Cognitive–affective pathways to AI-assisted informal English-speaking participation in translanguaging EMI contexts: an explanatory sequential mixed-methods study

Artificial intelligence (AI) has expanded opportunities for spoken English practice beyond the classroom, yet learner participation remains uneven. Using an explanatory sequential mixed-methods design, this study examined the associations of digital competence, cognitive load, and foreign language enjoyment (FLE) with Chinese EMI students’ AI-assisted informal speaking participation. Survey data from 386 students were analysed using structural equation modelling (SEM), followed by thematic analysis of interviews with 10 purposively selected survey participants. SEM indicated that digital competence was positively associated with AI-assisted informal speaking participation, both directly and indirectly through FLE; a sequential indirect association through cognitive load and FLE was also significant. Cognitive load was not directly associated with participation, and the indirect association through cognitive load alone was non-significant. The interview findings suggested that manageable and enjoyable AI interaction did not necessarily coincide with oral practice. Participants most often spoke when they had an immediate disciplinary need, whereas they could use AI to clarify concepts without sustained oral production. Translanguaging was described as supporting disciplinary meaning-making and preparation for English speaking but did not consistently culminate in oral rehearsal. Together, the findings highlight a distinction between speaking participation from broader AI use and identify cognitive, affective, and situated factors associated with such participation.