The states of India are not arranged by river, by rainfall or by soil. They are arranged by language, and this is unusual enough to be worth stating plainly: very few countries have redrawn their internal boundaries so that each unit is dominated by one language, and none has done it at this scale.
But the sentence "state borders follow language" is only half the story, and the missing half is the interesting one. Language borders are not found in nature. They are drawn, by committees, using categories that someone had to define first. So the map is a product of two decisions: a political one about representation, and a statistical one about what counts as a language.
The order of events
It is often assumed that India had natural linguistic regions and then organised states around them. The sequence ran the other way in an important respect. Before 1956 the internal map was largely inherited from colonial administration — multi-lingual provinces like Madras and Bombay, ruled from capitals that spoke a different language from much of the territory.
The change came from below. In 1952 a Gandhian named Potti Sriramulu began a fast in Madras demanding a separate state for Telugu speakers. He died after 58 days, and the resulting agitation was severe enough that the government created Andhra state in 1953. That concession established a principle: language was a legitimate basis for a state. Three years later the States Reorganisation Act applied it to the whole country, creating linguistic states across the union.
Read that sequence carefully. The linguistic state did not recognise an existing fact of nature. It was conceded after a death, then generalised. Which means the borders came first, and the sense of a language community as a political unit — with its own literature, its own school syllabus, its own film industry — was consolidated afterwards, partly by the institutions the new state built.
The statistics decide the answer
The 2011 census recorded 19,569 distinct mother tongues. These were then rationalised — grouped into languages, and the languages into scheduled and non-scheduled categories. The grouping is where the politics lives.
Hindi, as returned, is 43.6 per cent of the population. But that figure includes Bhojpuri (about 50 million speakers), Rajasthani dialects (about 25 million), Haryanvi, Chhattisgarhi, Magahi and several dozen more, all counted as Hindi because the census schedule says so. A speaker of Bhojpuri and a speaker of standard Hindi from Delhi cannot hold a conversation with any ease. In most European countings they would be separate languages. Here they are one, and the consequence is that the largest language in the country is nearly twice as large as it would be under a stricter rule.
This is not a criticism of the census. Any count has to draw lines, and drawing them the other way would produce its own distortions. The point is methodological: when someone quotes a language share in India, the number is a joint product of demography and administrative classification, and the second is not neutral ground.
What a linguistic state produces
Redrawing states along language lines solved one problem and created another. It gave each major language a unit where it was the majority and could run its own schools, universities and administration. It also turned every other language inside that unit into a minority — sometimes a large one.
Karnataka contains substantial Urdu and Marathi speaking populations. Maharashtra contains Konkani speakers whose language is closer to Goan Konkani than to Marathi, and who have their own recognition claim. Assam contains Bengali speakers who are a majority in some districts and a minority in the state. West Bengal contains Hindi-speaking districts on its western edge. None of this is a failure of the 1956 settlement; it is what any partition of a continuous linguistic landscape into discrete units must produce.
Why this belongs in a pillar about land
Because the language map is the one human map in this pillar that does not follow the physical one. Rainfall, soil and rivers produce gradients; gradients have no borders. Languages are distributed in gradients too — dialect continua, in which neighbouring villages understand each other and distant ones do not. But states need lines, and lines have to be drawn somewhere on a continuum where no natural line exists.
That is the structural fact this entry adds to the pillar. The other six entries describe constraints that people worked around. This one describes a constraint that people created, and then lived inside.
Sources and checks
- Census of India 2011 Language: Table C-16, mother tongues and languages returned, with the rationalisation schedule
- Constitution of India Eighth Schedule; Article 351; Official Languages Act, 1963
- States Reorganisation Act, 1956 Act 37 of 1956, and the report of the States Reorganisation Commission (Fazl Ali Commission), 1955
- Grierson, George Abraham Linguistic Survey of India, 11 volumes, 1903–1928
- Guha, Ramachandra India After Gandhi, chapters on the linguistic reorganisation of states. Picador, 2007
Last checked 30 October 2026 Yellowsea Huang Corrections