The words you are reading are shrapnel from a five-thousand-year-old cultural explosion. Archaeologists and linguists now think they know where it erupted and who set it off.
David W. Anthony
David W. Anthony is an associate professor of anthropology at Hartwick College in Oneonta, New York, and director of the Institute for Ancient Equestrian Studies.
In June 1994, after centuries of trading words with England, France closed its linguistic borders. Long the lingua franca of diplomats, French had been usurped by English in this century and then, most gallingly, had been forced to collaborate with its conqueror. As American movies, fast-food chains and scientific journals inundated France. Hundreds of blunt Americanisms made their way into delicate French mouths, signaling a "process of collective self-destruction," in the words of a statement signed by 300 French intellectuals. Then came la Résistance. According to a law passed in 1994, terms such as le microchip, le fast-food and le talk show are never again to appear in French publications. Car manufacturers are to replace les air bags with les sacs gonflables. Businessmen, instead of discussing le cash flow while flying in le jumbo jet, are urged to discuss la marge brute d'autofinancement while in le gros porteur.
To many Americans, accustomed to the hybrid vigor of their own language, the French law seems futile, if not faintly comical. By maintaining that their language is under attack from "franglais," the government implies that an older, purer French exists. Yet a brief review of linguistic history should only deepen their concerns. Modern French shares its source in Latin with more than ten other languages. Latin, in turn, is only one of myriad languages that sprang from a single, common linguistic stock: a language spoken more than 5,000 years ago, known as Proto-Indo-European (PIE).
In an age of multiculturalism and identity politics, Proto-Indo-European offers a welcome reminder of common origins. Ethnic roots, so often synonymous with ethnic differences, tend to converge in linguistic history: the deeper you dig, the closer they grow. The mother tongue itself has long remained beyond reach. But two centuries of scholarship is finally providing firm answers to some of history's most intriguing questions: Who were the speakers of PIE? Where did they live? What was their language? And how did its regional dialects become the dominant languages in so much of Europe and Asia?
Archaeological discoveries in Ukraine and Russia, together with new research methods in genetics and linguistics, now point independently to a region that has been a perennial candidate for the Indo-European homeland. If those findings are correct, the speakers of PIE lived east of the Dnieper River, in the grasslands of the Eurasian steppes. By introducing horseback riding and wheeled vehicles in the region, they revolutionized a pastoral economy. That revolution, I believe, set off a linguistic explosion that continues to expand with every century, echoing through voices from Scotland to China, shaping English and French (and Russian and Ukranian) and the words that pass from one to the other.
The Indo-European problem was first defined by William Jones, a justice in the high court of Calcutta. Jones's appointment, in 1783, signaled Calcutta's transition from a merchant colony to a seat of the British government. It did little, however, to dispel the city's reputation in England as a place of mythic exoticism. Fascinated with the complexities of Hindu law, Jones became a student of Sanskrit, an archaic language used throughout India in Hindu religious and legal texts. For three years he pored over Sanskrit texts, the oldest of which—the Rig-Veda—included hymns, accounts of rituals and stories of the heroic traditions of a people who called themselves Aryans. Then, in 1786, while presenting a paper at a meeting of the Asiatic Society of Bengal, Jones uttered the sentence now quoted in every introduction to historical comparative linguistics:
The Sanskrit language, whatever be its antiquity, is of a wonderful structure; more perfect than the Greek, more copious than the Latin, and more exquisitely refined than either, yet bearing to both of them a stronger affinity, both in the roots of verbs and in the forms of grammar, than could possibly have been produced by accident; so strong indeed, that no philologer could examine them all three, without believing them to have sprung from some common source, which, perhaps, no longer exists.
An accomplished linguist, familiar with Latin, Greek, Welsh and Gothic (an early Germanic language), Jones had put his finger on a truth wrapped in an enigma.
In the centuries since Jones's revelation, the search for his "common source" has proceeded in two disciplines: comparative linguistics and archaeology. Bit by bit, linguists have pieced together clues buried in ancient documents and in modern languages, defining the nature of the relations between the Indo-European languages with increasing precision. The Indo-European language family today comprises most of the languages of Europe, including Albanian, Bulgarian, Czech, Danish, Dutch, English, French, Gaelic, German, Greek, Italian, Latvian, Lithuanian, Norwegian, Polish, Portuguese, Romanian, Russian, SerboCroatian, Spanish, Swedish and Ukrainian (but not including Basque, Finnish, Hungarian or Turkish). In addition, Indo-European encompasses Armenian, Kashmiri, Kurdish, Persian and numerous tongues of India, including Bengali, Hindi and Urdu. Its extinct branches reach as far as Asia Minor and Syria, home of the Hittites, and into northwestern China, where Tocharian texts were discovered in the ruined caravan cities of Xinjiang. Those languages have been shown to share grammatical constructions and lexical roots that reveal their common ancestry.
Yet intriguing as it is, the linguistic evidence has long floated uncertainly in time and space, waiting for archaeologists to moor it to specific graves, settlements and material remains. Unfortunately, many archaeologists consider the Indo-European problem poisoned by propaganda. The Nazi myth of a Germanic "Aryan" superrace was based partly on the research of the German archaeologist Gustaf Kossinna, who argued in the 1920s that the Indo-European homeland was located in modern-day Germany. Today equally fanciful assertions are made in Russia and other Eurasian countries to justify racist beliefs and territorial demands.
Archaeologists not put off by the propaganda surrounding the Indo-European problem have split into three camps: those who distrust important parts of the evidence linguists have uncovered, particularly their ability to reconstruct the proto-language; those who root their theories in it; and those who dismiss the entire Indo-European problem as unsolvable.
The first camp is represented by the archaeologist A. Colin Renfrew of the University of Cambridge. In 1977 Renfrew proposed a scenario that remains popular among archaeologists: The speakers of PIE were pioneer farmers who lived in Anatolia (modern Turkey). Sometime before 6000 B.C. they moved northwest into Greece and the Balkans, and from there east into Romania, Ukraine and Russia, where they first established farms. Renfrew links the Indo-European linguistic expansion with a well-documented demographic and economic expansion from Anatolia into Greece and the Balkans. By placing the Indo-European homeland in Turkey, he also helps explain why certain words in Proto-Indo-European sound similar to words in both the Semitic languages of the Near East (Arabic, Hebrew and the like) and the non-Indo-European languages of the Caucasus (Chechen and Georgian, for example). His scenario is clean and simple—and therefore very appealing. It also suffers from a serious weakness.
To agree with Renfrew, archaeologists must dismiss most of what linguists have learned about the PIE lexicon in the past 200 years. That archaeologists have been willing to do so in great numbers demonstrates just how far apart archaeology and linguistics have drifted in recent years. Comparative linguistics is unreliable, archaeologists have charged. It relies too much on linear family trees, ignoring the ways languages can borrow, converge or create creolized versions of one another.
As an archaeologist of the second camp, I trust linguistic findings but not the way they are sometimes used. Some archaeologists have assumed that because PIE was a single language its speakers must have belonged to a single ethnic group. They have then associated telltale artifacts with that group—battle-axes, for instance—and have designated all sites that contain them as Indo-European. Such assumptions, though demonstrably false, cannot be blamed on the comparative method in linguistics, which is unconcerned with ethnicity. On its own terms, the comparative method has a proven record of success. It has predicted aspects of archaic languages (the sound of w and k pronounced simultaneously in Mycenaean Greek, for instance) well before their documentation in ancient inscriptions.
When comparative linguists try to explain changes in a language, they first look for unstable phonetic structures within the language. Only if that conservative approach does not work will they suggest that certain words, sounds or grammatical forms might have been borrowed from other languages. Although archaeologists, in their own work, are equally skeptical of external agents of change, they often take similarities between languages as evidence of borrowing. Nevertheless, no amount of linguistic borrowing or convergence has ever given rise to similarities as striking as the ones between Indo-European languages.
Languages are shaped by culture and by biology. Cultures invent and reinvent their own vocabularies, grammar and syntax, whereas the mechanics of speech set limits on cultural creativity. But words, once invented, tend to evolve in predictable ways. For example, p's often turn into f's (the Greek pyr became the English fire), but f's less often become p's.
When two words with a similar sound and meaning do appear in two or more languages, they may share a root, in which case they are known as cognates. The English word horn and the French word corne, for instance, each evolved from the PIE root word *krn-.(An asterisk preceding a linguistic form indicates that the form is inferred but unattested in any manuscript materials.) Or two such similar words may simply be the same word, recently introduced into both languages (café and coffee both are versions of the Turkish kahveh). Separating cognates from borrowed words is a principal goal of the comparative method in linguistics. To do it, one must know how a particular sound in one language should sound in a particular foreign cognate. Fortunately, the rules of sound change are quite regular. If the word coffee had come from a PIE root, its French cognate might be chef or cief—but certainly not café. Given the rules of sound change in French and English, no PIE word of any phonetic shape could have evolved into both coffee and café.
Linguists have compared thousands of cognates, reconstructing important pieces of PIE vocabulary and grammar. The PIE root word *kmtom, for example, was reconstructed from several cognates that mean "a hundred": simtas in Lithuanian, centum in Latin and satem in Avestan, an ancient Iranian language. Like pieces of a crossword puzzle in three dimensions, with words radiating from their every letter in every direction, PIE words have to satisfy several arcane criteria at once. For the k in *kmtom to be accurate, it must, according the rules of evolutionary sound change, lead to the first letters of numerous cognates spoken in far-flung countries. The longer the proto-word, the more exacting the proof. Only random pieces of the proto-language emerge via that method, but the authenticity of the ones that do is thought to be highly reliable.
Reconstructed PIE is a priceless treasure, providing a window into the religion, kinship and descent systems, technology, social structure and economy of its speakers. To begin with, the very fact that many PIE words and grammatical rules can be unearthed suggests that PIE was a single language, perhaps with regional dialects, rather than a group of languages. Particular word clusters, in turn, reveal rituals and beliefs beyond the reach of archaeological evidence alone. They show that the speakers of PIE probably practiced patrilineal descent and patrilocal (husband's family) residence, recognized the authority of chiefs who were associated with a residential-kin group, had formally instituted warrior bands, practiced ritual sacrifices of cattle and horses, drove wagons, recognized a male or father sky deity, and avoided speaking the name of the bear for ritualistic reasons. They even demonstrate two senses of the sacred: "that which is imbued with holiness" and "that which is forbidden."
Some of the more mundane words have enabled linguists to circle in on the Indo-European homeland. For example, the speakers of PIE probably did not live in Greece or Turkey: The linguistic evidence makes it clear that they were familiar with horses, and wild horses were either rare or absent from that region, depending on how one reads the archaeozoological evidence. The speakers of PIE had words for otter, beaver, birch and aspen, and they shared euphemisms for bear, so they must have lived in a temperate environment. (On the basis of the cognates for names of trees alone, the American linguist and anthropologist Paul Friedrich suggested the PIE homeland was in what are now Ukraine and western Russia.) Finally, the Uralic language family (which gave rise to modern Finnish and Hungarian) borrowed words from early Indo-European languages, and the Caucasian language families borrowed words from PIE itself, so the speakers of those languages were probably neighbors.
Viewed in an archaeological context, some PIE words also hint at when their speakers lived. Terms for domesticated sheep, pigs and cattle suggest that the speakers of PIE lived after 6000 B.C., when the earliest Neolithic economies were established in the temperate zone. Terms for wheel, axle and draft pole, and a verb meaning "to go or convey in a vehicle" suggest that PIE existed as a single language after 3500 B.C., when wheeled vehicles were invented. Proto-Indo-European must have begun to disintegrate before 2000 B.C.: by 1500 B.C. three of its daughter languages—Greek, Hittite and Indic—had become quite dissimilar. Altogether, then, the linguistic evidence points to a homeland somewhere between the Ural and the Caucasus mountains, in what are now Russia and Ukraine, in the centuries between 3500 and 2000 B.C.
West of the Urals and north-west of the Caucasus the Dnieper River runs through the center of Ukraine and spills into the Black Sea. On the western bank of that river, near the modern village of Dereivka, people of the Sredni Stog culture built a Copper Age hamlet and cemetery between 4200 and 3800 B.C. They lived in buildings probably made of reed and timber; cooked stews in rough, plain clay pots; buried their dead in formal cemeteries near settlements; and decorated their bodies with pendants of boar's tusk and strings of beads made from shells and animal teeth. The few copper ornaments they had were imported from the west, probably by traders from the sophisticated Tripolye culture across the river.
People at the Dereivka site herded cattle, sheep and pigs, caught fish, and hunted deer and wild horses, relying on the latter for most of their meat. In the case of one horse, however, they made an exception: its hide, with its head and the bones of one foreleg attached, was ritually buried at the edge of the settlement along with the heads and pelts of two dogs. The horse's premolar teeth bear the unmistakable signs of bit wear (90 percent of modern horses that are frequently bitted show bit wear). Since 1987 I have studied bit wear in ancient and modern horses with Dorcas R. Brown, my wife and colleague at Hartwick College in Oneonta, New York. We have noted the relative effects of wear from metal, bone, rope and leather bits. The horse at the Dereivka site, a stallion between seven and eight years old, was bitted with a hard bit, probably bone, for at least 300 hours. Because it lived around 4000 B.C.—about 500 years before the wheel was invented—the horse was probably ridden. If so, it is the earliest direct evidence of the practice of horseback riding anywhere in the world.
Nowhere would learning to ride a horse have been more useful than in ancient Eurasia. At the center of that continent, the dry steppe extends 3,000 miles, from the mouth of the Danube to Mongolia. Steppe dwellers began to carve a secure, productive life from that desolate landscape sometime before 5000 B.C., when they learned to domesticate sheep and cattle. Horseback riding, practiced at Dereivka 1,000 years later, dramatically improved the pastoral life. It enabled scouts to search for good pasture at great distances, increased long-distance trade, expedited large-scale herding and conferred an important military advantage. When wagons arrived on the steppe between 3500 and 3300 B.C., the pastoral life completed its transformation. Herders could now pack up their tents and supplies, leaving their river-valley crops behind to follow their herds through the steppe for a season.
That package—herding, riding and wagon driving—was first put together by the Yamna culture. Descendants of the Sredni Stog people at Dereivka and similar groups on the Don and Volga rivers to the east, the Yamna people lived between 3500 and 2500 B.C. Their cemeteries, made up of between four and twenty-five kurgans, or low burial mounds, are scattered across the vast steppes north of the Caspian and Black seas, between the middle Ural River (around Orenburg) in the east and the lower Prut River in the west. In the west, Yamna kurgans were built on the ruins of Tripolye towns—once the largest human settlements in the world—that were abandoned around the time the Yamna culture appeared. The core of the Yamna area lay between the Dnieper and the middle and lower parts of the Volga.
The Yamna people introduced a new way of life to the steppe. They were the first to build some of their cemeteries in the deep steppe, far from permanent water sources—probably to mark off their distant pastures. They were the first to extensively exploit steppe copper ores and to hunt the saiga, an antelope of the deep steppe. Finally, they were the first to possess wagons and carts on the steppe, valuing them so highly that they sometimes buried them with their owners—perhaps to carry the soul toward distant pastures in the afterlife. Some 248 wagon or cart graves have been unearthed in the steppes between the Ural and Prut rivers, the earliest dating to around 3000 B.C.
Neither an empire nor a unified polity, the Yamna culture was still powerfully influential. The societies that adopted its way of life in the Dnieper-Volga steppes between 3500 and 3000 B.C. seem to have had common rituals—as their graves demonstrate—and they probably had a common language. That language, in all likelihood, was Proto-Indo-European.
According to a scenario first proposed by the American archaeologist Marija Gimbutas and later revised by the archaeologist James P. Mallory of Queen's University in Belfast, Ireland, and by me, the Yamna culture was the first to carry PIE into Europe. Cultures descended from it then carried related tongues across the steppes to Iran and India. Equipped with horses and wagons, the Yamna people and their descendants used ideology, alliances and trade to dominate their neighbors or exploit divisions among them, transforming Eurasian pastoralism as they went. The Yamna region may have included the entire language family to which PIE belonged. In that case, PIE was a local language that gained high status and then gradually displaced its sister languages in the Dnieper-Ural steppes.
The Yamna people lived in the right place at the right time to plant the clues later discovered in the PIE lexicon. Unlike Renfrew's Anatolia, their homeland had a temperate climate, and it lay between the Ural and the Caucasus mountains. It was home to otter, beaver, bear, birch and aspen, and its location would explain the early ties between Indo-European languages, Uralic languages and the languages of the Caucasus. Our scenario, by setting Indo-European origins three millennia later than Renfrew does, would also explain how words for wheeled vehicles entered into PIE.
Renfrew argues that PIE never had terms for wheeled vehicles because it dispersed long before they were invented. Later Indo-European languages borrowed those terms from one another or from other languages, he maintains, when wheeled vehicles reached their cultures. For that to be true, Indo-European languages from India to Scotland would have had to pass along the same set of terms for wheeled vehicles that a single culture invented.
No archaeological evidence supports that idea, and much contradicts it. Wheeled vehicles appeared almost simultaneously in eastern Europe, the steppes and the Near East after 3500 B.C. No one knows where, when or by whom they were invented, but their spread, understandably, was quite rapid. The Indo-European words describing them fit seamlessly into their languages. The idea that terms describing them were borrowed between languages has never been supported by a linguistic study. On the other hand, numerous published studies have shown those terms to be true cognates derived from PIE. The speakers of PIE, the evidence shows, were clearly familiar with wheeled vehicles.
Linguists group Indo-European languages into at least nine sub-families—Albanian, Anatolian, Balto-Slavic, Celtic, Germanic (which includes English), Greek-Armenian, Italic, Indo-Iranian, Tocharian and perhaps Thracian—but they have long disagreed on when each subfamily split from the mother tongue. Recently, however, the linguists Ann D. Taylor and Don Ringe, and the information science specialist Tandy J. Warnow, all of the University of Pennsylvania in Philadelphia, have created a robust evolutionary tree for Indo-European. Using a modified form of cladistics (a mathematical method used to define relations between biological species), they have shown that pre-Celtic and pre-Italic separated from the Indo-European core quite early. Pre-Tocharian, pre-Greek-Armenian and pre-Indo-Iranian followed next. Pre-Germanic probably began to separate from Balto-Slavic around the time the speakers of pre-Greek-Armenian became isolated. After the speakers of pre-Germanic became fully separated from Balto-Slavic, they established contact with and began to borrow numerous words from the speakers of pre-Italic and pre-Celtic. Finally, pre-Baltic and pre-Slavic developed into two distinct languages.
The Yamna region gave rise to two migrations that help explain the shape of that evolutionary tree. The first migration, inferred on the basis of well-dated archaeological sites and vestigial genetic patterns in European populations, flowed from the western steppes into the lower Danube Valley, the Balkans and eastern Hungary between 2900 and 2700 B.C. Pockets of Yamna migrants seem to have coexisted, at least at first, with indigenous societies. Seventeen Yamna cemeteries have been mapped in Bulgaria, each including between five and twenty kurgans.
They continue up the Danube Valley and into the eastern part of Hungary, where one specialist has counted 3,000 kurgans. Kurgans in Bulgaria and eastern Hungary match those in the steppe in every detail: the occupants are laid on their backs with knees raised, their heads to the northeast; daggers and silver temple rings serve as grave gifts; mats painted with red and white stripes or other geometric forms lie on the grave floor or grave cover; red ocher or hematite lumps are placed at the body's hip or shoulder; and the grave pits are covered in timber and topped with carved stone stelae (broken and undecorated in Hungary, decorated in Bulgaria like the ones in the steppes). In Bulgaria one kurgan contains a wagon. In eastern Hungary the immigrants had to import hematite for their graves. When hematite was not available, they laid a lump of dirt next to the occupant's shoulder and painted it red. They went to great lengths, in other words, to cling to Yamna traditions.
The Carpathian Basin went on to become a crossroads for interregional trade, powerfully influencing the cultures of central and western Europe during the Bronze Age. The proto-Celtic languages, in particular, are widely thought to have been associated with the Hallstatt culture, which evolved in the mountains west of Hungary. The first Yamna migration took place at the right time and place to spawn Italic as well.
The second migration out of the Yamna region took place between 2200 and 2000 B.C., and it flowed east from the Volga-Ural region into the steppes east of the Ural Mountains. The Poltavka culture, which gave rise to it, grew directly from late Yamna at the northeastern boundary of the Yamna world. (I am now investigating Yamna and Poltavka sites in the lower Samara River Valley with support from the National Geographic Society.) East of the Urals, the migrants interacted with people of a different tradition, establishing the remarkable Sintashta culture. The Sintashta people built compact, fortified settlements, forged weapons and ornaments in bronze and buried chariots in kurgans with their dead warriors or chiefs. Aside from those burials, traces of Yamna and Poltavka ancestry can be seen in Sintashta weapons, pottery and head-and-hoof horse sacrifices.
Like the small figurine nesting within a wooden babushka, Sintashta gave shape to ever expanding cultures. First it joined with the related Petrovka complex to engender the Andronovo cultural horizon, which spread across the eastern steppes to the Tien Shan and the borders of Iran and Afghanistan. Andronovo helped shape the Indo-Iranian culture which, in turn, led to a Persian culture that retained some traces of Sintashta, including chariotry, horse sacrifices and the myth of the horse-headed human.
With Indo-Iranian, the path to PIE comes full circle. Sanskrit is also a daughter tongue of Indo-Iranian, so similar to Old Persian that it could have sprung only from the same source. Like the Persians, the Aryans kept some Sintashta customs alive—elaborate mortuary rituals and horse and cattle sacrifices, for example. But they considered Sanskrit too sacred to set it down in writing. The hymns and rituals of the Rig-Veda, with their gambling heroes, chariot races and paeans to plump cattle, were passed on orally for centuries, even after Sanskrit passed out of daily use. Transcribed, at last, and dispersed throughout India, they made their way to Calcutta and the curious mind of William Jones. Five thousand years after disappearing on the steppe, the mother tongue had begun to be found.
Reprinted with permission from The Sciences, January/February 1996, pp. 34–39. © 1996 by the New York Academy of Sciences. Individual subscriptions are $21.00 per year in the U.S. Write to: The Sciences, 2 East 63rd Street, New York, NY 10021.