ChatGPT’s Corrective Feedback on Korean College Students’ Writing and Its Pedagogical Implications

This study analyzed error patterns in the English writing of Korean EFL learners using ChatGPT and examined their pedagogical impli- cations. A total of 150 errors were identified from student writing samples, with the most frequent involving grammar errors followed by awkward phrasing, collocation, and structural errors. Many of these errors reflected Korean linguistic features such as the absence of articles and plural markers, as well as reliance on literal translation and limited exposure to authentic English. To address these challenges, the study proposes a multi- dimensional approach that combines explicit grammar instruction, contrastive analysis, data-driven learning, and context-based practice. A key contribution of this research lies in demonstrating the effectiveness of AI-powered tools like ChatGPT: it not only supported learners in recognizing and understanding errors but also provided instructors with diagnostic insights for targeted teaching. While AI cannot replace human feedback, it shows strong potential as a supplementary resource for improving accuracy in EFL writing.