学習支援に向けたai採点型スピーキングテストの活用―日本人中学生における予備的研究―

Drawing on principles of assessment for learning (Phakiti & Leung, 2024), this pilot study explores how test feedback and teacher support influence students’ awareness of their strengths and weaknesses, engagement in learning activities, learning outcomes, and goal setting. Ten second-year junior high school students took the AISATS twice—a speaking test that provides instant feedback generated by artificial intelligence (AI). After the first test, students received both AIgenerated test feedback and individualized guidance from a teacher. They subsequently reflected on their performance, set learning goals, and undertook targeted practice. Following the second test, which took place two days after the first, they engaged in another round of reflection based on the new feedback. Data from test scores, student reflections, and questionnaire responses were analyzed using a convergent mixed-methods design. Findings indicate that score feedback, teacher support, and other score-report-related activities contributed to improvements in students’ vocabulary, grammar, and fluency scores, as well as in their self-assessment ratings. Although the effects were relatively limited, students’ reflections and learning activities were associated with performance gains. These results underscore the indispensable pedagogical role of teachers in amplifying the impact of AI-generated feedback on student learning.