AI-assisted task-based language learning and English proficiency: a dual-pathway model of cognitive engagement and self-regulated learning with AI trust as a boundary condition

Purpose: Artificial intelligence is increasingly transforming language learning environments, yet the psychological mechanisms through which AI-assisted instructional practices influence language outcomes remain insufficiently understood. Although prior studies have examined AI-supported learning, limited research has explained how AI-assisted task-based learning activates distinct cognitive and behavioral processes that contribute to English proficiency in Chinese higher education contexts. Therefore, this study examines how AI-assisted task-based language learning influences English proficiency among university students in China by integrating cognitive engagement and self-regulated learning as parallel mediating mechanisms, with AI trust functioning as a moderating boundary condition. Methods: Data were collected from 412 undergraduate students enrolled in Chinese universities and analyzed using partial least squares structural equation modeling. Results: The results reveal that AI-assisted task-based learning significantly enhances both cognitive engagement and self-regulated learning, which subsequently improve English proficiency. Notably, self-regulated learning demonstrated substantially stronger direct and mediating effects than cognitive engagement, suggesting that AI-assisted learning environments may operate more effectively through behavioral regulation processes. Furthermore, AI trust strengthens the positive influence of AI-assisted learning on English proficiency. Implications: By integrating instructional design, learner cognitive processes, and trust in artificial intelligence within a unified analytical framework, the study advances a mechanism-based explanation of AI-enabled language learning and highlights how AI technologies can enhance educational effectiveness in higher education language learning environments.