14 April 1999

An evolutionary framework for understanding language (change)

William Croft

University of Manchester

An evolutionary framework for language change is outlined in this seminar, based on "The Evolution of Utterances" (Croft, forthcoming). The framework was developed by David Hull ("Science as a Process", 1988), based on research in evolutionary biology; Hull applies the framework to conceptual change in science.

The evolutionary framework is based on fundamental principles in biological evolutionary theory. Evolutionary processes apply to spatiotemporally bounded entities, that is, actually occurring, historical entities. Evolutionary processes characterize change via replication, not inherent change of an entity. Biological evolution requires the population theory of species: a species is defined as a population of reproductively isolated individuals, not as an "essentialist" type with a set of necessary properties. Finally, Hull develops a generalized theory of selection, which represents a synthesis of the debate between "gene selectionists" (such as Richard Dawkins) and "organism selectionists" (such as Ernst Mayr). Hull argues that both positions have converged on a single model of selection. Evolution occurs at two levels: replication of the structure of a replicator (e.g. a gene) and selection---differential replication of replicators----resulting from the interaction of an interactor (e.g. an organism) with its environment (the ecosystem). Hull applies his generalized theory of selection to the history of science, arguing that theoretical concepts are the replicators and scientists are the interactors, interacting with other scientists as well as with the empirical evidence for or against their ideas.

Applying the evolutionary framework to language requires that we examine real, historical entities. That is, a language grammar as an idealized (essentialist) system of rules, which changes inherently, is a fiction that must be discarded. Instead, we must examine actually occurring utterances and actual speakers' linguistic knowledge, and their replication. The populations that are most relevant to language (change) are the communicatively isolated population of speakers in a speech community and the population of utterances produced by those speakers. The two levels in Hull's generalized theory of selection correspond to innovation (= replication) and propagation (= selection) in language change. The replicator is the lingueme, which is a structured part of an actually occurring utterance (i.e., particular occurrences of phonemes, morphemes, words and constructions); the lingueme is altered in replication in the production of utterances. The interactor is the speaker; and the environment includes the speaker's interlocutors, the social context, and the experiences that the speaker wishes to communicate. The speaker selects lingueme variants as a result of interacting with her environment in communication.

A rigorous application of the evolutionary framework to language change reveals a number of interesting features, compared and contrasted with biological evolution. Linguemes are organized in terms of degrees of inclusiveness; but there is "top-down" as well as "bottom-up" organization of utterance structure---as with the genome. The lingueme pool represents frequencies of variants of linguistic variables---cf. allele frequencies in the gene pool. Grammaticalization (J. Greenberg, E. Traugott) describes a significant subtype of lingueme lineages. The central mechanism of lingueme replication, namely convention (D. Lewis, H. Clark), is always integrated with nonconventional devices for communication; and anyway, conventions are always in the process of being acquired, replaced or lost. Innovation occurs via reanalysis of the form- function mapping in utterances, while propagation occurs via social factors (acts of identity), etc. The population structure of speakers is far more complex than the population structures of organisms: speakers belong to multiple speech communities (H.Clark); communities are internally heterogeneous (J. & L. Milroy), and communicative isolation is never complete (language contact). Language phylogeny resembles plant phylogeny, not animal phylogeny, possessing patterns of hybridization as well as divergence.

Although the evolutionary framework is essentially a theory of language change, it has significant implications for theories of language in general. The evolutionary framework offers a coherent theory for understanding language, integrating structure, function, change, society and mind.