Japanese morphology and its theoretical consequences: Derivational morphology in distributed morphology.

摘要

DISTRIBUTED MORPHOLOGY (DM) (Halle and Marantz, 1993) is a research program in morphology which abandons the traditional generative Lexicon (Chomsky, 1965 and 1995, among many). Recent work argues that all generative processes, including derivational morphology, can be accomplished syntactically, the SINGLE ENGINE HYPOTHESIS (Marantz, 2001). In Chapter 1, I introduce the most recent work within DM which adopts and adapts Chomsky'S DERIVATION BY PHASE HYPOTHESIS to lexical-category formation. I then reanalyze some important and well-known data of Aronoff (1976) in order to show that the single engine hypothesis is motivated and explanatory. Chapter 2 proposes an analysis of two types of common deverbals nominalizations in Japanese. I argue that, actually, only one of the two types is deverbal; the other type is root-derived. Those root-derived nominalizations that contain apparent verbal transitivity markers, the focus of this chapter, raise a paradox for the single engine hypothesis because of their non-compositional semantics. I resolve it by adopting a proposal of den Dikken (1995)'s: anomalous transitivity markers are AFFIXAL PARTICLES. Chapter 3 concentrates on lexical causatives in Japanese. There is a widely-held view among linguists (Harley, 1995, 1996, Levin and Rappaport Hovav, 1995, Pinker, 1989, among many), that a lexical causative cannot be derived from a verb which has an agentive subject. Using observations of Matsumoto (1996) and data from idioms in Japanese I argue that no such semantic criterion applies in Japanese. Given the proper pragmatic reading, all verbs with agentive subject can have a mono-clausal causative partner. To put it another way, all verbs, regardless of their lexical semantics have lexical causatives in Japanese. This seemingly unique characteristic of Japanese is argued to be directly related to the fact that apparent transitivity markers in Japanese are affixal particles as argued in Chapter 2. Chapter 4 concludes with a comparison of transitivity marking in Turkish and Korean with Japanese. I argue differences support the affixal particle analysis for Japanese. The proposed analysis, under standard historical assumptions about Japanese, raises an issue about the diachronic direction of grammaticalizations. With Roberts and Roussou (2003)'s work on grammaticalizations as background, this issue is briefly discussed.