Get it on Google Play
Download on the App Store

5.4  Linguistic Arguments

5. Some new arguments


5.4. LINGUISTIC ARGUMENTS

5.4.1. East-Asian influences

Bernard Sergent traces practically all Indian language families to foreign origins.  He confirms the East-Asian origins of both the Tibeto-Burmese languages (Lepcha, Naga, Mizo etc.) and the Austro-Asiatic languages (Santal, Munda, Khasi etc.). Though many tribals in central and southern India are the biological progeny of India�s oldest human inhabitants, their adopted languages are all of foreign origin.  To Sergent, this is true of not only Austro-Asiatic and Indo-Aryan, but also of Dravidian.

The Himalayan branch of the Sino-Tibetan language family, distinct from Tibetan, already has a very long but inconspicuous presence in northern India.  Originating in China, this group of now very small languages once embraced parts of the northern plains.  Of greater historical importance is the Austro-Asiatic family, which Sergent describes as once the predominant one in a continuous area from Central India to Vietnam, but now reduced to a series of pockets in between the riverine population centres dominated by the immigrant Thai and Tibeto-Burmese languages (originating in western and ultimately in northern China) and in India by the Indo-Aryan languages.

He follows those scholars who consider the Central-Indian language isolate Nahali (assumed by its few students to be the original language of the western-Indian Bhils) as also belonging to the Austro-Asiatic family. This view is emphatically not shared by F.B.J. Kuiper, who lists 123 items of core vocabulary not reducible to Austro-Asiatic, Dravidian or IE roots, and calculates that �about 24 per cent of the Nahali vocabulary has no correspondence whatever in India�. If Kuiper is wrong, it would mean that as per the prevalent theories, not a single living language in the subcontinent (except for the peripheral languages Burushaski and Andamanese, at least for now) is indigenous.

Sergent is merely following in others� footsteps when he assumes that mayUra, �peacock�, gaja, �elephant�, karpAsa, �cotton�, and other Sanskrit fauna or flora terms are loans from Austro-Asiatic. In most such cases, the only ground for this assumption is that similar-sounding words exist in the Munda languages of Chotanagpur, languages which have not been committed to writing before the 19th century.  Chances are that in the intervening millennia, when these words were attested in Sanskrit but not necessarily in Munda, they were borrowed from Indo-Aryan ino Munda, or from an extinct language into both.  At any rate, the hypothesis of an Austro-Asiatic origin should only be accepted in case the term is also attested in non-Indian branches such as Khmer.

The alleged loans only start appearing n the 10th and youngest book of the Rg-Veda and really break through in the Brahmanas.  Sergent follows the classical interpretation, viz. that this shows how the Vedic Aryans gradually moved east, encountering the Austro-Asiatic speakers in the Ganga basin.  While I am not convinced of the existence of more than a few Munda terms in Sanskrit (more in the adjoining Indo-Aryan Prakrits: Hindi, Bengali, Oriya), I would agree that there are other Munda influences, notably in mythology, as we shall discuss separately.  Non-invasionists will have to account for this Munda contribution.

Here too, I suggest that chronology is all-important.  It is quite possible that Munda had not arrived in India at the time of the Rg-Veda.  When the Harappans migrated eastward (as demographically expansive populations do), or when the post-Harappans fled eastward from the disaster area which the Indus-Saraswati basin had become, the Munda-speaking people they encountered in eastern Uttar Pradesh and Bihar may have been re-cent immigrants.  All the same, it remains possible that for local flora and fauna, the indo-Aryans did adopt some Munda terminology.

Broadly, the Austro-Asiatic expansion from the agricultural civilization of Thailand can be compared with the gradual spread of the Old European Neolithic from Anatolia and the Balkans to the far corners of Europe, and with the spread of India�s Northwestern Neolithic to the rest of the subcontinent.  In that case, the Munda-speaking farmers in the eastern Ganga basin must have assimilated into the Indo-Aryan population, with only the peripheral populations in the hills retaining their imported languages.  This Munda contribution is by no means incompatible with a native status of IE, and even Hindu nationalists should welcome it as a factor of national integration across linguistic frontiers.

5.4.2. Is Dravidian native to India?

In one of his most innovative chapters, Sergent reviews all the evidence of Dravido-African and Dravido-Uralic kinship.  In African languages spoken in the entire Sahel belt, from Sudan to Senegal, numerous semantic and grammatical elements are found which also exist in Dravidian.  The similarity with the Uralic languages (Finnish, Hungarian, Samoyedic) is equally pronounced.  Sergent offers the hypothesis that at the dawn of the Neolithic Revolution (start of agriculture, some 10,000 years ago), the Dravidians left the Sudan, one band splitting off in Iran to head north to the Urals, the others entering India and moving south.

Within this scenario of a Dravidian immigration, it is tempting to speculate that upon entering India, the Dravidians first of all founded the Indus civilization.  Surprisingly, Sergent rejects this otherwise popular hypothesis, on the impeccably rational ground that there is no evidence for it.  Thus, except in coastal Sindh and Gujarat, geographical terms in the Indus-Saraswati area are never of Dravidian origin.  There is also no continuity in material culture between Harappan culture and the oldest known Dravidian settlements.

True to scholarly norms, Sergent pleads for a provisional acceptance of our ignorance about the identity of the Harappans.  However, as a concession to impatient readers who insist on having some theory at least, he gives one or two very slender indications that the Burushos (who preserve their Burushaski language till today in Hunza, Pak-Occupied Kashmir) may have played a role in it. However, he finds no Burushaski lexical influence on Indo-Aryan except possibly the word sinda, �river�, connected in one direction or the other with Sanskrit Sindhu, �river, Indus�, not otherwise attested in IE. He is also skeptical of David MacAlpin�s thesis of an �Elamo-Dravidian� language family: what isoglosses there are between Elamite and Dravidian can be explained sufficiently through contact rather than common origin.

Like many others, Sergent suggests that the early Dravidians can be equated with the �southern Neolithic� of 2500-1600 BC.  Their round huts with wooden framework are the direct precursors of contemporary rural Dravidian housing.  Two types of Hindu vessel have been discovered in southern Neolithic sites, including a beaked copper recipient still used in Vedic fire ceremonies.

Though the prehistory of the southern Neolithic is difficult to trace, it can be stated with confidence that the best candidate is the Northwestern Neolithic, which started in Mehrgarh in the 8th millennium BC.  It is, by contrast, very unlikely that it originated as an outpost of the Southeast-Asian Neolithic, which expanded into India at a rather late date, bringing the Austro-Asiatic languages.  According to Sergent, a link with the mature Harappan civilization is equally unlikely: neither in material culture nor in physical type is such a link indicated by the evidence.  The Dravidians were certainly already in the Deccan when the mature Harappan civilization started.  Sergent suggests that the Dravidians formed a pre-Harappan population in Sindh and Gujarat, and that they were overwhelmed and assimilated, not by the invading Aryans, but by the mature-Harappan population.

The picture which emerges is that of a multi-lingual Indus-Saraswati civilization with Dravidian as the minor partner (possibly preserved or at least leaving its mark in the southern metropolis of Mohenjo Daro) who ended up getting assimilated by the major partner, a non-Dravidian population whom we may venture to identify as Indo-Iranian and ultimately Indo-Aryan.

5.4.3. Afro-Dravidian kinship

One of the most remarkable findings related in some detail by Bernard Sergent, on the basis of three independent studies (by Lilias Homburger, by Tidiane Ndiaye, and by U.P. Upadhyaya and Mrs. S.P. Upadhyaya) reaching similar conclusions, is the multifarious kinship of the Dravidian language family with African languages of the Sahel belt, from Somalia to Senegal (Peul, Wolof, Mandè, Dyola).  As Sergent notes, all Melano-African languages have been credibly argued to be related, with the exception of the Khoi-San and Korama languages of southern Africa and the Afro-Asiatic family of northern Africa; so the kinship of Dravidian would be with that entire Melano-African superfamily, though it would be more conspicuous with some of its members.

Thus, between Dravidian and Bantu, we find the same verbal endings for the infinitive, the subjunctive, the perfect, the active participle or nomen agentis, related postpositions or nominal case endings, and many others.  In over-all structure, Dravidian and the Melano-African languages (as distinct from North-African and Khoi-San languages) form a pair when compared with other language families: �The tendency to agglutination, the absence of grammatical gender, the absence of internal vowel change, the use of pre-or postpositions instead of flection are some of the main traits which set the Negro-African and Dravidian languages jointly apart from the Indo-European and Hamito-Semitic groups.� Here I would say that this doesn�t prove much: the first trait is shared with some more, and the other ones are shared with most language families on earth; it is IE and Semito-Hamitic which stand out jointly by not having these traits.

But there are more specific similarities: �A simple system of five basic vowels with an opposition short/long, vocalic harmony, absence of consonant clusters in initial position, abundance of geminated consonants, distinction between inclusive and exclusive pronoun in the first person plural, absence of the comparative degree in adjectives, absence of adjectives and adverbs acting as distinct morphological categories, alternation of consonants or augmentation of nouns noted among the nouns of different classes, distinction between accomplished and unaccomplished action in the verbal paradigms as opposed to the distinction of time-specific tenses, separate sets of paradigms for the affirmative and negative forms of verbs, the use of reduplicated forms for the emphatic mode, etc.�

Sergent himself adds more isoglosses: �Preference for open syllables (i.e. those ending in vowels), the rejection of clusters of non-identical consonants, the generally initial position of the word accent in Dravidian and in the languages of Senegal�. The similarity in the demonstrative affixes is among the most striking: proximity is indicated by [i], initial in Dravidian but terminal in Wolof; distance by [a], intermediate distance by [u].

Knowing little of Dravidian and nothing at all of African languages, I don�t feel qualified to discuss this evidence.  However, I do note that we have several separate studies by unrelated researchers, using different samples of languages in their observations, and that each of them lists large numbers of similarities, not just in vocabulary, but also in linguistic structure, even in its most intimate features.  Thus, �the preposed demonstratives of Dravidian allow us to comprehend the genesis of the nominal classes, the fundamental trait of the Negro-African languages�.

To quite an extent, this evidence suggests that Dravidian and some of the African languages (the case has been made in most detail for the Senegalo-Guinean languages such as Wolof) have a common origin. At the distance involved, it is unlikely that the isoglosses noted are the effects of borrowing.  Either way, Proto-Dravidian must have been geographically close to the ancestor-language of the Negro-African languages.  Did it come from Africa, as Sergent concludes?  Should we think of a lost Saharan culture which disappeared before the conquests of the desert?  Note that earlier outspoken fans of Dravidian culture didn�t mind describing the Dravidians as immigrants: unlike the Aryans, they were bringers rather than destroyers of civilization, but they were immigrants nonetheless. Or should we follow Tamil chauvinists in assuming that the Dravidians came from Tamil Nadu and the now-submerged lands to its South, and took their language and civilization to Africa?

5.4.4. Additional indications for Afro-Dravidian

Bernard Sergent argues against the Indian origin of Dravidian.  One element to consider is that the members of the Dravidian family have not diverged very much from one another.  The relative closeness of its members suggests that they started growing apart only fairly recently: a thousand years for Tamil and Malayalam (well-attested), perhaps three thousand for the divergence of North- from South-Dravidian.  This would indicate that Dravidian was still a single language covering a small area in the early Harappan period, after having entered the country from the West.

That the �genealogical tree� of the Dravidian family seems to have its trunk in the coastal West of India, i.e. to the northwest of the main Dravidian area, has long been recognized by scholars of Dravidian. It also fits in with the old Brahminical nomenclature, which includes Gujarat and Maharashtra in the Pañcha-DraviDa, the �five Dravida areas of Brahminical settlement� (as contrasted with Pañcha-GauDa, the five North-Indian ones).  The northwestern coast was the first part of India to be dravidianized, the wellspring of Dravidian migration to the south, but also an area where Dravidian was gradually displaced by Indo-Aryan though not without influencing it.

Another indication for the Dravidian presence in Gujarat is the attestation in Gujarati Jain texts of inter-cousin marriage, typically South-Indian and quite non-Indo-European. The IE norm was very strict in prohibiting even distant forms of incest, a norm adopted by both Hinduism and Christianity. Linguists had already pointed out, and Sergent confirms, that Dravidian has left its mark on the Sindhi, Gujarati and Marathi languages (as with the distinction between inclusive and exclusive first person plural) and toponymy.  So, it is fairly well-established that Dravidian culture had a presence in Gujarat while spreading to South India.

It is possible that Gujarat was a waystation in a longer Dravidian migration from further west.  Whether the itinerary of Dravidian can ultimately be traced to Sudan or thereabouts, remains to be confirmed, but Sergent already has some interesting data to offer in support.  Africans and Dravidians had common types of round hut, common music instruments, common forms of snake worship and tree worship.  Thus, a South-Indian board game pallankuli closely resembles the African game mancalal; varieties of the game are attested in Pharaonic Egypt and in a pre-Christian monastery in Sri Lanka.

A point which I do not find entirely convincing is the distinction, based on Mircea Eliade�s research, between two types of Shamanism, one best known from Siberia and in evidence among all people originating in North and East Asia including the Native Americans and the Indian Munda-speaking tribes, another best known from Africa but also attested among some South-Indian tribes. This is a distinction between Shamanism properly speaking, in which the Shaman makes spirit journeys, despatches one of his multiple souls to the spirit world to help the soul of a sick person, etc.; and the religion of ghost-possession, in which the sorcerer allows the ghost to take him over but at the same time makes him obey.  The latter is perhaps best known to outsiders through the Afro-Caribbean Voodoo religion, but is also in evidence among South-Indian tribals such as the Saora and the Pramalai Kallar.

If anthropologists have observed these two distinct types, I will not disbelieve them. It does not follow that there must be a link between Africa and South India: Sergent himself notes that the same religion of ghost-possession is attested among the Australian aboriginals, who are related with the Veddoid substratum in India�s population. On the other hand, this theme of ghost-possession is but one of Sergent�s numerous linguistic and anthropological data which all point in the same direction of Afro-Dravidian kinship.

5.4.5. Uralic-Dravidian kinship

If Dravidian migrated from Africa to India through the Middle East, it could have left traces in Egypt and countries under Egyptian influence as well, explaining the data which led earlier researchers to the thesis of a Dravidian �Indo-Mediterranean� culture. Sergent links Indian forms of phallus worship with Sahel-African, Ethiopian, Egyptian and Mediterranean varieties of the same.  The Egyptian uraeus (�cobra�), the snake symbol on the pharaonic regalia, has been linked in detail with Dravidian forms of snake worship, including a priest�s possession by the snake�s spirit.  Dravidian cremation rituals for dead snakes recall the ceremonial burial of snakes in parts of Africa. Others have added the similarity between the Dravidian naga-kal (Tamil: �snake-stone�, a rectangular stone featuring two snakes facing one another, their bodies intertwined) and the intertwined snakes in the caduceus, the Greek symbol of science and medicine.

It has consequently been suggested that some Dravidian words may also have penetrated into the European languages.  Thus, Dravidian kal, �stone�, resembles Latin calculus, �pebble�, and Dravidian malai, �mountain�, resembles an Albanian and Rumanian word mal, �rock, rocky riverside�. But this hypothesis is a long shot and we need not pursue it here.

Far more substantial is the Dravidian impact on another language family far removed from the recent Dravidian speech area, viz. Uralic.  The influence pertains to a very sizable vocabulary, including core terms for hand, fire, house (Finnish kota, Tamil kudi), talk, cold, bathe, die, water, pure, see, knock, be mistaken, exit, fear, bright, behind, turn, sick, dirty, ant, strong, little, seed, cut, wait, tongue, laugh, moist, break, chest, tree; some pronouns, several numerals and dozens of terms for body parts. But it goes deeper than that.  Thus, both language families exclude voiced and aspirated consonants and all consonant clusters at the beginning of words.  They have in common several suffixes, expressions and the phonological principle of vocalic harmony.

As the Dravidian influence, like that of IE, is more pronounced in the Finno-Ugric than in the Samoyedic branch, we may surmise that the contact took place after the separation of the Samoyedic branch.  But the main question here is how Dravidian could have influenced Uralic given their actual distance.  Sergent suggests that a lost branch of Dravidians on the way from Africa strayed into Central Asia and got assimilated but not without influencing their new language.

He also rejects the theory that Dravidian forms one family along with Uralic, Turkic, Mongolian and Tunguz.  The latter three are often grouped as �Altaic�, a partly genetic and partly areal group which may also include Korean and Japanese, and all the said languages have at one time or another been claimed as relatives of Dravidian, with which they do present some isoglosses.  However, the isoglosses are fragmentary and mostly different ones for every language group concerned.  Moreover, some Dravidian influences are also discernible in Tokharic, or Arshi-Kuchi (Tokharic A c.q. Tokharic B) as Sergent appropriately calls it, which is obviously a matter of influence through contact.  So Sergent concludes that this is a matter or areal influence rather than genetic kinship: Dravidian was a foreign language entering Central Asia at some point in time to briefly exert an influence on the local languages before disappearing.

I am not sure this will convince everyone: if Dravidian is not genetically linked with all the said language groups, it might still be so with one of them, viz.  Uralic, at least on the strength of the data Sergent offers.  Tamil chauvinists may well be tempted to complete the picture by claiming that before the Indo-Europeans from India colonized Central Asia and Europe, it was the turn of the Dravidians to colonize Central Asia and, after mixing genetically and linguistically with the natives, to develop the Uralic languages.  At a time when subtropical Neolithic cultures had a tremendous technological and demographical edge over the hunter-gatherers in the inhospitable northern countries, it would not even be so far-fetched to imagine that a small wayward group of Dravidians could enter the vast expanse of Central Asia and completely change the linguistic landscape there.

At any rate, Sergent�s observations represent a clean break with earlier theories which had the Dravidians originate in the Uralic speech area and preceding the Indo-Aryans in an invasion of India from Central Asia.

5.4.5. Geographical distribution of IE languages

Since Bernard Sergent doesn�t take the Indocentric case for IE seriously, he doesn�t bring out all the linguistic data which to him support the Kurgan scenario.  One classical argument from linguistics is nonetheless developed at some length: �In Europe one finds the most numerous and geographically most concentrated IE language groups.  Such a situation is not unique, and invariably denotes the direction of history: the Indo-Iranian languages represent a branch extended to the east and south, starting from Europe and not the other way around.  It is obviously not the IE languages of Europe which have come from India�.

Thus early in his book (p.30 of 584 pp.), he is already so sure that �obviously� the central question of the Urheimat has been decided to the disadvantage of India.  That is a great pity, for it is the reason why he has not applied himself to really developing the argument against the Indian Urheimat.  If anyone is capable of proving the AIT, it must be Sergent.  Yet, because he assumes no proof is necessary, he gives the question much less attention than e.g. the much less contentious (though more original) question �of the geographical origins of Dravidian.

To be sure, the pattern of language distribution invoked by Sergent as �not unique�, is indeed well-attested, e.g. in sub-Saharan West Africa, there are about 15 language families, while in the much larger region of sub-equatorial Africa, a very large majority of the people speaks languages belonging to only one family, Bantu.  Though it is only a branch of a subfamily of the Niger-Kordofanian language family, Bantu easily outnumbers all the other branches of this family combined: �Africanists conclude that Bantu originated in a small area, on the border between Nigeria and Cameroon.�

But in fact, India is in this respect more akin to West Africa, and Europe more to sub-equatorial Africa.  India has more language families: Nahali, Andamanese, Burushaski, Dravidian, Austro-Asiatic (Munda and Mon-Khmer), Sino-Tibetan (Himalayan, Tibetic and Burmese) and IE (Iranian, Kafir, Dardic, Indo-Aryan, and possibly proto-Bangani).  Europe is almost entirely IE-speaking, with Basque serving as the European counterpart to the Khoi-San languages in subequatorial Africa, a left-over of the original linguistic landscape largely replaced with the conquering newcomer, IE c.q. Bantu; and Uralic (Finnish, Estonian, Hungarian) a fellow if perhaps slightly later intruder in the European landscape, vaguely comparable to the intrusion of an Austronesian language in a: part (viz.  Madagascar) of southern Africa.

Therefore, I reject the argument from the geographical distribution.  I have already pointed out another objection against it: if the spread of the IE languages to Europe was often a matter of assimilating divergent native populations, this process promoted the speedy diverging of the IE dialects into distinct language groups.  Though this is not a conclusive argument against the possibility of IE settlement in India being younger than in Europe, it at least terminates the impression that there was a compelling case in favour of that possibility.  So, even under Bernard Sergent�s hands, the fabled �linguistic evidence� has failed to decide the IE Urheimat question once and for all.
 

Footnotes: 
 

    Bernard Sergent: Genèse de l�Inde, p.31. The precarious situation of Nahali is described as follows by K.S. Nagaraja, reviewing Robert Parkin.  A Guide to Austro-Asiatic Speakers and Their Languages, University of Hawaii Press, Honolulu 1991, in Bulletin of the Deccan College 1996-97, p.342: �On the basis of my observation after visiting Tembi (Teli) village in November 1996, I can say that the Nahals there no longer speak Nahali language at all. (�) in the districts of Buldana in Maharashtra, in the village called Jamud, there is a big concentration of Nahals who actually speak this language (�) there are many settlements in the nearby villages where the language is still spoken.  The total number of speakers seems to be over three to four thousand.�

    F.B.J. Kuiper: Nahali, a Comparative Study, Amsterdam 1962, p.49.

    Bernard Sergent: Genèse de l�Inde, p.370.

    Bernard Sergent: Genèse de l�Inde, p. 138.

    Remark that the Iranian name Hindu for �Indus�, hence also for �India�, indicates that the Iranians have lived near the Indus.  If they had not, then Sindhu would have been a foreign term which they would have left intact, just as they kept the Elamite city name Susa intact (rather than evolving it to Huha or something like that).  But because Sindhu was part of their own vocabulary, it followed the evolution of Iranian phonetics to become Hindu.

    Bernard Sergent: Genèse de l�Inde, p.48, with reference to Bridget and Raymond Allchin and to Dharma Pal Agrawal.

    Bernard Sergent: Genèse de l�Inde, p.52.

    Bernard Sergent: Genèse de l�Inde, p.55, quoting from U.P. and S.P. Upadhyaya: �Dravidian and Negro-African�, International Journal of Dravidian Linguistics, 1976/5.1, p.32-64; my quotation is retranslated from the French translation (as quoted by Sergent), �Affinités ethno-linguistiques entre les Dravidiens et les Nègro-Africains�, Bulletin de l�Institut FranAais d�Afrique Noire 38.1, p. 127-157.

    That Hamito-Semitic (Afro-Asiatic) and IE stand jointly apart and may have a common origin in Mesopotamia, has been argued by B. Sergent: Les Indo-Européens, p.431-434. Critics (such as the reviewer in Antaios 10, Brussels 1996) have suggested that with this position, he is playing a political game.  This much is true, that by design or by accident, he is pulling the leg of far-rightist adepts of IE studies who consider the reduction of IE to sisterhood with Semitic as sacrilege.  All the same, Sergent�s position is quite sound linguistically.

    Bernard Sergent: Genèse de l�Inde, p.55, quoting from U.P. and S.P. Upadhyaya: �Dravidian and Negro-African�, International Journal of Dravidian Linguistics, 1976/5.1, p.32-64, retranslated from the French translation, �Affinités ethno-linguistiques entre les Dravidiens et les Négro-Africains�, Bulletin de l�Institut FranAais d�Afrique Noire 38.1, p.127-157.

    Bernard Sergent: Genèse de l�Inde, p.56.

    Bernard Sergent: Genèse de l�Inde, p-53.

    E.g. Father H. Heras: Studies in Proto-Indo-Meditarranean Culture (1953), and Alain Daniélou: Histoire de l�Inde (1983).

    A map showing this �tree� is given in G. John Samuel, ed.: Encyclopedia of Tamil Literature, Institute of Asian Studies, Madras 1990, p-45, with reference to Kamil Zvelebil, who locates the Proto-Dravidians in Iran as late as 3500 BC.

    Bernard Sergent: Genèse de l�Inde, p.51.

    This in contrast with Biblical Judaism and especially with Islam: Hindu converts to Islam were often required to prove their conversion by eating beef and, if possible, marrying a cousin or niece; half of the marriages in rural Pakistan are between cousins.  Note, however, that the Zoroastrians deviated from the IE standard by also practising marriage within the family.

    Bernard Sergent: Genèse de l�Inde, p.59.

    Bernard Sergent: Genèse de l�Inde, p.62.

    Bernard Sergent: Genèse de l�Inde, p.62.

    E.g. Father H. Heras: Studies in Proto-Indo-Mediterranean Culture, Indian Historical Research Institute, Bombay 1953.

    For all Sergent�s details about Dravidian snake-worship, which fits in well with the classical picture of snake-worship as an �aboriginal� or at least non-Aryan element in Hinduism, it is interesting to note that he (Genèse de l�Inde, p.482, n.607) deviates from the mainstream in his etymology of nAga, �snake�.  With reference to Manfred Mayrhofer, he links it quite regularly to Germanic s-nake; the prosthetic s- is quasi-onomatopoeic.  Personally, I suggest an even more regular link with Germanic naked (from PIE *nogwos/nogwodhos), which reveals the basic meaning: the snake is unhairy, sheds its skin, and exposes itself more deeply to its environment by not having limbs with which to keep objects or the ground at a distance, all forms of exposure or nakedness. NAgA SAdhUs are those Hindu godmen who walk naked.

    Mentioned in a long enumeration of pre-IE loans, but without reference to the Dravidian counterparts, in Sorin Paliga: �Proto-Indo-European, Pre-Indo-European, Old European Archaeological Evidence and Linguistic Investigation�, Journal of Indo-European Studies, Fall 1989, p.309-334.

    Bernard Sergent: Genèse de l�Inde, p.66-67.

    Bernard Sergent: Genèse de l�Inde, p.71-76.

    Bernard Sergent: Genèse de l�Inde, p.29-30.

    Bernard Sergent: Genèse de l�Inde, p.30.