Prompt Design as AI-Mediated Pedagogical Performance in Initial Teacher Education: A Case Study in Foreign Language Teaching Methodology

Generative artificial intelligence is creating new possibilities for educational resource development, making prompt design increasingly relevant to interactions with these tools. However, little research has examined how pedagogical criteria are represented in prompts developed within initial teacher education or how prompt design can be conceptualised in relation to teachers’ professional knowledge. This study analyses 15 group-produced prompts and their documented development processes, created by students enrolled in a Bachelor’s Degree in Early Childhood Education during a Foreign Language Teaching Methodology activity involving the creation of children’s stories in English using generative artificial intelligence. An exploratory case study design was adopted, using an eight-dimension analytical rubric addressing pedagogical, linguistic, narrative, and strategic aspects of prompt design, with the corpus independently assessed by three raters. The analysis combined descriptive scores across rubric dimensions with qualitative examination of prompt content and documented revision processes to identify recurring patterns of educational prompt design. Findings showed that educational criteria were generally represented explicitly across the corpus, although differences emerged across dimensions and cases. Strategic development and agency, target learners and language level, and language function and objectives were the most consistently represented dimensions, whereas formal control of the output showed comparatively lower scores and professional contextualisation showed the greatest variability. The combined quantitative and qualitative analysis revealed recurring cross-case patterns in the explicitness and integration of pedagogical criteria rather than discrete competence levels or developmental categories. The findings suggest that prompt design can be examined as a form of AI-mediated pedagogical performance in which educational criteria are translated into operational instructions for generative systems. Rather than demonstrating a wholly independent teaching competence, the study highlights prompt design as a potentially relevant context for mobilising and externalising pedagogical knowledge within initial teacher education. Given the small, context-specific corpus of collaborative prompts, the findings are exploratory and are not intended to support generalisation to individual prompting competence or to other teacher education contexts.