Integrating Large Language Models in Foreign Language Teaching: Prompting Advanced English Grammar

Abstract

This chapter discusses how Large Language Models (LLMs) like ChatGPT can be used in modern foreign language teaching. It focusses on teacher training in integrative approaches that do not see LLMs as supplementary learning, but as part of the central teacher–learner interaction, including autonomous learning in flipped classroom approaches. It is not an empirical study with students, but an empirical play with prompts, the decisive starting point of successful human–machine/artificial intelligence/chatbot interaction. It emphasizes the potential of LLMs for teaching and learning without neglecting ethical concerns and practical challenges. It provides examples from ChatGPT 3–5 and Bard Experiment, now Gemini, to illustrate how LLMs can be used to generate explanations, construct exercises with model answers and even correct exam tests for complex grammatical English constructions such as article usage and modal auxiliaries. This study illustrates where LLMs can be integrated into teacher training to help students in autonomous or teacher-mediated learning, but it also shows clearly that users always have to be well-informed and remain skeptical about LLM suggestions and that even LLMs in their current free versions only provide suggestions that may only be used in revised or adapted form.