referential-expression preferences in english and …aschafer/miyaoschaferschwartz-cuny2015.pdf ·...
TRANSCRIPT
Referential-expression preferences in English and Japanese discourse during reading
Mari Miyao1, Amy J. Schafer2 & Bonnie D. Schwartz1 Department of Second Language Studies1, Department of Linguistics2 & LAE Labs, University of Hawai‘i
Thanks to the dissertation committee members, Yunchuan Chen, an NSF Doctoral Dissertation Research Improvement Grant (BCS-1252235), the Linguistics Beyond the Classroom Program, the Hawai‘i English Language Program, and the participants.
Selected references
Clancy, P. (1980). Referential choice in English and Japanese narrative discourse. In W. Chafe (Ed.), The pear stories: Cognitive, cultural, and linguistic aspects of narrative production (pp. 127–202). Norwood, NJ: Ablex.
Gordon, P. C., Grosz, B. J., & Gilliom, L. A. (1993). Pronouns, names, and the centering of attention in discourse. Cognitive Science, 17(3), 311–347.
Hoji, H. (1990). Kare. In C. Georgopoulos & R. Ishihara (Eds.), Interdisciplinary Approaches to Language: Essays in Honor of Prof. S.-Y. Kuroda (pp. 287–304). Dordrecht: Reidel.
Kaiser, E., & Trueswell, J. C. (2008). Interpreting pronouns and demonstratives in Finnish: Evidence for a form-specific approach to reference resolution. Language and Cognitive Processes, 23(5), 709–748.
Noguchi, T. (1997). Two types of variables and variable binding. Language, 73(4), 770–797. Yang, C.-L., Gordon, P. C., Hendrick, R., & Wu, J.-T. (1999). Comprehension of referring expressions in Chinese. Language and
Cognitive Processes, 14(5), 715–743.
28th CUNY Conference on Human Sentence Processing 21 March 2015
University of Southern California
3. Methodology (sentence-by-sentence self-paced reading task)
Participants: 40 English speakers & 36 Japanese speakers
Materials
Task: Read one sentence at a time at a natural speed & answer a comprehension question presented after the third sentence (i.e., sentence [c])
Japanese Ø 30 test items (k=5, Latin-squared): 20 closely translated from English + 10 for Null conditions • 3 RE forms: Null vs. Overt vs. Name • 2 discourse types: [Continue] vs. [Shift]
Ø 51 filler items Ø 6 item lists + 6 lists in reversed-item order
a. 由香さんは 今朝 9時に 誠君を 起こしました。 Yuka-san-wa kesa 9-zi-ni Makoto-kun-o okosimasita. Yuka-HON-TOP this morning 9-o’clock-at Makoto-HON-ACC woke up
b. ø/彼女は/由香さんは 毛布を めくって 「起きて!」と 言いました。
[ø/Kanozyo-wa/Yuka-san-wa] moofu-o mekutte “okite!”-to iimasita. [null/She-TOP/Yuka-HON-TOP] blanket-ACC took off “Wake up!”-COMP said
b’. ø/彼は/誠君は 目覚まし時計を 見て ベッドから 飛び出しました。
[ø/Kare-wa/Makoto-kun-wa] mezamasi-dokei-o mite beddo-kara tobidasimasita. [null/He-TOP/Makoto-HON-TOP] alarm-clock-ACC looked at bed-from jumped out
c. 授業は 9時半に 始まります。
Zyugyoo-wa 9-zi-han-ni hazimarimasu. classes-TOP 9-o’clock-half-at start
a. Jane woke up Tom at 9 am this morning.
b. She/Jane took off the blanket and said, “Wake up!”
b’. He/Tom looked at the alarm and jumped out of bed. c. Classes start at 9:30 am.
English Ø 20 test items (k=5, Latin-squared) • 2 RE forms: Overt vs. Name • 2 discourse types: [Continue] vs. [Shift]
Ø 34 filler items Ø 4 item lists + 4 lists in reversed-item order
[Continue] (a-b-c)
[Shift] (a-b’-c)
Accuracy: 95% for English speakers; 94% for Japanese speakers
Data exclusion Ø Incorrectly answered items Ø RT outliers: 3.2% in English data; 2.4% in Japanese data RT conversion: Residual RTs adjusted for sentence length Significant RT differences Ø English speakers
[Continue] Overt vs. Name: t1(39) = –1.99, p = .05; t2(19) = –1.66, p = .11 Ø Japanese speakers
[Continue] Null vs. Overt: t1(35) = –3.97, p < .01; t2(29) = –3.41, p < .01 Null vs. Name: t1(35) = –2.28, p = .03; t2(29) = –2.39, p = .02 [Shift] Null vs. Name: t1(35) = 2.41, p = .02; t2(29) = 2.23, p = .03 Overt vs. Name: t1(35) = 2.74, p = .01; t2(29) = 2.39, p = .02
[Continue]
[Shift]
5. Discussion
Ø Both groups showed the RNP in the [Continue] condition à Overt < Name in English; Null < Name in Japanese
Ø Japanese speakers prefer Name in the [Shift] condition à Null dispreference is probably due to lack of gender features
Ø Status of Overt kare/kanozyo in Japanese • Overt dispreferred both for highly salient and for less salient entities • Overt falls outside the RE hierarchy (see Introduction above) • Overt in Japanese processed differently from Overt in Chinese due to: v infrequent use of Overt pronouns in Japanese? v sensitivity to factors other than syntactic role (form-specific
multiple-constraints approach, Kaiser & Trueswell 2008)? v cross-linguistic difference in the (syntactic/semantic) nature of
pronouns? Acknowledgments
1. Purpose
To investigate English and Japanese speakers’ online processing of referential expressions (REs: e.g., pronoun, repeated name) in discourse so as to compare RE processing across languages and by natives vs. learners (in progress: Chinese natives & Chinese/Japanese learners of English)
2. Introduction
Salience/Accessibility of discourse entities
Ø The salience of discourse entities is modulated by multiple factors (e.g., syntactic role) Ø Salience is inversely related to RE explicitness
Online processing of REs: English speakers
Ø Gordon, Grosz & Gilliom’s (1993) Exp. 4: sentence-by-sentence self-paced reading task
• Reading times (RTs) for sentences (c) & (c’): He never thinks… < George never thinks… She screamed… ≈ Debbie screamed…
• Use of repeated names for highly salient entities à the Repeated Name Penalty (RNP) Online processing of REs: Mandarin Chinese speakers
Ø Yang, Gordon, Hendrick & Wu’s (1999) Exps. 2&4: 3-sentence passages similar to Gordon et al.’s • 3 RE forms: null pronoun (“Null”), overt pronoun (“Overt”), repeated name (“Name”) • RT results in the [Continue] condition: Null ≈ Overt < Name à the RNP
RE forms in Japanese (null-argument language)
Ø Null pronouns predominantly used for discourse topics in elicited narratives (e.g., Clancy 1980) Ø Use of kare/kanozyo (“he”/“she”) is controversial • Not frequently used in production (e.g., Clancy 1980) • Hoji (1990): they share a deictic property with the demonstrative are (“that”) • Noguchi (1997): they can still be “pronouns”; they share, with English pronouns, “a feature that
sets a certain set of lexical categories apart from the others” (p. 783)
Research Questions
1. Will English speakers replicate Gordon et al.’s RNP? 2. Do Japanese speakers process REs as Chinese speakers do? (Will Null and Overt pronouns
pattern together in Japanese RE processing?)
Explicitness of REs
High Low Salience of discourse entities
Pronouns Full NPs
a. George jumped out from behind a tree and frightened Debbie. b. He was surprised at her hysterical reaction. c. He/George never thinks how others might feel. c’. She/Debbie screamed loudly and ran away. d. Practical jokes are not always fun for everyone.
[Continue] (a-b-c-d) [Shift] (a-b-c’-d)
4. RT results English speakers
-‐350
-‐250
-‐150
-‐50
50
150
Mean Re
sidu
al RTs (m
s)
Japanese speakers
-‐350
-‐250
-‐150
-‐50
50
150
Mean Re
sidu
al RTs (m
s)
Overt Name Null Overt Name
-‐350
-‐250
-‐150
-‐50
50
150
Mean Re
sidu
al RTs (m
s)
-‐350
-‐250
-‐150
-‐50
50
150
Mean RE
sidu
al RTs (m
s)