Abstract
Phonological recoding of English words and nonwords by Japanese ESL learners was examined. Forty-eight adults in three proficiency groups named English monosyllabic nonwords and words of two regularity types. It was expected that grapheme-phoneme recoding would lead to regular correct pronunciations on regular nonwords and body-rime recoding to irregular correct pronunciations on irregular nonwords. The results showed high accuracy rates on words and a higher incidence of error responses on both types of nonwords. However, error analyses of responses on nonwords revealed that less proficient readers dominantly used illegitimate sub-syllabic strategy using CV-type segmentation, which may have stemmed from their native language, Japanese.
