Ground zero? With tools from the study of virus out- breaks, a team traced Indo-European languages to an ancestral homeland chiefl y in modern (red).

from other studies on 103 ancient and mod- ern Indo-European languages. They con- sidered this data set analogous to molecular sequence data, with the rate of gain and loss akin to the rate of nucleotide substi- tution in viral evolution. Atkinson’s team also added published data on the geographical ranges of all 103 languages, plus historical dates for language divergence, such as the breakup of the Roman Empire, which triggered the evolution of from a type of Latin. The LINGUISTICS computer model then worked back in time, inferring possible ancestral relationships and New Method Puts Elusive Indo- patterns of diffusion, and generating possible homelands. Then the team compared “how European Homeland in often the origin locations fell into the range proposed for the Anatolian theory versus the Among all the world’s language families, substantive points,” says archaeologist David Steppe theory,” Atkinson says. The Anatolian Indo-European is by far the most global. Anthony of Hartwick College in Oneonta, hypothesis won hands down. Today its more than 400 languages are New York. “There are a lot of curious and sur- The analogy to virus evolution is “clever on November 8, 2015 spoken by some 3 billion people on every prising results that they don’t explain.” and insightful,” says linguist Paul Heggarty inhabited continent. But who were the fi rst Charles Darwin noted in 1871 that lan- of the Max Planck Institute for Evolution- Indo-European speakers, and where did they guages, like plants and animals, could be ary Anthropology in Leipzig, Germany, who come from? Researchers have argued fi ercely classed into related groups. Each language thinks the study will prompt other linguists over two major competing candidates: Neo- arose only once, in one place, and modern to try the method. He says the study will give lithic farmers who carried their agricultural languages descended with modifi cation from supporters of the Steppe hypothesis “plenty know-how with them as they pushed out ancestral ones. The proofs of language and of explaining to do.” from Anatolia, chiefl y in modern Turkey; or species evolution “are curiously parallel,” he Other researchers, however, take strong

Bronze Age horse masters from the Eurasian wrote in The Descent of Man. issue with the fi ndings. Anthony notes that www.sciencemag.org steppes, who spread into Asia and Europe Atkinson and colleagues applied Dar- Atkinson and his colleagues limited their with advances such as the wheel. win’s analogy to the Indo-European language study to vocabulary, just one of three sub- Now an international research team has family, which includes varied but related sets of linguistic data, “something you are borrowed a computational approach from tongues such as English, Italian, Alban- really not supposed to do,” he says. In addi- biology to shed new light on the problem. ian, Persian, and Hindi. Linguists have long tion, the authors rooted their model in geog- Using models originally created to trace sought clues to the origins and spread of raphy mainly using modern distributions of the origins of viral pathogens, such as avian these languages by analyzing their vocabu- languages. “The results don’t tell you much Downloaded from influenza, during outbreaks, evolutionary lary, sounds, and grammar, and by studying about the past,” Anthony concludes. psychologist Quentin Atkinson of the Uni- the of ancient migrations. Evi- The paper makes many inferences on versity of Auckland in New Zealand and his dence marshalled by Renfrew in the 1980s matters such as the rates of language change colleagues report on page 957 of this issue suggested an Anatolian homeland, the same and how languages diffuse, says Victor Mair, that they have found decisive support for the land from which the fi rst farmers spread 8000 a Chinese language expert at the University Anatolian hypothesis. “I think we’ve put for- or 9000 years ago. But recent archaeological of Pennsylvania. “There is so much about ward the best case yet for where the Indo- and linguistic data have pointed to an origin this paper that is arbitrary,” he says. By com- European languages came from,” Atkinson on the steppes north of the Black and Caspian parison, he says, the Steppe hypothesis “is says. “And we’ve also shown that languages seas, where seminomadic herders known as based heavily on archaeological data such as can be used to retrace human history in both the Yamnaya expanded into Europe and Asia burial patterns, which are directly tied to dat- , 6097(24AUGUST 2012)

time and space.” with domesticated horses and wheeled carts able materials.” 337

If the fi ndings stand, they hold key impli- beginning perhaps 5000 years ago. Atkinson counters that archaeologists cations for everything from human migrations Enter Atkinson and a research team struggle to link their fi nds to particular cul- SCIENCE, to the role of agriculture in ancient language drawn largely from the fields of biology, tures. “Peering back into human prehistory ,

expansion. “This is a major breakthrough,” computer science, and psychology. They is not easy,” he says. “It’s like holding a dim ET AL. says archaeologist of the Uni- focused on vocabulary, specifi cally the gain candle over a dark abyss, and you need to use versity of Cambridge in the United Kingdom. and loss of , or words in related every piece of information that you can.” But in this divisive fi eld, the team’s methods languages—such as “mother” in English and –HEATHER PRINGLE and fi ndings are already drawing fi re. “This “mutter” in German—that stem from a sin- Heather Pringle is a contributing editor at Archaeology

article raises more questions than making gle ancestral root. The team used cognates magazine. R. BOUCKAERT CREDIT:

902 24 AUGUST 2012 VOL 337 SCIENCE www.sciencemag.org Published by AAAS