Designing Immersive Translator Education: A Learning Analytics Case Study and Evidence-Led Virtual Reality Pedagogical Proposal

Background: Translator education requires learners to mobilise lexical, syntactic and semantic knowledge while regulating situated decisions. This study examined how classroom evidence could identify design targets for a prospectively testable virtual reality (VR) pedagogy. Methods: A single-cohort repeated-measures case study involved 26 first-year translation and interpreting students in 2026. Curriculum-aligned pre/post assessments, three digital learning capsules and task-proximal microanalysis were linked using multiplicity-controlled inference and an item-level robustness check using generalised estimating equations (GEE). Results: Total performance changed from 12.04 to 11.27 (mean change = −0.77, p = 0.275), and no individual domain showed a statistically significant pre-post change. Relative change nevertheless differed by linguistic domain, F(2, 50) = 4.43, p = 0.017, partial η2 = 0.151, with lexis more positive than syntax after Holm correction (p = 0.025). Because the parallel forms were blueprint-matched rather than psychometrically equated, and syntax/semantics began with higher baseline scores, the interaction is interpreted as a domain-differentiated pattern rather than a stand-alone learning-gain estimate. The pattern remained supported in an item-quality sensitivity analysis. Capsule performance and task-proximal regulation also differed by domain, and syntax/semantics capsule performance plus semantic microanalysis showed construct-matched associations with corresponding final performance that survived false discovery rate (FDR) correction. Conclusions: The case study identifies course-embedded patterns of difficulty and regulation that informed SCOPE-VR, linking immersive action to explanation, correction, non-VR transfer and a preregistered primary-outcome family. The VR intervention itself remains to be evaluated prospectively.