Digital resilience in wartime education: the role of online platforms in sustaining english language learning

Abstract

This study addresses a clear problem: wartime disruption in Ukraine forces English language instruction for engineering students to operate under power cuts, connectivity loss, displacement, and elevated cognitive stress, so digital resilience becomes a core educational capacity rather than a bonus. We examine how specific platform configurations and teaching routines sustain continuity, equity, and assessment integrity in this context. Using a mixed-methods design, we link LMS analytics, videoconference metadata, gradebooks, short student and instructor surveys, and reflection notes collected across 2022-2025. We operationalize outcomes through four indices: continuity index, engagement stability, outcome robustness, and equity gap. Quasi-experimental contrasts compare blackout-heavy weeks with more stable weeks, and interrupted time series capture the effect of interventions such as offline-first content bundles, compressed audio with transcripts, and tri-channel mirroring. Redundancy and offline-first design are associated with higher continuity and lower volatility in attendance and submissions, particularly during clustered outage windows. Assessment flexibility that combines versioned tasks, rubric-linked feedback, and brief oral verification preserves integrity without grade inflation and stabilises criterion-level scores. Short, predictable check-ins and rotating micro-orals sustain engagement more effectively than long live lectures under air-raid alerts and curfews. Device-light alternatives, 72-hour submission windows, and cautious data practices narrow equity gaps for phone-reliant, displaced, or sheltering students while respecting safety constraints. We conclude with a practical wartime playbook for technical universities that institutionalises redundancy, offline-first access, assessment flexibility, and privacy-by-design, and we outline a forward research agenda on causal effects under real outage telemetry, long-term retention, and ethical AI scaffolding for ESP tasks.