VietTESOL International Convention 2026

Generative artificial intelligence is reshaping language education, yet many pre-service teachers initially utilize it as a basic search engine rather than a collaborative pedagogical partner. This mixed-methods case study investigates how seven pre-service English teachers at a Vietnamese university conceptualize and practice prompt engineering to design communicative speaking tasks. Following an instructional workshop on prompting strategies, data collection incorporated instructional design tasks, chat logs, written reflections, and semi-structured interviews, guided by technological pedagogical content knowledge and the Include, Develop, Evaluate, and Apply prompt engineering frameworks. The findings reveal a profound cognitive shift: participants transitioned from a passive search mindset to identifying themselves as pedagogical architects responsible for actively orchestrating artificial intelligence. Quantitatively, a strong positive correlation was found between participants’ prompt engineering competence and the pedagogical quality of their resulting speaking activities. Qualitatively, while generative tools demonstrated remarkable efficiency in drafting materials and differentiating multi-layered scaffolding for mixed-ability learners, they consistently produced logical inconsistencies and generated scenarios lacking genuine communicative conflict. To ensure activities adhered strictly to task-based language teaching principles, the teachers had to exercise strong pedagogical agency. They accepted a critical trade-off, sacrificing generation speed to iteratively evaluate and manually refine the synthetic outputs. Ultimately, this study underscores that prompt engineering constitutes a core pedagogical act. Consequently, integrating formal artificial intelligence literacy into teacher education programs is imperative to foster professional autonomy, critical evaluation skills, and responsible technology integration.