Multi-dimensional word formation in Japanese.

摘要

Some constructions in a natural language behave quite irregularly. Among those are the constructions involving verbal nouns in Japanese which are exclusively treated in this dissertation. The theoretical framework employed in this dissertation is the Autolexical Grammar advanced by Sadock (1985, 1981). The goal of this dissertation is to demonstrate that those irregularities can be explained in terms of the interaction between lexical information and principles on which those grammatical components operate.;In Chapter 1, the overview of the framework is introduced. The basic assumptions, subcategorization frames, and case-marking system are sketched out. Then, the essential properties of verbal nouns (VNs) are discussed. Chapter 2 explores the VN-su-constructions, in which the VN is combined with a semantically empty verb su. The previous analyses about the VN-su-constructions are examined in detail. Then, an alternative approach to the constructions is proposed from the viewpoint of Autolexical Grammar. Chapter 3 reviews the observations by such authors as Miyagawa (1989), Tsujimura (1990), and Kageyama (1993) which test the unaccusativity of verbs, and onomatopoeic words. It is shown that the observations of onomatopoeic words pro ide important clues to expose the true nature of VN-su-constructions and the light verb constructions. Chapter 4 surveys the light verb constructions, in which a verbal noun appears as a direct object of the semantically empty verb su. From the parallels and differences between the light verb constructions and noun incorporation examined in this chapter, an interesting conclusion is deduced: There are two types of incorporation---morphological incorporation and syntactic incorporation. Chapter 5 discusses nominal clauses, which are such clauses that take a noun as a head, but have sentential cases on NPs. The head noun of nominal clauses is a deverbal nominal, which, Iida (1987) claims, is formed through nominalization. It is shown, then, that the deverbal nominals are not formed through nominalization. In Chapter 6, N+VN compounds, which are composed of a noun and a verbal noun, are examined in detail. Shibatani and Kageyama (1988) claim that the N+VN compounds are composed after they have undergone all syntactic operations. I argue against their claim, and conclude that they simply involve many mismatches among components.