Indo-European tongues traced back more than 8,000 years to present-day Turkey

By Bruce Bower

Web edition: August 23, 2012
Print edition: September 22, 2012; Vol.182 #6 (p. 10)

Enlarge

ANCIENT SPREAD

The map shows the timing and geographic expansion of Indo-European languages proposed in a new statistical analysis. The red area in what’s now Turkey is a possible birthplace of the Indo-European language family more than 8,000 years ago.

Credit: Remco Bouckaert et al.

Indo-European languages range throughout Europe and South Asia and even into Iran, yet the roots of this widespread family of tongues have long been controversial. A new study adds support to the proposal that the language family expanded out of Anatolia — what’s now Turkey — between 8,000 and 9,500 years ago, as early farmers sought new land to cultivate.

A team led by psychologist Quentin Atkinson of the University of Auckland in New Zealand came to that conclusion by using a mathematical method to calculate the most likely starting point and pattern of geographic spread for a large set of Indo-European languages. The new investigation, published in the Aug. 24 Science, rejects a decades-old idea that Kurgan warriors riding horses and driving chariots out of West Asia’s steppes 5,000 to 6,000 years ago triggered the rise of Indo-European speakers.

“Our analysis finds decisive support for an Anatolian origin over a steppe origin of Indo-European languages,” Atkinson says.

He and his colleagues generated likely family trees for Indo-European languages, much as geneticists use DNA from different individuals to reconstruct humankind’s genetic evolution. Many linguists, who compare various features of languages to establish their historical connections, consider Atkinson’s statistical approach unreliable (SN: 11/19/11, p. 22).

Atkinson’s group analyzed 207 commonly used words, including terms for relatives and numbers, in 103 ancient and modern Indo-European languages. The researchers produced possible language trees based on estimated rates at which languages gained and lost cognates, words with similar meanings and shared sounds, such as five in English and fem in Swedish.

The studied cognates are basic vocabulary terms that rarely get borrowed when speakers of different languages encounter one another, Atkinson contends. Thus, in his view, these words provide a valuable window into the evolution of separate branches on the Indo-European family tree.

The researchers combined their language trees with present geographic ranges of individual languages to identify the most likely location and age of the Indo-European family’s origins. An ancient Anatolian root emerged whether the researchers combined linguistic data or separately considered the 20 ancient languages and 83 modern ones.

As a further check, statistical simulations that assumed slow rates of language migration if people traveled along land routes or faster migration rates spurred by water crossings converged on a scenario in which Indo-European tongues originated among Anatolian farmers sometime between 8,000 and 9,500 years ago.

Farmers alone didn’t propel the evolution of different Indo-European tongues, Atkinson says. His team’s proposed trees suggest that new languages began to sprout within the five major Indo-European subfamilies from 4,500 to 2,000 years ago, after agriculture had spread across Europe. Kurgans or other expansionist Indo-European cultures could have instigated those later linguistic developments, Atkinson says.

Atkinson’s statistical reconstruction is unpersuasive, comments linguist H. Craig Melchert of the University of California, Los Angeles. Researchers can confidently rebuild trees of Indo-European languages extending back no more than about 7,000 years, he says.

Many linguists and archaeologists suspect that Indo-European languages originated in what’s now the southern Russian steppes, and that’s unlikely to change as a result of the new study, says linguist Joe Eska of Virginia Tech in Blacksburg. Cognate swapping across languages could have occurred more often than assumed by Atkinson, undermining his conclusions, Eska contends.