The main purpose of this study was to examine whether or not the reading of Katakana and Hiragana entails different types of information processing, and whether highly familiar Katakana words can be read visually, in a gestalt way. The study, consisting of two experiments, was designed to compare the reading performance between native Japanese and foreign readers. The results showed some fundamental differences in the structure of the internal lexicon corresponding to Kana syllabaries between Japanese and foreign students, and the tendency towards using phonetic codes in reading Kana words in Japanese subjects, but not in foreign ones.