画像生成AIを活用した初級日本語会話支援の考察

This study aims to compare and analyze the characteristics and educational effects of textbook-based conversation activities (Activity A) and conversation activities (Activity B) using image generation AI in beginner Japanese conversation classes. As a result, Activity A was conducted around a relatively immediate response while imitating the vocabulary and example sentences presented in the textbook, and feedback was mainly limited to verbal modifications by teachers and learners. On the other hand, Activity B was an activity that repeats the process of creating and modifying prompts until learners obtained the desired image, and the generated visual stimulation served as the starting point for language activities. This process induced learners' spontaneous and creative expression and encouraged the experience of constructing utterances while moving back and forth between language and thought. In addition, the multi-step feedback generated during the generation process provided an opportunity for learners to visually recognize their learning progress and sense of accomplishment. Furthermore, the process of creating and correcting prompts in Japanese after organizing thoughts in the native language promoted the application of expressive and grammatical knowledge and showed results that increased learners' subjectivity. From the above, it was revealed that conversation activities using image generation AI have effectiveness as a new teaching and learning model that can foster learners' creativity and subjectivity in beginner Japanese conversation classes.