I welcome Christiansen & Chater's (C&C's) contention that cognitive scientific approaches to language should take seriously the idea that language is adapted to its environment. Although C&C concentrate on the private cognitive and neural structures that each language user possesses, my commentary concerns a different aspect of the environment of language, one which is no less responsible for the shape language takes: namely, the public social-interactional setting in which language is learned and used, and which is the central conduit for the historical distribution and development of language in populations.
C&C consider the implications of a dual-inheritance model by which human genetic evolution proceeds in parallel with much faster processes of historical evolution of culture in domains such as kinship, social values, technology, and language (Cavalli-Sforza & Feldman Reference Cavalli-Sforza and Feldman1981; Durham Reference Durham1991; Richerson & Boyd Reference Richerson and Boyd2005; cf. Enfield Reference Enfield2005; Reference Enfield2008, for language). A great discrepancy between the high speed of language change and the relatively low speed of genetic evolution drives C&C's argument that language (evolving historically) is adapted to the brain (evolving phylogenetically), rather than the other way around. This requires that language change be analyzed in evolutionary terms. To this end, C&C draw a parallel between the language system and the organism, as others before them have done. However, this may not be the most apt analogy. In genetic evolution, the organism serves as a vehicle for the replication of genes. In language, the vehicle for replication of linguistic units (e.g., words or constructions; Nettle Reference Nettle1999) is not the language system as a whole. The larger system is an aggregate of interrelated linguistic items. It may be stored in individual brains, or in linguistic descriptions, but it is not a vehicle for replication of linguistic units.
The vehicle by means of which linguistic items are used and diffused is the speech act (or utterance; Croft Reference Croft2000). Through being used in speech acts, linguistic items maintain circulation in populations; and it is through this circulation that selection of linguistic variants takes place. Diffusion of linguistic variants involves not just brains, but a see-sawing process, from private mental states (conceptual representations, communicative intentions), to public states of affairs (speech acts), back to new brain states (interpretations of speech acts leading to new conceptual representations, new communicative intentions), and on to new states of affairs (more speech acts). It is a continual chain of transition from private to public to private to public, and so on (Sperber Reference Sperber, Enfield and Levinson2006) – as is the case in the historical evolution of cultural variants more generally (Richerson & Boyd Reference Richerson and Boyd2005; Rogers Reference Rogers1995).
Linguists of many stripes recognize the privileged status of something closely akin to the speech act as the basic shape for linguistic organization: variously described, for example, as the “clause” (Foley & Van Valin Reference Foley and Van Valin1984), “intonation unit” (Chafe Reference Chafe1994), “turn-constructional unit” (Sacks et al. Reference Sacks, Schegloff and Jefferson1974), “growth point” (McNeill Reference McNeill1992), and so on. Why should just this unit constitute the privileged shape for linguistic organization? The answer is that the speech act or utterance is a basic unit at the level of informational delivery; that is, an utterance conveys one idea at a time (Pawley & Syder Reference Pawley, Syder and Riggenbach2000), thereby effecting one increment of social action at a time (Heritage & Atkinson Reference Heritage, Atkinson, Atkinson and Heritage1984). Is there something about the brain that privileges linguistic units of just this shape? Or are there other reasons that the speech act (utterance, clause, turn, etc.) should be a privileged shape for linguistic structure?
There is reason to think that the one-speech-act-at-a-time structure of grammar is shaped by contingencies of the conversational settings in which language is learned, used, and diffused. Human social interactions are sequences of moves and counter-moves (Goffman Reference Goffman1964), each move being a response to an immediately prior move, and a stimulus for a subsequent move (Sacks et al. Reference Sacks, Schegloff and Jefferson1974). This pattern of rapid response and counter-response is what characterizes free conversation, the basic format for language acquisition and use in everyday life.Footnote 1 Once we view the use of language in human interaction as a type of joint activity (Clark Reference Clark1996b), and not just an event of information processing that happens to involve more than one brain (Hutchins Reference Hutchins1995; Reference Hutchins, Enfield and Levinson2006), then we see how language is shaped by the sequential contingencies of social interaction. Grammar's basic “chunking” delivers minimal units, such that each unit contributes to a sequence of social interaction as an adequate response to a prior unit (Schegloff Reference Schegloff, Enfield and Levinson2006). To be effective, any such unit increment must not only be successfully formulated by a speaker, but must also be successfully comprehended by a listener. The organization of language in chunks of “turn” size gives interlocutors the opportunity to monitor misfirings and misunderstandings as soon as they happen, and to correct them if necessary (Schegloff et al. Reference Schegloff, Jefferson and Sacks1977). This co-contingency of unit contribution and response may be argued to serve as a direct determinant shaping linguistic organization, including many of the properties of “Universal Grammar” that C&C are looking to explain without reference to an innate language faculty (cf. Tomasello Reference Tomasello1995; Reference Tomasello2004).
In sum, the acquisition and use of language involves not just the brain's private structures, but also the public patterns of co-dependent, interlocking contributions to sequences of social interaction. The preferred locus of grammatical organization – the speech act – is an optimal solution to multiple constraints of information-packaging in an environment where not just formulation of social action, but the possibility of timely monitoring of intersubjective understanding and appropriateness of response is as definitive of linguistic well-formedness as any arbitrary unit of sentence-level syntactic structure.