Baasha Seva — Imagining Multilingual Digital India

Below is an imaginative draft of India Stack based solution to “language problem” of India. Feedback welcome.


  • Even though there are 22 official languages of India, for practical purpose, English / Hindi are only 2 dominant languages offline.
  • Thanks to early digital adoption issues, English is only language used online.
  • Digital India cannot reach its milestones without localization and talking to people in their language, both online and offline.

Imagine an India, where digital India solves language puzzle for millions of migrants, linguistic diversity is encouraged through digital India, powered by India Stack, creating a stronger language economy.

Imagine the below possibilities.

  • Bihari Migrant worker in Kerala would be able to read, understand govt documents / private forms, marketing material in Hindi (language of choice), powered by India Stack.
  • A Telugu person traveling in TamilNadu gets a train ticket in Telugu, while same ticket can be checked by Odia ticket checker, if tickets are digital through Indian railways app, which has localization layer powered by India Stack.
  • An IT professional in Bangalore, gets to read local body election manifesto in English.
  • All state governments websites are bi-lingual / tri-lingual by default. Additional translation available on request.


  • Document translation (Forms, Tickets, Small metadata translation through semi-automated platform)
  • Translation through Translation service providers. (TSP)
  • Software / Web / App localization through Localization service providers. (LSP)

How? – Document translation

  • Digital documents – Certificates, tickets, forms which would be xml / json and have document IDs of govt / private[transactional] across all namespace must be posted to India Stack.
  • Any new govt document gets added is available for translation.
  • 1st pass automated translation through data dictionary.
  • 2nd pass verification by govt TSP.
  • Any app / site using the document, can get localizations in available languages calling the API of LSP specifying the language required.


  • Are set of language experts, can serve a particular language.
  • Can be individuals working from home / companies, creating digital jobs to bilingual.
  • Anyone can request translation of any content (any web page, newspaper article) via TSP, pay for the services through UPI.


  • Are software localization experts, provide / operate IT infrastructure for TSPs.
  • LSPs enable localization of digital goods for govt, private use.
  • Work on R&D on Indic computing / localization, new digital products like machine translation platform, voice recognition, OCR etc. Can use the data generated by TSPs for R&D.

Pilot Project

Possible Projects

  • truly local UPI (Localize NPCI layer of UPI using govt TSP, while mandating PSPs to localize, showing that TSP model is possible).
  • Digital Train tickets (IRCTC tickets localized in 22 languages)
  • Baasha Seva can be a vertical solution, cutting across India Stack products.


  • Banking / Financial / Telecom sector has existing mandatory localization regulations to inform customers about products in their own language. Encouraging TSPs will lower cost of localization as more players will enter market.
  • Pass a Digital India law to mandate *all consumer websites / software / app*, available as per 3 language policy.

Social Side effects

  • Job creation through Skill India, people need to have language + digital skills to be employed by TSPs. This can be on Uber model too, where people work as per convenience, get paid by UPI
  • Encouraging research investment on Indic computing through LSPs.
  • GoI, truly respecting language as right and all 22 languages as official language.


  • If a 1.0 is possible, TSPs can expand scope of translating academic texts, legal, medical, policy information into languages, making knowledge accessible in all languages.

Leave a Reply

Your email address will not be published. Required fields are marked *

%d bloggers like this: