Linguists have long assumed that most of the languages of Europe and the Indian subcontinent derive from a single ancient tongue. But researchers have fiercely debated just when and where this mother tongue originated. Now a study asserts that the common root of the 144 so-called Indo-European languages, which include English and nearly all the languages spoken in Europe and on the Indian subcontinent, was spoken more than 8000 years ago by Neolithic farmers in Anatolia, in central Turkey.
Evolutionary biologist Russell Gray and his graduate student Quentin Atkinson of the University of Auckland in New Zealand combined state-of-the-art computational methods from evolutionary biology with an older technique for dating languages, called glottochronology. Glottochronology uses the percentage of "cognates"--words with shared roots--to determine how long ago various languages diverged. For example, the Sanskrit and Latin words for "fire," agnis and ignis, show clear evidence of a common origin. But the technique has long been out of favor, in part because of its flawed assumption that words alter steadily over time. Gray and Atkinson applied their method to a database previously compiled by Yale University linguist Isidore Dyen, comprising 2449 cognate sets from 87 Indo-European languages.
No matter how they varied parameters such as the rate of word change, the answer came out pretty much the same: Indo-European languages initially diverged between 7800 and 9800 years ago, with the best guess being around 8700 years, they report in the 27 November issue of Nature. Moreover, the analysis showed ancient Hittite, an extinct Anatolian language, to be closest to the root of the language tree, providing a slam dunk for the so-called Anatolian hypothesis.
"It is almost too good to be true," says Margalit Finkelberg, a classicist at Tel Aviv University in Israel who has long favored the Anatolian hypothesis. But many linguists prefer a competing theory, which traces Indo-European languages to Kurgan horsemen in southern Russia about 6000 years ago. Some of these researchers challenge the new study. "I cannot possibly accept [their] results," says linguist Craig Melchert of the University of North Carolina, Chapel Hill, who adds that the paper "simply reconfirms the unreliability of any glottochronological model, no matter what improvements are made."