非情物主語のニ受動文関連性に基づく分析へ

Abstract

In Japanese passives, the agent can be marked with either ni or ni yotte. In contrast to the ni yotte-passive, however, the ni-passive can often not be used when the subject referent is inanimate. The goals of this paper are to examine under what circumstances inanimate nipassives become acceptable, and to propose a pragmatic explanation within the framework of Relevance Theory. First, accounts based on the assumption that the ni-passive always expresses the affectedness of the subject referent are considered and rejected. Next, the the basic principles of Relevance Theory are introduced, according to which Shibatani's crosslinguistic analysis of the passive voice, based on the concept of 'discourse relevance', is then reinterpreted. It is hypothesised that the function of the sentence subject is to indicate to the hearer that extending the context with assumptions related to the subject referent will lead to larger contextual effects, and thus to more relevance. It is then claimed that this hypothesis, combined with a view of the ni-passive construction as simply a grammatically marked method of choosing non-agentive event participants as subjects, allows for a natural explanation of how semantic features of referents, such as animacy, can influence the acceptability of ni-passives. The fact that inanimate ni-passives become acceptable when they imply an attribute of the subject or when they involve a 'latent affectee', as argued for mainly by Masuoka, is explained using the notion of implicature. In the course of the discussion, the present analysis is compared with the perspective-based analyses of Kuno and Takami, which turn out not to be able to offer the same level of consistency.