Peer Review or AI Feedback in EFL Writing Instruction: Experimental Outcomes in Student Writing Revision

Abstract

This mixed-methods research investigated the comparative efficacy of two main types of feedback, AI-generated feedback and peer review, in developing the writing revision among EFL university students. Grounded in sociocultural and feedback literacy frameworks, the research involved 80 intermediate-level (CEFR B1–B2) learners who completed argumentative essays and revised them after receiving either peer review or ChatGPT-5 feedback. Quantitative results showed that peer feedback led to significantly greater improvements in coherence and vocabulary, especially for B2 learners, while AI feedback yielded greater gains in grammatical accuracy. Gender also moderated insights: female students appreciated peer feedback’s affective and motivational support, whereas male students conveyed distinguished trust in AI’s reliability. Interviews further exposed that peer interactions raised engagement, while AI was valued for its direct, accurate corrections. The findings emphasized the corresponding strengths of each feedback modality and proposed a mixed method, supporting EFL practitioners to purposefully combine AI and peer feedback to support both language accuracy and higher-order writing skills. These results encourage functional, research-informed strategies for teachers seeking to develop the motivational impact and quality of writing instruction in varied EFL classrooms.