Abstract
In this study, letter search task was used to investigate the process to extract sound from Japanese characters. Participants were instructed to find a character with given pronunciation among eight characters placed randomly on the screen (character set) and to judge whether the character with given pronunciation is exist or not. Character sets were constructed in four conditions; (1) kanji characters with pronunciation which correspond to many kanji characters ( VH-kanji ), (2) kanji characters with pronunciation which correspond to few kanji characters ( VL-kanji ), (3) hiragana characters with pronunciation which correspond to VH-kanji condition (VH-hiragana), and (4) hiragana characters with pronunciation which correspond to VL-kanji condition (VL-hiragana). The results showed that (1) Mean RT is shorter for hiragana trials than kanji trials, (2) Mean RT is shorter for VL-kanji trials than VH-kanji trials, and (3) Mean RTs are not differ for VL-hiragana trials and VH-hiragana trials.
