L2 Self-Guides, Achievement Emotions, and Engagement in AI-Mediated Informal Digital Learning of English: Insights from Structural Equation Modeling and Psychological Network Analysis

The rapid integration of artificial intelligence (AI) into informal digital learning of English (IDLE) has expanded learning opportunities, yet little is known about how learners’ motivational self-concepts and achievement emotions jointly shape engagement in such contexts. This study, drawing from control-value theory, aims to investigate the dynamic interplay between L2 self-guides, positive emotions, and engagement among university EFL learners in AI-mediated IDLE. A mixed-analytic approach was employed, combining structural equation modeling (SEM) to test theoretically specified pathways with psychological network analysis (PNA) to uncover the dynamic interconnections among variables. Data were collected from 852 Chinese university EFL learners. SEM was conducted in AMOS 26.0, with normality testing performed in SPSS 27.0, while PNA in R 4.2.0 was used to further explore the network structure among the variables. SEM results showed that the ideal L2 self significantly predicted enjoyment, hope, pride, and engagement, whereas the ought-to L2 self failed to directly predict pride and engagement. Enjoyment, hope, and pride all positively predicted engagement. PNA further revealed engagement as the central hub of the system, closely connected with enjoyment, hope, and pride, with hope emerging as the strongest link. By contrast, the ought-to L2 self appeared at the periphery, exerting only weak associations with other variables. These findings highlight engagement as a central driver of AI-mediated IDLE, while demonstrating the broader applicability of control-value theory and offering pedagogical insights for sustaining learner motivation.