Negotiating Politeness and Agency in AI‑Mediated Discourse: A Digital Pragmatics Study of ESL Learners’ Speech Acts and Meta-Pragmatic Awareness

The fast adoption of the generative Artificial Intelligence (AI), especially ChatGPT, in higher education have altered the way ESL students communicate in the academic sphere. Though most of the past studies have mainly concentrated on writing results, little has been done on the practical processes of learner AI interaction. It is a qualitative research paper that examines how speech acts are built in ESL environment, to identify negotiation of politeness strategies and learner agency in a ChatGPT-mediated discourse by undergraduate ESL (BS English) students. The study is based on an interpretivist paradigm and qualitative case study research design based on the Speech Act Theory, Politeness Theory, and SLA views of pragmatic competence and agency. The information was gathered using AI mediated interaction tasks, qualitative Discourse Completion Tasks (DCTs) and semi structured interviews utilizing purposive sampling. Three qualitative tools were used in the analysis, which were speech act coding, politeness strategy analysis, and thematic analysis. The results suggest that the learners mostly employ directive speech acts, which are often reduced with modalization and hedging. The politeness was not only habitual but also strategic as it implied relational positioning and AI perceived authority. Recurrent re-prompting, critical evaluation, and responsibility perseverance provided through interview data revealed active agency of the learners and metapragmatic awareness. The paper concludes that communication based on AI transforms pragmatic practice without disfigurement of learner control and reflective participation.