An analysis of rhetorical structures in english language teaching research article introduction written by human authors and chat GPT

Abstract

This study analyzed the proficiency of one of the most famous generative AI tools (ChatGPT-4o) in generating research article introductions (RAIs) within the field of English Language Teaching (ELT). Using Swales' Creating a Research Space (CARS) model, the study compared the rhetorical structures of RAIs written by human authors and the ChatGPT-4o. The study used a qualitative design combining genre-based content analysis and rhetorical comparison. It analyzed two sets of 14 Research Article Introductions (RAIs): one from Scopus-indexed ELT journals and one generated by ChatGPT-4o using the same article titles. Through the qualitative research method approach, content analysis identified the moves and steps in both human author and ChatGPT-generated RAIs, followed by qualitative comparison. The results revealed that while ChatGPT adhered to the established rhetorical structure, it overemphasized Move 2 (establishing a niche) and exhibited nuances in Moves 2 and 3 compared to human-written RAIs.