For the last five decades, much linguistic research has adopted an interactionist position about the sources for Universal Grammar (UG). This approach partitions overt language universals into those determined by inborn linguistic structures and those shaped by language performance systems (e.g., learning and general cognitive factors). For example,
Many an aspect of adult … linguistic structure is itself partially determined by the learning and behavioral processes that are involved in acquiring and implementing that structure … some formally possible structures will never appear in any language because no child can use [or learn] them. (Bever Reference Bever and Hayes1970, pp. 279–80)
Christiansen & Chater (C&C) adopt an extreme version of the interactionist approach, eliminating the contribution of linguistically unique genetic bases for language in the brain. Their designated metaphor is language as an organism that has evolved to fit the human brain: “Language is easy for us to learn and use, not because our brains embody knowledge of language, but because language has adapted to our brains” (target article, sect. 1, para. 3).
Yet, mentioning “the brain” does nothing to enrich or support their claims. Unlike, for example, Deacon (Reference Deacon1997), C&C say nothing about which properties of the human brain are such that language has adapted to them. C&C restate much that has been said before about the influence of behavioral systems, while claiming – without demonstration – that they exhaust the constraints on linguistic universals. The critical claim is that the individual language learner/;user filters possible languages via general cognitive, communicative, and physiological processes. This is tantamount to the idea that language is a tool, ergonomically shaped to be maximally usable, no different in principle from a Boeing 707 or recipes for Salzbuger Nockerl. Yet no one would reasonably propose that airplanes or recipes are organisms that evolved to fit the human brain. C&C suggest that it is the alleged unconscious emergence of possible languages that makes language unique from other tools. Perhaps so, but they do not offer any theory of what the brain has to do with this that makes it reasonable to apply the organismic evolutionary metaphor only to language.
Characterizing language as evolving on its own also contributes unnecessary confusion because C&C restrict their notions of evolution to a classic model. For example, contrary to their interpretation, the adaptive complexity of biological organisms only marginally arises from random genetic variation, winnowed by natural selection. The many non-selective processes they cite (genetic drift, genetic hitchhiking, epigenetics, etc.), and others (e.g., evolutionary capacitors [Rutherford & Lindquist Reference Rutherford and Lindquist1998] and alternative splicing [Blencowe Reference Blencowe2006]), suggest that no complex organism, never mind language, evolved via selective winnowing of variants in the orthodox Darwinian model. Few biologists today adopt that pristine model, while some go as far as stating that “natural selection is just one, and maybe not even the most fundamental, source of biological order” (Gibson Reference Gibson2005; see also Carroll Reference Carroll2001; Sherman Reference Sherman2007; Wagner Reference Wagner2005).
After ignoring many lessons from current models of evolutionary processes, C&C proclaim it “astronomically unlikely” (sect. 4, para. 4) that non-adaptationist processes may have produced genuine evolutionary novelties. But, there are many “unlikely” cases in which non-selectionist mechanisms have been pivotal: the evolution of the genetic determination of sex (Quinn et al. Reference Quinn, Georges, Sarre, Guarino, Ezaz and Graves2007), the development of the eye (Sherman Reference Sherman2007), and the adaptive immune system (Hiom et al. Reference Hiom, Melek and Gellert1998). Appraised “likelihood,” of course, depends on the probability baseline that one adopts. Given a specifiable genetic configuration, combined with specifiable external and internal factors, the probability is close to one. If the assumed probability baseline is a random swirl of molecules (or a tornado in a junkyard, as Hoyle [Reference Hoyle1983, pp. 18–19] and Pinker [Reference Pinker1997] famously suggest), then indeed the probability of evolving a camera eye, or Universal Grammar – or a Boeing 707 – is vanishingly small. But why should one take anything like that as a relevant baseline?
C&C note further that, “Small genetic changes lead to modifications of existing complex systems … they do not lead to the construction of a new complexity” (sect. 4, para. 5). This statement is false when master genes (homeoboxes) are involved (see, e.g., Ronshaugen et al. Reference Ronshaugen, McGinnis and McGinnis2002; Sherman Reference Sherman2007).
In brief, insofar as one takes seriously “language evolving” as a meaningful metaphor, C&C's archaic view of biological evolution would render their arguments unlikely at best, and, most likely, wrong.
Secondly, the properties of cognition that C&C invoke to explain language occur in other species, including those that are less clearly present (e.g., hierarchical processing; Conway & Christiansen Reference Conway and Christiansen2001). Why did language not “evolve” in the service of these species? Humans must have innate equipment quite different from those species that also display their version of “general” human traits.
For “language as organism” to evolve under selective pressures enforced by the brain requires a suitable starting point: C&C suggest that a “proto-language” (sect. 7.1, para. 1) could fill the void. They reference computational models of the emergence of compositionality from a proto-language (cf. Smith et al. Reference Smith, Kirby and Brighton2003b), but ignore a critical feature of these models – the agents are equipped with a specific compositional mechanism for grammar induction. Thus, at a bare minimum, the brain must have contributed proto-language and such structural inductive systems as hierarchical processing, and recursion – abilities vastly undeveloped in primates (Conway & Christiansen Reference Conway and Christiansen2001).
Furthermore, C&C still face the problem of innovation from “proto-langauge.” What triggered the first truly structured sentence? The first recursive utterance introduced a transcendentally ungrammatical form. It is mysterious how this structure would ever be produced in the first place, let alone persist in the linguistic pool, if not for some innate capacity. At the very least, a possible pathway for the transition between unidentified cognitive mechanisms is needed, but C&C make no attempt to explicate one.
Finally, C&C report evidence of Broca's area involvement in an artificial grammar task and normal natural language processing (Petersson et al Reference Petersson, Forkstam and Ingvar2004) but omit drastically opposite data (Friederici et al. Reference Friederici, Bahlmann, Heim, Schibotz and Anwander2006; Musso et al. Reference Musso, Moro, Glauche, Rijntjes, Reichenbach, Buechel and Weiller2003).
