Sunday, 22 February 2015

Why Chomsky might be right about the evolution of language... but probably isn't. Part 1

This is a two-part blog post. Part 1 explains why Chomsky might be right, the second part will explain why I think he isn’t.

Chomsky says language didn't evolve - according to psychologist Frederick Coolidge, who recently wrote a blog post entitled "Why Chomsky is wrong about the evolution of language." Citing a 2014 paper by Bolhuis, Tattersall, Chomsky & Berwick as evidence, Coolidge makes two major claims. First, that Chomsky denies language evolved, but appeared suddenly and was not subject to natural selection. Secondly that Chomsky denies genetic evidence, comparative animal studies, neurophysiological evidence and childhood acquisition theories, all of which contradict his language origin theory. In this blog post I’ll deal with his first point and the second in my next post.

I'm a big fan of Coolidge's work in cognitive archaeology, in particular his collaboration with the archaeologist Thomas Wynn in their two excellent books: "The Rise of Homo sapiens - The evolution of modern thinking (2009)" and their more recent volume "How to Think Like a Neanderthal (2012)." Both books have a traditional cognitive psychological perspective and are strongly committed to the ‘mind as a computer’ metaphor, utilising for example a version of Baddeley & Hitch's model of working memory, complete with concepts such as executive control. However, they are excellent summaries of archaeological data (especially for a non-archaeologist such as myself) and they also produce some intriguing postulations, such as their theory that the transition of our ancestors from arboreal to terrestrial environments necessitated a change in sleeping patterns, which could in turn have played a role in imagination and the creation of new tools (I dream of electric stone hand-axes).

On the other hand I'm not a big fan of Chomsky's theoretical view of language and its origins. His argument that language must be innate and that it appeared suddenly and perfect "like a snowflake," as he said in this 2012 talk, has seemed to me to be both outdated and doggedly committed to a rationalist perspective that I instinctively distrust (let’s not dwell too much on my own contradictions here).

However, having read Chomsky et al. (2014) paper, I think Coolidge has misrepresented Chomsky's position somewhat. Chomsky does not deny that language evolved. Indeed the very first line states the evolution of language is uncontroversial and that the question is how language evolved. Chomsky defines language in this paper extremely narrowly, separating it into two parts. The first, which he has elsewhere called the broad language faculty (FLB), evolved, by natural selection, for communicative and social purposes. The second, which he has elsewhere called the narrow language faculty (FLN), he has reduced to a single property of merge, which is the mental ability to combine two syntactic elements into a novel set. Merge then exapted the pre-existing neural and multi-modal physical architecture of the FLB to give physical expression to this mental ability, while mappings to the conceptual system gave mental expression – or a ‘language of thought.’ This would include the ability to form recursive syntactic structures, but would not necessitate them, handily getting around the problem of some langauges such as the Piraha apparently not using recursion. This gave birth to modern humans, culture and everything this entails.

Natural selection would play an important role in spreading merge, because merge is such a powerful tool, it would have swept through ancient hominid populations in a similar way to adult lactose tolerance later sweeping through dairy populations. But the appearance of merge itself was sudden in evolutionary terms and probably the result of a chance mutation.

I’m sceptical about Chomsky’s definition of language, but if you accept his definition, I think his proposed evolutionary mechanisms are plausible – and not all that different from Wynn & Coolidge’s. Furthermore I think some work in embodied cognition supports this plausibility.

A major point in Roy Pfeiffers lecture on Robotics was that relatively simple mechanisms can produce very complex and varied behaviour, as long as the environmental constraints are conducive to this behaviour. For example, his Robot which walks bipedally, with such a natural gait, has no power or control mechanisms but must walk in a straight line along a small incline.

Merge then, could be seen as such a simple mechanism, which then interacts with the structure and constraints of the pre-existing hominin communication system, including the rest of the organism, other organisms, the physical environment, to produce fully modern representational thought and language. I’m not arguing that Chomsky has an embodied perspective here (he most certainly does not!), I’m just arguing that the principle of complex behaviour resulting from simple mechanisms could be applicable to his view of language and that applicability makes it more evolutionarily plausible than Coolidge gives credit for, because Chomsky is actually postulating the sudden appearance of a simple, rather than a complex mechanism.

In light of all this, why then does Coolidge claim Chomsky doesn’t think language evolved? I think part of the problem might be they have different understanding of what is meant by evolution. Coolidge appears to have a fairly standard ‘gradualist’ perspective on evolution and his theory on the evolution of the human mind is a detailed account of gradual change, containing multiple stages. Complex organs, like complex behaviours do not simply pop into existence, whether it is the complexity of an eye or a lung, the ability to make an Acheuleun handaxe or hold a conversation. They must have had transitional forms and these forms were shaped by natural selection.

Chomsky, however, relies heavily on the ideas of Gould/Lewontin (indeed, Lewontin is one of Chomsky's collaborators on other papers), who emphasised the importance of exaptation, where a feature which evolved for one purpose, is utilised for another. Feathers are a classic example, originally evolving for heat regulation, then exapted for flight. Natural selection is still central in this view, but it is not the only force shaping change, with genetic drift and chance playing a greater role.

In fact, Coolidge has proposed similar mechanisms in his own work. In their book: "The Rise of Homo sapiens - The evolution of modern thinking (2009)," Wynn & Coolidge postulate that the final major advance in modern human cognition was a mutation which increased working memory and enabled advanced executive functioning. They argue this small change resulted in a series of complex changes including advanced contingency planning and a transition from symbols used as markers, to the emergence of fully modern symbolic thought, as evidenced by the archaeological discovery of complex figurines. Is this so different, in terms of evolutionary plausibility, from Chomsky’s proposed emergence of merge?

However, although Chomsky's theory on the evolution of language is plausible I don’t think it's correct, for reasons I will elucidate in my next post.


  1. Interesting discussion. Thanks. Here's a relevant paper that argues that Chomsky has focussed inappropriately on inner speech:

    1. Thanks Fred, that’s an interesting paper and it makes some powerful points about the nature of inner speech and how it is not simply an internal homolog of vocal speech. I think Chomsky's Minimalist Thesis is flexible enough (some might say slippery enough) to survive some of these objections, but it’s given me a couple of good ideas for my next post.