Integrating teacher and AI-generated feedback to enhance EFL academic writing: A mixed-methods study of performance and perceptions

Abstract

Academic writing presents considerable challenges for English as a Foreign Language (EFL) learners, particularly in mastering higher-order skills such as idea development and organization, as well as lower-order skills including vocabulary, grammar, and writing mechanics. Although instructor feedback provides valuable contextual and rhetorical support, it is frequently constrained by limited time and heavy workloads. In contrast, AI-based feedback offers prompt and detailed language corrections but lacks the capacity for in-depth conceptual and rhetorical evaluation. This study investigates the impact of combining lecturer and AI-generated feedback on EFL students' academic writing achievement and examines their perceptions of this hybrid approach. An explanatory sequential mixed-methods design was implemented with 25 fifth-semester EFL students. Participants completed a pre-test, participated in an integrated feedback intervention, took a post-test, and subsequently responded to a questionnaire and semi-structured interviews. Quantitative findings revealed a statistically significant increase in overall writing scores from 67.56 to 79.48 (p < 0.001), with a very large effect size (Cohen's d = 1.78). Improvements were evident across all analytic writing components. Qualitative results indicated positive student perceptions, with lecturer feedback identified as conceptually clarifying and AI feedback as technically beneficial. Overall, the integrated feedback model effectively strengthened both linguistic accuracy and rhetorical development in EFL academic writing.