Journal Paper Published

Abstract

Generative artificial intelligence (GenAI) is rapidly expanding within L2 education and is increasingly positioned as a core resource for lesson planning. This study explores how 22 pre-service teachers mobilized multimodal GenAI tools, including DALL-E for image generation and ChatGPT for instructional support, to plan Grades 3–4 Chinese-as-a-second-language (L2) vocabulary lessons in Hong Kong primary schools. Using the TPACK framework, we examined how their pedagogical, content, and technological knowledge interact during this process. Lesson artefacts, reflective forms, and semi-structured interview responses were analyzed. Results showed that participants actively employed a range of multimodal GenAI functions across instructional stages, with image-generation tools being the most frequently used (34 instances). They were aware of the Chinese character features and pupils’ cognitive and cultural characteristics to produce culturally attuned teaching materials. Rather than static overlap, content, pedagogical, and technological knowledge continuously shaped one another through iterative negotiation. Theoretically, the study operationalizes the TPACK framework into a GenAI-supported, multilingual primary setting and illustrates how the three knowledge domains co-evolve through recursive negotiation. Practically, it underscores the value of collaborative design, systematic cultural vetting, and the deliberate integration of interactive chatbots to enhance L2 vocabulary instruction.

Keywords