한국어 쓰기 피드백을 위한 ChatGPT의 활용 가능성 연구

초록

Purpose: The purpose of this study is to compare and analyze the feedback generated by ChatGPT and instructors, seeking to explore ChatGPT’s potential and applications as a tool for Korean writing feedback. Methods: Nineteen informational texts and eighteen opinion-based texts written by academic-purpose learners were evaluated using both analytic and holistic feedback by three instructors and ChatGPT. Results: First, ChatGPT demonstrated high intra-rater reliability (ICC) through repeated evaluations, and also showed satisfactory inter-rater reliability (IRR) with an instructor having over ten years of experience. Second, ChatGPT assigned higher analytic scores than human raters for both text types. Specifically, significant differences were found in “content and task fulfillment” and “citations and references” for informational texts, and in “content and task fulfillment” and “language use” for opinion-based texts. Third, in holistic feedback, ChatGPT tended to provide generic comments, while instructors offered more specific guidance tailored to the learners’ writing levels. Conclusion: As an auxiliary tool for first-stage feedback, ChatGPT can enhance both the effectiveness and efficiency of Korean writing feedback.

키워드