A New Approach to Language and Archaeology: The Usatovo Culture and the Separation of Pre- Germanic David W. Anthony Hartwick College, Oneonta, NY Regional variants of Tripolye C2 in western Ukraine and Moldova might have played an important role in the origins of the northwestern Indo-European language branches. In particular, the Tripolye C2 Usatovo culture might have played a significant role as the intermediary between Proto-Indo-European and the Germanic branch. The influence of the Usatovo culture extended up the Dniester, and upper-Dniester Tripolye C2 cultures extended this chain of social interaction into southeastern Poland during the final centuries of the Trichterbecker or TRB culture, prior to the appearance of the Corded Ware horizon there. The Proto-Indo- European dialects that would ultimately form the root of Pre- Germanic might have spread up the Dniester from the Usatovo culture through a nested series of patrons and clients, eventually being spoken in some of the late TRB communities between the upper Dniester and the Vistula. These late TRB communities later evolved into early Corded Ware communities, and it was the Corded Ware horizon that provided the medium through which the Pre- Germanic dialects spread over a wider region. Any attempt to connect the archaeological evidence from prehistoric Europe with the linguistic evidence from the early Indo-European languages must explain how the connection is to be made. The need for each interpreter to explain his/her own approach shows that the underlying problem—how to connect archaeology and language—has not been solved, or at least no solution has been widely accepted. Nevertheless, many observers among both linguists and archaeologists agree that just from a geographic point of view the Corded Ware horizon of the North European Plain, 3100-2400 calBC, probably was related in some way to the origins of at least some of the north-western Indo-European branches: Celtic, Germanic, Slavic, and Baltic (Mallory and Adams 2006:452). But archaeologists have demonstrated that the Corded Ware horizon had mostly local origins in the pre-Corded Ware Volume 36, Number 1 & 2, Spring/Summer 2008 2 David W. Anthony cultures of Germany and Poland, and few linguists now believe that the Proto-Indo-European homeland could have been located there. Convincing archaeological evidence for a migration from any of the stronger Proto-Indo-European homelands into the North European Plain at the beginning of the Corded Ware period has not been found. The northwestern group therefore constitutes one of the hardest knots in the tangle of problems that continues to separate archaeological from linguistic evidence. This essay advances the hypothesis that the late Tripolye cultures of the forested uplands northwest of the Black Sea in what is today western Ukraine and Moldova, regional variants of Tripolye C2 as defined by archaeologists, played an important role in the origins of the northwestern Indo- European language branches. In particular, the Tripolye C2 Usatovo culture might have played a significant role as the intermediary between Proto-Indo-European and the Germanic branch. Usatovo was strongly influenced by the steppe Yamnaya horizon before and during its first appearance and early development in the steppes around the Dniester estuary. The influence of the Usatovo culture extended up the Dniester, and upper-Dniester Tripolye C2 cultures extended this chain of social interaction into southeastern Poland during the final centuries of the Trichterbecker or TRB culture, prior to the appearance of the Corded Ware horizon there. Fortified Tripolye C2 centers such as Brynzeni III and Zhvanets on the upper Dniester were in close contact with Usatovo in the steppes (Brynzeni III-type pottery is the basis of the Usatovo painted-ceramic repertoire) and with late TRB fortified towns in southeastern Poland, notably Gródek Nadbu(cid:3)ny and Zimne (Bronicki, Kadrow and Zakoßcielna 2003; Koßko 1999; Movsha 1985). The Proto-Indo-European dialects that would ultimately form the root of Pre-Germanic might have spread up the Dniester from the Usatovo culture through a nested series of patrons and clients, eventually being spoken in some of the late TRB communities between the upper Dniester and the Vistula. These late TRB communities later evolved into early Corded Ware communities, and it was the Corded Ware horizon that provided the medium through which the Pre-Germanic dialects spread over a wider region. Material culture is not correlated one-to-one with language in the modern world, has not been so correlated in The Journal of Indo-European Studies The Usatovo Culture and the Separation of Pre-Germanic 3 historical or ethnographic societies, and is not correlated in this hypothesis. Languages expand or contract not with changes in cooking pots, but with shifts in the sources of power and prestige (Kulick 1992; Wardhaugh 1987; Fabian 1986). The spread of a language should be tracked archaeologically, if it can be tracked at all, through the geographic expansion of a novel prestige system, a new fashion in displaying and attracting social power, particularly one that accompanies a novel way of accumulating wealth and/or food. Among tribal societies prestige, wealth, and food usually were directly connected. The introduction of cultivated plant foods and domesticated animals by pioneering farmers has almost always carried their language with their introduced economy (Bellwood and Renfrew 2002) not just because of demographic advantages, but because a prestige system based on the seasonal feasts enjoyed by all human societies (Dietler and Hayden 2001) was tightly connected with a novel and more productive way of accumulating food. Military conquest, another vector of language expansion, carries the language of the conquerors only when it provides the defeated access to the new prestige system at a relatively low social cost—in other words, when there is both little negative shame or humiliation for a person’s family if that person cooperates with the conqueror, and widespread positive public recognition or reward for most who cooperate, including the opportunity for their children to advance to higher social positions (Anthony 2007; Mallory 1992; Atkinson 1994,1989; Barth 1972). Rome provided those opportunities at a low social cost, particularly in the Roman army, and as a consequence many conquered peoples adopted Latin. The Norman lords in England did not, and their language was not widely adopted. Imperial conquest and colonization by farmers were not the only vectors of language expansion in the ancient world. Among prehistoric tribal societies that lacked empires and standing armies, yet lived in a landscape already occupied by a variety of farming and herding cultures, we must be able to identify other causes of language expansion. I would suggest that even in the cases of imperial conquest and agricultural colonization the underlying social process that attracts speakers to an expanding language is the expansion of a novel prestige system associated with a new economy. It seems to me that this kind of model is already held even if unarticulated by Volume 36, Number 1 & 2, Spring/Summer 2008 4 David W. Anthony many observers, and that it is one underlying attraction of the Corded Ware horizon as a possible material indicator of a significant episode of language expansion across northern Europe: the Corded Ware horizon represented the rapid spread between about 3100-2700 calBC of a new, more pastoral and mobile economy connected with a new prestige system represented by new varieties of weapons and a new culture of competitive elite drinking parties (Sherratt 1997). Language shift follows prestige, wealth, and social power, and the Corded Ware horizon seems to have introduced fundamentally new sources of prestige and new kinds of pastoral wealth. Regardless of the specific connection with north-western Indo-European, we might expect some language shift to have followed those Corded Ware chiefs whose behavior and language was seen as epitomizing the new standard. The Pre-Germanic phase Linguists do not use the prefixes pre- and proto- in a consistent way, so I should be clear about what I mean by Pre- Germanic. Proto-Germanic was the language that was immediately ancestral to the known daughter languages in the Germanic branch. The sound changes that defined Proto- Germanic, summarized under Grimm’s Law and Verner’s Law, probably were still spreading and becoming established in Scandanavia and northern Germany at the time of Julius Caesar. But Proto-Germanic occupied just the later portion of an undocumented period of linguistic change that must have occurred between it and Proto-Indo-European. The intermediate language stage was Pre-Germanic. Pre-Germanic represents not a language but an evolutionary period defined by Proto-Germanic at one end and Proto-Indo-European at the other. The earliest phase of Pre-Germanic was a western dialect of Proto-Indo-European, and therefore, ironically, we can say more about it; the latest phase was an evolved set of dialects that were transformed by Grimm’s Law and are almost unknown. Pre-Germanic probably was spoken near Pre-Baltic and Pre-Slavic, given the network of borrowings between them; and also exhibited borrowings with Pre-Celtic and Pre-Italic. These branches constitute a north-western sub-group within Indo-European (Mallory and Adams 2006: 78-80). They absorbed elements from non-Indo-European substrate The Journal of Indo-European Studies The Usatovo Culture and the Separation of Pre-Germanic 5 languages spoken earlier in northern Europe. Schrijver (2001) summarized the evidence for at least three different extinct non-Indo-European languages or language families with different phonological systems that were in contact with the north-western Indo-European languages: 1. Krahe’s ‘language of Old European hydronomy’, preserved principally in river names, now thought by many linguists to be non-Indo- European (alternatively, if Indo-European, these names could be remnants of Pre-Germanic itself); 2. the ‘language of bird names’, preserved in the names of several kinds of birds, including the blackbird, lark, and heron, and also in some other terms borrowed into early Germanic, Celtic, and Latin, including the terms for ‘ore’ and ‘lightning’; and 3. the ‘language of geminates’, Kuiper’s A2 substrate, which survives only in a few odd sounds quite atypical for Indo-European, borrowed principally into Germanic, but also into a few Celtic words, including doubled final consonants and the word-initial kn-, as in ‘knob’ (see also Krahe 1954; Huld 1990; Polome 1990; Venneman 1994; Kuiper 1995). The now-extinct languages that produced these sounds competed with and influenced Pre-Germanic dialects for millennia, so Pre- Germanic did not evolve in isolation, and it cannot be assumed that Pre-Germanic replaced competing languages rapidly. The expansion of Pre-Germanic occurred with the expansion in prestige and power of chiefs who spoke it. My subject in this essay is just the initial separation of Pre- Germanic from Proto-Indo-European at the beginning of this long and complicated history. Of course at this initial stage it was only Pre-Germanic in hindsight. At the time, it was just a western Proto-Indo-European dialect that became established in northern Europe. The time and place of the homeland It is impossible to discuss the archaeological evidence related to the detachment of Pre-Germanic if the time and place of the Proto-Indo-European homeland is not established first. My views on this subject are defended at length in a recent book (Anthony 2007) and have been discussed elsewhere (Anthony 1991, 1995). I generally agree with Mallory (1989) and Mallory and Adams (2006); and in terms of location and general time-frame, with Gimbutas (1977, 1991). The homeland was in the Pontic-Caspian steppes between Volume 36, Number 1 & 2, Spring/Summer 2008 6 David W. Anthony about 4500-2500 calBC. I will briefly defend this argument, which in my view is now so strong that it can be accepted as the most probable. GERMANIC BALTIC SLAVIC *kwekwlos wheel *rot-eh- wheel 2 *hih s- th i ll 2 *aks- axle CC NNII *wegheti convey in a vehicle MMAA BBBAAALLLTTTIIICCC II RR AA NN II AA CCCEEELLLTTTIIICCC GGEERRCC SSSLLLAAAVVVIIICCC NN TTOO CC HH AARRIIAANN TOCHARIAN II TT CCEELL IITTAALLIICC RREEEEKK AAANNNAAATTTOOOLLLIIIAAANNN IIRRAA NN II AA NN DDIICC GG NN IRANIAN II CELTIC ITALIC GREEK I NaDndIC ANATOLIAN Fig. 1. The geographic distribution of the shared cognates for wheeled vehicles in the Indo-European languages. After Anthony 2007. On the question of time, the vocabulary for wheeled vehicles in Proto-Indo-European (Figure 1) indicates that it was spoken after wheeled vehicles were invented, or certainly after about 4000 calBC, and probably after about 3500 calBC (Bakker, Kruk, Lanting and Milisauskas1999). The appearance of separate and distinct Anatolian, Greek, and Old Indic daughters in inscriptions between 1900-1400 calBC indicates that Proto-Indo-European had broken up into its major daughter branches by about 2500 calBC (Mallory and Adams 2006; Anthony 2007). By that date the phonetics and grammar of the classic reconstructed parent had evolved into an intermediate, evolved set of distinct late Indo-European languages and dialects that were no longer sharing innovations. Proto-Indo-European probably was spoken for five to ten centuries between 4500-2500 calBC. The principal alternate hypothesis, an Anatolian homeland dated about 6500 calBC, is contradicted by the wheeled-vehicle vocabulary. The Anatolian homeland is The Journal of Indo-European Studies The Usatovo Culture and the Separation of Pre-Germanic 7 premised on a first radiation of Proto-Indo-European with the first farmers from Anatolia to Greece about 6700-6500 calBC and from Greece into temperate Europe about 6200-6000 calBC. But the presence of a vocabulary for wheeled vehicles in Proto-Indo-European suggests that Proto-Indo-European was spoken much later than this. Renfrew (2001) pulled the first- farmer dispersal hypothesis toward the wheeled vehicle vocabulary by suggesting that Proto-Indo-European remained a single language for 3500-3000 years, from 6500-3000 calBC. Early Proto-Indo-European, in this hypothesis, could have existed in 6500 calBC and late Proto-Indo-European could have been the medium through which the new wheeled- vehicle vocabulary spread from the Rhine to the Volga between 3500-3000 calBC. This is unlikely. It requires that Proto-Indo-European persisted as a unified dialect chain for 3000-3500 years between the initial pioneer Neolithic colonization of Greece and the invention of the wheel-and-axle principle. In the first millennium after 6500 calBC Greek Neolithic villages generated a dozen regional archaeological cultures spread across several distinct climate zones that interacted with a variety of indigenous European foragers speaking unrelated languages; and this was followed by 2000-2500 years during which those Early Neolithic cultures evolved into hundreds of very different later Neolithic and Eneolithic cultures—in this hypothesis, without developing distinct languages. What is worse, this frozen state during the 3500 years of the Neolithic and Eneolithic would have to have been followed by a much more rapid rate of language change during the Bronze and Iron Ages in order to account for the hundreds of Indo- European daughter languages, divided into twelve branches (or perhaps more), that had come into existence 3500 years after wheels were introduced, or by about 500 CE. The rapid real rate of diversification evident in the Indo- European languages in the 3500 years after the wagon was invented finds a real-world parallel in the spread of the Bantu languages. Proto-Bantu was spoken by cattle-herders who introduced a new pastoral economy and cattle-based prestige system across eastern and southern Africa about 2500-2000 years ago (Phillipson 2002). In just two millennia it evolved into more than 500 modern Bantu languages assigned to 19 branches interspersed with enclave languages still belonging Volume 36, Number 1 & 2, Spring/Summer 2008 8 David W. Anthony to non-Bantu language families. Similarly, the languages of the pioneer farmers who carried the first farming-and-herding economies into prehistoric temperate Europe would have evolved into a dense bush of daughter languages by 3500-3000 calBC. Even if the language of the original immigrant Greek Neolithic farmers in 6500 calBC began as a simplified, fairly homogenous dialect stripped of variation through the social processes typically associated with long-distance colonization (Anthony 2007), its daughters 3000 years later when wheels were invented would have included hundreds of languages, already divisible (if only a linguist had been there to classify them) into perhaps dozens of branches, interspersed with pre- Neolithic language families of profoundly different types— such as Schrijver’s language of bird-names. Proto-Indo- European simply cannot have been spoken in Neolithic Greece and remained frozen in its Greek-Neolithic proto-form for 3000 years until wagons were invented. Yet there can be no doubt that the wagon and wheel vocabulary existed in Proto-Indo-European before it broke up into its daughter branches. The wagon vocabulary contains at least five classic reconstructions based on cognates spread across the Indo-European-speaking world, including ancient Old Indic and Mycenaean Greek (but possibly excluding Anatolian). Wheeled vehicles appeared in four quite different media—written signs for ‘wagon’, two-dimensional images, three-dimensional models, and archaeological wagons in graves and bogs—after 4000 calBC. Certainly most of this evidence, arguably all of it, is dated after 3500 calBC (Bakker, Kruk, Lanting and Milisauskas1999). Proto-Indo-European still existed at this date. Therefore Proto-Indo-European did not spread with the farming economy. Its first dispersal occurred much later, certainly after 4000 calBC, probably after 3500 calBC, in a Europe occupied by many different kinds of farming and herding societies, not by a scattered population of foragers. This is not a minor difference of opinion over chronology. It means that the key to understanding the spread of early Proto-Indo-European should be found principally in the methods of socio-linguistics, including social anthropology, comparative religion, and the economics of political power—not principally in agronomy, genetics, and demographics. The social modeling of the process of Indo- The Journal of Indo-European Studies The Usatovo Culture and the Separation of Pre-Germanic 9 European language expansion remains underdeveloped (Mallory and Adams 2006: 458-59), perhaps because we have tried to rely on the more easily quantifiable methods of genetics and demographics rather than the qualitative methods of sociolinguistics, where the answers probably lie. To b ol UR AL Mts. Ui Kama Volga SSS RRR EEE GGG RRRAAA Samara OOO FFF DnieEEper DoneDPPPt o s RRRn OOO TT T OOO -- - III NNNHHH DDDOOOOOOMMM---EEEEEELLVLUUUoAAARRlRgNNNaOOO DDDPPP EEE AAA NNN Ural R. FFFEmOOObaRRRAAAGGGEEERRRSSS CARPDATnHiIeANst Merts.PSr.uBtugTTTRRIIPPOOLLYY A Z OSOFEVA MMM AAAIIIKKKOOO PPP CA S CAUCASUS Mts. PI A Danube BLACK SEA Kura N SE Araxes A LEGEND Halys 500 km Steppe boundary PIE homeland Fig. 2. The Proto-Indo-European homeland. After Anthony 2007. If Proto-Indo-European did not disperse with the first farming economies, how and from where did it disperse? The location of the homeland is determined best by borrowings Volume 36, Number 1 & 2, Spring/Summer 2008 10 David W. Anthony between Proto-Indo-European and its neighbors (Figure 2). Proto-Uralic borrowed Proto-Indo-European roots for the words water, give or sell, price, bring or lead (possibly meaning marry), plait, drill, fear, wash, and sinew; and the two language families exhibit a similar, possibly shared inherited vocabulary for pronouns (see several papers in Carpelan, Parpola and Koskikallio 2001). These borrowings indicate that the two language families shared a border or perhaps even a distant common ancestor, which pulls the Proto-Indo-European homeland toward a broad region centered on the Ural Mountains. The presence of a farming and herding vocabulary in Proto-Indo-European and the end date of 2500 calBC together eliminate a homeland east or northwest of the Urals, where foraging economies persisted in northern Kazakhstan, western Siberia, and the western-Ural Russian forest zone until after 2500 calBC. The combination of Proto-Uralic/Proto-Indo- European borrowings, time restriction to a period before 2500 calBC, and the farming vocabulary in Proto-Indo-European together make a homeland southwest of the Urals most likely. Indirect borrowings between Proto-Indo-European and a language ancestral to Proto-Kartvelian, possibly through a third intermediary language, also pull the homeland toward the southwest, toward the Caucasus Mountains, although this relationship is weaker than the Uralic one (Nichols 1997: Appendix). A homeland between the Urals and the Caucasus, in the Pontic-Caspian steppes of Russia and Ukraine between 4500-2500 calBC, satisfies all these requirements. Sequence of branch separations The sequence of separations among the Indo-European daughters (Figure 3) provides the final and most demanding test for any proposed homeland. All linguistic studies of sequences, whether by cladistics or more traditional means, agree that Pre-Anatolian separated first (Blazek 2007; Ringe, Warnow, and Taylor 2002; Gray and Atkinson 2003). Both cladistic (Ringe, Warnow and Taylor 2002) and traditional (Starostin as presented in Blazek 2007) methods agree that Pre-Tocharian separated next, though it also showed some traits that might be considered later, particularly in its similarities to some aspects of Germanic. The next branching event separated Pre-Celtic and probably Pre-Italic from the still-evolving core. Germanic has some archaic traits that suggest an initial separation at about the same time as Pre- The Journal of Indo-European Studies