A Typology of Social Evolution and Language Change
William Croft, University of Manchester

Joseph H. Greenberg made major contributions to the field of linguistics in genetic classification of languages and the typology and universals of language. Greenberg also made important links between language, culture and society and between language and evolutionary theory in biology. I will attempt to build on Greenberg's legacy in integrating these areas of research in order to address issues raised in historical linguistics. I will combine the classification of societies from social anthropology (Service, Diamond) with research in sociolinguistics (inter alia, Fishman, Gumperz and the Milroys) and social psychology (H. Clark) on the communicative stucture of society, and propose a typology of language contact situations and their linguistic outcomes correlated with the types of societies in contact. Then I will turn to the evidence for the prehistory of human societies and suggest some tentative conclusions about the nature of language change since the origin of modern human language.

Social anthropology has classified human societies into bands, tribes, chiefdoms and states, though there are of course no sharp divisions between the types. Bands number a few dozen members and are nomadic, while tribes number in the hundreds and are sedentary, in a single village. Bands and tribes are largely egalitarian. Chiefdoms represent larger societies, numbering in the thousands and residing in multiple villages. Economic specialization and social stratification emerge in chiefdoms. Finally, states represent even greater population densities and specialization, and also the integration of multiple societies into a single unit, usually by force.

Historical liguistics always ultimately reflects underlying social history. The family tree model of genetic classification of languages reflects the pattern of fission in societies. Fission commonly occurs in all types of societies. Language contact on the other hand involves some degree of reticulation of the family tree, at least with respect to individual linguistic traits. These fall into two broad types: borrowing of actual forms (both words and inflections), and convergence of structural patterns (both sound structure and grammatical structure), arising from language shift or long-term multilingualism. New languages also arise: pidgins, typically from trade or migrant labor (including slavery), and the creoles they can evolve into; and a few instances of so-called mixed languages.

Social contacts that may have linguistic consequences include marriage, trade and politicial integration. Among bands, exogamy can lead to sometimes significant borrowing, but often the spouse's band speaks the same language (due to fission) and then exogamy reinforces the integrity of the language. Trade can lead to multilingualism in all societies, but to lingua francas probably only with chiefdoms and states, and trade pidgins probably only with states. This sort of multilingualism leads to specialized borrowing and structural convergence. Symbiotic relationships between chiefdoms may also lead to structural convergence in their languages. Among bands, tribes and chiefdoms, political alliances are largely opportunistic and functionally restricted, with linguistic consequences not unlike trade. It is with states that the greatest range of contact situations occurs. States engage in trade, conquest, colonization, enslavement, and political integration, leading to lingua francas and trade pidgins, migrant labor pidgins and creoles, structural convergence, and on rare occasions stable mixed languages.

I take the view that modern human language arose the with emergence of behaviorally modern humans approximately 150 to 70 thousand years (ky) ago. Archaeological evidence suggests that early modern humans were largely organized as bands. Tribal organization may have developed in ecologically favorable environments, but such environments were rarer then due to the last Ice Age. There is some archaeological evidence that stratification (chiefdoms) may have emerged in Ukraine 26-14ky ago and in SW France 15-11ky ago. The first states emerged around 6ky ago, well after the end of the last Ice Age. Thus, it appears that the types of language contact situations that cause the greatest potential difficulties for linguistic genetic classification have probably occurred only in the most recent period of language evolution. However, borrowing particularly in the context of band exogamy must be carefully considered.