Chinese university EFL learners’ essay revision and perceptions of the feedback

A comparison between teacher vs. ChatGPT feedback

Abstract

Revising is an important part of the writing process, especially when writing in a foreign language (FL). FL learners who routinely revise inadequacies in their texts after receiving feedback tend to develop better writing skills than those who do not (Boubekeur, 2015). To help FL learners achieve a desirable revision quality in writing, providing various types of feedback about their writing is of great importance (Lee, 2017). However, as providing detailed and high-quality writing feedback to individual students requires a considerable amount of time and effort on FL teachers, it is not always practical in FL instructions (Myer et al., 2023). In particular, in the context of College English teaching in China, where an English teacher normally teaches 60 to 80 learners, it is not always feasible for English teachers to provide writing feedback to every student on a regular basis. The recent development of generative artificial intelligence (GenAI) such as ChatGPT may either supplement or complement teacher feedback in helping Chinese learners of English as a FL (EFL) revise their English writing (Escalante et al., 2023). The present study aimed to compare: 1) the effect of teacher versus ChatGPT feedback on Chinese university EFL learners’ English essay revision; and 2) Chinese university EFL learners’ perceptions of teacher versus ChatGPT feedback in terms of usefulness and ease of use.

A quasi-experiment was conducted among 73 Chinese university EFL learners. The students came from two intact College English classes (n=36 and n=37 respectively) taught by the same English teacher. A pre-test of essay writing task found that the two classes did not differ on English writing proficiency (F(1, 71)=0.17, p=.68, ?²=.05). One class received teacher feedback whereas the other class received ChatGPT feedback on narrative essay writing. Students were required to revise their essays using the feedback they received during one English class. Upon completion of the essay revision, they were surveyed about their perceptions of the usefulness and ease of use of the feedback by filling out a Likert-scale questionnaire. The results of one-way ANOVAs showed that students who received ChatGPT feedback obtained significantly higher scores of their revision than their peers who received teacher feedback (F(1, 71)=8.74, p<.01, ?²=.11). However, students in the teacher feedback class perceived that the feedback was more useful (F(1, 71)=6.50, p<.05, ?²=.08) and easier to use (F(1, 71)=14.36, p<.01, ?²=.17) than those in the ChatGPT feedback class. To encourage students to use ChatGPT feedback in helping them revise their English writing, it is important to foster positive perceptions of the ChatGPT feedback. Some useful strategies may include: 1) teacher modelling how to use ChatGPT feedback for students, to make students familiar with the format of ChatGPT feedback, which may enhance students’ positive perceptions of ease of use. 2) guiding students to explore different aspects that ChatGPT feedback can cover, which may foster students’ positive perceptions of the usefulness of ChatGPT feedback. The present study only adopts quantitative methods, future research may use semi-structured interviews to provide a rich description of Chinese EFL learners’ experience of using teacher and ChatGPT feedback in their English essay writing.