At the cutting edge

In the first of a series in which top academics discuss their ground-breaking research, archaeologist Colin Renfrew explains how our genes hold the key to a linguistic riddle 8,000 years old

The Indo-European language family has been recognised by historical linguists for more than 200 years - it includes most of the languages of Europe as well as Iranian and many of the languages of north India and Pakistan. But what is not agreed is how such a vast area of linguistic affinity came about.

Most linguists agree that there must have been an ancestral language, "Proto-Indo-European", from which all the others are derived, and that it must have been spoken at least as early as 3000bc in a homeland located somewhere in the midst of the great modern distribution of Indo-European languages. One widely accepted suggestion is that "Proto-Indo-European" was spoken in the steppe lands of what is now the Ukraine around 3000bc, and that with the domestication of the horse, it was carried east and west by mounted-warrior-nomads who came to dominate most of Europe and beyond.

But a decade ago, in my book Archaeology and Language, I argued that this theory was not persuasive. There is no evidence that the horse was ridden for military purposes in Europe before about 1200bc. Instead I suggest that Proto-Indo-European speech came to Europe with the coming of farming, around 6000bc from Anatolia, the modern Turkey.

Many linguists found my suggestion very unsatisfactory in linguistic terms, yet most archaeologists agree with me that the old mounted-warrior-nomad theory is an insufficient explanationfor how the language came to be dispersed. At the McDonald Institute for Archaeological Research in Cambridge I and my colleagues are working on a project, "The Prehistory of Languages'', which operates in the difficult overlap area between linguistics, archaeology and molecular genetics. When evidence is brought to bear from genetics on the Indo-European problem some fruitful discoveries emerge.

Study of gene frequency maps shows that it is likely that there was strong gene flow from Anatolia to Europe in prehistoric times. At first this was seen as strong support for the farming dispersal theory for Indo-European origins. But work on human mitochondrial DNA by Brian Sykes and his colleagues in Oxford, suggests that this gene flow may have been in the Palaeolithic period - far earlier than most linguists would accept for Proto-Indo-European. The matter is one of increasing controversy.

Our project is supported by a grant from the Alfred P. Sloan Foundation of New York, which has the policy of supporting research into subjects which are at the very "limits of knowability". For the archaeologist and the prehistorian one of the most challenging problems is to explain the great range of cultural diversity in the world today. An important part of that diversity is linguistic: more than 5,000 languages are spoken. Some are clearly related - French and Spanish, for instance. But other resemblances - for example those between Latin and Greek - must have a much earlier origin, apparently lost in the mists of prehistory.

Today, however, the application of molecular genetics is beginning to reveal much about the histories of populations. A synthesis may be emerging between the disciplines of prehistoric archaeology, historical linguistics and molecular genetics which may take us beyond the limitations of any one of these areas towards an understanding of cultural, linguistic and genetic diversity.

The problem is that no one of the three subject areas in isolation can provide all the answers.

In my view it is necessary to clarify the processes of population history and language history. Impressive research has been undertaken in recent years on the languages of the Pacific, notably the Polynesian languages. There, farming dispersal seems an important part of the story. The Afroasiatic and Indo-European language families have been classified by a number of Russian scholars as belonging with several other language families (eg Uralic, Dravidian) within a still larger linguistic phylum or macrofamily called Nostratic - a claim disputed today by many linguists. But how does one test the Nostratic hypothesis that all these languages have a single, common origin? We have begun by publishing a monograph by the linguist Aharon Dolgopolsky entitled The Nostratic Macrofamily and Linguistic Palaeontology which will be discussed at a symposium in July. Later we hope to have a meeting on linguistic and genetic diversity in the Americas, and then to submit the "farming dispersal" hypothesis for the origin of various language families to critical scrutiny.

In a few years we may be a little clearer as to whether these are problems capable of solution, or whether they do indeed lie beyond the limits of knowability.

Professor Colin Renfrew is director of the McDonaldInstitute for Archaeological Research, Cambridge University.