This thesis is a study of clause architecture within a theory of generative grammar. It discusses four major syntactic hypotheses that have crucial bearing on the design of phrase structure: the Agreement-based Case theory, the internal subject hypothesis (ISH), the uniform three-level X-bar theory, and a hypothesis about string vacuous head movement.In Chapter 2 through Chapter 4, I discuss object positions in three typologically and genetically different languages, i.e. English, Zarma, and Japanese, and argue that they all possess Object Agreement Phrases (AGRoP). A consideration of learnability suggests that the presence of AGRoP in these languages, particularly in Japanese, is not learnable from the data available to children. Thus, it is highly likely that UG is so construed that every language has AGR (THE UNIVERSAL AGR HYPOTHESIS).In Chapter 5, I turn to subject positions. There is conflicting evidence regarding the base-position of the external argument. Some data indicate that the external argument originates in a position lower than its surface position, as is expected under the internal subject hypothesis. Other data suggest that the base-position of the external argument is outside AGRoP and the VP that dominates the main verb and the internal arguments. If the ISH is correct in that all arguments of a predicate category (e.g. V) originate within the maximal projection of this category, then the apparently contradicting data suggest that there are two verbs, hence two VPs, in a single clause (THE SPLIT VP HYPOTHESIS). One verb is above AGRoP, and the other is below AGRoP. The so-called external argument is an argument of the upper V, and "internal arguments" are arguments of the lower V.Chapter 6 is concerned with X-bar theory. I argue, contrary to the wide held view, that some functional categories allow more than one specifier position within their projections (LAYERED SPECIFIERS), and that the number of specifiers is different across categories. If this is correct, the X-bar schema as such cannot be a part of Universal Grammar, as already suggested in Fukui (1986) and others. Our claim, however, is crucially different from Fukui's (1986) in that the specifiers of functional categories do not necessarily (but sometimes do) "close off" their projections.In Chapter 7, I discuss string vacuous verb raising in head final languages. In particular, I present evidence that verbs in Japanese raise out of the VP in overt syntax. Its consequences are also explored to various aspects of syntactic theory such as the Proper

No abstract

The present study investigated scrambling effects on the processing of Japanese sentences and priority information used among thematic roles, case particles and grammatical functions. Reaction times for correct sentence decisions were significantly prolonged for scrambled active sentences with transitive verbs in the first experiment and with ditransitive verbs in the second experiment. Errors were made with scrambled sentences more than canonical sentences in both experiments, which suggested that scrambling effects were apparent in active sentences. Passive sentences in the third experiment indicated that canonical order defined based on case particles, not thematic roles, was more quickly and accurately identified than scrambled order. Potential sentences in the fourth experiment and causative sentences in the fifth experiment indicated that the processing of scrambled sentences based on grammatical functions, but not on case particles, required longer reaction times and resulted in higher error rates than canonical sentences. Consequently, scrambling effects in the present study indicated that neither thematic roles nor case particles can provide fully-satisfactory information for canonical phrase order, and that only grammatical functions offer satisfactory information in all types of sentences.

No abstract

No abstract