Prompt Engineering as Mediation: Investigating AI Chatbot-Assisted Writing Process From an Activity Theory Perspective

Current discussions on Artificial Intelligence (AI) in foreign language learning have shifted from evaluating tool utility to exploring effective pedagogical practices. Drawing on Activity Theory (AT), this study investigates how Chinese EFL (English as a Foreign Language) undergraduates strategically deploy prompt engineering to mediate chatbot-supported expository writing tasks when autonomously completing assignments. Data triangulation includes student-chatbot interaction logs, writing artifacts, and stimulated recall interviews. Findings reveal that prompt engineering acts as a critical mediator in dynamic activity systems, shaping writing outcomes through iterative human-AI co-construction. Multilingual prompting strategies, found to be a common practice in the study, mainly involve using L1 for prompt engineering and making translation requests to address L1-L2 disparities. By situating AI chatbot-assisted writing within activity systems, this research advances theoretical understanding of technology-mediated language learning and offers practical insights into integrating prompt engineering into EFL writing curricula.