[PS-2.52] Online Resolution of Empty Arguments in Japanese

Fujii, T. 1 , Ono, H. 2 & Yoshida, M. 3

1 Yokohama National University
2 Tsuda College
3 Northwestern University

Arguments are often empty in Japanese. In (1a), the because-clause has a missing dative argument (?__?). The missing dative in (1a) might be analyzed as a gap left by movement of the dative NP that follows. This analysis should fail because a gap must be c-commanded by its antecedent (Saito 89).
(1) [Taro-ga ? doku-o noma-seta node] {a. Jiro-ni/ b. Jiro-o} ...
T-nom poison-acc take-made because a.J-dat / b. J-acc ?
?Because Taro made ___ take a poison, ? John... ?
When Jiro bears accusative [(1b)], no possibility of movement arises due to case mismatch, where ?__? should be interpreted as a pronoun.
We investigated the online reading of the sentence type shown in (2). Embedded-Verb (Dat- vs. Acc-Verb) x Case of the potential antecedent (Dat- vs. Acc-NP) were independent factors.
(2) [Taro-nom poison-acc {take-made/took}-because] Jiro-dat/-acc NP-nom Adv Verb
A significant interaction of the two factors (t=2.10, p<.05) was found at Adv, and Adv was read significantly slower in the Dat-Verb/Dat-NP condition than in the Dat-Verb/Acc-NP (t=2.55, p<.02). This suggests that the parser first analyzed "?___" as a movement-gap and then recognizes the illicitness of the configuration.