Effectiveness of XR in vocabulary learning: A meta-analysis

Extended reality (XR), including virtual reality (VR) and augmented reality (AR), has increasingly been applied in vocabulary acquisition. This meta-analysis synthesized empirical evidence on the effectiveness of XR technologies for second and foreign language vocabulary learning. Following PRISMA guidelines, searches were conducted in Web of Science, Scopus, and ERIC for studies published up to 2026. From 2,160 initially identified records, 29 experimental or quasi-experimental studies met the inclusion criteria. Effect sizes were calculated as Hedges’ g and analyzed using random-effects and three-level meta-analytic models. The primary random-effects model showed a significant moderate-to-large positive effect of XR-based instruction on vocabulary learning, g = 0.775, 95% CI [0.545, 1.005], p < .001. A three-level model produced a comparable estimate, g = 0.773, confirming the robustness of the findings. Publication bias diagnostics indicated funnel asymmetry, but trim-and-fill analysis showed that the adjusted effect remained significant. Moderator analyses revealed that exposure frequency significantly influenced learning outcomes, with sustained or repeated XR exposure producing stronger effects than short exposure. Technology modality and assessment timepoint were not significant moderators. These findings suggest that XR can support vocabulary acquisition and may facilitate vocabulary retention over time, particularly when integrated through repeated instructional engagement.