On the role of holistic processing in the recognition of Japanese Kana words

This study examined the possibility of holistic processing of Japanese Kana words by using three different tasks: lexical decision, naming, and letter-search. Twenty to twenty-four university students participated in each task. All of the words used in the experiments were foreign loan-words normally written in Katakana. In order to disrupt holistic processing in word recognition, the Katakana words were transcribed into either Hiragana or a mixture of Katakana and Hiragana. For the lexical decision task, the disruptive effects were larger for words than nonword controls. However, for the naming task, the script type manipulation had an equal effect on words and nonwords. Lastly, for the letter-search task, the word superiority effect on reaction time for the Hiragana stimuli was approximately the same magnitude as that for the Katakana stimuli. These findings can be interpreted within the framework of Besner and Johnston's (1989) processing model, and suggest that Kana words are identified primarily on the basis of letter level information rather than their holistic property.