1994 Volume 65 Issue 5 Pages 377-382
Katakana words (e.g., _??__??__??_=ka me ra) were modified in three different ways to create pseudowords that were presented to 36 Japanese readers in a lexical decision task. In Conditions 1 and 2, the first Kana symbol was changed, but in the former, the pronunciation of the initial consonant was kept intact (e.g., _??__??__??_+ko me ra), while in the latter the whole mora was changed (e.g., _??__??__??_=so me ra). In Condition 3, two morae were changed (e.g., _??__??__??_=so ki ra), creating a pseudo-word which was two Kana symbols different from any real word. If the pseudo-word gains access to lexical information, it will interfere with the pseudo-word classification response. Such a result was observed for Condition 1 (the mean RT=770ms) which significantly differed from both Condition 2 (735ms) and Condition 3 (720ms). Therefore, the interference effect was a phonological one, which indicates that Japanese readers must be sensitive to phonemic units when processing Kana. Thus, the smallest unit of phonological processing in Japanese is not necessarily the mora. The results are explained within the framework of interactive-activation model whereby phoneme and mora units are both activated when processing Katakana pseudo-words.
