The Impact of ChatGPT on L2 Writing Skill Development and Learners’ Perceptions

Abstract

Following the pandemic, technology was inevitably incorporated into foreign language classrooms, which resulted in significant modifications to the way English is taught to non-native speakers. In particular, practitioners use a range of tools for various purposes when evaluating the written texts produced by L2 learners. The aim of the current study is to investigate whether written feedback generated via one of such tools, namely ChatGPT, could be used to foster L2 writing skill development and to figure out what L2 English learners consider about receiving feedback from ChatGPT instead of a human rater. For the purposes of this research, an embedded mixed design was administered to 46 participants who were recruited via a convenient sampling procedure. Five data collection tools, including a demographic form, an English proficiency test, pre and post writing proficiency tests, a semi-structured focus group interview, and an adapted questionnaire, were utilized to collect both quantitative and qualitative data. Throughout the intervention, participants received feedback for their essays without knowing the source of feedback. Thus, by employing several prompts prior to the study, feedback generated by AI was made indistinguishable from human. Meticulous analyses revealed that ChatGPT had positive impacts on L2 writing proficiency, whereas participants’ perceptions as to the integration of it into assessment were divergent. Despite the fact that they haven’t distinguished between ChatGPT and human rater feedback in many areas, like clarity and engagement, L2 English learners ultimately prefer human rater judgments due to motivational concerns and perceptions regarding the necessity and efficacy of the latter.