Indo-European languages traced back to Turkish region
A 'family tree' of languages adds weight to the theory that a large group of languages, including English, originated from the region that is now Turkey.
The controversial research challenges the long-held view that these languages originated in the grassy steppe region around Ukraine and Russia 5,000- 6,000 years ago.
Using techniques for tracing virus outbreaks to map the evolution of languages, the study published in Science suggests Indo-European languages emerged 4,000 years earlier from Anatolia.
"We hope we've provided the most convincing case yet that this family of languages came from the Anatolian region 8,000-9,500 years ago, around the time agriculture was beginning to spread," says study co-author Quentin Atkinson, an evolutionary anthropologist at the University of Auckland.
Atkinson says the research provides insight into the development of our culture and adds weight to the theory that agriculture has been crucial in shaping the world's language and cultural diversity.
The Indo-European language family is one of the largest groups in the world, and includes English, German, French, Spanish, Russian, Persian, Hindi, Sanskrit and ancient Greek. While they sound very different, linguists have known for centuries they all came from a single origin.
But with no written record of the common 'ancestor' language, linguists aren't sure where the languages originated.
The 'DNA' of language
The most widely held steppe theory is based on traditional linguistic evidence, which requires a link between archaeological evidence and language. For example, if an early word for wheel existed at a time where wheels were only known to be found in one location, you can assume the language was spoken there.
But previous work by Atkinson and colleagues indicated that Indo-European languages date back much earlier than predicted by the steppe theory, supporting the opposing Anatolian theory.
To compare the two theories, the researchers turned to a method used to trace virus outbreaks to find where the languages come from.
"With a virus outbreak, you collect samples from around the world, extract DNA and work out how all the strains are related," says Atkinson. "We can do the same thing with language."
Just like species or viruses, languages evolve via descent and modification - so when a population splits into two, the language will begin to diverge as it's passed down and eventually change altogether.
But, just like species, they retain common elements known as 'cognates', which are like the DNA of language, he explains.
"Cognates are words that linguists can identify as being related due to common ancestry," says Atkinson. For example, water in English is linked to the German word wasser, and 'mother' is linked to 'mutter' in German and 'madre' in Spanish.
The researchers looked at cognates across 103 languages to construct the Indo-European family tree, then used their knowledge of where modern languages are spoken and the distribution of ancient languages to map the branches by space and time.
Using that information, they could predict how likely it was that the tree had its roots in either the Pontic-Caspian steppe or Anatolia - and the evidence for the Anatolia theory was extremely strong, says Atkinson.
But not everyone is convinced.
Andrew Garrett, a historical linguist and expert in Indo-European languages from the University of California, Berkeley, says the study doesn't take into account several factors, such as how much easier it would have been for language to spread across the grassy, flat steppe compared to the deserts of Anatolia.
"The paper is an interesting and provocative contribution to an important debate. It will certainly stimulate discussion, particularly with its new methodology, but specialists will not be convinced by the claims," says Garrett.
Robert Mailhammer, a linguist from the University of Western Sydney, also worries that the researchers have been cavalier in their disregard of the previous evidence.
"The paper seems to just brush the linguistic evidence off very simplistically," Mailhammer says.
"Of course linguistic methodology isn't foolproof, nothing is. But if they want to prove their theory they need to disprove the evidence that has come before it, and they haven't done that."
But Mailhammer hopes in future to use the traditional linguistic methods alongside the new method, known as the phylogeographic approach, to explore the debate further.