現在，文章生成や画像生成をはじめとする生成AIと呼ばれるサービスが注目され，様々な用途での活用が期待されている．本研究では，日本の中学校を対象とした英語教育支援のためのロボットサービスと教師業務知識ルールベースを組み合わせたシステムにより，英語の授業時における対話練習を通じて各単元で習得すべき能力を学習者がより意欲的に身に着けることができる学習環境の実現を目指している．本稿では，従来は教師，および学習者自らが作成していた対話練習用の対話文について，大規模言語モデルによる対話練習用の候補文章生成と学習者レベルに合わせた適切な言い換えを比較的小規模なドメイン特化型の言語モデルによって行う対話練習教材の作成支援手法を示す．

Currently, services called generative AI, such as text generation and image generation, are attracting attention and are expected to be utilized in various applications. In our study, a system that combines a robot service and teacher business rule bases has been developed to support English education for junior high schools in Japan to help learners acquire the skills they need to acquire in each unit through dialogue practice during English classes. Then, the system aims to create a learning environment where students can become more motivated. In this paper, a method to support the creation of dialogue practice materials by integrating a large language model and a domain-specific language model, although the sentences for dialogue practice are ordinary created by teachers and learners themselves in current classroom situations. In this method, to generate general dialogue sentences according to the teacher’s demand, a large-scale language model is utilized firstly. Then, to adjust the level as the suitable for the leaners, a relatively smaller scale language model is used with paraphrasing the generated sentences.