5ZM-07
Improving the communication methods of GPT based on the language proficiency of learners
○張 子涵,木實新一(九大)
GPT exhibits significant potential in interactive language learning. Referencing the communicative scenarios of daily life and adjust communication methods not only makes conversations more natural but also provides learners with content matched to their proficiency level. This paper examines the effectiveness of different design prompts and fine-tuning techniques to enable GPT to identify and analyze users’ language proficiency automatically, thereby adjusting communication methods such as sentence length and complexity of grammar or vocabulary. This approach provides learning content more suited to the learners’ current proficiency level.