Based on the scope of negation in Japanese, the dissertation establishes a generalization that all vP-internal phrases move out of NegP in Japanese. The main argument concerns reconstruction, which allows a moved element to be interpreted at the original site rather than the surface position. It is shown that reconstruction applies to moved elements quite generally except for those that occur with adjoined focus particles in both Japanese and English. A semantic explanation for this is provided on the basis of Fox’s (2003) trace conversion, an interpretive mechanism that serves to interpret movement chains. Given the mechanism, it is shown that a number of differences between Japanese and English with respect to scope relations between arguments and negation can be derived from a single structural difference: Only in Japanese, objects obligatorily move to a position above negation. The analysis of the anti-reconstruction property of focus particles presented in the dissertation is extended to provide explanations for the Neg-split phenomenon in English, the fact that a contrastive topic is interpreted as the focus of negation in Japanese, and the interpretive properties of focus particles in German and various kinds of disjunction phrases in English, Japanese, and Korean. Regarding the reason for the movement of vP-internal phrases in Japanese, it is shown that it must occur due to the morphological merger of heads. It is argued that T, Neg, v and V are combined into a word by morphological merger in Japanese, and that morphological merger is subject to a structural adjacency condition that prohibits overt elements from intervening between T and V. It follows then that all vP-internal phrases must move out of NegP for morphological merger to apply. The analysis is shown to make correct predictions for a number of properties of constructions with a variety of complex predicates, such as causatives and potentials. A further consequence of the analysis is that Japanese does not have the process of su-support that parallels English do-support.
Stored only when both papers are in the corpus — 40 citations exist upstream.
Imported from OpenAlex (CC0) — nothing here is generated.