Language Vitality Beyond Speaker Counts: A Large-Scale Application of the LAR-II Framework across Fifty Indian Languages

India is home to an estimated 780 spoken languages, of which the vast majority are underdocumented, unscheduled, and endangered at various degrees of severity. The standard instrument for assessing which languages are in danger — speaker population count — turns out, when systematically tested, to be a deeply misleading guide to actual structural vitality. A language with 10 million speakers can be more fragile than one with 200,000. A language officially categorised as 'Safe' can be on the verge of generational rupture. And a language with fewer than 50,000 speakers can exhibit structural conditions more resilient than several of its million-speaker neighbours. These counter-intuitive but empirically grounded observations form the analytical centre of this paper. We present the largest application of the LAR-II quantitative endangerment framework to date, covering 50 Indian languages across five ecological and demographic zones: North-Himalayan (10 languages), Northeast (10), East-Central tribal belt (10), West-Central (10), and South (10). All variables are drawn from Census of India 2011 data with transparent proxy documentation. The LAR-II formula — EV = (NoS / Lv) × (Literacy / 100) × (1 + Growth / 100) × LTI / Area — produces a continuous vitality index for each language that is then normalised to a 0–100 scale and inverted into a Risk Score. The results demonstrate three principal findings. First, the Pearson correlation between speaker count (NoS) and LAR-II vitality index is 0.84 (p < 0.001), which appears strong but is overwhelmingly driven by extreme-case anchoring at the distribution tails: very large scheduled languages at one end, small critically endangered languages at the other. The Spearman rank correlation is substantially lower at 0.588 (p < 0.001), and the partial Spearman correlation between LTI and vitality controlling for NoS is 0.707 (p < 0.001) — meaning that once the scale effect of speaker count is removed, intergenerational transmission is a considerably stronger predictor of vitality than population size alone. Second, 23 of the 50 languages (46%) exhibit Speaker-Rich Distortion: they have large or very large speaker populations but receive LAR-II Risk Scores above 80, identifying them as structurally fragile despite their demographic bulk. The most extreme case is Bhili/Bhilodi — with 10.4 million speakers, India's most linguistically diverse tribal language by variation count — which receives a LAR-II Risk Score of 98.62, near the highest in the dataset. Third, 41 of 50 languages (82%) receive Risk Scores above 90, and 29 of 50 (58%) fall below a vitality index of 5 on the 0–100 scale, demonstrating that the Indian indigenous and tribal language landscape as a whole is in a condition of structural precarity that speaker-count-based monitoring dramatically understates. The paper argues that the three structural dimensions of language vitality — demographic size, variation fragmentation, and intergenerational transmission — are not interchangeable, not predictable from each other, and not capturable by any single categorical label. Effective language planning in India requires all three dimensions to be tracked independently and simultaneously.