The Language Myth: A Review

Evan’s ‘The Language Myth’ is an excellent reply to the accepted dogma that language acquisition can only be explained by postulating an innate domain specific language faculty. The book covers a lot of ground and evaluates virtually all of the arguments for linguistic Nativism, yet despite agreeing with the conclusion that there is no language instinct, I was not impressed with aspects of the book; at times it reads like a wild caricature of generative grammar. Such caricatures will serve to turn off generative grammarians from reading past the first chapter. Generative grammarians are no strangers to caricature. Chomsky’s review of Skinners ‘Verbal Behaviour’ was a complete mess and didn’t really deal with any of the facts as Skinner presented them. Pinker attacks those who do not agree that the evidence supports a language faculty by claiming they hold a blank slate view of human nature, this is despite the fact that these supposed blank slate theorists use innate architecture in their theories. If you disagree with Pinker and Chomsky their first reaction is to ridicule and misrepresent you. But I don’t think the correct response to caricature is further caricature. This approach makes dialogue between the disciplines impossible. And it creates an army of followers who believe the caricatures and refuse to look at empirical data relevant to the subject. The tone and caricatures used by Evan’s are counterproductive and other criticisms of the idea of a language instinct are far superior; for example, Sampson’s ‘The Language Instinct Debate’ and Lappin and Clark’s ‘Linguistic Nativism and Poverty of Stimulus Arguments’. Pinker and Chomsky have never replied to Lappin, Clark, or Sampson presumably because they know the empirical evidence provided by Sampson et al refutes their belief in a language instinct. I expect them to reply to Evans and to point out the caricatures in the book. This is a good tactic from them refuting a few caricatures presented by Evans is easier than dealing with the mountains of empirical evidence which counts against their theories.

Evans made four claims in Chapter 1 which were inaccurate:

  • He claimed that Generative Grammarians argue as though all languages are like English. This is false. Rather they claim all languages including English follow universal principles which deviate because of parametric variations. They set themselves the task of empirically discovering what the universal principles and parametric variations are. They study languages from all over the world including Hebrew (one of the first languages studied by generative grammarians), Japanese, Chinese, Irish, Piradha etc.
  • He argued that generative grammarians do not do comparative studies of human and non-human communicative systems. This is false. See, for example, Chomsky, Fitch and Hauser (2002) and the follow up papers.
  • He attacked Chomsky’s language is a mutation theory, and implies that this view is the standard one in generative grammar. This is not true. There is much debate on the evolution of language within Generative Grammar. See Jackendoff and Pinker (2005) Berwick (2009). Etc it would be much fairer to address the diversity in the literature instead of focusing on Chomsky’s minority position.
  • He called Poverty of Stimulus Arguments nothing but appeals to incredulity. The idea being that since I cannot think how language is acquired then it must be innate. This is a caricature. Poverty of Stimulus arguments are usually precisely structured and purport to show that since it is impossible for a child to learn x because they have no experience of x, and they don’t try false approximations of x, and even if they did they would not be corrected. Therefore the child cannot learn x by induction and trial and error. This is a perfectly sensible approach and not an appeal to incredulity. However we now know that children are exposed to x in their primary linguistic data, they do try false approximations of x, and they are corrected for these approximations. So the poverty of stimulus arguments are empirically refuted, though there is nothing wrong with that mode of argumentation.

In chapter 2 Evan’s sets out to refute what he claims is a key myth of the language instinct paradigm:

“Myth: Language is the preserve of humans, and humans alone; it cannot be compared to anything found amongst non-humans, and is unrelated to any non-human communicative activity” (The language Myth p. 27)

Chomsky does spend a lot of time arguing for the uniqueness of human language. And to some degree this is justified. There are clearly massive differences between human language and the communication abilities of other creatures whether bees, monkeys or starlings. Evan’s of course doesn’t deny any of this:

“Despite the range of communicative systems evident in the animal kingdom, the complexity of human language, the nature and type of texts it allows us to signal, how it is acquired, and crucially the range of functions it facilitates, are both a different quality and a different level of complexity from any other” (ibid p. 63)

Yet despite agreeing that human language is far more complicated and richer than any other animal language Evans spends the entire chapter discussing various different linguistic abilities that animals share with humans. This is because he wants to show how human language evolved out of the communicative abilities of the common ancestor that humans shared with Bonobos and Chimpanzees. He wants to contrast his position with Chomsky’s argument that language emerged suddenly as a result of a random mutation. He argues that this Chomskian position makes language a kind of singularity. Evans thinks that the empirical studies he brings together show that aspects of human language are shared with a variety of other different species. Thus he speaks of the bee waggle dance, of monkey’s who use suffixes to alter the meaning of sounds, and of Starlings who can master recursion.

The fact of Starlings mastering recursion is relevant because of claims by Chomsky et al (2002) that recursion a unique and universal feature of human language. It is obviously not unique if Starlings can master it, while Dan Everett’s research with the Piranha indicates that recursion may not be a universal feature of natural language either. Now while I think Evans makes some good points against Chomsky’s views on the evolution and uniqueness of human language he doesn’t make it clear enough that Chomsky’s views on this topic are a minority. The minimalist programme is not really accepted by the majority of generative linguists. Chomsky et al (2002) divided language into the faculty of language narrow (Recursion) and the faculty of language broad the conceptual intentional system and sensory motor system. This division seems to me to have resulted from accepting Chomsky’s minimalist programme and I am not sure there is much reason to divide up the supposed language faculty in the way Chomsky proposes. So I don’t think refuting Chomsky’s speculations on evolution and minimalism really affects the idea of a language instinct, not as the instinct is understood by the majority of linguists. But that said I do think Evan’s arguments to show a lot of weakness’s in Chomsky’s position.

It is hard to know what to make of this chapter. He does show, by means of comparative data of other species, that Chomsky’s claims about recursion being a unique feature of language are wrong. He also shows that Hockett’s supposedly unique design features of human language are actually not uniquely human and the features are shared by many other animal species. However the whole point of discussing what communicative abilities humans and animals share is to indicate that language emerged gradually from our ancestor species and didn’t require a sudden leap/mutation to produce this entirely new thing. But theorists like Pinker and Jackendoff do not support this sudden leap story, they focus on selection for communication. Evans will have to deal with their arguments in detail. In fact if he wants to critique generative grammarian’s views of the evolution of language he will need to do a detailed literature review of the main papers on the topic and not merely focus narrowly on Chomsky’s minority position.

In chapter 3 Evan’s addresses the question of whether language is innate. A lot of his arguments are supported by empirical evidence from Tomasello. Evan’s cites approvingly the following Tomasello quote:

“The conclusion in the case of individual differences and the language acquisition process is thus that input does matter. Children learn what they hear, and different children hear different things and in different quantities. What this suggests is that language acquisition is not just triggered by the linguistic environment, as proposed by generative grammarians, but rather the linguistic environment provides the raw materials out of which young children construct similar their linguistic inventories. The fact that most adults end up with fairly similar (though not identical) linguistic inventories does not negate the obvious fact that early in development children can only learn what they are exposed to. It is also useful in this context to note that when pattern finding computer programs are given CDS as input, they are able to group together, by means of distributional analysis, linguistic items in a way that would seem to be psychologically realistic for young children.” (Tomasello ‘Constructing a Language’ p. 110)

The paper Tomasello is referring to which shows computer programmes learning to group together linguistic items in a psychologically realistic way is: ‘Distributional Information: A Powerful cue for acquiring syntactic categories’ Redington et al. (1998).

It may seem that he is question begging against Nativists who claim that we don’t need exposure to learn certain syntactic constructions. However Evans (and others) have provided evidence to show that the amount of linguistic exposure a child receives is directly proportional to their linguistic abilities.

Some of this evidence is:

  • Hart and Risely (1995) ‘Meaning Differences in Every Day Experience of Young Children’
  • ‘Semantic generality, input frequency and the acquisition of syntax’ Theakston et al. (2004)
  • Naigles, L. R. And E. Hoff-Ginsberg. (1998) ‘Why are Some verbs learned before other verbs? Effects of input frequency and structure on children’s early verb use’
  • Huttenlocher, J. M. Vasilyeva and P. Shimpi. (2002) ‘Syntactic Priming in Young Children’

These papers are merely the tip of the ice-berg of empirical research, and clearly show that despite what generative grammarians claim, a child doesn’t just have their universal grammar triggered by a bit of linguistic exposure. Rather the linguistic environment a child grows in will be directly relevant in determining the richness of the language a child will learn. Evans does well in bringing together empirical data which seriously undermines the claims of those who support the principles and parameters approach.

One area of his discussion of innateness that I think could have been improved was his discussion of poverty of stimulus arguments. Evan’s does mention Pullum and Schulz’s empirical research on a child’s primary linguistic data.  But I thought a more careful discussion of the issue was warranted. Chomsky (and later Pinker), used Auxiliary  Inversion in question formation as evidence that children could learn the rule for structure dependence without receiving any evidence from their linguistic environment. In his 1975 ‘Knowledge of Language’ Chomsky has argued that a child could go much or all of his without receiving evidence relevant to learning the rule. Pullum and Scholz (unlike Chomsky or Pinker) actually looked at the linguistic exposure that children actually receive, and they showed that Chomsky and Pinker were wrong. Children are routinely exposed to enough evidence in their linguistic environment to learn the relevant rule (Sampson replicated this finding). Lappin and Clark in their ‘Linguistic Nativism and Poverty of Stimulus Arguments’ showed that domain general machines could learn the rule from less exposure to the linguistic data than a child receives. This is clear evidence against the claims of linguistic Nativists. Their reaction to this refutation has been both comical and sad at the same time. The reply by Chomsky et al has been that the rule wasn’t really meant as a poverty of stimulus argument, so the fact that it was refuted is irrelevant. This reply is strange because this poverty of stimulus argument is the most cited one in cognitive science. Pullum and Scholz cite eight different occasions that Chomsky uses the example (Chomsky 1965, 55-56; 1968, 51-52; 1971, 29-33; 1972, 30-33; 1975, 153-154; 1986, 7-8; 1988, 41-47). They also cite other Chomskian thinkers (including linguists such as Lightfoot, 1991, 2-4; Uriagereka, 1998, 9-10; Carstairs-McCarthy, 1999, 4-5; Smith, 1999, 53-54; Lasnik, 2000, 6-9; and psychologists such as Crain, 1991, 602; Macrus, 1993, 80; Pinker, 1994, 40-42, 233-234) who here endorsed the claim. However the second it is shown to be false Nativists  claim that it is not the real poverty of stimulus argument and argue there is some other piece of evidence that represents the poverty of stimulus argument.  In her ‘What’s Within’ Fiona Cowie spends a lot of time demonstrating a number of occasions that poverty of stimulus arguments have been refuted.  She then noticed the following depressing pattern:

The nativist- say, Chomsky- articulates a version of the argument. The empiricist counters it by pointing to its evidential short-falls and/or its failure to do justice to empiricism’s explanatory potential. But no sooner is one rendition of the APS cut down than myriad other variations on the same argumentative theme spring up to take its place. For every non-obvious rule of grammar (and most of them are non-obvious), there is an argument from poverty of stimulus standing by to make a case for Nativism. And for every such argument (or at least for the ones I have seen), there are empiricist counter examples of exactly the kinds we have reviewed in this chapter, waiting, swords at the ready, to take it on.  ( Cowie: 1999, 203)”

This shows how difficult it is to get a Nativist to accept any kind of refutation. I think that Evans book would have been much better if he addressed this issue.

One area where I disagree with Evans (and Tomasello) is that I disagree with their contention that children are not corrected for grammatical mistakes. Contrary to their views I think that implicit correction does indeed play a role in helping the child to learn his first language. There is a lot of empirical research into this issue that supports the view that correction plays a role in a child acquiring their first language:

(1) Nelson, K.E., et al. ‘Maternal Input Adjustments and Non-Adjustments as Related to Children’s Linguistic Advances and to Language Acquisition Theories’ (1984)

(2) Demopoulos, M.J. et al. ‘Feedback to first Language Learners: The Role of Repetitions and Clarification questions’ Journal of Child Language 13 (1986)

(3) Saxton, M. ‘The contrast theory of negative input.’ Journal of Child Language 24 (1997)

(4) Saxton, M. ‘Negative Evidence and Negative feedback: Immediate effects on the grammaticality of child speech’ First Language 20 (2000)

(6) Choinard and Clark ‘Adult Reformulations of Child Errors as Negative Evidence’ Journal of Child language (2003).

Again this data is just the tip of the iceberg but the above papers shows both that the child is corrected for ungrammatical speech and makes use of these corrections. There is a lot of empirical data on both sides of this particular debate and I don’t mean to imply that all the evidence supports the view that children make use of corrections in learning their language. Rather I just mean to point out that Evan’s complacent assertion that correction plays no role in them acquiring a language needs a much more robust defence than he supplies.

Having dealt with the idea of innateness he goes on to attack the idea of universal rules that all that the estimated 7000 languages in the world share. This chapter is a bit quick but he does nicely make the point that no two generative grammarians seem to agree on what these supposed universal rules are. Furthermore as we study more and more of the world’s languages the less sense empirical support exists for these supposed linguistic universals.

He attacks the modularity of the mind thesis proposed by Fodor and supported by Pinker in his ‘The Language Instinct’ (1994). Pinker claimed that language and general intelligence were clearly different modules and this can be shown by the strange case of double disassociation between language and intelligence in Specific Language Impairment (SLI) and in Williams Syndrome.  In the case of Specific Language Impairment Pinker discussed the case of the KE family who allegedly had normal IQ but who had deficit in the ability to understand and produce grammatical sentences. He contrasted this case with the case of people suffering from Williams Syndrome, who Pinker claimed are linguistic savants with an extremely low IQ.

Evans does a good (but too brief) job of debunking this myth. He shows how the members of the KE family did in fact have IQ deficits, and their supposed grammar impairment could be mostly explained in terms of problems with motor control resulting from damage to the Foxp2 gene. While in the case of Williams Syndrome, Evans examines a variety of different empirical sources to show that people with Williams Syndrome are not the Savant’s Pinker claims. In fact Evans manages to demonstrate a variety of semantic, pragmatic, and syntactic difficulties with their speech. Overall he does a good job of critiquing this particular argument of Pinker. That said I think that people with Williams Syndrome do still seem to have excellent linguistic abilities relative to their general intelligence. So I plan to dedicate my next blog to discussing this issue in more detail than Evan’s manages to.

Evans also attacks the idea of mentalese proposed by Pinker. His attacks on the Computational Theory of the Mind which he incorrectly equates with Pinker’s Mentalese hypotheses are pretty weak.  He notes that there are three problems with the mentalese hypothesis:

  • How meaning arises from computation (Searle’s Problem)
  • How are symbols in mentalese interpreted? (The infinite Regress of homunculus (Ryle’s Problem))
  • The third problem is that mentalese theories are too syntacticocentric. (Chomsky’s ‘Colourless ideas sleep furiously’ doesn’t show syntax and semantics are separate)

A few points need to be made here. Firstly by equating mentalese (a language of thought) with the computational theory of mind Evans is really showing an outdated and unnecessarily restrictive conception of computation. The subject has moved on since the 80s (though apparently Pinker and Fodor aren’t aware of this), Bayesian and Connectionist models are far superior to the old symbol crunching computation and don’t rely on any idea of a language of thought.

Secondly, one would have thought that if he was going to bring in Searle’s Chinese Room Argument into the book he would have responded to the systems objection. The systems objection reveals the Chinese room argument to be an intuition pump which misdirects our intelligence in trying to understand what is going on with the machine. As far as I can see systems objection nicely refutes Searle. Why Evans doesn’t even mention the objection is a mystery to me.

Finally he sums up by defending the Sapir Whorf hypothesis against the criticisms made against it by Pinker.   He notes that Pinker creates a caricature and then destroys the straw man he has set up. Evans explicitly quotes Whorf to show that he does not hold the views that Pinker attributes to him. He then reviews some recent empirical research which demonstrates that a weak form of Linguistic relativism is possible. But this version doesn’t rely on anything like the linguistic determinism that Pinker attacks.

Over all I found ‘The Language Myth’ interesting and thought it offered some good criticisms of the idea of linguistic Nativism. However I thought his overblown rhetoric and caricatures of generative grammar took away from the central message of the book. I thought his attacks on the computational theory of the mind weak (and could have been made much stronger by incorporating the work of Evan Thompson and Terrence Deacon). But thought that his emphasis on embodied cognition and behavioural and cross linguistic data did show many weakness of linguistic Nativists approach to language acquisition.


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 )

Google+ photo

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

Connecting to %s