Language-Aware Learning Paths from Open Video: Curating YouTube Lessons for Hindi-Medium and Multilingual Learners in India — Methods and Early Evidence from YUNOVA

This preprint describes YUNOVA (https://yunova.in), a free web platform that turns open YouTube videos into ordered, language-aware learning paths for Hindi-medium and multilingual learners in India, and reports descriptive data from its curriculum and first recorded activation funnel. It documents a skill–module–topic curriculum model with a rule that repeated videos cause topics to merge; a two-track design separating "just learning" from a certificate track with Elementary, Intermediate and Expert levels and a camera-monitored final test; and a spoken-language check that reads the language of YouTube's automatic caption track instead of uploader tags. Across 22 language-teaching skills, 6,004 videos were checked (5,386 with an ASR track), and uploader tags disagreed with ASR spoken language in 38.3% of tagged placements. The catalogue holds 266 skills and 17,358 topics; trending guides offer 243 lessons in six languages. A single-week funnel is presented descriptively, with no causal claims.