Artificial intelligence-enhanced self-paced learning in English as a foreign language: A nonlinear dynamic and self-determination framework for optimizing postgraduate linguistic proficiency—a mixed-methods randomized controlled trial

This empirical study examines the effectiveness of artificial intelligence-enhanced personalized language learning interventions—tailored content adaptation, self-paced progression, and targeted weakness intervention—on academic writing proficiency, vocabulary acquisition, and grammatical accuracy among 540 Chinese postgraduate learners of English as a foreign language. Utilizing a pretest–posttest randomized controlled design, participants were assigned to three experimental groups (each receiving a distinct artificial intelligence-driven intervention) or a control group (traditional instruction). Quantitative analyses (multivariate analysis of covariance/analysis of covariance) demonstrated significant improvements across all learning outcomes for artificial intelligence groups compared with controls ( p < .001), with the most pronounced effect sizes observed in the self-paced learning group (Cohen’s d = 2.71–16.43). Posttest indicated this group were substantially higher (academic writing: M = 7.09; vocabulary: M = 57.36; grammar: M = 48.21) compared with controls ( M = 5.16, 29.72, 18.22, respectively), accounting for 68.3–98.0% of variance (partial η²). Thematic analysis of qualitative data (surveys, interviews) corroborated these findings, with 89% of participants in the self-paced group reporting enhanced autonomy and reduced cognitive load, in alignment with self-determination theory. Triangulation revealed robust associations between quantitative gains and qualitative themes of motivation, adaptability, and learner satisfaction. These results highlight the superiority of artificial intelligence-driven self-paced learning over static interventions, challenging the prevailing emphasis on content adaptation as the primary driver of proficiency. The study offers practical recommendations for integrating AI to optimize learner-centered language education while addressing technical and pedagogical challenges.