SINO-PLATONIC PAPERS Number 129 December, 2003 Linguistic Evidence for Cultural Exchange in Prehistoric Western Central Asia by Michael Witzel Victor H. Mair, Editor Sino-Platonic Papers Department of East Asian Languages and Civilizations University of Pennsylvania Philadelphia, PA 19104-6305 USA [email protected] www.sino-platonic.org SINO-PLATONIC PAPERS FOUNDED 1986 Editor-in-Chief VICTOR H. MAIR Associate Editors PAULA ROBERTS MARK SWOFFORD ISSN 2157-9679 (print) 2157-9687 (online) SINO-PLATONIC PAPERS is an occasional series dedicated to making available to specialists and the interested public the results of research that, because of its unconventional or controversial nature, might otherwise go unpublished. The editor-in-chief actively encourages younger, not yet well established, scholars and independent authors to submit manuscripts for consideration. Contributions in any of the major scholarly languages of the world, including romanized modern standard Mandarin (MSM) and Japanese, are acceptable. In special circumstances, papers written in one of the Sinitic topolects (fangyan) may be considered for publication. Although the chief focus of Sino-Platonic Papers is on the intercultural relations of China with other peoples, challenging and creative studies on a wide variety of philological subjects will be entertained. This series is not the place for safe, sober, and stodgy presentations. Sino- Platonic Papers prefers lively work that, while taking reasonable risks to advance the field, capitalizes on brilliant new insights into the development of civilization. Submissions are regularly sent out to be refereed, and extensive editorial suggestions for revision may be offered. Sino-Platonic Papers emphasizes substance over form. We do, however, strongly recommend that prospective authors consult our style guidelines at www.sino-platonic.org/stylesheet.doc. Manuscripts should be submitted as electronic files, preferably in Microsoft Word format. You may wish to use our sample document template, available here: www.sino-platonic.org/spp.dot. Beginning with issue no. 171, Sino-Platonic Papers has been published electronically on the Web at www.sino-platonic.org. Issues 1–170, however, will continue to be sold as paper copies until our stock runs out, after which they too will be made available on the Web. Please note: When the editor goes on an expedition or research trip, all operations (including filling orders) may temporarily cease for up to three months at a time. In such circumstances, those who wish to purchase various issues of SPP are requested to wait patiently until he returns. If issues are urgently needed while the editor is away, they may be requested through Interlibrary Loan. You should also check our Web site at www.sino-platonic.org, as back issues are regularly rereleased for free as PDF editions. Sino-Platonic Papers is licensed under the Creative Commons Attribution-NonCommercial- NoDerivs 2.5 License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/2.5/ or send a letter to Creative Commons, 543 Howard Street, 5th Floor, San Francisco, California, 94105, USA. Michael Witzel, "Linguistic Evidence for Cultural Exchange in Prehistoric Western Central Asia," . Sino-Platonic Papers, 129 (December, 2003) Linguistic Evidence for Cultural Exchange in Prehistoric Western Central Asia Michael Witzel q Abstract Recently discovered evidence suggests that there is a body of loan words preserved independently from each other in the oldest Indian and Iranian texts that reflects the pre-Indo-Iranian language(s) spoken in the areas bordering N. Iran and N. Afghanistan, i.e. the Bactria-Margiana Archaeological Complex. These loans include words from agriculture, village and town life, flora and fauna, ritual and religion. They were taken over and then exported to Iran and N. India by the speakers of the various Old Iranian and Old Indo-Aryan (Vedic) languages, as well as by a western off-shoot, the Mitanni Indo-Aryan of Syria/Iraq and by the language of related tribes indicated by some Indo-Iranian words in Kassite. All these represent series of intrusions by Indo Iranian speakers into the world of the great Mesopotamian, Bactro-Margiana, and Indus civilizations and their acculturation. § 1.1. Introductionlt Over the past few decades archaeologists have discovered an increasing number of sites of the great Oxus Civilization, perhaps better known nowadays as the Bactria Margiana Archaeological Complex (BMAC), as well as its Neolithic and Chalcolithic predecessors. 1 While they have filled in a large gap between the great civilizations of Mesopotamia and the Indus, so far no written documents have been found, with the exception of the seal from"Anau reported by F. Hiebert.2 However, little to nothing is known about the language(s) spoken in the areas east of Mesopotamia (Hurrite, ~ This study is a detailed follow-up on earlier " notes (Witzel 1995: 103, 1997b: xx-xxiv), lectures = (Erlangen, Indogermanische Gesellschaft, Oct. 1997 Witze12000c; Philadelphia, Contact and Exchange in the Ancient World, May 2001 - 3rd Harvard Round Table on the Ethnogenesis of Central and South Asia, May 2001, preprint: http://wwwlas.harvard.edu/-sanskrit/images/C._ASIA_.pdf.Leiden.Third = IntI. Vedic Workshop 2002 forthc. b), and investigations (Witzel 1999a: 58-60, 1999b, 1999c: 388-393; Witze12000a, 2000c). - Special thanks are due to John Colarusso: he has suggested, just before printing, a substantial number of additions and corrections, especially from Caucasian; they are quoted below as "J. Colarusso, pers. comm." " ) Sarianidi 1992, 1998a, 1998b Dani 1992, Francfort 1989, 1990, 1994, 1998, 1999, 2001, Hiebert 1988, 1992,2001. 2 Hiebert 2002, Colarusso 2002, however, see Mair 2001. A few Elamite seals have been found in S. Turkmenistan. 1 Michael Witzel, "Linguistic Evidence for Cultural Exchange in Prehistoric Western Central Asia," Sino-Platonic Papers, 129 (December, 2003) . Akkadian, Sumerian, Elamite/, and those west of the Indus area. The language(s) of the Indus civilization also are by and large unknown, that is if we neglect the materials that can be distilled from the materials contained in the earliest texts in Indo-Aryan, the Vedas,4 but which have unfortunately been overlooked for that purpose. Nevertheless, these serve as a guide of what language(s) may have been present in the subcontinent in c. 2000 BCE.5 " The picture can be enlarged and projected back in time by using the oldest Iranian counterparts of the Vedas, the Avestan texts of the Zoroastrians, as well as the Old Persian inscriptions. Since Old Iranian and Old Indo-Aryan (Vedic) are so closely related, items common to both languages can be used to reconstruct the common Proto-language, Indo-Iranian. Otherwise, we have virtually no evidence for the areas between the great civilizations and those north of Greater Iran as they are too distant from the Near Eastern, Indian, and Chinese cultures to have been discussed or described in details in 6 their texts. However, the seal recently discovered at Anau should alert us to the possibility that early writing might be found in the area after all.7 In the meantime all that we can establish for the languages used in the western Central Asian area comes from early Near Eastern and Indian (and also Old Iranian) sources. There are some references in the Sumerian and Akkadian documents of the 3rd to 1st mill. BCE, but they deal just with the border areas of Mesopotamia8 and furnish only some vague references such as that to Aratta, probably Arachosia.9 Similarly, we have only a few vague reminiscences in the earliest Indian texts (~gveda) composed in the Greater Panjab (c. 1200 BeE-lOOO BCE) 10 which seem to refer back to the ar~a along the Volga (Rasa) and secondly, to the people along the River Sindes (Tacitus' name for the Merw or Tedzhen river): the Dasa or O.P. Daha (whom the Greeks called Da{h]ai), the Arii, 3 Languages known from barely more than the names given to their speakers in Mesopotamian sources, such as those of the Guti and Lullubi at c. 2250 BCE, are neglected here. For the contemporary situation on the Iranian plateau, see Vallat 1980, 1985, 1993, Steinkeller 1982, 1989, Blazek 1999. For (possible) connections between Blamite and Vedic names see BlaZek 2002. 4 For such (loan) words see Witzel 1999 a,b. 5 For a discussion see Witzel 1999 a, b, 2001b, and forthc. a. 6 For some such data see, however, §2, where the linguistic boundaries ofW. Central Asia are discussed. 7 See, however, V. Mair 2001; yet note Proto-Elamite seals close by, at Tepe Hissar, as well as at Shahdad, Shahr-i Sokhta, etc., and recent finds to the west of Tehran at Tepe Uzbeki. 8 Such as the Guti and the Lullubi; similarly, the texts of the Hittite and Urartu realms for the boundary areas of Anatolia, NW Iran. 9 Lapis lazuli is found in the nearby Chagai Hills (just south of Arachosia/Aratta) and in Badakhshan. Note Steinkeller 1982: 250 with details about a green variety, "carnelian with green spots," possibly turquoise, from Marh-aiL However, the blue Badakhshan variety is more famous, until today. For Aratta see Steinkeller 1982, Vasil'kov and Gurov 1995, Gamkrelidze and Ivanov 1989: 36, Witzel 1995: 320-2, BlaZek 2002b: 215-218. 10 The lowest date depends on the date of iron, c. 1000 BCB; see Possehl and Gullapalli 1999. For present purposes, "Greater Panjab" indicates the area from Gandhara (Peshawar) and Swat in the west to Delhi and the Upper Doab in the east, from the lower PamirI Himalayan ranges in the north to the borders of Sindh and the Bolan in the south; however, the clear center of the ~gvedic area is western and eastern Panjab/Haryana. 2 Michael Witzel, "Linguistic Evidence for Cultural Exchange in Prehistoric Western Central Asia,lI Sino-Platonic Papers, 129 (December, 2003) and the *Parna eYed. Pa1J.i, cf. Ptolemy, Geogr. 6.10.2 Parnoi, Daai/Parn~ Dacae; otherwise Dahae). Pinault (2003) connects PatJi/*Parna, as loan word from the west, with Common Toch. *paniya 'that which belongs to wealthy people" > Toch. B peniyo, A paf# flsplendor" and takes the Gr. form Parn-oi as reflecting a local variant of Ved. Pat).i with "intrusive" -r- (cf. Kuiper 1991:70-81), however see below §5. {BlaZek (2002: 219-226) compares Vedic dasyu with Elamite tassu-p "people", *taSsu *"man"; note Romani das flnon-Gipsy" < RV dasa). In addition, after the sparse attestation found in the old Iranian and the much later Middle Iranian sourcesll most of these areas became Turkish speaking after about 1400 years ago; this has obliterated much if not most of the older Iranian and IIr. • 12 record, frequently even that of topographical names. Even in this unfortunate situation, we can retrieve, based on the records of neighboring Indo-Iranian peoples and on old loan words, an increasing amount of details of the pre-Ur./Iranian languages of the area, notably that of the BMAC (c. 2400-1600 BCE)13 and of Greater Afghanistan. However, it is precisely these Indo Iranian sources that have largely been neglected so far. 14 For some years (1995-2002) I have drawn attention, mostly in brief and passing fashion, to a common body of words in Old Indian and Old Iranian texts that do not seem to be of Proto-Indo-Iranian (thus, Proto-Indo-European) origin. These words represent the non-lIr. languages spoken in Iran and in the northwestern part of the Indian subcontinent at the time these texts were composed, that is late in the second and early on in the first millennium BCE. As such, they are invaluable materials for the study of the language(s) preceding the introduction of Indo-Aryan (Vedic) and Old Iranian (O.Persian, Avestan). More importantly, both hieratic texts share a common substratum that can only be that of S. Central Asia. As will be seen below, it cannot come from elsewhere as both Vedic and Old Iranian individually imported it into their particular habitat, the Greater Panjab and IranI Afghanistan. Such substrate words are quite common in. languages that have occupied the territory of an earlier people speaking a different language.15 In English, for example, such common words as sheep (Dutch schaap, German Schaf) belong to the Neolithic sub stratum1 6 of the North Sea coast of Northern Germany and Denmark, the homeland of Anglo-Saxon. 11 A few texts in Bactr~an, Khorezmian, and Parthian as well as (frequently mythical) data in Pahlavi, and in the Graeco-Roman sources. 12 We may note the proliferation of Central Asian place names ending in Turk. -su, -kul, -kum, etc. See the paper by P. Golden in Mair (forthc.) for information concerning the rise and the spread of the Turks. Needless to say we do not have adequate etymological dictionaries of Turkmen, Uzbek, or even for the Ir. languages (with the exception of one for older Turkic by G. Clauson, the dated one ofP. Hom for Persian, and G. Morgenstieme for Pashto, H. Bailey for Khotanese Saka). 13 Based on new carbon dates, see Francfort and Kuz'mina 1998: 468; 2400-1500 BCE (post-urban: 1800- 1500 BeE) in Francfort 2001: 152. "Greater Afghanistanll signifies the territory covered by this country , and some adjacent surrounding areas. 14 For initial suggestions see Witzel 1995, 1999a,b; see below n. 158, 195,204. 15 Exceptions are the territories of Australia, Polynesia, and the Americas when first settled. 16 ef. Huld 1990, Polome 1986, 1990, Vennemann 1994, 1998. 3 Michael Witzel. "Linguistic Evidence for Cultural Exchange in Prehistoric Western Central Asia," Sino-Platonic Papers, 129 (December, 2003) § 1.2. Sources In order to evaluate the scarce materials at our disposal properly, a brief look at our sources is in order.17 The Vedas were composed (roughly, between 1500-500 BCE) in parts of present day Afghanistan, northern Pakistan, and northern India. The oldest text at our disposal is the ~gveda (RV); its is composed in archaic Indo-Aryan (Vedic • Sanskrit). It is followed by a number of other Vedic texts, usually listed as Sruphitas, BrahmaJ;,las, .Ar~yakas, and Upani~ads. Linguistically, however, we have to distinguish five distinct levels: ~gveda) other Sarphitas (Mantra language), Yajurveda Sazphita prose, Brahmat:las (incl . .A.rat:lyakas and Upani~ads), and the late Vedic Sntras (Witzel 1987, 1997a) 18 The language of the RV is an archaic form of Indo-European. Its 1,028 hymns are addressed to the gods and most of them are used in ritual. They were orally composed and strictly preserved by exact repetition through rote learning, until today. It must be underlined that the Vedic texts are "tape recordingsl119 of this archaic period. Not one word, not a syllable, not even a tonal accent were allowed to be 20 changed. The oral texts are therefore better than any manuscript, and as good as any well-preserved contemporary inscription. We can therefore rely on the Vedic texts as contemporary sources for names of persons, places, and rivers (Witzel 1999c), and for loan wordl1 from contemporary locallanguages.22 23 The ~gveda was composed in the Greater Panjab and is to be dated before the introduction of iron in the northwestern subcontinent around 1000 BCE (Possehl and Gullapalli 1999). Later texts cover all of northern India up to Bengal and southwards towards the Vindhya hills. Some 40/0 of the words in the ~gvedic hymns that are composed in an archaic, poetic, hieratic form of Vedic, clearly are of non-IE, non-Indo-Aryan origin. In other words, they stem from pre-IA substrate(s).24 17 For the sparse Mesopotamian sources, see below (Steinkeller, Vallat). 18 For abbreviations of the names of texts see attached list. 19 The middle/late Vedic redaction of the texts has influenced only a very small, well-known number of cases, such as the development Cuv > Cv. 20 They even preserve very special cases of sentence intonation, see Klein 1997, Witzel2001a. 21 Summary and discussion for RV words by Kuiper 1991; for post-RV texts, see Witzel 1999a,b. 22 The Vedas are followed by the ancient Tamil"Sangamll (Cankam) texts from the beginning of our era, all virtually unexplored for substrates and adstrates. On the Iranian side, there are sources such as the Pahlavi and early New Persian texts (Sah Nameh, etc.), all beyond the scope of the present paper. For place names, see Eilers 1982, 1987, Savina 1964, Schmitt 1995. Such investigations, however, are largely lacking for Afghanistan (note, however, Gryunberg 1980, Pakhalina 1976, Rozenfel'd 1953 for the northeast). For the toponymy of present day Iran, see the useful web site at Tokyo Gaikokugo Daigaku: http://www.aa.tufs.ac.jp/-kkami/AbadlranE.html. 23 See Witzel 19 97a, 200la: roughly, from Eastern Afghanistan, Gandhara, Panjab up to Delhi and even up to the Ganges (twice mentioned); and fro~ the Pamirs/Himalayas southwards to the Bolan area. 24 See Kuiper 1991, Witzel 1999a,b. This situation is remarkable: if one were to apply it to a Near Eastern context, it would mean that an ancient Jerusalem temple ritual might contain Philistine, Lebanese, Akkadian, Egyptian, or other IIheathen" words. The Indian situation also differs remarkably from that of the Hittite empire, where the preceding non-IE language, Hattie, was actually used as the ritual language. . . 4 Michael Witzel, r'Linguistic Evidence for Cultural Exchange in Prehistoric Western Central Asia," Sino-Platonic Papers, 129 (December, 2003) The situation is similar but not quite as beneficial as far as the Old Iranian texts are concerned. Only about a quarter of the original Avesta has been preserved. The oldest parts are Zarathustra's RV -like poems, his 5 long Ga:6a.s (Yasna 28-53), and his(?) contemporaneous ritual text embedded among the Gai}a.s, the Yasna Hapta1)haiti, a collection of Mantras used for fire worship. The rest of the Avestan texts is post-Zoroastrian and composed in Young Avestan language. However, the initial oral o tradition of the Avesta has been converted in Sasanide times (c. 400 CE) into a written tradition whose surviving earliest manuscripts are not older than a thousand years and 25 have been corrupted by centuries of decline during the early Islamic period of Iran. Nevertheless, the philologically restored Avestan texts offer some data from Greater Afghanistan as Zarathustra's homeland was probably situated in northwestern Afghanistan (near the Kashaf River)26 and much of the later Avesta was composed or redacted in southern Afghanistan (Sistan, Arachosia). However, in spite of being geographically closer to the Mesopotamian cultures with datable historical information, the Avestan texts are even less amenable to absolute dating than the Vedic ones. Mesopotamia (or early China) simply do not figure in all these texts. The older Avestan texts (Gai}as/Yasna HaptalJha.iti) point to a copper/bronze age culture quite similar to that of the RV. The younger texts might overlap with the expansion eastwards of the Median realm (c. 700-550 BCE).27 The few Old Persian inscriptions that have survived date from 519 BCE onwards. However, other than is the case with old Indian texts,28 the "foreign" words in the Old Iranian texts have not been evaluated so far. Researchers apparently were of the opinion that only a few could be found; the matter simply has been neglected (see n. 14, 158, 195, 204). ! § 1.3. Loan words and substrate languages At this stage, a few words about linguistic substrates are in order. "Words from substrate languages" are defined here as all those words in early Vedic and O.Ir. that do not conform to Indo-European/Indo-Iranian word structure (including sounds, root structure and word formation) and have no clear IE/Ilr. etymology.29 We have to distinguish various types of loans (Anttila 1989: 154 sqq). Some are due to cultural and economic contacts, such as the modern guru or karma (from India), or the slightly older coffee (from Arabia), cocoa, chocolate (from Meso America), or tea (French the, etc.) whose.origin can be traced to S. Chinese (Arnoy t'e), 2S Modem recitation depends on these written texts and cannot be used in the same way as Vedic recitation. 26 Humbach et al., 1991. 27 Discussion by Skjcerv~ 1995. However, the YAvest. local name of Bactria (BaxbJ) is attested earlier, in the Atharvaveda, see Witzel 1980. Current estimates for Zoroaster range from the 14th to the 7th c. BeE. However, an early date is indicated by the name of Ahuramazda: O.Avest. mazda ahura (or ahura mazda), Y.Avest. ahura mazda, and in Old Persian (519 BCE) already one word, A[hluramazda. For the transfer of Zoroastrianism into Persis (the modern province of Fars, i.e. southwestern Iran) see K. Hoffmann 1992. 28 Note the ongoing debate, since the mid-19th century, especially S. Levy, Przyluski, Kuiper, and the relevant summaries in Mayrhofer, KEWA and EWA; last update in Witzel 1999a,b. 29 Lubotsky (2001) adds also some less indicative features: limited geographical distribution, specific semantics, i.e. a category which is particularly liable to borrowing. 5 Michael Witzel, "Linguistic Evidence for Cultural Exchange in Prehistoric Western Central Asia," Sino-Platonic Papers, 129 (December, 2003) while the Russian, Indian and Japanese chai/cha, Nep. chiya are from a N. Chin. dialect. The Indian word has thus come overland and not by sea. The example would also be instructive if we did not know the history of transmission: linguists would be able to pinpoint the, origin of the loan in two areas of E.Asia. Similar examples will be found below tor Central Asian words. This kind of introduction of loan words is from an "adjoining" language, an adstrate. Examples abound in multi-lingual societies (India) or of societies in close contact (ancient and modern W. Europe, with cases such as street < Latin (via) strata, Kaiser < Caesar, castle < castellum, cellar < cellarium, cella; etc.). Loans stemming from previously existing languages, upon introduction of a new, dominant language, are different (Anttila 1989: 171 sq.). The new language may function as superstrate, properly used and understood only by a minority at first (such as Latin in Celtic France), but it then spreads by assimilating an often large number of local words from the previous language, the substrate (note the Celtic place names in England, below). Sometimes the superstrate does not become dominant (as Norman French failed to do in England); in this case we may still expect a large number of words from the superstrate in the persisting local language (French beaute > beauty, ancetre ,.., ancestor, where the English form reveals the older French one, with -st-). Even if the source of the loan remains unknown, many loan words from "foreign" (substrate/adstrate) languages can be easily detected by linguistic means, and even if they belong to a long disappeared language. The reason is that all languages follow certain patterns, allowing only certain sounds or groups of sounds while others that are difficult to pronounce must be substituted by local ones. A typical example from English is that, until fairly recently, German and Yiddish words beginning with the sound sh- (schnitzel, strudel, to shlep) would have been impossible as English allowed only s-, as in snit, strut, slip). By now, these sounds have been accepted and are pronounced correctly. Similarly, even today words beginning in ng-, mf- etc. are not allowed (though by now a few African names have been locally adopted, such as Mfume).30 Words with such uncharacteristic sounds or'sound clusters therefore indicate a certain cultural influence, even if the native speaker (or a latter day scholar) may not know where these words had come from originally. This is especially true when we have to deal with toponyms and hydronyms that have come down to us from prehistory. It is well known that place names, especially names of (larger) rivers, are very conservative. Even today they may reflect languages spoken many thousands of years ago. For example, we have the Rhine (Lat. loan word Rhenus < Celtic *Renos < IE *reinos), Danube (Lat. Danubius - N. Iran. Dan-), Don, Gr. Tanais (from pre-W. Circassian t(ana/t'ane "Don", J. Colarusso, pers. comm.), Tigris (Latin, Greek < O.P. Tigra, cf. o.P. tigra "quick"; Arab. [Nahr all Dijlat, both < Akkad. (I)di-iq-Iat / Sumerian Idigna, all from a pre-Sumer. substrate!), Euphrates (cf. Arab. (Nahr all Furat) which has been taken over from Greek < C.P. [h}Ufratu (close to [hlu-fratar good brother)", Sum. Buranuna I Akkad. Purattum / II Elam. u-ip-ra-tu-iS, all from a pre-Sum. substrate more than 5,000 years ago. 30 Not all loans are as easily discernible as the Amerindian loan words tipi, squaw, papoose, Manitou, etc.; note however, the more difficult words moose < moosu, chipmunk < sitomu, or woodchuck (Marmota monax) from Algonkian otchek, ochig, odjik "fisher, weasel"; nevertheless, the English folk etymology gives the word away (Witzel 1997b). 6 Michael Witzel, "Linguistic Evidence for Cultural Exchange in Prehistoric Western Central Asia," Sino-Platonic Papers, 129 (December, 2003) The early river names of most of Europe belong to one and the same old 31 system. A different prehistoric system is found in Greece and the Aegean area, with the typical pre-Greek -s(s)-, -nt-, -mn- suffixes. This phenomenon has been extensively discussed for much of the 20th century.32 The detailed investigation of both regions mentioned just now can serve as a guide -- and as warning post -- for the following deliberations . • It is important to keep in mind that names taken from a previous language (or from an ads trate ) have more often than not lost their original meaning. If the source language is little known or unknown we can only analyze and compare the outward form of the names involved. This includes the sound system as well as typical suffIXes and prefIXes that frequently indicate the type of name, such as "rive.r, place, mountain, plain" or that describe the item in question, such as "quick/slow, white/black" (river), "high/low" (place). However, these names have often been adjusted or re-interpreted by later languages, frequently by popular etymology (see above, Tigris as tigra "quick"). As may be seen in the discussion of the pre-Hellenic and Old European place/river names, these conditions may lead to many pitfalls. Some may appear in this exploratory paper as well. The particular situation of Central Asia may be approached by a comparison with that of place names in England. We know that the early form of English, an Old Saxon dialect (a part of the Germanic branch of IE) has overlaid, in the middle of the first mill. CE, the Celtic (and Latin) languages of Britain. Both Celtic and Latin have left a number of loan words in Old English as substrate words, such as London < Celtic Lugodunum "town of the god Lug," -chester < Latin -castrum "fortified settlementll • Later on, English saw the superimposed (superstrate) influences of the Viking language (N. Germanic, with words such as egg, they, she, he, place names in -vik, -ay), then of Norman French with a large number of loans (beauty, ancestor, -ville, etc.), and fmally an equally huge amount of learned, newly formed Graeco-Latin words, as well as various minor adstrate influences from the neighboring languages such as Dutch (words such as dike, boss, mate, etc.). Most interestingly for our purpose, Old Saxon and Germanic in general can be shown to have a large percentage of non-IE substrate words (such as sheep, eel, roe, boar, lentil, land, delve, prick) derived from a 33 long-lost prehistoric Northern European language. The situation in the Greater Panjab (the area of the earliest Vedic texts) and in Greater Iran (the area of the Avestan and O.P. texts) is quite similar. A brief, simplified summary would look like this. 31 Explained, since H. Krahe, as an "old European" layer of IE (summary by W. P. Schmid 1995); this layer of river names has several elements that seem to differ from, and to predate PIE; note also that many Germanic words or names in the North Sea/Baltic area belong to a pre-IE substrate, see Polome 1990, Huld 1990. 32 Summary by Th. Lindner 1995. 33 Cf . Hamp 1998: 328, Huld 1990, Vennemann 2001. 7 Michael Witzel, ItLinguistic Evidence for Cultural Exchange in Prehistoric Western Central Asia," Sino-Platonic Papers, 129 (December, 2003) -------------------------------------------------------------------------------------- GREATER PA NJAB GREATER IRAN -------------------------------------------------------------------------------------- English loan words Urdu/Perso-Arabic superstr. loans influx of learned (Sanskrit) words (development to New Indo-Aryan) (development to modern Iranian) influx of learned (Sanskrit) words Arabic superstrate loans (dev. of various Pralqts) (dev. to Middle Iranian) Old Greek loan words some Old Greek loans Old Persian/Iranian loan words . Old Persian /Later Avestan (development from Vedic to MIA) Later (~g)V edic lOlA dialects Old Avestan Dravidian ads trate Old Iranian superstrate <----immigrant Old Indo-Aryan <-----immigrant OIA in Iran *SarasvatI, Sarayu, etc.) Central Asian substrate Harappan language (see below) Indo-Iranian in C. Asia, south of Urallc, • Ket (Yen.) unknown local language ( s) <------Indo-European § 2. Triangulating the Central Asian Area As has been indicated earlier, we know even less of C. Asia than about the substrate situation of Iran and Northern India since we do not have any old Central Asian written or other traditional records, such as the orally transmitted Avestan texts. In addition, in most of these areas, people have been speaking Turkic languages for the past 1000-1500 years, which has obliterated much of the older Iranian, I1r. and pre IIr. local record. Yet, even there we can make out, based on the written records of neighboring peoples and on old loan words, some details of the pre-IIr./lranian languages of the area, notably of the BMAC (Oxus civilization) region. As western Central Asia and the lands south of it were later on occupied by speakers of the various Indo-Iranian languages such as Saka, Avestan, Median, Old Persian, Nuristani, Vedic, etc., many of which have left us texts, it is best to begin with this language family. The original speakers of Proto-Indo-Iranian (PlIr., sometimes also called Aryan) have been located in various areas, such as the southern Urals and northern Kazakhstan, the Ukraine and the Caucasus area, or in recent Indian revisionist writing, even in Northern India. However the combined data of the reconstructed PIlr. language allow us to pinpoint the general area where the still united Indo-Iranian proto-language was 8