The syntax of subjunctive complements: Evidence from Japanese

要旨

This dissertation aims to present a syntactic analysis of seemingly problematic, but intriguing properties of the subjunctive complements in Japanese concerning Tense (T), Case for subjects, and the locality of A-scrambling, under the Minimalist Program pursued by Chomsky (1995, 1998, 1999). ^ Chapter 2 provides basic descriptions concerning the categorial status and the distribution of subjunctive clauses. I establish that the two forms, -yoo(ni-(to)) and - koto, are the subjunctive complementizers introducing finite clauses. ^ Chapter 3 discusses the Case properties of complement subjects and their control property. It is indicated that subject-to-subject raising takes place out of a certain type of subjunctive complement. It is also demonstrated that while only nominative Case is licensed for subjects in some other types of subjunctive complement, both nominative Case and null Case are actually available in the rest. I also discuss how the control property of empty subjects is related to interaction between the semantics of the governing verbs and the aspectual property of complement predicates. ^ The tense property of subjunctive complements is deeply investigated in Chapter 4. It is observed that T in a certain group of subjunctive complements is defective with respect to tense morphology, licensing of temporal adverbs, and tense interpretations of nonpast stative predicates. I point out that nominative Case and null Case are correlated with [+ tense] feature and defective [− past] feature, respectively. ^ Chapter 5 presents two analyses of the fact that long-distance A-scrambling is allowed out of subjunctive CP complements, but not out of non-subjunctive CP complements. One analysis accounts for the case in which defective T occurs in the complement. Based on Chomsky's (1998, 1999) hypothesis concerning strong phases, I propose that CP of which head selects defective T does not count as a strong phase. The other analysis explains the case of the complement headed by -koto, in which complete T appears. Based on - koto's rich nominal property, I assume that the Spec of - koto is an A-/L-related position. Furthermore, it is suggested that V-to-T movement is not a necessary condition on A-scrambling and that A-scrambling is movement of a different kind from A′-scrambling. ^

主題