$ 1.7. Greater Sindh In contrast to the clear picture of the Panjab in Rgvedic times, the situation in Greater Sindh is much more vague and the following results must remain tentative. The RV does not mention this area as such, yet there are some indications that Sindh and neighboring Baluchistan were known. First of all, the bhalAnas tribe took part in the Ten Kings' Battle (RV 7.18) that settled the suzerainty of the Bharata chieftain over the Panjab tribes. The bhalAnas are identified with the bolAn pass and river near Quetta in Baluchistan. Unfortunately, southern local rivers are not mentioned anywhere in the RV south of the gomatI (Gomal River). However, data from RV book 8 may supplement our scanty information. Book 8 has long been connected with Eastern Iran: K. Hoffmann (1940 = 1975: 1 sqq.) has pointed to Iranian looking names such as kazu ~ Avest. kasu- (EWA I 330), kazu caidya 8.5.37, kanIta ~ Scythian kanitEs, cf. further tirindira 8.6.46 ~ tiridatEs ~ Avest. tIrO.nakatha, kRza 8.59.3 ~ k@r@sAspa, parzu 8.6.46 ~ O.Pers. pArsa 'Persian', paktha RV 8.22.10 (mod. Pashto, Paktho), varo suSAman 8.60.18 (with unusual Sandhi), arzasAna 8.12.9, 2.20.6, etc., anarzani 8.32.2 ~ Iran. @rSan-? All such names, if Iranian, belong to pre-Iranian tribes that spoke a dialect close to the one that later developed to E. Iranian (cf. the similar case of the Mitanni-Aryans, below). Book 8 also knows of camels (uSTra 8.4.21-24, 31, 46-48, O. Iran. uStra, as in zarath-uStra), that are first attested archaeologically in S. Asia in the bolAn area, at Pirak, c.1700 BCE. Now, apart from RV 3 and 7, Drav. words occur first in the Middle RV book 8, more specifically in its kANva section (RV 8.1-48, and 8.49-59, 60-66); they include kuNDa- 8.17.13, mayUra 8.1.25, naDa/naLa 8.1.33 (see above); note also the many words in RV 8 with retroflexes (Kuiper 1991: 17, Hoffmann 1941, 1975:16, Kuiper 1967: 84 n. 18, 86 n. 26). If one locates at least the kANva sections of book 8 in East Iranian lands, that is in (S.W.) Afghanistan and Baluchistan, one can also adduce the very name of this clan of poets. K. Hoffmann (and I) have connected the name with kR 'to act magically, to do sorcery' (Hoffmann 1975: 1 sqq., Witzel 1983-5). Kuiper (1991: 80) has correctly objected there also is pra-skaNva, with the common Indus prefix pra- *[p@r-] (contra, with insufficient reasons, Oberlies 1994: 341). This may mean that the Indus language extended to Eastern Iran, especially to the area west of Sindh, to Baluchistan, and to Makran with its many Indus settlements. Book 8 would then represent an amalgam of Dravidian and Para-Munda influences (including some pre-Iranian?) Dravidian influence in Middle Rgvedic (the time of king sudAs) can be traced back, with some probability, to the areas from Arachosia to Sindh as well. It is here that Drav. place names are assumed to appear first (cf. L.V. Ramaswamy Iyer 1929-30). These names (showing MIA development p > v) extend from Sindh via Gujarat and Maharastra to the South: Sindhi -vali, Gujarati -wArI/warI (Sankalia 1949), Mar. -oli, all from a Drav. word for 'village' (Tam. paLLi 'hamlet', Kan. paLLi, haLLi, Tel. palli 'village', Kur. pallI DEDR 4018, CDIAL 7972, see Parpola 1984, 1994: 170 sqq., 1997; Southworth 1995: 271, see further, below; -- Panjabi -wAlA, wAlI rather looks like the common Hindi etc. suffix, as in jAne-wAlA, petrol pump-walla, etc.). A similar view has been proposed, on the basis of linguistic and archaeological observations, by Zvelebil (1972, 1990: 48, 123), Southworth and McAlpin,(fn. 3) and Fairservis (1992: 17, 21). It has to be underlined, however, that McAlpin's reconstruction of an Elamo-Dravidian language family has not been accepted by Dravidologists. Fairservis and Zvelebil think of an immigration by Drav. speaking tribes at c. 4000/3500 BCE, from the mountainous lands of East Iran into the Indus valley. Both underline data that characterize the Dravida as originally pastoral hill tribes. In sum, we may reckon with early Drav. pastoralists (Fairservis 1992, 1997) in Baluchistan and later on, after a period of acculturation with the Indus people, we may encounter Drav. farmers (Southworth 1979, 1990, 1995) who practiced intensive rice (Kenoyer 1998: 178, Jarrige 1985) and millet cultivation in Sindh. $ 1.8. The languages of Sindh In addition to these western (Dravidian, pre-Iranian) elements there also are local 'Sindh' ones. First of all, it is precisely in this area that rice was first introduced into the Indus civilization. It occurs first as odana 'rice gruel' in the (partly E. Iranian) kaNva book (RV 8) in the emuSa myth, which clearly smacks of 'foreign' origin: RV 8.69.14, 8.77.6-11, 8.77.10, (cf. also 8.96.2, 1.61.7; summary and discussion by Kuiper 1991: 16 sqq.) He had explained it earlier on (1950) as Austro-Asiatic, but is more cautious now (Kuiper 1991: 18f., cf. below). On closer observation, we can notice a mixture of an IA, Austro-Asiatic and possibly Drav. myth. Kuiper (1991) now shows that the kaNvas, non-IA local sorcerers, introduced this myth into the RV. At any rate, the motif is unusual for the RV. Its hero is a divine bow shooter (probably seen on an Indus copper plate, only at Mohenjo Daro, in Sindh, Parpola 1997: 39; cf. also Avesta, Yt. 8.6,37 @r@xSa, kRzAnu RV 4.27.3, Rudra, and murukan in S. India; for 'bow' see KS dAlbhUSI, MS drumbhUlI; with PDrav -R- > [l~] / [.Z.], Kuiper 1991: 26). This bow shooter splits a mountain, finds the odana rice gruel and kills the boar emuSa. The myth is an imitation of the well known Rgvedic Vala myth (splitting the mountain cave containing the cows/dawns), but is otherwise completely alien to the RV. Now, the suffix -uSa (Kuiper 1991) of emuSa clearly indicates a name taken from the (Para-Munda) Indus language. This points to a late myth (because a latecomer, rice, is important), adopted from the local southern or southwestern Indus region and from beyond. (fn.4) Second, the word for 'rice' occurs in a Sindh and a Panjab variety (see below). The Sindh version, closer to Dravidian, has been transmitted further west, along the southern trading route to Fars and has entered western languages from there (Greek oryza). Whether rice was otherwise known to the Rgveda is doubtful. Rice was introduced towards the end of the Indus civilization in its southern areas, in Sindh (Kenoyer 1998: 178, in Pirak, along with newly introduced sorghum and millet, and also horse, donkey, camel). In this case, we have again to reckon with a (West-)Munda word: odana is connected with oDi(kA) 'wild rice' (lex., CDIAL 2546) and Santali hoRo, huRu 'rice plant' (EWA I 280) and explained as Munda loan (Berger 1963: 420, Kuiper 1950: 179; but cf. Zide and Zide 1973: 8-9 on Mundari kode, Kharia kuDa 'millet, ragi'). Together with the introduction of rice its charter myth (Malinowski) may have been taken over as well. As has been mentioned, the Dravidians originally had neither a word for 'rice' nor for the staple food of the Indus civilization, wheat. In sum, it can be said that we may have to reckon with a combination of several factors in the southern Indus area: with the (Para-Munda) Indus language, with some more eastern Munda influences, with immigration from E. Iran in the person of VasiSTha (RV 7) and of (pre-)Old Iranian tribes into Baluchistan and the neighboring Kachi plain of the Indus valley (e.g. at Pirak, 1700 BCE), and with Dravidian immigration. As mentioned above, Zvelebil (1970, 1990) is of the opinion that the Dravida entered South Asia from the Iranian highlands. Their oldest vocabulary (Southworth & McAlpin) is that of a semi-nomadic, pastoral group, not of an agricultural community. They are thus not expected to have their own word for 'wheat'. Wheat, however, was the staple of the Indus civilization, and was called in Dravidian by an adaptation of a local word: *gO-di 'low red plant' (Southworth 1988, 1979, 1990) which is quite different from the Panjab word *go-dum > Vedic godhUma 'cow smoke' (details below). If the Dravidians acquired agriculture only in the hills bordering S. Asia, they may very well have been inhabitants of Baluchistan at the time. At any rate, neighboring Sindh, just as Gujarat and Maharastra, show place names that are explainable from Dravidian *paLLi (see above). Then, according to archaeology, a large section of the population of Sindh left this area towards the end of the Indus period. They moved further east, to Gujarat, where we find a late, local phase of the Indus civilization (Rangpur phase IIb, IIc, see Allchin 1995: 32 sqq., Kenoyer 1998: 173 sqq.), and, again, Drav. place names. It is indeed possible that the Dravida constituted a first wave of central Asian tribes that came to Iran before the IA, just as the Kassites came to Mesopotamia before the Mitanni-IA. In that case they knew the horse already in Central Asia, but would not have taken it over directly from the Indo-Iranians (as may be indicated by Brahui (h)ullI, O.Tam. ivuLi 'horse', etc., different from IIr. ac'va). In other respects as well, they have not been influenced by the Indo-Iranians. One can even assume that the early testimony of the introduction of horse and camel from the Iranian plateau into Sindh (Pirak and Kachi plain in western Sindh) is due to the Dravida (c. 1700 BCE, Kenoyer 1998: 178; Allchin 1995: 31). In that case, it must be investigated why they apparently did not preserve a word for 'camel'. In this fashion, that is through the mediation of the Dravida in Sindh, Drav. *variJci 'rice' must have reached Iran (> M.Pers. brinj), that is not, as otherwise common, via the northwestern Khaiber Pass, as in this region another form of the word is found, with *vrijhi > Pashto wrizE, etc. (see below). This may mean, on the one hand, that the Dravida themselves were immigrating at the time of the older RV, or that they only influenced the Panjab in the later, Middle Rgvedic period, coming from Sindh. This is perhaps supported by archaeological facts, for Sindh was practically deserted by its population in the post-Indus phase (Allchin 1995: 31 sqq.) It is from this Southern basis that they suddenly appear in mid-level RV, with names such as kavaSa 'straddle legged' (k. ailUSa RV), cf. zailUSa ''dancer, singer" VS (EWA II 655, Kuiper 1991:20, 25, 42) which Kuiper 1991: 24 explains with reference to Dravidian: initial c- is often dropped in South(!) Dravidian; further examples in RV are : zirimbiTha : irimbiThi EWA II 639, cf. also ziriNA 'hiding place, night?' : irINa 'salt pan, hiding place (for gambling)' (Witzel 1999). ailUSa is important, as it was this poet who was an important priest on the side of the opponents of the Bharata. (These opponents included the bhalAnas). His great-grandson tura kAvaSeya, however, is an important priest of the Kuru realm that succeeded the Bharata 'kingdom'; he developed the agnicayana ritual (Th. Proferes, Harvard Ph.D. thesis 1999). This case shows the inclusion of a Dravidian into the fold, and underlines the important role a new 'convert' to Arya religion could play in its very development (that of the post-RV, classical zrauta ritual, see Proferes). Further, he was not classified as zUdra but obviously as a Brahmin who had learned to compose RV hymns in the traditional poetic IA language! All of this is indicative of a high degree of amalgamation and language acquisition at this time, during the middle and late Rgveda period (see below). $ 1.9. The Southern Indus language: Meluhhan However, there are indications that another language was prevalent in Sindh before the immigration of the Dravida. The trade of the Indus civilization with Sumeria and later Mesopotamia has left us a number of words that are not Dravidian. It is perhaps best to call this language "Meluhhan" after the name the Sumerians gave to the country, meluHHa. Its language was also sufficiently different from Elamite or Sumerian to require a 'translator from meluHHa' (Possehl 1996a: no. 2), whose name is u-iliSu (Parpola 1994: 132). In fact, "the language of MarhaSi [Bampur area, just west of Iranian Baluchistan] is different from that of the simaSkian [Tepe Yahya in southern Central Iran], and only very partially Elamite-related." (Vallat 1985: 52). This indicates that there was a language boundary, somewhere to the west of the present Iran-Pakistan border, probably in a southwards prolongation of the Iran-Afghanistan border. Possehl identifies the area of meluHHa (1996, 1997) as having a center in the hills and mountains of Baluchistan, closer to the population center of the early Indus civilization, which allows for a hypothetical identification of the marhaSi language with that of meluHHa and makes a thorough investigation of the data of RV 8 (see $ 1.7.) even more important. There are men with meluHHa as a personal name, thus apparently 'the meluHHan'; several persons, among them urkal and ur-dlama, are called 'the son of meluHHa'. There also is a 'village of meluHHa', from where a person called nin-ana comes. The products of meluHHa include giS-ab-ba-me-lu-HHa (abba wood, a thorn tree), mesu wood ('of the plains'), ships of meluHHan style (magilum boat), (Possehl 1996a). In total, there are some 40 "Indian" words transmitted to ancient Mesopotamia, some of which may have been coined by Dilmun (Bahrain) traders. They include: Sindh wood sinda (si-in-da-a, si-in-du), date palm, the 'red dog of MeluHHa', zaza cattle (zebu?), elephants, etc. (cf. Landsberger, Die Welt des Orients 3. 261, Possehl 1996a). As coming from Dilmun (Bahrain) we may add the Meluhhan(?) trees giS-Ha-lu-ub or Haluppu wood, giS-mes-makan or msu wood of Magan, and the giSgiSimmar wood (cf. above *zimmal in zimbala, zalmali 'Salmalia malabarica'!). A slightly later(?) loan-word relationship is seen in Sumer. ili 'sesame', Akkad. ellu/Ulu 'sesame oil', which is only found in South Drav. with eL, eLLu 'Sesamum indicum' (D. Bedigian 1985); the word can be compared, however, with Ved. tila and jar-tila 'sesame' which shows the typical Para-Munda prefix C@r- (cf. Kuiper 1955: 157 for a Munda origin). The ultimate source, **(t)il, however is unclear, cf. further, on Sumer. loan words, Blazek and Boisson 1992. The word meluHHa is of special interest. It occurs as a verb in a different form (mlecha-) in Vedic only in ZB 3.2.1, an eastern text of N. Bihar where it indicates 'to speak in barbarian fashion'. But it has a form closer to meluHHa in Middle Indian (MIA): Pali, the church language of S. Buddhism which originated as a western N. Indian dialect (roughly, between Mathura, Gujarat and the Vindhya) has milakkha, milakkhu. Other forms, closer to ZB mleccha are found in MIA *mliccha > Sindhi milis, Panjabi milech, malech, Kashmiri bri.c.hun 'weep, lament' (< *mrech-, with the common r/l interchange of IA), W. Pahari mel.c.h 'dirty'. It seems that, just as in other cases mentioned above, the original local form *m(e)luH (i.e. m(e)lukh in IA pronunciation, cf. E. Iranian bAxdhI 'Bactria' > AV *bahli-ka, balhi-ka) was preserved only in the South (Gujarat? > Pali), while the North (Panjab, Kashmir, even ZB and Bengal) has *mlecch. The sound shift from -HH-/-kh- > -cch- is unexplained; it may have been modeled on similar correspondences in MIA (Skt. akSi 'eye' ~ MIA akkhi, acchi; kSetra 'field' ~ MIA khetta, chetta, etc.) The meaning of Mleccha must have evolved from 'self-designation' > 'name of foreigners', cf. those of the Franks > Arab farinjI 'foreigner.' Its introduction into Vedic must have begun in meluHHa, in Baluchistan-Sindh, and have been transmitted for a long time in a non-literary level of IA as a nickname, before surfacing in E. North India in Middle/Late Vedic as mleccha. (fn. 5) Further examples of the Southern Indus (Sindh) language include the designations of plough, rice, wheat, and millet. Plough The old agricultural word lAGgala 'plow' (RV, 4.57.4, a late hymn) is found, in a divergent form, in Tam. JAJcil, nAJcil, Kan. nEgal, Gadba nAngal (DEDR 2907). Southworth (1988; 1979: 200, 205; 1995: 268, cf. Kuiper 1948: 127, 1955: 156, Przyludski BSL 24, 118 sqq., cf. Parpola 1994: 168) assumes a popular etymology PDrav. *JAn-kal, *JAn-kel 'earth stone' and traces the term back an Austro-Asiatic source, Munda *Ja-kel, Jan-kel (Zide & Zide 1973: 5), Santali nahel, Khasi lynkor [l@nkor] < *le~nkol, Khmer aGkal; cf. also the Austronesian forms, Malay tengala, Makassar naGkala (Bagchi 1929, 9). V. BlaZek and C. Boisson, (1992: 17-19) think of a Sumerian, and ultimately perhaps even Afro-Asiatic origin of this widespread word of culture: Sumer. ni'G-Gala or ni'G-Gl 'sickle' (!) and Afro-As. *nigal 'to reap; reaping sickle.' However, the Munda words do not agree with Ved. lAGgala, though one can easily assume dissimilation of n-l. The word underlying RV lAGgala must have come from an intermediate language, in short, the Panjabi form of the Indus language (Para-Munda), with *laGgal. This form cannot have been that of the Southern Indus language (Meluhhan) as this has resulted in Drav. *JAnkal, JAnkel. While the difference is small here (g/k, n/l), it is more substantial in other agricultural words. Rice The word for 'rice' shows a difference between a Northern form, approximately **(@)@rij, versus a southern one, *vari, (v)ariki, variJci. Note that this indicates the same difference in tenuis/media as met with in the word for 'plough': N. *laGgal, *v@riji :: S. *naGkal, *variJci/variki. Still another form exists in Proto-Munda *@-rig; it has provided Dravidian *(v)ari, variki > Tam. arici, ari, Kan. akki (DEDR 215), and also Tam., Tel. vari (DEDR 6565). Though rice is indigenous to S. Asia, the domesticated version can be traced back to S.E. Asia and S. China. (fn. 6) It has been found in India since the 3rd millennium BCE (Glover & Higham 1996, Kajale 1991), and appeared late in the southern Indus civilization, at Pirak c. 1700 BCE. However, it appears first (as vrIhi) only in post-RV texts (AV, c. 1200 BCE), though it probably was an ingredient in the RV offerings puroDAza 'rice cake' and odana 'rice gruel'. The older IA grain is only yava 'barley', but later on we have 7 or 10 agricultural products: in the yajurveda saMhitAs, the 'seven agricultural plants' (sapta' grAmyA' o'SadhayaH); ZB 14,9,3,22 has even ten: vrIhi' Oryza sativa L.; ya'va Hordeum vulgare L. subsp. hexastichum (L.) Schinz et Kell.; ti'la Sesamum indicum L.; m'Sa Phaseolus mungo L. var. radiatus = Phaseolus Roxburghii; a'Nu Panicum miliaceum L.; priya'Ggu Setaria italica (L.) Pal. Beauv. = Panicum italicum L.; godhU'ma Triticum aestivum = Triticum sativum Lam.; masU'ra Lens culinaris Medic. = Ervum lens L.; kha'lva Phaseolus radiatus L. a variety of Phaseolus mungo L. = mASa(?); khala'-kula Dolichos biflorus L. (W. Rau, in: Witzel 1997: 203-206). Southworth (1979, 1988: 659-660) supposes an Elamo-Dravid. origin: *var 'seed, grain', Elam. bar 'seed', PDrav (stage 1, c. 2000 BCE) *vari 'rice grain'. (McAlpin 1981, Tyler 1968, Southworth 1988). Achaemenid Elam. umi 'grind (grain)', *um 'to process grain', PDrav1 *um 'husk, chaff' DEDR 637; (this should be compared with *gant-um-a, gandh-um-a!). However, the Elamo-Drav. family has not been proven to the satisfaction of Dravidianists (McAlpin (et al.) 1975, Krishnamurti 1985, Zvelebil 1985), and the N. Drav. language Brahui, seen as a link by McAlpin, is a late-comer to Baluchistan (Elfenbein 1987). Southworth (1988: 664) stresses the difference between northern (Gangetic) and southern rice, which might have been dry land rice. On the other hand, Southworth later on mentions that PDrav *(v)ariki DEDR 215, has been taken over from PMunda at c. 1500 BCE: *@rig 'millet, Panicum militare' (Zide & Zide 1973: 8) --> *arik(i) 'staple grain' (Southworth 1988: 660), because the South Drav. sound change k > c took place only between the second and third stage of Drav. (Krishnamurti 1969); thus: Munda *@rig --> Drav. *(v)ariki > Tamil ari, arici. This South Dravidian form arici has been transmitted westwards, probably by sea trade, Greek o'ryza, o'ryzon and Arab. ruz, Engl. rice etc. (Southworth 1979: 202, cf. EWA II 598). Southworth also reconstructs PDrav. *vari, *variJci DEDR 5265. This, too, was transmitted westwards, but via the Baluchistan-BampUr trail, to Old Iranian as *brinj, M.Iran. brinj, N.Pers. birinj). It must have been this form that was the basis of the word in the late Southern Indus civilization. The northern track westwards is attested by Ved. vrIhi < pre-IA *vrijhi- and reflected in the E. Iran. (and N. Iran.?) languages: Pastho wriZE, (but Khotan. rrIysua [rIzua]!), Nuristani wrI.c, rI.c (cf. Fussman 1972). The Northern Indus dialect had *vrij > Ved. *vrijhi > vrIhi, Nuristani wrI.c., Pashto wriZE. The Southern dialect is indicated by M.Pers. brinj, N.Pers. birinj, going back to *v@riJji, Dravidian *variJci, a form with "infixed" -n-, found in central Dravidian: Gondi wanjI (Pengo verci(l), Gadba vasil, DEDR 5265). The form with -n- points to Munda origin and to a relatively far-reaching influence or expansion of the Munda in this early period (cf. Kuiper 1955: 140, 1962: 14, 51, 1991: 39f.) Again, this distribution also suggests a difference between, on the one hand, northern or north-western form, including the northern Indus language, and on the other, the southern Indus language and the rest of the subcontinent. However, these forms have to be reconciled with Tibetan 'bras [@bras] > mod. Tib. [je], Purik bras, with the neighboring, linguistically isolated Burushaski bras (Kuiper 1962: 40, 1955: 143 n. 17, Tikkanen, 1988: 303-325), Dumaki bras, and even with some Austronesian forms such as Malay b@ras--> Somali bari`s?; cf., however, Dayak bari, Malegasy vare, vari --> Bantu wari, wali (Nurse 1983, Southworth 1988: 664, Witzel 1995) and O.Jpn. uru-shine, (cf. mod. Jpn. uru-chi < *uru-ti). Both bras and pre-Vedic *vrijhi must go back to a source such as **@w@rij (Witzel 1997). In the study of the Asian words for 'rice' we have to take into account words from S., S.E. and E.Asia: - S. Asia: Ved. vrIhi < *vrijhi, Burushaski bras, (fn. 7) Tib. 'bras, (fn. 8) Drav. *arici, *variJci; (fn. 9) Munda *@-rig, Tib.-Burm. *dza- (fn. 10) < Austr. *Csamaq Kusunda cusum 'rice in husks', kAdiyun 'cleaned rice' - S.E. Asia: Munda *rung-kug (Zide & Zide 1973: 17) Austr. *Csamaq Austrones. *pajay; Austrones. *i-may Thai *xau > khaw (Haudricourt, in Shafer 1966-7: 522) Austro-Thai *kru-may (> Jpn. kome) - E. Asia: Chin. *mi@r, Tib.-Burm. *may (fn. 11) The distribution of the various words for 'rice' points to an old (South-)East Asian word of culture. Just as in the modern spread of the E. Asian word 'tea', several routes of distribution have to be distinguished: 1. an approximate reconstruction of the S.(E.) Asian word *@vrij(h)i/*@bras, probably < **@w@rij, (fn. 12) which is spread out in a wide arch between 2. E. Asian *may, *xau, *krumay (< *kru-*may?) ((fn. 13) and 3. S. Asian *@-rig, (fn. 14) *rung-ku(g). PMunda *rung-ku(g) (Zide & Zide 1973: 17, *(r)-(n)-ku, Kuiper 1962) may be an Austro-Asiatic form with prefix r-. This might be connected, via metathesis, with Benedict's Austro-Thai-Japanese *krumay (> Jpn. kome, kuma-shine), a word that may be composed, if Sino-Tib. (Benedict 1972: no. 65, 128, 149, 192, 193) *may, Austrones. i-may and Thai *xau are compared, of *kru-*may. In the end, one may think of a Proto-form **kru as the ultimate source for 'rice' in S.E. and E. Asia (Sino-Tib., Austro-As., Austro-Thai). Wheat Further dialect differences between the northern (Panjab) and the southern (Sindh) forms of the Indus language can be observed in the designation of 'wheat'. Though some claim that wheat, the staple of the Indus civilization, is a local domesticate (cf. Allchin 1995: 46, cf. Allchin & Hammond 1978, Kenoyer 1998), it is a western import, as it originated west of the Zagros and south of the Caucasus. In S. Asia it is found as early as the 7th millennium BCE. This leaves several thousand years before the attestation of the S. Asian words for 'wheat', Ved. godhUma, Kan. gOdi etc. These are clearly related to Near Eastern ones, e.g. Old Egypt. xnd, Hitt. kant, PSemit. *HanT. The individual track of the loan word differs, however, just as in the case of the word for 'plough'. A form *gant-um that has entered via the northern Iranian trade route (Media-Turkmenistan-Margiana/Bactria-Aratta/Sistan) has resulted in Avest. gantuma and the later Iranian forms: M.Pers. gandum, Baluchi gandIm, Pashto ghan@m < *gandUma?, Yigdha gondum, Shugni Zindam; Khotanese ganama < *gandama, etc. (see Berger 1959: 40f, EWA II 498). The Iranian form has also been taken over by the Drav. newcomer in the region, Brahui: xOlum < IA *gholum (CDIAL 4287), from Bur. according to Berger (1959: 42). However, Bur. guriG, gureG (pl.), gha'rum < *ghor-um < **ghund- (Berger), seem to have been borrowed from the Indus language. (Berger thought of a loan from Bur. into the Panjab area languages; cf. also Bur. gur 'barley, wheat colored', bur 'buck wheat' Berger 1959: 43) When this word entered the Panjab it must have changed its initial syllable (*gan-) to go-, thus *godum, a change echoed by the Southern Indus language (*godi). Vedic has godhUma and similar continuants (Turner, CDIAL 4287). This is a clear folk etymology: the unfamiliar *gantum/gandum > *godum was analyzed as go-dhUma 'cow smoke'. Another form of the Near Eastern word that has come via the Southern route (Elam/anSan-simaSki/Tepe Yahya-marhaSi/Bampur) has resulted in Meluhhan *gOdi. This is retained in Drav. *gOdi (Kan. gOdi, Tam. kOti, cf. DEDR 1906). The change from -an- > -o- is not unfamiliar in Sindh (see below). A pre-Iranian *gantum must have become *go-tum or *go-dum in Sindh. The Drav. word, too, seems to be a popular etymology of the unfamiliar *godum: 'low red plant'', reconstructed by Southworth (1988: 658, 660) as PDrav. 3 at c. 1000 BC as *kO-tumpai. Maybe he thought of DEDR 3334 Tam. tumpai etc. 'nettle, weed' etc. (cf. Tam. kOtumam, Mal. kOtambu?). The exact development from *tumpai > -di would then not clear; (at this supposed late date kOtumpai could even be based on RV godhUma!) Obviously, in this case both the Northern and Southern Indus language have changed -an- > -o, while the Northern language otherwise retains -an- (see below). The northern form, based on Pre-Iranian *gantum would have resulted in Vedic **gan-dhUma or perhaps **gandha-dhUma "perfume smell', cf. CDIAL 4020 Skt. (lex.) gandhAlu 'fragrant rice', Pashai ganda'r 'a kind of grain'. The Southern (Meluhhan) *godi must have influenced a northern *gantum/gandum that facilitated a later Vedic popular etymology as 'cow smoke'. The mechanism of this influence is unclear. It may be due to Dravidian influence on the Panjab in the Middle/Late Rgvedic period; note that godhUma appears only in early post-RV texts. In short, the inhabitants of the northern Indus region (Panjab) thus must have called their wheat something like *godum and those in the Southern Indus region (Sindh), *godi. $ 1.10. Further dialect differences The strange sound change *an > o is not isolated. It also occurs in the migrant word of culture for 'hemp': Ved. zaNa (AV 2.4.5, PS 2.11.5 zaNa), M.Pers., N.Pers. San, Khotanese Saka kaMha (but gAndhArI > Niya Pkt. SaMNa), Osset. goen, goenoe, (Greek ka'nnabis, EWA II 605; Engl. hemp, etc.). It appears, again, in Dravidian, with popular etymology, as Tel. gOnu, gO~gu, cf. gOGgUra, Kan. gOgi, 'hibiscus cannabinus' (DEDR 2183). The original northwestern form is guaranteed by the North-Iranian (Ossete), Greek and Germanic forms of the loan word: kanna-bis, hemp, etc. The northwestern dialect has preserved *-an-, for example in the Rgvedic, yet certainly pre-Indo-Aryan tribal name of the gandhAri (and in the later Vedic country gandhAra). The northwestern name zambara (in the Afghan. hills), too, has not been changed to *zobara, but note the name of a poet in the more southern RV 8, sobhari kANva. We have a clear distinction between N. Indus -an- and Southern Indus -o-. (Note that original *-an- appears in post-RV texts further east and south, in Dravidian, as -o-). This is again a point that may turn out to be of importance for the decipherment of the Indus script which indeed has several features (special signs) that are different in Harappa (N) and Mohenjo Daro (S), (see B.Wells 1998). This is the opportune moment to briefly discuss another northwestern peculiarity, the interchange of k/z in Vedic. This has occasionally been observed, even one hundred years ago in the case of karkoTa/zarkoTa, but it has not been put into proper relief (Kuiper 1991: 41, 42, 44 as Proto-Munda, cf. KEWA III 309, Witzel 1999). The interchange of k and z is not related at all to the well-known Indo-Ir. development of IE *k' > Ved. z, as the present variation occurs only in 'foreign' words. The name of the snake demon zarkoTa (AV) appears also as karkoTa(-ka) RVKh 2.14.8, and locally especially in Kashmir and Nepal; cf. Bur. hergin (Berger hargi'n) 'dragon' or rather gha'rqa (Berger gha'rqas: CDIAL 3418?) 'lizard', Skt. karkaTa 'crab', Mundari kaRkom etc. (Pinnow 1959: 341 $483d). The prefix zar-/kar- can be connected with [s@r-] of the '300 foreign words' (Kuiper 1991: 40-1, 1948: 121), for example in sRbinda (Kuiper 1939 = 1997: 3 sqq.), ku-sur(u)-binda, bainda (Bind tribe), post-Vedic vindh-ya. Further materials include kambala/zambara 'blanket/name of a demon', kabara/zabara, kIsta/zISTa 8.53.4 (with var. lect. zIST-, zIrST-, zIrSTr-, see above), kimIdin/zimidA- 'demon/a demoness', kambu/zambu 'shell' (Kuiper 1955: 182), cf. KU-zAmba, Kau-zAmba 'name of a person', cf. ki-zora 'filly' AV, 'youth' CDIAL 3190 : zi-zu 'baby', zi(M)-zu-mAra 'Gangetic dolphin', zizUla 'dolphin' RV (EWA II 641-2; Le'vy, in Bagchi 1929: 121 sqq.), kirAta/cilAda 'a mountain tribe', kiknasa 'ground grain' AB: cikkasa 'barley meal' lex., Bur. Son ~ Ved. kANa 'blind' RV. The realization [k'] or [z] of an unknown phoneme (probably k') would easily unite such words as zam-bara : kam-bala, zabala : kabara; it would also offer a better candidate for Pinnow's unexpected reconstruction for the Munda and Mon-Khmer self-designation *Sqawar > zabara AB, and in the tribal names > Sora, Hor, Kora, Kherwar, Koro/Korku, Khmer etc., Pinnow 154 $311); rather from * k'awar, *k'amwar. In consequence, Vedic loan words with the interchange of z / k may go back to a phoneme K' with realization close to [k'] or [z] in the Indus language. Millet Another dialect difference can be observed in the "new" import at the time of the Indus civilization, millet. This domesticated plant has originated in China and another variety in Africa (Southworth 1988: 665, Randhawa 1980: 504; Nurse 1983, summarized by Cavalli-Sforza 1995, see now Meadow 1998). The Chinese words have no similarity to the Indian ones (Karlgren 1923), and the source of the Indian words has not been established so far: any language between the Sahel belt and Baluchistan is possible. It has to be noted, that in the case of this comparatively late import, -an-, -am- has been preserved both in Proto-Munda *gaGgay, Dravidian DEDR 1084 kaGgu (Tam. kaGku), DEDR 1242 kampu, Ved. priyaGgu, OIA dialects *kaGkuna, *kaGguna, *taGguna (which may provide some indication of the time frame for the words discussed above). Even though comparisons between the various words for 'millet' can be made, they cannot be traced back, as is the case with many widely spread loan words, to a single source. Hindi kaGgnI can be compared with OIA *kaGkunI CDIAL 2606, with Tamil kampu DEDR 1242 and with Munda *gaG(-)gay (Southworth 1988: 660, Zide & Zide 1973: 8). The source of these words may have had a form such as **kaG-CV. From this, Ved. priyaGgu (EWA II 190) can be derived as well, as it seems to have been changed by popular etymology, like several other agricultural terms: prefix *p@r- (Kuiper 1991: 42f.) > *priya+gu 'dear cow'. Other IA designations of millet are: Ved. aNu and *aNuni CDIAL 195. All of this points to a contamination or cross of *kaGgu and *-(k/g)aGgu --> IA aNu; (*al 'to mill' EWA I 55; rather a Munda change, Pinnow 1959: 198f., k/*q > 0 typical for Sora, Kharia k : Sora 0; thus: kaGgu : *aGgu --> Ved. aNu, cf. Kuiper 1991: 38). In short, all major language families of S. Asia have taken over the word from an unknown, but not exactly the same source. Nevertheless, a clear difference between Northern and Eastern/Southern forms is visible: PDrav. *kampu is opposed to PMunda *gaGgay (Zide & Zide 1973), while the IA forms stand in between the two. The usual IA form is Ved. aNu (cf. O.Indo-Aryan *aNunI, Turner, CDIAL 195). However, based on Ved. pri-yaGgu < *p@r-gaGgu? and the reconstructed OIA forms *kaGkunI, *kaGgunI, *taGgunI (CDIAL 2606), a northwestern Indian *kaGkun, a central-northern *kaGgun, a more eastern North Indian *taGgun can be reconstructed for the pre-Vedic period, while the Southwest must have had, next to Drav. *kampu DEDR 1242 (= Skt. kambU, in hemAdri) also a form *kaGgu CDIAL 2605, DEDR 1084. The northern Indus language should have had *kaGku(n), its southern dialect (Meluhhan), *kaGgu. The modern languages also do not agree: In Hindi (Masica 1979: 76 sqq., 135f.) we find various terms for the many varieties of millet: kaGgnI (*kaGkunI CDIAL 2606); kuTkI (Masica from Skt. kuTakA, not found in the dictionaries; cf. kuTaka 'a kind of tree' KauzS.); kodoN (CDIAL 3515 kodrava 'grain eaten by the poor' Mbh., cf. koradUSa 'idem' Suzr., -ka KZS; DEDR 2163 Tam. kural, Kan. koRale, korle; Konda koren 'a grain'); khil (Masica: from Skt. khiD), junhAr, j(u)wAr) (*yonAla > yavanAla > juAr, < Drav. *coN2N2el, DEDR 2359, DEDR 2896, CDIAL 10437); bAjrA (Vedic: HZS varjarI, CDIAL 9201 *bAjjara); ma(N)RUa (CDIAL 9728 < maDaka 'the small grain Euleusine corocana'); sANwAN (Ved. zyAmaka VS, CDIAL 12667). Some of them belong to the c. 30% of agricultural vocabulary in Hindi that comes from Masica's "Language X". Finally, as pointed out above, the word for 'peacock' must go back to a northern Indus form *mayur > Ved. mayUra RV level II, and to a southern form *mayil/r > Drav. Tamil mayil, Irula muyiru, Tulu mairu, Konda mrIlu, miril etc. In summing up, it can be stated that in the north-west and also in the Panjab, as represented by loan words in most of the RV, original northwestern *-an- is opposed to southern -o-. The same relationship is also found in north-western z : subcontinental k, north-western -J- : subcontinental zero in the word for 'rice'. We can discern a clear difference between the Panjab (-->Vedic) and Sindh/Gujarat (--> Dravidian) forms of the Indus language. Dialect differences between Panjab and Sindh seem even to be indicated in the Indus inscriptions themselves. Seals and plates from Harappa (Panjab) differ in a number of items from those found at Mohenjo Daro (Sindh), for example in the sign for 'container, quantity' which looks like a V; this is almost only found at Harappa (B. Wells 1998). The same applies to some 'suffixes' in the inscriptions (Wells, by letter 1999). It can be concluded that the Meluhhan variety of the Indus language was the 'original' language of Sindh. Was it also the Indus trading language? In that case, it has disappeared, just like Sumerian and Elamite, and traces may at best be found in Sindhi -- a step that has not been taken. There is no etymological dictionary of Sindhi. $ 1.11. Dravidian immigration The observations about the early linguistic evidence from Sindh, made above, indicate that speakers of Dravidian were not a primary factor in the population of the Indus civilization, even of Sindh, and that they were immigrating into the Panjab only in middle Rgvedic times. But when could they have entered South Asia? Earlier scholars (Heine-Geldern 1964, Pinnow 1954: 15) thought that they entered S. Asia (sometime as late as the early 1st millennium BCE) and proceeded via Baluchistan, Sindh and Gujarat to S. India (Zvelebil 1970, 1990: 48, 123). Indeed, their tracks are still visible in certain place names in Sindh, Gujarat and Maharashtra (see above). According to Southworth and McAlpin, however, the semi-nomadic speakers of Dravidian who even had contacts in Iran with the pre-immigration Indo-Aryans (Southworth 1979: 203, 228 f., 1990: 222-3, 1995), came to S. Asia relatively late, but early enough to participate in the Indus civilization, from which they acquired agriculture and the accompanying vocabulary. This scenario, if applied just to Sindh, explains why the c. 300 foreign words of the RV (in the Panjab) with their (agricultural) vocabulary are relatively free of Drav. influence. According to the indications given above, the Dravidians apparently were just as foreign to Sindh and its agriculture as the Indo-Aryans to the Panjab. As the Northern Indus language (Para-Munda) differs considerably from the Southern one (Meluhhan), it seems likely that the speakers of Indo-Aryan entered the Panjab and acquired local words from the Northern dialect (zaNa, lAGgala, vrIhi, godhUma, kaGgu, gandhAra), and that the Dravidians entered Sindh at or about the same time and acquired such words from the southern dialect (gOnu, JAJcil, variJci, godI, kaGku/kampu). It may even be the case that the first who made horses statues at Pirak (1700 BCE) were Dravidians, not IA bhalAnas. For the first use of horses must not necessarily be linked to speakers of an IA language. The Drav. words for 'horse' underline this: DEDR 500 Tam. ivuli, Brah. (h)ullI, 1711 Tam. kutirai, Kan. kudire, Tel. kudira, etc., 3963 Tam. pari 'runner', 4780 Tam. mA 'animal' (horse, elephant), Tel. mAvu 'horse, (cognates mean 'deer' etc. in other Drav. languages), cf. Nahali mAv 'horse'. These words are quite different and independent of IA azva 'horse' and various words for 'runner' (arvant, vAjin, etc.), etc. On the other hand, the technical terminology for chariots is IA and IE. It has been taken over into Drav.: akSa 'axle' RV > Parji-Kolami accu 'axle'; ANi RV > ANi 'lynch pin', ara RV > Ar 'spoke' (cf. Southworth 1979: 230 n. 14). Note that the earliest IIr *ratha 'chariot (with two spoked wheels)' (Gening 1977, Pigott 1992, Anthony u. Vinogradov 1995, cf. Littauer u. Crouwel 1996) is found about 2000 BCE, near the Volga (North Iran. *rahA > Greek rha~ = Avest. ranghA, Ved. rasA). The IIr word for 'chariot', however, is old enough to have resulted in the archaic compounds Ved. rathe-SThA, Avest. rathaE-Sta- 'chariot fighter', cf. Old Avestan rathI, RV rathI 'chariot driver.' Dravidian has nothing of this, but possesses words for 'wagon' or 'bullock cart'. An early wave of Dravidian speakers might very well have preceded the IAs into Iran and S. Asia and some may have stayed on in SE Iran. (Note the strange absence of the western Baluchistan country of Maka in the Avestan record of "Aryan countries" in V. 1, cf. Herodotos 3.94). A few IA loans in Proto-Drav. would settle the case, but culturally decisive words, such as for the newly introduced horse, the chariot, or other pastoral terminology do not exist. The Dravidians hardly had any previous contact with the Indo-Aryans while still in Iran. Contra Southworth (1979: 196f.), there is little secure evidence for early loans from IA into Drav.; such words can have been taken over any time between the RV (1200 BCE) and the earliest attestation of Tamil at the begin of our era (see above, on Drav. evidence in Vedic). There are only a few questionable loans that might have come from the pre-immigration period, that is from hypothetical contact when still in Iran; these remain speculative; perhaps one can think of a common source for Ved. gar-da-bha EWA I 473, Drav. kal~u-tai DEDR 1364 'donkey', similar to Ved. khara, Avest. xara. $ 2. Eastern Panjab and Upper Gangetic Plains $ 2.1. The Kuru realm We return now to the epicenter of post-Indus developments, the area of Eastern Panjab-Haryana-Uttar Pradesh, in other words, the lands from the Pakistani border up to Allahabad. In the early post-RV texts, its hub is kurukSetra, northwest of Delhi. This is the realm of the middle Rgvedic Bharata and the late Rgvedic Kuru (Witzel 1997). The Bharata tribe and its successor, the new tribal union of the Kuru, represent a new wave of IA immigrants from the other side of the Indus (vasiSTha RV 7, JB 3.238-9 $204), which brought new linguistic traits with them (kuru for older kRNu, sarva for vizva, etc., Witzel 1989). The Kuru dialect is remarkably more modern than the language of the bulk of the RV. However, RV book 10 often reads already like the next level, that of the AV and other Mantra texts of the Kuru period. The Kuru confederation, supplanting the 50-odd Rgvedic clans and tribes, became the center of linguistic (Witzel 1989), religious and social (Witzel 1997) development. They formed, together with partly IA- acculturated Indus people (Arya-tribes such as the anu-druhyu, yadu-turvaza) and with the new addition of Dravidian speakers, a new society with a new elite kit (Ehret). This included pastoralism (cattle, horse, sheep, goat), IA ritual and acculturated customs, IA religion and ritual, but also post-Indus type agriculture (barley, wheat, rice, millet) and local artisans (potters, etc. see below). The new culture, Vedic orthopraxy and social system (with four classes) then spread eastwards into the Gangetic plains, and ultimately to Bihar. Because of the amalgamation of the three groups (IA, Para-Munda, Drav.) discussed above, we have to suppose a large degree of bilingualism and even trilingualism, and the forming of pidgins. (Kuiper has a forthcoming paper on a 'bilingual' Vedic poet). A Vedic pidgin must have been used at home, and proper Vedic Sanskrit was learnt 'in school', at the time of initiation of boys. While the lingua franca was a form of late/post-Rgvedic IA, pockets of the Para-Munda Indus language, of the newly arrived Dravidian as well as some remnants of the Gangetic Language "X" must have survived as well. Among the post-Rgvedic texts, especially the AV is full of non-IA, 'popular' words of plants, animals, demons, local deities, and the like. Their character still is, by and large, Para-Munda, with some words from the 'local' language ("X"), and with some Drav. words included; all of which is clearly visible in the increase of words with retroflexes. The linguistic situation is reflected, among other items, in the mixture of IA and other river names in the area. The famous sarasvatI is also called vaizambhAlyA / vaizampAlyA / vibalI; these names and that of the nearby vipAz < *vipAL/vipAZ all seem to go back to a local word, *vi-zam-paZ/-paL, (Witzel 1999). However, and typically, there are no Dravidian river names in the whole Kuru area. A hint of how Drav. influence on Vedic was exerted is contained in the name of the zUdra. From the late RV (10.90) onwards, this designates the fourth, non-Arya class; it was added to the three 'Arya' classes of Brahmins, kSatriya (nobility) and vaizya ('the people') only at this time. However, Greek sources of Alexander's time still place a tribe, the sudroi, at the confluence of the Panjab rivers with the Indus; this may still indicate their origin in Sindh/ Baluchistan. Drav. words first appear in Middle and Late Rgvedic, in RV 3, 7, and 8, especially in the kANva section. Interestingly, it is tura kAvaSeya, the great-grandson of the Drav.-named kavaSa 'straddle legged', a priest on the 'wrong side' in the great Bharata battle (RV 7.18) who becomes an influential priest in the Kuru realm and who developed the new, post-Rgvedic (zrauta) rituals (Proferes 1999). It has been stressed by Burrow (1973 : 386) that the post-Vedic texts have more Dravidian words; indeed, the evidence of Para-Munda words, too, is not diminishing but increasing during the Vedic period. This is the case right from the Mantra texts, and includes the yajurveda saMhiTAs whose territory can be easily established (Witzel 1987, 1989, 1997) as that of the area between E. Panjab (Lahore), Allahabad and the Chambal River area (Ujjain). A complete discussion of the c. 200 longer or shorter Vedic texts must be postponed to a separate paper (for some lists, see below). In the mean time, one can compare the word index to the AV (Whitney 1881), or Vishva Bandhu's Vedic Word Concordance (in Devanagari script), in conjunction with EWA, KEWA (and DEDR). The new tribal union of the Kuru (and their more eastern allies, the paJcAla), with their new social set-up and ritual expanded, incorporating the surrounding tribes, eastwards into the Gangetic plains, in a partly military, partly peaceful fashion until it reached northern Bihar (Witzel 1995, 1997). The eastern tribes were at first regarded as half-barbarian (JB 1.337 $115) or 'asurya' (demonic). The same is seen in archaeology: late Harappan people emigrated towards the Upper Gangetic plain (the only movement of people the archaeologists allow for the whole period under discussion here, Shaffer 1995: 139, cf. Allchin 1995: 33-35), a fact reflected in the Vedic texts as well. The emigration was possible due to a new type of agriculture, permitting cultivation of rice during the monsoon as well as wheat and barley in winter, resulting in a food surplus. The settlement at first occurred along the river banks, in half-nomadic treks (grAma, Rau 1997). This is reflected by the Painted Gray Ware culture, with their clear elite pottery whose regional motifs indicate the split into western Kuru and more eastern paJcAla, something that is also seen in the Vedic dialects they use (Witzel 1989). Not everybody is included: The non-IA kIkaTa (3.53) or the paNi are clearly described as foreigners (late RV hymn 6.45.31), and even later, in the Mantra and YV saMhitA period, the niSAda in the Chambal area (MS 2.9.5 etc.) and other dasyu 'enemies' (JB, Witzel 1997: n.161, 163, 278); in RV 10.61.8 as well the South (i.e. the area south of kurukSetra) still is the land to banish someone. $ 2.2. The substrates of kuru-paJcAla Vedic. As has already been indicated, the features of the Rgvedic substrate language are also found in post-Rgvedic texts that were composed further east in the kurukSetra and in western Gangetic plains, as well as in the Chambal area. These words are not just the same as found in the RV, but there are many new ones. In the Mantra period, starting with YV (MS, KS, TS) and AV/PS, we can clearly distinguish all three linguistic elements: * Indo-Aryan with some already incorporated north-western elements such as Nuristani kAca 'shining piece of jewelry' or Burushaski kilAy ~ RV kIlAla 'biestings, sweet drink', Bur. Son ~ RV kANa 'blind in one eye', Bur. bus ~ RV busa 'chaff, mist', (cf. Pinnow 1959: 39), etc.; * The Indus substrate (Para-Munda), that also is found in the Ganges area (next to some elements of language 'X'), such as RV kuzika, karaJja, kaGkata, ziMzapA, ziMzumAra, puSkara, puSya, especially the words with prefix C@r (p@r/k@r/s@r-), kar-koTa-ka RVKh ~ zar-koTa AV, tila AV: jar-tila KS, kalmaza MS, KS, kal-mASa PS, kul-mASa Up. : mASa AV, with the -Ta, -zA/Sa -suffixes, and with -ND-: ka-maNDalu : maNDa-la, kaNTha? PS, etc. * The Middle and Late Rgvedic Drav. element also is found in the Ganges area: godhUma AV (Hindi gehu~ etc., Kusunda gabun), kuNapa AV, kurkura AV, cUDa ZB, coDa TS, eDaka JB, arka ZB, bilva AV 20 (Kuiper 1991:66), -nIra- ZB, etc. In short, the upper class IA language (of the Vedic priests) used in the upper Gangetic plains contains the same substrate elements as seen in the late Rgvedic period of the Panjab. However, due to the increasing stratification of society and increasing specialization among occupations, many words from the sphere of the artisans and from technology were added; furthermore many names of persons, localities and rivers. Their affiliation can still be ascertained to some extent. With regards to agriculture, Kuiper's RV list (Kuiper 1991: 8, 21, 96, see already Kuiper 1955) contains quite a number of such terms (kInAza, lAGgala, bIja, etc.) Especially among the artisans there is an increasing number of non-IA designations; many of them first appear in the azvamedha (MS kevarta, kaivarta TB). (fn. 15) Some of them are, in line with the increasing specialization, new Indo-Aryan formations (anucara 'servant', grAma-NI 'leader of a trek, wagon train' etc.), but especially those of fishermen (kevarta/kaivarta, dAza, dhIvan, daivara, puJjiSTha, pauJjiSTha, bainda, mainAla) are non-IA (often until today). Furthermore, non-IA specialists are: musicians (talava, ADambara-AghAta, dundubhy-AghAta (cf. dundubhi RV), vINA-gAthin, vINa-vAda, cf. vINA KS (EWA II 568), artisans (kaNTakI-kArI, bidala-kArI, also kulAla, and the pAlAgala 'messenger' (cf. pAlAgalI 'fourth wife of a chieftain'), gaNaka 'astrologer' (cf. gaNa RV) and 'usurer' (kusIdin, kusIda KS). Such words come up not only in the eastern parts of North India (Bihar, area of VS/ZB) but also everywhere from the Panjab (RV) and the Delhi area (MS, KS) eastwards, e.g. kInAza RV, gaNa RV, dundubhi RV, vINA KS, kusIda KS. The newly attested words have the same 'foreign' grammatical formations as seen in the RV: prefixes (ke-/kai-, dun-dubhI?), retroflexes (ADambara, kaNTakI-), initial b- (bidala), suffix -Ala (pal-Ala, main-Ala, cf. Oberlies 1994:341). Similar data could be supplied for the spheres of material culture and the surrounding nature: agriculture and domesticated plants, local animals and plants, many items of food, illnesses and poisons, implements and utensils, and ornaments; this would lead to far afield in present context (see the lists in MacDonell-Keith, Vedic Index, Delhi 1967 [1912] 517-92). For more examples, one can consult Mayrhofer, EWA and for non-IA details especially KEWA; these may serve, in connection with CDIAL, DEDR, Kuiper 1948, 1955, 1991 and Pinnow 1959 as a first orientation. $ 2.3. The Para-Munda substrate. Prefixes with ka- are found in the AV, YV and the brAhmaNas (here follow only a few proposals for etymologies; it is to be expected that not all of the following words can be divided in the way proposed below; ultimately, this depends on a fitting etymology): kapaTu AV, PS, cf. with Sora pud-@n, Sant. od etc. (Pinnow 1959: 121 $237; kapAla AV; kapiJjala PS; kapola RVKh, cf. Sant. puTi 'to swell', Kharia poTki 'to sprout' etc. (Pinnow 1959: 173 $378) ~ puTa 'bundle, bag' MS, BZS; kaphauDa AV, see Kuiper 1948: 44; kamaNDalu KS cf. maNdala etc.; karIra MS, KS; karIS-in AV; karuma AV; karUkara AV; kalApin ZS; kaliGga AB, cf. Skt. tri-liGga, etc., see Kuiper 1948: 45; kavaca PS (but see above, Zvelebil's no. 13); kazambhUka SuparN.; kazipu AV; kazIti JB; kazoka AV; kazmaza? AV; kaSAya ZB; kaSkaSa? AV; kasarNIla AV, cf. sarNika TS/sRdIka MS (cf. sRdAku?); kasAmbu AV, etc.; kastUpa, kastUpa-stopinI PS, cf. stupa KS/stuka RV; kahoDa ZB, JB. With 'double prefix' C@r-/C@l- there are the following words in which the many variants of the prefix in k@r- stand out: karkandhu MS, KS; karkI? AV; karkoTa-ka RVKh ~ zarkoTa AV, PS, cf. Mundari kar-kom (Pinnow 1959: 341 $483d), Kuiper 1991: 41, 44, 1948:121, Bur. gharqas 'lizard'; kardama KS, cf. Munda ko-dil, @-dil 'dirty' (Pinnow 1959: 87 $101); karpAsa Suzr., kArpAsa ZS; karzapha AV, PS : zapha?; garmut TS, gArmuta MS (Kuiper 1948: 146, CDIAL 4063: Sindhi gamu 'a sort of grass'); kalkuSI PS; ZB, kalmali AV; kalmAza MS, KS, kalmAza- ZS, PS; kArSmarya KS; kharjUra 'date palm' KS; gulma? SaMh.; jar-tila 'wild sesame' KS : tila 'sesame' AV; jarvara PB; jalASa PS (or -ASa suffix); palala SU., palAlI AV; palAva AV; palIjaka AV; barjahA, barkara ZS; barbara KS; barhiNa ApDhS; bharUji AV; marIca ApDhS; markaTa KS/markaTaka ApZS; zarkara AV, cf. Bur. ghoro?; zarkoTa AV, PS (see above karkoTa); sardigRdi TS. Double prefix C@n-/C@m- in: kaGkUSa AV, PS ~ zaGku; kaNTha? PS, (saha)-kaNTh- AV, cf. Kharia konko, Khmer ko, Mon ka "possibly old compound", Pinnow 1959: 132 $ 276; kANDa? AV, cf. Kharia koNDen 'bamboo', (Pinnow 1959: 132 $275); kaNDUy-? KS; kandhara Up; kambala AV ~ zambara?; kambUka AV ~ zambUka; kamboja PS, cf. Greek Ambautai; kAmpIla- KS; jAmbila KS, TS; taNDula AV; talAza? AV (if not with -Aza suffix); parUSaka ZS; palANDu ApDhS; palAza TB (if not with -Aza suffix); palIjaka AV; palpUlana AV; palvala SU; pAlAgala ZB, -I ZB; barza? KS, barzva? KS; balAsa PS, balkasa ZB; balbaja RV; balbUtha RV; bhalAnas RV. From the post-Rgvedic materials come words with other prefixes in C@r- and with other vowels, etc.: kirika YV, girika MS; kirmira VS, etc.; kul-mASa Up. cf. mASa AV; ku-Taru YV, etc.; sRdAku 'lizard', etc., lex., sRdAku/-gu MS, sRdara 'snake', etc. Mayrh. ZDMG 110, 6189 Munda prefix sR- + dak 'water', see KEWA s.v. sRdAku, etc.; kazmaza? AV; kaSkaSa? AV; jASkamada AV; maSnAra AB; masUra? KS, masura TS; etc.; prakubrata ZB, prakudrata ZBK, pramota AV etc.; tilvaka ZB, tailvaka MS, etc.; tumbara KauzS etc. Further Vedic words which are suspected of a Para-Munda origin are, among others: me-khala AV: zR-G-khala Skt.; khaDga MS, EWA 443, cf. N.Pers. karka-dAn, Arab. karkaddan, Aelianus karta'zOnos (*kargazOnos) 'Indian rhinoceros', cf. Kuiper 1948: 136 sqq.; karta/garta to be compared with Kharia gaRha 'river', Mundari gaDa, gaRa 'pit, trench, grave, water course, stream, river'; Sant. gaDa 'hollow, pit, excavation, trench, river'; etc. (Pinnow 1959: 351f. $ 498); tittira KS, MS cf. Korku titid, Santali sengel titi 'Guinea fowl': Kharia khontheD, Sora on-tid-@n (Pinnow 344 $ 488a); probably also: musala AV; jala? RVKh, PS; dhUkSNa/dhlukSNa/dhLkSNa PS, jhaSa ZB : jaSa AV, TS : caSa VAdhB; drumbhUlI MS / dAlbhuSI KS / class. dambholi, see Kuiper 1991: 26 (cf. p. 18, 47, 61, 75). Para-Munda suffixes. In order to characterize the substrate, certain typical suffixes can be used. Kuiper (1991: 45 sqq.) has isolated the following in the substrate of the RV: -Ala, -ASa,-ISa,-USa/-Aza,-IzA,-Uza, -Ta, -nas, -ya, -ra, -za/Sa, -ha. Among the suffixes are to be underlined in this context are those often found in personal and tribal names, in -Ta (KIkaTa, kRpITa, birITa, kevaTa RV / avaTa SV), and the ones in -Ala/-Ara (kIlAla, caSAla; mainAla VS, cf. IA karmAra RV 'smith'; GandhAri RV, GandhAra, AbhisAra etc., cf. Witzel 1999). Such suffixes also appear in post-Rgvedic time in the texts of the Mantra period and in the Yajurveda-SaMhitAs, e.g. kalmASa 'spotted' VS, TS; niSkASa 'scraping' MS, KS; yevASa 'an insect' AV, evaSa MS 4.8.1:107:16, yavASa KS 30.1, KpS 46.6 (vRSaz ca yavASaz ca); RjISa a name of Indra, RV, 'residue of Soma' AV; uSNISa 'turban' AV; karISa[-ja] PS, 'dung', karISin AV, karISa ZB, (cf. the frequent purISa 'dung'); cf. also tUSa 'border of garment' KS; later also: palAza 'leaf' TB, ZB, ni-palAza ZB, zirISa 'Acacia sirissa' SaDv.B, etc.; cf. also jhaSa 'a certain large fish', ZB jaSa AV, TS, caSa VAdhB. Para-Mundas in KurukSetra and in the Gangetic plains. The words mentioned above clearly show that also in post-Rgvedic, i.e., in the Mantra texts (AV, SV, RVKh, YV), in Yajurveda Prose, and in the BrAhmaNas, such Para-Munda words can still appear for the first time. Therefore, they had either already existed in Vedic colloquial speech or they entered Brahmanical High Vedic at that particular point in time from the sphere of village life or of the artisans. The area of the early post-Rgvedic texts (Mantra texts, YV Prose) can be localized fairly well (Witzel 1987, 1989): it contains KurukSetra (i.e. more or less, modern Haryana) and the western gaGgA-yamunA-doAb (i.e. the Gangetic plains of western Uttar Pradesh). In these areas, where no modern groups of Munda speakers survive, the same Rgvedic substrate with its typical prefixes can be found. That means Haryana and Uttar Pradesh once had a Para-Munda population that was acculturated by the Indo-Aryans. If the late Vedic texts (such as the JaiminIya Br. and zatapatha-Br.) are added, the area in question is further enlarged to include the regions south of the Ganges and east of Uttar Pradesh. Here, new Munda words appear as well; however, these regions include those where even today Munda languages are spoken. In short, a strong Austro-Asiatic substrate is found both in the early Panjab (RV, c. 1500 BC) as well as later on in the Ganges valley (YV saMhitAs, brAhmaNas, c. 1200 v. - 500 BC.), a fact that can also be shown in the names prevailing in these areas (Witzel 1999). As examples, I mention the river names gaGgA (popular etymology of Munda ga(N)D), gaNDak-I (see below), narma-dA, and tribal names such as maraTa, vibhindu (and vibhindukIya, cf. nAr-ka-vinda PS 12.2.3, sR-binda RV (Kuiper 1991: 40-43, 1997), ku-suru-binda TS, TB, SB, ku-sur-binda JB, bainda VS, cf. Munda bid 'insert, plant, sow', Pinnow 1959: 143 $285), zabara (*Zqawar, cf. Pinnow 1959: 154 $31; rather from *k'awar/zawar), puNDra, aGga/vaGga (cf. also gaGgA?; further: pra-vaGga), kaliGga (cf. teliGga/triliGga, see S. Le'vy in Bagchi 1929: 100, cf. Shafer 1954: 14, 122 as Tib.-Burm.; Kuiper 1948: 45 compares kuliGga 'fork-tailed shrike' Mbh., and *liG in Munda, Khasi, Mon, Khmer, Malay); ikSvAku (RV, emigration from the Panjab eastwards, Witzel 1997: 307 sqq., 321, 1989: 237), niSAda/ *nisadha/naiSadha, mucIpa/mUtIba/muvIpa, magadha (cf. pra-maganda), zaphAla cf. zAvasa, vasa etc. However the truly eastern words (Uttar Pradesh, Bihar) are, next to some remnants of language "X", of Munda nature: there are many personal and place names (Witzel 1999), e.g. that of the river gaNDak(I), or even that of the Ganges, with popular etymology: gaGgA, a sort of intensive formation of gam 'to go' (if not modeled after the tribal names aGga, vaGga). Pinnow (1953-4) has pointed out many river names, from the gaNDakI to the narma-dA which contain the Munda element -*da', *-da'k 'water' (Pinnow 1959: 69), for gaNDa(kI) cf. Santali gADa, Ho gaDa 'river' (Pinnow 1954: 3). The gaNDakI is not attested in Vedic, and is referred to as sadAnIra 'always having water'. Apart from the Epic, it appears in local context, the early Licchavi inscription (467 CE), Sanskritized as gaNDakI and in other Skt. texts: kAla-gaNDikA, gaNDArikA, apara-, pUrva-; the shorter version, gaNDI, appears from the Epic onwards, and several times early on in Nepal as gaNDi-(gulma-viSaya) (998, 1092, 1165 CE, see Witzel 1993). The gaNDaka appear as people in Mbh. as well. Further, tribal names such as pulinda/Pali bUli, Pali moriya (from Skt. mayUra 'peacock') and also mara-Ta (PS), (from Munda mara 'peacock'), kunti from Munda kon-ti'd 'bird', cf. RV za-kunti, Epic za-kuntalA, etc. (contrast the IA matsya 'fish' (RV), a tribe just west of the Kunti), mUtiba (mUcIpa), zabara (mod. Saora?), puNDra (Bengal), the aGga, at the bend of the Ganges, and the neighboring vaGga (Bengal). The prefix change in aGga (AV) / vaGga (AB) is indicative of a Munda formation (Kuiper 1991: 43). Mundas may also have lived in the hills and valleys of the Sub-Himalayas, for example in the Kathmandu Valley (see below, Witzel 1993). Other typical words of the Gangetic plains are, from west to east: sardigRdi TS, palAza TB, palANDu ApDhS, tumbara KauzS, kazIti JB, kirmira VS, kaSAya ZB, pra-kudrata ZBK, pra-kubrata ZBM, ka-hoDa ZB, JB, kul-mASa Up. etc. Especially informative for regional dialect features of the substrate, from W. to E.: jaSa AV, TS : caSa VAdhB : jhaSa ZB 'a certain large fish'. The Rgvedic substrate thus has the same grammatical structure as the words in the yajurveda-saMhitAs and the brAhmaNas that newly appear from the substrates of the KurukSetra (Haryana) and Ganges regions (doAb, Uttar Pradesh). It is of great importance that we can detect the same Indus substrate as found in the RV. In other words, the Rgvedic Panjab as well as the post-Rgvedic Gangetic Plain were largely settled by speakers of Para-Munda (including remnants of Masica's 'Language X'). They had been joined, in the early Rgvedic period, by speakers of Indo-Aryan and, in the later Rgvedic period, by those of early Dravidian (see above). Dravidian In the new IA speaking, culturally Vedic "eastern territories" of the Gangetic plains some Drav. words occur for the first time in literature, e.g. nIr 'water' in the name of the eastern river sadAnIrA, the modern gaNDak (Witzel 1987), or the verb 'to speak in barbaric fashion', mleccha-ti. However Drav. nIr is not found in the neighboring N. Drav. languages (Malto, Kurukh), but is only found in Baluchistan (Brahui dIr, DEDR 3690). This may be accidental, but it may also indicate that Brahmanical educated speech of the Kuru with their IA-Drav.-Munda symbiosis and acculturation had incorporated some Drav. words which appear only now in the texts. The word mlecch has been discussed above. Its appearance in the eastern context is not surprising. From the point of view of the Brahmins, the easterners are 'foreigners', mleccha. The word may at first have designated only the southern (Sindh) foreigners, and later on all others. These central and eastern North Indian territories, however, have no Dravidian names; the river names belong to other substrates. A study of present and medieval north Indian places names has not been undertaken in earnest. We will have to account for such names as that of the town of goND(A) in Uttar Pradesh, some 180 km north of Allahabad. The name goND appears nowadays only on the Central Indian Vindhya mountains, and is not known in U.P. from medieval and classical sources. (For some supposedly Drav. river names such as sadA-nIrA from Drav. nIr 'water' see above, and for the varaNAvatI at Benares, see Witzel 1999.) There are, as always, wrong leads, such as the river name kankai in the Eastern Nepal Terai, which looks like the Tamil form of the name gaGgA (Witzel 1993); there are, however, no traces of an earlier S. Drav. occupation in the area. The Dravidian Kurukh living in the Terai now have recently been imported as laborers from Central India (K.H. Gordon, Phonology of Dhangar-Kurux, Kathmandu 1976) where they are known as Kurukh or Oraon. For a different view of early Dravidian settlements in N. India, see R. Shafer 1954, Parpola 1994: 168, and Burrow 1973 : 386. Burrow points to the fact that most of the Drav. loan words are found in post-RV texts and concludes: "the influence took place in the central Gangetic plain and the classical madhyadeza." Therefore, "the pre-Aryan population of this area contained a considerable element of Dravidian speakers." If that had been the case, we would expect some Drav. river names in the Gangetic plains. However, only Munda (and Tib.-Burm.) names are found (Witzel 1999). =========================================================================== Michael Witzel Elect. Journ. of Vedic Studies Harvard University www1.shore.net/~india/ejvs --------------------------------------------------------------------------- my direct line (also for messages) : 617- 496 2990 home page: www.fas.harvard.edu/~witzel/mwpage.htm