The notion of Universal Grammar (UG) is, or was, one of the most uneasy concepts in cognitive science. Not only has it ultimately failed to capture the enormous diversity of human language, but it also lacks credibility in evolutionary terms. Chomsky himself was driven to conclude that language could not be the outcome of natural selection, once declaring that “It would be a serious error to suppose that all properties, or the interesting structures that evolved, can be ‘explained’ in terms of natural selection” (Chomsky Reference Chomsky1975, p. 59) – although he seems more recently to have had something of a change of heart (see Hauser et al. Reference Hauser, Chomsky and Fitch2002). Christiansen & Chater (C&C) have done cognitive and evolutionary science an enormous service by pointing out the deficiencies of UG as a plausible concept, and giving breathing space to alternative, more naturalistic approaches.
I think, though, that their title is slightly misleading, as the brain is scarcely mentioned in their article. Rather, they make a good case for supposing that language is shaped by human knowledge, learning capacities, and culture, and is not some encapsulated system with its own rules. I would like simply to add some points that I think support their general position.
In an endnote, C&C suggest that their arguments have equal force regardless of whether language has its origins in vocalizations or in manual gestures (see Note 2 of the target article). It may be useful, though, to note that even present-day language can be either spoken or signed (Emmorey Reference Emmorey2002), and is indeed typically a combination of both (McNeill Reference McNeill1992). C&C do appeal to evidence from newly emerging sign languages to support their arguments concerning processes of grammaticalization, but for the most part they equate language with speech. Adding signed languages and manual gestures provides a better appreciation of the full repertoire of human language, and can help differentiate those aspects of language that depend on the particular medium of expression from those that may reflect more cognitive aspects. Thus, in section 6.2, C&C note that “The seriality of vocal output … forces a sequential construction of messages” – the transformation known as linearization. But this constraint is much less severe in the case of signed languages, which can allow some degree of parallel output, and spatial mapping.
The notion of a “mother tongue” originating some 100,000 to 200,000 years ago, and giving rise to present-day languages, also presumably applies to speech rather than to language itself. Present-day signed languages appear to have emerged independently of any spoken language. One example, mentioned by C&C, is Al-Sayyid Bedouin Sign Language (ABSL), which bears little relation to the other languages, spoken or signed, of the region (Sandler et al. Reference Sandler, Meir, Padden and Aronoff2005). I agree with C&C that this is not proof of an innate UG, but, equally, it cannot be construed as a descendent of any mother tongue. It also allows one to speculate that language itself goes back much further than the last 100,000 or 200,000 years ago, as proposed by Bickerton (Reference Bickerton, Christiansen and Kirby2003) or Cavalli-Sforza and Feldman (Reference Cavalli-Sforza and Feldman2003). I see no reason why our large-brained forebears of 500,000 or 1,000,000 years ago could not have invented something like ABSL, though perhaps with some vocal accompaniment (Armstrong & Wilcox Reference Armstrong and Wilcox2007; Corballis Reference Corballis2004; Reference Corballis2006).
One cognitive domain that may have influenced, and perhaps even shaped, the evolution of language is mental time travel – the ability to mentally relive events in the past (episodic memory) or imagine events in the future. This is arguably uniquely human (Suddendorf & Corballis Reference Suddendorf and Corballis1997; Reference Suddendorf and Corballis2007). Language seems almost tailor-made to allow individuals to share past experiences and future plans, with adaptive consequences (Corballis & Suddendorf Reference Corballis, Suddendorf and Pasternak2007). The features of language that may derive from mental time travel, or that perhaps co-evolved with it, include the following:
1. Symbolic representations allowing reference to actions, actors, objects, and so forth, that were part of episodes at times and places other than the present.
2. Ways of referring to different points in time, including past, present and future, and for making other temporal distinctions, such as action completed versus action ongoing. Languages have different ways of doing this. Many languages, such as English, have systems of tense, whereas others, like Chinese, have no tense but use other kinds of time markers (Lin Reference Lin2005). Culture may play a hand here; thus, the Pirahã seem merely to distinguish between events that are in the present from those that are not (Everett Reference Everett2005).
3. Generativity. The imagining of future events includes the generation of different possible scenarios – and even episodic memory appears to involve an element of construction – to the point that “memories” are often false (e.g., Roediger & McDermott Reference Roediger and McDermott1995). Closely related to mental time travel, I suspect, is fiction – the deliberate construction of stories that are not based on fact, but that nonetheless serve as surrogates for experience. Correspondingly, the generativity of human language is perhaps its most distinctive property.
If we marry the emergence of mental time travel with a language that was based on manual gestures, we can look nostalgically back to an era when we had time on our hands.
