JB: Which dimension is producing the interesting work?

RENFREW: In a way genetics is the most obvious area of progress. Molecular genetics is making enormous strides. It makes sense to compare different loci both for the nuclear genome, and the mitochondrial genome. I think that work is bowling along very merrily. There' are several controversies, but there deserve to be; there are problems in interpretation, questions about the constancy of mutation rates, especially for mitochondria, etc. That's science that is progressing impressively, and we know that in 5, 10 years time we shall see that much more clearly.

In the linguistics field it's a very different situation. I find it very difficult to get through to see it clearly. Most of the well-respected linguists are specialists perhaps in a single language family. If you're an Indo-Europeanist, you know a lot of the Indo-European languages, but you may not know so much about other language families. There are not very many people who like to look at this language family and that other and geographically distant language family to see if they have a similar pattern. One of the people who does like to do that is Joseph Greenberg at Stanford, and he is very much a synthesizer among linguists, but he's also very much criticized by linguists.

Aharon Dolgopolsky at the University of Haifais another synthesizer; he is very much a member of that Russian school, the so-called Nostratic School. They take the view that if you look at the major language families, primary language families of Eurasia like Indo-European, and Afro-Asiatic, Uralic and so on, then you can see some relationships between some of those, that make them feel that at a greater time depth there was a broader linguistic grouping, a macro-family, the Nostratic macro family, as they name it, which embraces these families. That takes you, if you believe it, if you accept it, to a greater time depth. It takes you to proto-Nostratic. Only perhaps a 50th of the languages of the world would be described as Nostratic.

There's little clear evidence that there was a single ancestral language, though there may have been at a very much earlier point in time. But even this Nostratic business, and Greenberg's enterprise, which is an analogous one, don't perfectly harmonize. But the goal is to look at larger family units, which would therefore be earlier family units. The problem there is that many of the best linguists, certainly many of the most careful linguists, say that you simply can't talk in these terms. Many believe that it's very difficult to go beyond a time depth of four or five thousand years ago.

As a non-linguist, my problem is that I can understand what the generalists, the synthesizers are talking about. I read what they're saying, their ideas seem to me interesting. The criticisms made of them seem to me sometimes not very valid- they're expressed as a principle which is that you can't go beyond 5,000 years. But what principle is that? It doesn't make any particular sense at all.

It's an extraordinary situation that I can't, really, find any dialogue between these two groups of linguists. We've actually got a grant for my Institute from the Alfred P. Sloan Foundation, to seek to go beyond where we are now, and see what may be possible. We're hoping to make some progress in that. I'm not a linguist myself, but I do see where the archaeology comes in; I think that's my contribution, and I also do see where the genetics comes in, although I'm not a geneticist.

As I was saying earlier, the real problem is the interface between these three fields of archeology or culture-history, genetics, and historical linguistics. And nobody's a master of all these fields, so I don't feel too diffident; one's always an amateur if you're going between them.

JB: Ten years from now will there be a new synthetic field that will merge these specialties?

RENFREW: There might be. I can certainly see how the archaeologists and the geneticists talk the same language. For example, if we're asking whether humankind came out of Africa then it's the genetic evidence that's very powerful there. With the linguists, I really don't know how one gets through this barrier separating the lumpers and splitters, as they're called.

The lumpers are those who like to synthesize; the splitters are those who say, oh no, we can't join them together; we know about this, and we know about that, but we don't think that they're linked. And of course they may be right. It's very difficult to see how the lumpers and splitters are going to come into a coherent dialogue. Until a way forward is found to see issues in historical linguistics with a clearer eye, it's going to be difficult. I'm not a linguist, although I read some of these things, so I have my ideas of what makes sense, what doesn't. We probably need one or two bright young linguists, whose linguistic capability, and knowledge of their subject, is undoubted, who can really begin to reconcile some of these differences amongst an older generation.

JB: To what extent has computation affected this area of research, in terms of genetics, and linguistics, and archeology?

RENFREW: In the field of genetics it certainly has, and indeed it's interesting that most of the difficulties in interpretation rise not from the analytical techniques in the molecular biology, but once you get your measures of similarity and distance, from how you handle that. Some of these have been not so much computational, although computation is important, as interpretational problems. And you probably recall the controversy about 3 or 4 years ago, when the Out of Africa tree, the dendrogram, produced by Stoneking, Cann and Wilson was then criticized as not being the most parsimonious tree. The data are from mitochondrial DNA, so they're molecular genetic data, but the controversy was in data handling, which is a computational problem, as you say, or an interpretational problem. And there it turned out that they didn't have the most parsimonious tree, and their argument actually collapsed rather dramatically, because a whole range of roughly speaking equally parsimonious trees were found, which were open to other interpretations. However, the Out of Africa idea is still sustained, because you come to similar conclusions using other loci and using nuclear DNA as well as mitochondrial DNA..