Does word order influence non-verbal event description by speakers of OS language? Hiromu Sakai1, Takuya Kubo1, Hajime Ono2, Manami Sato1, Masatoshi Koizumi3 1Hiroshima
University, 2Kinki University, 3Tohoku University author:
[email protected]
1Corresponding
Abstract Although several studies examined non-verbal event description task (gesture without speech) have shown that the natural order of event description is ActorPatient-Action order, the languages investigated so far are limited to SO languages in which the subject NP (Actor) precedes the object NP (Patient) (e.g., English, Spanish, Turkish…). →We conducted a gesture production task in a VOS language and argue that Actor-Patient order is preferred regardless of the word order of languages.
Does word order influence Prior Studies: NO ([1], [2])
non-verbal event description ?
• To describe a simple transitive event using gesture without speech, Actor-Patient-Action order is predominantly produced even though the basic word order of their native language is SVO (e.g., English, Chinese, Italy, Spanish). → Actor-Patient-Action order is the natural order of event description independent of their native language.
However, since these studies only tested SO languages in which the subject precedes the object, it is unclear if the Actor-Patient preference is due to the SO word order. We conducted a gesture production task in a VOS language (Kaqchikel) and an SOV language (Japanese) to verify the effect of word order on nonverbal event description.
About Kaqchikel
Prediction
p One of the Mayan languages spoken in Guatemala. p Constituent order ・Canonical: VOS [3] ・Most frequent: SVO ・Limited: VSO [example] X-∅-u-ch’öy ri ak’wal ri xtän COM-erg.3s-abs.3s-slap the boy the girl The girl slapped the boy.’
Experiments
If the natural order of non-verbal event description … p is completely independent of word order, → Actor-Patient-Action order would be most frequently produced in both Kaqchikel (VOS) and Japanese (SOV). p is influenced by word order, → more Patient-Actor order would be produced in Kaqchikel speakers than Japanese speakers.
Results
Materials used in the task
① Order of Actor and Patient ■ Kaqchikel (N=32)
p Pictures manipulating the animacy of patient
■ Japanese (N=10)
100%
100%
80%
80%
60% 40%
84.2%
Patient-Actor
93.4%
88.4%
Actor-Patient
20%
Human-Human
Human-Animal
Human-Inanimate object
• • • • •
Picture description task. 60 native speakers of Kaqchikel 24 target pictures (8 × 3) Participants verbally described target pictures using a simple sentence Percentages of SVO active sentences were calculated based on the sum of SVO and VOS sentences.
Analysis: mixed logistic regression was conducted with patient animacy 100% 80% 60% VOS 40%
87.1%
SVO
81.1% 60.9%
20% 0%
p SVO order was most frequently produced. p VOS sentences were produced more often with the inanimate object than with the other conditions (p < .05, respectively).
Human-Human Human-Animal Human-Object
Experiment 2: Non-verbal description • Gesture production task. • 18 target pictures (6 × 3) • Participants described target pictures to a collaborator who pretended not to have seen the pictures before. References: [1] Goldin-Meadow et al. (2008). Proceedings of the National Academy of Sciences, 105(27), 9163-9168. [2] Langus & Nespor (2010). Cognitive Psychology, 60, 291-318. [3] Matzar et al. (1999). Gramatica del Idioma Kaqchikel. PLFM. La Antigua Guatemala. [4] Kubo et al. (2012). Poster presented at the 25th Cuny Conference, New York. [5] Bock & Warren (1985). Cognition, 21, 47–67.
Acknowledgement: The authors thank Yoshiho Yasugi, Lolmay P. García, Feliberto P. Majzul, and Juan E. A. Sián for their support for conducting experiments in Guatemala, and other project members for valuable comments. This research was supported by Grant-in-Aid for Scientific Research (S) #22222001 (PI: Masatoshi Koizumi).
100.0% 40%
70.6%
81.6%
20%
0%
0% Human-Human Human-Animal
Experiment 1: Verbal description [4]
60%
Human-Object
Human-Human Human-Animal
Human-Object
• Actor-Patient order was more frequently produced than Patient-Action order in both languages. • For Kaqchikel speakers, production of Patient-Actor is more frequent in human-patient condition than in the other conditions (mixed logistic regression; p <.05, respectively). • Japanese speakers showed the same tendency as Kaqchikel (Fisher's exact test (two tailed), p < .01). • Differences between two languages were not significant (Fisher's exact test (two tailed), p > .1).
② Distribution of constituent orders (Ar=Actor, P=Patient, A=Action) 70% 60% 50%
Kaqchikel 40%
Japanese
30% 20% 10% 0% ArPA
ArAP
AArP
PArA
PAAr
p Differences between the two languages were significant (χ2(5) = 42.195, p <.05). • Kaqchikel speakers produced more ActorAction-Patient order than Japanese speakers. • Japanese speakers produced more ActorPatient-Action order than Kaqchikel speakers.
APAr
Discussion & Conclusion p Agent and Patient order • Regardless of word order of a given language, there was a clear tendency to produce gestures with Actor-Patient order. • Kaqchikel speakers showed the opposite tendency between verbal and gestural description with respect of animacy effect.
→ These results support the general preference of Actor-Patient order independent of their languages in non-verbal event description. p Action and Patient order • Departing from the prior studies ([1][2]), we observed Actor-Action-Patient order in Kaqchikel compared to Japanese. • Gestural description by Kaqchikel speakers might be influenced by VO-initial order of their language, though further investigation is needed. 34th ANNUAL MEETING OF THE COGNITIVE SCIENCE SOCIETY (Sapporo), August 4, 2012