Several studies show that different types of GenAI can be successfully used in language learning. GenAI is important, but human direction, intervention and quality assurance are perhaps more needed than ever before.

GenAI is everywhere, is affecting everything, and it is not going away. Furthermore, GenAI is a powerful tool in second-language teaching if used intelligently. GenAI is bringing about radical changes in the way we teach students in second-language classes, but we need new theoretical frameworks for how we apply GenAI. This article presents a lexicographically inspired AI prompting framework, which helps second-language learners and second-language teachers use GenAI in specific AI-assisted learning activities.

So, the question is whether lexicography combined with human teaching, interaction, empathy, feelings, and world knowledge still have a role to play in second-language learning and teaching? As will appear from this article, this is very much the case. In fact, it is argued that the human element (staged and facilitated by lexicography as an organizing and architectural science) is a powerful catalyst in AI-assisted second-language learning and teaching.

The research objective of this paper is to present and discuss a GenAI prompting framework illustrating how teachers and students can use GenAI in second-language classes with a particular focus on L2 text production.

The paper draws on empirical data from two studies. The first study from 2021 investigated how students and professionals worked with a first generation GenAI tool. The descriptive-analytical study involved a total of seventy test subjects. The study yielded both quantitative data and qualitative comments, which were thematically analysed by means of Nvivo resulting in a multitude of insights based on the replies by the test subjects.

The second study from 2023 investigated how ten professional translators used ChatGPT 3.5. The ten professional translators first tested ChatGPT 3.5 and were then interviewed by means of a semi-structured open research interview technique. The qualitative interview data were transcribed an analysed by means of Nvivo resulting in several useful insights on the training needs for future communication and translation specialists.

The empirical data and theoretical considerations from three overall theoretical veins of thinking led to the development of a lexicographically oriented GenAI prompting framework designed as a decision support tool that helps teachers teach students how to use GenAI in different learning activity situations. In addition to this decision support tool the paper also presents a model for AI-assisted, lexicographically oriented language exercises to be used in second-language teaching. Finally, the paper presents a RAG-based model for how teachers effortlessly can build their own RAG-based GenAI tool for second-language teaching using own resources.

Based on the analysis of the empirical data and the theoretical considerations, this paper makes the case for a lexicographically oriented model focusing on the strengths of both humans and GenAI. In conclusion, there is no divide between machines and humans. However, machines need operators. Humans.

Keywords: GenAI, prompting framework, AI-assisted learning activities, second-language teaching.