Indic dialect geography nests linearly inside each language's subspace

Kumar J et al. fine-tune Whisper-base and Wav2Vec2.0-base on 60 Indic languages/165 districts under joint versus language-only supervision [kumarj-etal-2026-analyzing-language-and-geographical-variation-in-speech-representations-across-60-indic-languages] Language-conditioned logistic-regression probes decode district structure within each language's subspace (Whisper-base mean F1 91.29 joint vs 80.89 language-only; Wav2Vec2 87.14 vs 77.05) [kumarj-etal-2026-analyzing-language-and-geographical-variation-in-speech-representations-across-60-indic-languages] A normalized conditional mutual-information analysis reveals an asymmetric hierarchical geometry (AUC_D|L=2.68 vs AUC_L|D=5.43) with district clusters nested inside language structure [kumarj-etal-2026-analyzing-language-and-geographical-variation-in-speech-representations-across-60-indic-languages]