Did the First Farmers Spread Their Languages?

This chapter has brought together what might appear to be a rather bewildering quantity of information, often rather vague and contested, on the history of the major agriculturalist language families. Does this information support the three correlations between farming and linguistic homelands suggested at the start of this chapter? These three correlations were as follows: language families should have their putative homelands located within or close to agricultural homelands, they should intersect geographically within or close to agricultural homelands, and they should have dispersal histories that commence chronologically within agricultural homeland areas and then become progressively younger with distance away from such homelands.

My feeling at this point is that radially patterned dispersals of language families and early farming systems from agricultural homeland areas can be traced to a degree much greater than we could expect from chance alone. IndoEuropean, Afroasiatic, and Elamo-Dravidian radiate from around the Southwest Asian region of early farming. Likewise, Sino-Tibetan, Austroasiatic, and Austronesian, not to mention Tai, Hmong-Mien, and Altaic, all emanated from the Chinese region, albeit not all during the Neolithic. New Guinea has its Trans New Guinea Phylum, while West Africa spawned the Niger-Congo languages. In the Americas, we have just reviewed some rather hazy evidence that Quechua, Aymara, Arawak, and Panoan (with Tupian being less specific) developed in or fairly close to the Andes of Peru. Mayan, Chibchan, MixeZoque, Otomanguean, and Uto-Aztecan developed within Mesoamerica. Algonquian, Iroquoian, and Siouan perhaps had some connection with the early development of seed cropping in the Eastern Woodlands.

These radial patterns of genesis and spread of agriculturalist language families and early forms of farming do not occur at random all over the world. There is no evidence for them, for instance, in Europe, central Asia, southern Africa, the lower Amazon or the lower Mississippi. No one, to my knowledge, has ever claimed that Proto-Indo-European was spoken in the United Kingdom, that rice cultivation began in Sulawesi, or that maize cultivation began in New York (or

if they have, I haven't read their works). I am unable to prove that the suggested correlations can only be explained in the manner suggested, but there is clearly sufficient correlation within the spatial and chronological patterns to make the farming/ language dispersal hypothesis worthy of a great deal more attention than it has received so far.

Chapter 11

Genetics, Skeletal Anthropology,

and the People Factor

Critics of the early farming dispersal hypothesis usually raise the crunch-time question: never mind the languages and archaeology, what about the people? Did languages and farming lifestyles spread through unmoving populations of hunter-gatherers simply by cultural diffusion (i.e., adoption, borrowing, acculturation), or did they spread because of the demographic expansion (via demic diffusion) of populations whose ancestors already possessed the relevant languages and farming lifestyles? For some regions of the world, biological data are too few to inform usefully and directly on such questions. For other regions, such as western Eurasia and Southeast Asia/ Oceania, the debates are quite heated and schools of opinion can be discerned emerging from the steam. This is because both kinds of movement are, of course, logically possible, and can be eloquently argued for, even though the biological data do not clinch the matter with the finality that some protagonists might hope for.

The phenomenon of actual population spread is referred to in this chapter as demic diffusion, following the terminology used in their classic study of Neolithic Europe by archaeologist Albert Ammerman and geneticist Luca Cavalli-Sforza (1984:6). Demic diffusion in prehistoric contexts would have occurred, at least in optimal continental circumstances, via a wave of advance, in which continuous demographic growth in non-circumscribed frontier situations would have prompted a population boundary to expand outward by gradual or saltational means. The inward (or backward) direction would have been demographically "full," settled by populations that had reached optimal size after the wave of advance had moved on beyond them. Ammerman and Cavalli-Sforza calculated, using archaeological C14 dates, that a Neolithic wave of advance spread from southeast to northwest through Mesolithic Europe at an average rate of around 1 kilometer per year (see also Fort and Mendez 1999; and Fort 2003 for comparable calculations for the Pacific). They noted also that

actual rates of spread on the ground would have varied greatly because of environmental barriers and differing carrying capacities.

Ammerman and Cavalli-Sforza most emphatically did not suggest that the Neolithic farmers simply drove all the Mesolithic foragers in Europe into extinction - critics of the wave of advance model sometimes jump unfairly to this conclusion without good reason. Instead, they advocated continuing genetic admixture between the two groups as the wave of advance progressed (1984:128-130), such that the ultimate cultural and linguistic descendants of a "Middle Eastern" population commencing with 100 percent "farmer" genes, upon reaching the Atlantic limits, could have found themselves (had they noticed!) with almost 100 percent "native forager" genes, at least in theory (Krantz 1988:93; Cavalli-Sforza 2003). Such a process would have established what geneticists term a demic cline - a geographical gradient away from an origin in the occurrence of one or more genes. The essence here is a continuous watering-down of an expanding genetic configuration, away from its original state, by inexorable local input from an ever-unfolding native genetic landscape (Renfrew 2001c, 2003). It need hardly be stressed that Neolithic farmers could not have spread in the way 19th-century European colonists migrated to North America and Australasia, with constant back-up from source in the form of flotillas of ships carrying thousands of people. Neolithic expansion was more of an insistent creep, with occasional jumps, than a sudden and overwhelming sonic boom.

We return to genes and clines later, first making the point that this is not a chapter on the technicalities of population genetics, biochemistry, and mathematical computation. Interest here is in the historical observations that paleoanthropologists and geneticists can make about ancient human populations, rather than in the analyses and computations of the genetic data themselves. Archaeologists in search of population history should at least attempt to understand the historical reasoning processes of linguists and biologists, just as reverse accommodations should also take place.

Are There Correlations between Human Biology and

Language Families?

In 1774, on Cook's Second Voyage to discover the Great Southern Continent, Johann Reinhold Forster mused "What occasions the inhabitants of O-Taheitee to be so much distinguished from the Mallicolese?" (Thomas et al. 1996:175). He had noted that the Tahitians of Polynesia and the Malekulans of Vanuatu (Melanesia) spoke languages (now known as Austronesian) with many words in common, yet the people were so different in physical appearance. Austronesians are physically very varied, just as are speakers of Afroasiatic (e.g., Arabs and Ethiopians), Indo-European (e.g., Bengalis and Norwegians), and Altaic languages (e.g., Turks and Mongolians), to name just a few of the more obvious cases. The speakers within these language families seem unlikely to have descended entirely from single founder biological populations in the few millennia since the linguistic dispersals occurred, at least not in terms of our current understanding of rates of biological change in humans. But do these patterns reflect language movements through unmoving but already highly differentiated populations, or do they reflect population movements leading to a kaleidoscope of admixture between incoming and indigenous groups?

We might begin by asking if there are any demonstrable geographical correlations today between patterns of linguistic and biological variation on a worldwide canvas. If there are, then it stands to reason that such correlations could result from past episodes of demic diffusion, when population expansions occurred that carried both genes and languages. In 1988 and 1994, Luca Cavalli-Sforza and his colleagues addressed this question from a worldwide perspective, utilizing genetic distance analyses of 42 modern populations in terms of their frequencies of 120 different classical genetic markers. They claimed that the resulting genetic tree of populations resembled closely a tree of populations classified by language family membership, a claim recently restated (with qualifications) by Daniel Nettle and Louise Harriss (2003), using the same gene frequency data. Basically, genetic distances and languages correlate

quite closely in Europe, East and Central Asia, but rather less well in West Africa, Southwest and Southeast Asia. Nettle and Harriss rightly point out that strong cases of correlation could result from situations of past demic diffusion, whereas weaker situations allow for more population mixing. Furthermore, by controlling for the influence of purely geographical factors in their analysis, they are able to refute the complaint that such correlations simply reflect geographical proximity, rather than factors of correlated genetic and linguistic origin.'