AI-Supported Translanguaging Processes in Primary School: Empirical Insights into ChatGPT’s Role in Multilingual Interactions

Abstract This study investigates how artificial intelligence (AI) can support multilingual learners’ translanguaging processes and the activation of multilingual resources in primary education. Drawing on the Design-Based Research project KiKo-Prim, the article analyses interaction sequences between multilingual learners, teachers, and ChatGPT. Combining Design-Based Research with interpretive interaction analysis, the study provides exploratory insights into student–teacher–AI interactional dynamics, focusing on how AI was embedded as an integral component of the instructional design, how it scaffolded multilingual explanation practices, and how it mediated mathematical problem-solving and digital media production activities. The dataset comprises 462 conversational turns and 339 min of small-group video recordings from n = 13 learners in Grades 3–4. The findings suggest that ChatGPT can function as a 'multilingual buddy', supporting vocabulary retrieval, the refinement of explanations, and the production of multilingual learning artefacts, even in groups without shared multilingual repertoires. Productive AI use, however, depended on guided prompt-engineering practices, teacher facilitation, and the purposeful integration of AI into subject-specific learning environments. Based on these results, the article proposes initial design principles and pedagogical conditions for AI-supported translanguaging, with implications for inclusive, linguistically responsive teaching in digitally mediated primary classrooms.