Comparing Peer Feedback and Generative Artificial Intelligence Feedback in Japanese English as a Foreign Language Speaking Context: Impacts on Motivation, Engagement, and Writing Self‐Efficacy

ABSTRACT This study explores the comparative effectiveness of peer feedback and generative AI (GenAI) feedback on Japanese university students learning English as a Foreign Language (EFL), specifically examining their motivation, engagement, and writing self‐efficacy during script‐writing preparation for speaking tasks. Acknowledging the close relationship between speaking and writing, as well as the essential role of feedback in second language acquisition (SLA), the research contrasts the impact of peer feedback during the first semester with GenAI feedback, particularly through ChatGPT, in the second semester. The primary objective was to assess whether students experienced increased motivation, enhanced writing self‐efficacy, and greater engagement after transitioning to GenAI feedback, while also investigating their perceptions of using ChatGPT. A survey was conducted with 147 first‐ and second‐year university students at three intervals throughout the academic year to evaluate these factors. Results revealed a statistically significant rise in students’ ideal L2 self following the introduction of GenAI feedback. Writing self‐efficacy improved significantly after peer feedback and remained high after the shift to GenAI feedback. Additionally, students reported notably higher affective and behavioral engagement with GenAI feedback compared to peer feedback, although cognitive engagement did not show significant differences. Qualitative responses highlighted benefits such as immediate error detection and improvement suggestions, alongside concerns regarding over‐reliance and the absence of human interaction. The findings indicate that GenAI feedback can effectively enhance motivation and both affective and behavioral engagement in EFL speaking preparation, serving as a complementary tool to traditional feedback methods. This study underscores the pedagogical implications of thoughtfully integrating GenAI tools into EFL instruction, emphasizing the importance of careful prompt design and instructor support while recognizing the limitations of the current study's design.