Horse, Wagon & Chariot: Indo-European Languages and Archaeology
Anthony, David W., Antiquity
The current relationship between archaeology and traditional historical linguistics can be described as a trial separation. In recent publications by archaeologists, historical or comparative linguistics has been criticized as a narrow, self-contained discipline that cannot reliably reconstruct elements of proto-languages or proto-lexicons because unrealistic assumptions guide the comparative method (Renfrew 1987: 78-86, 100-118; 1988; 1992: 448-53; Robb 1991; 1993; Zvelebil 1994). Its worst flaw is said to be its disregard of and inability to account for processes of linguistic borrowing, convergence and creolization as agents of language change; it is said to rely overmuch on linear, 'family-tree' models. The comparative method also has been linked, some would say inextricably (Robb 1993: 748), to outdated theories equating language, ethnicity and material culture, an equation that is demonstrably false in many cases. As a consequence, the lexical reconstructions produced by historical linguistics are considered misleading except at the most general level, and 200 years of comparative research on the Indo-European protolexicon has been been thrown into doubt by a wave of criticism from archaeologists. I find myself in the odd position of pleading for the comparative method before a court of archaeologists so that I can use historical linguistics in Indo-European archaeology without seeming naive.
In defence of the comparative method
The comparative method is not at all concerned with ethnicity or its relationship to material culture. Rather, it is a tool for classifying languages and determining the kinds and degree of relationship between them. The results of the comparative method have been used inappropriately in simplistic scenarios of mass ethnic migration by both linguists and archaeologists, encouraged, in Robb's (1993: 748) view, by the 'family tree' diagrams used to illustrate genetic relationships between languages. Linguists are very much aware that diagrams are simplistic summaries of complex relationships. Ann Stewart (1976) has written an entire book on alternative methods for diagramming linguistic relationships. Evolutionary biologists have the same problem, but no one has suggested that evolutionary theory should be discarded because it is difficult to diagram. It is up to archaeologists to use the results of comparative linguistic studies appropriately.
Language is a complex mix of the cultural and the mechanical - culture imposes arbitrary rules that govern vocabulary, grammar and syntax, making every language an entirely cultural invention; and the human vocal apparatus imposes mechanical constraints that guide historical phonological change in predictable directions. Because the rules and lexicon that define any language are cultural and arbitrary, linguists assume that systematic similarities between languages cannot arise by accident, but must reflect cultural contact. A principal goal of the comparative method is to determine if languages are related genetically (that is, if they are differentiated daughters of a common mother language). It is not true that the method assumes that genetic relationships are the only kind of historical linguistic relationship. It is true that genetic relationships are the only kind of relationship that the method tests. The result of a comparative analysis of two or more language lineages is either a verifiable genetic relationship or no verifiable genetic relationship.
A genetic relationship between two or more language lineages is demonstrated by a large cognate vocabulary united through regular rules of sound change to a shared proto-lexicon and to a shared grammar; the cognate terms fit into the grammar in such a way that cognate terms belong to the same arbitrary grammatical classes across the daughter languages. Multi-layered homologies like these can be explained systematically - that is, by reference to a standard set of rules and principles - only by proposing a genetic relationship. …