Advancing Interactive English Learning through Mixed Reality

Abstract

Traditional English instruction has long been constrained by limited interactional forms, the absence of authentic learning contexts, and low efficiency in skill transfer, thereby hindering second language acquisition (SLA). Mixed reality (MR), with its capacity for multimodal interaction, offers a potential solution to these limitations. This study investigated the scenario adaptability and skill specificity of MR-based interaction, drawing on SLA theory and multimodal interaction theory. A randomized controlled trial with a mixed-methods design was conducted with 90 non-English-major university students over a 16-week intervention. One control group received traditional multimedia instruction, while three experimental groups engaged in MR-based interaction designs: (a) virtual characters with multimodality, (b) spatial scenarios with peer collaboration, and (c) personalized, multimodal interaction. Data were collected through language proficiency assessments, interaction behavior coding, immersion scales, and semi-structured interviews and were analyzed using repeated-measures analysis of variance (ANOVA) and thematic coding. Results showed that MR-based interaction significantly improved English proficiency, learning motivation, and perceived immersion compared with traditional instruction, with effects strengthening over time. Distinct patterns of interaction–skill adaptability emerged: virtual character interaction yielded the greatest gains in speaking, spatial scenarios most enhanced listening, and personalized adaptation produced the strongest improvements in writing. Immersion and depth of meaning negotiation fully mediated the relationship between MR interaction and learning outcomes. Learners with lower initial proficiency benefited more, and designs with lower cognitive load further amplified learning effects. A three-dimensional interaction format–scenario–skill adaptability model was developed, offering a mechanistic account of how multimodal interaction operates within SLA and providing a precision-oriented design paradigm for technology-enhanced language education. The findings also deliver theoretical and practical foundations for the scalable deployment of MR in educational settings.