For some mysterious reason, I find myself penetrating a little bit deeper into the far part of Germany, the erstwhile German Democratic Republic, every time I go there. This time I took the train to Dresden, past Eisenach, Erfurt and Weimar which I already visited the latest times around, but got off before the end station, at the city of Leipzig. It is here that Napoleon was defeated in the Nations Battle exactly 200 years ago.
On 2-3 December 2013, I participated in the interdisciplinary workshop on Indo-European, “Talking Neolithic”, which brought together geneticists, archaeologists and linguists. It was surprisingly well-organized and the papers were original (sometimes revolutionary) and at a high level. Socializing was also fun, both meeting new people and finally getting to see some famous people I only knew from their publications, like James Mallory, whose dictionary of Indo-European I often use though he turns out to be an archaeologist. I had hoped to meet Joanna Nichols, who had at first been announced, if only to hear out her erstwhile Bactrian homeland thesis, but it seems health problems kept her back. Nonetheless, I had a very good time.
I haven’t taken notes and will not do a full report, but a few papers have made such an impression that I have to mention them here. It goes without saying that the responsibility for any misrepresentation of the speakers’ ideas, no matter how involuntary, will be mine. My apologies to the speakers who are left undiscussed here.
State of the art
Co-organizer Guus Kroonen, a Dutch linguist teaching in Copenhagen, introduced the workshop by summarizing the state of the art. Ever new data or deeper knowledge about existing data confirm that the Indo-Europeans in Europe borrowed a lot of their agricultural and botanical terminology from lost indigenous languages. This contrasts with the terminology from pastoralism and wheeled locomotion, which is fully Proto-Indo-European. Therefore, many scholars have suspected that the Indo-Europeans in their steppe homeland knew no agriculture at all. As pastoralists, they established a hierarchical relation with the native European agriculturalists, so that for reasons of prestige, the common folk abandoned their mother tongue and adopted the Indo-European language of the elite, transforming it along the way into the daughter languages known from history. However, Kroonen’s own work finds that the Indo-Europeans did know at least some agriculture, though they adopted new cultivars and techniques from the Old Europeans.
Nonetheless, he also mentioned the consequential fact that east of the Dniepr in Ukraine, and in the vast stretch of land where the home of the Proto-Indo-Europeans is usually situated, no trace of agriculture has been found prior to 2000 BCE. I deduce that this area may fit the putative homeland of PIE pastoralists, who did not know any agriculture, but cannot possibly fit a PIE population of whom it has been established that it knew at least some agriculture. Whether this is really the case, this workshop was meant to verify.
The next speaker, James Mallory from Belfast, gave a historical overview of how the agricultural question had been treated in two centuries of IE studies. Very entertaining, and what I have taken away from it, is that around 1900, the homeland was often situated in Bactria as the northern rim of the original wheat cultivation zone, stretching from the Euphrates to the Indus valleys. This makes sense: Bactria, like Northwestern India and Mesopotamia, was an area propitious to agriculture and demographic growth. If in such an area a famine or a conflict forces a part of the population to leave, it may be a small percentage in its homeland, but a mighty big group in the endless steppelands, capable of imposing its language. The unparalleled success of the Indo-Europeans in spreading or imposing their language is far better explained by postulating a thickly-populated homeland rather than the bare and backward steppes.
He confirmed the surprising lack of names for cultivars common to Indo-Iranian and the European branches. While the currently most favoured explanation is that different groups of pastoralists from Russia separately entered Europe and India, it was thought then that the absence of agricultural terms in Indo-Iranian could be explained by a division of labour: the European-to-be tribes lived in the Oxus valley proper tilling the fields and took their agricultural terms with them from there, while the eastern-oriented Indo-Iranians trekked through the Hindu Kush and Pamir hill slopes with their herds.
My own paper
The choice to introduce my own paper at this point, though in reality it was scheduled only on the second day, is not inspired by a lack of modesty. I simply need to clarify where I stand on this matter, so that the reader can see how I had to interpret the other papers.
Incidentally, organizational and technical gave me only 25 minutes to read my paper, so I hurried through it and had no time left for questions, which I had kind of counted on. While I had a fairly original message on the homeland question, I found that during the subsequent lunch and over drinks, only a few people had any questions about it, let alone objections. This was mostly not a matter of politely hiding their scepticism (as Germans are wont to do, but the many Dutch and Anglophone participants are more forthright), but of a profound disinterest in the matter. Either they have interiorized the SW-Russian homeland theory during their schooling and simply cannot imagine Indo-European history otherwise (to the extent that quite a few have not even heard of an Indian challenge), and consequently are deaf to or immune against any alternative; or they have vaguely heard that this is a wild conspiracy theory promoted only by Hindu nationalists, which can safely be ignored.
In my paper, I argued that agricultural terminology in Germanic are usually taken to confirm the established picture. Words corresponding with Indo-Iranian terms, hence fully Indo-European, have an agricultural meaning in Germanic, but not in Indo-Iranian (Dutch tarwe/”wheat” but Sanskrit durva, “grass”, apparently the original meaning; or harvest but kṛpāṇi, “sword”, both from a root meaning “cut” but only in Europe specialized to an agricultural setting). The usual reading of this is that in their Russian homeland, the Proto-Indo-Europeans were pastoralists without agriculture, that the Indo-Iranians took this level of culture along with them to Afghanistan and India, while the European branches showed a quick immersion in an agricultural society, attested either by the specialization of Indo-European terms to agricultural meanings, or by borrowing agricultural terms from the Old European natives.
I then revisited a paper from 1979 by Colin Masica, as well as his 1991 standard handbook of Indo-Aryan linguistics, to evaluate his list of Indo-Aryan agricultural, botanical and pastoralist terms. Of these, over 9 % are deduced from Dravidian, mostly because corresponding Dravidian words are attested, sometimes just because they look Dravidian; likewise, over 5% of Munda. Some 24% are from Persian or otherwise foreign, and need not detain us here. Various Indo-Aryan or even Indo-Iranian groups amounts to some 44%, and the rest is unexplained. In fact, quite a few of the other categories are not securely explained either. Now, many of these words look perfectly Indo-Aryan, either because Hindi-speaking mouths have successfully integrated them, or because they simply are. The criterion that a word can only be Indo-European if it occurs in at least two branches, is conventionally assumed but in fact arbitrary. An Indian homeland would almost necessitate that numerous terms for Indian plants were forgotten along with the plants themselves by all non-Indian branches as people left India and never encountered, say, a banyan tree again.
According to Michael Witzel, some 4% of Rg-Vedic terminology is non-Indo-European. That would make the Rg-Veda the purest Indo-European text by far. And even that 4 % is a generous estimate, for the words he lists as borrowed have a disputed origin and some may just be Indo-European. Witzel lists bīja, “seed”, which exists also in Iranian, as borrowed from the purely hypothetical “Bactria-Margiana Archaeological Complex” language, but Masica opines that it is from Dravidian. As for parṣa, “sheaf”, I learned from another speaker that it is one of the few agricultural terms attested both in India and in Europe (Greek), viz. as part of the name of grain goddess Demeter’s daughter Persephone < perso-gwhen- = Skt. parṣa-han-, “sheaf-beater”, i.e. thresher. In Germanic, the number of words deemed loans has been halved in recent decades by identifying cognates in other branches, e.g. swim turns out to have cognates in Baltic and Celtic, though meaning “move” or “hunt”; it is not impossible that such cognates will yet be found for Sankrit too.
But not that there is anything against loans into Sanskrit or Indo-Aryan. Munda was the language of a rice-cultivating society around the mouth of the Ganga, Dravidian that of the Gujarat part of the Harappan area, dimly bordering on the Sankrit speech area (separated by the Rajasthan desert), which moreover was to assimilate many Munda- and Dravidian areas into the expanding Indo-Aryan community. Moreover, South-Indian kings invited Brahmin communities to settle and give Vedic prestige to their dynasties, upon which they integrated Dravidian words into later Sanskrit, e.g. mīṇa, “fish”, or kāṇa, one-eyed. This expansion is attested by well-known historical developments and doesn’t require, let alone prove, an Indo-Aryan invasion from abroad. As for the European branches, they borrowed words but often also gave a new agricultural meaning to general IE words, e.g. Skt. ajra, “empty land” but Latin ager, “field”. But this doesn’t mean that Sanskrit had no agricultural terms, e.g. sītā, “furrow” (from PIE seh, “sow”), kṣetra, “field”.
Masica considers his findings as incompatible with the Indian tradition of an Indian homeland. We think we have shown otherwise. Of the other arguments, he knew a few but more have been added since: (1) linguistic paleontology: agreed here not to prove a cold homeland or any homeland at all; (2) common developments with Greek, and generally the geographical distribution of the isoglosses, is better explained by an Indian “extreme” homeland than by radiation from the Russian centre; (3) Finno-Ugric has hundreds of Iranian loanwords but imparted no words to Iranian or Indo-Aryan (the seeming exception of Guṅgu, “moon”, can either be a coincidental homonymy, date from an earlier Nostratic period, or was somehow a loan from Indo-Iranian), which is typical for a colonial situation, with Scythian Iranian imparting words and also borrowing some but not communicating them back to the homeland; (4) Mitannic can be shown to belong to the youngest layer of the Ṛg-Veda, so allowing for the language to emigrate and to become a dead substrate of Hurrian (and the similar case of Kassite), the Ṛg-Veda must have been complete by 1800 BC or so; 5) geographical distribution has the homeland typically in a far corner (Amerind superfamily, Bantu/Austronesian family, Turkic group, Arabic/Russian language), not in the centre (IE: Volga), which also is not the zone of greatest diversity, on the contrary; (6) Vedic literature contains a few astronomical passages datable because of the precession (ca. 1° in 71 years), e.g. Kauṣītaki Brāhmaṇa 2300 BC instead of ca. 1000 BC, Vedāṅga Jyotiṣa 1300 BC instead of ca. 400 BC, incompatible with an invasion scenario ca. 1500 BC; (7) kentum substrate in Bangani, India, e.g. dokru instead of expected daśru, “tear”; (8) Vedic and Puranic literature refers several times to emigrations, never to early immigrations, and the Northwest was not venerated as an area of origin; (9) the agricultural terminology proves, contra Masica, to be compatible with an Indian homeland.
Hans-Jürgen Bartelt, teaching in Hamburg, analyzed the way the label “interdisciplinary” is misused by scholars, so that linguists hide behind findings supposedly done by archaeologists or vice-versa. Genetics is another mute science with a tremendous promise but hitherto of limited or largely false use in linguistic reconstruction.
Ranko Matasović from Zagreb, Croatia, asked if there are loans in Indo-European from North-Causasian, and after surveying some candidates, he concluded that there weren’t any. If the Proto-Indo-Europeans had been their northern neighbours in their Russian homeland, they should reasonably have taken some loans with them to their respective historical habitats. I conclude that, like with Uralic, the homeland had no point of contact with North Causasian, and therefore was not in the peri-Caspian steppe zone.
Paul Heggarty from Cambridge analysed linguistic paleontology to extinction. The basic supposition of linguistic paleontology is that from the vocabulary, you can deduce all about a society and its environment. But words can undergo semantic shifts, so that a general word at an unidentifiable moment gets specialized towards a specific technological meaning. As for glottochronology, the calculation of the time at which a certain rate of linguistic change has taken place, it is an outrageous failure. This is shown by the rate of change calculated for languages known from their beginnings onwards, e.g. Portuguese and Italian are calculated to have separated less than 1000 years ago. This was a plea against the old and hazy methods and in favour of modern precision. Applying Bayesian phylogenetics, on the basis of linguistic facts furnished by the linguists themselves, gets better results. Between the Russian and the Anatolian homelands (an Indian homeland is not taken seriously by the researchers, or not even known), it judges strongly in favour of the Anatolian one. Nevertheless, most people present thought, on other linguistic grounds, that Anatolia was definitely not the homeland, among other reasons because the Hittite language spoken there has all the marks of a result of foreign conquest, with a very large substratal element both in the language and in the mythology. Note also that with the new fancy methods, the split between Indian and Iranian is estimated at 2800 BC, a more realistic date than the 1600 BC or so required by the Aryan Invasion Theory.
Robert Mailhammer from Sydney discussed six words for wheeled locomotion, which are as consistently present in all Indo-European branches as the words for relatives or the numbers. We give a Germanic example plus the Sanskrit: wagon/vahati (cart, conduct); rad/ratha (wheel, cart); wheel/cakra; ahs/akṣa (axle); ar/īṣā (pole, rudder); yoke/yuga. It seems that Indo-European had a full vocabulary for wheeled locomotion because this technology had just been developed in the Proto-Indo-European community. Indeed, I deduce that wheeled technology was part of the reason for the successful expansion of Indo-European. This technology gave them as edge over others, and by the time their neighbours had acquired it, the Indo-European language family was already spread from India to Ireland. Colin Renfrew may have been wrong with his Anatolian homeland theory, but he has the merit of explaining the Indo-European expansion: the shift to agriculture gave the population concerned a great demographic edge, and as the people spread, their language spread with them. If we deem Renfrew wrong, we have at least to think of another reason for this success.
Question time after Mailhammer’s lecture became a debate with Heggarty, who, for my taste, went a bit too far in his doubt that a vocabulary can teach us anything at all. Thus, the existence of words for “wheel”, “yoke” and “axle” in Proto-Indo-European does not strictly prove that the Indo-Europeans had wheeled technology, e.g. the words for “wheel” may simply have meant “turn, revolve”, a type of motion that also exists in the natural world. But in that case, it is unlikely that all branches of later Indo-European would all have adapted the same terms for natural phenomena to name the freshly invented wheel technology. Like most linguists, I believe that in this case, the evidence of the vocabulary is so strong and so consistent as to allow the inference that the cart technology was alive already in the Proto-Indo-European stage.
Joseph Salmons from Wisconsin applied the suppositions of the Indo-Europeanists to the present-day language situation in America. While American English does have words of Amerindian origin (moose, squirrel, skunk), the southern dialects do not have any words specifically from the local Cherokee language, nor the Wisconsin dialects from the local Algonquian (only the Algonquians themselves intersperse their English with some of these). Dialectal American English does have many local formations unknown to any outsiders, even in this audience consisting mostly of linguists. Some of these are variations on English expressions, other are onomatopoeic or of unknown origin. If we already don’t understand the finesses of these local variations on the language in which the workshop was conducted, what are our chances in a language situation thousands of years before its first writing? At any rate, the test of linguistic mechanisms posited for PIE (Grimm’s law, Venner’s law, etc.) is whether they could explain linguistic processes today.
I do not remember who it was that cited a couple of leading linguists as affirming that the homeland must have been a mountainous area with thunderstorms and heat, since Proto-Indo-European had words for “mountain”, “thunder” and “hot”. Well, what could be the homeland where people would never experience thunderstorms or feel hot? Yes, Antarctica might fit the bill (though it inconveniently has a few mountains), and we might all be found willing to scrap Antarctica from the list of possible homelands of Indo-European. For me, the point of this example, or of some names in Mallory’s historical survey, is that this academic discipline has thrown up quite a few people in positions of authority who in seriousness held theories that could not stand the test of common sense.
Hrach Martirosyan, an Armenian working in Leiden, showed that Armenian shared many isoglosses (common linguistic innovations, starting with common new vocabulary) with Greek. These had had to have shared most of the way from the homeland, wherever it lay, to their historical habitats in Armenia and the southeast corner of the Black sea coast, c.q. in the southern Balkans and the Aegean Sea. This pleads against a Pontic (steppe) homeland, as Greek would probably have to go west through the Balkans while Armenian would have to go south through the Caucasus. By contrast, in an Indian homeland scenario, Greek and Armenian (as well as the extinct and little-attested Phrygian) would follow the same route, thus fitting these languages’ isoglosses.
From Adam Hyllested, who spoke about the roots for “carrot”, I retained that unlike all other branches, Slavic had not been searched for substrate words because, as the language of the homeland, it had had no non-Indo-European substrate language. In fact, another explanation is possible: Slavic was preceded by other Indo-European languages who cleared the way and assimilated the substrate populations. Unlike even its Baltic twin, it was not in touch with non-Indo-European. It may have adopted substrate words, but through other Indo-European language, so by definition in common with other branches and hence classed as Indo-European so far. Note also that even in the Russian homeland theory, it was not the language of the homeland: united Slavic had Slovakia or so as its habitat, Central Europe and not the steppes, which for more than a thousand years were the realm of the Iranian-speaking Scythians.
Lucien van Beek from Leiden showed that while names of grains are either pan-Indo-European or at least common to the European branches, the names for five species of pulses are undeniably borrowed from a substrate layer. Moreover, it turns out that they are cognate in the Greek and Albanian branches, and often also in the other European branches. The borrowing from this Old Balkanic language group sometimes produced doubles, e.g. orobos and erebinthos, “pea” c.q. “chick pea”, borrowed at different times or places from different variants of the native language. The strengthening (as opposed to the denying) prefix a- and the suffix inth- are instances of features regularly appearing in Greek that must stem from this Balkanic language.
Two etymologies he stated drew my attention. Odysseus and his men are at one point locked up by a one-eyed giant, the Cyclops. Folk-etymologically, this means “(he with the) circular eye”, but the speaker quoted Paul Thieme as saying that this Ky-klops came from pku-klop-, “cattle (Latin pecus, German Vieh) thief (cfr. kleptomania, ‘compulsive stealing’)”, the image of violent and lawless raiders that the Old European farmers must have had of the pastoralist intruders. The other is Persephone, Perso-gwhen-, “she who beats the sheafs”, which makes a hole in Witzel’s theory that Skt. parṣa, “sheaf”, is a BMAC loan into Indo-Iranian.
Rosemarie Lühr from Berlin was an old-school professor who did not bring a power-point but read her whole lecture quite literally from her paper. Nonetheless, her message put the final nail in the coffin of the “Northwest block” theory. This language, also known as “Belgian”, was a hypothetical Indo-European language spoken in the low countries. Like Phrygian, Anatolian and Tocharian, it died out, i.c. around the time of the Roman conquest. However, it left traces in Germanic, mostly the hard-to-explain words starting with initial p-, such as pool and plough. She argued that within Germanic itself, mechanisms are available to explain these words without having to bring in the uneconomical hypothesis of an extra language. Other linguists agreed but thought that, as far as they were concerned, the theory had already been abandoned.
When the Neolithic people talked
Peter Schrijver, who teaches Celtic languages in Utrecht, ended the workshop with a revolutionary paper, exploring the possibility that Sumerian and Hattic, and perhaps even Minoan (linear-A), were dialects of the same language. When you compare classical Sumerian, which was studied for many centuries after its disappearance as a living language ca. 2000 BC, you will find little that resembles Hattic. But when you go to the oldest Sumerian available, many resemblances start to appear. The differences that remain, on the other hand, prove a great time-depth: the original language must have broken up already several thousands of years earlier. This chronology makes it entirely possible that theirs was the language of the Neolithic Revolution, expanding from Syria to southern Mesopotamia (Sumerian), Anatolia (Hattic) and Crete (Minoan), before Semitic and Indo-European took over.
Most participants at the workshop agreed that Colin Renfrew had been wrong with his proposal that Indo-European was spread along with agriculture from an Anatolian homeland. The two developments, agriculturization and Indo-Europeanization, had been separated by at the very least a thousand years, and the Indo-Europeans found an agriculture-based economy ready-made when they entered Europe. But the earlier spread of Hattic-related languages along with agriculture may be a defensible proposal.
We concluded with an informal drink. Everyone applauded when Paul Heggarty proposed a toast to William Jones, the founder of Indo-European linguistics. Note that Jones worked as a judge in Kolkata at the time, in the far corner of the Indo-European expansion zone, east even of a putative Indian homeland in the western Ganga basin. In India, many know of his depiction on a frieze displayed in Oxford Chapel, where he is seen taking notes while being fed information by native informers. Without the tradition of Sanskrit learning, his scholarship might not have been that rich, and he might have missed his breakthrough to the recognition of a kinship between the languages now known as Indo-European.
Unlike at most conferences I have attended, all the papers read here were interesting to most participants, and certainly to me. They came not only to rattle off their own findings, but to listen and learn. In spite of what I hear in India, where comparative-historical linguistics is lambasted as a “pseudo-science” dealing in “ghost languages”, this discipline is really making progress.
Except for myself, everyone present seemed to believe the Russian (or at any rate a non-Indian) homeland theory. Yet some of their findings strengthened the case for an Indian homeland.