A conservative interpretation of the reflexive zibun by Japanese children
Naho
1
Orita ,
Hajime
2
Ono ,
Naomi
1
Feldman ,
Jeffrey
1
Lidz [1:
å
Experiment
Learning Problem
• “John thinks that Bill painted himself.” (himself = Bill, himself ≠ John)
Akira-ga
zibun-ni
penki-wo
nut-ta
to
omot-teiru.
Taro-TOP
Akira-NOM
zibun-DAT
paint.N-ACC
paint.V-PAST
COMP
think-PRS.PRG
interpretation 2
long-distance
• No long-distance zibun in child-directed speech.
Learning long-distance zibun requires
projecting beyond the input.
0
0
0
3
31
12
zibun-wa/ga (zibun-TOP/NOM)
zibun-wo/ni (zibun-ACC/DAT)
Total
5
0
37
0
0
8
0
0
0
1
0
4
6
0
49
MiiPro Corpus, Arika (2;11-5;0): 40,412 utterances in total from her mother.
Previous
Method: Truth
studies
Value Judgment task
• TVJT: A puppet makes a statement about a story. A child’s task is to reward or
correct the puppet based on the statement.
• Previous studies: Children allow the long-distance zibun interpretation.
- Otsu (1990): disproportional salience of potential antecedents (zibun’s picture)
- Okabe (2008): failure to satisfy Condition of Plausible Dissent (Crain & Thornton 1998)
References
60.94%
AgeGroup
Adults
Children
50
25
20.83%
Condition
Experiment 2: kare
!
Taro thinks that Akira painted kare. → True
Children can access
the matrix subject.
100
75
73.33%
60.42%
AgeGroup
Adults
Children
50
25
Two possibilities
• zibun is local-only
• unable to access the matrix subject
→ To distinguish these possibilities,
we ran a kare condition.
Local Long−dist
Total
1
7
The long-distance
interpretation of zibun is
true at the end of the
scenario.
66.67%
Acceptance %
3rd, extra-sentential
30
2
Scenario example
a significant interaction between age and
condition (p < 0.01)
75
0
zibun-de (by zibun)
zibun-no (self’s)
1. Painting game
Children have local-bias
83.33%
“Taro thinks that Akira painted zibun.”
“Taro thinks that Akira painted zibun.”
2nd, local
3rd, local 3rd, non-local,
intra-sentential
Taro thinks that Akira painted zibun. → True
Taro thinks that Akira painted zibun. → True
100
Acceptance %
Taro-wa
local
1: local & long-distance zibun
Local zibun
Long-distance zibun
• Japanese zibun can be bound across clause boundaries. interpretation 1
University of Maryland, 2: Tsuda College]
Discussion
Summary
Children incorrectly rejected the
long-distance antecedent for
zibun, despite being able to
access this antecedent for kare.
Children’s knowledge of zibun
• Local-only?
• Ambiguity resolution difficulty?
Acquiring LD zibun beyond the input
0
Matrix
Condition
2. Akira painted Akira (→ False) 3. Local zibun (→ False)
• The form of intensifiers
• Occurrence in subject position
4. Long-distance zibun (→ True) 5. A puppet’s statement
!
!
Crain, S., & Thornton, R. (1998). Investigations in universal grammar: A guide to experiments on the acquisition of syntax and semantics.
Cambridge, MA: MIT Press. Nisisawa, Y., & Miyata, S. (2010). Japanese -MiiPro- Arika corpus. PA: TalkBank.Okabe, R. (2008). Child
causatives: Acquisition of bi-clausal structures in Japanese. Ph.D. dissertation, UCLA. Otsu, Y. (1992). Zibun futatabi [zibun revisited]. In
Ninchi Gengo no Seiritsu: Ningen no Kokoro no Hattatsu, (pp. 113-122). Tokyo: Kuba Pro.
Acknowledgments
We thank Ashiya Aiko preschool and Higashi-Toyonaka St. Michael nursery school for collecting child participants’ data; Tomo
Fujii, Hideki Kishimoto, and Taisuke Nishigauchi for collecting adult participants’ data; and Akira Omaki and Shevaun Lewis for
advice on the experiment setup. This research was partially supported by UMD through an International Graduate Research
Fellowship and through international student support supplementary to an NSF IGERT award.