A test of a hypothesis of automatic phonological processing of Kanji words

This paper aims to examine a hypothesis that phonological processing should occur in Japanese Kanji words as well as in Kana words, but automatic in Kanji words. In Experiment 1, subjects performed concurrent articulation or finger tapping during a semantic processing task of Kana words with 2 to 4 phonemes, for which phonological processing should be indispensable. Concurrent articulation was found to eliminate the phoneme number effect found in the control condition, but finger tapping was not. This result indicates the effectiveness of concurrent articulation in articulatory suppression. In Experiment 2, the materials were changed to Kanji words, for which phonological processing had been supposed to be dispensable, and subjects performed the same task. The results showed no phoneme number effect even in the control condition, and the reaction time in the concurrent articulation condition was delayed significantly. These results suggest the existence of automatic phonological processing of Kanji words. Finally, the pertinence of this hypothesis was confirmed by applying it to the explanations of 3 kinds of Japanese dyslexia.