How Chinese as a foreign language learners use generative AI for oral script-writing: A qualitative perspective on cognitive scaffolding in project-based learning

This study investigates how learners of Chinese as a Foreign Language (CFL) interact with generative artificial intelligence (GenAI) tools to support oral script-writing within a project-based learning (PBL) environment. Positioned at the intersection of cognition, motivation, and AI mediation, the research examines both learner perceptions and the challenges of integrating GenAI into classroom practice. A qualitative approach was adopted, drawing on focus group interviews, AI-assisted student scripts, and teacher observation logs. Thematic analysis revealed four core themes: GenAI as a cognitive scaffold, GenAI-induced cognitive load, creativity and efficiency in script preparation and over-reliance and usability challenges. Results indicate that learners' proficiency levels shaped their engagement with GenAI. Higher-proficiency students engaged with outputs criticall evaluated AI outputs, refining and creatively adapting suggestions, while lower-proficiency peers relied more heavily on AI's linguistic accuracy yet often struggled to adjust content for contextual and communicative appropriateness. Teacher observations highlighted peer collaboration as an essential compensatory mechanism, particularly for less proficient learners, in refining AI-generated scripts. Nonetheless, a persistent gap remained between AI-assisted writing and spontaneous oral performance, underscoring the limitations of GenAI in fostering real-time communicative competence. This study contributes to the growing body of research on AI in language education by demonstrating how GenAI can simultaneously scaffold and constrain learning within PBL contexts. The findings emphasize the importance of differentiated scaffolding, the development of AI literacy, and instructional designs that combine AI-assisted writing with opportunities for authentic oral interaction.