Of these Tamil and Malayālam can be considered as two dialects of one and the same language, which is, in its turn, closely related to Kanarese. Tulu, Kodagu, Toda and Kōta can be described as lying between Tamil-Malayālam and Kanarese, though they are more nearly related to the latter than to the former. The same is the case with Kuruχ and Malto, while Kui and Gōndī gradually approach Telugu, which latter language seems to have branched off from the common stock at an early date. Finally, the Brāhūī dialect of Baluchistan has been so much influenced by other languages that it is no longer a pure Dravidian form of speech.
The Dravidian languages have for ages been restricted to the territory they occupy at the present day. Moreover, they are gradually losing ground in the north, where they meet with Aryan forms of speech. If we compare the caste tables and the language tables in the Indian census of 1901 we find that only 1,125,479 out of the 2,286,913 Gōnds returned were stated to speak the Dravidian Gōndī. Similarly only 1505 out of 17,187 Kōlāms entered their language as Kōlāmī. Such tribes are gradually becoming Hinduized. Their language adopts an ever-increasing Aryan element till it is quite superseded by Aryan speech. In the north-eastern part of the Dravidian territory, to the east of Chanda and Bhandara, the usual state of affairs is that Dravidian dialects are spoken in the hills while Aryan forms of speech prevail in the plains. The Dravidian Kui thus stands out as an isolated island in the sea of Aryan speech.
This process has been going on from time immemorial. The Dravidians were already settled in India when the Aryans arrived from the north-west. The fair Aryans were at once struck by their dark hue, and named them accordingly kṛiṣṇa tvac, the black skin. In the course of time, however, the two races began to mix, and it is still possible to trace a Dravidian element in the Aryan languages of North India.
The teaching of anthropology is to the same effect. Most speakers of Dravidian languages belong to a distinct anthropological type which is known as the Dravidian. “The Dravidian race,” says Sir H. Risley, “the most primitive of the Indian types, occupies the oldest geological formation in India, the medley of forest-clad ranges, terraced plateaus, and undulating plains which stretches, roughly speaking, from the Vindhyas to Cape Comorin. On the east and west of the peninsular area the domain of the Dravidian is conterminous with the Ghats, while farther north it reaches on one side to the Aravallis and on the other to the Rajmahal hills.”
This territory is the proper home of the race. A strong Dravidian element can, however, also be traced in the population of northern India. In Kashmir and Punjab, where the Aryans had already settled in those prehistoric times when the Vedic hymns were composed, the prevailing type is the Aryan one. The same is the case in Rajputana. From the eastern frontier of the Punjab, on the other hand, and eastwards, a Dravidian element can be traced. This is the case in the valleys of the Ganges and the Jumna, where the Aryans only settled at a later period. Anthropologists also state that there is a Dravidian element in the population of western India, from Gujarat to Coorg.
It is thus probable that Dravidian languages have once been spoken in many tracts which are now occupied by Aryan forms of speech. The existence of a Dravidian dialect in Baluchistan seems to show that Dravidian settlers have once lived in those parts. The tribe in question, the Brāhūīs, are, however, now Eranians and not Dravidians by race, and it is not probable that there has ever been a numerous Dravidian population in Baluchistan. The Brāhūīs are most likely the descendants of settlers from the south.
There is no indication that the Dravidians have entered India from outside or superseded an older population. For all practical purposes they can accordingly be considered as the aborigines of the Deccan, whence they appear to have spread over part of northern India. Their languages form an isolated group, and it has not been possible to prove a connexion with any other family of languages. Such attempts have been made with reference to the Munda family, the Tibeto-Burman languages, and the dialects spoken by the aborigines of the Australian continent. The arguments adduced have not, however, proved to be sufficient, and only the Australian hypothesis can still lay claim to some probability. Till it has been more closely tested we must therefore consider the Dravidian family as an isolated group of languages, with several characteristic features of its own.
The pronunciation is described as soft and mellifluous. Abruptness and hard combinations of sounds are avoided. There is, for example, a distinct tendency to avoid pronouncing a short consonant at the end of a word, a very short vowel being often added after it. Thus the pronoun of the third person singular, which is avan, “he,” in Tamil, is pronounced avanu in Kanarese; the Sanskrit word vāk, “speech,” is borrowed in the form vāku in Tamil; the word gurram, “horse,” is commonly pronounced gurramu in Telugu, and so on. Combinations of consonants are further avoided in many cases where speakers of other languages do not experience any difficulty in pronouncing them. This tendency is well illustrated by the changes undergone by some borrowed words. Thus the Sanskrit word brāhmaṇa, “a Brahmin,” becomes barāmaṇa in Kanarese and pirāmaṇa in Tamil; the Sanskrit Dramiḍa, “Dravidian,” is borrowed by Tamil under the form Tirāmiḍa. Dramiḍa, which also occurs as Draviḍa, is in its turn developed from an older Damiḷa, which is identical with the word Tamiṛ, Tamil.
The forms pirāmaṇa and Tirāmiḍa in Tamil illustrate another feature of Dravidian enunciation. There is a tendency in all of them, and in Tamil and Malayālam it has become a law, against any word being permitted to begin with a stopped voiced consonant (g, j, ḍ, d, b), the corresponding voiceless sounds (k, c, ṭ, t, p, respectively) being substituted. In the middle of a word or compound, on the other hand, every consonant must be voiced. Thus the Sanskrit word danta, “tooth,” has been borrowed by Tamil in the form tandam, and the Telugu anna, “elder brother,” tammulu, “younger brother,” become when compounded annadammulu, “elder and younger brothers.”
There is no strongly marked accent on any one syllable, though there is a slight stress upon the first one. In some dialects this equilibrium between the different parts of a word is accompanied by a tendency to approach to each other the sound of vowels in consecutive syllables. This tendency, which has been called the “law of harmonic sequence,” is most apparent in Telugu, where the short u of certain suffixes is replaced by i when the preceding syllable contains one of the vowels i (short and long) and ei. Compare the dative suffix ku, ki, in gurramu-ku, “to a horse”; but tammuni-ki, “to a younger brother.” This tendency does not, however, play a prominent rôle in the Dravidian languages.