No abstract

No abstract

No abstract

This paper argues that wh-in-situ in Japanese in fact involves S-structure movement of an invisible entity and therefore that the Subjacency effects discovered and used as evidence for LF Subjacency by Nishigauchi (1986Nishigauchi ( , 1990 and Pesetsky (1987) are due to tiffs S-structure movenlent. This conclusion is forced on us by the facts that (1) in multiple questions where one of the wh-phrases is inside an island and the other is outside of it, there are no Subjaceney effects, contrary, to the expectation of the LF Subjacency hypothesis, and that (2) an indirect question constitutes a whisland for overt movement even though the visible part of the wh-phrase is still insitu at S-structure.Since the work by Huang (1982), a common assumption has been that S-structure movement is constrained by Subjacency and the ECP, whereas LF movement is only sensitive to the ECP. There is, however, a growing body of literature including argues that Subjacency applies to LF movement as well. Thus, there is an apparent conflict here between the evidence that motivates the lack of Subjacency at LF and the one that displays LF Subjacency effects. This paper attempts to resolve this paradoxical situation by focusing on the properties of wh-movement in Japanese. Specifically, it will be argued, by bringing in new data, that Subjacency should constrain only the mapping from D-structure to S-structure. This paper basically presupposes the Barriers framework of Chomsky (1986) and some subsequent work, where Subjacency subsumes the Condition on Extraction Domain (CED), the Complex NP Constraint, and the wh-island Condition. The main logic of the discussion is the following. We first observe that there are two levels of movement in Japanese interrogatives, only the first of which obeys Subjaccncy. On the ground that there is a wh-island effect for overt movement in Japanese, it follows that the first level of movement is identified as S-structure movement, since the account of the wh-island effect requires the CP Spec of a whisland to be filled at S-structure.Section 1 clarifies the issues surrounding the LF Subjacency hypothesis and then looks at further data which are not examined in the past literature, arguing that apparent LF Subjacency effects do not manifest themselves uniformly. Section 2 attempts to establish that Japanese wh-

No abstract