ANRV355-AN37-13 ARI 14 August 2008 11:11 ANNUAL Evolutionary Linguistics REVIEWS Further Click here for quick links to Annual Reviews content online, William Croft including: Department of Linguistics, University of New Mexico, Albuquerque, New Mexico 87131; • Other articles in this volume email: [email protected] • Top cited articles • Top downloaded articles • Our comprehensive search by 142.177.59.32 on 06/04/09. For personal use only. Annu. Rev. Anthropol. 2008. 37:219–34 Key Words First published online as a Review in Advance on replicator, selection, phylogeny, comparative method June 17, 2008 Annu. Rev. Anthropol. 2008.37:219-234. Downloaded from arjournals.annualreviews.org The Annual Review of Anthropology is online at Abstract anthro.annualreviews.org Both qualitative concepts and quantitative methods from evolutionary This article’s doi: biology have been applied to linguistics. Many linguists have noted the 10.1146/annurev.anthro.37.081407.085156 similarity between biological evolution and language change, but usually Copyright c 2008 by Annual Reviews. have employed only selective analogies or metaphors. The development All rights reserved of generalized theories of evolutionary change (Dawkins and Hull) has 0084-6570/08/1021-0219$20.00 spawned models of language change on the basis of such generalized theories. These models have led to the positing of new mechanisms of language change and new types of selection that may not have biolog- ical parallels. Quantitative methods have been applied to questions of language phylogeny in the past decade. Research has focused on widely accepted families with cognates already established by the comparative method (Indo-European, Bantu, Austronesian). Increasingly sophisti- cated phylogeny reconstruction models have been applied to these fam- ilies to resolve questions of subgrouping, contact, and migration. Little progress has been made so far in analyzing sound correspondences in the cognates themselves. 219 ANRV355-AN37-13 ARI 14 August 2008 11:11 INTRODUCTION and language change) but do not imply an over- arching generalized theory. Evolutionary models have come to be employed An isolated analogy from evolutionary biol- Phylogeny: a graph in several areas of the study of language in ogy that has proven to be useful in explaining structure, usually a the past two decades. The use of evolutionary language change is Lass’s application of exap- tree, representing the models is naturally found in historical linguis- evolutionary history of tation to historical linguistic phenomena (Lass tics and also in the study of the origins of lan- a set of individuals 1990). Exaptation in biology is the employment guage. In the latter case, however, the employ- (organisms, species, of a phylogenetic trait for a function different languages) ment of evolutionary models is handicapped from the one for which it was originally adapted; by the absence of data regarding the transition Selection: the process Lass slightly changes the definition to apply to by which the from our primate ancestors to the emergence of linguistic structures that have lost their func- interaction of the modern human language, which is found in all tion but have come to be employed for another interactor with its societies. All that we can go by is the archaeo- environment causes function. logical record and the comparison of the social- differential replication A recent example of the employment of a cognitive abilities and communication systems of the relevant biological metaphor is Blevins’s theory of evo- replicators of humans and other animals, particularly lutionary phonology (Blevins 2004). Evolution- nonhuman primates. Because the study of the ary phonology proposes to account for syn- origin of human language does not depend on chronic phonological patterns as the result of linguistic data, it is not discussed in this article. phonetically motivated changes in the transmis- Even so, the area under review is vast and sion of sound systems from adult to child over growing, and therefore this review is restricted time. It uses the notions of inheritance (via the to research in which qualitative concepts and child learning the adults’ language), variation quantitative methods from evolutionary bi- generated by “errors” in replication (mecha- ology have been applied to the analysis of nisms by which the listener alters what he hears language, in particular language change and from the speaker), and natural selection (cer- language phylogeny. by 142.177.59.32 on 06/04/09. For personal use only. tain sound changes are more/less likely in par- ticular phonetic contexts). However, because of EVOLUTIONARY THEORY disanalogies between biological evolution and AND THEORIES OF language change, Blevins explicitly rejects an LANGUAGE CHANGE evolutionary approach to sound change that is Annu. Rev. Anthropol. 2008.37:219-234. Downloaded from arjournals.annualreviews.org In historical linguistics, the parallels between more than metaphorical. biological and linguistic evolution have been Although analogies or metaphors between observed since Darwin himself first took notice biological evolution and language change can (for a historical survey, see Atkinson & Gray be fruitful, one does not know which parallels 2005). However, the differences in the domains between the two domains are legitimate to draw of biology and language appear to have out- and which are not, or even more important, weighed the similarities, and Darwinian evo- which parallel structures must be present for lutionary theory has developed over time. In the analogy/metaphor to make sense. In partic- the meantime, the advent of structuralism and ular, it is common to assume that the mech- generative grammar has led to the dominance anisms that cause variation and selection in of an ahistorical approach to the study of biological evolution must be the same in other language (Croft 2002). As a consequence, lin- domains such as language change, yet the mech- guistics has rarely used models from evolution- anisms are domain specific. What is required ary biology. Nevertheless, the similarities be- is a generalized theory of evolutionary change tween the two have led historical linguists to that subsumes biological evolution, language employ evolutionary analogies or metaphors. change, and other phenomena of evolutionary Analogies/metaphors indicate similarities be- change such as cultural evolution. Researchers tween the two domains (biological evolution have derived models of cultural evolution from 220 Croft ANRV355-AN37-13 ARI 14 August 2008 11:11 biological evolution (Boyd & Richerson 1985, ties. Hence concepts are replicators and minds Cavalli-Sforza & Feldman 1981, Durham 1991, or brains of individuals are vehicles. Memeti- Richerson & Boyd 2005). But the generalized cists generally use a parasite-host model for the Replicator: an entity theories of evolutionary change that have at- relationship between memes/concepts and the that is copied and tracted the most attention in historical linguis- mind or brain of the possessor: Memes are para- preserves most of its tics are those developed by Dawkins (1989, sites that use the brain (the host) as their vehicle structure in copying 1982) and Hull (1988, 2001). for replication. Meme: a cultural The most crucial feature of a generalized The most extended analysis of language replicator theory of evolutionary change is that evolution- change in Dawkinsian memetic terms is by Ritt Interactor: an entity ary change is change by replication, a process (2004). Ritt, a historical phonologist, focuses on that interacts with its by which some entity is copied in such a way phonological change. Following Dawkins, lin- environment in such a that most or all of the structure of the replicate guistic memes are concepts in the mind; specif- way as to cause differential replication is the same as that of the original. The replica- ically, they are some type of replicable brain of the relevant tion process is cumulative and iterative, lead- structure. Ritt argues that phonemes, mor- replicators ing to lineages. Second, evolutionary change phemes, phonotactic patterns, metrical feet, is a two-step process: the generation of varia- and phonological rules, or more precisely tion in the replication process, and the selection their conceptual representations, are memes. of variants via some mechanism. Dawkins’ and However, linguistic signs (form-meaning pair- Hull’s models have these properties, as do the ings) are not replicators because, in Ritt’s view, models of cultural evolution mentioned above. they do not preserve enough structure in repli- In the context of language change, Lass notes cation. Instead, signs are the result of an alliance that these properties are necessary to under- of replicators. stand languages as historical entities (Lass 1997, In the Dawkinsian model, the linguistic be- pp. 109–11), although he does not develop a de- havior that a speaker produces on the basis of tailed theory of language change on this basis her conceptual memes exists for the purpose by 142.177.59.32 on 06/04/09. For personal use only. (see also Nettle 1999, Wedel 2006). Evolution- of replicating the memes, not for communi- ary theory also rejects any notion of progress. cation (Ritt 2004, p. 231); this is the selfish Dawkins’ and Hull’s models are related but gene/meme theory. The replicators are repli- differ in important respects. Dawkins gener- cated across speakers by an imitation process alizes the concept of a gene as a replicator.