Share of OTT content (films + series, high/medium confidence, top 150 by language) where caste is a central or significant plot element. Bengali is the reference category.
Tamil OTT content foregrounds caste at 10.6× the Bengali rate. Hindi at 8.3×. Even Telugu and Marathi, whose caste-themed OTT output is modest, are 4.7–4.9× more likely to carry a caste narrative than Bengali productions. Bengali is the only language where zero items were coded "yes" — caste as a central theme — in the entire top-150 corpus.
Classification by Claude Haiku 4.5 via AICredits. Prompt explicitly excludes surname inference and incidental occupational mention; requires a plot event (conflict, discrimination, violence, untouchability, inter-caste marriage) as evidence. Manual overrides applied for 24 confirmed caste-themed titles from academic literature and critical reviews where TMDB synopses were insufficient.
Caste theme rate (yes + partial) broken down by media type within each language. Bengali maintains zero for web series and near-zero for films across both formats.
Web series show systematically higher caste-theme rates than films in Hindi and Tamil, suggesting that the shorter-format OTT series model has been a vehicle for more socially critical content in those industries. Bengali web series — the format most directly comparable to the Hoichoi/Zee5 Bangla ecosystem — show zero caste-themed content across the entire corpus.
Annual caste-theme rate per language, 2014–2024. Years with fewer than 3 items are excluded to avoid single-item noise. The Bengali line is flat at zero throughout.
Tamil shows the most pronounced inter-annual variation, with peaks in 2018, 2020, and 2022–2023 corresponding to waves of Pa Ranjith and Mari Selvaraj-adjacent anti-caste content on OTT platforms. Hindi shows modest activity concentrated around 2018. Bengali shows no movement in any year across the full decade. The flat Bengali line is not a statistical artifact of small samples — the corpus contains 7–14 Bengali items per year from 2014 to 2023.
Breakdown of caste narrative types among content coded "yes" or "partial". Bengali has one entry: "upper_caste_power" — a background signal, not a thematic event.
| Language | Type | N |
|---|
Tamil leads on dalit_protagonist (6 codings), caste_discrimination (5), and caste_violence (4) — the narrative forms most directly associated with assertive Dalit cinema in the Pa Ranjith tradition. Hindi's top type is upper_caste_power (6), reflecting the crime-drama framing of series like Paatal Lok and Mirzapur. Marathi shows untouchability and honour_killing, consistent with the Fandry/Sairat tradition. Bengali's single caste_type coding ("upper_caste_power", partial confidence) does not represent a narrative about caste — it represents a Brahmin family whose privilege is background texture, not the subject of the story.
Detailed breakdown of Bengali OTT content across the corpus. Every Bengali item, its classification, and what the LLM found.
Across 134 Bengali OTT productions spanning 2014 to 2024, not one item was classified as having caste as a central narrative theme. The single "partial" coding — 0.7% — represents a film where an upper-caste family's social position is background texture rather than the subject of narrative scrutiny. Bengali cinema and web series, across both film and serial formats, across the full decade of the OTT expansion, produced no content in which caste relations, caste discrimination, caste violence, inter-caste conflict, or Dalit experience is foregrounded as a plot driver. This silence is not random noise. It is structural.