<![CDATA[The increasing use of generative artificial intelligence (GenAI) in English writing has shifted attention from AI-generated products to the ways learners interact with AI during the writing process. However, limited research has examined prompting as an evolving discourse practice in authentic academic writing. This study investigated university students’ prompting strategies, their linguistic and sequential development, their communicative functions, and the ways they reflected self-regulated learning and AI literacy. A qualitative design was employed using Discourse Analysis (DA) and Computer-Mediated Discourse Analysis (CMDA). Data were collected from the ChatGPT interaction histories and semi-structured interviews of 20 master’s students over a three-week academic writing period. The findings identified six recurring prompting strategies: formulation, revision, refinement, adaptation, clarification, and evaluation or confirmation. Students’ prompts became increasingly explicit, contextualized, and sequentially dependent as interactions progressed. Five communicative functions were also identified: requesting information, directing text production, clarifying meaning, negotiating revisions, and evaluating responses. The interaction patterns further revealed recurrent processes of monitoring, evaluation, clarification, and strategic prompt adjustment associated with self-regulated learning and emerging AI-literacy practices. These findings suggest that prompting should be understood not merely as a technical input mechanism but as an evolving discourse practice through which learners negotiate meaning, regulate writing processes, and manage human-AI interaction. The study highlights the importance of integrating prompt literacy into AI-assisted English writing instruction by developing students’ ability to formulate, evaluate, and strategically revise prompts while critically engaging with AI-generated responses.]]>
