A Study on the Comparison of English Learners’ Composition, ChatGPT’s Feedback-based Revised English Composition, and Native English Teachers’ Feedback-based Revised English Composition: A Coh-Metrix Analysis

The purpose of this study is to investigate the differences among students' original English compositions, native English teachers’ feedback-based revised English compositions, and ChatGPT’s feedback-based revised English compositions in terms of vocabulary, syntax, discourse, and readability using Coh-Metrix analyses. The study involved fifteen students who provided their original English compositions. These compositions were then revised after feedback from both native English teachers and ChatGPT to analyze the variations in the revised versions. The research results indicated statistically significant differences among the three types of compositions in three vocabulary indices and one syntax index. However, there were no statistically significant differences among the three types of compositions in all indices of discourse and readability. The findings highlight precise areas where human revisions and ChatGPT-assisted revisions both diverse and converge. Pedagogical implications on these findings will be provided. (Pusan National University)