抄録
This study conducted exploratory practical research focusing on conversational strategies used by sixth-grade elementary school students at a national elementary school and a public elementary school when interacting with “School AI”. As a result, at School A, a national elementary school where the average vocabulary size was relatively high, students tended to sustain interaction with AI by using English as much as possible without requesting support in Japanese, and they were able to respond flexibly to the AI’s improvised utterances. In contrast, at School B, a public elementary school where the average vocabulary size was at a standard level, students repeatedly used formulaic expressions from the textbook. The repeated use of “School AI”, appropriately tailored to the unit objectives, appeared to support the steady consolidation of basic vocabulary and expressions and may have contributed to increased confidence in presentations and enhanced learning motivation. These findings suggest that by designing prompts tailored to learners’ actual proficiency levels, it is possible to facilitate AI-mediated conversations aligned with unit or lesson objectives. Furthermore, through interaction with generative AI, students were observed to sustain dialogues with AI by making full use of their available linguistic resources, regardless of vocabulary size. In addition, AI-generated feedback was found to encourage a certain degree of learner autonomy. This implies that, with teachers proactively adjusting prompts, generative AI can function as a supplementary pedagogical tool rather than a replacement for instruction.
https://ror.org/00khh5r84 