Distribution of Senses of a Polysemous Word in Japanese Text

Using “Balanced Corpus of Contemporary Written Japanese”

Abstract

In this paper, we report how the senses of a polysemous word are distributed in a given text. Using the “Balanced Corpus of Contemporary Written Japanese,” (BCCWJ) we received the following results. (1) Polysemous words are likely to share the same sense when occurring more than once in a given text. The tendency is not so strong as to admit no exceptions. (2) The gaps between instances of a polysemous word with the same sense are shorter than gaps with different senses and vice versa. (3) This phenomenon is also observed between a polysemous word and its synonyms and antonyms to some extent. These results suggest that the lexical cohesion positively affects the distribution of the senses of polysemous words as well as related terms to a lesser extent.