Tone, engagement, and positioning in EFL writing feedback: Human touch versus AI responses

Abstract

Feedback plays an essential role in English as a Foreign Language (EFL) education, particularly in academic writing. As artificial intelligence (AI) language models like ChatGPT become increasingly integrated into EFL education, it is important to explore the extent AI-generated feedback can mimic feedback produced by teachers, especially in terms of tone, engagement and learner positioning. The present qualitative study draws on the Appraisal Theory by Martin and White (2005) to analyze and compare the linguistic characteristics of feedback provided by EFL teachers and ChatGPT on student academic essays. More specifically, the research explores authorial attitude (affect, judgment, appreciation), and engagement (dialogic openness and reader positioning). The findings reveal that teacher feedback is richer in digressions, comments and the variety of engagement features, resulting in more interactive and engaging feedback. The study contributes to discussions on the effectiveness of AI-generated feedback, offering suggestions for pedagogical implications for integrating ChatGPT into EFL writing instruction.