“What's in a name?” asked Juliet of Romeo. “That which we call a rose by any other name would smell as sweet.” A real-life Juliet probably would have spoken to Romeo in an obscure medieval Italian dialect rather than Shakespeare's English. Nevertheless, her word for the sweet-smelling flower would have shared the same linguistic root (rosa, in modern Italian) as the English version does and indeed as many other languages spoken throughout Europe do—Rose, capitalized in German fashion, or the lowercase French rose. Croatian? An aromatic rua. To the nearly 60,000 Scots who still speak the ancient Scottish Gaelic, this symbol of passionate love is a ròs.
Why do such geographically diverse languages use similar words for the same flower? All these tongues, along with more than 400 others, belong to the same family of languages—the incredibly far-flung Indo-European language family—and have a common origin. Indo-European languages, which include Greek, Latin, English, Sanskrit, and many languages spoken in Iran and on the Indian subcontinent, are the most dominant linguistic group in the history of humanity. They account for about 7 percent of the world's estimated 6,500 languages but are nonetheless spoken by three billion people—nearly half the world's population. Understanding how, why and when they spread so readily is key to understanding the social, cultural and demographic changes that created today's diverse populations in Europe and much of Asia. As Paul Heggarty, a linguist at the Max Planck Institute for the Science of Human History in Jena, Germany, puts it: “We have to explain why Indo-European was so outrageously, overpoweringly successful.”
Because words and languages do not fossilize, the task of tracking their movements across time and space was left for more than a century to traditional linguists and a small number of archaeologists. Recently, however, the search for Indo-European origins has gone high tech, as biologists and experts in ancient DNA have gotten into the act. Armed with new theoretical and statistical approaches, these investigators have begun to transform linguistics from a paper-and-pencil exercise into a field that uses powerful computers and methods borrowed from evolutionary biology to trace language origins.
You might think that this attempt to modernize linguistics would bring researchers closer to an understanding of where and when the Indo-European languages arose. But in many ways, the opposite has happened, and the question is in even greater dispute. Everyone agrees on one key point: the Indo-European languages descend from a common ancestor, a mother tongue called Proto-Indo-European, or PIE. But as to why this particular language produced so many linguistic offspring or where it originated, there is no accord.
Researchers have fallen into two warring camps. One camp, which includes the majority of traditional linguists, argues that Central Asian nomads, who invented the wheel and domesticated the horse, spread the mother tongue throughout Europe and Asia beginning about 6,000 years ago. The other camp, led by British archaeologist Colin Renfrew, credits early farmers from more than 500 miles to the south in what is now Turkey with disseminating the language at some point after they began spreading their agricultural know-how 8,500 years ago.
Over the years first one idea then another has had the upper hand. Evolutionary biologists published a series of studies in 2003 that concluded that the Indo-European family tree originated in the Middle East at least 8,000 years ago, based on the idea that the evolution of words can mimic the evolution of living organisms; their results are consistent with the farmer hypothesis. In the past year or two some linguists, archaeologists and geneticists struck back, using rival computational analyses and samples of DNA from ancient skeletons to support the nomad hypothesis. And so the pendulum continues to swing.
The Horse, of Course
Scholars did not have to wait for high-speed computers to recognize connections among the Indo-European languages. That realization dawned as early as the 1700s, after Europeans had begun to travel far afield. Some of the parallels among widely distributed tongues are now seen as dead giveaways. Thus, the Sanskrit and Latin words for “fire,” agní- and ignis, clearly indicate their Indo-European family ties.
By the 19th century, linguists were sure there must be a common ancestor for all Indo-European languages. “There was a sense of shock that the classical languages of European civilization sprang from the same source as Sanskrit, an exotic language spoken in India, on the other side of the world,” says David Anthony, an archaeologist at Hartwick College and a fierce advocate of PIE's nomadic origin.
So linguists set about reconstructing this ancestral tongue. Sometimes this was not too difficult, especially if the original word had not changed unrecognizably. For example, linguists could take the English word “birch,” the German Birke, the Sanskrit bhrjá and other Indo-European words for this slender tree and, by applying basic linguistic rules of language change, extrapolate backward to figure out that the PIE root was something like *bherh1- (the asterisk indicates that this is a reconstructed word for which there is no direct evidence). Other reconstructions are not as obvious. Thus, the PIE word for “horse”—áva- in Sanskrit, híppos in Greek, equus in Latin and ech in Old Irish—was determined to be *h1éwo (the subscript 1 refers to a sound made in the back of the mouth).
But when some linguists tried to identify the peoples behind the language, things became trickier. These scholars began linking certain cultures with PIE, an approach called linguistic paleontology. They noticed that PIE contained many terms for domesticated animals, such as horses, sheep and cattle, and began postulating a pastoral Indo-European “homeland.”
That approach eventually led to trouble. In the early 20th century German prehistorian Gustaf Kossinna proposed that a group of Central European settlers, who created intricately engraved pottery called Corded Ware starting 5,000 years ago, were in fact the first Indo-Europeans. Kossinna argued that they later spread out of what is today Germany, carrying their language with them. That idea was music to the ears of the Nazis, who resurrected the term “Aryan” (a 19th-century term for Indo-Europeans), along with its connotations of racial superiority.
The Nazi endorsement gave Indo-European studies a bad name for many years. Many researchers give credit to Marija Gimbutas, an archaeologist who died in 1994, for making the subject respectable again, starting in the 1950s. Gimbutas situated the origins of PIE in the so-called Pontic steppes north of the Black Sea. For her, the prime mover of PIE was the Copper Age Kurgan culture, which can first be identified in the archaeological record about 6,000 years ago. After a millennium of roaming the barren steppes—in which the nomads learned how to domesticate the horse—Gimbutas argued, they charged forth into Eastern and Central Europe, imposing their patriarchal culture as well as the strongly enunciated vowels and consonants of their native Indo-European language. More specifically, Gimbutas identified the Yamnaya people, who lived in the Pontic steppes between about 5,600 and 4,300 years ago, as the original PIE speakers.
Other researchers also found evidence to support such a view. In 1989 David Anthony began working in Russia, Ukraine and Kazakhstan, focusing on horse teeth that had been earlier excavated by Soviet archaeologists. Anthony and his colleagues confirmed previous suggestions that there was bit wear on teeth dated as early as 6,000 years ago, pushing back the earliest evidence for horse domestication—and horse riding—by about 2,000 years. Their studies also provided evidence to link several technological developments—including the use of wheeled vehicles such as chariots—to the Yamnaya people. These finds supported the idea that the steppe pastoralists had the necessary transportation and technology to fan out rapidly from their homeland and spread their language in all directions.
The steppe hypothesis, also known as the kurgan hypothesis, after the kurgans, or burial mounds, in which the pastoralists buried their chiefs, was rarely questioned until the 1980s. Then Renfrew put forth a radically different idea, called the Anatolian hypothesis. (Anatolia, from the Greek for “sunrise,” refers to present-day Turkey.) Renfrew, the dean of British prehistorians, who now sits in the House of Lords, had spent years digging in Greece and was struck by how much the artifacts he unearthed, especially the carved female figurines, resembled those from earlier archaeological sites in Turkey and the Middle East.
Archaeologists already knew that farming spread from the Middle East to Greece first. Renfrew wondered if there might be a continuity of language in addition to culture. Thus, the first PIE speakers, he posited in lectures and a book, might be the farmers who moved from Anatolia to Europe 8,500 years ago, bringing their words along with their agricultural practices.
Traditional linguists, who had spent decades working painstakingly with paper and pencil to reconstruct PIE by tracing modern Indo-European words back to their original roots, were outraged. Most dismissed the Anatolian hypothesis, sometimes with bitter invective. One University of Oxford professor called the idea “rubbish,” and another skeptic declared that “a naive reader would be grossly misled by the simplistic solutions that the author offers.”
Renfrew and his supporters fought back, arguing that the steppe hypothesis cannot explain the broad expansion of PIE from wherever it began across both Europe and Asia. Researchers know that PIE-derived languages were spoken as far west as Ireland and as far east as the Tarim Basin, in what is now northwestern China, and down into India. A key question is how PIE would have gotten from the steppes to East Asia if the kurgan hypothesis were right. Did it spread to the north around the Black and Caspian seas, as in the steppe hypothesis? Renfrew sees no archaeological evidence for this route. Or did PIE take a southerly and earlier path to the east from Anatolia? He thinks it more likely that PIE spread south around the Black Sea from Turkey and then along early trade routes through Iran and Afghanistan.
Thus, Renfrew believes, only an Anatolian origin can account for PIE's simultaneous spread to the east and west because the peninsula offers the best historical evidence of movement between the European and Asian continents. And the only sociotechnological driver powerful enough to propel the language so far in opposite directions, he adds, was the advent of agriculture, which appeared in the Fertile Crescent—just south and east of modern Turkey—roughly 11,000 years ago. This transition of human society from hunter-gatherers into settled farming communities marked the so-called Neolithic Revolution and was “the only big thing that happened on a Europe-wide basis,” Renfrew says. “If you wanted a simple theory for the coming of the Indo-European languages, the Neolithic was the best thing to hang it on.”
Emblematic of the linguists' objections to Renfrew's Anatolian hypothesis is the origins of the word “wheel.” The reconstructed PIE root is *kwékwlo-, which became cakrá- in Sanskrit, kúklos in Greek and kukäl in Tocharian A, an extinct Indo-European language of the Tarim Basin. The earliest evidence for wheeled vehicles—depictions on tablets from ancient Mesopotamia (modern-day Iraq)—dates to about 5,500 years ago. The actual remains of wagons and carts show up in kurgans beginning about 5,000 years ago.
Many linguists have argued that the PIE root for “wheel” could not have arisen until the object was invented, and so PIE cannot be much earlier than 5,500 years—or about 5,000 years after the invention of agriculture. “That doesn't mean the PIE speakers invented wheels,” Anthony says, “but it does mean that they adopted their own words for the various parts of wheeled vehicles.”
But Renfrew and others counter that the word *kwékwlo- derives from a much earlier root meaning “to turn” or “to roll” and only later was adapted as a name for the wheel. “There was a whole language about rotation before the wheel was invented,” Renfrew says.
Andrew Garrett, a linguist at the University of California, Berkeley, and proponent of the steppe hypothesis, agrees that the PIE word for “wheel” has an earlier derivation, the root *k
Renfrew's Anatolian hypothesis was facing an uphill battle when, in 2003, a bombshell landed in the middle of the debate from an entirely unexpected direction—the field of evolutionary biology. Russell D. Gray, a biologist who had made his early reputation studying bird cognition, and Quentin D. Atkinson, then his graduate student at the University of Auckland in New Zealand, used state-of-the-art methods from computational biology to date the origins of PIE. Gray and Atkinson adapted an earlier linguistic technique called glottochronology, which compared the proportion of cognates—words with shared roots—in different languages to determine how long ago they diverged. Glottochronology had long been out of favor because it required linguists to assume that words change their form steadily over time—something they knew was not true. Gray and Atkinson employed a new and improved version of glottochronology, along with other statistical techniques used to determine the evolutionary trees of living organisms. Their database included cognates from 87 Indo-European languages, including Hittite, an extinct language that had been spoken in Anatolia.
The results were a slam-dunk for the Anatolian hypothesis. No matter how the pair crunched the numbers, the divergence of Indo-European languages from PIE came out no later than about 8,000 years ago—or nearly 3,000 years before the apparent invention of the wheel. Despite howls of objection from some linguists that words do not change the way that living organisms and genes do, the paper was highly influential and gave a big boost to the Anatolian hypothesis. Gray, who is now co-director of the Max Planck Institute for the Science of Human History (where Heggarty works), says that he and Atkinson were simply bringing linguistics into the 21st century.
Moreover, although Gray and Atkinson found that the initial spread of PIE tracked the spread of farming, they also detected a second divergence 6,500 years ago, which led to the Romance, Celtic and Balto-Slavic languages. The Anatolian and steppe hypotheses “need not be mutually exclusive,” they concluded.
Indeed, subsequent analyses favored the Anatolian hypothesis so strongly that some younger linguists began to call for older linguists to drop their objections. “Traditional linguistic objections to the Anatolian hypothesis are now wearing a little thin,” Heggarty wrote in a June 2014 commentary in Antiquity.
Calls for advocates of the steppe hypothesis to surrender, however, may have been premature. Beginning in 2013, Garrett and Chang launched their own analysis, using Gray's methodology. But the Berkeley researchers made an assumption that Gray's team did not: They “constrained” certain languages to be ancestral to their descendants, based on what they insist is solid historical evidence. Thus, they assumed, for example, that Classical Latin was directly ancestral to Romance languages such as Spanish, French and Italian. Gray and Atkinson, in contrast, allowed for the possibility that some as yet unidentified form of popular Latin spoken in the streets of Roman cities was the true ancestor of the Romance languages.
Garrett and Chang's results, published last year in Language, were also a slam-dunk—but for the steppe hypothesis, not the Anatolian hypothesis. Despite this apparent new life for the steppe hypothesis, Heggarty argues that Garrett's team is wrong to assume that some ancient languages are directly ancestral to others. Even small differences in Classical versus “Vulgar” Latin could throw off Garrett's estimates, Heggarty argues.
Garrett remains unconvinced. “For many of these languages we know quite a bit about the speech communities and the history of the languages,” he says. “Best understood are Greek and Latin. It isn't likely that there are other varieties of Greek and Latin floating around that we don't know about.”
Gray, for his part, calls the Language paper that used his own methods against the Anatolian hypothesis “a lovely piece of work” that really engages with the methods “rather than just saying [that Atkinson and I] are wrong.” Yet since turnabout is fair play, Gray's team has now started recrunching Garrett's data, but letting the data decide whether some languages are ancestral to others rather than assuming it. Although this work is preliminary and still unpublished, Gray and his colleagues are finding that the numbers again come up trumps for the Anatolian hypothesis.
New Clues from DNA
If the words themselves cannot tell us who is right, perhaps more evidence from outside the field of linguistics could help tip the balance. The latest genetic studies, at least, seem to favor the steppe hypothesis. Anthony and an international team of ancient DNA experts sequenced samples of genetic material from 69 Europeans who lived between 8,000 and 3,000 years ago, including nine skeletons from Yamnaya sites in today's Russia, and compared the DNA samples with those from four skeletons of the later Corded Ware culture of Central Europe.
Amazingly, the Corded Ware people, whose culture spread across Europe as far as Scandinavia, could trace three quarters of their ancestry to the Yamnaya people, and this Yamnaya genetic signature is still found in most Europeans today. So the Yamnaya, along with their genes and possibly their language, did indeed sweep out of the steppes in massive numbers, probably about 4,500 years ago. These results are a “smoking gun” that such massive migrations did take place out of the steppelands, says Pontus Skoglund, an ancient DNA expert at Harvard Medical School who was not involved in the paper but works in the laboratory of one of its authors. They “level the playing field” between the two hypotheses, he adds.
Unless, of course, this migration was a “secondary” wave that carried later Indo-European languages with it but not the original mother tongue, Proto-Indo-European. Such an interpretation, the pro-Anatolian researchers counter, would fit with the conclusions of Gray's 2003 study that pointed to the possibility of a later migration out of the steppes.
Will we ever know who is right? New evidence from ancient DNA for the spread of steppe peoples eastward into Siberia around 4,700 years ago could potentially overcome one of Renfrew's key objections to the steppe hypothesis, but it offers no proof about which languages went with them. One thing is sure: researchers will continue to debate the issue in whatever language their ancestors bequeathed them.