[Combinability of 2965 kanji characters of Japanese Industrial Standard on formation of two-kanji compound words].

Japanese two-kanji compound words are composed of front- and rear-single kanji (e.g., "counsel" comprises/soo/and/dan/). Kanji characters are likely to vary in pronunciation according to the combination of front and rear characters (e.g., "partner" comprises/ai/and/te/). The purpose of this study is to address the combinability of 2 965 Japanese Industrial Standard kanji based on two different corpus scales: 78 426 compound words in the large corpus and 5 209 compound words in the small corpus. Results indicate that there are fewer companions of front kanji than those of rear kanji in both corpora. The results are discussed in terms of how the combinability of kanji would affect recognition of two-kanji compound words.