Professional Documents
Culture Documents
Austronesian Languages
D.Victoria Rau, Chun-Chieh Wang, and Hui-Huan Ann Chang
NATIONAL CHUNG CHENG UNIVERSITY, TAIWAN
framed languages or prefer using path verbs, they claimed that Proto-Austronesian was
probably a path-salient language. To corroborate Huang and Tanangkingsings claim, it is
necessary to find another Austronesian language with sufficient language documentation
to both conduct a comparative study and fine-tune the current methodology for applica-
tion to future investigation of motion events. Although the Frog stories (Slobin 1996,
1997, zaliskan and Slobin 1999) used in Huang and Tanangkingsing (2005) comprised
35 narratives totaling 230 minutes and 8 seconds of data (Huang and Tanangkingsing
2005:312), a good corpus in an Austronesian language with representative genres can also
provide rich data on the expression of motion events. In addition, a wider range of motion
events, not restricted to the Frog stories, can provide a solid foundation to sharpen the tool
for conducting comparative research.
This paper aims to provide quantitative evidence from Yami (Rau and Dong 2006) to
provide an in-depth analysis of the characteristics of another Austronesian language.
Yami, spoken on Orchid Island in Taiwan, has long been recognized as a sister of
Itbayaten, Ivatan, and the other Bashiic languages, possibly as a result of back-migration
from the Philippines within the last millennium or so (Reid 2009).
Similar to other Philippine languages (Reid and Liao 2004), the Yami case marking of
the regular transitive and intransitive verbs (table 1) is ergative in that the A is marked
with the genitive case and the S and O are marked with the nominative case. Yami intran-
sitive verbs include dynamic verbs with the traditionally called AF (Agent Focus) affixes
(-om-, mi-, ma-, maN-, maka-, maci-), stative ma- verbs, and involuntary ka--an verbs.2
2. The abbreviations of the morpheme glosses not found in, or differing from. the Leipzig Gloss-
ing Rules are as follows: AF, Agent focus; Co-, company; CON, conjunction; H, hesitation
marker; IF, Instrumental focus; LF, Locative focus; LIN, linker; NF, nominal affix; ORD, ordinal
number; P, plural; PA, perfective aspect; PAR, particle; PF, Patient focus; PLN, place name; PN,
personal name; RED, reduplication; SUB, subjunctive; SV, stative verb; TOP, topic linker; VF,
verbal affix.
INVESTIGATING MOTION EVENTS IN AUSTRONESIAN LANGUAGES 3
3. The twenty texts can be obtained either from http://yamiproject.cs.pu.edu.tw/yami or Rau and
Dong (2006).
4 OCEANIC LINGUISTICS, VOL. 51, NO. 1
The verb rolled in (1) is a manner verb in the context, describing how the pencil moved
down from the table. However, the verb blew is a cause verb in the context, which
implies the pencil is moved down from the table by some other cause instead of moving
in that manner by itself.
In this study, a motion event is determined by the potential realization of its four com-
ponents: Figure, Ground, Path, and Motion. Therefore, we focused only on verbs that
show the trajectory of the figure with a potential ground that can be inferred from the
motion. In other words, even though it is possible to identify prototypical motion verbs,
coding of motion events can only be properly determined within clauses, similar to Hop-
per and Thompsons coding of semantic transitivity (1980).
Our distinction between path and manner verbs or V-languages and S-languages also
follows Talmy (2000). As mentioned before, in V-languages, path is encoded in the main
verb while manner can be introduced outside the verb, such as Spanish entrar, salir,
subir, and bajar, meaning go in, go out, go up, and go down without any coding of
manner. In S-languages, the manner information is encoded within the verb itself, while
the path information is encoded in the satellite, such as The dog ran (manner verb) across
(path satellite) the street in English. One justification for recognizing the satellite as a
grammatical category is that, for V-languages, it is the characteristic site for the expres-
sion of the core schema (Talmy 2000:102).
S-languages and V-languages have distinct narrative styles. According to Berman and
Slobin (1994:11819), S-languages tend to have greater specification of manner but also
allow for detailed description of path within a clause. The manner tends to be encoded in the
verb, with the path incorporated and encoded by a particle. The satellite also tends to be
expressed by a prepositional phrase. In V-languages, path and manner are elaborated in sep-
arate clauses. The manner is added as a separate adverbial phrase/clause. If path and manner
occur in the same clause, the manner is generally optional and more compact in form.
Slobin (2004) claimed that in addition to V-languages and S-languages, a third type of
languages should be distinguished in motion events, the so-called equipollently-framed
languages (E-languages), where path and manner are expressed by equivalent grammati-
cal forms, including serial, bipartite, and generic verbs. Based on Talmy and Slobins
research, Huang and Tanangkingsing (2005) identified four distinct patterns to encode
path and manner verbs (constituent order ignored), as in (2):4
(2) A. Satellite framed language: MANNER VERB + PATH SATELLITE
B. Verb framed language: PATH VERB + MANNER ADJUNCT
C. Macroevent language: [MANNER PREFIX + PATH ROOT] VERB
D. Serial-verb language: PATH VERB # MANNER VERB
or MANNER VERB # PATH VERB
Satellite- and verb-framed languages have been illustrated with English and Spanish,
respectively. A macroevent language (2C) can be illustrated with Tsou, where the path
verb root is combined with a manner prefix, as in (3).
4. The coding symbols + and # follow Huang and Tanangkingsing (2005). The symbol + in X+Y
indicates the constituent order of X and Y should be ignored, while the symbol # in X#Y indi-
cates X precedes Y.
INVESTIGATING MOTION EVENTS IN AUSTRONESIAN LANGUAGES 5
ing trajectory of the moved object and express Path information implicitly. This decision
also followed Huang and Tanangkingsings coding (2005).
In summary, following Talmy (2000)s definition, a motion event should contain the
four components Figure, Motion, Path, and Ground. However, not all four components
must be explicitly encoded in each clause before a motion event can be determined.
Sometimes, the Figure or the Ground may not be expressed in the clause to be consid-
ered, but can be interpreted anaphorically from the previous context. Therefore, a motion
verb is determined by its potential realization of the four components in a clause.
* The pattern M#P#D is illustrated in the Chinese example (4). The former M means a
manner verb, middle P is a path verb, and latter D is deixis.
10 OCEANIC LINGUISTICS, VOL. 51, NO. 1
style it is 16. As a preliminary observation, we noticed that the path verbs are much more
frequent than either manner verbs or serial verbs M#P or P#M in the Frog story.
In tables 3 and 4, the percentages of path expressions and manner expressions in
seven Austronesian languages are presented. The percentages in Yami were calculated
by combining the written with the spoken forms. Thus, the percentages are not the same
as shown in table 2. The Path expressions included Path verbs, MP, M#P, P#M, M=P and
M#P#D patterns, as categorized in Huang and Tanangkingsing (2005). The six patterns
all contain a Path component with the manner verbs excluded. The total number of the
Yami path expressions in table 3 is 29 (72.5 percent), with 25 Path verbs (62.5 percent), 1
P#M (2.5 percent), and 3 M#P (7.5 percent).
Table 4 presents manner expressions, including all six types with a manner component,
that is, Manner verbs, MP, M#P P#M, M=P and M#P#D. The path verbs were excluded
from the table. The total number of the Yami manner expression in table 4 is 15 (37.5 per-
cent), with 11 Manner verbs (27.5 percent), 1 P#M (2.5 percent) and 3 M#P (7.5 percent).
Comparing Yami with the six WAN languages of Huang and Tanangkingsings study
(2005), we find that Yami is a path-salient language based on the very small data set from
the Frog story. However, the limited tokens with skewed distribution from the seven lan-
guages do not allow us to generalize to any of these languages, let alone to all Austronesian
languages. Therefore, our next step was to code motion events from a Yami corpus with
texts on various topics to investigate whether Yami is really path-salient.
fully selected from texts collected by Rau and Dong since 1995. The 20 texts were gath-
ered from 4 males and 5 females in Yayo and Iranomilek villages. The transcription and
annotation of the texts are very reliable, as they are based on over a decade of collabora-
tive work between a linguist and a Yami language consultant.
The dependent and independent variables used for the logistic regression to account
for the variation of Yami path and manner verbs are as follows:
(20) Dependent Variable:
FG1: Production of path and manner verbs
1 = path verb
0 = manner verb
Independent Variable:
FG2: Figure
f = Figure marking
z = Non-figure marking
FG3: Ground
d = Locative marker do and deictic dang there
n = Personal pronoun Locatives (e.g., jimo 2S.LOC)
z = No
As mentioned earlier, the basic motion event consists of the Motion and Path of the
moving FIGURE to the located object, that is, GROUND (Talmy 2000: 25). We would
like to know (i) what type of motion verb is most frequent in Yami, and (ii) to what extent
FIGURE and GROUND can account for the choice between path and manner verbs.
Based on our preliminary observations, we noted that path verbs were more frequent than
manner verbs in the Yami Frog story. In addition, the Ground tends to be marked by the
locative case marker do. We wonder if the nature of the location might trigger the use of
different motion event, as proposed by Iacobini & Vergaro (2010).5 However, the deter-
mination of Yami semantic categorization of GROUND is a topic beyond the scope of
this research. We decided to limit our study to the encoding of GROUND by different
forms. Thus, this quantitative study tested the following hypotheses:
(i) Yami is a path salient language because the use of path verbs is more frequent than
manner verbs.
(ii) The occurrence of FIGURE and GROUND is related to the choice between path
and manner verbs.
(iii) The encoding of GROUND is related to the choice between path verbs and manner
verbs.
5.1 DATA AND CODING. The study focused only on clauses that contain motion
events (that is, path verbs or motion verbs).6 In other words, all the non-SVC clauses that
are motion events were labeled either Path or Manner and included in the analysis. Table
5 summarizes the distribution of the four constructions of motion events we identified
5. There are many culturally different ways to categorize the GROUND. For example, Nikitina
(2008) divided GROUND into two categories: (1) containers locations with a well-
defined boundary (for example, rooms, boxes, buildings), and (2) areaslocation that lacks
a clear boundary (for example, forests, sea).
6. Constructions with other verbs were excluded.
12 OCEANIC LINGUISTICS, VOL. 51, NO. 1
among Yami narratives. The total number of tokens is 709 clauses. We decided to focus
only on the 339 path verbs and 282 manner verbs, excluding the 59 occurrences of go
and 29 serial verbs with the linker a from the VARBRUL analysis. Strictly speaking, the
grammaticalized go verb can be analyzed as a type of SVC with the linker a optionally
deleted. Although it was suggested by one anonymous reviewer that the grammaticalized
go be included in the analysis as the motion event denoted by the second verb may still
be assigned either Manner or Path, the inclusion of the limited tokens of SVCs would
complicate the current research design without much gain. Future studies need to investi-
gate motion events in SVCs separately.
Some might think it better to run a VARBRUL analysis with all the excluded catego-
ries (that is., go verbs, serial verbs, and nonprototypical path or manner verbs) included
for an additional full VARBUL analysis. But this would not meet the requirement of a
logistic regression, where confounding variables should be identified, controlled, and/or
excluded before any VARBRUL runs to avoid interpretation problems.
5.2 RESULTS. After the first VARBRUL run was performed, there was no knock-
out group, and a binomial up-and-down run for path/manner verb application suggested
all groups are significant factors. The results of numbers and percentages of the construc-
tions of motion events in Yami are shown in table 6.
As indicated in table 6, path verb tokens constitute 55 percent of the total 621 tokens,
while manner verb tokens constitute 45 percent. Path verbs with Figure marking (56 per-
cent) are slightly higher than manner verb with Figure marking (44 percent). As for the
* The second verb in this example happens to be a deictic verb functioning only to indicate
whether the motion is toward or away from a person. But as shown in examples (14) and
(15), the two verbs in SVCs can have either a Manner or Path assignment.
Ground group, path verbs tend to cooccur with Locative marker do (70 percent) more
frequently than do the manner verbs (30 percent).
The final inferential statistics from the VARBRUL analysis are presented in table 7,
which shows that the input probability of path verbs has a value of 0.55. This means that
our first hypothesis that Yami is a path-salient language can be supported, but note that
path verbs are only slightly more frequent than manner verbs (55 percent vs. 45 percent).
As the total Chi-square (goodness-of-fit) has a value of 3.19, less than 7.82 (df = 3, p <
0.05), it indicates the two independent variables, Figure and Ground, are independent
from each other. Furthermore, both Figure and Ground account for the choice between
path and manner verbs. This result supports our second hypothesis, confirming that Figure
and Ground are two significant factor groups that form the most parsimonious model to
account for the variation between path and manner verbs.
We can further interpret the VARBRUL weights (values) from table 6 to find out the
influence of the factors.7 Overall, Ground has a slightly higher impact (Range = 22) on
the choice of path verbs than does Figure (Range = 8). While explicit Figure marking
only slightly favors path verb production, with a value of 0.52, zero Figure marking
favors manner verbs. Furthermore, location marker do or deictic dang there as Ground
marking favors the path verb production with a value of 0.66. On the other hand, Ground
with personal pronoun locatives such as jimo toward you and zero GROUND marking
slightly disfavor the production of path verbs (Pi = 0.46 and Pi = 044, respectively).
Alternatively, it also means that manner verbs are slightly favored in pronominal and zero
Ground marking. Overall, these results also support the third hypothesis, confirming the
importance of Ground marking with a locative marker do or deictic dang to account for
the choice of path verbs.
From the quantitative analysis, we confirm that Yami is a path-salient language,
although path verbs are only slightly more frequent than manner verbs. Most of the patterns
in motion events (87.5 percent) are either marked with pure path verbs or pure manner
verbs. When path verbs are used, they almost never cooccur with manner. When manner
follows a path verb, it is usually the second verb of a serial verb construction.8 Contrary to a
7. Note that there is a standard formula to interpret the VARBRUL weights. For each factor, there
is a value/weight ranging from 0.00 to 1.00. VARBRUL factor values of more than 0.5 indicate
a favorite factor while values of less than 0.5 signify a disfavorite factor. A value of 0.5
denotes that the factor has no significant effect on the choice between path and manner verbs.
8. A serial verb construction can either involve or not involve grammaticalized go, as demon-
strated in examples (14) and (19), respectively. There are also SVCs in Yami not involving
motion events, for example Ya mateneng a maganam o mavakes ito That woman is good at
dancing. As they do not pertain to motion events, no further discussion is provided.
14 OCEANIC LINGUISTICS, VOL. 51, NO. 1
REFERENCES
Berman, Ruth A. and Dan I. Slobin. 1994. Relative events in narrative: A crosslinguistic
developmental study. Hillsdale, NJ: Lawrence Erlbaum Associates.
Hopper, Paul J. and Sandra A. Thompson. 1980. Transitivity in grammar and discourse.
Language 56(2):25199.
Huang, Shuanfan. 2002. Tsou is different: A cognitive perspective on language, emotion,
and body. Cognitive Linguistics 13(2): 16786.
Huang, Shuanfan, and Michael Tanangkingsing. 2005. Reference to motion events
in six western Austronesian languages: Toward a semantic typology. Oceanic
Linguistics 44: 30740.
Huang, Shuping. 2011. Linguistic representation of macro-events: Lexical prefixes in
Isbukun Bunun. Paper presented at the NTU Workshop on discourse and grammar
in Formosan languages. Taiwan, Taipei: National Taiwan University. January 8.
Iacobini, Claudio, and Carla Vergaro. 2010. The role of inference in the directional reading
of manner of motion verbs: evidence from Italian and English. Paper presented at the
4th International Conference on Intercultural Pragmatics, Madrid.
Johnson, Mark. 1987. The body in the mind: The bodily basis of meaning, imagination,
and reason. Chicago: University of Chicago Press.
Nikitina, Tatiana. 2008. Pragmatic factors and variation in the expression of spatial
goals: The case of into vs. in. In Syntax and semantics of spatial P, ed. by Anna
Asbury, Jakub Dotlail, Berit Gehrke, and Rick Nouwen, 175209. Amsterdam:
John Benjamins.
zalkan, eyda, and Dan I. Slobin. 1999. Learning how to search for the frog:
Expression of manner of motion in English, Spanish, and Turkish. In Proceedings
of the 23rd Annual Boston University Conference on Language Development, vol.
2, ed. by Annabell Greenhill, Heather Littlefield, and Cheryl Tano, 54152.
Somerville, MA: Cascadilla Press.
Rau, D. Victoria and Maa-Neu Dong. 2005. Yami reduplication. Concentric: Studies in
Linguistics 31(2):5787.
. 2006. Yami texts with reference grammar and dictionary. Language and Linguistics
Monograph A-10, Institute of Linguistics, Academia Sinica, Taipei.
Reid, Lawrence A. 2009. The reconstruction of a dual pronoun to Proto Malayo-
Polynesian. In Discovering history through language: Papers in honour of
Malcolm Ross, ed. by Bethwyn Evans, 46177. Canberra: Pacific Linguistics.
Reid, Lawrence A. and Hsiu-Chuan Liao. 2004. A brief syntactic typology of Philip-
pine languages. Language and Linguistics 5(2):43390.
Sankoff, David, Sali Tagliamonte, and Eric Smith. 2005. Goldvarb X: A variable rule
application for Macintosh and Windows. Department of Linguistics, University of
Toronto. http://individual.utoronto.ca/tagliamonte/goldvarb.htm
Slobin, Dan. 1996. Two ways to travel: Verbs of motion in English and Spanish. In
Grammatical constructions: Their form and meaning, ed. by Masayoshi Shibatani
and Sandra A. Thompson, 195220. Oxford: Clarendon Press.
. 1997. Mind, code, and text. In Essays on language function and language
type, ed. by Joan Bybee, John Haiman, and Sandra Thompson, 43767.
Amsterdam: Benjamins.
. 2004. The many ways to search for a frog: Linguistic typology and the expres-
sion of motion events. In Relating events in narrative: Typological and contextual
perspectives, ed. by Sven Stromqvist and Ludo Verhoeven, 21957. Mahwah, NJ:
Lawrence Erlbaum Associates.
Talmy, Leonard. 2000. Toward a cognitive semantics. 2 vols. Cambridge, MA: MIT Press.