and
the function-to-form approaches (Bernini and Giacalone Ramat 1990,
Giacalone Ramat 2003).2 Investigations of the meaning associated with
early morphology led to discover the emergence of a rst temporal and
aspectual contrast between a generic, temporally unspecied, present-like
base form and a compound perfect form labeled passato prossimo in the
Italian tradition. The latter form mostly expresses a temporal/aspectual
meaning of pastness and perfectivity. In all the learners observed (Giacalone
Ramat 1992, 2003), the functional opposition between the perfective aspect
expressed by the passato prossimo and the imperfective aspect represented
by the imperfect was acquired later than the present-passato prossimo
opposition. Even later, the means were acquired for the morphological
expression of the future. These analyses were couched in functional terms,
looking for the means to express semantic functional categories of tense
and aspect. They were inuenced by the learner varieties approach, which
obtained relevant results on the acquisition of temporality in the languages
of Europe (Klein 1986; Klein and Perdue 1992).
The interest to investigate the role of lexical aspect originated in the
eld of rst and second language acquisition (starting from Antinucci and
Miller 1976). The formulation based on Andersens work, the so-called
Aspect Hypothesis (Andersen 1991, Shirai and Andersen 1995, Andersen
and Shirai 1996), was broadly supported by results concerning learner
Italian (Giacalone Ramat 1995, 2003). Lexical aspect or Aktionsart has
indeed turned out to be a crucial factor in the acquisition of tense and
aspect morphology, past tense and perfective aspect being mainly used
with telic verbs, imperfect with statives and activities. Some problems,
however, were sorted out, which demanded a revision of the model. In
particular, it was shown that durative verbs (activities) like studiare to
study, giocare to play, and lavorare to work are endowed with perfective
marking to encode events that have come to an end, and are not found in
the imperfect:

2. Detailed information on the corpus of learner Italian can be found in Andorno


and Bernini (2003). The data discussed in this section belong to low-level pro-
ciency uninstructed learners.
402 Anna Giacalone-Ramat and Stefano Rastelli

(6) Loro han studiato quello di commercio.


They have attended a commercial school (in China).
(S 4, L1 Chinese) (Giacalone Ramat 1995)
Further analyses on Italian learner languages (Giacalone Ramat 2003b)
conrmed that durative predicates emerged with perfective marking, con-
trary to Andersens predictions, whereas stative predicates like be, have,
and modals received an imperfect marking, somewhat later than perfective
marking. What can be suggested is that the Aspect Hypothesis is too strong,
although the use of verbal morphology is clearly inuenced by lexical aspect
(see the discussion in Bardovi-Harlig 2000).
A later study on Italian data (Giacalone Ramat 2002) especially aimed
at the comparison of the acquisition of passato prossimo and imperfect in
two groups of learners in order to gain insights on the role of the source
languages. The group of German learners used the passato prossimo with
all classes of verbs, showing many cases of overextension of the passato
prossimo in place of the imperfect, as in the following excerpt, where the
learner uses the forms e venuto has come and ha chiesto has asked to
refer to habitual, imperfective situations:

(7) ma non era possibile andare da sola in giro


but not be-IMP possible go-INF alone around
per sempre qualcuno e venuto
because always someone come-PASS-PROSS.3SG
e ha chiesto cosa fai oggi?
and ask-PASS-PROSS.3SG what are you doing today?
(AN, L1 German, 02)
When I was in Perugia it was not possible to walk around alone
because somebody would always come and ask what are you doing
today?
The use of the imperfect proved to be more dicult for German learners,
probably also because of the many uses this form has in Italian. In any
case, results from this comparative study provided evidence in support of
the claim that lexical aspect or actionality has an eect in the development
of acquisition, states and activities receiving imperfective aspect marking
in more advanced learners. The English learners also showed problems
with the imperfect and also with the progressive periphrasis, although
they are familiar with progressive from their L1, due to the dierent con-
Data analysis: The qualitative analysis of actionality in learner language 403

ditions of use in the two languages. Results from this study showed, on the
one hand, how the notions of tense and aspect exhibit considerable cross-
linguistic variation as far as codication strategies are concerned, and on
the other hand, they suggested that the L1 inuences in subtle ways the
acquisition of tense/aspect distinctions in the target language. Transfer
should anyway be conceived as a conceptual operation according to which
learners may assume that the target languages have the same conceptual
distinctions as their L1. Transfer is not a central issue for the Aspect
Hypothesis, since it is assumed that it is in principle universal (see however
Chapter 8). Results from L2 Italian conrm that the codication of tense
and aspect as was said before is far from being uniform in languages,
and thus learners have to undergo a reorganization of their L1 knowledge.
Several explanations for the observed eects of the Aspect Hypothesis
have been put forth. A cognitive and semantic principle was proposed by
Giacalone Ramat (1995) called the principle of selective association,
stating that features that are semantically congruent, such as telicity, per-
fectivity, and pastness, tend to be associated. Andersen (2002) suggested
a relevance principle and a congruence principle to account for the
anity of certain morphological markers with particular lexical aspectual
classes. These principles have in common the semantic cognitive dimen-
sion, and they are meant to guide the learners discovery of the form/
meaning relation encoded by the tense/aspect markers. A further relevant
factor to account for the learner acquisition path is the prototype eect.
In this view, Shirai and Andersen (1995: 758) stated: children acquire a
linguistic category starting with the prototype of the category, and later
expand its application to less prototypical cases. The same developmen-
tal path should hold true for L2 learners. The application of the notion of
prototype to acquisition is, however, in need of qualication: if learners
are able to infer a prototypical meaning for each inection, where does
this ability come from? Are prototypes in a learners mind as part of their
cognitive equipment (licensing an innatist view of prototypes), or are they
learned by children and adults from the available input, as functional
theories of language acquisition would suggest (Giacalone Ramat 2009:
265)? Descriptions of how learners expand prototypes from their proto-
typical contexts to new ones were provided by Andersen and Shirai (1996)
for the English progressive and by Giacalone Ramat (2009) for the present
tense in Italian. According to Dahl (1985: 78), the prototype of perfective
(which coincides with the prototype of past) associates telic predicates,
perfective aspect, and past tense; and this association, as showed above, can
404 Anna Giacalone-Ramat and Stefano Rastelli

account for preferred choices in early learner data. However, the emergence
of the perfective prototype in learner languages is usually blurred and
delayed because rst verbs, including both telic verbs and durative verbs,
appear in a present-like form also when a clear past meaning would be
expected:
(8) Da Asmara primo io vado Keren novanta
From Asmara rst I go-PRES.1stSG Keren ninety
duo kilometri.
two kilometres
I went from Asmara to Keren; it makes ninety two kilometres.
(MK, L1 Tigrinya, telling how he reached Italy)
(9) Interviewer: tu vivevi in campagna o in citta in Cina?
Were you living in the country or in a city in China?
Vivo in una grande citta Shanghai.
I-live-PRES-1SG in a big city Shanghai
I used to live in a big city, Shanghai.
(WE, L1 Chinese)

At any rate, it can be safely stated that the acquisition and use of proto-
typical past for telic verbs requires a certain amount of time (which varies
in individual learners), and that the spread of the past prototype to activities
and states takes even more time. This opens up the possibility that proto-
types are formed starting from the input, as suggested by Bertinetto and
Noccetti (2006) for L1 Italian. Thus, temporal reference, aspect, and action-
ality appear to be initially underspecied in learner productions: learners
struggle with the task of disentangling the various functions of verb inec-
tions and of assembling the appropriate features.
The role of the input available to learners accounts for the Distribu-
tional Bias Hypothesis proposed by Shirai and Andersen (1995) and
Andersen (2002) as an alternative interpretation to the cognitive pre-
disposition to recognize events with an endpoint. According to this prin-
ciple, the distribution of tense/aspect markers with particular classes of
verbs is already present in adult languages (or in native speakers for L2
learners), so that children and learners might simply imitate the input.
This does not seem to be the case, however, because learners appear to
gradually grasp the temporal aspectual and actional properties of the
forms they hear in the input, and initially they do not acquire all mean-
ings of verbal forms, but use them in a central or prototypical meaning.
Data analysis: The qualitative analysis of actionality in learner language 405

Andersens assumption (2002: 81) is that learners are cognitively predis-


posed to nd real realized unitary bounded events encoded in language and
thus recognize that meaning of the past or perfective form and not the
broader ranges of meaning that the form has in adult native speaker use
(emphasis in original). This would imply that the Aspect Hypothesis takes
for granted that learners know from scratch the actionality category,
namely if a certain verb is categorized as an achievement or an activity.
As shown in Section 1.3, empirical evidence from learners of Italian reveal
a dierent picture: beginning learners, especially, seem to use actionally
underspecied verbs. Following this idea, Giacalone Ramat and Rastelli
(2008) proposed a revision of the Aspect Hypothesis, focusing in particular
on the acquisition of actionality in early learner productions. In order to
analyze in more detail the development of early interlanguage, a new
research design was sketched. Whereas in previous research on L2 Italian
spontaneous learners were recorded on the basis of personal narratives
and lm retelling in which no referent control was possible and no
repeated task was carried out, new data were collected from instructed
learners using more controlled tasks. The qualitative analyses deriving
from this research design described in the following section were
meant to suggest new ways for the study of tense, aspect, and actionality.

3. The study

3.1. Subjects, task, and the satisfaction of qualitative analysis criteria


In this section, we apply the qualitative principles presented in Section 1.2
to learner Italian data in order to discuss how qualitative analyses contribute
to a better understanding of the L2 acquisition process. In the current study,
two dierent groups of Chinese beginner/low intermediate, tutored learners
of Italian were asked to recount orally the drawings of Mayers (1969)
Frog, where are you? (for further information on the study, see Biazzi and
Matteini 2010). Group 1 was originally formed by 11 students. Six of
them recounted the story twice (rst time and 5 months later), whereas
only 3 of them recounted the same story 3 times (rst time and 5 and 11
months later). Group 2 was formed by 12 students, who recounted the
story twice (rst time and 1 month later). Table 1 summarizes the break-
down of subjects and hours of instruction for each session:
406 Anna Giacalone-Ramat and Stefano Rastelli

Table 1. Number of subjects and amount of class instructions (hours) throughout


the re-telling sessions
L1 1st session 2nd session 2nd session 3rd session
1 month 5 months 11 months
Group 1 Chinese 11 (200) 6 (400) 3 (560)
Group 2 Chinese 12 (300) 12 (560)

(a) Requirement of referent-check and within-subject variability. Sentences


were selected for analysis only when there was an element in the picture
that clearly forced the choice towards one verb in the pair while excluding
the other. When referring to that element, the native speakers choice is
always mutually exclusive. For example, there is no doubt that a native
speaker of Italian would say looking at the boy from Mayers story
that he sta cercando is searching his frog and that non lha ancora
trovata he has not found it yet. Our longitudinal design should allow us
to see how the same learner expresses that single element in the picture
over time. Unfortunately, subjects in longitudinal analyses often drop out
of the study, or they may happen to focalize on a dierent element in the
same scene (e.g., not on the boy), and thus the number of subjects to be
scrutinized decreases accordingly.3
(b) The requirement of comparing L1 and L2 against the choice of verb pairs.
Firstly, verbs under scrutiny are the most frequently used by learners when
recounting the frog story. This ensures that instances of the same verb are
likely to be repeated over time, allowing researchers to observe changes.
Secondly, as far as the target language is concerned (Italian), the traits
that oppose the members of each pair are carefully checked. Guardare to
look and vedere to see in Italian are two transitive verbs. There exist
dierent semantic parameters according to which transitivity may vary
(Lazard 2002; Hopper and Thompson 1980), for example, in the number
of participants, in [edynamicity], [econtrol], and [eagentivity]. Guardare

3. This condition stresses again the dierences between quantitative and qualita-
tive analyses already sketched in Section 1.2. In no way our subject can be
considered as a sample of the population of L2 learners of Italian, and in no
way will we generalize the outcome of our qualitative analysis. Rather, our
aim is to move research away from anecdotal in order to narrow the range of
hypotheses to the most reasonable ones and, nally, to prepare the eld for
always more interlanguage-aware quantitative generalizations.
Data analysis: The qualitative analysis of actionality in learner language 407

and vedere dier in many respects, mainly for [eagentivity], secondly for
the properties of the NP object that is being seen or looked at, and nally
for their subcategorization frame. Generally speaking, vedere to see is an
event perception meaning something that appears in the visual spectrum,
whereas guardare to look is an agentive, controlled event of perception,
as it is clearly demonstrated by the mutually exclusive compatibility of these
two verbs with the adverbials di proposito on purpose and per caso by
chance:
(10) Ti ho guardato di proposito.
I looked at you on purpose.
(11) *Ti ho visto di proposito.
I saw you on purpose.
(12) *Ieri per caso ho guardato una rapina.
Yesterday by chance I looked at a robbery.
(13) Ieri per caso ho visto una rapina.
Yesterday by chance I saw a robbery.
The whole picture is much more complex though, because the [eagentiv-
ity] feature is not the only one to come into play. Agentivity may correlate
with the semantic and grammatical characteristics of the direct object NP.
For instance, the lack of duration of the event represented by the object
lampo lightning selects vedere and excludes guardare, but some objects
that refer to durative events like la partita in TV TV match allow
both verbs, irrespectively of their [eagentive] features. Yet, the feature
of specied quantity triggers the acceptability of lampi lightning as the
possible object of guardare (with an iterative reading):
(14) Ho guardato i lampi.
I looked at the lightening. (pl.)
The phrasal context and the aspectual perspective on the event conveyed
by the verb add some other restrictions too. When used with perfective
tenses expressing the experiential past, vedere is largely preferred:
(15) Hai mai visto una giornata cos bella?
Have you ever seen such a ne day?
(16) ??Haimai guardato una giornata cos bella?
Have you ever looked at such a ne day?
408 Anna Giacalone-Ramat and Stefano Rastelli

From a syntactic point of view, vedere but not guardare allows the
complementizer che that to ll the head position of a CP:
(17) Ho visto che Maria e arrivata.
I saw that Maria has arrived.
(18) *Ho guardato che Maria e arrivata.
*I looked that Maria has arrived.
Phasal pair verbs cercare to search and trovare to nd are both transi-
tive too. The former is an activity, atelic verb (as such, it is incompatible
with in x-time expressions), whereas the latter is telic, even though it is
fully compatible either with the progressive (19) or with for x-time expres-
sions (20):
(19) Sta trovando molte dicolta.
He/she is encountering-PROGR many diculties.
(20) Ho trovato tracce di animali per molti giorni.
I found animal footprints for many days.
When selecting actional and phasal pairs in learner data, not only the
actional template of verbs, but also their relationship (and the nature)
with direct object NPs is important. For example, a dierence when com-
paring an actional pair (such as guardare and vedere) to a phasal pair
(such as cercare and trovare) is that only the latter allows all kinds of NP
objects. In theory, anything that can be searched for can also be found,
whereas this does not necessarily hold for actional pairs (something that
can be looked at might not also be seen). In choosing verbs as candidates
for qualitative analysis, it is important that the direct object NPs are
neither over-preferred (collocated), nor outranked for one verb or another
in the pair. In this way, external factors for verb selection are minimized.
In the Frog story, the NP rana frog is compatible as the NP object of
either verb in the pair.
As far as learners L1 is concerned, verbs that enter phasal pairs to
search/to nd and actional pairs to look/to see in Chinese share an
identical morpheme but dier for the presence or absence of a resultative
complement (Li and Thompson 1981). To search corresponds to zhao,
whereas to nd corresponds to zhaodao, which is formed from zhao
plus the resultative dao (which literally means to reach, to arrive). The
presence of dao (resultative complement) conveys the meaning that the
process of searching eventually turned into nding something. Similarly,
Data analysis: The qualitative analysis of actionality in learner language 409

to look can be translated with kan, and to see with kanjian, where jian
(which also means to see) resolves the (potentially) endless process into
an achievement (the same resultative complement jian can also be used
with other verbs of apperception such as wen to smell and tng # to listen
to). We wonder whether these L1 features are likely to be transferred to
the target language and whether they can interfere with the lexical choice
of either verb in the pairs. Three facts are relevant to us in the following
cases: (a) in the learners L1 the verbs in each pair dier in their meaning;
(b) the dierence concerns aspect; and (c) in Chinese, the dierence is not
lexical, but rather it is overtly marked by a morpheme. The overall conclu-
sion is that nothing, both in the L1 and in the L2, represents an obstacle
or delaying factor for either verb in the pairs to be learned in the target
language.

(c) The requirement of scope-widening. If actional membership is assigned


on the basis of lexical properties of the verb, then it is very hard to dene
it in a rst language, let alone in a second language, where actional tests
(for instance, those listed by van Valin 2005: 3442) are of little or of no
help (i.e., misleading). For instance, Giacalone and Rastelli (2008) pro-
vided examples of learner sentences where telic verbs are compatible with
expression of duration (see Section 1.1). Actional categories are highly elu-
sive and the judgement of a verb belonging to a class or another is far
from being clear-cut even for native speakers. As far as L1 Italian is
concerned, the actional shift (or actional hybridism as it is called in
Bertinetto 1986) seems to be the norm rather than the exception. Under
the appropriate conditions, almost every Italian verb may be regarded as
belonging to two, sometimes three and even four dierent actional classes.
For instance, among 3,429 verbal tokens of 33 verb types extracted from
the TreSSI corpus (a corpus of written modern Italian), the cases of
unique actional assignment amount to about 20% (Lenci and Zarcone
2009). Our hypothesis is that learners may nd it easier to express the
actional content of verbs at a sentence level rather than lexically, and
that the actional value of an L2 predicate seems to be a matter of com-
positional induction rather than lexical deduction. In other words, since
we do not credit initial learners with the capacity to derive the actional
content from the meaning of the verb, we nd it more likely that they
express the actional content of verbs by using everything at hand in the
sentence (preposition, adverbs, adjuncts). Consequently, we took into account
the whole sentence (i.e., not just the verb alone).
410 Anna Giacalone-Ramat and Stefano Rastelli

3.2. The data


3.2.1. Guardare to look and vedere to see
In the description of one of the main scenes of the story (i.e., the boy and
his dog hear a noise, look behind a hollow trunk lying in a pond, and
nally see the frog together with its family), learner LUN uttered sentence
(21) the rst time he narrated the story, and sentences (22) and (23) after
5 and 11 months respectively when the story was narrated again. In the
rst narration (sentence 21), the sequence of events expressed by looking
and seeing merge into a single underspecied guardare to look. In the
second narrative, represented in sentence (22), the rst element in the
sequence is skipped, but the second element, vedere to see, is used in a
target-like manner. Only in sentence (23) is guardare to look used in a
way that seems to depict the verbs semantic [agentive] and actional
[duration], [telic] traits:
(21) Guardano e rana e il tua mamma e insieme.
(They) look (and) it is the frog and your (its) Mom is together.
(22) (Ragazzo) ha visto rana con tua madre.
(The boy) has seen the frog with your (its) Mom.
(23) Guardato dietro albero ma e molto felice perche lui trovato tuo
animale prediletto.
Looked behind a tree but he is very happy because he found his
favorite pet.
While searching for his frog, the boy asks many animals if they have seen
it by chance. In sentences (24), (25), and (26), our interest focalizes on the
passage between verb omission (24), guardare (25), and nally vedere (26):
(24) Chiede la talpa lo so la mia amica rana?
He asks the mole: Do you know where my friend is?
(25) Ragazzo chiede gufo sei guardato la mia rana?
The boy asks the owl: Have you looked at my frog?
(26) (Il ragazzo) chiede talpa: hai visto mia animale?
The boy asks the mole: Have you seen my animal?
The opposite process (from vedere to guardare) is what possibly drives
informant SOF to end up merging the two predicates into a single lexical
entry (28):
Data analysis: The qualitative analysis of actionality in learner language 411

(27) Percio il bambino e il cane vedere altri.


So the child and the dog see other (frogs).
(28) Loro hanno guardato il ramo e loro guardato questa rana.
They have looked at the branch (trunk) and they have looked
(seen) this frog.
SOF strains in a similar way the verb guardare over vedere:
(29) Questo bambino e guardato il ramo ma non guardato la rana.
This child looked at the branch-tree but did not look (see) at
the frog.
Learners from Group 2 were not any dierent with regards to the unex-
pected, non target-like choices of verbs. For instance, two months after
having oscillated between guardare and vedere, participant IVA eventually
chose the latter:
(30) Vada dietro un albero e vede e guarda due rane.
(He) goes behind a tree and (he) sees and looks at two frogs.
(31) Dietro lalbero lui e il cane hanno visto due rane.
Behind the tree he and his dog saw two frogs.
The opposite path seems to be the one followed by subject LUC when
retelling the same event 2 months after the rst narrative:
(32) Loro guardano dietro albero e vedono la rana.
They look behind the tree and see the frog.
(33) Dietro lalbero loro hanno guardato e hanno guardato la rane.
Behind the tree they have looked at and they have looked at
the frogs.
When retelling another scene (in which the boy and the dog see a swarm
of bees), SOF just 2 months after substituted guardare with the target-
like vedere:

(34) Il cane e molto felice per guardare molto api.


The dog is very happy to see many bees.
(35) Allinizio loro hanno visto li api.
At the beginning they saw bees.

When LIL recounted the scene in which the boy sees that the bowl in
which the frog was kept is empty, he rst used the target-like vedere, but
2 months after, he seemed to backslide to the non target-like guardare:
412 Anna Giacalone-Ramat and Stefano Rastelli

(36) Al mattino Luigi non vede la rana nella botilia.


In the morning Luigi [a fantasy name] does not see the frog in
the bottle.
(37) La mattina di domani il ragazzo con il cane guardano la rana non ce.
The morning after the boy and the dog look (that) the frog there
is no more.
To sum up, when looking at these utterances, one is driven to think that
these learners miss the actional distinction between guardare and vedere
and that their preference for either verb does not seem to follow a clear
developmental path.

3.2.2. Cercare to search and trovare to nd


Cercare and trovare occur more frequently than guardare and vedere
because this is exactly what the Frog story is all about: a frog that is lost,
looked for and eventually found in a pond with its family of frogs. In our
data, almost every learner in every group exchanges, overstrains, or totally
blurs the dierences between these two verbs on many occasions, and
inconsistently, in the same paragraph or even in the same clause. For
instance, in all three occasions in which (in a time span of almost a year)
LUN recounted the scene in which the boy searches for his frog every-
where in his bedroom, she exclusively used the verb trovare to nd:
(38) Trova scarpe letto molto posta ma non trova la rana.
She nds (searches in) the shoes the bed many places but he does
not nd the frog.
(39) Ragazzo con il tuo cane trovare tutto la camera ma non ha trovato.
The boy with the dog nd (search in) everything in the room but
he has not found.
(40) Enrico trovato la stivale, bicchiere tutte le camera, ma non ce piu.
Enrico found (searched in) the boot, the glass, everything in the
room, but there is no more.
The same scene was recounted exactly in the same way (after 5 months
from the rst time) by ORL:
(41) Poi il bambino trova la rana nella scarpa.
Then the boy nds (searches) the frog in the shoe.
(42) Poi il bambino trovare sotto la letto e nella scarpa.
Then the child nds (searches) under the bed and in the shoe.
Data analysis: The qualitative analysis of actionality in learner language 413

There are cases in which the target-like form is only apparently acquired
over time. JAC used trovare, but 5 months later, after a brief wavering, he
eventually seemed to know which verb to choose in the pair:
(43) Loro usciano da casa [. . .] per trovare il rana.
They exit the house to nd (search for) the frog.
(44) Il bambino con il suo amico escono a casa per trova per cerca- per
cercare la rana.
The boy and his dog exit the house to search for the frog.
This distinction was only apparently attained, because a few seconds
later JAC while describing the search seemed to lapse and says:
(45) Il bambino trova il rana sul albero.
The child nds (searches) the frog on the tree.
Inconsistency of choices was frequent within and across subjects. LUI
lapsed from cercare to trovare when referring to the boy searching for the
frog in his boot while his dog searches inside a glass bowl:
(46) Poi lui cerca i scherpi e il cane cerca il bicchiere.
Then he searches (in his) shoes and the dog searches (in the) glass.
(47) Il bambino cerca il suo calve [. . .] il cane trova nella botttiglia.
The boy searches (in his) ?shoes [. . .] the dog nds (searches in)
the bottle.
As in the case of guardare and vedere, these sentences show that our learners
not only blur the distinction between cercare to look for and trovare to
nd, but the examples also reveal that the speakers choices are inconsis-
tent over the whole time span of recordings.

4. Discussion

Three facts emerged from the qualitative analysis of our data. First, some
frequent and basic verbs (as Viberg 2002 calls them) are used instead of
other actionally dierent verbs. As far as the verbs under scrutiny are con-
cerned, the verbs guardare to look and vedere to see despite being ac-
tionally dierent are exchanged very often by learners. A further nding
is that learners lexical choices seem to be unpredictable, in that they do
not seem to follow a consistent developmental path (e.g., from atelic to
telic). Second, verbs that enter phasal pairs are often used indierently
414 Anna Giacalone-Ramat and Stefano Rastelli

and inconsistently over time. In our data, trovare to nd and cercare to


look for are exchanged over the whole time-span of recordings. As in the
case of basic verbs, the random choice of either verb does not seem to
reduce over time. Third, frequency in the input and strength/exclusiveness
of prototypical associations do not ensure that learners automatically
know the properties of actionality, as the case of MKs uses of the very
frequent verbs imparare to learn and studiare to study has possibly
shown. Our working hypotheses are that facts (1) and (2) reviewed above
may entail that actionality and general meaning of L2 verbs can dissociate
and be acquired at dierent times, and that fact (3) may entail that
something more than statistical learning (rather, a qualitative change in a
learners grammar) is expected to take place in a learners actional com-
petence. In Section 4.1 we ask ourselves whether the phenomenon of
actional underspecication of L2 verbs is observable also in non-Chinese
learners of Italian, whereas in Section 4.2 we discuss why actional tests
fail to shed light on learners semantic representation based on the idea
tha L2 actional competence is defective.

4.1. Cross-linguistic evidence of actional underspecication


In Section 3.1, it was questioned whether the learners L1 (Chinese) may be
the reason for the exchange in actional pair of the basic verbs guardare/
vedere and in the phasal pair cercare/trovare. Actually, the same phenomena
are common to learners with dierent L1s, as reported in other surveys of
Italian learner corpora (see Giacalone and Rastelli 2008; Rastelli 2008,
2009). To give some examples, in sentence (48), the beginning learner MK
(see Section 1.3) is talking about a friend who moved from Italy to England.
MK thinks that he will regret this now that he can see how life is over there:
(48) Adesso lui guarda come e la vita di Inghilterra.
Now he looks (sees) how is life in England.
Another Tigrinya learner, AB, uses vedere to see instead of guardare to
look up a dictionary:
(49) Vedo dizionario.
(I) see (look up at) (the) dictionary.
In a survey of written retellings of American undergraduate students of
Italian, Rastelli (2006) found many instances of overextension of trovare.
In sentence (50), the English-speaking learner NIK is recounting a lm
scene where a woman was left alone in an auto grill (restaurant) by her
trip companions, and the coach did not go back to pick her up:
Data analysis: The qualitative analysis of actionality in learner language 415

(50) Lautobus non puo ritornare a trovarla in quel momento.


The coach cannot go back to nd (search for) her in that momento.
MK uses trovare instead of cercare also in periphrases such as cercare di
try to.
(51) Trovero per continuare la scuola.
(I) will nd (try) to continue school.
Similar phenomena are found in L2 Italian independently from the learners
L1. Whether and to what extent these phenomena correlate with learners
prociency is outside the scope of the present study.

4.2. Actional tests and the L2 defective actional competence


In Section 1.1, we reported recent studies that stressed the need for tapping
more directly into the actional competence in the interlanguage. In light of
our ndings, we claim that unlike Shirai (Chapter 8) actional tests
designed for L1s do not represent a reliable procedure to achieve the goal
of coding aspectual representations. In fact, these actional tests are at risk
of committing the comparative fallacy in that they do not take into
account the fact that the actional competence of L2 learners is defective
in at least three ways:
(1) The temporal frame is uncertain. Time expressions (adverbials such
as quando when or in/per unora in/for one hour) are often used by
learners in an underspecied way, so that the relationship between ac-
tional membership and the outcome of actional tests is established on
slippery grounds. For instance, sentence (52) was uttered by Chinese
learner GIO and refers to the scene when, after waking up, the boy realizes
the frog is gone from the glass jar where it was kept by the boy (who has
been given the name of Mario):

(52) Quando Mario ha dormito questa rana ha corso.


When Mario has slept this frog has run.

Literal interpretation of (52) is not allowed in native Italian when describ-


ing that scene, because quando when in this sentence is punctual, and, as
such, it indicates the point in time when the process of sleeping is over and
not the time-span of sleeping during which the frog escaped. In sentence
(53), the Chinese learner MIN utters a sentence where the opposite phe-
nomenon seems to occur:
416 Anna Giacalone-Ramat and Stefano Rastelli

(53) Un giorno mentre si sono addormentati la la rana e saltata fuori


dalla bottiglia.
One day, while they were falling asleep, the frog jumped out of the
bottle.

In (53) the [telic] and punctual character of the achievement verb addor-
mentarsi to fall asleep conicts with the conjunction mentre while,
unless we admit that one of the two (or both) is underspecied in a learners
interlanguage.

(2) Learners may blur the terminative-bounded opposition and use gram-
matical encoding to express telicity structurally (Rastelli 2008, in press).
Bertinetto and Deltto (2000: 191), comparing dierent mature languages,
proposed to use the pair terminative-non terminative to refer to the
aspectual domain and the pair bounded-non bounded to refer to the
actional domain. They stressed that the presence of an endpoint (or its
atelicity, duration, etc.) in a mental representation of that predicate sur-
vives the fact that the speakers may present the event as concluded or
in course. This distinction is meant to set up a basis for interlinguistic
comparisons among languages that overtly mark verb actionality and
aspect versus languages that do not. Italian overtly codes the terminative-
non-terminative distinction in the past, but it does not code overtly the
bounded-unbounded distinction in predicates (actionality is a matter of
lexical distinction). We think that learners should not be credited with
native representational awareness about a distinction between aspect and
actionality, and that the separation between aspect and actionality is a
rather weak descriptive tool. Rastelli (2008, in press) advances the idea
that learners might use the temporal/aspectual [Perfective] morpheme
to add telicity to atelic verbs (thus blurring the distinction between action-
ality and aspect) and that they upgrade adjuncts (typically, AdvP and PP)
to the rank of arguments to gear the actionality of underspecied verbs to
a telic interpretation. This happens as long as L2 Italian learners do not
check [ePerfective] features in the functional head T(ense) rather than in
Asp (which is either their L1s or default parameter value).
(3) Learners may not be aware of the interactions between lexical and
grammatical aspect. Learners should not be credited to know from scratch
in which circumstances and for which verbs the perfective-imperfective
alternations alone may cause the actional content of the same verb to
Data analysis: The qualitative analysis of actionality in learner language 417

shift.4 If researchers take for granted that L2 learners are aware of this
shift, this would be enough to make some actionality tests fail and, thus
mislead researchers. For instance, in sentence (54), despite the use of the
perfective passato prossimo, the referent-check on the lm scene convinces
us that the English-speaking student wants to say that the woman was
wearing green sunglasses (stative) and not that she wore sunglasses
(achievement).
(54) Ha indossato occhi di sole verdi.
She wore (was wearing) green sunglasses.
In native Italian, the verb indossare to wear can be telic or atelic depend-
ing on the aspectual morphemes (perfective or imperfective). In sentence
(54), the specic use of the verb indossare is telic if we consider its (per-
fective) form, but it is atelic if we consider the supposed meaning (what
the learner probably meant when trying to describe the lm scene). This
form-meaning mismatching leads us to believe that another piece of
actional competence is likely to be missing in a learners interlanguage.

5. Conclusion

At the beginning of this section we expressed the idea that actionality too
is learned. Now, we can dene this idea more precisely. There are converg-
ing cues that there could exist a developmental stage in which learners
actional competence is in re-construction alongside with the whole tense
and aspect system. At this stage, basic verbs (similar in meaning) are easily
exchanged, phasal verbs are easily exchanged, actional distinctions are
blurred, and nally the interactions between lexical and grammatical
aspects are largely ignored. Actionality is learned, as long as the actional
properties are acquired gradually by learners. Qualitative analysis should
warn researchers against carelessly labeling telic or atelic (durative or non-
durative) all verbs in the interlanguage based on accuracy percentages in
mandatory contexts. The procedure of counting up target-like and non
target-like occurrences (in mandatory contexts) of prospective telic and
atelic verbs should not be used as evidence of developmental primacy of
prototypical associations. In fact, as long as we do not know whether a

4. Bertinetto (1986: 109) quotes an example where an aspectual shift between


perfective and the imperfective alone causes the actional shift: Luca mi volto
le spalle Lucca turned his back on me (achievement) Luca mi voltava le spalle
Lucca turned/was turning his back to me (stative).
418 Anna Giacalone-Ramat and Stefano Rastelli

certain verb is telic or atelic in a learners mental representation, we


cannot conclude that a prospective (what appears to be) telic verb is
more likely to attract prospective perfective morphemes. In this respect,
we do not believe that a careful denition of the procedures used in
actional tests (Chapter 8) is enough to prevent us from committing the
comparative fallacy. In fact, these procedures are based on how we expect
that things should work in the interlanguage, assuming that temporal
frames (adverbs, prepositions, and adjuncts) and verb morphemes are in
place and do their part of the job exactly the way we expect them to do,
which is not always the case.
Qualitative research stresses the need to nd alternative research methods
that can tap more directly onto the learners semantic representations. The
necessity of investigating the developing semantic competence in a second
language is addressed for instance in Slabakova (2006). Qualitative analysis
of elicited narrative and naturalistic data alone are inconclusive with respect
to discovering the nature of learners representations. Nevertheless, a quali-
tative analysis can identify some possible indicators of the development of
a learners actional competence, which can be tested further with experi-
mental methods. These methods could include psycholinguistic and neuro-
imaging studies, alongside with behavioral data. Adapting the fruitful idea
of mixed methods research suggested by Dornyei (2007), we suggest that a
qualitative analysis could provide a preliminary step for experimental
studies (see also Chapter 12). For instance, once basic verbs and phasal
pairs are identied in a learners interlanguage by means of qualitative
longitudinal analysis, they could be used to test whether and when learners
develop intuitions for actional distinctions going beyond the general mean-
ing of verbs. Phasal and actional pairs could be used as stimuli in MRI,
PET, ERP and eye-tracking studies in order to be manipulated for the
[etelic] or [edurative] conditions. In conclusion, we suggest that future
research on L2 actionality will benet from these methods more than
from quantitative studies in which an impressive number of verb occur-
rences are labeled for actional membership not according to learners repre-
sentations, but perhaps to researchers L1 biased intuitions.

Acknowledgments

This work is the result of a continuous exchange of ideas between the two
authors. However, Stefano Rastelli is responsible for Sections 1, 3, and 4,
and Anna Giacalone Ramat is responsible for Sections 2 and 5.
Data analysis: The qualitative analysis of actionality in learner language 419

References

Andersen, Roger
2002 The dimension of Pastness. In M. R. Salaberry & Y. Shirai,
(Eds.), The L2 acquisition of tense-aspect morphology, 79105.
Amsterdam/Philadelphia: John Benjamins.
Andersen, Roger & Yasuhiro Shirai
1994 The primacy of aspect in rst and second language acquisition.
In W. Ritchie & T. J. Bhatia (Eds.), Handbook of second lan-
guage acquisition, 527570. San Diego: Academic Press.
Andersen, Roger & Yasuhiro Shirai
1996 The primacy of aspect in rst and second language acquisition:
The pidgin/creole connection. In W. C. Ritchie & T. K. Bhatia
(Eds.), Handbook of second language acquisition, 527570. New
York: Academic Press.
Andorno, Cecilia & Giuliano Bernini
2003 Premesse teoriche e metodologiche. In A. Giacalone Ramat (Ed.),
Verso litaliano, 2736. Rome: Carocci.
Bardovi-Harlig, Kathleen
2000 Tense and aspect in second language acquisition: Form, meaning,
and use. Oxford: Blackwell.
Bertinetto, Pier Marco
1986 Tempo, aspetto e azione nel verbo italiano: Il sistema dellindica-
tivo. Florence: Accademia della Crusca.
Biazzi, Michela & Isabella Matteini
2010 Referential and topic movement in Chinese learners of Italian:
A longitudinal account. In M. Chini (Ed.), Topic, struttura
dellinformazione e acquisizione linguistica, 137157. Milano:
FrancoAngeli.
Bley-Vroman, Robert
1983 The comparative fallacy in interlanguage studies: The case of
systematicity. Language Learning 33. 117.
Cruse, Alan
1997 Lexical semantics. Oxford: Oxford University Press.
Dahl, Osten
1985 Tense and aspect systems. Oxford, Blackwell.
Dornyei, Zoltan
2007 Research methods in applied linguistics. Oxford: Oxford Univer-
sity Press.
Ellis, Rod & Gary Barkhuizen
2005 Analysing learner language. Oxford: Oxford University Press.
Fellbaum, Christiane
1998 Wordnet: An electronic database. Boston: The MIT Press.
Gass, Susan & Larry Selinker
2008 Second language acquisition: An introductory course. New York:
Erlbaum.
420 Anna Giacalone-Ramat and Stefano Rastelli

Giacalone Ramat, Anna


1990 Presentazione del progetto di Pavia sullacquisizione di lingue
seconde. Lo sviluppo di strutture temporali. In G. Bernini & A.
Giacalone Ramat (Eds.), La temporalita nellacquisizione di
lingue seconde, 1338. Milano: FrancoAngeli.
Giacalone Ramat, Anna
1992 Grammaticalization processes in the area of temporal and modal
relations. Studies in Second Language Acquisition (11)5. 297
322.
Giacalone Ramat, Anna
1995 Tense and aspect in learner Italian. In P. M. Bertinetto, V. Bianchi,
O. Dahl & Squartini (Eds.), Temporal reference, aspect and action-
ality, vol. II, Typological perspectives, 289307. Torino: Rosenberg
and Sellier.
Giacalone Ramat, Anna
2002 How do learners acquire the classical three categories of tempo-
rality? In M. R. Salaberry & Y. Shirai (Eds.), Tense-aspect mor-
phology in L2 acquisition, 221248. Amsterdam: John Benjamins.
Giacalone Ramat, Anna
2003a Typology and second language acquisition. New York/Berlin:
Mouton de Gruyter.
Giacalone Ramat, Anna
2003b Verso litaliano. Percorsi e strategie di acquisizione. Rome:
Carocci.
Giacalone Ramat, Anna
2009 Typological Universals and Second language Acquisition. In S.
Scalise, E. Magni & A. Bisetto (Eds.), Universals of language
today, 253272. Rotterdam: Springer.
Giacalone Ramat, Anna & Stefano Rastelli
2008 Learning actionality: An investigation on data of L2 Italian. In
B. Ahrenholz, M. Rost-Roth & R. Skiba (Eds.), Empirische
Forschung und Theoriebildung. Beitrage aus der Soziolinguistik,
Gesprochene-Sprache-Forschung und Zweitspracherwerbsforschung.
Eine Festschrift fur Norbert Dittmar zum 65. Geburtstag, 239250.
Frankfurt: Peter Lang.
Hopper, Paul & Sandra Thompson
1980 Transitivity in grammar and discourse. Language 56. 251299.
Klein, Wolfgang
1986 Second language acquisition. Cambridge: Cambridge University
Press.
Klein, Wolfgang & Clive Perdue
1992a Utterance structure. In C. Perdue (Ed.), Adult language acquisi-
tion: cross-linguistic perspectives. Vol. 2: The results, 340. Cam-
bridge: Cambridge University Press.
Data analysis: The qualitative analysis of actionality in learner language 421

Klein, Wolfgang & Clive Perdue


1992b Utterance structure. Developing grammars again. Amsterdam/
Philadelphia: John Benjamins.
Lardiere, Donna
1998 Case and Tense in the Fossilized steady state. Second Language
Research 10. 2548.
Lardiere, Donna
2003 The comparative fallacy revisited: A reply to Lakshmanan and
Selinker (2001). Second Language Research 19. 129143.
Lakshmanan Usha & Larry Selinker
2001 Analysing interlanguage: How do we know what learners know?
Second Language Research 17. 393420.
Lazard, Gilbert
2002 Transitivity Revised as an example of a more strict approach in
typological research. Folia Linguistica XXXVI/34. 141190.
Lenci, Alessandro & Alessandra Zarcone
2009 Un modello stocastico della classicazione azionale. In Giacomo
Ferrari, Ruben Benatti & Monica Mosca (Eds.) Linguistica e
modelli tecnologici di ricerca. Proceedings of the XL international
congress of the Societa di Linguistica Italiana, 125148. Rome:
Bulzoni.
Li, Charles & Sandra A. Thompson
1981 Mandarin Chinese: A functional reference grammar. San Diego:
University of California Press.
Moens, Marc
1987 Tense, aspect and temporal reference. Edinburgh, Scotland: Uni-
versity of Edinburgh, Ph.D. dissertation.
Rastelli, Stefano
2006 ISA 0.9 Written Italian of Americans: Syntactic and semantic
tagging of verbs in a learner corpus, Studi Italiani di Linguistica
Teorica e Applicata (SILTA). 73100.
Rastelli, Stefano
2008 A compositional account of L2 verb Actionality and the Aspect
Hypothesis. Lingue e Linguaggio 7. 261289.
Rastelli, Stefano
2009 Lexical Aspect too is learned: Data from Italian learner corpora.
In A. Saxena & A. Viberg (Eds.), Multilingualism, Serie Linguis-
tica Upsaliensia 8, 272281. Uppsala.
Rastelli, Stefano & Mirta Vernice
In press The building blocks of telicity in L2 Italian. International Review
of Applied Linguistics and Language Teaching.
Salaberry, M. Rafael & Yasuhiro Shirai
2002 Tense-aspect morphology in L2 acquisition. Amsterdam/Philadel-
phia: John Benjamins.
422 Anna Giacalone-Ramat and Stefano Rastelli

Shirai, Yasuhiro
2007 The aspect hypothesis, the comparative fallacy, and the validity
of obligatory context analysis: A reply to Lardiere (2003).
Second Language Research 23. 5164.
Shirai, Yasuhiro & Roger Andersen
1995 The acquisition of tense-aspect morphology: A prototype account,
Language 71. 743762.
Slabakova, Roumyana
2006 Is there a critical period for semantics? Second Language Research
22. 302338.
Starren, Marianne
2001 The second time: The acquisition of temporality in Dutch and
French as a second language. Utrecht: LOT.
van Valin, Robert
2005 Exploring the syntax-semantics interface. Cambridge: Cambridge
University Press.
Verkuyl, Hans
2005 Aspectual composition: surveying the ingredients. In H. Verkuyl,
H. De Swart & A. van Hout (Eds.), Perspectives on Aspect, 19
39. Dordrecht: Springer.
Viberg, Ake
2002 Basic verbs in lexical progression and regression. In P. Burmeis-
ter, T. Piske & A. Rohde (Eds.), An integrated view of language
development, papers in honor of Henning Wode, 109134. Trier:
Wissenschaftlicher Verlag.
Wul, Stefanie, Nick Ellis, Ute Romer, Kathleen Bardovi-Harlig & Chelsea
LeBlanc
2009 The acquisition of tense-aspect: Converging evidence from corpora,
cognition, and learner constructions. Modern Language Journal
93. 354369.
Chapter 12
Integrating the analyses of tense and aspect across
research and methodological frameworks

M. Rafael Salaberry, Llorenc Comajoan, and


Paz Gonzalez

1. Introduction

In the last decade, the study of the acquisition of tense and aspect among
second language (L2) learners has become a signicant area of research (e.g.,
Ayoun and Salaberry 2005; Bardovi-Harlig 2000; Rocca 2007; Salaberry
2008; Salaberry and Shirai 2002; Slabakova 2001). Some researchers have
gone as far as arguing that there are too many studies on the topic: an
embarrassment of riches as proposed by Slabakova (2002: 186). Despite
the abundance of studies on tense-aspect, there are two recurrent problems
in the current literature: (a) the lack of an open dialogue across theoretical
frameworks (cf., Shirai 2007: 61), and (b) the lack of consistent method-
ological procedures (even when acknowledging disparate theoretical per-
spectives). That is, there has not been a systematic integration of ndings
across theoretical frameworks and methodological procedures. Overall,
such lack of integration leads researchers to base their analyses on only a
limited number of studies, among the many that are available for scrutiny.
Consequently, even if there is an abundant literature on the topic of tense
and aspect, we may not be using all relevant information to gather our
conclusions, thus leading to an inecient analysis or, even worse, an
incomplete one.
For instance, while some theoretical positions consider the role of
adjuncts (in particular, adverbials) as part of a denition of grammatical
aspect (e.g., Salaberry 2008), others regard the role of adjuncts to be
outside of the scope of grammatical knowledge and, more properly, to be
part of world knowledge (e.g., Slabakova and Montrul 2007). Whereas, in
principle, both denitions of the given theoretical construct are valid, the
analysis of ndings from both perspectives can be protably reconciled at
a supra-level in order to obtain an overall interpretation of tense-aspect
phenomena across a wide range of theoretical descriptions. At a minimum,
424 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

researchers will be led to reconcile the ndings from studies carried out
within distinct theoretical paradigms (see Chapters 5 and 6). A similar
situation arises with the analysis of research methodologies. For instance,
starting with the rst studies that investigated the acquisition of tense-
aspect, there have been discrepancies about the best procedure to code
the data (e.g., coding for lexical aspect, discourse grounding, etc.) and,
more importantly, how to interpret the ndings produced by such diverse
types of coding criteria (see Chapters 8 and 9). Along the same lines, some
data collection procedures are based on grammaticality judgments, whereas
other studies are based on the analysis of data obtained from open-ended
narrative tasks (see Chapters 7, 10, and 11).
Whereas some studies have already provided an integrative view of the
state of the art theoretical approaches to the study of the L2 acquisition of
tense and aspect phenomena (e.g., Ayoun and Salaberry 2005; Bardovi-
Harlig, 2000; Montrul and Salaberry 2003), such proposals have been
limited in scope, mostly restricted to summaries of studies. On the other
hand, to the best of our knowledge, the issue of research methodologies
applied to the analysis of the acquisition of tense-aspect has been restricted
to brief analyses presented as part of the background to specic studies.
The integration of ndings across studies is potentially important for
the overall objective of drafting a comprehensive picture of the acquisi-
tion of tense-aspect phenomena. Furthermore, the integration of analyses
carried out across methodological paradigms may lead to a more ecient
process of theorization. Given the previous analysis, the present volume
has addressed the following questions:

(a) What are the most relevant theoretical constructs (e.g., iterativity versus
habituality in Slabakova and Montrul 2007; cognitive grammar in
Putz, Niemeier, and Dirven 2001) that can be integrated into a pan-
theoretical perspective that would make research on tense-aspect more
ecient and, more importantly, advance our theoretical understanding?
(b) What are the various research methodologies (e.g., production versus
comprehension tests, quantitative versus qualitative approaches) that
could be used the purpose of integrating the various ndings into a
single coherent whole?
(c) What are the results from dierent theoretical frameworks and meth-
odologies that could provide evidence for one or many theories of the
acquisition of L2 tense and aspect?
Integrating 425

In this nal integrative chapter, we will address the previous questions in


the context of the arguments presented in this volume in an eort to (a)
provide a comprehensive and up-to-date review of the possible answers
to the questions above, and (b) expand those analyses and consider the
perspective of mixed methods (Tashakkori and Charles 2003) and language
teaching.

2. Theoretical frameworks on L2 tense-aspect research

In general, theoretical frameworks are broadly dierentiated among each


other depending on the way that both dependent and independent variables
are dened. First, not surprisingly, the way that the object of study (i.e.,
aspect as a theoretical construct) is dened has a clear eect on how we
frame the questions that can be asked about the acquisition of tense-aspect
concepts. In this respect, specic linguistic theories (e.g., functional or
formalist theories) provide a broad theoretical perspective that guides the
characterization of the object to be studied (e.g., the inclusion of adjuncts
as part of the main denition of aspectual constructs, the interactional
contexts of use of the target item). Second, the selection and denition of
independent variables that can account for the acquisition of tense-aspect
(e.g., the eect of the L1, the type of data collection procedure) provides
an additional layer of information that has a signicant inuence on how
we frame the questions and analysis of the dependent variable.

2.1 Theoretical framing of the dependent variable


The most basic way to frame the analysis of the acquisition of tense-aspect
is to examine how dierent approaches view the object to be learned per se
(i.e., the theoretical construct). In practice, the focus is on dening the
boundaries between lexical, grammatical, discursive, and pragmatic-
contextual representations of aspectual knowledge. For instance, syntactic
approaches mostly draw the boundary at internal and external arguments,
whereas other approaches move the denitional boundary beyond internal
and external arguments up to the level of adjuncts and context. These dis-
tinctions are not trivial, and one possible consequence is that the analyses
of aspect within a narrow specication of the construct leaves out of
the realm of investigation the eects of discursive and pragmatic context
(whether the latter are necessary or not is part of the theoretical debate).
The importance of the topic of circumscribing the denition of aspectual
426 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

meanings is highlighted by the fact that a common thread in the chapters


in this volume has been their emphasis on establishing the boundaries
across levels of aspectual meanings (see Chapters 5 and 6).
The classical division of lexical aspect versus grammatical aspect (cf.
Andersen 1986), or situation aspect versus viewpoint aspect (cf. Smith
1991/1997) highlights the fact that the denition of aspect can have more
than one single interpretation depending on how much contextual infor-
mation we bring to bear on this denition. The above-mentioned contrasts
(e.g., lexical and grammatical aspect) are clearly represented in the various
denitions of aspect. As one example, let us review two well-known dis-
tinct denitions provided by Comrie (1976) and Klein (1994) respectively
(added emphasis):1

Aspect represents a way of viewing the internal temporal constituency of a


situation (Comrie 1976: 3)
Aspect concerns the dierent perspectives which a speaker can take and
express with regard to the temporal course of some event, action, process, etc.
(Klein 1994: 16)

That is, whereas Comries denition focuses our attention on the linguistic
characterization of the features of the situation in reality, Klein focuses
our attention on the speakers perspective as opposed to the situation
in reality or the linguistic characterization of such reality. It is obvious
that Kleins denition is the one that introduces a complex interpretation
of the meaning of aspect given its emphasis on the subjective interpreta-
tion of situations by speakers and hearers. Such contrast in meanings
has been discussed in great detail by Smith (1991/1997) and reminds us
that aspect is a complex topic given by the various layers of interpretation
possible.
Despite the apparent subjectivity of how we use aspectual markers to
highlight specic construals of situations in reality (i.e., Kleins deni-
tion above), there are, however, conventions of use (cf. Smith 1991/1997)
of grammatical aspect that provide limits to the arbitrariness of the sub-

1. Granted, both Comrie and Klein oer a more extensive (and nuanced) deni-
tion of aspect. However, we believe that the basic gist of their denitions is
adequately represented in the quotes presented above.
Integrating 427

jective selections. For instance, Smith (1997: 7) points out that [s]tandard
choices focus on aspectual properties that are salient at the basic level;
non-standard choices focus on other properties. Accordingly, Smith con-
tends that sentences (5) and (6) are two possible linguistic descriptions of
the same event (both examples from Smith).
(1) The ship moved.
(2) The ship was in motion.
According to Smith, sentence (1) conveys the conventional interpretation
of the event according to the properties that are salient perceptually and
functionally. In contrast, sentence (2) is representative of the less conven-
tional description of the same eventuality, because it depicts the event as
a state. Either representation has obvious consequences that speakers
may or may not choose to highlight (e.g., option (2) freezes the motion
of the ship). And even if speakers select the unconventional depiction,
Smith notes that receivers will not be misled by unconventional choice.
Knowing a language includes the standard, basic associations of verb con-
stellation with verb type. Moreover, there are cases in which speakers are
even more constrained by the choices, as in the following examples from
Salaberry (2008) (adapted from Langackers (1982) examples in English)
and Doiz-Bienzobas (1995):
(3) a. Sally fue (PRET) de Phoenix a Tucson.
Sally went from Phoenix to Houston.
b. *Esta ruta fue (PRET) de Phoenix a Tucson.
That route went from Phoenix to Houston.
(4) a. La carta deca (IMP) hola.
The letter said hello.
b. *La carta dijo (PRET) hola.
The letter said hello.
As can be seen in both sets of examples, the perfective form is ungram-
matical, thus showing very concrete limits to the range of subjectivity to
be exercised by speakers.
Even though previous classications of theoretical perspectives on the
acquisition of tense-aspect are not necessarily restricted to the identication
of the object of study, there is a clear correlation of theoretical perspective
and the object of study (i.e., the dependent variable). Thus, Montrul and
Salaberry (2003) listed ve theoretical characterizations of the development
of tense and aspect morphology in Spanish broadly dened according to
428 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

the description of the dependent variable as follows: (a) lexical semantics,


(b) generative, (c) context-based, (d) communicative, and (e) cognitive-
perceptual. For instance, the rst perspective, lexical semantics, has become
associated with the Lexical Aspect Hypothesis (LAH).2 This hypothesis
is based on the concept that the lexico-semantic characteristics of verbal
predicates (i.e., lexical aspectual classes such as states, activities, and telic
events) are a major contributor to the emergence of tense-aspect morphology.
More precisely, the LAH, in its original formulation, advanced the notion
that in the early stages of acquisition, verbal morphology used by L2 learners
marks inherent aspectual distinctions instead of tense or grammatical aspect.
Although the LAH is primarily focused on the beginning stages of acquisi-
tion of grammatical aspect, it does restrict its analysis to the lexical nature
of aspectual marking (see Chapter 8).
Similarly, the generative perspective is based on the minimalist theory
of language, and thus it is focused primarily on the syntactic frame of
reference to aspectual contrasts, or at most the syntax-semantics interface.
More precisely, it focuses on how the acquisition of tense-aspect morphology
correlates with the acquisition of syntactic features: Giorgi and Pianesi
1997 was the principal theoretical reference used by the earliest studies on
the acquisition of aspect from the minimalist point of view (e.g., Slabakova
and Montrul 2002: 122). From such perspective, grammatical aspect is a
functional category that is located in a dierent position than lexical
aspect in the clause structure (syntactic trees). Cross-linguistic variation
occurs when, for instance, the Spanish e perfective features are checked
overtly through inectional morphology, whereas in English only the
perfective feature is checked with Simple Past (verbs in English are
perfective). More importantly, generative perspectives typically restrict
the denition of aspect to the combined meanings of main predicate and
its arguments (see Chapter 4).
The third perspective does not focus on semantics or syntax as the two
previous ones, but rather it gives prominence to contextual factors such as
type of discourse and text. This viewpoint is closely associated with the
Discourse Hypothesis and the Distributional Bias Hypothesis. The dis-
course hypothesis (Bardovi-Harlig 1998) proposes that learners will put
to use their developing morphology to mark specic types of information
in specic texts, namely foreground in narratives (see Chapters 7 and 9).
On the other hand, the Distributional Bias Hypothesis (Andersen 1994;
Andersen and Shirai 1994) states that the distribution of verbal morphology

2. The terms Lexical Aspect Hypothesis (LAH) and Aspect Hypothesis (AH) are
used indistinctively by dierent authors in the chapters of the volume.
Integrating 429

in dierent types of verbal predicates (cf. the Lexical Aspect Hypothesis)


mimics the distribution found in the linguistic input received by L2 learners.
The fourth perspective is cognitive-perceptual and focuses on the role
of perceptual saliency and frequency of verbal endings (e.g., regular versus
irregular) to account for the acquisition of L2 tense-aspect. Emergent verbal
endings will be those that are more frequent and irregular (see Chapter 3).
The nal theoretical perspective is labeled communicative because it makes
a distinction between semantic meaning and pragmatic implicature for the
acquisition of tense-aspect in the sense that the way a speaker construes a
situation (and uses verbal morphology accordingly) depends on how the
speaker views the situation from a communicative-pragmatic point of view.
As already pointed out by Montrul and Salaberry, some of these pro-
posals have received more attention than others, at least in terms of the
number of empirical studies carried out within any given perspective.
More specically, the largest proportion of the empirical studies has been
dedicated to the analysis of data that can help us reject the hypotheses
provided by the lexical semantics, generative, and contextual approaches.
It appears that this trend continues to be the case as the dierent chapters
in the current volume attest. In addition, even within one theoretical
approach, there has been some further elaboration of the original theoret-
ical claims. This is clearly the case of the generative approach (the dier-
ences in theories in Chapter 4) and the cognitive approach (see Chapters 1
and 3).

2.2. Theoretical framing of the independent variables


Another way to examine the dierent theories of the L2 acquisition of
tense and aspect is to focus on the factors that act as independent variables
to account for such acquisition. Salaberry (2008) listed ve variables: (a)
lexical aspect, (b) discourse structure, (c) syntactic structure, (d) L1 transfer,
and (e) perceptual saliency. Each one of these variables is closely related
to the theoretical perspectives proposed in Montrul and Salaberry (2003),
although the latter classication gives less prominence to the potential
explanatory variables that account for the acquisition process, while focus-
ing more on the theoretical description of the actual object of study (i.e.,
aspect). In particular, the classication scheme provided by Montrul and
Salaberry (2003) largely embeds the eect of independent variables, such
as L1 or cognitive factors, within the macro level of theoretical approaches
to the acquisition and development of knowledge about aspectual concepts
430 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

in the L2. In contrast, Salaberry (2008) describes the explanatory power of


the above-mentioned ve independent variables, given that these variables
are the ones that have been the most prevalent in the research design of
previous studies on the L2 development of tense-aspect.
The analysis of the eects of both lexical aspect and discourse ground-
ing has been central to the analysis of the acquisition and development
of tense-aspect construals since the 1980s, when Andersen and Bardovi-
Harlig published their rst studies on the eect of both inherent lexical
aspect and grounding. There are several studies that trace the history of
the studies that used these two independent variables (e.g., Bardovi-Harlig
2000; Labeau 2005; Salaberry 2000). More recently, syntactic structure
has become a prominent factor on the analysis of tense-aspect acquisition
(see Chapter 4). Interestingly, whereas all of these variables (lexical, dis-
course, syntax) have been associated with dierent types of universal pre-
dispositions, the role that the native language of the learners may play
as an independent variable has remained relatively less studied (albeit
indirectly considered by the majority of studies). Similarly, general cogni-
tive predispositions have also been indirectly assessed in previous studies
(e.g., the role of saliency of morphological markers eected through fre-
quency eects).
The role of the learners L1 is especially important for the notion of a
possible general marker of tense that overrides any marker of aspect. For
instance, in the case of L1 English speakers learning L2 Spanish, the
simple past in English marks only tense, but not aspect (cf. Giorgi and
Pianesi 1997). Thus, Salaberry (1999, 2000) and Wiberg (1996) predicted
that during the rst stages of L2 development, learners will mark tense
rather than aspectual distinctions. This proposed tense-aspect dissociation
brought about by cross-linguistic eects can be substantiated along com-
plementary lines of theoretical analysis, including the eect of linguistic
factors (e.g., lexical aspect, grounding), and cognitive factors (e.g., percep-
tual saliency through instruction or prototypicality of past tense marking).
For example, Salaberry (2011) analyzed data that seem to show that the
eects of lexical aspect and discourse grounding can be integrated into
one single developmental process that is inherently tied to the prevalent
marking of tense through the eect of the L1. Similarly, McManus (2011:
206) analyzed the eect of form-meaning pairs from L1 German and L1
English on the acquisition of L2 French among university students. In his
analysis, McManus concluded the following:
Integrating 431

English low group learners show stability in using the IMP to mark progres-
sivity but not habituality. This is arguably because habituality needs mapping
but progressivity does not. German low group learners indicate diculty in
marking habituality and progressivity with the IMP because both meanings
require remapping. Therefore, it is claimed that when dierences exist in
how the L1 and the L2 express meaning, it is these L1-L2 dierences that
constitute diculties in L2 development.

Thus, the L1-L2 mapping can be considered an important variable that


deserves further study (cf. Carroll and von Stutterheim, 2003; von Stutter-
heim, Carroll, and Klein, 2009 for the importance of the role of L1-L2
mapping and how dierent types of discourse are organized in inter-
language).
Cross-linguistic inuences are also important to weigh the relevance
of some of the proposed independent variables. For instance, Giacalone
Ramat and Rastelli (Chapter 11) summarize several ndings from their
studies on the acquisition of perfective and imperfective past tense forms
in L2 Italian (i.e., Passato Prossimo and Imperfect) that point in the direc-
tion of strong L1-specic eects that override other factors such as lexical
aspect. For instance, L1 German learners used the passato prossimo with
all classes of verbs, showing many cases of overextension of the passato
prossimo in place of the imperfect (402). Similarly, durative predicates
emerge with perfective marking, contrary to Andersens predictions, while
stative predicates like be, have, and modals receive an imperfect
marking, somewhat later than perfective marking. In their analysis,
Giacalone Ramat and Rastelli conclude that [w]hat can be suggested
is that the Aspect Hypothesis is too strong, although the use of verbal
morphology is clearly inuenced by lexical aspect. Similarly, in his analysis
of German-speaking and English-speaking students learning L2 French,
McManus (2011) concluded that the eects of lexical aspect increase in
association with increasing prociency, also providing evidence against
the strict interpretation of the Lexical Aspect Hypothesis.
Findings and claims such as the ones summarized above have prompted
a healthy debate on the relevance of old and new theoretical proposals.
Shirai (2004), for instance, argued that the various results discussed above
(inter alia) are debatable, given the possible eect of methodological
factors. Thus, Shirai (2004: 91) claimed that the ndings that support or
reject the main tenet of the AH are confounded by the research method-
ology used in various studies. Arguably, on the one hand production data
produce results that often go against the AH, and, on the other hand,
paper-and-pencil tests often show patterns consistent with the hypothesis.
432 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

McManus (2011: 214), however, points out that if the AHs predictions
are not empirically supported (i.e., not reected in spontaneous produc-
tion), then its claim for universality rings hollow. How can the AH be a
valid theory of L2 development if it fails to convincingly predict actual
patterns of L2 development? As we can see, it is dicult to disentangle
the eects of theoretical factors from methodological ones. Indeed, there
are enough methodological dierences in the various studies carried out
to test the eect of the various proposed independent hypotheses that it is
worth considering the relevance of methodological factors.

3. Trends in methodology to study L2 tense-aspect

The eld of SLA has developed in such a way in the last 20 years that
methodologies are not only dierent in the various theoretical frameworks
(e.g., UG versus LAH), but also within the frameworks. For instance,
Chapters 7, 9, and 11 in this volume provide a review of methodologies
that could be considered within one framework (functional-discourse).
Nonetheless, these chapters also provide evidence that within each frame-
work one can choose from plenty of elicitation tasks and methods. For
instance, Comajoans review of 19 studies (Chapter 9) that have used the
foreground-background distinction in L2 studies shows that their deni-
tions vary considerably, and consequently their way of analyzing the data
varies as well. Along the same lines, Bardovi-Harlig (Chapter 7) mentions
three major types of open-ended tasks for tense and aspect L2 research:
narratives, description, and conversation, and within narratives, she studies
conversational and elicited narratives and within the latter she discusses
personal, impersonal, and personalized narratives.
In designing a method to investigate the acquisition of L2 tense and
aspect, at least the following seven factors need to be taken into account
(Montrul and Salaberry 2003; Salaberry and Ayoun 2005; Salaberry 2008):

1. Type of data and data collection procedures


2. Selection and use of operational tests to determine categories (e.g.,
lexical aspectual; discourse, etc.)
3. Language-specic characteristics of tense-aspectual contrasts
4. The eect of learning environment
5. Types of input
6. Subject-related factors (age, level of prociency of L2, motivation)
7. Denition of ultimate attainment in L2 tense-aspect acquisition
Integrating 433

The number of factors listed above attests to the complexity of studying


L2 tense-aspect and the diversication of theories. However, some of the
most pressing issues regarding methodology in L2 tense-aspect, which are
often cited in the current volume and elsewhere, concern primarily factors
1 and 2 above. These factors are especially important given that researchers
have direct control over them as part of research design.3 While in princi-
ple it is possible for researchers to make methodological decisions about
factors 36, their eects are primarily beyond the control of research
design once some basic decisions have been made (i.e., source and target
language, subjects background). With regards to the more ne-grained
decisions associated with factors 1 and 2 above, there are some specic
methodological concerns that deserve special attention, and as such are
often mentioned in the chapters in the current volume: (a) the use of multi-
variate approaches to study how dierent variables interact, and (b) the
selection of types of data analyses.

3.1. Multivariate analyses


The complexity of most tense-aspect systems in language and its interaction
with other variables, as explained earlier, warrants the need to analyze the
L2 acquisition of tense-aspect from a multivariate perspective. Even though
it is expected that in an early period of study and development of research
within the discipline of SLA single variables will be studied, after two
decades of research in the L2 acquisition of tense-aspect and the recurrent
nding that dierent linguistic levels (e.g., phonology, morphology, syntax,
and discourse) interact in the acquisition of tense-aspect, further research
projects that integrate the interaction of two or more variables need to be
developed. In line with this argument, Bayley (Chapter 10) reminds us of
Young and Bayleys (1996: 253) principle of multiple causes: it is unlikely
that any single contextual factor can explain the variability in the [inter-
language] data. Indeed, in Chapter 10, Bayley summarizes a list of factors
that aect the use of tense-aspect markers such as: perceptual saliency
(e.g., position of the aspect marker), contrastive morphological saliency
of present-past forms, L1 prosodic transfer, frequency and prototypicality,
social factors, learner characteristics (L1, other languages, L2 prociency,

3. Factor 7 would also fall within research design, but it has already been ana-
lyzed in the previous section.
434 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

type of learning environment, etc.), discourse context, and task type. The
factors discussed by Bayley can be divided into two main groups, those
that are related to theoretical constructs of the tempo-aspectual system of
languages and those that are related to tasks to elicit data. For this reason,
the adoption of a multivariate analysis perspective can be considered an
issue that is both related to the theoretical framework and to the specic
methodology.
The most common multivariate analyses in L2 tense-aspect have been
related to the study of the LAH and the DH (see summaries of previous
studies in Bardovi-Harlig 2000 and Salaberry 2008). However, in the
analyses of the conspiring factors from the two hypotheses, results tend
to be discussed as if there were two competing forces and one had to play
a more important role than the other, when it is very likely that one may
subsume the other, that both simply interact and are in constant interac-
tion, or that at dierent periods of acquisition one is more important
than the other (see Salaberry 2011 for empirical data that support the
latter argument). In addition, the studies that have examined the interaction
of the two hypotheses have not adopted a multivariate statistical analysis
(like the one argued by Bayley in Chapter 10), but rather statistical techni-
ques that may be more appropriate for quantitative univariate analyses.
Current research is also investigating how dierent variables interact in
the L2 acquisition of tense-aspect from perspectives that go beyond the
LAH and the DH. For instance, Ellis (Chapter 3) investigates how fre-
quency (and its many implications) and salience play a decisive role in
the acquisition of L2 tense-aspect and argues for multivariate analyses
not just from a linguistic perspective (e.g., how are dierent verb forms
marked), but rather from a cognitive perspective that studies how human
beings learn in general and how they learn an L2 specically: we need
models of learning, language, meaning, usage, interaction, development,
and emergence that take all these factors into account dynamically.

3.2. Data analysis


There are two main factors related to data analysis that are amenable to a
great deal of control on the part of researchers: the coding of data and the
use of interpretation- or production-based data.The rst factor relates to a
clear denition and use of operational tests to code and analyze data, as
argued in several chapters of the current volume (e.g., Chapter 6, 8, and
9). Providing the specic tests and stating how data were coded will allow
Integrating 435

researchers to consider how conrming or conicting results relate to the


methodology used in a study. In this view, the guidelines in the chapters
by Shirai and Comajoan regarding the coding of lexical aspect and dis-
course grounding, respectively, may prove useful for designing new studies.
The second issue that any study that deems itself to be rigorous needs to
include is a discussion of task selection and how it relates to the acquisi-
tion of L2 tense-aspect features. As clearly shown, for instance, in the
chapters by Bardovi-Harlig and Giacalone Ramat and Rastelli, dierences
in task design may have an important eect in the elicitation and produc-
tion of learners interlanguage. Consequently, studies need to be clear on
their motivations for selecting the type of task in one study as opposed to
other tasks used in similar studies, and assess how such methodological
decision may aect the results obtained.
The second factor associated with data analysis that deserves special
attention for its implications in building a theory of L2 acquisition of
tense-aspect is the use of interpretation versus production tasks. Production-
based studies tend to focus on the building of a large enough database that
can provide researchers with information about the development of tense-
aspectual concepts as speakers interact in real world use. The specic focus
on development in particular is consistent with the underlying assumption
that learners are guided by a general problem-solving, cognitive-based L2
system when they put to use their knowledge of the theoretical construct.
The best examples of this kind of study are the studies carried out by the
European Science Foundation in the early 1990s (e.g., Dietrich, Klein and
Noyau 1993). In contrast, interpretation-based studies tend to focus on
the identication of target-like selections of grammatical choices, irrespec-
tive of the general cognitive constraints that may aect such target-like
interpretation.
The arguments about interpretation- and production-based procedures
tend to make associations between the above-mentioned research proce-
dures and specic theoretical approaches. Thus, some researchers have
dismissed the contributions of functionalist-based approaches due to their
reliance (or over-reliance) on interactional language use. Rothman and
Iverson (2008: 308), for instance, claim that Insofar as it is justied to
assume that target-like interpretation provides better evidence for assess-
ing underlying competence than looking at morphological production
alone . . . it is fair to claim that generative approaches are better equipped
to diminish the inherent opaqueness of determining linguistic competence
(emphasis added). Notwithstanding the advantages of interpretation-based
436 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

protocols of data collection, the procedures based on the analyses of data


production are no less important in their own right. That is, meaning-
oriented production tasks have some inherent advantages over interpreta-
tion tasks. Most importantly, (a) they increase processing demands, thus
making it less likely that learners can eciently access conscious applica-
tion of monitored rules, and (b) they increase the likelihood that con-
textual factors that shape decisions about tense-aspect marking will not
be left out of consideration. By their very nature, interpretation tasks
require that researchers create and delimit the context that will help
determine the judgments of acceptability of the various morphosyntactic
choices selected for analysis. That is, the fact that research design circum-
scribes the problem space increases the chances that empirical results may
be an artifact of the research procedure (see Chapter 6).
Apart from the specic features of each procedure of data collection,
there are also discrepancies related to the analysis of data (however it
was obtained). More specically, the assessment of the research value of
interpretation- and production-based procedures is sometimes obfuscated
by the confusion between data and data analysis. For example, Rothman
and Iverson (2008: 309) claim that usage/experience based approaches
that employ a functionalist view of morpho-phonological use . . . face an
explanation problem since, in line with others, our data showed robust
semantic properties of target competence in this L2 domain. The fallacy of
the previous argument, however, is that we confuse data with data analysis.
The fact that some data may be indicative of a certain outcome does not
necessarily invalidate alternative explanations given that there is no indepen-
dent criterion that adjudicates the outcome (it is theoretically-biased given
that there is more than one theoretical option).

4. Towards a mixed methods methodology?

The development of SLA research has not been an exception to the chasm
in theories and methodologies that has taken place in other academic dis-
ciplines. For instance, the early debate about the proliferation of theories
in SLA seen as benecial or not to theory construction and development
of the eld (see a summary in Jordan 2004) and the recent one about
social versus mentalistic perspectives of SLA (see a summary in Laord
2007) are proof of the diversity of voices and opinions in SLA and to an
extent to a chasm that has not been solved yet. The divergence of voices is
Integrating 437

also evident in the study of L2 tense-aspect, as witnessed in the dierent


theoretical frameworks and methodologies described in the current volume.
As mentioned earlier, dierences in how studies are designed (operational
tests, task design, and so on) do arise and discussions related to how the
subdiscipline of L2 tense-aspect acquisition develops are present as well
(e.g., Lardiere 2003; Shirai 2007; Slabakova 2002). In this respect, we
would like to suggest a way out of some of the discussions regarding meth-
odology by advocating for a methodology that in other academic disciplines
has proven fruitful, namely mixed methods (Denzin 2010; see a review in
Tashakkori and Teddlie 2006).
Dornyei (2007) provided an introduction to mixed methods in applied
linguistics and we use his discussion as a point of departure. Dornyei
dened mixed methods following the foundational work of Strauss and
Corbin (1998):

Qualitative and quantitative forms of research both have roles to play in


theorizing. The issue is not whether to use one form or another, but rather
how these might work together to foster the development of theory.
Although most researchers tend to use qualitative and quantitative methods
in supplementary or complementary forms, what we are advocating is a true
interplay between the two. The qualitative should direct the quantitative
and the quantitative feedback into the qualitative in a circular, but at the
same time evolving, process with each method contributing to the theory in
ways that only each can (43).

For the purpose of our discussion, we highlight three fundamental ques-


tions that Dornyei posed to L2 researchers who want to adopt a mixed
method methodology: (a) Why do we want to mix methods?, (b) Can
dierent methods be really integrated?, and (c) What is the best way of
mixing methods?, In the following section, we analyze in further detail
each question and answer provided by Dornyei, in order to provide more
detail on the potential benets of such an approach for the analysis of
research in tense and aspect.
The rst question is: Why do we want to mix methods? Dornyeis
answer to this question is that mixed methods are useful to (a) expand
our understanding of a complex issue, (b) corroborate ndings through
triangulation, and (c) reach multiple audiences. Indeed, reviews of research
have often mentioned that tense-aspect represents a complex topic, both
from the perspective of the learner as well as the researcher or the lan-
guage instructor (e.g., Bardovi-Harlig 2000; Salaberry 2008, inter alia).
438 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

Thus, the goal of expanding the understanding of a complex issue seems


an appropriate one. Such an expansion of our understanding can fulll
four dierent functions (complementary, development, initiation, and expan-
sion) (Dornyei 2007). First, quantitative and qualitative studies can be
complimentary in the traditional sense that they assist each other to study
dierent aspects of one phenomenon (often, this is done carrying out
qualitative research to explore a topic, and then, after a hypothesis is
posited, to test it in a quantitative perspective). Second, qualitative and
quantitative methods can be used sequentially in a way that the ndings
of the rst method are used to develop the second method. Giacalone
Ramat and Rastelli (Chapter 11) advocate for this type of methodology
when they suggest that a qualitative method (verb pairs) be used to gen-
erate hypotheses that will inform the development of further quantitative
studies. Third, sometimes it may be useful to use dierent methods to elicit
dierent results purposefully and thus be able to study the eect of methods
themselves and posit new hypotheses. Finally, the two types of methods
may be used to expand the topic or eld of study. The combination of
qualitative methodologies to investigate the learning of tense and aspect
in the classroom with quantitative studies such as those mentioned in the
article by Niemeier (Chapter 1) could be considered an example of this
function.
The second question refers to the integration of methods: Can dierent
methods be really integrated? More precisely, is there a danger that the
paradigmatic collision of methods may cancel out any potential benets?
Dornyeis answer to this question is that mixed methods is not an any-
thing goes approach. One can use multiple methods as long as one is
consistent with the methodology. Dornyei, for instance, points out that
the key process is principled mixing: Collect data so that the resulting
mixture or combination is likely to result in complementary strengths and
non-overlapping weaknesses (167). Another manner to examine the inte-
gration of methods and its advantages is to relate it to the integration of
analyses at the micro- and macro-levels. There is no solid tradition of in-
tegrating methods in research in tense and aspect. However, there have
been constant calls for an integration of dierent hypotheses (the Aspect
and Discourse Hypotheses mainly) with their corresponding methods
and lately of research paradigms (Salaberry 2008). For instance, as
mentioned earlier, Giacalone Ramat and Rastelli (Chapter 11) advocate
for mixing methods even though they do not focus on how they can be
integrated. Similarly, even though Bardovi-Harlig (Chapter 7) does not
refer to mixed methods explicitly, she also argues for the use of dierent
Integrating 439

types of tasks that can provide complementary results to the study of tense
and aspect. Finally, Comajoan (Chapter 9) reviews dierent approaches
to the study of narrative discourse grounding while highlighting current
studies in which dierent methodologies are integrated (e.g., Rosi 2009).
In his third question, Dornyei asked, What is the best way of mixing
methods? There are multiple combinations depending on two typological
principles: sequence (what methodology is rst? quantitative or qualitative?)
and dominance (which dominates?). Dornyei provided a variety of exam-
ples that combine mostly questionnaires (quantitative) and interviews (qual-
itative) (also self-reports, observation, and experiments). Research in tense
and aspect has used both qualitative and quantitative methods, but very
few researchers have combined them. For instance, Liskin Gasparro (1996)
used retrospective interviews to study how advanced learners of Spanish
argued for the use of preterite and imperfect forms. Other studies examined
the same object but using dierent data collection tasks (e.g., video retell-
ings, story book retellings, and conversations in Comajoan 2001), but there
have been few studies that make an eort to combine methodologies and
tasks (an exception is Rosi 2009, who provided quantitative and qualitative
analyses).
Dorney wondered why researchers do not mix methods more, and
pondered about three possible main reasons. First, researchers in some
elds do not have enough knowledge and training about method mixing.
Second, researchers do not have enough expertise to put into practice
research designs based on mixed methods. And third, publication pressure
may encourage researchers to publish dierent studies with dierent meth-
odologies separately instead of using one single study that combines both
types of method. Those disciplines that have adopted a mixed method per-
spective have soon discovered that such an approach is not the solution to
all problems (see a review in Hesse-Biber 2010; Tashakkori and Teddlie
2006). A major concern has been how to avoid placing one (commonly,
quantitative) rst and relegate the other (often, qualitative) to a subsidiary
position (what Howe 2004 calls mixed-methods experimentalism). As dis-
cussed above, the combinations of methodologies are multiple, and de-
pending on the goals of the study one can be more appropriate than the
other. For instance, Giacalone Ramat and Rastelli (Chapter 11) explicitly
refer to the relationship between the two types of methodologies and argue
for carrying out qualitative studies rst that would inform later quantita-
tive studies later (see Howe 2004 for a detailed discussion of positioning
qualitative studies rst in the eld of educational studies).
440 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

In sum, mixed methods are not the solution to all the theoretical and
methodological concerns raised in the volume, but incorporating such a
new perspective, which has been adopted in other disciplines and proven
productive, may contribute to the development of research in L2 tense-
aspect. Ultimately, in regard to theory and methodology, we agree with a
number of Jordans (2004: 115116) guidelines for the construction of a
rational theory of SLA, which can adequately be applied to research in
L2 tense-aspect: (a) research is inseparable from theory as has been
clearly demonstrated in the chapters in the volume and in the current
chapter, (b) we cannot formalize the scientic method, that is, there is
no one method to carry out research in L2 tense-aspect and, as argued
by Jordan, there is no clear mark between science and non-science,
but rather multiple methods whose combinations will contribute to the
development of theory and methodology, and (c) there is no need for
paradigmatic theories, in the sense that there should be no concern for
the proliferation of theories that may contribute to the development of
the eld as long as theories are well constructed.
In line with the claims advanced by Jordan (2004), the analysis of
the compatibility of the dierent theoretical perspectives can be framed
by a number of broad discussions (which are not necessarily dierent
from previous general discussions in SLA). In the end, it is clear that a
multiple methods approach is best. Zyzik and Gass (2008: 385), for
instance, argue that . . . multiple methods of data collection . . . enrich
our understanding of a variety of linguistic structures. In particular, Zyzik
and Gass highlight the fact that recent studies conducted within generative
approaches have started to include more contextualized procedures to
gather evidence to test proposed hypotheses (e.g., sentence-selection tasks
as in Bruhn de Garavito and Valenzuela 2008 or paragraph interpretation
as in Rothman and Iverson 2008). In light of this new trend in tense-aspect
studies, our proposal is to encourage the use of mixed method method-
ology in tense and aspect research.

5. The eect of explicit instruction on the acquisition of tense and aspect

As it is well known by any researcher who is also a second language


teacher or learner, the explicit instruction of tense and aspect marking
is central to most second language courses, and as such they clearly play
an important role in grammatically focused pedagogical materials. Both
Niemeier (Chapter 1) and Doiz (Chapter 2) make reference to several
Integrating 441

instructional eects, both potential and actual, that can have an eect on
the acquisition of tense-aspect contrasts. In this section, we discuss three
main principles to implement a way of teaching informed by current
research and methodology in the acquisition of L2 tense and aspect.
Before embarking upon the discussion of the principles, we tackle the issue
regarding the eect of instruction on the acquisition of second languages.
Ellis (1997) addressed a number of problems in the teaching and learn-
ing of grammatical items in a second language, and he argued for the
Weak Interface Hypothesis: the kind of explicit knowledge which results
from formal grammar instruction can become, through practice, the kind
of implicit knowledge that is required for use in communication. To be
able to prove whether explicit knowledge can be converted into implicit
knowledge, one has to prove that the learners awareness and use of the
particular grammatical structure improves after instruction. There have
been a number of empirical studies that have demonstrated that learners
who receive instruction outperform those who do not, both with respect
to the rate of acquisition and ultimate level of achievement (Long 1983,
1988; Ellis 1985, 1990, among others). Thus, on the basis of the research
on the eects of form-focused instruction on accuracy, there is enough evi-
dence to show that form-focused instruction can result in denite gains in
accuracy (Ellis 2002). For instance, the European Science Foundation
carried out a number of studies on L2 teaching including target languages
such as English, Dutch, French, German, and Swedish. Dietrich et al.
(1995) conducted dierent studies with learners of each of the ve target
languages. Some of the learners received instruction, whereas some did
not, which allowed for a comparison between instructed and uninstructed
learners. One of the key ndings was that instruction on its own was not
responsible for success, but rather it was a tool to accessing and grasping
the target grammatical phenomenon; and, as such, it had a positive eect
on its acquisition. In sum, eective instruction may have an impact on the
acquisition of L2 tense and aspect. What we dene as eective instruction
has to do with the three principles discussed next.
The rst principle has to do with understanding the complexity of the
instruction and learning of L2 tempo-aspectual systems. Blyth (2005),
using the criteria for determining the diculty of grammatical struc-
tures adapted from Ellis (1997, 2002) showed that Romance aspect is
particularly dicult for English-speaking students because all eight criteria
(formal complexity of the system, functional complexity of the system, relia-
bility of rules, metalanguage of rules, L1-L2 contrasts, frequency of forms
442 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

in the input, redundancy in the relationship between form and meaning,


and saliency in the input) contribute to making L2 tense-aspect highly
complex and dicult to learn. Thus, it is important that both L2 teachers
and learners are made aware that teaching and learning a new tempo-
aspectual system takes time and often requires adopting a new perspective
on how time is encoded and viewed in dierent languages. A possible
new way to raise such awareness may be done adopting cognitive gram-
mar as a method of describing the grammar and designing new teaching
materials.
The second principle focuses on the teaching and learning of appropriate
form-meaning relationships. Learning a language means both learning
new forms and new meanings. In fact, the learning challenge is even more
dicult because learners also need to reorganize or reconceptualize already
known notions. Thus, one of the tasks of the L2 instructor is to help
learners become aware of the broad range of form/meaning aspectual asso-
ciations of the target language (Blyth 1997). Since classroom learners seem
to be better at learning verbal forms than verbal use or meaning (Bardovi-
Harlig 1995), it is necessary to focus on verbal meaning and use when
teaching new forms (Larsen-Freeman 1990). Thus, eective instruction
should avoid the still common practice of teaching aspectual contrasts
(e.g., between perfective and imperfective forms in Romance or Slavic
languages) with isolated sentences that do not include enough contextual
information to make a decision on what form to use. It is debatable, how-
ever, whether such broad contextualization should occur from the begin-
ning stages of instruction, or whether learners need to go through an
initial developmental stage in which frequency and salience determine
that learners rely on prototypical choices. That is, although the avail-
able empirical data do show that learners follow a prototypical to non-
prototypical sequence of acquisition (e.g., Salaberry 2011 for a wide range
of data supporting this claim), it is not clear whether this eect is created
by current pedagogical practices (e.g., preterite is typically presented before
the imperfect), or whether learners can handle a more complex representa-
tion of aspect as soon as they start marking aspectual contrasts.
The issue of providing appropriate form-meaning relationships to lan-
guage learners has to do with their access to linguistic input in the L2.
Classroom L2 acquisition is dierent from naturalistic language learning,
mainly because classroom learners do not receive random or spontaneous
language, with an abundance of pragmatic and contextual clues, but a lan-
guage that is manipulated (mostly through simplications and reductions
Integrating 443

of context) in order to facilitate the learning of the particular phenomenon.


However, even if classroom interaction incorporates more free conversa-
tional tasks, it is doubtful that the functional needs of true communicative
interaction can be successfully recreated. The type of input that learners
are exposed to is also of interest because previous studies have shown,
for instance, that untutored learners mark perfective and imperfective
meanings with pragmatic means, whereas classroom learners mark them
with morphosyntactic means (see review of studies in Salaberry 2000).
Therefore, if classroom learners are to develop an appropriate representa-
tion of grammatical aspect and understand the dierence between several
aspectual forms, then other strategies on top of the contextual ones must
be presented. For instance, it may prove fruitful to train learners to dis-
cover how grammatical aspect interacts with predicational aspect (Gonzalez
2003) and to raise their awareness that form-meaning associations are
complex and not just a matter of learning isolated forms (see a review of
other techniques in Blyth 2005).
Finally, the third principle focuses on the evaluation of the background
knowledge (both general and tempo-aspectual) that learners bring to the
learning task. Thus, when learning new form-meaning aspectual associations
we have to keep in mind that learners do not start from scratch, but rather
they already possess knowledge about temporality and how to mark it in
their L1. In order to link what they are learning with what they already
know (Ausubel 1968; Slagter 2000), learners need to be made aware of
both their L1 and L2 tempo-aspectual systems. This is no easy task as the
level of awareness that learners have may vary and the metalanguage
to describe the systems may be highly complex to the extent that both
learners and teachers are not prepared for accurate descriptions of the
tempo-aspectual system. A possible way to make learners aware of the
systems is to resort to a new way of describing grammatical relationships
through cognitive grammar, such as the descriptions by Niemeier (Chapter
1) and Doiz (Chapter 2) for English and Spanish, respectively. As argued
by Niemeier, due to its usage-based nature and its focus on (conceptual)
meaning, cognitive grammar may oer foreign language learners a de-
scriptively adequate and intuitively comprehensible account of grammar
(11). Precisely, as mentioned in the previous principle, tense and aspect are
so closely related to language use that only a description (or theory) of
grammar that gives use such a prominent role may prove adequate for
language learners. Furthermore, cognitive grammar makes an eort at
providing intuitive explanations that capture the generalities of grammar
rules rather than the idiosyncratic exceptions to the rules (often in the
444 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

form of lists). As mentioned by Niemeier and Doiz, there is to date little


research that has fully applied cognitive grammar to classroom instruc-
tion, but new teacher training materials and grammar manuals (e.g.,
Alonso et al. 2011 and Lopez 2005 for L2 Spanish) are a welcome addi-
tion to the development of studies in cognitive grammar in SLA. How-
ever, it must be acknowledged that cognitive grammar will not be the
solution to all the diculties in describing grammar in the L2 classroom,
as teachers may not have the appropriate background to cope with the
conceptual turn required for the adoption of a cognitive approach to
grammar teaching or the terminological complexity of some cognitive
accounts. For this reason, further research needs to investigate how founda-
tional texts on cognitive grammar and current studies within the cognitive
approach can be introduced in the classroom and be made accessible to lan-
guage teachers who are not necessarily experts in SLA (for the relationship
between research and practice, see Ellis 1997; Kumaradivelu 2006).

6. Conclusion

This chapter opened with three main questions that the current volume
has attempted to answer: identifying the most relevant theoretical con-
structs that can contribute to a pan-theoretical perspective, integrating
research methodologies, and, nally, analyzing the results derived from
the theories and methodologies. The answers provided in the current
chapter are not conclusive, although they arguably contribute to a better
understanding of the analysis of the acquisition of temporality in the L2
by way of a comprehensive investigation of the path that research in tense
and aspect has followed, the obstacles we have found on the road, and
the hurdles we may still nd in the future. As can be gathered from this
chapter and the ones in the volume, research in L2 tense and aspect has
developed both in the number of theories and methodologies, a fact that
can be considered a strength or a weakness. On the one hand, it is positive
to have a pluralistic perspective of viewpoints on how tense and aspect are
learned in a second language because it contributes to the development of
the SLA discipline in general. On the other hand, the proliferation of
theories and methodologies can be considered problematic when theories
and methodologies become incommensurable among themselves and
results can only be understood as a sort of consequence of having faith
in such and such theory or method. We have suggested that a middle of
Integrating 445

the road approach, which circumvents the problematic aspects of such


a proliferation of theories and methods, is possible if researchers adopt
rigorous methodologies (which include the provision of explicit theoretical
assumptions, the specication of coding criteria, and the critical analysis
of ndings within a broad perspective) that may combine quantitative and
qualitative methodologies and foster dialogue among researchers involved
in the many faces of the study of tense and aspect.

References

Alonso, Rosario, Alejandro Castaneda, Pablo Martnez, Lourdes Miquel, Jenaro


Ortega & Jose Placido
2011 Gramatica basica del estudiante de espanol. Edicion revisada y
ampliada. Barcelona: Difusion.
Andersen, Roger
1986 The need for native language comparison data in interpreting
second language data. Unpublished manuscript. Forum lecture
1986 TESOL Summer Institute, University of Hawaii.
Andersen, Roger
1994 The primacy of aspect in rst and second language acquisition.
In William Ritchie & Tej J. Bhatia (Eds.), Handbook of second
language acquisition, 527570. San Diego: Academic.
Andersen, Roger & Yasuhiro Shirai
1994 Discourse motivations for some cognitive acquisition principles.
Studies in Second Language Acquisition 16. 133156.
Ayoun, Dalila & M. Rafael Salaberry
2005 Tense and aspect in Romance languages. Amsterdam: John
Benjamins.
Ausubel, David
1968 Educational psychology: A cognitive view. New York: Holt,
Rinehart and Winston.
Bardovi-Harlig, Kathleen
1994 Reverse-order reports and the acquisition of tense: Beyond the
Principle of Chronological Order. Language Learning 44. 243
282.
Bardovi-Harlig, Kathleen
1995 The interaction of pedagogy and natural sequences in the acqui-
sition of tense and aspect. In Fred Eckman et al. (Eds.), Second
language acquisition theory and pedagogy. Hillsdale: Erlbaum.
Bardovi-Harlig, Kathleen
1998 Narrative structure and lexical aspect: Conspiring factors in sec-
ond language acquisition of tense-aspect morphology. Studies in
Second Language Acquisition 20. 471508.
446 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

Bayley, Robert
1996 Competing constraints on variation in the speech of adult
Chinese learners of English. In R. Bayley & D. R. Preston
(Eds.), Second language acquisition and linguistic variation, 97
120. Amsterdam: John Benjamins.
Blyth, Carl
1997 A Constructivist approach to grammar: Teaching teachers to
teach aspect. The Modern Language Journal 81. 5066.
Blyth, Carl
2005 From empirical ndings to the teaching of aspectual disctinc-
tions. In D. Ayoun & M. R. Salaberry (Ed.), Tense and aspect
in Romance languages, 211252. Amsterdam: John Benjamins.
Bruhn de Garavito & Elena Valenzuela
2008 Eventive and stative passives in Spanish L2 acquisition: A matter
of aspect. Bilingualism: Language and Cognition 11. 323336.
Carroll, Mary & Christiane von Stutterheim
2003 Typology and information organisation: perspective taking and
language-specic eects in the construal of events. In A. Giacalone
Ramat (Ed.), Typology and second language acquisition, 365
402. Berlin: Mouton de Gruyter.
Comrie, Bernard
1976 Aspect. Cambridge: Cambridge University Press.
Comajoan, Llorenc
2001 The acquisition of Catalan L2 past morphology: Evidence for the
aspect and discourse hypotheses. Bloomington, IN: Indiana Uni-
versity, Ph.D. dissertation.
Denzin, Norman
2010 Moments, mixed methods, and paradigm dialogs. Qualitative
Inquiry 16. 419427.
Dietricht, Rainer, Wolfgang Klein & Colette Noyau
1995 The acquisition of temporality in a second language. Amsterdam:
Benjamins.
Doiz-Bienzobas, Aintzane
1995 The preterite and the imperfect in Spanish: Past situation versus
past viewpoint. San Diego, CA: University of California-San
Diego, Ph.D. dissertation.
Dornyei, Zoltan
2007 Research methods in applied linguistics. Oxford: Oxford Univer-
sity Press.
Ellis, Rod
1985 Understanding second language acquisition. Oxford: Oxford Uni-
versity Press.
Ellis, Rod
1990 Instructed second language acquisition. Oxford: Blackwell.
Integrating 447

Ellis, Rod
1997 Second language acquisition research and language teaching.
Oxford: Oxford University Press.
Ellis, Rod
2002 Does form-focused instruction aect the acquisition of implicit
knowledge? Studies in Second Language Acquisition 24. 223236.
Giorgi, Alessandra & Fabio Pianesi
1997 Tense and aspect: From semantics to morphosyntax. Oxford:
Oxford University Press.
Gonzalez, Paz
2003 Aspects on aspect: Theory and applications of grammatical aspect
in Spanish. Utrecht, The Netherlands: University of Utrecht,
Ph.D. dissertation.
Hesse-Biber, Sharlene
2010 Emerging methodologies and methods practices in the eld of
mixed methods research. Qualitative Inquiry 16. 415418.
Howe, Kenneth
2004 A critique of experimentalism. Qualitative Inquiry 10. 4261.
Jordan, Geo
2004 Theory construction in second language acquisition. Amsterdam:
John Benjamins.
Klein, Wolfgang
1994 Time in language. London: Routledge.
Kumaravadivelu, Bala
2006 Understanding language teaching: From method to postmethod.
Mahwah: Lawrence Erlbaum.
Labeau, Emmanuelle
2005 Beyond the aspect hypothesis: Tense-aspect development in ad-
vanced L2 French. Bern: Peter Lang.
Laord, Barbara
2007 Second language acquisition reconceptualized? The impact of
Firth & Wagner (1997). The Modern Language Journal 91.
735756.
Langacker, Ronald
1982 Remarks on English aspect. In P. Hopper (Ed.), Tense-aspect:
Between syntax and pragmatics, 265304. Philadelphia: John
Benjamins.
Lardiere, Donna
2003 The comparative fallacy revisited: A reply to Lakshmanan and
Selinker (2001). Second Language Research 19. 129143.
Larsen-Freeman, Diane
1990 On the need for a theory of language teaching. In J. Alatis (Ed.),
Georgetown University Round Table on Languages and Linguistics
1990, 261270. Washington, DC: Georgetown University Press.
448 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

Liskin Gasparro, Judith


1996 Narrative strategies: A case study of developing storytelling
skills by a learner of Spanish. Modern Language Journal 80.
271286.
Long, Michael
1983 Does second language instruction make a dierence? A review of
the research. TESOL Quarterly 17. 359382.
Long, Michael
1988 Instructed interlanguage development. In L. Beebe (Ed.), Issues
in second language acquisition: Multiple perspectives, 115141.
New York: Newbury House.
Lopez, Angel
2005 Gramatica cognitiva para profesores de espanol L2. Madrid: Arco
Libros.
McManus, Kevin
2011 The Development of aspect in a second language. Newcastle,
England: Newcastle University, Ph.D. dissertation.
Montrul, Silvina & Rafael Salaberry
2003 The development of Spanish past tense morphology: Developing
a research agenda. In B. Laord & M. R. Salaberry (Eds), Stud-
ies in Spanish second language acquisition: State of the science,
4773. Washington, DC: Georgetown University Press.
Montrul, Silvina & Roumyana Slabakova
2002 The L2 acquisition of morphosyntactic and semantic properties
of the aspectual tenses preterite and imperfect. In A. T. Perez-
Leroux & J. Munoz Liceras (Eds.), The acquisition of Spanish
morphosyntax, 115151. Dordrecht: Kluwer Publishers.
Putz, Martin, Susanne Niemeier & Rene Dirven (Eds.)
2001 Applied cognitive linguistics I: Theory and language acquisition.
Berlin: Mouton de Gruyter.
Rocca, Sonia
2007 Child second language acquisition: A bi-directional study of
English and Italian tense-aspect morphology. Amsterdam: John
Benjamins.
Rothman, Jason & Michael Iverson
2008 Beyond morphological use: What semantic knowledge tells us
about aspect in L2 Portuguese. Language Acquisition 15(4).
270314.
Rosi, Fabiana
2009 Learning aspect in Italian L2. Corpus annotation, acquisitional
patterns, and connectionist modeling. Milan: FrancoAngeli.
Salaberry, M. Rafael
1999 The development of past tense verbal morphology in classroom
L2 Spanish. Applied Linguistics 20(2). 151178.
Integrating 449

Salaberry, M. Rafael
2000 The development of past tense morphology in L2 Spanish.
Amsterdam/Philadelphia: John Benjamins.
Salaberry, M. Rafael
2008 Marking past tense in second language acquisition: A theoretical
model. London: Continuum Press.
Salaberry, M. Rafael
2011 Assessing the eect of lexical aspect and grounding on the acqui-
sition of L2 Spanish Preterite and Imperfect among L1 English
speakers. Bilingualism: Language and Cognition 14. 184202.
Salaberry, Rafael & Yasuhiro Shirai
2002 Tense-aspect morphology in L2 acquisition. Amsterdam/Philadel-
phia: John Benjamins.
Shirai, Yasuhiro
2004 A multiple-factor account for the form-meaning connections in
the acquisition of tense-aspect morphology. In J. Williams, B.
VanPatten, S. Rott & M. Overstreet (Eds.), Form-meaning con-
nections in second language acquisition, 91112. Mahwah, NJ:
Erlbaum.
Shirai, Yasuhiro
2007 The aspect hypothesis, the comparative fallacy, and the validity
of obligatory context analysis: A reply to Lardiere (2003). Second
Language Research 23. 5164.
Slabakova, Roumyana
2001 Telicity in the second language. Amsterdam: John Benjamins.
Slabakova, Roumyana
2002 Recent research on the acquisition of aspect: An embarrassment
of riches? Second Language Research 18(2). 172188.
Slabakova, Roumyana & Silvina Montrul
2007 L2 acquisition at the grammar-discourse interface: Aspectual
shifts in L2 Spanish. In J. Liceras, H. Zobl & H. Goodluck
(Eds.), Formal features in second language acquisition, 452483.
Mahwah, NJ: Lawrence Erlbaum.
Slagter, Peter Jan
2000 Learning by instructing. Utrecht, The Netherlands: University of
Utrecht, Ph.D. dissertation.
Smith, Carlota
1991 The parameter of aspect. Dordrecht: Kluwer.
Smith, Carlota
1997 The parameter of aspect, 2nd edition. Dordrecht: Kluwer Aca-
demic Press.
Strauss, Anselm & Juliel Corbin
1998 Basics of qualitative research: Techniques and procedures for
developing grounded theory, 2nd edition. Thousand Oaks: Sage.
450 M. Rafael Salaberry, Llorenc Comajoan, and Paz Gonzalez

Tashakkori, Abbas & Charles Teddie


2003 Handbook of mixed methods in social and behavioral research.
Thousand Oaks: Sage.
von Stutterheim, Carol, Mary Carroll & Wolfgang Klein
2009 New perspectives in analyzing aspectual distinctions across
languages. In W. Klein & P. Li (Eds.), The expression of time,
195216. Berlin: Mouton de Gruyter.
Wiberg, Eva
1996 Reference to past events in bilingual Italian-Swedish children of
school age. Linguistics 34. 10871114.
Zyzyk, Eve & Susan Gass
2008 Epilogue: A tale of two copulas. Bilingualism: Language and Cog-
nition 11. 383385.
Author biographies

Dalila Ayoun (Ph.D., University of Florida) is Professor of French Lin-


guistics and Second Language Acquisition and Teaching at the University
of Arizona in Tucson. Her research focuses on the second language acqui-
sition of morphosyntax by adult learners. Her most recent publications
include two edited volumes on French applied linguistics: Studies in French
applied linguistics and French applied linguistics. She is also a coeditor for
the Studies in Bilingualism book series (John Benjamins).
Kathleen Bardovi-Harlig (Ph.D., University of Chicago) is Professor and
Chair of Second Language Studies at Indiana University, Bloomington.
Her research investigates second language acquisition of tense-aspect and
temporal expression through a variety of research frameworks that link
form and meaning, interlanguage pragmatics, and the interface of grammar
and pragmatics, including most recently the role of conventional expressions.
Her work on tense and aspect includes Tense and aspect in second language
acquisition: Form, meaning, and use and has appeared in Language Learning,
Studies in Second Language Acquisition, Modern Language Journal, The
Oxford Handbook of Tense and Aspect, and many edited volumes. She is a
former editor of Language Learning and a former president of AAAL.
Robert Bayley (Ph.D., Stanford University) is Professor of Linguistics at
the University of California, Davis. He has conducted research on variation
in English, Spanish, American Sign Language, and Italian Sign Language
and ethnographic studies on language use in U.S. Latino communities.
His publications include Language as cultural practice: Mexicanos en el
norte (with S. R. Schecter), Sociolinguistic variation: Theories, methods,
and applications (with C. Lucas), and the Oxford handbook of socio-
linguistics (with R. Cameron and C. Lucas).
Llorenc Comajoan (Ph.D. Indiana University, Bloomington) is Associate
Professor at the University of Vic (Spain), where he teaches at the Depart-
ment of Philology and Language and Literature Teaching, and he is a
member of the University Centre for Sociolinguistics and Communication at
the University of Barcelona. He has conducted research in second language
acquisition (tense and aspect), educational sociolinguistics (longitudinal
studies on the learning of Spanish and Catalan), and language teaching
(grammar teaching and the role of motivation). He has published his
research in Language Learning, Hispania, Catalan Review, Caplletra, and
452 Author biographies

Articles de Didactica de la Llengua i la Literatura. He directs the Educa-


tion, Language, and Literature research group at the University of Vic.
Aintzane Doiz (Ph.D., University of California, San Diego) is Associate
Professor at the University of the Basque Country UPV/EHU (Spain),
where she teaches English language and applied semantics in the Depart-
ment of English and German studies, Translation and Interpretation. Her
research interests are cognitive semantics, contrastive linguistics, and applied
linguistics. Her latest research focuses on multilingualism in higher education
and the acquisition of an L3 in CLIL. She is a member of the Language
and Speech research group at the University of the Basque Country.
Nick Ellis (Ph.D., University of Wales) is Professor of Psychology, Pro-
fessor of Linguistics, Research Scientist in the English Language Institute,
and Associated Faculty at the Centre for the Study of Complex Systems
at the University of Michigan. Before his research appointment to the
University of Michigan in 2004, he was Professor of Psychology at the
University of Wales, Bangor. His research interests include language acqui-
sition, cognition, emergentism, corpus linguistics, cognitive linguistics,
psycholinguistics, and applied linguistics. Two recent books on these themes
are: Language as a complex adaptive system (with Diane Larsen-Freeman),
and Handbook of cognitive linguistics and second language acquisition
(with Peter Robinson). He is an adviser to the Oxford University Press
Applied Linguistics series and serves as the General Editor of Language
Learning.
Anna Giacalone Ramat (Ph.D., University of Florence) is Professor Emerita
of Linguistics at the University of Pavia (Italy). She has been Professor of
Linguistics since 1975 and of Sociolinguistics in the 19821991 period. She
is a former President of the International Society for Historical Linguistics,
the Societa Italiana di Glottologia, and the Societas Linguistica Europaea.
Her main areas of research are historical linguistics, grammaticalization
theory, languages in contact, plurilingualism and minority languages, and
second language acquisition.
Paz Gonzalez (Ph.D., Utrecht University, The Netherlands) is Assistant
Professor at Leiden University (The Netherlands), where she teaches Spanish
language and Linguistics at the Department of Latin American Studies,
and she is a member of the Leiden University Centre for Linguistics. Her
research interests are second language acquisition, bilingualism, language
Author biographies 453

variation, and language instruction, all viewed from a crosslinguistic per-


spective. She has published Aspects on aspect: Theory and applications
of grammatical aspect in Spanish and several articles in international
journals.
Susanne Niemeier (Ph.D., University of Duisburg) is Professor of Applied
Linguistics and Foreign Language Teaching Methodology at the English
department of the University Koblenz-Landau, Campus Koblenz (Ger-
many). In 2004 she achieved her Habilitation from the University of
Bremen (Germany). Her main areas of research are applied cognitive
grammar, grammar teaching, the connection between language and cul-
ture, bilingual education, and early English instruction. She is the current
president (2012-present) of the German Cognitive Linguistics Association.
Stefano Rastelli (Ph.D., University of Pavia) is Assistant Professor at the
University of Pavia (Italy). His main area of interest is second language
acquisition. He has published works on the following topics: auxiliary
selection, passive voice, null subjects, aspect, and verb actionality.
Jason Rothman (Ph.D., University of California, Los Angeles) is Associate
Professor at the departments of Spanish and Portuguese Studies and Lin-
guistics at the University of Florida. He is co-executive editor of Linguistic
Approaches to Bilingualism, associate editor of Hispania, and editor of the
book series Issues in Hispanic Linguistics published by John Benjamins.
His areas of research are: the cognitive processes of adult second language
acquisition, adult multilingual acquisition, heritage language acquisition,
and child rst language acquisition. Recent articles have appeared in
Applied Linguistics, Bilingualism: Language and Cognition, International
Journal of Bilingualism, International Review of Applied Linguistics, Second
Language Research, and Studies in Second Language Acquisition, among
other journals.
M. Rafael Salaberry (Ph.D., Cornell University) is Professor of Second
Language Acquisition in the Department of Spanish and Portuguese at
the University of Texas in Austin. His main areas of research are the
acquisition of tense-aspect among adult second language learners, L2
teaching in the university setting, and bilingual education. His latest book
publications are: Language allegiances and bilingualism in the US, Mark-
ing past tense in second language acquisition: A theoretical model, The art
of teaching Spanish (with Barbara Laord), and Tense and aspect in
Romance languages: Theoretical and applied perspectives (with Dalila
Ayoun). He has published his work in Applied Linguistics, Bilingualism,
454 Author biographies

The Canadian Modern Language Review, Language Learning, Language


Learning and Technology, and The Modern Language Journal, among
other journals. He is on the Editorial Board of Language@Internet.
Yasuhiro Shirai (Ph.D., University of California, Los Angeles) is Professor
in the Department of Linguistics, University of Pittsburgh, and he is
currently Invited Scholar at the National Institute of Japanese Language
and Linguistics and Invited Professor from Overseas at Sophia University,
both in Tokyo. His research interests include rst and second language
acquisition of tense-aspect morphology, and cognitive models of language
acquisition/processing. He was a Japan Foundation Fellow (20012002),
is editor of Studies in Language Sciences, an associate editor of First Lan-
guage, and serves on the editorial boards of Studies in Second Language
Acquisition, IRAL, and Journal of Cognitive Science. He is currently Presi-
dent of the Japanese Society for Language Sciences.
Subject index

Accomplishment 4, 18, 60, 68, 98 Boundedness 27, 54, 66, 6870, 82,
100, 123, 132, 137, 160, 176, 177 187, 216
9, 2717, 2834, 287295, 299,
306, 339, 3634, 366, 3778, 381 Coding 45, 23, 77, 81, 834, 125,
Achievement 4, 18, 60, 68, 98100, 165, 171, 211, 260, 268, 271, 280
123, 132, 1345, 137, 142, 160, 2, 285, 2878, 295, 2899, 300,
161, 176, 1789, 201, 2712, 274 32532, 336, 33841, 3535, 358
8, 2834, 287299, 339, 3634, 9, 362, 368, 392, 396, 4156, 424,
366, 3778, 382, 392, 398, 405, 4345, 445
409, 4167, 441 Cognitive grammar 11, 58, 442
Actional pairs 3967, 408, 414 Conceptualization 1214, 29, 53, 58,
Activity 26, 60, 70, 92, 989, 131, 160, 114, 188, 194, 199, 217, 293,
268, 27276, 27779, 281, 283, 296
28690, 29599, 315, 370, 3778, Construal 11, 14, 17, 22, 58, 64, 65,
381, 392, 3979, 405, 408 69, 98, 107, 262, 315, 345, 349,
Actual event 356, 71, 74, 7678, 80 426, 430, 446
2, 198 Construction grammar 2, 52, 90, 106,
Actual occurrence reading 57, 6365, 108, 110, 112, 217
69, 71 Constructions 3, 7, 30, 48, 523, 72,
Actual past situation 79, 84 905, 98, 1012, 104, 1067, 109,
Aktionsart 153, 160, 185, 295, 303, 1114, 117, 129, 207, 2101, 213,
401 215, 389, 422
Aspect hypothesis 6, 98101, 119, Contingency of form-function mapping
137, 15960, 1756, 181, 258, 92, 94, 97, 102
263, 271, 27880, 298, 302, 304
7, 322, 358, 362, 37880, 3823, Default past 119, 159, 175, 177, 178,
392, 398, 40003, 405, 42122, 183, 190, 280, 304
4289, 431, 447, 449 Discourse grounding 263, 30910,
Atelic see telic 3156, 3189, 3214, 326, 328
30, 3401, 345, 392, 424, 430,
Background 3, 5, 1213, 30, 37, 41, 435, 439
44, 47, 667, 82, 103, 129, 137, Discourse hypothesis 45, 101, 119,
193, 197, 213215, 222, 230, 231 180, 258, 309, 350, 353, 355, 428
3, 236, 23940, 2458, 2501, Durative 18, 138, 159, 16170, 175,
253, 2589, 265, 268, 309, 315 17981, 193, 200, 207, 214, 272,
18, 320, 346, 349, 353, 361, 364 294, 317, 320, 342, 363, 391,
7, 36970, 3778, 380, 424, 432 399, 4012, 404, 407, 4178,
3, 4434 431
Basic verbs 393, 4134, 4178, 422 Dynamic 2, 14, 60, 99100, 108, 110
Bounded-unbounded 1722, 248, 34, 1, 1145, 117, 123, 148, 163, 230,
39, 60, 648, 70, 82, 85, 1646, 233, 267, 2713, 276, 2857, 296,
16971, 249, 320, 399, 405, 416, 318, 329, 363, 378, 389, 394, 406,
418 434
456 Subject index

Foreground 5, 1214, 44, 47, 66, 82, 135, 141, 163, 171, 175, 1778,
87, 213, 222, 2303, 236, 23940, 182, 184, 188, 190, 193, 209, 260,
2458, 24953, 256, 259, 265, 265, 271307, 309, 324, 32831,
268, 309, 31520, 346, 349, 353, 339, 3412, 3445, 349, 362, 365,
3646, 370, 3778, 392, 428, 432 370, 372, 376, 378, 3834, 387,
Frequency 1, 3, 6, 21, 51, 8994, 96 391, 392, 394, 398, 4013, 421,
99 10107, 111, 292, 298, 365, 424, 42632, 435, 445, 449
367, 385, 396, 398, 414, 42930, Logistic regression 5, 358, 370, 376,
4334, 441, 443 382

Generative paradigm 145 Mental spaces 402, 501


Grammatical aspect 17, 19, 99, 114, Minimalist 1, 119, 123, 1256, 147, 428
119, 1334, 141, 15962, 165, Mixed methods 418, 425, 436
164, 169, 171, 1745, 17984, Modality 7, 16, 29, 30, 34, 38, 402,
193, 219, 262, 277, 280, 283, 286, 48, 53, 88, 116, 119, 129, 144, 249,
289, 294, 304, 328, 284, 365, 386, 254, 262, 265, 26768, 300, 302,
417, 423, 426, 428, 443, 447, 453 317, 3778

Habitual 3, 24, 36, 767, 1056, 115, Narrative 4, 37, 40, 42, 445, 47, 67,
130, 132, 152, 168, 1723, 188 71, 100, 103, 109, 119, 1312,
90, 1924, 1989, 20002, 2078, 134, 13940, 177, 1845, 215,
2147, 249, 251, 273, 279, 2823, 2203, 22969, 275, 280, 301,
285, 28990, 2956, 317, 320, 303, 304, 30945, 3535, 3602,
402, 424, 431 364, 3679, 3767, 3805, 398,
Heritage learners 14143 405, 4101, 418, 424, 428, 432,
439, 445, 448
Image schema 12 Non-accidental 116, 191, 194, 1967,
Imperfect (IMP) 1889, 192196 2026, 212
Imperfective predicates 60, 64, 68
Imperfectivity 22, 82, 133, 148, 169, Outer/inner aspect 119, 128, 13335,
192, 365 141
Implicational scale 379, 382, 387
Interface hypothesis 12, 1423, 145, Perfective predicates 60, 64, 65, 68,
155, 441 701
Interpretable features 126 Perfectivity 124, 128, 133, 148, 192,
Iterative 12, 235, 1056, 115, 152, 217, 317, 322, 365, 401, 403
188, 190, 198202, 213, 215, 217, Phasal pairs 3967, 400, 408, 4134,
2823, 28990, 2945, 297, 299, 4178
317, 407 Plus principle 1634
Predication eect hypothesis 15960
(Language) prompts 3, 187, 189, 200, Predicational aspect 1604, 167, 171
2026, 2079, 212, 2412, 252, Preterite (PRET) 57, 6371, 7685,
254, 2567 427
Lexical aspect 19, 99101, 10305, Property reading 57, 634, 667, 71,
109111, 116, 119, 1289, 133, 85
Subject index 457

Prototype 7, 12, 15, 3437, 49, 54, 91, Telic-atelic/telicity 7, 66, 98100,
96, 99100, 117, 2756, 27980, 1034, 107, 12330, 132, 138,
301, 306, 310, 343, 388, 403404, 148, 153, 1612, 173, 1768, 184,
422 1889, 1913, 201, 209, 2134,
Prototypicality of meaning 92, 96, 99, 216, 2739, 3201, 324, 327, 329,
102 3634, 370, 380, 3914, 396, 398
Psycholinguistics 110, 113, 114, 155, 401, 4034, 40810, 413, 4168,
302, 452 421, 428, 449
Punctual 18, 247, 168, 193, 202, Temporality 45, 110, 127, 139, 161,
2145, 2725, 279, 2949, 303, 183, 219, 221, 232, 254, 2634,
3167, 320, 327, 3545, 358, 363 267, 307, 316, 322, 3256, 3412,
4, 370, 380, 399, 4156 3456, 389, 401, 420, 422, 443,
444, 446
Radial network 15 Terminative 159, 1634, 16671, 175,
Recency 92, 95, 206 17981, 416
Redundancy 92, 967, 107, 442 Type and token frequency 93, 101

Salience 367, 41, 467, 92, 95, 978, Universal Grammar (UG) 119, 120
1067, 109, 111, 315, 319, 349, 24, 1256, 1289, 14345, 154,
434, 442 204, 432
Semantic entailments 122, 141, 144 Unboundedness 25, 54, 66, 82
Semelfactive 24, 176, 2723, 299, 370, Uninterpretable features 132, 137, 142
399 Universal grammar 119, 14647, 149,
State 1718, 20, 23, 32, 68, 989, 152
123, 132, 1378, 153, 160, 170,
1756, 1789, 181, 24950, 271, VARBRUL 358, 365, 367, 3701,
272, 277, 287, 290, 2956, 305, 376, 382, 389
339, 362, 364, 380, 402, 404, 428 Virtual event 57, 712, 74, 76, 77, 79,
Structural plane 191, 1989 81
Virtuality 76, 82, 856
Task 219260, 280, 336, 341, 366,
395, 424, 432469, 434 Zipfs law 94, 101

You might also like