Evidence page · counted from the dictionary
A Dravidian language is spoken in Balochistan. It shares 191 roots with Tamil.
Brahui is spoken in the hills of Balochistan in Pakistan — roughly two thousand kilometres northwest of Tamil Nadu, beside the Indus. It is a Dravidian language. That single fact does more to unsettle the standard geography of South Asian language history than any argument, and it is checkable: counted directly from the Dravidian etymological dictionary, 191 of Brahui's 262 recorded roots have Tamil cognates. This page publishes the full comparison, and the caveat that keeps it honest.
The count.
shared etymological roots with Tamil, by language
Each Dravidian language in the dictionary is indexed to shared root numbers. Counting how many of each language's roots also appear in Tamil gives a direct measure of shared inheritance — no reconstruction, no interpretation, just set overlap.
the 191 roots, sortable
| Language | Region | Roots | Shared | % of its roots found in Tamil | |
|---|---|---|---|---|---|
| Malayalam | Kerala | 2,444 | 2,162 | 88% | |
| Toda | Nilgiris | 1,014 | 837 | 83% | |
| Koḍagu | Kodagu | 898 | 732 | 82% | |
| Kota | Nilgiris | 1,188 | 953 | 80% | |
| Brahui | Balochistan, Pakistan | 262 | 191 | 73% | |
| Tulu | coastal Karnataka | 2,158 | 1,531 | 71% | |
| Telugu | Andhra / Telangana | 2,526 | 1,787 | 71% | |
| Kannaḍa | Karnataka | 2,791 | 1,964 | 70% | |
| Gondi | central India | 899 | 540 | 60% | |
| Kui | Odisha | 919 | 490 | 53% | |
| Malto | Rajmahal hills, Jharkhand | 693 | 358 | 52% | |
| Kuṛux | Jharkhand / Bengal | 751 | 376 | 50% | |
The page's number is contested by its own caveat. Putting both in one inspectable table is the honest presentation.
Counted across 26 language sheets. Full table and the underlying root sets ship with the site as dedr_roots.json.
The caveat that has to come with it.
why the percentage misleads and the raw count does not
Read naively, the table says Brahui shares a higher proportion of its vocabulary with Tamil than Telugu or Kannada do. That is a sampling artefact and should not be repeated.
Brahui has only 262 recorded roots here against Telugu's 2,526. A small recorded inventory for a language only about 15% of whose vocabulary is of native Dravidian origin — against roughly 20% Balochi, 20% Indo-Aryan and 35% Persian and Arabic — is weighted toward basic, stable, pan-Dravidian vocabulary — the words most resistant to replacement. Telugu's much larger inventory includes centuries of internal innovation that Tamil never shared. Tested directly: 45% of Brahui's roots fall in the core shared by Tamil, Kannada, Telugu and Malayalam together, against 41% of Telugu's — the two are comparable, and the apparent Brahui "closeness" is an effect of what survives to be recorded.
The meaningful figure is the raw one: 191 roots, shared across two thousand kilometres and an unbroken stretch of Indo-Aryan and Iranian speech. That number needs an explanation regardless of how the percentage is read.
What the distribution means.
a residue, not an origin
Dravidian languages today are concentrated in the south, with outliers scattered north and east: Gondi in central India, Kui and Kuwi in Odisha, Kuṛux and Malto in Jharkhand and Bengal, and Brahui in Balochistan. Two readings are possible, and they are not equally economical.
| Reading | What it requires |
|---|---|
| Dravidian is southern, and Brahui migrated north | A movement of speakers, or of a language, roughly 2,000 km northwest, through and past every intervening population, leaving no trace between origin and destination. |
| Dravidian was once spoken far more widely, and the modern map is what remains | A formerly continuous or near-continuous distribution across much of the subcontinent, subsequently displaced in the centre and northwest — leaving the south as the largest surviving block and Brahui as a stranded pocket of speech. |
The scattered outliers — Jharkhand, Odisha, central India, Balochistan — are what the second reading predicts and the first does not. And it bears directly on a question this platform treats carefully: the deep south's Dravidian character is not evidence that Dravidian originated there. The languages survive there in strength because that is where they were not displaced.
Brahui is spoken beside the Indus region, and this is frequently offered as evidence that the Indus civilisation spoke a Dravidian language. The inference does not follow on its own. The presence of a Dravidian language in Balochistan today is consistent with a Dravidian Indus; it is also consistent with a later arrival, and the dating of Brahui's presence there is genuinely disputed among specialists.
What the root count establishes is narrower and firmer: a Dravidian language exists two thousand kilometres from the Dravidian heartland, sharing basic vocabulary with Tamil. Whatever explains that is a story about a much wider former distribution — which is the necessary precondition for the Indus hypothesis, without being evidence for it. The wider question is treated here.
Roots recorded in Tamil alone.
105 entries, and what they are
A second result falls out of the same count: 105 root numbers appear in the Tamil data and in no other Dravidian language in this dataset. Their content is striking — they are overwhelmingly local flora, fauna and material culture:
| DED | Tamil | Meaning |
|---|---|---|
| 150 | அந்து | a small grey-winged insect found in stored paddy |
| 204 | அரணை | a green house lizard |
| 210 | அரளி | oleander |
| 257 | அல்லி | waterlily |
| 258 | அல்லியன் | a stray elephant separated from the herd |
| 280 | அரிஞ்ஞில் | sage-leaved alangium; the buttress tree |
| 365 | ஆநூணி | bread-fruit tree; jungle jack |
| 238 | அலகு | number, calculation; cowries as counting tokens |
The tempting reading is substrate — words for local species absorbed from an earlier, non-Dravidian population of the deep south. That is a real hypothesis and this kind of vocabulary is exactly where linguists look for it.
But the sober reading is at least as likely: these are simply words for things that only exist where Tamil is spoken. A word for the jungle jack fruit will not appear in Brahui because the tree does not grow in Balochistan. Ecological restriction explains the pattern without requiring a substrate at all, and a set of 105 items dominated by trees, insects and lizards is precisely what ecological restriction looks like. This platform records the set as a starting point for that question, not as an answer to it.
Limits.
what is claimed
- Machine-verified here: all root counts and overlaps, computed by set intersection on shared etymological index numbers across 26 language sheets. Output
dedr_roots.jsonships with the site. - Source: a digitised edition of the Dravidian etymological dictionary data. It is a secondary compilation, and its coverage is uneven by design — the small hill languages have far fewer entries than Tamil or Kannada, and low counts reflect documentation effort as much as linguistic reality.
- NOT claimed: that Brahui is more closely related to Tamil than Telugu is. That reading is a sampling artefact, tested and rejected above.
- NOT claimed: that the Indus civilisation spoke a Dravidian language. The Brahui distribution is a precondition for that hypothesis, not evidence for it, and the date of Brahui's arrival in Balochistan is disputed — Elfenbein (1987) places it as late as the tenth century CE.
- NOT claimed: that the Brahui are a genetic residue of a Dravidian population. They are not. Y-chromosome haplogroups R1a and J account for more than 65% of Brahui male lineages while the Dravidian-associated H and L are rare; autosomally the Brahui are indistinguishable from neighbouring Balochi speakers and genetically distant from central Dravidian speakers; and a 2025 study looking specifically for Northern Dravidian ancestry found no signal of Oraon-related ancestry in Brahui. In Pakistan, genetic relationships track geographic proximity rather than linguistic affiliation, and the Brahui are the standard illustration. What is stranded in Balochistan is a language, not a population — which is what this page’s own argument requires and does not currently say.
- NOT claimed: that the 105 Tamil-only roots are substrate vocabulary. Ecological restriction is the simpler explanation and is not excluded.
- Method limit: shared index numbers indicate cognacy as judged by the dictionary's compilers. This page counts their judgements; it does not independently verify sound correspondences, and a set-overlap count is not a subgrouping analysis.
What this page has not read
the Brahui arrival date turns on a source nobody here has opened
The page's Indus paragraph rides on Elfenbein (1987, 1989), unread. That is the most important fact about its evidence and it should be visible.