ChatGPT를 활용한 영어글쓰기 문법 오류분석: 한국대학생 사례 연구 = The Use of ChatGPT for Grammatical Error Analysis in EFL Writing: A Case of Korean University Students

다국어 초록 (Multilingual Abstract)

This study examines the use of ChatGPT in analyzing grammatical errors in English writing. The purpose is to explore the effectiveness and accuracy of ChatGPT’s error analysis in comparison to human raters. Seventy-seven paragraphs written by Korean university students were analyzed by ChatGPT and two human raters based on fifteen error types. The results showed disparities in error frequencies between ChatGPT and human raters. The only similarity in error frequency identified was that both ChatGPT and human raters marked word choice error as the most frequent. However, ChatGPT detected sentence structure and informal language errors more frequently than human raters. Furthermore, ChatGPT’s analysis on individual students’ error frequency differed from that of human raters. In addition to the quantitative result of error frequency, the researcher conducted a qualitative analysis of ChatGPT’s error analysis. The results demonstrated that ChatGPT sometimes identified errors that did not exist in the students’ original texts and failed to consider the context of the original text. In many instances, ChatGPT exhibited inconsistencies in determining the appropriate error types. The findings of this study suggest that, while ChatGPT represents state-of-the-art technology, educators should be mindful of its strengths and limitations when integrating it into English writing instruction.