Dennett and the Evolution of Language


“To my own surprise, I’ve come to believe that there is an element of truth in the apparently less plausible Platonic story that’s easy to miss, one that seems to be almost completely obscured by the paradox that both Quine and Plato have described. It isn’t that our languages were deliberately invented by particular groups of people, legislators of syndics in the formal sense of these words, sitting around particular tables, at particular times in the past. It seems to me that they’re more like our dogs, our wolfhounds and sheepdogs and dachshunds, our retrievers, and pointers and greyhounds. We didn’t invent them exactly, but our ancestors did repeatedly make deliberate more or less rational choices in the process that made them what they are today, choices among a long series of slightly incrementally different variants, unconsciously shaping the dogs into precisely what their human breeders needed them to be. ( Cloud: The Domestication of Language p. 8)


Dennett’s chapter on the evolution of language is very vague. Dennett notes from the outset that discovering the origins of language is one of the major unsolved problems in science on a par with explaining the origins of life. His explication of the origins of language is meant as a general sketch of how a naturalistic account of the evolution of language could occur without any of what he calls “a skyhook”. By “a skyhook” he typically means some kind of miracle. In that sense his theory of language while vague is truistic as every adult doing research on the evolution of language rules out miracles. But Dennett also classes Chomsky’s theory as a skyhook; as he views it as an improbable theory or hopeful monster with little evidence to support it. In this regard I think Dennett’s reasoning is very weak he sketches his own vague theory about how the evolution of language occurred and offers a critique of Chomsky that is little more than a caricature.


Dennett segments the chapter on the origin of language into two sections


(1) The Chicken-Egg problem:


In the first section he argues that all he needs for the purpose of his book is a bird’s eye view on various different takes on the evolution of language. He notes that language may originally have been an ungainly mess that was later streamlined by evolution. As always in thinking about the origin of an entity or behaviour Dennett asks the question who benefited from the proliferation of languages? He notes that scholars traditionally would claim that people were the beneficiaries of language but Dennett argues that when we take the memetic point of view another approach becomes possible. Dennett argues that in the beginning language may have been more of a hindrance than a help.


            He gives a list some of the key functions that language served: (1) Communicative Utility (2)Productivity (3) Digitality (4) Displaced Reference (5) Ease of Acquisition. He briefly considers the possibility a key precursor to language would have been the unique way humans co-operate (a possibility discussed in detail in Tomasello 2014) and goes onto discuss the work of Boyd and Richardson on the precursors to language. Dennett notes that in our search for precursors we should open minded to precursors that weren’t necessarily immediately beneficial to humans:


“Instead of looking only at the prerequisite competences our ancestors needed to have in order for language to get underway, perhaps we should also consider vulnerabilities that might make our ancestors the ideal hosts for infectious but nonvirulent habits (memes) that allowed us to live and stay mobile long enough for them to replicate through our populations.” ( Dennett ‘From Bacteria to Bach and Back’ p. 254)


He asks us to consider the possibility that language was at first like a virus that infected us. He argues that the memes which stood the constant selection pressures of their environment; the ones who survived, would be structured in such a manner that they would fit the brains of their human hosts better than the other memes that didn’t pass selection muster. As a result of this he claims:


“Innovations in memes that made them more effective reproducers in brains that were not yet well designed for dealing with them could provide the early “proof of concept” that would underwrite, in effect, the more expensive and time-consuming genetic adjustments in brain hardware that would improve the working conditions for both memes and their hosts” ( ibid p. 255)


Given that culture is such a key feature of what made homo-sapiens special Dennett notes that any sensible theory will need to show why homo-sapiens developed culture but other mammals, as well as fish and birds have not. There must be some threshold that they needed to pass but didn’t; Dennett considers a variety of different possible innovations that could act as a threshold e.g. Bipedality, Social intelligence, Imitation etc. His discussion is curiously vague he considers a variety of different theories and notes that there isn’t enough data to solve the problem.


While Dennett’s sketch is thus far very unspecified it does bare a resemblance to other theories that were sketched in more detail e.g. Deacon’s theory as sketched in his Symbolic Species, and Christiansen and Chater’s theory as written in their “Language as Shaped by the Brain”. Dennett notes that his position is similar to the position held by Christiansen and Chater’s but he argues that they misconstrue the nature of memetics and that they exaggerate the case against genetic evolution (ibid. p.279)


(2) The Winding paths to human language:


In this section he discusses how once we have our proto-language in place it could have developed into full blown language: he considers three routes: (A) It began as a proto-language where short utterances like the Vervet Monkey’s alarm calls which were appropriate to the situation but which lacked productivity and any distinction between imperatives and declaratives (B) Perhaps a gesture language came first used for attention grabbing and emphasis (ibid p. 266) (C) Perhaps an auditory “peacock’s tail” arms race for vocal signals and improvisations. After this he goes on to explicate Hurford’s work on the origins of language he does a good job of describe Hurford’s work ( interestingly when I read Hurford I was critical of him from a partially Dennett point of view But Dennett endorses his views.


Dennett’s discussion of the fact that; though we can describe people’s linguistic behaviour interms of rules; this doesn’t mean they represent those rules in the brain (they may be free floating rationales) was important. Though it should be noted Quine and Davidson made similar points before him. To see the importance of considering free-floating rationales and the idea of competence without comprehension in relation to language evolution it is worth having a detour and considering a conjecture of Dan Everett’s in his recent paper ‘Grammar Came Later: Triality of Patterning and the Gradual Evolution of Language’.


 According to Everett there is evidence that language of some sort has been around 1 million years. Acheulean tools were not developed 1.76 million years ago. Chomsky and Berwick speculate that long lag between language development may have been a result of the fact that language did not exist at this time. Everett admits that this is a possibility but that it is just as likely that it resulted from ‘saticificing’ we look for good enough not for perfection. (Everett). Homo Erectus immigrated from Africa to Europe around 900,000 years ago, Everett argues plausibly that it strains credulity that such a feat would have been possible without some kind of language. Surely some kind of complex communication would have been necessary in order for boats to be built. Everett’s claim is plausible; it is indeed hard to imagine how a person could learn to build boats to sail across the continent without some kind of linguistic skills. However despite the plausibility of Everett’s claim there is reason to doubt it. Dennett’s uses the slogan “competence without comprehension” throughout his book, he gives examples of creatures who are competent at certain behaviours but do not internally represent reasons for those behaviours. He notes the behaviour of Antelope Stotting where the Antelope’s who can jump high when being chased are much less likely to be attacked by Lions. There is an evolutionary reason we can give for the Antelope’s behaviour as Dennett notes it is almost like the Deer is saying “Don’t bother chasing me I am hard to catch; concentrate on one of my cousins who isn’t able to stot-much easier meal” (Dennett ‘From Bacteria to Bach p. 91). One can also imagine the Lion using similar logic as a reason to guide his behaviour towards the stotting Antelopes. But of course implausibly attributing such complex propositional attitudes to the Antelopes and the Lions is not needed. The logic of natural selection dictated that stotting Antelopes survive better than their non-stotting cousins and Lions who attack stotting Antelopes are less like to feed and hence to survive. There is a reason why the Antelope behaves as it does but the Antelope doesn’t need to represent these reasons. Similar considerations apply to the behaviour of newborn Cuckoo’s kicking the eggs of non-Cuckoo’s out of the nest as soon as she is born. There is a reason for the Cuckoo’s behaviour but the Cuckoo’s doesn’t have to represent those reasons. Similar considerations apply to Australian Termite castles; the Termites build the complex structures without speaking to each other and instructing each other (ibid p. 238)


            A defender of Everett’s thesis could argue that the case of building a termite castle is not remotely comparable to building boats to sail to another continent. But this is not necessarily the case. Dennett cites a passage from Rogers and Ehrlich (2008):


“Every boat is copied from another boat…Let’s reason as follows in the manner of Darwin. It is clear that a very badly made boat will end up at the bottom after one or two voyages and thus never be copied…one could then say, with complete rigor, that it is the sea herself who fashions the boats, choosing those which function and destroying others” (ibid p. 214)


Now this is a striking passage where the sea plays a role of “designer” of the boat. Various different shapes are tried but only the shapes that work are selected.  Foresight is not a necessary condition of building complicated ships rather trial and error “decides” what the best design is. People can copy designs that appear to work and modify them in ways that may or may not work. Furthermore the people who built boats that (with the help of the sea or natural selection), obviously wouldn’t have to represent where they want to go. An instinct for discovery would do the job to get them sailing, and luck as well as “design”  would play an important a role in whether these people ended up in another continent. Everett could be right that language played role in the Homo-erectus emigrating to Europe but he hasn’t established this position nor ruled out the alternative theories. One of the strengths of Dennett’s book is that he gets us to search for ‘Free-Floating Rationales’ for evolutionary cheap tricks which give creatures the competence to perform a task without the creature necessarily having comprehension of why they are doing what they are doing. Dennett offers us an alternative to Everett’s proposal and thus gives us further theory to test. Everett’s story still has some plausibility to it but considerations of Dennett’s idea of ‘Free-Floating Rationale’s’ show that things are not as clear cut as Everett seems to think.


But while there are strong points to Dennett’s book his discussion of Chomsky was a non-engagement; he didn’t even try to understand where Chomsky was coming from. Dennett has long been a critic of Chomsky’s views on the evolution of language. In his ‘Darwin’s Dangerous Idea’ Dennett discussed how he attended a conference on AI and psychology in Tufts in 1978 and was shocked to see Chomsky reject engineering approaches to the faculty of language. Chomsky seemed to be fixated on language being something which could either be described interms of fundamental laws of physics or failing that as something which we could only describe in a similar manner that literary theorists describe human behaviour.  Dennett whose life work has involved trying to reverse engineer the mind obviously rejects this dichotomy of Chomsky’s noting that Chomsky is ignoring engineering solutions. I think there is a real degree of truth to this claim. But it is not the case that Chomsky ignores engineering solutions he rather adopts a different philosophy of science than Dennett. Linguist and Chomskian Cedric Boeckx discussed Chomsky’s attitude to science in his (2006) ‘Linguistic Minimalism’. Boeckx correctly noted that that Chomsky’s attitude to biology is different to a lot of main stream biologists. Chomsky as a formalist seeks to abstract from messy details and construct formal models which he thinks can capture certain core competencies. But a lot of evolutionary biologists seek to explore messy detail using functional explanations and reverse engineering. Both methods are scientific and Darwin used similar methods of a lot of evolutionary biologists while Newton’s used the methods preferred by Chomsky. Boeckx cites Freeman Dyson’s paper ‘From Manchester to Athens’ (1982) where he distinguishes between two styles of scientists the unifiers and the diversifiers. These different styles are best exemplified Newton and Darwin. Boeckx notes that looking back over the history of science we see that the greatest successes in science have been made by the unifiers. Boeckx is probably correct on this point but it is no indication that this pattern will continue nor is it true of evolutionary biology where the greatest success have been achieved by the diversifiers. Obviously in the debate between Chomsky and Dennett the unifier is Chomsky and the diversifier is Dennett. The point is that both positions are perfectly respectable positions to take in scientific theorising so Chomsky not adopting the engineering approach that Dennett recommends is not automatically a black mark against Chomsky. A scientific theory is to be judged by its successes in prediction and explanation and in this sense Chomsky’s theory has more than paid its way. Dennett’s conjectures on the evolution of language are just conjectures that may in the future pay their way but as of yet have not done so.


 Dennett’s criticisms of Chomsky in both his ‘Darwin’s Dangerous Idea’ and his ‘From Bacteria to Bach and Back’ centre on what Dennett perceives to be Chomsky’s uncomfortableness with language being tool created by natural selection:


The Idea that words have evolved by natural selection was obvious to Darwin, and one would that Noam Chomsky, who pioneered the idea of an innate Language Acquisition Device in the brain, would look with favour on evolutionary accounts of how languages in general, and words in particular, came to have their remarkable features, but he has disparaged evolutionary thinking in linguistics in almost all regards” ( ‘From Bacteria to Bach’ p. 187)


But Chomsky made the claim suspect in many quarters through his adamant resistance to any attempt to account for the design of the LAD by natural selection!” (ibid p. 277)


If Dennett was making this claim in 1995 he would have some justification as Chomsky’s few remarks on the evolution of language were pretty sceptical of the possibilities of us being able to explain the evolution of language. But since that that date Chomsky has written dozens of articles and co-authored a book explicitly designed to demonstrate how language evolved. So Dennett’s remark is simply incorrect and has little to justify it. Chomsky and Berwick in their recent book on the evolution of language ‘Why Only Us’ do argue that merge that was selected for its use in thinking. They call this the faculty of language narrow. But they also discuss the faculty of language broad which consists of the sensory motor system and the conceptual intentional system. Both the sensory motor system and the conceptual intentional system are according to Chomsky adaptations built up by tinkering processes of natural selection. So it is unfair to say that Chomsky is adverse to explanations interms of adaptations he uses them when he thinks they are appropriate. Furthermore Chomsky presents evidence as to why he thinks merge was a mutation by citing archaeological evidence that indicates a great leap forward about 60,000 years ago which indicates a sudden leap in intelligence. Now many theorists such as Everett, Tomasello, and McNeill have critiqued this but they addressed Chomsky’s actual argument and evidence. Dennett doesn’t do this instead resorting to a weak just-so story:


“What’s more, we could plausibly conjecture that Merge itself was no fortuitous giant step, no saltation in design, but a gradual development out of more concrete versions of Merge that can be seen in the manipulations of children (and adults) today: put block on block; use your big hammer stone to make a smaller hammer stone; put berry pile in bigger pile, put bigger pile in still bigger pile; put pie in cup in bowl in bag, and so forth. But are any of these processes real recursion? That is a misguided question, like the question: are the hominins real Homo sapiens? We know that gradual transitions are the rule in evolution, and a gradual emergence of (something like) real recursion-real enough for natural language would be a fine stepping stone, if we could identify it ( ibid pp. 279-280)


And then offering rhetoric:


“And notice that if something like Merge eventually proves to be a hard-wired operation in human brains, as Chomsky proposes, it wouldn’t then be a skyhook. That is, it wouldn’t be the result of a chance mutation that, by cosmic accident, happened to give our ancestors an amazing new talent. The idea that a random mutation can transform a species in one fell swoop is not a remotely credible just so story; it has more in common with comic book fantasies like the Incredible Hulk and all the other action heroes whose encounters with accidents grant them superpowers” (ibid p. 280)


Dennett is here engaging in nothing more than rhetoric and ignoring Chomsky’s actual arguments.


A further problem with Dennett is that while he is dismissing Chomsky’s views on the evolution of language (with little evidence) he sits on the fence on whether Chomsky’s proposed theory on the nature of language is correct:


I am not taking sides in the controversies, even where I have convictions; I have more than enough to defend without defending how much bias, how much “Universal Grammar,” as the Chomskyans say, must be genetically installed in the “Language Acquisition Device” nor do I stipulate in any way what form this biasing must or might take. One way or another, without deliberate (let alone conscious) theory construction of any kind, the infant homes in on habits that it will share with its parents, responding sensitively to  semantic information present to its senses.” (ibid p. 194 )


Dennett’s approach here is simply wrong; one cannot abstract from which theory of the nature of language  is true when you are sketching a theory of how language evolved. Noam Chomsky has long argued that any theory of the evolution of language will need to be informed about what language actually is. It is pointless in speculating about how language evolved without understanding the nature of language itself. Chomsky is surely right about this. Any theorist goes into a discussion of the evolution of language with a theory about what the nature of language is. A false theory about the nature of language may send one down a wrong path trying to discover how it evolved. Thus if you think that a key feature of language is that it is an internal computational procedure used for thinking primarily, or if you think that language is primarily a shared system symbols used to communicate meanings; these different theories about the nature and function will have serious effects on how you will understand what the archaeology is telling you about the nature of language. Thus  if we consider some facts about the evolution of our ancestors we will see how these facts will appear depending on the theory of language one accepts. In his (2016) ‘Grammar Came Later’ Dan Everett notes some key facts about our evolutionary history. Over 6 million years ago a new type of ape arrived on earth, this bipedal ape was called Australopithecus. According to Everett (2016) Australopithecus could recognise iconicity. To support the claim that they recognise iconicity Everett points out that they collected pebbles, he cites the example  Makpansgat Manuport pebble 3million years old which was an icon shaped like a face collected by Australopithecus. Everett goes on to note that up to  2.7 million years ago we have evidence of an icon shaped like an phallus called the Erfoud Manuport. Aside from our use of icons Homo Erectus was using crude Oldowan tools at 2.6 million years ago (Everett 2016).


            Now obviously for someone like Everett who thinks that the evolution of symbols is the key factor in the evolution of language these facts will be pure gold. Everett following Peirce notes that the movement from index to icon to symbol is one of the key features of our linguistic capacities. So the fact that Australopithecus was using icons 3 million years is clear evidence that our ancestors were starting on a process of developing a language 3 million years ago. Though obviously the use of an icon is not a linguistic practice but it is evidence of cognitive capacities which may have offered an entering wedge into language. However for a Chomskian who defines a language as the ability to use merge; language is an all or nothing capacity. So a Chomskian will be less than impressed with these empirical details and will believe that they are irrelevant to the details of what Chomsky thinks of as the key feature of language: Merge. We can see from the different ways that Everett and Chomsky treat archaeological details that your evolutionary theory of language is deeply connected to the theory you hold about the nature of language. Dennett being vague in his theory of how language evolved be an untended result of his not taking sides on debates on the nature of language.


Overall while Dennett’s discussion of plausible candidates for a naturalistic explanation of language explanation was interesting he left too much detail out. His competence without comprehension thinking tool is useful as a way of reminding linguists to not just assume that complex behaviours Must be evidence of comprehension. However his treatment of Chomsky was much too cursory to be of any real value.


1 thought on “Dennett and the Evolution of Language

  1. Ashok

    Dennett is bringing an important point which David also is missing… sense and thought and therefore language is indistinguishably inseparable..
    Language is actually an inhibiting process..analogous to the predominant inhibition activity of the brain… language as an item itself is only in human awareness as a form which actually is not an existent, separable from sense.. Wittgenstein is bringing this point to the front..I guess


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s