1981 年 52 巻 5 号 p. 266-273
Two experiments were conducted with Japanese subjects to investigate whether in reading Japanese phonological recoding is an obligatory stage or the direct access to semantic representation is the general rule. The first experiment used a word reading-out task and the second, a sentence judgement task. In the first experiment Kana (Japanese characters) were read out faster than Kanji (Chinese characters), but in the second experiment with the silent reading condition Kanji were judged faster than Kana. These results suggest that Kana is superior to Kanji in access to the phonemic codes, but this does not imply that the meaning is comprehended more rapidly in Kana. In contrast, Kanji takes longer for reading out (decoding) than Kana, but makes rapid access to semantic codes possible. This seems to indicate that in the silent reading of Kanji the direct proccessing from visual (graphemic) codes to meaning (semantic codes) is possible, whereas in Kana the relation of graphemic codes to meaning is mediated by the phonemic system.