<![CDATA[This study investigates the linguistic characteristics and pedagogical depth of written corrective feedback (WCF) provided by a human lecturer versus ChatGPT-4 on undergraduate EFL writing tasks. Employing a descriptive-comparative content analysis method, the study evaluated five student argumentative essays alongside corresponding manual lecturer annotations and standardized ChatGPT feedback logs. The findings indicated a substantial qualitative disparity, the lecturer’s feedback functioned primarily as direct corrective feedback, relying on visual shorthand markup (strikethroughs, insertion arrows) to deliver localized, surface-level fixes without metalinguistic explanations. Conversely, ChatGPT functioned as an interactive formative writing tutor, offering detailed metalinguistic explanations, full-sentence reconstructions, formal vocabulary substitutions, and macro-level discourse guidance, including active voice recommendations, redundancy removal, and transitional flow improvements. The study concluded that while ChatGPT excels in comprehensive, line-by-line syntactic and stylistic feedback, it lacks the human empathy and contextual intuition of an instructor. Consequently, the paper advocates for a blended, hybrid feedback model in EFL pedagogy: students utilize ChatGPT as an automated pre-submission screening tool for lower-order mechanical corrections, allowing human instructors to focus their feedback on higher-order priorities such as argument coherence, critical thinking, and authentic voice.]]>
Copyrights © 2025
