Published online 27 November 2003 | Nature | doi:10.1038/news031124-6


Language tree rooted in Turkey

Evolutionary ideas give farmers credit for Indo-European tongues.

Languages, like genomes, encode information.Languages, like genomes, encode information.© Corbis

A family tree of Indo-European languages suggests they began to spread and split about 9,000 years ago. The finding hints that farmers in what is now Turkey drove the language boom - and not later Siberian horsemen, as some linguists reckon.

Russell Gray and Quentin Atkinson, of the University of Auckland in New Zealand use the rate at which words change to gauge the age of the tree's roots - just as biologists estimate a species' age from the rate of gene mutations. The differences between words, or DNA sequences, are a measure of how closely languages, or species, are related.

Gray and Atkinson analysed 87 languages from Irish to Afghan. Rather than compare entire dictionaries, they used a list of 200 words that are found in all cultures, such as 'I', 'hunt' and 'sky'. Words are better understood than grammar as a guide to language history; the same sentence structure can arise independently in different tongues.

The resulting tree matches many existing ideas about language development. Spanish and Portuguese come out as sisters, for example - both are cousins to German, and Hindi is a more distant relation to all three.

All other Indo-European languages split off from Hittite, the oldest recorded member of the group, between 8,000 and 10,000 years ago, the pair calculates1.

Around this time, farming techniques began to spread out of Anatolia - now Turkey - across Europe and Asia, archaeological evidence shows. The farmers themselves may have moved, or natives may have adopted words along with agricultural technology.

The conclusion will be controversial, as there is no consensus on where Indo-European languages came from. Some linguists believe that Kurgan horsemen carried them out of central Asia 6,000 years ago. "No matter how we [changed] the analysis or assumptions, we couldn't get a date of around 6,000 years," says Gray.

"This kind of study is exactly what linguistics needs," says April McMahon, who studies the history of languages at the University of Sheffield, UK. It shows how ideas about language evolution can be tested, she says: "Linguists have always been good at coming up with bold hypotheses, but they haven't been terribly good at testing them."

“Linguists have always been good at coming up with bold hypotheses, but they haven't been terribly good at testing them”

April McMahon
University of Sheffield

But the technique is still fraught with difficulties, McMahon warns. There is lots of word-swapping within language groups. English took 'skirt' from the Vikings, for example, but 'shirt' is original. Linguists must separate the shared from the swapped, as any error will affect later studies.

The Kurgan might not be out of the picture entirely, says McMahon - they may have triggered a later wave of languages. "This isn't going to knock the debate on the head," she says.

Biology and linguistics can learn a lot from each other, comments geneticist David Searls of GlaxoSmithKline Pharmaceuticals, based in King of Prussia, Pennsylvania. "There may be some fundamental principles of evolution of complex systems, such as languages and organisms," he says. 

University of Sheffield

  • References

    1. Gray, R. D. & Atkinson, Q. D. Language-tree divergence times support the Anatolian theory of Indo-European origin. Nature, 426, 435 - 439, doi:10.1038/nature02029 (2003). | Article | ISI | ChemPort |