2018 Sat Poster 6409

Saturday, November 3, 2018 | Poster Session II, Metcalf Small | 3:15pm

The Anti-reconstruction Property of Cleft Constructions in Child Japanese
H. Shimada, A. Ohba, R. Okada, K. Yamakoshi

Introduction & Research Question

In Japanese cleft constructions (hereinafter JC), it has been widely assumed that the subject/sentential negation in the presupposition clause does not c-command the focus position. Further, JC show some anti-reconstruction properties: negative polarity items(NPIs) such as nanimo ‘anything’ or sika ‘only’, which appear in the focus position, cannot be licensed by reconstruction to the presuppositional clause, as shown in (1) ([1], [2]). However, it is not always the case that focus phrases cannot be undone. According to [3], the availability of the distributive reading in (2) indicates that the focus phrase “3- kyoku” is c-commanded by the quantified subject by reconstruction at LF.

How do Japanese children learn the anti-reconstruction properties of JC? If children learn grammatical knowledge concerning JC based on their experience, they need to receive “Negative Evidence,” which seems to be unavailable in the input of adults ([4] [5]). Hence, under such a scenario, it is expected that Japanese children should be insensitive to the anti-reconstruction properties of JC.

Experiment
In order to address this issue, we examined 27 children using the Truth Value Judgment Task ([6]). The children were divided into two groups. As for the target group, we investigated whether Japanese children disallow the reconstruction of a focus phrase in JC with negation. The test sentence in (3) was given under situation in (4). If the children correctly disallow the reconstruction of the focus phrase, they should assign the “all>not” reading to the sentence, and hence, they should reject (3). Regarding the control group, the test sentence in (5) was given under the same situation. (5) has the canonical word order and is ambiguous. If the children can correctly access the “not>all” reading, they should accept (5). We also investigated whether the participants in the target group correctly allow the reconstruction of a focus phrase without negation in (6), compared with the control group’s performance in terms of the canonical sentence in (7). (6) and (7) were given under the situation in (8). In our experiment, there were two trials for each item.

Discussion
The results are shown in Table 2. With regard to the target group, the incorrect acceptance rate of (3) is 28.57%(8/28), which indicates that most children correctly disallowed reconstruction of the focus phrase in (3). In contrast, they correctly accepted the distributive reading in (6) at 85.71%(24/28) of the time. This means that they have grammatical knowledge of reconstruction of a focus phrase, but most of them did not apply it in (3)_. In the control group, they also correctly allowed the distributive reading in (7) at 84.62%(22/26) of the time. In addition, they correctly accepted the canonical test sentence (5) with negation and all. The acceptance rate was 80.77%(21/26), which indicates that children can access “not>all” reading if the universal quantifier is c- commanded by negation. The difference in the acceptance rate between (3) and (5) is statistically significant(F(1,25)=11.127, p=0.003). Therefore, Japanese children seem to be sensitive to the anti-reconstruction properties of JC, despite the lack of negative evidence.

References

[1] Mihara & Hiraiwa 2006. [2] Hiraiwa & Ishihara 2012. [3] Nishigauchi & Fujii 2006.

[4] Brown & Hanlon 1970. [5] Chomsky 1981. [6] Crain & Thornton 1998