Linking Big Five personality traits to L2 willingness to communicate with GenAI chatbots: A mixed-methods study of the mediating role of regulatory focus.

Generative artificial intelligence (GenAI) chatbots are increasingly used for second language (L2) speaking practice, yet learners' willingness to communicate (WTC) with them remains insufficiently understood. Building on MacIntyre et al.'s (1998) Pyramid Model, this study examined how the Big Five personality traits and regulatory focus jointly shape L2 WTC in GenAI-mediated interaction. Survey data from 426 Chinese university learners were analyzed using Partial Least Squares Structural Equation Modeling, complemented by qualitative thematic analysis. Openness emerged as the only trait directly predicting WTC and was positively associated with both promotion and prevention focus. Agreeableness positively predicted prevention focus, whereas Extraversion showed a negative association; Conscientiousness and Neuroticism were unrelated to regulatory focus. Prevention focus, but not promotion focus, significantly predicted WTC. Indirect effects indicated that Openness and Agreeableness enhanced WTC through prevention focus, while Extraversion exerted a negative indirect effect. Qualitative analysis identified four themes, with psychological safety and controllability explaining why prevention-focused learners favored GenAI. Together, the findings suggest that GenAI's de-socialized, low-risk interaction environment reshapes how personality and motivational dispositions translate into L2 WTC, offering implications for both L2 pedagogy and the design of GenAI-supported speaking tools.