I have much sympathy with the general view advocated by Christiansen & Chater (C&C) in the target article. But I find some disturbing problems with its particular emphases.

In classic generative grammar, the explanandum is the distinctive features of human language, and the explanans is the postulated LAD (language acquisition device) or UG (Universal Grammar). Further, UG is said to reside in the human brain. Thus, classic generative grammar explains why language is the way it is because of innate properties of the human brain. How does this view differ from the position taken in the target article, that “language has been shaped to fit the human brain” (Abstract)? In both cases, the brain has certain properties first, and then language grows to fit those properties. There are differences between the two positions, but these differences are rather subtle.

The first difference to notice is the perfect tense in “has been shaped,” which invokes a diachronic, evolutionary process. The classic generative position accords no explanatory power to evolution, or to any diachronic processes (outside of ontogeny). In the classic view, languages grow their characteristic form, in toto once in each human lifetime, with every person's ontogeny, just as an organ grows. The slow, cumulative historical process advocated by C&C is surely plausible. But then the comparison with a bodily organ can no longer hold, because successive generations do not share the same physical organs. So the thing that grows, a language, has to be something outside of individuals. This risks getting Platonic, and C&C do not address this danger. It can be averted by stating that there is enough faithful, mind-external record of the whole structure of a language in the total ambient experience of each child. And this total ambient experience is provided by a society of other individuals who are all themselves at various stages of acquiring their own internalized representations of the language, in response to their own experiences of it. The “it” here is the language, a slowly moving target, as the authors point out. The classic I-Language/E-Language distinction is not undermined here. But the E-Language is given a status as an evolving social entity, which has a real causal influence, worth studying and not impossible to study, on the successive I-Languages of its speakers.

C&C argue against the view that human genes have adapted to a specifically “linguistic” environment. Throughout section 3.2 of the target article, there is a subtle vagueness about what is meant by “linguistic” and “language,” the entity to which UG may have adapted. It would be absurd to suggest that during the course of recent human biological evolution, fully-fledged languages like French, Arabic, and Tamil existed, and that such languages with their rich structure constituted the environmental challenge to which the human genome adapted. To attack this view is to attack a straw man, as no one advocates recent adaptation to the detailed structure of fully modern languages.

But the remote ancestors of human languages were only in the barest ways like modern human languages. (This excludes the remote possibility that modern human languages suddenly appeared, fully formed, as a result of some macro-mutation or saltation.) Just around the dawn of our species, there would have been rudimentary genetic dispositions, different from those in our ape ancestors, biasing us toward acquiring communication systems of some rudimentary sort, the remote forerunners of modern languages. Now these rudimentary forms were not yet “linguistic” in the modern sense, but they differed from anything previously existing. The target article only considers in a last brief section (sect. 8.2) the possibility of a kind of niche-construction in the specific domain of the unique type of communication embarked upon by proto-humans. Here C&C concede that there could have been natural selection for functional aspects of language. (Well, of course, whoever heard of natural selection for anything non-functional?)

Given that human language is spectacularly unique, and spectacularly functional, it would not be surprising to find some overlap between the unique aspects and the functional aspects. And this would give us domain-specificity of functional aspects of language. In this case, general cognitive capacities, used in non-linguistic domains, such as navigation, planning, and recall of past events (all without the aid of inner speech), cannot be summoned up to explain the relevant features of language. I will be little more specific than C&C about what these unique-to-language features might be. But a few reasonable candidates are: (1) the ability to acquire a massive inventory of form-to-meaning mappings (vocabulary); (2) the ability to manage a communication system with duality of patterning (double articulation), that is, a system with two combinatorial layers, phonology and syntax; and (3) the ability to manage nested long-distance dependencies, where the co-dependent items are in no way represented outside the head of the producer while speaking. All human languages have these properties which must be mastered in a few years by the child; and, barring pathology, children achieve mastery spectacularly well. These properties are arguably specific to language only, and there could well have been runaway biological adaptation, with a feedback loop, to the embryonic beginnings of systems with such properties. One force driving such adaptation would have been the peculiar hominid social setup, with large groups of mutually cooperative individuals. This all needs to be argued in detail, but the idea is as well developed as any alternative suggested by C&C in the target article.

Finally, ironically, the target article mentions the difficulty of adapting to a moving target. The innate UG that the authors attack is itself a notoriously moving target. Their target is “the extremely rich and abstract set of possibilities countenanced by the principles of UG” (sect. 3.2.3, para. 1). The generative enterprise based on this formulation has been turned upside-down by the more recent insistence on minimalism and the proposal that all that is distinctive about human language is recursion. So maybe C&C are pushing at an open door, or flogging a dead horse. The issues raised by the authors are important, and always have been. These issues need further work – and in a positive empirical direction, please.