Construction of a Japanese Language Learning Support System that Enables Word Accent Learning

In second language learning, learners learn four skills: reading, writing, listening, and speaking. However, compared with the other three skills, speaking skills have rarely been studied in environments wherein they can be learned. Therefore, we proposed a method for recognizing Japanese word accents using the pitch average of the whole word and that of each mora as part of speaking skill learning. We demonstrated that this method outperforms the pitch drop detection method. Using the proposed accent recognition method, we embedded a function that enabled learners to learn the accents of target words into the feedback phase of an existing Japanese language learning support system. This allowed learners to speak the target words and receive feedback on whether their accents were correct.