Assessing ChatGPT Feedback for EFL Learners’ Engagement and Writing Performance

ABSTRACT Grounded in engagement theory and feedback literacy, which emphasize the interplay of emotional, behavioral, and cognitive dimensions in learning to write, the present study examines the comparative effectiveness of peer feedback and ChatGPT‐generated feedback in the context of English as a foreign language (EFL) writing. We utilized a quantitative method to examine the comparative effectiveness of peer feedback and ChatGPT‐generated feedback in the context of EFL writing. The participants were 174 university students enrolled in a mandatory writing course in China. The control group, consisting of 90 participants, utilized only conventional peer feedback, whereas the treatment group, with 84 participants, incorporated feedback generated by ChatGPT, employing specifically crafted prompts to assist in their writing revisions. A validated survey on engagement was conducted at the start and conclusion of the semester. The participants also completed a writing test. They also completed an internal writing test, focusing on five sections: filling in blanks, paraphrasing, sentence construction, paragraph writing, and essay composition. The findings revealed that students in the ChatGPT group experienced significant improvements in affective (estimate = 0.45) and behavioral engagement (estimate = 0.29), as well as notable advancements in their writing performance (“group × time” interaction estimate = 4.19). The study concludes with a discussion of the pedagogical implications for incorporating AI‐generated feedback tools into EFL writing instruction.