- 京都大学言語学研究 (ISSN:13497804)
- no.26, pp.59-75, 2007
This paper deals with the formation of Japanese compound verbs and claims that possible combination of verbs, as well as their argument structures, can be predicted by assuming the mechanism of superimposition of meaning. It has been pointed out that lexical verbs in natural languages, whether simplex or complex, are subject to inherent semantic constraints, which can be captured by the following schema in the framework of Lexical Conceptual Structure (LCS): (i) [x ACT(-ON y)] CAUSE [BECOME [y BE-AT z]] In previous studies, however, it has been often assumed that the LCSs of compound verbs have more complex structures, in which the LCSs of their constituent verbs are combined with additional predicates such as BY and WHILE. This paper takes another approach. It is assumed that every compound verb strictly follows the structure shown in (i), and its content is obtained by superimposing the LCSs of the two constituent verbs. It is shown that this assumption accords well with basic properties of Japanese compound verbs, and some of their peculiar behaviors can also be seen as direct consequences of the assumption. This paper further argues that possible combinations of verbs can be properly constrained by taking into account three additional constraints: the temporal iconicity condition, the subject identity, and the theme identity. The transitivity identity condition, another constraint proposed in the literature, automatically follows from the conditions proposed in this paper.