Developing preservice teachers’ TPACK for ChatGPT integration in CLIL: a longitudinal study in authentic classroom settings

This study investigates how an 18-week cyclical teacher education program supported Taiwanese Content and Language Integrated Learning (CLIL) preservice teachers’ (PSTs’) development of Technological Pedagogical Content Knowledge (TPACK) for integrating ChatGPT into authentic elementary classrooms. Grounded in Mishra et al. TPACK framework, this multi-method case study involved 48 PSTs who participated in three iterative teaching cycles. These cycles focused on using ChatGPT for lesson planning (teaching support) and facilitating students’ direct use of ChatGPT for CLIL task completion and inquiry (learning support). Data sources included lesson plans, ChatGPT interaction logs, reflective journals, student work, and classroom observations. Findings revealed a clear developmental trajectory in ChatGPT integration, progressing from lower- to higher-order tasks, from one-way tutor-like use to collaborative partner-like engagement, and from text-based to multimodal learning experiences. TPACK analysis showed a strong emphasis on technological pedagogical knowledge (TPK), with PSTs demonstrating increasing confidence in leveraging ChatGPT’s generative and social affordances. However, limited development was observed in contextual knowledge (XK), particularly in addressing ethical, sociocultural, and institutional considerations. The study contributes to emerging GenAI scholarship by providing longitudinal, classroom-based evidence and demonstrating that structured training supports the meaningful integration of ChatGPT into teaching, with clear implications for future teacher education.