The design of tasks to induce learning from audiovisual input is vital for second/foreign language teachers and has attracted a good deal of attention in Computer-Assisted Language Learning (CALL) research. This study presents a new contribution to research in this area by investigating learner engagement in the context of vocabulary learning through dual-subtitled video viewing and examining the correlations between different types of learner engagement and vocabulary gains across different learning conditions. This study investigates 72 university EFL learners' behavioural, emotional, and cognitive engagement in four different learning conditions, which feature the presence/absence of vocabulary preview and/or postviewing interventions. Both between-group and within-group ANOVA results revealed significant differences in the level and type of engagement across the four learning conditions. Behavioural and cognitive engagement were the strongest in learning conditions requiring vocabulary preview and postviewing L2 production in the learner-generated content (LGC) condition, while the strongest emotional engagement was exhibited in learning conditions involving vocabulary preview and postviewing L2 production in the teacher-generated content (TGC) condition. Learner engagement at all three levels was moderately low in learning conditions where only vocabulary preview was included and was even lower in learning conditions when both previewing and postviewing interventions were absent. The article ends with discussions on how understanding learner engagement manifested in viewing dual-subtitled video conditions under different learning conditions may guide pedagogical decisions in the context of L2 learning through such audiovisual materials.