2013 年 50 巻 p. 1-21
The study builds upon the results of a previous experiment assessing Japanese adolescent's ability to decode pronunciations in an L2 with a much less regular orthography (English) to obtain empirical data that may offer implications for the English-language education of Japanese children. Because English orthography is relatively opaque, learning how to decode it takes more time than for many languages, and instructions are reported to be effective on the development of decoding skills in LI-English children. However, study of Japanese EFL learners is sparse. Here, children (fifth and sixth grades) and adolescents (eighth and ninth grades) read words containing target spellings (<ai>, <au>, <ou>, <u>), heard the pronunciations, and repeated them in training sessions. Decoding automaticity and errors before and after training were measured by naming pseudowords containing the target spellings. Improvements were seen in the adolescents' automaticity, indexed by coefficient of variation of naming latency, and in both groups' error rates. A larger decrease in romaji-based errors after the application of a corrective stimulus suggests that the ability to use one decoding system facilitates the learning of another. While training of this sort seems more effective for adolescents, other strategies may be needed to raise accuracy and automaticity.