PURPOSE: To present KANJI (Chinese character) candidates desired by a user with the highest preference and to improve the automatic selection efficiency of homonyms by giving the preference to the KANJI candidates which are consistent among sentences.

CONSTITUTION: An independent word storage part C5 is provided together with an adjunct storage means C6, an independent word table C7, and an independent word deciding part C8 which refers to the table C7 to decide whether the independent words forming each KANJI candidate are included in the part C5 or not when the KANJI candidates connected to the specific affixial words are listed to the input KANA (Japanese syllabary) characters and gives the preference to the KANJI candidate whose independent words are included in the part C5. Then the independent words included in the converted KANJI- KANA sentences are stored. Then it is decided whether the independent words approximate to the KANJI candidates connected to the specific adjuncts in terms of meaning are already used or not. The preference is given to the KANJI candidate corresponding to the used words. Thus the preference is given to the KANJI candidates which are consistent among sentences. Then the automatic selection factor is improved for homonyms.