三好 伸芳

This paper clarifies the differences in the distribution of nonrestrictive relative clauses among different text genres in Japanese by using a parsed corpus with syntactic and semantic information, the NINJAL Parsed Corpus of Modern Japanese (NPCMJ). Conventional corpora based on morphological information are difficult to quantify by specifying an arbitrary syntactic environment. In contrast, the NPCMJ, a corpus with syntactic and semantic information, can be used to specify the syntactic environment of the clause and the lexical properties of the head noun and to search for adnominal constructions. The results of this survey demonstrate that nonrestrictive relative clauses in Japanese occur more frequently in expository texts than in literary texts, and in the former text genre, they tend to be associated with the function of adding information to the head noun (rather than to the main clause as a whole). This distribution is explained by the characteristics of expository texts: the frequent occurrence of discourse-new proper nouns.