Reframing Hybrid Feedback in Japanese University EFL Writing: Complementary Roles of Peer Feedback and Generative AI

Abstract

The present classroom-based exploratory study re-examines hybrid feedback in a Japanese university EFL writing course for science majors. Thirty-six students used both peer feedback and ChatGPT-based feedback during a 12-week semester. Drawing on end-of-course questionnaires and a focused analysis of drafts, peer comments, AI feedback, and revisions, the study asked how students perceived the roles of peer and AI feedback and what pedagogical implications follow for hybrid feedback design. The findings suggested a complementary rather than competitive relationship. Peer feedback primarily supported content development, organization, reader awareness, and collaborative engagement, whereas generative AI primarily supported grammar, wording, and immediate local revision. When combined, the two sources appeared to encourage iterative revision and confidence; however, some students also experienced overload and difficulty prioritizing multiple comments. The paper argues that the value of hybrid feedback lies not in replacing human interaction with AI but in assigning distinct pedagogical roles to different feedback sources. It further suggests that the next stage of research should focus less on broad source comparison and more on comment quality, uptake, and AI-supported comment coaching. The study contributes a cautious classroom-based account of how generative AI can be integrated into EFL writing pedagogy in ways that preserve the educational value of peer interaction. Based on these patterns, the paper proposes a provisional four-stage framework for coordinating peer, AI, student, and teacher roles.