Not so fast? A comparative study of pre-service teachers’ lesson design using corpora and generative artificial intelligence

Abstract

The integration of corpora and generative artificial intelligence (GenAI) in language teacher education presents both opportunities and challenges. While corpus-based approaches have long been promoted for data-driven learning (DDL), their adoption remains limited due to complexity issues and time-demands. In contrast, GenAI tools offer immediate, user-friendly access to linguistic data, yet raise concerns about authenticity and reliability. This study compares pre-service teachers’ use of corpora and GenAI in pedagogically oriented language analysis, lesson planning, and materials development. Conducted within a graduate-level course, the study examines student teachers’ approaches to corpus-based and AI-based lesson design, focusing on their ability to retrieve and analyse linguistic data, plan lessons, create learning materials, and reflect on the effectiveness of these tools. Findings indicate the considerable potential of both corpora and GenAI for supporting data-informed, inductive approaches to language learning and teaching. Yet, the results also reveal that while pre-service teachers demonstrated operational proficiency in using both tools, they struggled to extract meaningful linguistic insights and integrate their findings into cohesive pedagogical frameworks. The study highlights the need for targeted training to develop teachers’ analytical and pedagogical skills in working with both types of resources. Ultimately, it argues that rather than replacing corpora, GenAI should complement data-driven learning, reinforcing the importance of linguistic accuracy and pedagogical soundness in technology-enhanced language teaching.