Why do Europeans speak Indo-European?

Why do Europeans speak Indo-European?

Andrew Duncan
May 2012

1. The occasion of this essay or aide-memoire is my loss of belief in a theory (of IE origins) which I believed from 1987 to 2012. I found this loss upsetting. But I wanted to write about the circumstances. The theory I lost was Renfrew's idea that the spread of IE languages in Europe was part of the spread of farming, from the south-east. The total replacement of older languages was because they were not spoken by farmers but by foragers, the Mesolithic population, who did not densely populate the continent and were relatively easily diluted and washed out by the dense peasant populations as they gradually spread. So the first Indo-Europeans lived in Anatolia, in an area (perhaps the Konak plain) which took up farming, and their language family spread through Europe from the south-east as its speakers diffused, bearing farming techniques, from the 7th millennium on. The languages recorded in the early second millennium in Anatolia, Indo-European but of an aberrant type (the ‘Anatolian’ languages) would be a relic of this folk movement. This theory replaced a much older theory, in which the Indo-Europeans were not the first farmers, but invaders from the horse-riding world of the Steppes, whose languages replaced in Europe those spoken by an already dense peasant population, originating at least partly in south-west Asia. The ‘vector’ for their amazing expansion, despite small numbers, was military superiority, giving rise to ‘feudal’ societies in which the upper class spoke an Indo-European language which was gradually copied by the peasant majority. The nature of this superiority was speculated on, and consisted either of rude barbarian vices or pristine barbarian virtues. This is the Steppe Origins model.
The problem with the 'peasant-Anatolian' theory was set out by Benjamin W Fortson III in his text-book on the IE languages (Indo-European Language and Culture. An Introduction), at p. 43, where he says that the terminology for parts of a wagon is as old as the unified IE language and so the speakers of that language were still in one place up to the invention of vehicle technology and the domestication of the horse. These innovations can be dated from archaeological evidence, and this dating is incompatible with the Renfrew thesis. I read Fortson in about May 2011 and have been unable to shake off the logic of what he says. I therefore now think that the version of the Indo-European spread set out in DW Anthony's The Horse, the Wheel, and Language (2007) is much more probable. (The dating of the first horse riding has changed considerably in recent years, due to research by Anthony and associates.) This aide-memoire sets out the reasons for this.

1a. The finding of a source for the first Indo-Europeans is a free area of intellectual activity. That is, because it is almost insoluble, or at least unprovable. Language historians have long speculated on the source area for the language whose descendants we have so very many records of. 26 such essays are collected in Scherer’s 1967 anthology (Herman Scherer, ed. Die Urheimat der Indogermanen, 1967.) Notoriously, if you read one explanation of a complex event you feel satisfied, and if you read three explanations you feel dissatisfied and in need of more information. The 26 essays leave behind pervasive dissatisfaction. Written texts point us to the forms of earlier stages of the language, but not to places where those forms were, prior to writing, spoken. If we find related languages in Greece and the Punjab, we can project back to a common source, but we can't get a directional fix on that source. The radius of the 'projection lines' is already then so long that the source could be almost anywhere in Eurasia. Of course west Africa, let's say, or Ireland, are out of the frame - but have not genuinely been eliminated. There is no good reason to think that a speech area of 2000 years before the first (relevant) written record can be proved to be right even if it is right.
(Scherer includes some fringe material, e.g. by Kossinna and Neckel, but gives us the mainstream rather than the really speculative or crackpot stuff. I mean, Neckel comes across as a psychotic scholar rather than as a mere out-patient.)

2. The Steppe Origins model calls for the new language to spread on a massive scale from the incomers to the existing peasant cultures of Europe. That is, a language which was a marker of a pastoral way of life and Steppe origins in around 4000 BC came, at some later date, to be spoken by farmers on a large scale, and in areas from modern Belarus westwards as far as Ireland and Spain. Its spread over the majority of the continent could only follow it having become the language of the majority of individuals in a dense agricultural population. That is, demographic weight was the basis for its ability to spread. The spread occurred in two strokes: first as the language of horsed pastoralists, then as the language of a farming culture, although with pastoral components as well as field cultivation. We would want traces of the spread of the new culture across Europe. An archaeological complex which fits that role, whether well or badly, is the Corded Ware culture. The unity of this culture across huge parts of Europe matches the unity of the IE languages and is an adequate material entity to carry the theorised language community. This culture crosses many national borders and archaeologists do not use standard notation, so it is also referred to as Single Grave culture and Battle-Axe culture. (In German, variously, Schnurkeramik, Streitaxtkultur, Einzelgrabkultur.) This brings us to another problem, namely that archaeologists of a certain generation hate the idea of migrations and so write up evidence which is wholly compatible with migration and with the spread of new languages with their speakers in any other way, so long as words like migration, invasion, dominance, replacement do not appear. This is a sort of protest against the facts of European history - a traumatic response. Christopher Hawkes labelled this deviation 'immobilism'.

2a Renfrew has not published a second book on his 1987 theory, but the papers he has published on the subject (indo-europeanization, language reconstruction, etc.) since 1987 must amount to several hundred pages, so that the 'Renfrew Theory' is significantly different now from the classic statement then. In fact, the summary came before the detailed and specialised studies. This makes it unusually hard to refute.

3. The Corded Ware culture of roughly 2900-2350 BC was identified in significant papers by Kossinna (1902) and Tadeusz Sulimirski (1931) as the product of the first Indo-Europeans, the point from which they disseminated. Kossinna's view, overwhelmingly coloured by nationalist urges, is that Corded Ware spread from North Germany and that here was the first home of the Indo-Europeans. It follows that heroic Germani rode their horses all the way to north India and bequeathed the Vedic language to their descendants. Also, that they impetuously flooded across the Balkans to Greece and founded Greek culture.
This heritage of fantasy inspires caution, and what Anthony says about the Corded Ware culture at pp.367-8 is cautious. This horizon covers most of North-Central Europe and seems to have been a culture mainly of farmers which nonetheless has some share of individuals coming from the pastoral steppe world, and which derives a large share of its economic wealth from pastoralism. It therefore fits the desired shape of a culture which has the numerical strength to influence an entire continent, and yet spoke a language deriving from the remote east - the steppes north of the Caspian, on the Volga and Don rivers. Anthony is not assertive about this. He seems to find the origin of the Corded Ware material kit too early for it to fit the migrations he is mapping. He speaks more of eastern Corded Ware cultures speaking languages like proto-Germanic, but tacitly allows for other Corded Ware cultures speaking other languages. The cohesion of Corded Ware (or Schnurkeramik) culture offers pathways which would have facilitated the spread of an intrusive language (and some of its speakers) from an ‘entry point’ (or, ‘entry strip’) over a very wide area.
The Schnurkeramik culture starts too early to be the wave of Steppe horsemen which Anthony describes. Yet the extent of the culture matches the wide-scale homogeneity which we find for Indo-Europeans at the dawn of history, and which we (it follows) want to identify in the Europe of the 3rd and 2nd millennia BC, well before that dawn. (The cordlike decoration on the pots has been seen as an imitation of the ropes by which the pots were, frequently, suspended for carrying or storage.) Anthony speaks of a merger with an existing culture, but surely if we have a complete change of language we want to see a complete change of culture. If Corded Ware People = (Western) Indo-Europeans then the 'steppe invasion' theory is wrong. Corded Ware begins in the Middle Neolithic and is a peasant culture, the product of the primary farmers who pioneered field cultivation in a Europe of untouched fertility. If later on it acquires a much stronger pastoral component, while retaining its artefact culture, that can be thought of as a mixture of eastern incomers with the existing western pattern, but it does not sustain a change of language - from the old languages to Indo-European. It does not explain why the old languages died - and were never written down. Yet the Schnurkeramik is too large-scale, or too late, and too strong in pastoral components, not to be part of the story of Indo-European.
Corded Ware does not fit the bill, for Anthony. That is, it ‘kind of’ does fit the bill. But it is not a good match. This is a real problem with the Anthony model and it has to do with him being a historian of the Steppe world. The area where the posited Indo-European population flows into Central and Western Europe is outside his territory, and not even a feature of his outstanding book, so not one we could criticise. I think a new story needs to be told here. He is very clear that the Corded Ware group is not the ancestor of the Indo-Iranian or the Anatolian languages.
Frustrated, I drew down some information about the Corded Ware culture from Wikipedia.

>>There have been many different views concerning the origin of the Corded Ware culture. There is broadly a division between archaeologists who see an influence from pastoral societies of the steppes north of the Black Sea and those who think that Corded Ware springs from central Europe. [...] The distribution of the Corded Ware Culture coincides in part with the earlier Funnel Beaker Culture, with which it shares a number of features, such as cord impressions on pottery, and the use of horses and wheeled vehicles, that can be ultimately traced to the cultures of the European steppe<< (Wiki) http://en.wikipedia.org/wiki/Corded_Ware_culture

>>On most of the immense, continental expanse that it covered, the culture was clearly intrusive, and therefore represents one of the most impressive and revolutionary cultural changes attested by archeology. The degree to which cultural change generally represents immigration is a matter of debate[.]<<

>>In summary, Corded Ware does not represent a single monolithic entity, but rather a diffusion of technological and cultural innovations of different, contemporaneous peoples, living in close proximity to each other and leaving different archaeological remains.<<

That is, an impact zone of continental magnitude. Well, this is what we are looking for to explain the linguistic geography as eventually trapped by written records. The Corded Ware culture was linked with the Urheimat already in the 19th C, so this is not some heresy. Interestingly, the source says that the Corded Ware era replaced the Funnel Beaker (Trichterbecher) layer and that both covered huge areas with significant similarities of material culture. This suggests that the process by which a single language came to be spoken over a huge area, and the process by which Indo-European spread from the east and washed away older languages, may be quite separate. That is, there was a niche for languages which would cover such a wide area as North Germany or Northern France, and the successful candidates were Indo-European ones. The process by which the radius of a language came to be a couple of hundred miles long needs separate investigation from the geographically local source of languages such as Proto-Germanic and proto-Italic. If single languages came, during the Late Neolithic or at other points, to have very long radiuses for their area of domination, then most of the older European languages or language-groups would have died off anyway.
In this area of intellectual reflection, the suggestion that the Funnel Beaker territory did not coincide with the extent of a language, but could perhaps have included a dozen different languages, is of great interest.

4. Anthony's model definitely has the incoming cavalrymen winning armed struggles with the existing population, and after those occupying positions of lordship in which the sedentary population were in client relationships to them. This is how he interprets the new mixed cultures. He repeatedly describes a legal situation in which the farmers paid dues to ‘lords’ who were the Steppe incomers, and where political power was wielded by the lords. This correlates with burial evidence in which a very few individuals had rich burials whose equipment correlated with the material culture of the Steppe. He is keen to suggest that local people could aggregate to this status group - via a client arrangement involving oaths.
The incomers were on horses. The Peasants were not mobile because they were on foot and also tied to their fields, as their principal means of subsistence. The peasants could not easily abandon their crops and the investment of labour represented by the clearing of the ground, the buildings, fieldworks, and so on. A likely outcome was that they would remain with their fields but be subject to the distraction of the surplus, as taxation, feudal dues, etc., by an exploitative group who can accurately be described as 'lords', as feudal superiors.
This situation bears an amazing resemblance to what we learnt about in primary school as the feudal system. Indeed, the proposal is that this combination of knights with mastery of horses and of metallurgy as applied to weapons, and of peasants who produced the wealth but did not have full enjoyment of it, was prevalent in say 800 AD as a reminiscence of a social order present in the 3rd millennium BC. The technical debt of Western European feudalism to eastern sources, in the steppe world, may have been considerable and tends to be forgotten - although noted by historians such as Lynn White and Hugh Trevor-Roper.

5. The model has horsemen flooding in with their herds and occupying discrete areas in between terrain densely populated by the peasants. They went for empty areas which could be made productive by horse-based pastoralism. Their early settlements were separated from each other. This was a diffraction diffusion, not a smooth diffusion. This phase of separation gives us a context in which the oldest Indo-European language could break down into separate families. Thus, the three different language groups of Western Europe, Celtic, Italic, and Germanic, may record three discontinous areas of early IE settlement in eastern Europe, and different routes to the west. The persistence of coherent blocs of farmers allows for a formative isolation of early Western Indo-European groups.

6. The geography of the early IE languages is very distinctive and puts strong limits on the possible explanations for the data pattern. Reconstruction has shown that early IE was very coherent. This shows a point origin. The ‘point’ could be a larger area if the speakers were mobile, which would be typical of pastoralists in dry and not very productive pastureland, such as the Steppes. However, soon after this ’point’ stage the language had spread over a huge area and begun to disintegrate. In general, a society of horseback pastoralists, with the ability to undertake very long treks in search of lush or undefended pasture grounds, fits very well the puzzling geographical pattern which has related languages spoken all the way from Bengal to Ireland. The combination of single point origin, mobility, and loss of continuity or cohesion, puts strong constraints on the explanatory model. A story in which pastoral tribes overran pre-existing peasant populations, achieved power and settled down, and were then separated by clumps of the languages these peasants spoke, is the best model. This is the diffraction diffusion. When much later the IE languages mopped up the older languages, to produce a map of Indo-European languages without gaps where other languages were still spoken, they had already evolved away from each other. The process whereby each recorded language family became de-indo-europeanised is of singular importance. The lack of intermediate languages between German and Polish, for example, or Latin and Gaulish, is of great interest. It is quite possible that other languages were spoken in say 1000 BC and then died out before the arrival of literacy brought them to written record.

7. Cultivator societies tend to reach much higher densities of human per acre than pastoralists. This is why various Steppe invasions did not lead to cultural replacement in eastern Europe or Northern China. The Russians and Ukrainians did not start to speak Mongol. We should not project this back onto prehistory. Cultivators whose ploughs were drawn by oxen and had no metal parts, who had no horses or wheeled vehicles, were not going to reach high population density. The gap between them and the steppe pastoralists was not so great then.
A key difference between a simple invasion + exploitation story and Anthony's narrative is that he points to ways in which the migration from the Steppes could have improved the productive powers of the (wholly rural) economy in Europe and raised the carrying capacity of the land. Thus, the peasant cultures of before 4000 BC did not have wool. This was a new product made possible by specialised breeding of sheep (to have a long staple). Extensive grazing of poor pastureland was newly possible because of the invention of riding, for shepherds and cowboys (sic), and so made great areas productive which were not suited to arable farming. In fact, the modern pattern of European farming, with mixed arable and pastoral, only became possible with the innovations from the east. (However, horses were not used for ploughing - this only became possible with the invention of a working plough collar, in the 12th century AD.) Because the steppe invaders had strong contacts with the North Caucasus and its metallurgy, they were expert in bronze working. Indeed, the transition from Neolithic to Bronze Age in Europe is possibly connected with the arrival of Indo-European languages - and not with greater productivity allowed by bronze tools. Bronze was only of use in certain areas of endeavour, and was not a huge advance on stone tools. Yet the end of the Neolithic was a great transition and upheaval in the European world.

8. The second stage of IE expansion is from a speech society of peasants or mixed farmers of Indo-European speech with the population density associated with intensive land use. This stage two source society was not less numerous or dense than the societies who acquired a new IE language from them. This is easy to model. The first stage, where a peasant society takes over a language from a tribe of pastoralists who were less densely implanted in a territory, is more difficult to model and so of greater interest at this point. Anthony is very clear in explaining that pastoralism is also a source of wealth and that the arrival of 'mixed farming' may have been a much more efficient and indeed wealthy way of life than the peasant farming before the arrival of Steppe innovations in the form of horses, wool, wagons, new breeds, etc. That is, the new language may emphatically have recruited very large numbers of people from the peasant culture - acquiring new impetus and mass along the way.

9. Looking at the scenes proposed for the control of peasant farmers by horsemen brings an unmistakable impression that the ability to herd and move and instruct grazing animals was training for doing the same to humans - and that this shaped the new social organisation.

10. The breakdown into separate languages, the parents of Punjabi, Gaelic, Latin, etc., was correlated with geographical diffusion from the zone of origin. The distance in formal terms that various languages have travelled from the reconstructed Indo-European ancestor language may be an indicator of what kind of journey this diffusion involved. It could even tell us where the point of origin was - but the difficulties of translating changes in language structure into geographical space are obvious. The distance between Celtic and Germanic, as types, and the reconstructed ancestral language with its morphological richness in verbs and noun endings, its panoply of tenses signalled by endings, is striking and asks for an explanation. Why are Latin and Greek so much more conservative? were they the results of less radical sociological changes and journeys? This would be compatible with the source language being in the east - as Celtic and Germanic are found in the westernmost region, as well as in the northern part of Central Europe.
Many of the major modern works of synthesis about the period of European history in which the IE languages spread are by Barry Cunliffe. Cunliffe describes the history of language in terms largely agreeing with Renfrew. He refers to the original language 'giving rise, by creolization, to ... Slavic, Baltic, Germanic, and Celtic', in a process of contact-induced language shift (Europe Between the Oceans 9000 BC - 1000 AD, at p.138). Meanwhile Anthony says that no one has ever found any traces of creolisation in the IE languages. The contrast is amusing. Evidently Cunliffe thinks that creolisation occurred in Slavonic because anything else is incompatible with the 'Anatolian' model of diffusion. But no one who has learnt Russian could ever think this language had undergone creolisation. The paradigms of endings go on forever. Russian is an archaic Indo-European language close to the mother language. This is quite in accord with IE origins being in the steppe world, where today Russian and Ukrainian are spoken. It is not compatible with the Renfrew thesis.
Celtic and Germanic are the groups which are under suspicion of showing either creolisation or partial creolisation. This is controversial. However, it would fit with them being at the furthest stretch from a point of origin on the Volga. Sigmund Feist described a theory of Germanic as a creole in 1932. Understandably, German scholars have not been happy with this idea! It remains under-researched (and unrefuted). The suspicion remains that archaeologists take ideas about language which suit their preconceptions and leave all the other ideas or facts alone, regarding them as 'technical' or something like that. (I believe Marek Zvelebil has written about ‘creolisation’ of early Neolithic farmers and Neolithic foragers as part of the origin of Indo-European, but this must be some technical and non-linguistic sense.)
The modifications to Celtic do not much look like the changes typical of a creole. The complex verb endings are still there in modern Irish - I am looking at Teach Yourself Irish. This is the style of the oldest Celtic and our best window onto Bronze Age Celtic. However, Celtic has gone a long way away from the original Indo-European type and this is compatible with language mixture. (Actually, the case endings of Celtic nouns vanished very early, although there is a magnificent range of endings as plural markers.) The development of Welsh and of Scots Gaelic in post-mediaeval times looks quite similar to creolisation, but we know this was a late development because of the kind of language we see in the very ample manuscripts of Old Irish. It has no relevance to the Bronze Age.
Teach Yourself Irish, possibly a rather conservative textbook, discusses the existence of a dual in nouns in 20th C Irish. It is remarkably difficult to imagine a creole Celtic which would retain this notably archaic, elaborate, and unnecessary feature. Of course, within any language area, especially one covering most of Western Europe, there could be groups speaking a creolised language in some district.
The more geographically extreme languages, like Old Irish, look like the most ancient Indo-European, the more recent the break-up of the ancestral language must be and the more rapid and hard to explain is the expansion.

11. Domination. There are two discrete Left responses to the inequality of power which has marked so much of the past of Europe (although presumably not the whole past, uniformly). One is to focus on it and to regard silence about it as dishonest, collusive, abject, and basically unworthy. The other is to regard mention of it as impolite. Archaeologists since 1960 have plumped for the latter approach. This is why someone trying to trace the expansion of a language, i.e. of the territory inhabited by its speakers, is baffled by the modern archaeological literature. The information is there but it is buried under interpretative phobias and collective denial. Anthony is not an archaeologist but an anthropologist. His knowledge of archaeology comes mainly from scholars of what used to be the Soviet Union - who have a quite different set of prejudices than their western colleagues. (These prejudices were a quite different set of wishes and intellectual practices than Marxist ideology, even if their publications up to 1989 made obeisance to the grandiose errors of Engels and his pals.)

12. If we accept that the written record of the Middle Ages was dominated by the literate, a tiny minority, then we can guess that the picture which the written records tell reflects the wishes, including the unfulfilled wishes, of that minority. The literate were either part of the dominant group or at least worked for them. It would follow perhaps that the image of feudalism is not historical reality in its full extent but the projection of the wishes of the wealthy. This gap can be overrated, but I just want to observe that the idea of oppression complete in its nature and geographically 'flawless', i.e. without regions or groups for whom it hardly applied, is too simple.

13. Cunliffe has also published radical new theories, developed with John W Koch, in a book about Celtic from the West. This depends on a new theory, advanced by Koch and not yet proved, that the inscriptions of roughly 800-600 BC found in the region of Seville and associated with ‘Tartessus’ and the higher culture of the Phoenicians bringing East Mediterranean innovations, were in a Celtic language. The claim piled on top of this is that this first inscription was also the point of origin of the Celtic language family, and that this family spread along the shores of the Atlantic, without major migrations, from south-west Spain. Hence ‘Celtic from the West’. The inscriptions are all very short, and in a mainly alphabetic, partly syllabic, script whose phonetic values we know now, because they are similar to those in other Spanish inscriptions in other languages. The problem with Koch’s reading of the brief funerary inscriptions is that he does not know what they mean. The recognition of ‘Celtic’ roots as known from Welsh is only correct if the meaning is correct - but this is derived entirely from the glint of recognition. This method has been acknowledged as effective by the majority of scholars for certain classes of inscriptions, but these have the common factor that they are long texts. You can be blinded by glint.
Tartessian may be the oldest written version of a Celtic language. This does not make it the zone from which Celtic speech radiated - over territories which had no writing at the time (so that their speech could be Celtic, Slav, or anything you like). The oldest IE languages are the Anatolian ones. No one regards them as the point of origin for all the other families. They are too aberrant and have too many structures missing. First recording in writing is not the same as being the first to be spoken and the oldest. Celtic may have been a discrete entity in 2000 BC or even 3000 BC - in the stage of diffraction and loss of continuity.

14. Cunliffe and Koch, fascinated by the anti-migration fashion, suggest that Celtic could have spread from the region at the mouth of the Guadalqivir to Scotland as a trading language which people on the coast learnt in order to trade and then adopted as their daily speech. Then it spread from the coast inshore, to the hinterland over huge areas, flowing like water. None of this involved migration or any human movement except a few merchants and sailors who did not settle. Thus Gaul and the whole of the British Isles abandoned their languages to adopt this new speech, the littoral-mercantile languages. None of this is credible without population transfers as well. I am not aware of any examples in real history. Surely it is more plausible that Celtic came from the east, as one of the waves of Indo-Europeanisation which brought Italic and Germanic too, that it was already the speech of the Atlantic littoral before the first writing down of Tartessian words, and that assimilation along the coast only involved some details, names for imported objects, rather than a complete language transfer. The Celtic languages, probably, do not record the common elements of the cultures along the Atlantic coast, and the importance of these did not guarantee a common language and stands even without that. Cunliffe is insistent on the continuity of the Atlantic Littoral, and connecting linguistic discontinuity, i.e. indo-europeanisation, with archaeology is indeed a problem. Cunliffe does not explain, though, how Celtic reached the Guadalquivir valley.
Merchants learn the language of their merchant partners from other ports. Fishermen develop pidgins to use in ports where they land for trade or other purposes. But is it normal for peasants to learn a trading language and to abandon their own speech for it? has this ever happened in history? Long-distance trade does not seem to have been of great importance in the late prehistoric Atlantic world. It was a subsistence economy, not a commercial one. Trade from Tartessus, near Africa, to Ireland, would have been a tiny trickle.

15. The Steppe model of Indo-European origins involves things like the invention of warfare, armies, conquest, class inequality, lordship, and feudal dues. These techniques may have been revolutionary in their time - and moved history on to the next stage, even if the word ‘progress’ hardly seems to fit. In some circumstances, history is not only ‘the propaganda of the victor’ but the story of the victor. The abolition of feudal status in much of Europe dates to the period 1789-1861. The synchronism with the rise of Indo-European studies, and the readers who formed a market for them, is obvious, however indirect the connections. This ‘political subconscious’ has long bedevilled the field of archaeology and of language history.

16. Braudel is the most famous historian of the AESC school, rejecting the history of events in favour of long-term changes and unconscious structures. Famously, he developed his ideas as a prisoner of war in a German camp, a situation where there was very little change from day to day and the long term offered the only gleam of hope. Anti-event historians underrate invasion as opposed to la longue durée. Braudel would not have been where he was at all without the events of May 1940 - a few weeks which dramatically changed the lives of millions. Perhaps his version of history eliminates foreground in favour of things that are inconspicuous and unimportant. Only looking at what is unimportant. Only listening to what is silent. Many wars simply reverse the results of previous wars, just as the defeat of May 1940 was swept away by the victory of May 1945; but that is not true of all wars, in fact they can be decisive and much of history or prehistory may be the story of wars that decided quite fundamental things -such as what languages were spoken in Europe.
Fairly obviously Renfrew is presenting us 'prehistory without events' and Anthony, along with Schrader, Gimbutas, etc., is presenting us prehistory with bloody and decisive and dynamic events. There is an opposition between kinetic explanations of history and processual ones. There is an obvious affinity between historians who reject the study of events and modern archaeologists. Both want to be ‘structural’ and both want the actions of rulers, and politics in general, to be unimportant. The idea that history might be about winning and losing control of territory and of populations is not of interest to them.
Kinetic refers to something like the edge of a sword - even more acutely, the edge of a sword at the end of the arm of a man on the back of a galloping horse. (The man on horseback can be seen as a kind of ballistic process.) A prehistoric sword may have been the agent in a process whereby one person lost their life and another became the lord, and the whole social structure flowed out from this cut. Or it may have been a display object. Or it may have been used, in wars which were hastily entered into, very numerous, and indecisive. As for a plough, this may show a social order, as structure flowed out from the economic basis, the source of calories. Or the workers in the field may have been largely excluded from the political process - as conducted by a nobility. Interest in agriculture leads us towards processual explanations and interest in armed warriors leads us towards kinetic explanations.

17. The pre-Indo-European peoples still here today include Finns, Basques, Estonians. None of them very numerous. The percentage of the European population speaking old European languages is less than 3 by my count. (Magyar is a later arrival.) The Steppe contribution is 3% of the biological material but accounts for 97% of the speakers, i.e. of the humans. We can produce models to explain this. For example, elements of dress show downwardly mobile patterns, a cascade where they start as aristocratic fashions and diffuse downwards to the population as a whole, through however many intermediate stages. Particular words have followed this path of downward diffusion via imitation of the prestigious model. The idea that a whole language could diffuse downwards is coherent, although it is very hard to find in history. The spread of Latin over much of Western Europe might not be a good example, because the arrival of the Romans involved wars destroying large parts of the human material of the native societies, and mass migrations of people from Italy to be bestowed with land grants on arrival (‘colonies’). So ‘replacement without migration’ is hard to find. All the same Gaul and Spain lost their ‘old’ languages (excepting Basque) completely.
Archaeologists seem very happy with ‘dominance diffusion’ theories of language spread, whereby a dozen or so incomers can lead to the complete replacement of the native language spoken by hundreds of thousands of people already there. This version of language change, unfamiliar to linguists, is necessary to support the theory that migrations are so rare in prehistory as almost to be banned. This theory may be wrong. The idea that prestige is vital to the language choice of individuals, so that if 99% of a population speak a language but regard it as inferior, the stigma of a social group who carry out inferior roles and lack political influence, then in successive generations the percentage may be 89%, 79%, 69%, 59%, finally 0%, as individuals gallantly fight to be upwardly mobile. Concrete examples of this without massive populations of the incoming language to provide a ‘pole of attraction’ are apparently completely missing. But if you have the preset goal of proving that the Anglo-Saxon invasions were only a few dozen people, you need to fantasize that the spread of English was due to the Romano-British giving up their own language to acquire something more prestigious, and that 3 million Romano-British (6 million, in one version) acquired a strange language from a few boatloads of Saxons (who must have had a gift as language teachers).
The idea that 3% of the genes correlate with 97% of the speakers relies on a ‘dominance diffusion’ model of much this kind. It holds that the mere ‘dominance’ was followed by a stage of ‘fusion’, in which a sharp difference between lords and peasants was replaced by a coherent culture in which everyone spoke the same language. The incoming 'knights' or 'chivalry' took over political power enough to impose their language in the new society, but adopted the territorially bound way of life enough to merge with the peasant population. Exactly how Indo-European languages could have spread to countries which were nowhere near the Steppe Zone and also not part of the Corded Ware area in the 3rd millennium is not clear, and this is an area yet to be developed rather than a weakness of the Steppe Origin theory. The further spread may not have had anything to do with horse-borne raiders and herders.
If a dozen warriors land on some foreign shore, risking their lives, they can out-fight and replace the existing elite. However, the next step may well be to marry native women. They would tend to be a male group. Their children then have mothers who speak the older language. Why should the children speak the language of their fathers? especially if hardly anybody understands it? Political power is not something you can keep locked up in a casket - it grows when you use it. Marrying the local maidens was a way of making alliances, to balance the feuds you probably have. In order to exercise power over a population, you presumably want to speak to it - and be understood. Reasons for abandoning the 'new' language, the incoming one, are thus very weighty. Tiny conquering strata can be seen, at many points in history, to adopt the language of the conquered - if only to exercise their power and make their prestige audible to all. Language replacement is more compatible with large-scale migrations - enough to make the minority language useful rather than ceremonial and symbolic. The spread of IE over almost every part of Europe west of the Carpathians is an indication of the migration of large numbers of people in many periods, not of a kind of acquisition of long-distance prestige goods by mimicry.
I should point out that if we are talking about a few aristocratic figures moving to be chieftains five miles away from home, and there is a large population that close by speaking their ‘noble’ language, linguistic assimilation of the population in the zone they are arriving in is quite credible, at least over the long term. However, the spread of the IE languages took place rapidly over great distances and we have to look at mechanisms that account for long-distance movements of languages.
I had to look at one point for records of the Frankish spoken in Northern France, in order to compare it with Anglo-Saxon, testing the theory (based on place-names) that the Saxons came from the South Netherlands rather than from Saxony. But, as I found, there are no records of Frankish, the language of Charlemagne. The relics are fragmentary, or late and radically altered (the law words written into Latin texts as marginal glosses, the so-called Malberg Glosses in the Lex Salica.) (See http://en.wikipedia.org/wiki/Lex_Salica ). Otherwise there is one late translated text and a few runes. The Franks were a 'drowned elite', they abandoned their language to adopt the Latin-derived speech of their conquered subjects. How come Frankish disappeared and left no texts when the Frankish Empire was the central institution of the West and the basis for everything that happened in Western Europe after that? The facts are strange. But factual. ‘Dominance-diffusion’ may be a clever idea that doesn't happen. It is not the norm in human sociology.
Fieldwork in sociolinguistics has turned up interesting evidence for the solidarity of the people of lower status. Someone living among the ordinary people, which could mean 80% or 90% of the population, may wish to sound like the people they live among. The idea that everyone is striving upwards is not borne out by field observations. The force of solidarity, and the psychological wealth it implies, may induce people to abandon prestigious minority speech patterns. The idea of perpetual striving to imitate the language of power is crude and undifferentiated. It is by no means the case that the oppressed majority generally admire the wealthy minority and their manners. While dialect has for centuries been dying out in favour of standard English, it is by no means true that the prestigious speech norms of the 17th C came to be the ones used by everyone in subsequent centuries. People acquired the speech of large towns, on moving into them, but learnt this speech from the bulk of the population dwelling there, not from the noble elite, with whom their contracts will have been exiguous. Urban speech levelled down dialects because the cities included people from many diverse dialect zones. The levelling was not upward mobility.
Quite how a few thousand steppe horsemen induced a rather numerous peasant population, in the early contact zones of Yamnaya Culture and Corded Ware, to adopt a new language, is a tantalising question. I think we are speaking of a numerous population, using good grazing terrain for a pastoralism which was outstandingly successful and nourishing, and comprising a whole pastoral society, rather than a few mobile and aggressive warriors. We are also speaking of migration.

18. Attila got as far as Châlons on the Marne before being defeated or at least fought to a standstill. This was ‘the Battle of the Goths and the Huns’ in 451 AD. The Huns were certainly a mounted people from the Steppes. Famously, the Ostrogoths fought in Attila’s ranks while the Visigoths fought on the Roman side. Attila’s army was a colourful mixture of a dozen different peoples. This quality of ‘rolling up’ allies and clients was presumably a characteristic of great steppe leaders, victorious gamblers whose personal power increased with every winning throw. The ability to win over or intimidate client groups was presumably part of the ‘political technology’ of such leaders. Most of the warriors in Attila’s army spoke a Germanic language, so if he had won we would guess that the linguistic result would have been a spread of Germanic - not of Hunnic. Today, we cannot find a single village that speaks Gothic - or Hunnish. The idea of Indo-European coming into Europe west of the Steppes relies, I think, on a military alliance which did not succumb to defeat and disintegration, but rather sustained itself over centuries. In this long run, only, would the client groups have come to adopt the language of the dominant clan as well as its political hegemony. Otherwise the incoming language would have disappeared like Hunnic and Gothic - and Scythian.
A valid model in recorded history for the success of Indo-European could be the spread of Turkish. Two-thirds of Anatolia speak Turkish today, yet the number of ‘original Turks’ who came from the Altai region must have been quite small. We find a number of regions where a Turkic language is today the majority language of a dense population. These regions do not have the same history, however those long and perhaps puzzling histories offer a mirror where we can examine the theorised spread of Indo-European from an origin in arid steppes that did not support millions of people. One point of interest is the ‘skipping’ pattern, in which a mobile group of people passed through a whole region and colonised the one beyond it. Thus we have Iran and Kurdistan separating the Turkic homeland from the settlements in Anatolia. The Turkic realm is a discontinuous one, and this is what we expect from Bronze Age Europe too.

19. This more specific version of how the IE languages spread has implications for substrate studies, the research which deals with elements of IE languages which are not part of the inherited IE vocabulary and which belong perhaps to submerged pre-IE languages. This involves mainly lexical items but can also extend to things like prosody and the location of word-stress. If the incoming languages had extensive contact with peasant (Corded Ware) peoples already around 3000 BC, and in eastern Europe (north of the Carpathians), then the fascinating substrate words may come from a long way to the east and the substrate layer in Dutch or English may come from the Dniestr-Bug region rather than from the shores of the North Sea. This may not be very useful if we don’t have any supplementary sources for the ‘submerged’ languages either in Western Ukraine-Rumania or by the North Sea.
Anthony sketches a theory whereby the language of the peasants in south-eastern Europe was ‘Afro-Asiatic’. This needs decoding somewhat - it is the name of the language group which includes Hebrew, Arabic, Aramaic, and Ancient Egyptian. The idea that the spread of farming went along with the slow diffusion of the people who developed farming implies that the language of these first farmers- also in Anatolia and the Balkans and the Danube basin- was a language of the Fertile Crescent. The spread began in the early 7th millennium BC, so although the arrival of writing was hardly earlier than the end of the 4th millennium, we find it very easy to believe that the peasants of 6700 BC spoke languages related to the first written languages: such as Akkadian, Babylonian. These are Semitic languages, hence the term ‘Afro-Asiatic’ applied to early European peasants, which is the most diffuse and permissive way of indicating ‘Semitic or somehow related to Semitic’. It is a little difficult to conceive of the peasants of Bulgaria and Rumania speaking Egyptian, and the ‘Afro-Asiatic’ languages in south-west Asia are, all, Semitic, so we are really talking about Semitic languages in Europe. Or, maybe, ‘peri-Semitic’, or ‘Pre-pre-Semitic’, or ‘peri-pre-pre-Semitic’. There are however early languages in Anatolia and in the Fertile Crescent which are not Semitic.
Anthony points to ‘tawr-’ (bull) as a possible loan word in western Indo-European (as a loan from an Afro-Asiatic language). This was first suggested in 1832, but is abidingly an ‘outlier’ for language historians. As a European loan-word, (not known in Indo-Iranian), this could be a trace of contacts with the European farmers who transmitted the earliest domesticated cattle to groups roaming the Ukrainian or Moldavian steppe. However, this word exhibits the phenomenon of ‘s mobile’, so that German Stier (English steer, ‘bull’), with an initial s, corresponds to Latin taurus, Irish tarbh, etc. This phenomenon is generally regarded as belonging to the earliest stratum of Indo-European - and, less certainly, the initial s- is seen as original. This would mean that the root is ‘stawr’ rather than ‘tawr’. Tangible evidence for a possible pre-Semitic language being spoken in south-eastern Europe is non-existent. Which means that there is no evidence against it. It is quite a beautiful idea. Indeed, if farming spread across Europe from the south-east well before the waves of Indo-European, we can readily imagine pre-Semitic being spoken in France, Belgium, England, or Scotland. I like this idea.

20. Why do modern archaeologists hate migrations? or, why did archaeologists up to 1960 or so love migrations and produce beautiful maps showing so many of them? In the 1950s, carbon-14 dating came along and allowed an objective chronology to be developed, for the first time, for the parts of the world which were not very close to Egypt and other Near Eastern civilisations, with their written historical records. Unfortunately, archaeologists had not allowed this lack of basis to stop them from dreaming up dates and chronologies for the whole of Europe, and to some extent for ancient prehistory - anyway for the Bronze Age and Iron Age. Carbon-14 isotope measurements demolished a whole swathe of official knowledge which was part of the legitimating property of learned institutions. No real knowledge was destroyed, but a lot of embarrassment was caused, and people learnt something valuable about the sociology of knowledge, and legitimation or exclusion of knowledge. (The assimilation of the raw data provided by carbon-14 came after a delay, after 1960 or even after 1965.) A major casualty of this revolution was the elaborate scaffolding of Diffusion, by which everything in western and northern Europe was a delayed imitation of world-leading processes taking place in the Near East. Objective dating made it likely, if not wholly certain, that many of these advances had been made locally and autonomously. An extension of this was a critique of the migrations which had been postulated by so many older archaeologists as the bearers of cultural change. Each of these could be discussed as local imitation of prestige goods and customs, paid for by local elites who were acquiring new display objects and not themselves being replaced at all. Rather than have a million people migrating, and a million others dying, you have ten top craftsmen migrating and producing ‘top goods’ which are prestigious because they refer to the elites of other countries. It is more like buying a high-performance car than being conquered. It is more economical and so a better explanation. There is no longer any “Bell Beaker people” but a fashion, a linked set of prestige goods.
This was one of the great ideas of the 20th century (personal view!) and most of its results are not going to be rolled back. Regrettably, this hypothesis became too admired and swelled to the level of dogma. It seems likely that some processes in prehistory were in fact migrations, and that archaeologists who have graduated since 1960 are phobic about this real possibility.
In the popular media, as well as in learned circles, there was a levelling emotional investment in invasions, a kind of aesthetic sensibility which was interested in duels and battles and victories and conquests before all else, and which saw other processes in history as essentially tedious interludes which happened monotonously and in infeasibly large quantities. This was heavily kinetic and anti-processual. It was related to the nationalism which came to a partial halt in 1945, and its mythology of armies, heroes, acquisition of territory, triumphs, humiliation of national enemies, etc. Today, this repetitive imaginative world belongs to the neo-conservatives. It is understandable that archaeologists should have reacted against this since 1945 - or really since 1933, when the implications of militarism were already apparent to the educated strata in European countries, observing Germany. However, it is possible that European history and prehistory really were full of wars, victories, conquests, and subsequent migrations and creations of new elites, even if this offends the feelings of modern scholars. The peaceful life-styles of modern university scholars are not genuine arguments to prove that people in prehistory could not have been militarists and that certain Bronze Age societies could not have had ruling strata who were foreign invaders.
Having said that this was a great idea, it would be churlish not to credit it - to Renfrew and to JGD Clark.

21. Kossinna.
Actually, some archaeologists were militarists. You have to know the history of the discipline enough to realise how far into racist and militarist fantasy many of its leading scholars, in the period of high nationalism, say 1870-1945, plunged. Nazism was a product of the imagination and that imagination shaped the past according to its predilections. The sponsors of Nazism and the tier of educated Germans were not two distinct and segregated groups of people.
Gustaf Kossinna (1858-1931) was the scholar who linked archaeology and racism, cunningly and inextricably. His 1910 paper on the origin of the Indo-Europeans already has the essentials of the Nazi view of the world. It is impressive as a work of ideology, insistent, disguising itself behind accumulations of true facts, difficult to unravel from the facts. He was one of those people whose emotionally chosen ideas are so extreme and so unpopular among the scholarly community that they overwork, tirelessly and accumulatively, gathering facts to weave into their prejudiced programme so as to defy criticism. Indeed, advances in scholarship, which mostly demand more work and more time, are often inspired by unpopularity - someone who is likely to lose, given the opposition of much of the audience, can develop new methods and a new level of detail in order to crush their rivals and become the winning side. Kossinna’s wildly subjective fetishization of the Germanic people, seen as the master race and as the most authentic Aryans, gave him a faith which enabled him to carry out staggering amounts of work - as worship of his ideal. He was a loser but an obsessional one. This moment of flamboyant fantasy has left traces which affect the silent rules of the profession to this day. If we see a small number of IEs conquering huge tracts of Europe and winning with unnatural consistency, we tend to flash back to Kossinna’s world-view, and recognise the scenario. But, this military success may really be why a language spread from one obscure and not very fertile or rainy area to the whole of Europe, and beside that to much of India and to Iran. History can be the story of the victors.
If you look at the linguistic pattern which probably prevailed across Eurasia in 4000 BC, there would have been hundreds of language groups. The infant Indo-European was just one of these. As time rolled by, IE won and won and won and constantly eliminated its rivals, as we were still seeing in the 20th C. This ‘loser dies, winner reproduces’ pattern is familiar from population genetics, and the analogy with families of humans reproducing is almost inevitable. This analogy attracted extremists who already believed in a ‘master race’, but it is also possible that it has put modern scholars off: we don’t want history to be the story of the winners. But linguistically it really was the history of the winners, and still is. A language which has millions of speakers is just a more precious asset to a skilled speaker than one with 5000 speakers, and populous languages develop irresistible attractive power.
Kossinna identified (in 1902) the Corded Ware people with the earliest Indo-Europeans, which allowed him to place the source of the latter in North Germany. He was trained as a linguist and possibly saw archaeology mainly as a support tool for reconstructing the prehistory of languages, identified with the ethnic groups which spoke them. He spent a lot of time in museums but possibly never led an excavation. The idea that a shift in aristocratic accoutrements was a matter of fashion would have infuriated Kossinna, it destroyed more or less his whole life’s work. Without migrations, no master race. His proposal that material cultures could be equated with population groups, i.e. peoples, was accepted, some time after he developed it, by the majority of European archaeologists. It is part of the ideological armoury of most amateur archaeologists, the readers of books about archaeology, to this day. However, a large proportion of academic archaeologists who have been trained since the radical surge of 1968 no longer accept it, and are in search of less suggestive terminology and a sparser kit of assumptions. It is an outstanding example of an assumption which had more social support, i.e. among people living in the 20th C, than strictly evidential support.
V Gordon Childe took his basic ideas on ’material culture = people’ from Kossinna. His vision of prehistory through migrations and warfare strongly resembles Kossinna’s. It does not correlate with his conscious political views, which were Marxist.
I am not aware of a militarist party in archaeology or philology at the present day. This whole line seems to have dried up in May 1945. However, my impression is that such a party may exist outside the professional world of specialists - thriving among amateurs and on Internet forums. There is for example a publishing genre of ‘archaeology for wargamers’.

22. Marija Gimbutas wrote about the relationship between the Steppe world and the peasant culture of south-eastern Europe, representing a continuity between the tradition of scholars linking the IEs with the Steppes from the 1890s on, and the new version of Steppe origins, based on modern Russian archaeology. I mention this because several recent works attribute the Steppe Theory to Gimbutas, as the latest adherent of it in around 1990, counter to the usual practice of attributing a theory to the scholar who first formulated it. The ‘Urheimat’ was placed in South Russia by Schrader already in 1883, also by Uhlenbeck in 1895, and in the Kirgiz Steppe by Brandenstein in 1936. Koppers saw their culture as closely related to that of the earliest Turks. Gimbutas follows in the line of this older German scholarship in placing the Indo-Europeans at a point from where access to India was as credible as access to central Europe. Renfrew has described this as the SCGM theory, for 'Schrader, Childe, Gimbutas and Mallory'. The model she offers is radically unoriginal, as it strongly resembles the Nazi interpretation of early European history, with a master race seizing all the key points in the fabric of the state and the economy and sweeping everything aside. I think her work is obsolete and weighed down by lavish psychological fantasies. The problem with it, even as sentimental fiction, is that it casts the whole of European history since 4000 BC as the achievement of the violent and aggressive and domineering. This is simply unacceptable. The point that a new culture arose in eastern Europe at the start of the Bronze Age, that it had mixed origins and that it was not suffering from some sort of ancestral curse, has to be accepted. Gimbutas' theory coincides too much with Nazism: she is saying that all the achievements of European history are due to the Aryans, the Triumph of the Will. No, we can accept that warfare was a preoccupation of the European past but not that militarism was the force which made history progress. This is an emotional response but Gimbutas' work is overwhelmingly guided by literary impulses and not by scientific ones. Are we supposed to say No to metal, horses, and the wheel?
There is a line of sentiment which finds the whole of European history to be evil and power-seeking, but which attributes these problems to a minority, while underneath them the Innocent went on leading peaceful lives. There are other fringe archaeologists who have promoted these ideas. HJ Massingham, who promoted Diffusionist ideas, also wrote in the 1920s (in ‘Downland Man’, 1926) about how evil the Celts were, obsessed with militarism and expert in the use of metal, obviously for military purposes - in contrast to the gentle and peaceful Neolithics. This was a reaction to the trauma of the First World War. He simply saw the Celts as the Prussian Junkers projected back into the first millennium BC. I think everyone loves peace, but the search for a small group of people who make wars happen is in vain, as such a group does not exist. However, an account of European prehistory and history which passes over the existence of warfare is inherently flawed.
The wish to eject guilt from history expresses itself by ejecting the low-status majority from history and proclaiming their innocence of the great events which led to conquests and dynasties. My impression is that fringe archaeology in general has a high density of people who wish to say no to history and to identify with an ‘innocent majority’ who are permanently abused by a violent minority. This is the view of Marxists but it is much more widespread than that. The wishful thought here is to have a European militarism which can be separated from the history of European politics in general - as if the occupation of the land could be separated from the wish to possess land.
Part of the readership for books about archaeology prefers not to hear about the mighty and the wealthy and their feats and procreation. People who dislike the landowning class of the 20th century may well dislike their forerunners in the third millennium BC. This is what we mean by a ‘political unconscious’. It is very noticeable that the hippies loved the Neolithic and wanted to reject the changes which arrived with metallurgy and so on. If they also loved the Celts, it was from screening out the data from Classical historians about Celtic love of warfare and so on.

22a. According to Bruce Trigger, (A History of Archaeology pp.168-70), one of the main themes of JL Myres' The Dawn of History, in 1911, was 'his belief that all hierarchical societies developed when politically dynamic, pastoral peoples, such as the Semites and the Indo-Europeans, were forced by drought to leave their homelands and to conquer and to rule politically less innovative peasant cultures. According to Myres, the Indo-Europeans, whom he believed to be nomads from the steppes of central Asia, were particularly adept at imposing their language, beliefs and social customs on conquered peoples, while adopting the latter's material culture.' The link between the apparently rapid expansion of IE and the dynamism of steppe horsed pastoralists in times for which we have narrative histories was made very early in the history of the discipline. I understand that Adolphe Pictet's work of 1859 already locates the Indo-European homeland in the steppes to the east of the Urals. Actually, the major questions and the most likely answers were formulated a very long time ago. We are in a postclassical age in that sense. The really new element today, or in the last 20 years, is the maturing of Russian-Ukrainian archaeology and the effective grasping of the subsoil archive of a cultural area which had few permanent settlements and which was of huge extent. This allows theories about the location of early Indo-Europeans in time and space - and in steppe excavations - to be made exact - so that they can be tested properly - so that we can reach some level of confidence in them. Renfrew famously pointed out that Dumézil never said, over 400 publications, where or when his beloved indo-Europeans lived. Anthony’s virtue is less originality than precision - because he can be proved wrong he can be proved right.
Even the link between the invention of riding and the spread of the new language was made very early - by Childe in 1926, if I am not mistaken. Childe's testament in 1957 recommends more research into this riding/Indo-European link. In all this time, no-one has ever produced a clinching reason why the Steppe Origin theory should be wrong. The rider is that, because nomads had no need for writing, records of languages spoken in the steppes are very late and there is no evidence in this region of a date relevant to the Indo-European dispersion.

22b. The Steppe pastoralist theory was not the consensus theory in the period 1890-1970. Why not? I think the reason is simply that so many scholars in the field were German nationalists, and they all wanted the Indo-Europeans to have come from Germany. This is the legacy problem with the Corded Ware theory - it was used by Kossinna in 1902 to relocate the Fore-Homeland to North Germany. The historic populations of Iran and North India would thus have migrated from Germany. Populations? actually just the ruling class, the 'bearers of culture'. But outside German and Austrian scholars of a deutschnational mentality, Steppe Origin was the most popular theory. There is nothing anticlassical about believing in it now.
My impression from reading Scherer's anthology of 'solutions' is that the language historians failed to produce a convincing result where there were no language records - so archaeology had to take over. Locating the 'Urheimat' at a point where both Europe and the Punjab were accessible was a speculation based on geography, and language history had nothing to say about it. A substantive account was held up by the inadequacies of Soviet archaeology - and followed the development of that discipline with the advent of prosperity in the Soviet Union. A further delay, probably, was due to a lack of knowledge of Russian among Western archaeologists.

23. The disappearance of all the old languages of Europe, with so few exceptions, is alarming. European history seems to open with the population surrendering their culture and folklore. Oral literature is bound to the language in which it is memorised and, from observation, tends to be lost when the speakers change their language. Translation is possible for some genres but not for others. Tunes would have survived, and musical instruments. Linguists are happy to sift through fragments of lost memory and detect possible non-Indo-European languages which survived to the threshold of literacy, towards 0 BC; to Basque, Estonian, Finnish we can possibly add Etruscan, Rhaetic, Iberian, and perhaps other, more shadowy, tongues and their speakers. But as literacy arrives and gives us historical narrative we see grand political events replacing one IE language with another. Thus, Latin became the speech of Iberia, but several of the languages it replaced there were already Indo-European, to be exact Celtic.
The demise of the Old European languages is a startling process. It calls for study and I can imagine people writing entire books about it in the future. There is no direct evidence for it but it is something we can use our leisure hours to think about. It seems like a remarkably thorough process. Every 'accepted idea' about peasants is that they are obstinate and conservative, they are the deposit of archaic words and customs. Somehow in the Bronze Age this was not true.

24. Anthony uses as his model (p.118-9) for people changing their language, in the phase of mixture between incoming steppe pastoralists and European farmers, in the Corded Ware period, a piece of research by the great Norwegian anthropologist Fredrik Barth. Barth (b.1928) did fieldwork (in the Fifties?) in Pakistan, among two groups speaking different languages: Pathan and Beluchi. The Beluchi were pastoral nomads, the Pathans cultivators. Barth recorded Pathan people who had lost land or position in their birth society moving over and becoming Beluchis. Among the Beluchis you could find sponsors who would become your patron and lend you cattle, acquiring you as a human asset without regard to genealogy. This cattle-herding society was more receptive to individuals and success in managing the herd could lead to a rapid rise in social status. Arable land was a non-elastic asset which was merciless for individuals who were born into the wrong family or lost inherited plots through misfortune. However, this does not show the peasants losing their language. Pushto is not dying out. The Pathans, or Pushtuns, are very numerous and over the past 30 years have been a major factor in world politics as the pool of recruits for the Taliban and the insurgency in Afghanistan. This model does not demonstrate the features which it is supposed to demonstrate. It does not illustrate a process by which a new Indo-European language could have spread over a more numerous peasant population.
Anthony's description of how one language could slowly take over the whole huge but interconnected Corded Ware area is convincing and has many features. But an observed model for this prehistoric process is missing. We know of dramatic processes in the Steppe world where huge areas were rapidly fused into one polity under the authority of successful khans, so that the whole Steppe world became a unity - above all a military one. The leaders were risk-takers and were rarely successful in the long run. Their followers were not at all loyal in the case of failure. The outcome, once such a heady unifier overreached and fell, seems to have been that the different ethnic groups were still speaking their own languages. The process of acculturation was much less rapid than the process of building allegiances. Yet, language shift did eventually follow the force lines of allegiances - 'dominant' languages did spread and 'absorb' client clans if the dominance lasted long enough.
Anthony is very good at explaining how the dynamic of wealth in stockbreeding may have produced a different kind of society in the semi-arid grazing lands. Flocks, in contrast to arable land, can be multiplied at great speed so long as you have access to grazing land. Military prowess may have guaranteed this access. The combination of mobility and of rapid accumulation of the basic asset of animals may have produced a much more fluid society in which, critically, client-patron relations partially replaced genealogical ones. Such a society had the potential for expansion - by flocks breeding, by migrating on horseback, and by acquiring new individuals or families by patronage. This last may have extended to taking over entire villages in what we would see as 'feudal' relations.

25. Dušan Boric (in Prehistoric Europe. Theory and Practice, edited Andrew Jones, 2008, p. 132) refers to 'Around mid-5th millennium cal. BC this continuous development [in Serbia] was temporarily disrupted in most of the areas covered by the Vinca and Tisza cultures of the central and north Balkans. It seems that this change is related to new cultural formations [...] from the upper reaches of the Tisza river. These mobile herders might have brought an end to some tell settlements, such as Vinca." Thus 'mobile herdsmen' were intruding into Vinca Culture cultivator lands in the mid-5th millennium BC. They were 'mobile' by foot power, because there were no riding horses at that date. They could not be Indo-European at this early date (following Anthony’s analysis). They came from the north-east - the Hungarian plains, roughly. This unexpected 'preview' of the Indo-European expansion perhaps shows us a social-military technique being developed - perhaps by 2500 BC this was a piece of political knowledge which the steppe chieftains were thoroughly familiar with. Any such early intrusion leaves the possibility that parts of the peasant world, in east or south-east Europe, may already have spoken Indo-European dialects prior to 3000 BC.
The history of peasant: pastoralist contacts over a long border (in the Danube-Dniepr area) was very long even before the Yamnaya-era expansion which Anthony is describing. In particular, Anthony explains the Anatolian languages (Hittite and Luwian) as the record of an expansion from the (eastern) steppe world well before the arrival of horse-riding and the wheel. This accounts for the puzzling relations between Hittite and related Anatolian languages and the 'classical' Indo-European languages and their verb structure (and so on). Archaeologists frequently read these contacts as pastoral intrusion onto settled cultures, with raiding and violence.
Because we have pastoral groups on flat and grassy areas of Europe so early, the Indo-Europeans would have had to contest these resource niches and displace (or absorb?) the earlier pastoralists. In the model, the key innovations of horse riding and wheeled transport were the advantage which made this possible.

26. The Danish archaeologist (working in Sweden) Kristian Kristiansen produced a map (in ‘Europe before History‘) which shows 8 culture areas in Europe in the Bronze Age. This is a major work of synthesis and such generalisation is the product of a vast amount of detailed work. It is irresistible to try to match these up with the language groups recorded much later, as writing arrives in these parts. The matching is interesting because it is not obvious, the fit is intriguing but not good. Cunliffe’s ‘Celtic from the West’ theory says simply that the ’Atlantic zone’ on this map is the homeland of Celtic, i.e. of a language ancestral to Gaelic and Welsh. Scholarship up until 2001 or so placed a Celtic language in the Hallstatt-La Tène area, which following Cunliffe we now think was the product of an eastwards migration of the Gauls, the one recorded by historians like Livy. The preceding language in the region of Bavaria and Austria was therefore not Celtic. So what language was spoken in the ‘North Alpine’ culture area in the Bronze Age? We seem to have lost this one. It looks like carelessness on a grand scale. But inscriptions from this area are remarkably few. Logically, therefore, it could have been the home of a non-Indo-European language for which we have no testimony. The testimony we have is Celtic and this may well be late and intrusive.
The culture zone located in south and eastern Spain may match up with a language known from inscriptions and referred to as ‘Iberian’. It was non-Indo European. So perhaps this culture zone was non-IE in the Bronze Age and perhaps it, too, represents a surviving language of Old European peasants.
Otherwise, three of the zones on the map can be fitted quite comfortably onto the areas of Germanic, Italic, and Greek. There is a ‘Nordic’ zone and a ‘Lusatian’ culture area, either of which could be the area of Germanic. If Germanic is the 'Nordic' culture region - then Lusatian is something else.
The ‘Carpathian-Danubian’ zone is problematic in terms of our grasp of the later languages of this area, which we think were diverse - Illyrian, Dacian, Thracian, etc. The evidence of say 0-500 AD is very poor and the languages died after that, so it is not surprising if the matching does not work very well. Few scholars would accept that this was a single language area, even though a single culture area. Perhaps we are saying that the Balkans were already balkanised.
I think that there is considerable room for further speculation about the relation of these cultural realms and boundaries to linguistics - which I look forward to.

27. Some scholars suggest that Rhaetic, spoken in perhaps east Switzerland, Austria, and the Veneto, was a non-Indo-European language. Wikipedia says:

>>Raetic (also Rhaetic, Rhaetian) is an extinct language spoken in the ancient region of Raetia in the Eastern Alps in pre-Roman and Roman times. It is documented by a limited number of short inscriptions (found through Northern Italy and West Austria) in two variants of the Etruscan alphabet. Its linguistic categorization is not clearly established, and it presents a confusing mixture of what appear to be Etruscan, Indo-European, and uncertain other elements.<< (from: http://en.wikipedia.org/wiki/Raetic_language)

One line of opinion says that Etruscan was not a migratory language from the Aegean, but an old European language - and that it was part of a family with Rhaetic. So this is another candidate for survivals of the pre-Indo-Europeans. Perhaps it was the language of the Bronze Age North Alpine realm, and of the bearers of the Hallstatt culture? the evidence is minute. But we have speculation to fall back on.

28. Anthony describes a cluster of five terms to do with vehicles which can only date from the invention of the wheel. Fortson adds another, the word ‘nave’. The word in this list which means ‘pole’ or ‘wagon shaft’ has a descendant in English, the word ‘oar’ - I suppose this means ‘long wooden member used for transmitting energy for propulsion’. It does not mean part of a wagon, though. It is the coherence of this group of words which is convincing. Darden adds another, the one for 'harness' which appears in Hittite as turijo. He discusses this one but is not convinced.
As an afterthought, a comment on the incompetence of European archaeologists before carbon-14. Archaeology arose from the study of antiquities of the parish, a matter of local pride. As it matured, it involved intense and rather muddy engagement with particular local sites. Most of the profession were very attached to the sites and the finds. The number of people who got into long-distance comparisons was quite few, and I think this accounts for the flimsiness of some of the theories which unified data across great distances. For example, in the 1920s Childe was possibly the only archaeologist in Britain who knew a whole range of European languages and had the energy to read, every year, publications from most of the continent and to form them into syntheses. Indeed, archaeologists in different countries did not use an 'international' nomenclature, so that the same phenomena in different regions were written up using quite different descriptive names, and comparison was and is rather difficult.

29. Anthony cites an article named 'Pokhovannya dobi eneolitu-ranu bronze na pravoberezhzhi Pivdenno Bugu'. The Pivdenno means 'south (the South Bug river), and corresponds to Russian 'polden'. The word means 'midday' (etymologically 'full day'), and south is the position of the sun at midday - in Polish, 'midnight' also means 'north'. This article must be in Ukrainian. The correspondence between Russian pol and Ukrainian piv is bewildering but is a feature of many words. This is why the find-site is 'Tovsta Mohila' and not 'Tolstaya Mogila'. Anthony is willing to read articles in Ukrainian - this is how high the bar is, I'm afraid. I saw a posting which argued that because evidence for something was not on the Internet there was no evidence. I think this is a pathological error. Very little of the evidence or the arguments in historical linguistics is on the Net. To get with the Steppe Theory, you need to read recent archaeology published in Russian and Ukrainian. At a pinch you could spare the Ukrainian. People who do not have access to university libraries may find this exclusive - it shuts them out. Anthony’s work is available in paperback, in English, and it transmits the results of an immense amount of archaeology from east of the Bug.
This set of circumstances also justifies me writing amateur notes that can be read on the Net.
(Darden quoted from his paper in Robert Drews, ed., Greater Anatolian and the Indo-Hittite Language Family , 2000)