Effects of Domain, Modality, and Linguistic Complexity on Cognitive Load and Self-Efficacy in Learning Academic Content in L2

Learning academic content in a second language (L2) depends on how content domain, presentation modality, and linguistic complexity jointly shape cognitive and motivational states. Using a repeated-measures design, we examined how task domain (science versus history), presentation modality (video versus text), and linguistic complexity (simple versus complex) jointly shaped cognitive load, self-efficacy, and performance accuracy in 36 Chinese-speaking learners of Japanese. Mixed-effects models revealed a key asymmetry: cognitive load was primarily driven by task features, whereas self-efficacy reflected both task features and disciplinary expertise. Key three-way interactions revealed that expertise moderated the effects of intrinsic load, with only science majors benefiting from low linguistic complexity on science tasks. This expertise also created a motivational buffer, as only science majors’ self-efficacy was sensitive to in-domain linguistic complexity. Furthermore, interactions showed that the effects of modality were conditional on the task’s combined conceptual and linguistic demands. Section-level analyses showed that momentary increases in load were associated with immediate drops in self-efficacy, especially for the science passages. Performance analyses showed a more negative load-accuracy slope for complex than simple syntax, and self-efficacy predicted accuracy chiefly for history and simple syntax. Bayesian mediation suggested a small negative indirect effect overall; the effect was credible for simple passages but not complex. These findings highlight the need for adaptive instruction that manages cognitive demands and supports motivation, calibrated to domain, modality, linguistic complexity, and learner expertise.