Researchers have adapted a method normally used for analyzing the evolutionary history of genes to construct a family tree of Western European languages. The method could become an essential tool for unraveling the histories of the world's other languages.
The languages of Western Europe are the most thoroughly studied of any, but the Celtic languages have remained mysterious. About 2000 years ago, most people in Western Europe spoke Celtic, but then it was nearly exterminated by a conquering spree of the Roman Empire. A few Celtic pockets along the Atlantic coast resisted the spread of Latin, eventually becoming the modern languages of Breton (in France), Welsh (in Wales), and the Gaelic of Ireland and Scotland, but the relationship among these offshoot languages is a puzzle. To explain the differences between Welsh and Gaelic, for example, the leading theory holds that Wales and Ireland were colonized by continental Celts on two occasions. But it has been difficult to test the idea because the archaeological evidence is unclear and few written records of ancient Celtic remain.
Taking a fresh approach to the Celtic conundrum, Peter Forster, a geneticist at the University of Cambridge, U.K., teamed up with Alfred Toth, a linguist at the University of Zurich, Switzerland, to analyze the words as if they were genes. As with genes, the basic words of a language rarely undergo drastic changes, and comparing such changes between languages can reveal their historical relationship. Drawing on ancient inscribed artifacts, Forster and Toth compiled 35 words in English, Latin, ancient Greek, the modern Celtic tongues, and Gaulish, an extinct Celtic language once spoken in France. They then adopted a statistical method used by geneticists to build a family tree.
Contrary to the accepted model, the results indicate that Celtic jumped from the continent over the English Channel just once, then split into the languages now spoken in Wales, Ireland, and Scotland. What's more, by estimating the linguistic "mutation" rate and calibrating against the known historical timeline, the method allowed Forster and Toth to extrapolate dates for the origins of the languages. Their tree, published 1 July in the Proceedings of the National Academy of Sciences, pushes the origin of Indo-European, the original ancestor of all European languages, from 4000 B.C. to about 8000 B.C.
The genetic approach could be "a breakthrough in methodology for historical linguistics," says Colin Renfrew, a linguist at the University of Cambridge. He cautions that the dates have a wide margin of error because the data are so scarce.