Professional Documents
Culture Documents
1 䉬 2012
Preview Article
Authors in the process of writing a book explain what motivated them to
write it, summarize its proposed content and indicate what contribution
they think it will make to Applied Linguistics.
Introduction
Though the concept of working memory (WM) can be traced back to as early
as William James’ 1890 demarcation of ‘primary memory’ and ‘secondary
memory’, the most influential model of WM in modern time is that proposed
by Alan Baddeley and Graham Hitch (1974). In this seminal model of WM,
Baddeley and Hitch abandoned the unitary account of short-term memory
(STM) usually characterized in previous models and postulated instead a
multiple-component system which they labeled WM. WM as conceptualized
in this way subsumes a supervisory attentional system (SAS) and two domain
specific slave/buffers (Repovs and Baddeley 2006). The SAS system, that is,
the central executive, is purported to be responsible for various executive
functions such as controlling and allocating attentional resources among
cognitive processes involved in higher level cognition. The phonological loop,
one of the two domain specific slave/buffers, is thought to temporarily store
and hold speech-like information via rehearsal mechanisms while the
visuospatial sketchpad, the other domain-specific slave/buffer, is responsible
for handling visual and spatial information.
Simple as it may appear, this tripartite WM model, since its inception in
1974, has proved to be an extremely useful framework for handling a wide
range of questions regarding higher-level human cognition (Baddeley and
Hitch 2001; Hambrick, Kane, and Engle 2005; Baddeley 2007; 2010). More
importantly, it has served as the catalyst for active research programs in a
whole range of disciplines within cognitive science, making it unequivocally
one of the “100 most influential works in cognitive science” (Conway, Jarrold,
Kane, Miyake, and Towse 2007: Preface).
Baddeley and Hitch’s work on WM is seminal also in the sense that it
triggers considerable enthusiasm towards conceptualizing the WM construct
among many cognitive psychologists with different research perspectives
who subsequently formulate a dozen other models (Miyake and Shah 1999).
These diverse theoretical conceptions of WM have been successfully pooled
together in two collected volumes recently edited by Miyake and colleagues
(Miyake and Shah 1999; Conway et al. 2007). The result of such a research
synthesis has been encouraging in that despite controversies on the source of
individual differences in WM, most current WM models have a lot more in
common concerning the nature of the WM construct. Based on this emerging
consensus, an all-encompassing definition of WM can be derived, which
refers to “those mechanisms or processes that are involved in the control,
regulation, and active maintenance of task-relevant information in the service
of complex cognition” (Miyake and Shah 1999: 450). More importantly, three
characterizations of the WM construct are beyond dispute among these WM
models. First, there is structure in WM (thus rebutting the completely unitary
view). Second, it has limited capacity (though the source of this limited
capacity is still debatable). Third, WM is closely linked to long-term memory
(LTM), thus making it a gateway to LTM.
It is conceivable that such characterizations of WM based on those unified
theories are extremely crucial when the WM construct with its applications
has straddled so widely a whole range of disciplines in cognitive psychology
and cognitive neuroscience (Conway et al. 2007). Such a comprehensive
account has a lot to offer to any inquiry into the role of WM in high-level
cognitive activities. Among the many activities of human cognition, language
is arguably the most complex and most intriguing of all (Gathercole 2007).
Therefore, it is argued that any attempt to investigate the role of WM in
language learning should not ignore these unified characterizations of the
construct of WM. Indeed, numerous studies have looked into the different
The executive component of WM (i.e. the EWM, or the central executive as was
originally conceptualized by Baddeley and colleagues) is the most important
but the least understood component in Baddeley’s early WM model (Baddeley
2000a). Baddeley and his colleagues consider it to play various executive
functions, such as co-ordinating the two slave systems, focusing and switching
attention, and activating representations within long-term memory, but claim
that it is not involved in temporary storage of information (Baddeley and Logie
1999). Later, this storage function is purportedly taken over by a newly
proposed component, namely, the episodic buffer (Baddeley 2000b). As already
discussed in the previous section, the consensus reached by various WM
models has been that all groups agreed that there is this homunculus-like
element and its function is to control and allocate/regulate attentional
resources (Miyake and Shah 1999). What follows is a review of some
conventional procedures to measure the EWM.
The most widely recognized EWM measure is Daneman and Carpenter’s
(1980) reading span task, in which subjects need to read aloud increasingly
longer sequences of sentence, and to further state the plausibility of each
sentence. At the end, they also need to recall the final words of all the sentences
in each sequence (Waters and Caplan 2003). For example, at the two-sentence
level (Level 2), participants may be asked to read aloud the following sentences
(Miyake 2001):
At the end of this trial, participants are expected to recall the two sentence-
final words, abruptly and all. Results from such reading span tasks have
correlated reasonably high with language comprehension scores (Daneman
and Merikle 1996). In essence, Daneman and Carpenter’s reading span tasks
are different from traditional short-term memory tasks in that they essentially
impose dual-task demands on participants (i.e. simultaneous processing and
storage), thus reflecting the contemporary views on models of WM that
require more than a passive storage of target memory items (Miyake and Shah
1999).
Later, Daneman and Green also developed a spoken version of the reading
span task, i.e. the speaking span task that was ‘modeled after the reading span
test in that it taxed processing while simultaneously imposing a storage
task’ (Daneman and Green 1986: 11). The fundamental difference between
the reading span task and the speaking span task is that the former taxes
comprehension processes while the latter taxes production processes.
Besides the reading span task and its variants (the listening span task, the
speaking span task for example), there is also another equally well-known
measure of EWM in cognitive psychology, namely, the operation span task.
In contrast to the reading span task and its variants that reflect a so-called
domain-specific view on WM, the operation span task was devised from a
domain-general view of WM (Turner and Engle 1989). Such a task normally
involves two processes: (a) mathematical operations; and (b) memorizing
words, in which subjects perform some simple mathematical operations while
maintaining words for later recall (Engle 2001). Though correlations have
been found between the two measures in studies by cognitive psychologists
holding a domain-general view on WM (e.g. Turner and Engle 1989; Engle
et al. 1992), it is argued that the reading span tasks and its variants are more
directly relevant for the integrated framework of WM for SLA in that they
are particularly associated with language processing. This issue relating to the
domain-specificity and domain-generality of WM measures will figure again
in the discussion of the integrated framework itself.
these SLA scholars also subscribe to the view that WM plays an important role
in SLA. Such a view is, for a large part, characterized by two theoretical
assumptions derived from the perceivable fundamental difference between
L1 learning and SLA.
On the one hand, some SLA-cum-cognitive-psychology researchers (Nick
Ellis, Brian MacWhinney, and John Williams among others) subscribe to a
connectionist perspective and hold the view that SLA, similar to L1 learning
is in essence a process of learning linguistic sequences or chunks (Ellis 1996).
They further speculate that WM plays an instrumental role in the chunking
process of linguistic sequences, thus placing WM in a critical position for the
acquisition of vocabulary and morpho-syntax (Ellis 1996; Ellis and Sinclair
1996; Williams and Lovatt 2005).
On the other hand, other SLA researchers subscribing to the more
traditional information processing view on language (Harrington 1992;
McLaughlin 1995; Miyake and Friedman 1998), believe that WM plays an
important role in SLA in that SLA processes by contrast are usually
characterized by more controlled processing (as opposed to the more
automatic processing taking place in L1 learning). This kind of controlled
processing, they argue, naturally demands more cognitive resources, thus
relying more on WM. In a similar vein, Skehan (1998; 2002; also see Dörnyei and
Skehan 2003) also postulates that different WM components (together with
other aptitude constructs) are functioning distinctively in different stages or
cognitive processes (such as noticing, pattern recognition, automaticity, etc.) of
SLA, thus putting the construct of WM at the very core of SLA.
Drawn from these two arguments, it can be summarized then that similar
to the close relationship between WM and L1 learning (Gathercole and
Baddeley 1993), a certain kind of theoretical association also exists between
WM and SLA. Indeed, such an assumption has motivated some sporadic but
increasing research studies in the SLA field that are devoted specifically to
investigating the WM effects. Mostly, they have emulated the research
paradigms in cognitive psychology to explore the relationship between
various WM components and different aspects of SLA stages or cognitive
processes. This group of WM/SLA studies can be broadly classified into
six categories depending on the specific L2 activities they have targeted
(see Table 1 for more details). Again, preliminary results from them have
been encouraging in that they do seem to corroborate the assumption that
WM plays an essential role in various SLA stages and processes. Pooled
together, these existing WM/SLA studies are lending further support to the
association between WM and SLA.
However, despite this generally positive picture, I need to point out that,
a closer look at these preliminary WM/SLA studies has also revealed some
intractable issues that deserve more caution on the part of SLA researchers
aspiring for this inquiry (see also Juffs and Harrington 2011 for a similar
discussion). These caveats as well as their potential threats to the SLA field are
discussed below.
Vocabulary Critical for the long-term learning of Involved in interpreting the Service (1992); Cheung (1996)
acquisition phonological forms of new words semantic characteristics
of new words?
Grammar/syntax Instrumental for chunking process Not clear yet Ellis (1996); Ellis and Sinclair (1996);
acquisition of linguistic sequences Williams and Lovatt (2005)
Language Used to maintain a phonological Involved in processing Harrington and Sawyer (1992);
comprehension record that can be consulted syntactic and semantic Berquist (1997); Miyake and
during off-line language information and storing Friedman (1998); Walter (2004);
processing products of processing Leeser (2007)
Language Not clear yet Involved in noticing Mackey et al. (2002; 2010); Sagarra
processing feedback. (2007); Trovimovich et al. (2007)
Written Involved in translating ideas and Involved in planning, Abu-Rabia (2003)
production reading process. translating,
programming, reading
and editing.
Speech Promotes narrative ability at early Not clear yet Fortkamp (1999; 2003); Payne and
production stage; promotes accuracy at later Whitney (2002); O’Brien et al.
stage? (2006; 2007); Guará-Tavares (2008)
The first and most imperative issue relates to the confusing use of the term
‘working memory (or WM)’ in the SLA literature. In this respect, the confusion
emerges from some of these WM/SLA studies when they use the same broad
term of ‘WM’ to actually mean different things, or to be more exact, to mean
different aspects or components of the same construct. Such a practice, it is
further argued, can create formidable difficulties when their results and
findings are to be synthesized for such purposes as conducting a meta-analysis
of WM effects (e.g. Watanabe and Bergsleithner 2006). Of course, a more
detrimental consequence resulting from such inconsistent uses of the three
terms is that it sometimes can lead to methodological pitfalls with WM
measures. For example, the two separate scores of the participants’ PWM and
EWM are combined in some SLA studies (e.g. Mackey et al. 2002; Winke 2005;
Sagarra 2007) to arrive at a so-called ‘composite Z score of WM’ for statistical
analysis. As far as I see it, there are some worrying elements for this practice.
First, such a composite score may fail to reflect the multiple functions of WM
adequately. Second, it may obscure the distinctive functions of the two WM
constructs (i.e. the PWM and the EWM). Intuitively, even when a person has
high PWM it does not necessarily mean that he/she also has high EWM. Last
but not least, such a practice is not common in cognitive psychology either. For
these reasons, it seems that a clear distinction is desirable for the multiple
components of WM (particularly PWM and EWM). This becomes one of the key
tenets of the integrated framework of WM for SLA.
If the first caveat is related to theoretical taxonomy at the conceptual level,
then the second caveat inflicting these previous WM/SLA studies relates to
the methodological considerations adopted by some of them. The first issue
here concerns the domain-specificity vs. domain-generality nature of WM
measures. This becomes a particularly more acute issue when SLA researchers
need to choose among the variant formats of the reading span task from
cognitive psychology. With such a wide variety of choices, they may experience
great difficulties in making decisions for a specific WM measure. Then there is
also a second issue which concerns the scoring procedures of WM span tasks.
With respect to scoring, based on previous literature on WM research, two
scoring measures have been adopted for WM span tasks (Miyake 2001): first,
the so-called ‘maximum set size’ score, in which the participants are given
credit for a set if they repeated correctly all the items in a set. Besides this, there
is also the so-called ‘total performance score’ in which participants are given
credit for all the items they have successfully recalled. But then, what are the
advantages and disadvantages for these two scoring measures? It is fair to say
that most previous WM/SLA studies have not addressed these issues seriously
enough when deciding on the WM measures and scoring procedures to be
implemented (also see Dekeyser and Juffs 2005; Juffs 2006; and Juffs and
Harrington 2011 for a similar argument). In view of this, it is high time to bring
in a kind of standardized guidelines for SLA researchers to select appropriate
WM measures and scoring procedures. This will translate into another key
tenet of the integrated framework of WM for SLA.
and sometimes even helps to make changes in LTM, thus rendering it a gateway
to LTM (Characterization 3). This again will have significant implications for
SLA.
Second, concerning the structure of WM for SLA, it is suggested that in
WM/SLA studies it is better to focus only on those WM components that are
most directly implicated in the SLA process though not ignoring the effects of
other possible WM components. Based on previous WM research in cognitive
psychology regarding its effects on L1 learning and based on preliminary
WM/SLA research regarding its effects on SLA (Table 1), the phonological
WM component (i.e. PWM) and the executive WM component (i.e. EWM)
become the most obvious candidates. Other components, such as the visuo-
spatial sketchpad or the newly proposed component of the episodic buffer, are
not included here because their direct relevance to SLA has not been firmly
established. For these reasons, I propose in the framework that the structure of
WM for SLA mainly involves the phonological WM (PWM) and the executive
WM (EWM).
Third, concerning measuring WM for SLA, it is suggested that the
nonword repetition span task be used for measuring the PWM component
and the complex memory span tasks (the reading span task and its variants)
for measuring the EWM component. The rationale for this is again based on
insights from WM research in cognitive psychology regarding WM measures.
For measuring the PWM, it is generally believed that the nonword repetition
span task is more appropriate than the nonword recognition span task in that
the former engages both purported mechanisms of the construct to a larger
extent (i.e. the phonological store and the articulatory rehearsal) while the
latter fails to represent the articulatory rehearsal function (Gathercole 2006).
On the other hand, it is also suggested that the complex memory span task is
a suitable measure for the EWM in that it purportedly taxes the dual functions
of storage and processing that are subsumed by the executive component of
WM (Jaeggi, Buschkuehl, Perrig, and Beat 2010).
At this juncture, it is hoped that the above integrated framework of WM
for SLA have taken into account of the theoretical advances in WM research
from cognitive psychology and also successfully incorporated results and
findings from preliminary WM/SLA studies. Provided that such a principled
approach to theorizing and measuring WM for SLA has been firmly established
and the main engaging WM constructs clearly identified together with their
measurement procedures, the next issue becomes how to apply this framework
in actual WM/SLA research. To facilitate future studies to probe the WM effects
in SLA, I further suggest some general principles that hopefully can serve as a
starting point for designing empirical studies to investigate the WM/SLA
nexus.
That is to say, the two WM constructs of PWM and EWM identified earlier in the
integrated framework should be treated separately in future WM/SLA studies.
Building on this principle, it can be further argued that a clear demarcation is
desirable for the three closely related terms of PWM, EWM and WM. In this
regard, it is further proposed that the term PWM should be used to refer to the
phonological aspect of WM only, and EWM for the executive control aspect of
WM, while WM be used as the umbrella term covering all these specific
components constituting the whole cognitive construct.
That is to say, when a WM/SLA study is set out to investigate the effects of WM
on a certain L2 sub-skill (e.g. listening, speaking, reading or writing), then it is
better to construct the corresponding domain-specific complex memory span
tasks for measuring the EWM, for example, a listening span task, a speaking
span task, a reading span task or a writing span task. Through this procedure,
it can be expected that the WM measure will tax the underlying construct more
closely. Applying this principle, it is not difficult to predict why some previous
WM/SLA studies should have failed to record significant effects of WM on
specific L2 activities; for example, Payne and Whitney (2002) and Mizera (2006)
used the reading span task to investigate L2 speech.
begin speaking immediately after they finished watching the video prompt.
Third, the independent variable of task structure was operationalized as
whether the narrative had a clear story plot line (Skehan and Foster 1999), that
is, the structured narrative task had a clear or tight story plot line and the
unstructured task did not have a clear story plot line.
The dependent variables (DV) of the study were eight L2 speech
performance measures adopted from well-established L2 task-based speech
planning research (Foster et al. 2000; Ellis and Barkhuizen 2005; Skehan 2009).
These include words per minute (speech rate), reformulation, false start,
replacement, conventional global accuracy measure (Accuracy Type I), length
of clauses that are produced accurately (Accuracy Type II) (Tavakoli and Skehan
2005), syntactic complexity (Complex) and lexical diversity (D) (Malvern and
Richards 2002; Skehan 2009). In terms of research design, the study adopted a
psychometric between-groups design that has been commonly used in the
cognitive psychology research of WM (see Swets, Desmet, Hambrick, and
Ferreira 2007). Speech samples from 40 participants entered the empirical
study, with 10 participants in each planning condition. All 40 participants
were administered both WM tasks (the nonword repetition span task and the
speaking span task). Subsequently, several rounds of statistical procedures
were conducted to demystify both the main effects and the interaction
effects of both PWM and EWM on L2 task-based speech planning and
performance.
As regards the results and findings of the empirical study, the main effects
of pre-task planning and task structure were both recorded significantly. More
intriguing were the results and findings with the differential effects of the two
WM constructs. Specifically, it was found that PWM particularly advantaged
some speed fluency measures (speech rate and replacement) and the more
conventional measure of global accuracy (Accuracy Type I). EWM, in slight
contrast, seemed to have affected more on repair fluency measures (e.g. false
start and reformulation), and the length of clauses that could be produced
accurately (Accuracy Type II). Finally, it was also found that both PWM and
EWM were sensitive to both the syntactic complexity and the lexical density
(D) measures of the L2 learners’ oral speech. The study thus reiterated that
due to their distinctive effects on different aspects of L2 task-based speech
planning and performance, the two WM constructs (PWM and EWM) should
be distinguished in future WM/SLA studies.
Based on these findings concerning the relationship between the two
WM constructs (PWM and EWM) and the various measures indexing the
three different dimensions of L2 speech (fluency, accuracy and complexity),
theoretical links between them could be forged correspondingly (as shown in
Figure 1). These hypothetical links in turn gave rise to a series of theoretical
proposals for understanding and predicting L2 learners’ speech performance
under different planning conditions and/or task features (Table 2).
Despite their rudimentary and sketchy nature, these theoretical links
clearly demonstrate the distinctive effects of the two WM constructs on
Conclusion
At this juncture, it is hoped that the empirical study reported in the book has
been able to demonstrate to some extent that the integrated framework of WM
for SLA thus proposed has a lot to offer to applied linguistics. The final
chapters of the book are devoted to discussing specifically how such a
principled approach can shed light on further studies exploring various
aspects of the WM-SLA nexus. For example, building on the empirical study
reported in the book, more research can be conducted to further demystify the
interaction effects of WM constructs (e.g. PWM and EWM) with other
planning conditions (such as task repetition, within-task or on-line planning,
post-task planning, etc.) or other task design features (Tavakoli 2009), on L2
task performance. Equally interesting should be studies identifying the
so-called threshold effects of WM (e.g. Sunderman and Kroll 2009), on various
Acknowledgements
I wish to thank Professor Peter Skehan for his supervision of the empirical study
reported here and for his Foreword written for the book. I am also greatly indebted
to the two anonymous reviewers of this journal for their constructive comments.
Needless to say, I am solely responsible for all remaining errors and shortcomings of
this preview article.
Note
Zhisheng Wen’s book Working memory and second language learning: An integrated
framework for theory and research will be published by Multilingual Matters in late 2011.
References
— and A. D. Baddeley (1993) Working Memory and Language. Hove: Lawrence Erlbaum
Associates.
— C. S. Willis, A. Baddeley and H. Emslie (1994) The children’s test of nonword
repetition: A test of phonological working memory. Memory 2.2: 103–27.
Guará-Tavares, M. G. (2008) Pre-task planning, working memory capacity and L2
speech performance. PhD thesis. Universidade Federal de Santa Catarina, Brazil.
Hambrick, D. Z., M. J. Kane, and R. W. Engle (2005) The role of working memory in
higher-level cognition. In R. Sternberg and J. E. Pretz (eds.), Cognition and
intelligence: identifying the mechanisms of the mind. Cambridge: Cambridge
University Press. 104–21.
Harrington M. (1992) Working memory capacity as a constraint on L2 development. In
R. J. Harris (ed.), Cognitive processing in bilinguals. Amsterdam: North Holland.
123–35.
— and M. Sawyer (1992) L2 working memory capacity and L2 reading skill. Studies in
Second Language Acquisition 14.1: 25–38.
Jaeggi, S. M., M. Buschkuehl, W. J. Perrig and M. Beat (2010) The concurrent validity of
the N-back task as a working memory measure. Memory 18.4: 394–412.
James, W. (1890). Principles of psychology, Vol. 1 (1902 edn.). London: McMillan &
Company.
Juffs, A. (2006) Working memory, second language acquisition and low-educated
second language and literacy learners. LOT Occasional Papers: Netherlands
Graduate School of Linguistics 89–104.
— and M. Harrington (2011). Aspects of working memory in L2 learning. Language
Teaching 44.1: 137–66.
Leeser, M. J. (2007). Learner based factors in L2 reading comprehension and
processing grammatical form: topic familiarity and working memory. Language
Learning 57.2: 229–70.
Kormos, J. and A. Sáfár (2008) Phonological short term-memory, working memory and
foreign language performance in intensive language learning. Bilingualism:
Language and Cognition 11.2: 261–71.
Mackey, A., J. Philp, T. Egi, A. Fujii and T. Tatsumi (2002) Individual differences in
working memory, noticing of interactional feedback and L2 development. In
P. Robinson (ed.), Individual differences and second language instruction. Philadelphia:
Benjamins. 181–201.
— R. Adams, C. Stafford and P. Winke (2010) Exploring the relationship between
modified output and working memory capacity. Language Learning 60.3: 501–33.
Malvern, D. D. and B. J. Richards (2002) Investigating accommodation in language
proficiency interviews using a new measure of lexical diversity. Language Testing
19.1: 85–104.
McLaughlin, B. (1995) Aptitude from an information processing perspective. Language
Testing 11.3: 364–81.
Mizera, G. J. (2006) Working memory and L2 oral fluency. PhD dissertation. University of
Pittsburg, PA.
Miyake, A. (2001) Individual differences in working memory: Introduction to the
special section. Journal of Experimental Psychology: General 130.2: 163–8.
— and N. P. Friedman (1998) Individual differences in second language proficiency:
working memory as language aptitude. In A. F. Healey and L. J. Bourne (eds.),
Foreign language learning: psycholinguistic studies on training and retention. Mahwah,
NJ: Lawrence Erlbaum. 339–64.
— and P. Shah (1999) Models of working memory: mechanisms of active maintenance and
executive control. New York: Cambridge University Press.
O’Brien, I., N. Segalowitz, J. Collentine, and B. Freed (2006) Phonological memory and
lexical, narrative, and grammatical skills in second language oral production by
adult learners. Applied Psycholinguistics 27.3: 377–402.
O’Brien, I., N. Segalowitz, J. Collentine, and B. Freed (2007) Phonological memory
predicts second language oral fluency gains in adults. Studies in Second Language
Acquisition 29.5: 557–82.
Payne, J. S. and P. J. Whitney (2002) Developing L2 oral proficiency through
synchronous CMC: Output, working memory, and interlanguage development.
CALICO Journal 20.1: 7–32.
Rai, K. M., L. C. Loschky, R. J. Harris, N. R. Peck, and L. G. Cook (2011) Effects of stress
and working memory capacity on foreign language readers’ inferential processing
during comprehension. Language Learning 61.1: 187–218.
Repovs, G. and A. D. Baddeley (2006) The multi-component model of working
memory: explorations in experimental cognitive psychology. Neuroscience 139.1:
5–21.
Robinson, P. (2005). Aptitude and second language acquisition. Annual Review of
Applied Linguistics 25.1: 46–73.
Sagarra, N. (2007) From CALL to face-to-face interaction: The effect of computer-
delivered recasts and working memory on L2 development. In A. Mackey (ed.),
Conversational interaction in second language acquisition: a series of empirical studies.
Oxford: Oxford University Press. 229–48.
Sawyer M. and L. Ranta (2001) Aptitude, individual differences, and instructional
design. In P. Robinson (ed.). Cognition and instructed second language learning.
Oxford: Oxford University Press. 319–53.
Service, E. (1992) Phonology, working memory and foreign-language learning.
Quarterly Journal of Experimental Psychology, A 45.1: 21–50.
Skehan, P. (1998) A cognitive approach to language learning. Oxford: Oxford University
Press.
— (2002) Theorising and updating aptitude. In P. Robinson (ed.). Individual differences
and instructed language learning. Amsterdam/Philadelphia: John Benjamins. 69–
94.
— (2009) Modelling second language performance: integrating complexity, accuracy,
fluency and lexis. Applied Linguistics 30.4: 510–32.
— and P. Foster (1999) The influence of task structure and processing conditions on
narrative retellings. Language Learning 49.1: 93–120.
Sunderman G. and J. F. Kroll (2009) When study-abroad experience fails to deliver: the
internal resources threshold effect. Applied Psycholinguistics 30: 79–99.
Swets, B., T. Desmet, D. Hambrick, and F. Ferreira (2007) The role of working memory
in syntactic ambiguity resolution: a psychometric approach. Journal of Experimental
Psychology: General 136.1: 64–81.
Tavakoli, P. (2009) Assessing L2 task performance: understanding the effects of task
design. System 37.3: 482–95.
— and P. Skehan (2005) Strategic planning, task structure, and performance testing. In
R. Ellis (ed.), Planning and task-based performance in a second language. Amsterdam:
John Benjamins. 239–73.
Trovimovich P., A. Ammar, and E. Gatbonton (2007) How effective are recasts? The
role of attention, memory, and analytical ability. In A. Mackey (ed.), Conversational