Speaker
Description
Language teachers have built interactive tools for their own learners for decades, and generative AI now allows teachers with no programming background to produce working applications from natural-language prompts. Research on this practice, however, is scattered across constructs that rarely refer to one another, such as pedagogical authorship, teachers as end-user developers, tool-making confidence and educational content sovereignty, and most studies report teachers’ perceptions or learners’ outcomes; few describe the making itself. Drawing on a structured evidence map of 128 research reports (2010–2026) and a close reading of 28 studies, this paper proposes an analytic framework for describing and comparing teacher-developed interactive tools. The framework has four dimensions: agency of making (who decides how the tool behaves, and who carries out those decisions), specification capability (how pedagogical intent becomes behaviour a tool can execute), model involvement (whether AI takes part in making the tool, sits inside the product, or both), and artefact lifecycle (what happens to the tool after it is built). Two contrasting cases, one built with a human programmer in 2016 and one built with AI coding assistants in 2026, illustrate how the framework separates questions that the literature tends to merge. The paper closes with four propositions for research on teacher agency in AI-mediated tool development.