Investigating the association between AI-driven personalized learning platform quality and English vocabulary acquisition among Chinese EFL university students

This study examines the association between AI-driven personalized learning platform quality and English vocabulary acquisition among Chinese EFL university students, with learner engagement as a mediator and vocabulary-learning self-efficacy as a moderator. A one-group pretest-posttest design was combined with a post-intervention survey. The 356 undergraduates completed a vocabulary pretest, used AI-supported tools for four weeks, completed a posttest, and responded to measures of platform quality, engagement, self-efficacy, and vocabulary acquisition. Partial least squares structural equation modeling assessed reliability, validity, direct effects, mediation, and moderation. Composite reliability ranged from 0.901 to 0.916, average variance extracted ranged from 0.646 to 0.726, and heterotrait-monotrait ratios were below 0.85. Platform quality was associated with engagement (β = 0.610, p < 0.001) and acquisition (β = 0.339, p < 0.001). Engagement was associated with acquisition (β = 0.360, p < 0.001) and mediated the platform-acquisition relationship (β = 0.219, p < 0.001). Self-efficacy showed a direct association (β = 0.240, p < 0.001) and a moderating effect (β = 0.075, p = 0.032). Vocabulary scores increased, but the design does not establish causation. The findings position AI personalization as a delivery mechanism for retrieval, spacing, feedback, and contextual vocabulary use.