Distribution of the Dravidian languages. All maps in this post are from Wikipedia.
There is not a consensus on the linguistic roots of the eighty-five or so Dravidian languages, all but a few of which are restricted to the Deccan Peninsula of India and the adjacent island of Sri Lanka. The three main divisions of the Dravidian languages are Southern Dravidian exemplified by Tamil, Central Dravidian exemplified by Telugu, and Northern Dravidian comprised of Brahui, Kurukh and Malto and related languages. The Northern Dravidian languages are spoken in linguistic islands remote from the core Dravidian area. But, there are indications that these pockets are the result of migrations, rather than a much greater historic range of the Dravidian languages. As Wikipedia explains:
Only two Dravidian languages are exclusively spoken outside India, Brahui and Dhangar, which is related to Kurukh. . . . The Brahui, Kurukh and Malto have myths about external origins. The Kurukh have traditionally claimed to be from the Deccan Peninsula, more specifically Karnataka. The same tradition has existed of the Brahui. They call themselves immigrants. Many scholars hold this same view of the Brahui such as L. H. Horace Perera and M. Ratnasabapathy. . . . The Brahui population of Balochistan has been taken by some as the linguistic equivalent of a relict population, perhaps indicating that Dravidian languages were formerly much more widespread and were supplanted by the incoming Indo-Aryan languages. However it has now been demonstrated that the Brahui could only have migrated to Balochistan from central India after 1000 CE. The absence of any older Iranian (Avestan) loanwords in Brahui supports this hypothesis. The main Iranian contributor to Brahui vocabulary, Balochi, is a western Iranian language like Kurdish, and moved to the area from the west only around 1000 CE.
This is not to say that that Dravidian languages didn't once have a wider geographical range. From the same source:
Dravidian place-names throughout the regions of Sindh, Gujarat and Maharashtra suggest that Dravidian languages were once spoken throughout the Indian subcontinent. [citing George Erdösy (1995), The Indo-Aryans of ancient South Asia: Language, material culture and ethnicity, p. 271 and Edwin Bryant, Laurie L. Patton (2005), The Indo-Aryan controversy: evidence and inference in Indian history, p. 254.]
Maharashtra and Gujarat are the two states of India on the map above which are contiguous with the Dravidian linguistic area of modern India on the subcontinent's Western coast.
Sindh is the most Southeastern province of Pakistan and is immediately adjacent to Gujarat.
There is borrowing from Dravidian into Vedic Sanskrit by at least the middle Rig Vedic period and there is likewise evidence that some Indo-European Indic languages spoken in India were influenced by large numbers of Dravidian language speakers who shifted linguistically to Indic languages.
While the Dravidian language are sometimes speculated to be native to India, they do not have the time depth to support a wide distribution for a very long time. As explained in the Wikipedia article on the Dravidian Urheimat (i.e. homeland):
Historical records suggest that the South Dravidian language group had separated from a Proto-Dravidian language no later than 700 BCE, linguistic evidence suggests that they probably became distinctive around 1,100 BCE, and some scholars using linguistic methods put the deepest divisions in the language group at roughly 3,000 BCE. Russian linguist M.S. Andronov puts the split between Tamil (a written Southern Dravidian language) and Telugu (a written Northern Dravidian language) at 1,500 BCE to 1,000 BCE.
Southworth identifies late Proto-Dravidian with the Southern Neolithic culture in the lower Godavari River basin of South Central India, which first appeared ca. 2,500 BCE, based upon its agricultural vocabulary, while noting that this "would not preclude the possibility that speakers of an earlier stage of Dravidian entered the subcontinent from western or central Asia, as has often been suggested."
Some estimates put Dravidian family divergences from Proto-Dravidian as late as 500 BCE.
Both Egypt and Sumeria have written documents that go back to 2500 BCE, but India was further afield than those records covered. The earliest Dravidian written records that survive date to the 7th century BCE. The absence of earlier written records, however, suggest that any outsiders who brought cultural innovation to India around the time that the Dravidian languages arose probably did not have a command of either the Egyptian or Sumerian writing systems. Since those systems were restricted to a narrow professional class of priest-scholar-bureaucrats at that time, rather than a widely literate population, the absence of writing that early in a group of people who may have come from afar by sea may not mean much.
There are other indications of a young time depth for Dravidian languages as well: "The Dravidian languages form a close-knit family – much more closely related than, say, the Indo-European languages. There is a fair degree of agreement on how they are related to each other."
Dravidian may very well predate Indo-Aryan languages in South Asia, but probably only by 1,000 to 1,500 years, and probably predates Indo-Aryan languages by a shorter time in Western India and Sindh.
The Genetic Evidence
The Dravidian languages are far younger, for example, than the autosomal genetic component described as "Ancestral South Indian" (ASI), which suggests genetic links to the Onge people of the Andaman Islands probably from 20,000 years ago or more, and far younger than the private South Asian mtDNA and Y-DNA haplogroups. Neither "Ancestral North Indian" (ANI), nor Ancestral South Indian automsomal DNA signatures show any greater affinity to African autosomal genetics than any other non-African population.
Distribution of Y-DNA Haplogroup T
As I have noted earlier in a post at Wash Park Prophet, the only apparent genetic marker with a distribution that is suggestive of the geographic range of the Dravidian languages, centered on the Proto-Dravidian area while being largely absent from most areas that lack evidence of a Dravidian substrate, and seems to be intrusive to India rather than autochronous, is Y-DNA haplogroup T.
While dating the origins of patrilineal and matrilineal haplogroups is a dicey proposition, the concurrence of phylogeny, the distribution of descendants of what appear to be autochronous Indian haplogroups, and mutation rate based dates for both Y-DNA and mtDNA haplogroups on both a relative and absolute basis, all strongly suggest that many of those haplogroups have origins in South Asia from the Upper Paleolithic era.
Distribution of Y-DNA haplogroup L
Notably, Y-DNA haplogroup L, a somewhat close relative of Y-DNA haplogroup L which has much greater time depth than the Dravidian languages, has a range that closely approximates the Indus River Valley but is virtually absent in the proto-Dravidan area at about the midpoint of the modern Dravidian range. Y-DNA haplogroup R2 has a distribution (and time depth) similar to that of Y-DNA haplogroup L.
This genetic evidence suggests that Dravidian is probably not a product of Harappan colonization. An absence of Harappan colonization is further suggested by relatively thin trade ties, through a couple of Eastern India coast trade outputs, between South India and Harappans when they were contemporaries, and by an absence of Harappan crops in early South Indian agriculture. This is also suggested by Witzel's finding that the oldest Rig Vedic texts do not show evidence of a Dravidian substrate, despite the fact that they do show familiarity with the Harappan region.
Y-DNA haplogroup T is very common in the Horn of Africa, found in North Cameroon Fulani language speaking populations, in Egyptians and the Sudanese, in Iraq, and at low frequencies in much of the Middle East and Europe, and in Jews, as well as in certain African populations that claim Semitic roots. A best guess to Y-DNA haplogroup T's origins would place it in Mesopotamia from a time period no more recent than the Neolithic and perhaps older. In Africa, haplogroup T looks like a comparatively recent arrival compared to African-specific haplogroups and with only a few exceptions like the Lemba and North African Cameroon Fulani, is found almost exclusively in Afro-Asiatic language speakers.
In the Horn of Africa, where haplogroup T is present at a high frequency, the climate is appropriate for Sahel crops, Bantu expansion was present at or not too long after the proto-Dravidian period, and maritime trade with India would be plausible, the principle African substrate Y-DNA haplogroups are A and B, which are strongly East African, as well as having associations with Khoisan and Pygmy populations, but are not characteristic of Niger-Congo speaking populations which are predominantly Y-DNA haplogroup E1b1a. Around 2500 BCE, the Horn of Africa was probably predominantly Afro-Asiatic language speaking, unless early proto-Swahili or very early wave Bantu expansion population (or isolated pre-Bantu colonies of Niger-Congo peoples) were present.
However, there is no evidence of mtDNA L, M1 or V haplogroups, which are predominant in contemporary Africa, that are not traceable to migrations to Indian in the historic era, long after the Dravidian languages developed. Likewise, there is no evidence of Y-DNA haplogroups A, B, E, or R1b-V88 which are predominant in Africa, in the Dravidian era that are not traceable to historic era migrations.
More Linguistic and Cultural Clues
There was maritime trade between Sumeria and the Indus River Valley in the time period when Dravidian probably originated, although the deep water Austronesian sailors who settled Madagascar had not arrived on the scene at the time and sailors ca. 2500 BCE probably attempted to stay within sight of the coast in shallow waters. But, the possibility of long distance sea travel to the East coast of India from the Horn of Africa, Egypt, or Mesopotamia is not precluded by technology. There is not, however, any evidence of maritime trade between West Africa and the Mediterranean or around the Cape of Good Hope from West Africa to the Indian Ocean at that time, and there is not strong evidence for maritime trade along the North African coast in that era.
Bernard Sargent has noted that Dravidian culture has a number of similarities, in Neolithic era home construction methods, musical instruments, religious ideas, inheritance rules and even games, in addition to its founder crops, with Niger-Congo language speaking Sahel farmers. He has also called attention to linguistic similarities, in terms of both root word cognates and grammatical elements, between Niger-Congo languages and Dravidian languages.
A look at the Internet based World Atlas of Language Structures (WALS) shows that Dravidian languages are indeed greatly different by a wide variety of measures from the other major language families of Europe, India and Asia: Indo-European, Austro-Asiatic, Tibeto-Burman, Austronesian, Thai-Kadai, and Hmong-Mien.
There are some similarities to Niger-Congo languages, but the similarities are greatest not to the core and exemplary Niger-Congo languages. Instead, two of the most similar Niger-Congo languages are on the fringe of the Niger-Congo linguistic area bordering and probably receiving areal influences from Afro-Asiatic languages. These are Wolof, of Senegal, and Swahili (a Bantu language with considerable Afro-Asiatic influence as a regional trade language), of East Africa. Sargent cites Mande, another Niger-Congo language on the fringe of the Niger-Congo linguistic area in West Africa that borders Afro-Asiatic linguistic areas as showing similarity to Dravidian. This is suggestive of language learner or near creole effects that dampened particularly distinctive Niger-Congo language related aspects of a putative Niger-Congo source for proto-Dravidian.
Efforts to link Dravidian linguistically to Sumerian, which is well attested in writing in the time period of proto-Dravidian, or Afro-Asiatic languages have also failed. There have been more fruitful efforts to link Dravidian languages to the Uralic languages, but those connections are controversial.
One possibility that could make sense to fit this evidence is that Y-DNA T bearing male colonists brought an early Bantu language, perhaps proto-Swahili, that they acquired culturally through language shift together with the Sahel Neolithic package, but that these men were few enough in number in the early population that their descendants spoke a creolized version of this language that was influenced by a substrate that might have had some structural similarities to Uralic and might have borrowed words through a BMAC language perhaps (or alternately, and perhaps more plausibly, Uralic might have some distant roots in this substrate language).
Indeed, since the Dravidian language expansion appears to have been perhaps somewhat less complete in its replacement of prior cultural strata than Indo-Aryan languages, which even when they were influenced by prior strata were sometimes influenced themselves through Dravidian, proto-Dravidian may still provide the strongest clues to the nature of the pre-Neolithic languages of most of India, even though Dravidian itself probably has a strong intrusive component to its origins.
In my view, Dravidian probably expanded as a component of the South Asian Neolithic which featured African Sahel founder crops and cultural parts of the Sahel Neolithic package beyond crops. This Neolithic package could thrive in conditions where the Harappan civilization of the Indus River Valley did not expand because its Fertile Crescent origin crops did not thrive there. This package probably brought by a predominantly male group of individuals who were the bearers of Y-DNA haplogroup T to India who probably arrived midway up the eastern coast of India by sea.
Dravidian shows signs of being a relatively young language that had many new language learners in its formative period. There is strong evidence that as a proto-language, it had intrusive elements and was not predominantly native to India, although there may have been an autochronous substrate. But, a narrative that can make sense of just what the nature of those intrusive elements were, or that can connect these intrusive elements to a specific historic cultural community, is elusive.
One of the most promising avenues for finding these links would be to study at a high level of detail, subhaplogroups of Y-DNA haplogroup T and determine which world populations are most strongly phyologenetically linked to the Indian forms of this haplogroup. This work has largely been accomplished for non-African bearers of haplogroup T, but not for haplogroup T in India.