Noun Phrases in Japanese and English: A Study in Syntax, Learnability and Acquisition

In Japanese, the genitive Case marker no is inserted after NP and PP prenominal modifiers, but not after relative clauses. Saito (1982) and Fukui (1986), among others, propose a no-insertion rule to account for this fact. Harada (1980) and Clancy (1985) observe that some Japanese speaking children, at around 2 years old, insert no even after relative clauses. ^ This thesis presents a hypothesis as to what the overgenerated no is, and as to how children retreat from this overgeneration. In order to attain answers to these questions, I first discuss in Chapter II the syntactic properties of several types of no in Japanese. In particular, I argue that nos of categories N and C exist, contrary to Kitagawa and Ross (1982), and Fukui (1986), but as suggested in Okutsu (1974) and Hoji (1990). This sets up three candidates for the overgenerated no: the genitive Case marker, no as N, and no as C. ^ Chapter III deals with the syntax of Japanese relative clauses. I first extend Perlmutter's (1972) analysis along the lines of Saito's (1985) analysis of topicalization, and claim that Japanese relative clauses may but need not involve movement. Based on this assumption, I draw two conclusions: (i) Japanese has pro PP for time and place adjuncts, and (ii) Japanese relative clauses are IP (the IP hypothesis). ^ With the syntactic analysis in Chapters II and III, Chapter IV addresses the questions concerning the overgenerated no in question. On the basis of further acquisition studies, I first draw the conclusion that the overgenerated no is of the category C. The focus, at this point, moves to the learnability question: why and how they retreat from it. Here, I rely crucially on the conclusion drawn in Chapter III that Japanese relative clauses are IPs. Based on this conclusion, I propose that Japanese children make the initial hypothesis that relative clauses are CPs, and lexically realize the head C as no. They later attain the knowledge that Japanese relative clauses are IPs, and hence, cease to overgenerated no. It is shown finally that this hypothesis meets the learnability criterion. On the basis of the positive evidence on the structure of pure complex NPs, Japanese children infer that all prenominal sentential modifiers, in particular, relative clauses are IPs. Thus, the IP hypothesis receives support from the studies in syntax, learnability and acquisition. ^