CHOMSKY AND QUINE ON ANALYTICITY PART 2
SECTION 1: Types of Analyticity
In Part 2 I discussed Quine’s arguments against the notion of analyticity. It was noted that Quine was attacking three notions, (1) Analyticity as a defence of the a priori (2) Analyticity as a form of necessity, (3) Analyticity as a purely semantic truth independent of collateral information. Chomsky’s claim that an analytic/synthetic distinction was a fact of natural language was then discussed. Throughout the discussion I treated Chomsky’s arguments as pertaining to semantic analyticity. By semantic analyticity I mean analytic truth as being true in virtue of the meaning of the terms and independent of collateral information. In this Part I will consider whether Chomsky’s positive conception of analyticity can support a priori knowledge, and our grasp of necessary truths.
As we have seen above, when Chomsky is discussing language and mind he is doing so in a thoroughly naturalistic manner. Our language results from an innate module of the mind which he attempts to explain in a purely naturalistic manner. Chomsky tells an evolutionary story about the origins of language (See for example Chomsky, Fitch, and Hauser 2002). He argues that any analytic connections in natural language result from the internal structure of the innate language faculty. He also claims that human concepts have idiosyncratic features which were of no particular use to our ancestors so it is unlikely that they were formed through natural selection. He assumes that if our innate conceptual abilities were not formed through natural selection then they must have resulted from a random mutation. Obviously, conceptual abilities which resulted from a random mutation will in no way provide us with any kind of a priori justification.
When discussing the rules which govern our language, Chomsky has been quite explicit on the fact that we are not justified in applying the rules of natural language. He argues that we just automatically implement them in the way a computer implements a programme:
In short, if I follow R, I do so without reasons. I am just so constituted. So far, these conclusions offer no challenge to the account discussed earlier. I follow R because S` maps the data presented into S*, which incorporates R; then I apply R blindly. There is no answer to the Wittgenstein Sceptic and there need be none. My knowledge, in this instance, is ungrounded….I have no grounds for my knowledge in any useful sense of the term and no reasons for following the rules: I just do it. If I had been differently constituted with a different structure of my brain, I would have followed a different rule. (1986b, 225)
In the above passage, Chomsky is responding to Kripke’s rule following paradox. He is claiming that he does not have a solution to the paradox. He cannot say why calculating according to the plus function is justified whereas calculating according to the quus function is not justified. His answer is to point to the fact that we do just calculate one way as opposed to the other way, and to explain this fact by invoking innate constraints on the type of mathematical rules we are capable of following. The innate structure of our brain means that we interpret ‘+’ in terms of ‘plus’, however if our brain was structured differently we may well interpret ‘+’ in terms of ‘quus’. I will not here discuss Chomsky’s solution to the rule following paradox; my main aim is to show that his answer to Kripke shows the folly of reading Chomsky’s conception of analyticity as an epistemological conception.
Chomsky has explicitly claimed that people are born with innate concepts some of which will as a matter of empirical fact yield analytic connections. He has furthermore claimed in a number of places, that concepts of natural language do not refer to entities in the mind-independent world in the manner in which people like Quine, Frege, and Tarski believe they do. This being so, it is clear that on the Chomskian picture concepts like BACHELOR and UNMARRIED MEN do not mean the same thing because they pick out the same mind-independent entity, rather, they both mean the same thing because they both pick out the same concept.
This view of course puts Chomsky in a very weak position. It means that we cannot justify our sentences as being true by virtue of the meaning of the terms involved. Instead we just automatically categorise them so, and if our brain was constituted differently we would categorise them differently. A number of points follow from this immediately. The types of analytic connections which Chomsky defends have no epistemological significance. The analytic connections he is concerned with are not connections which justify us in holding certain beliefs true. So it could be argued that the analytic sentences which Chomsky is concerned with are of a philosophically uninteresting type.
However such an interpretation is complicated by the following quote from Chomsky.
These are truths of meaning, not of fact. The a priori framework of human thought, within which language is acquired, provides necessary connections among concepts, reflected in connections of meaning among words and, more broadly, among expressions involving these words… (2000b, 63)
When Chomsky talks about an a-priori framework of human thoughts and about necessary connections amongst concepts he seems to be making claims which his explicit theory would not allow. No contingent fact about our brains would justify us holding a belief true, while it is likewise highly unlikely that a contingent feature of our brain structure would somehow put us in touch with a necessary truth about the universe. So it seems that Chomsky’s claim about humans being born with analytic connections amongst concepts is inconsistent with his claim that analytic truths provide a priori justification or knowledge of metaphysical necessity. Throughout the remainder of this section I will endeavour to reconcile these seemingly inconsistent views.
Given that Chomsky has claimed elsewhere that our syntactic rules are not justified but are merely followed blindly, it is obvious that we should not interpret Chomsky’s claims about an a priori framework of human thought epistemically. Rather we should instead interpret an a priori framework of thought to mean an innate framework of thought. But this of course leads to the question of what Chomsky means by necessary connections amongst concepts? Consider the concepts: 2, 4, +, = these can be combined to create the necessary truth 2+2=4. Now philosophers such as Hume and Ayer have claimed that the truths of arithmetic are analytic truths. Given Chomsky’s various discussions of analyticity, and his views on concepts expounded in his reply to Kripke, it seems that Chomsky also considers the truths of arithmetic as analytic, in the sense of being true based on the meaning of the terms involved. But it seems that this type of analyticity is not strong enough to support metaphysical necessity.
In order for our analytic truths to yield metaphysical necessity, there would have to be some sort of correspondence between the truths of meaning which are encoded in the brain and necessary truths which exist in an unchanging mind-independent world. If the connection were a mere contingent one, then it would have been just the result of time and chance. To grasp the distinction between contingent and necessary truths using the framework of possible worlds is sometimes helpful. According to Chomsky, it is a fact of the human brain that some of the concepts we are born with will automatically yield analytic connections. Surely, though, this being so, it is also possible that the human brain could have been constructed differently. It is obviously possible that humans could have been born with innate concepts but no analytic connections between their concepts: call such creatures Fodorsapiens. If we think of the analogy between male and females who are both members of the species Homosapien, it is also possible that there were two different types of Homosapiens, who were distinct from each other but could reproduce with each other; call them Chomskysapiens and Fodorsapeins. Both strands of Homosapiens are descended from the same ape ancestors and they just diverged from each other because of a random mutation which did not affect their ability to reproduce with each other. We can imagine in our story that Fodorsapeins eventually died out because their slightly different DNA made them unable to fight of the common cold. The point is that there are many possible worlds which can be constructed in which it was Chomskysapiens who died out, or in which a random mutation changed which sentences the Chomskysapien finds analytic. The truths of meaning which Chomsky claims are component of the mind brain seem to be nothing more than contingent results of our evolutionary history, and his claim that they provide necessary connections remains unargued.
So it would seem that since the brain is the contingent product of its evolutionary history, then any concepts it is innately born with will correspond with a mind-independent reality, if at all, only coincidently. Here, of course, the type of analytic truths will be of paramount importance. If we are to count the truths of arithmetic as analytic, then we can sketch a story of how an innate number sense would be vital to survival. We could cite how a creature who could not calculate whether the number of foes was equal to the number of friends would not survive very long. We could point to creatures such as monkeys and dogs who indicate a grasp of a number sense to show how this sense is probably a universal feature of any creature who wants to survive in a hostile environment.
When one tries to explain other analytic truths such as ‘All bachelors are unmarried men’, or ‘All dogs are animals’, sketching a story in terms of survival value seems more difficult. The idea of ‘All bachelors are unmarried men’ being a necessary truth or a truth which is justified independent of the facts seems unintelligible. However the idea that the sentence is analytic because ‘bachelor’ and ‘unmarried’ correspond to the same mentalese concept is obviously intelligible. So the question we are faced with is the following: since analytic connections are not limited to the language faculty, what is their place in the overall taxonomy of the mind which Chomsky has constructed so far?
In a personal communication Chomsky tried to clarify some of these questions for me. He claimed that:
In this context, “a priori” means what Konrad Lorenz called “biological a priori,” that is, part of the cognitive equipment with which an organism organizes and approaches external data, mapping it into experience, and then interpreting it. The issue of justification doesn’t arise, any more than it does in the case of “book” rhyming with “took” (in my speech). This doesn’t conflict with Boghossian. He is talking about something different. (Chomsky: personal communication).
So Chomsky here agrees with my analysis of his views on analyticity. His views offer nothing towards helping us clarify such notions as a priori knowledge and necessary truths.
Even if Chomsky is correct that people are born with an innate system of concepts which yield analytic connections, this fact will still leave him and Quine in agreement on some central points. Quine was criticising how philosophers had used analyticity as a philosophically neutral explanation of a priori knowledge and necessary connections. He wanted to show that the only type of analyticity he could make sense of, stimulus analyticity, could not explain the philosophically meatier conceptions of the a priori and necessity. Chomsky’s alternative conception of analyticity, like Quine’s, does not purport to explain our grasp of necessary truths, nor any supposed a priori knowledge. So both thinkers are in agreement on this point at least. However, as we have already seen, Quine was not only arguing against analyticity in the sense of an explication of the a priori and necessity. He was also arguing against the semantical notion of analyticity. So one clear area of disagreement between both thinkers centres on whether there are analytic connections in the sense of sentences whose truth is fixed purely in terms of meaning independent of extra-linguistic fact? Quine gave arguments against this in the first four sections of ‘‘Two Dogmas’’, and these are arguments which Chomsky never managed to answer. Furthermore, Chomsky has offered little evidence in support of his positive conception of analyticity. So it is clear that Chomsky has not really answered Quine’s objections against the idea of an analytic/synthetic distinction.
Since Chomsky’s conception of innate analytic connections in natural language and Quine’s conception of stimulus analytic connections rest on such radically different pictures of how language is learned it would be foolish to say that they are in agreement on the nature of analyticity. However, it is significant that both thinkers are in general agreement as to the fact that analytic sentences do not explain a-priori knowledge, or necessary truths. So the debate between them on analytic sentences is an illusory one when it comes to metaphysics and epistemology. There is, however, a substantive debate between the two thinkers as to the nature of analytic sentences as a part of natural language and I will return to this point later in the essay.
SECTION 2: FACULTY PSYCHOLOGY AND ANALYTICITY
Chomsky has at various different times spoken about the language faculty, a science forming faculty, a folk physics faculty, and a moral faculty. To say that we have a faculty is to say that we have domain specific knowledge of something, and that this knowledge has not resulted from a general inductive procedure. A clear indication of a faculty operating in the brain is when as result of a disease or an accident, a person loses a certain ability (e.g. linguistic ability) but their general intelligence remains unimpaired. So, for example, people such as Marc Hauser have argued for the existence of a moral faculty based on the fact that selective brain damage results in people losing their moral sense but retaining their overall general intelligence. Likewise, Chomsky has discussed cases of people losing their linguistic abilities without it having any effect on their overall general intelligence.
Another key reason to believe that some abilities are the result of a faculty is that the ability arises at set times in the person’s development. Furthermore the ability arises in people at more or less the same time, independently of their general intelligence. Again Hauser argues for the existence of a moral faculty based on the fact that children uniformly answer tests in a similar way when at the same age of their development.
There is strong evidence that children come equipped to the world with intuitive folk physics and folk psychology. The fact that these two different domains of knowledge are not commensurate indicates that they may belong to different faculties of the mind. And there is independent evidence that folk physics and folk psychology are faculties of the brain. A person’s folk psychology does develop at set times in the child’s development independently of general intelligence. So, for example by the age of four children can typically pass the false belief test with the notable exception of autistic children, most of whom fail the test at this age. The fact that autistic children fail this test is instructive because they typically show no deficit in general intelligence. Furthermore, children who do in fact have a lower IQ, such as children with Down syndrome, manage to pass the test by the age of four. This shows that we may be justified in claiming that folk psychology is a distinctive faculty of the human brain.
Overall I would claim that there is some reasonable evidence (though far from overwhelming) to support the postulation of a faculty for folk physics, folk psychology, morality, language, amongst others. When Chomsky claims that humans are born with innate concepts which will automatically yield analytic connections he fails to detail the nature of these concepts, and whether these concepts are unique to particular faculties. Consider the following purported analytic statements: (1) All ducks are animals. (2) If John persuaded Bill to go to school then Bill decided/intended to go to school and did so without being under duress. (3) Three times five is half of thirty. It is a fair bet that most people would consider these statements to be analytic, though it must be admitted that some people may answer 3 by saying it is a priori synthetic. Let us assume for a minute that most people would consider all three sentences analytic. The question is: where, for a Chomskian, does their analyticity derive?
The subject matter of 1 is biology, 2 folk psychology, 3 is mathematics. All three statements are, to use the traditional definition, true by virtue of the meaning of the terms and not because of extra-linguistic facts. For Chomsky, as we have seen, the meaning which creates these analytic connections is something we are born with, not something which is true by convention. The three different statements contain concepts of agency, number, and biological entities. What we do not know is if those concepts grow in the brain separately from the different faculties, or whether they grow internal to the different faculties. So, for example, if a mathematical truth such as 2+2=4 grows internal to a faculty of mathematics, then damage to this faculty would not have any significance for our ability to understand analytic truths which grew internal to other faculties. This is a question which Chomsky has not in general been very clear on. He correctly notes that more empirical research is required before we can decide on what the correct answer to the question is, and that it is empirical research that is necessary, not a priori philosophical speculation.
Whatever the exact structure of the various faculties are, Chomsky thinks that if there are analytic connections in different domains they will emerge internal to the faculties. Chomsky is explicit on this point in a personal correspondence with me:
Given my language faculty, it’s a necessary truth that “book” rhymes with “took” and that “chase” entails “follow.” Horwich is quite right. This has no bearing on, say, whether arithmetic truths are synthetic a priori… On the Hume questions, we have to distinguish what our innate sense of number determines, and what is the truth about numbers (if we regard them as entities that exist independently of our cognitive capacities, as Platonic entities of some sort). The two presumably coincide, but we can imagine an organism in which they would not (and maybe that’s us, as a matter of fact). (Chomsky: personal communication)
One important point to note about this communication is that he here admits a kind of scepticism about the truths of mathematics. He claims that while it is possible that our innate number sense does correspond to mind-independent Platonic entities, we cannot say for certain that they do. Now, given that our best scientific theories about the world are couched in mathematical formulae, this is a startling admission. Chomsky is claiming that we have no justification for acting in accord with our innate number sense, and that we have no way of knowing whether our innate number sense corresponds with any facts in the objective world.
Furthermore, Chomsky is arguing that different types of analytic connections will emerge internal to our various different faculties. So we will have analytic truths derived internal to our innate number sense, and analytic truths derived internal to our language faculty etc. This means that Chomsky’s view on the nature of analyticity is parasitic on his views on the various different faculties of the mind. It also shows that truths which are typically grouped together by philosophers as analytic derive their analyticity internal to particular faculties. So Chomsky’s justification for arguing that analytic truths exist depends on accepting his controversial views on (1) innate concepts, (2) faculty psychology, (3) potential radical scepticism about the truths of mathematics. Chomsky’s conception of analyticity seems to ask more questions than it answers.
SECTION 3: CHOMSKY AND QUINE ON ANALYTICITY
In the previous sections we established that neither Chomsky nor Quine have a conception of analyticity which is rich enough to provide a priori justification and evidence of necessity. However, I also noted that, despite having this much in common, their views on the nature of analyticity diverge radically. In this section, I compare the two views and consider which view deals with the available data better.
Having already explained Quine’s negative critique of analyticity at the beginning of this chapter, I will not repeat the material here. In this section I will focus on how Quine explains away people’s intuitions of analyticity.
As is well known, Quine criticised the notion of analyticity as it was used by philosophers from Descartes to Ayer. Quine, of course, did not rest with such negative criticisms; he also explained why people held such intuitions of analyticity in the first place. Quine was well aware of experimental work which showed that people had an intuitive conception of analyticity. However, for Quine, such intuitions do not provide evidence that analytic connections actually exist in natural language. He notes that analytic intuitions typically set in where people have difficulty figuring out what the person who denies the truth of such a sentence may mean by its denial. This reaction is perfectly consistent with Quine’s web of belief story. A person who denied something which is deeply embedded into our web of belief would indeed result in an interpreter having difficulty figuring out what the person meant by the denial.
We saw above that Putnam criticised Quine’s blanket criticism of the notion of analyticity. Putnam correctly argued that the web of belief story does not account for trivial cases of analyticity. In Word and Object Quine told a story about analyticity which, he argued, was largely consistent with the story told by Putnam.
Quine noted that terms like ‘‘Indian nickel’’ and ‘‘Buffalo nickel’’ are not stimulus synonymous while ‘‘bachelor’’ and ‘‘unmarried men’’ are. Quine explains this disparity by noting how the different terms are learned. He argues that terms like ‘‘Indian nickel’’ and ‘‘Buffalo nickel’’ are learned by association with non-verbal stimuli. Terms like ‘‘bachelor’’ and ‘‘unmarried men’’ are learned by association with verbal stimuli. So, for example, one learns the meaning of ‘‘bachelor’’ by learning to associate it with ‘‘unmarried man’’. Obviously from a social point of view the stimulus meaning of the term ‘‘bachelor’’ will vary from person to person. A child can be told the meaning of the word ‘‘bachelor’’ by ostensive definition; however, obviously, the particular person who is being pointed to will not be representative of the class of people in the world who are bachelors. So a child who learned the meaning of ‘‘bachelor’’ this way would be unable to generalise the term using this technique. Even if they pointed to a vast number of different bachelors and named them the child would not be able to generalise the term because there are no particular characteristics shared by bachelors alone other than being unmarried men. If an adult wants to teach a child what a bachelor is he can do no better than tell the child that bachelors are unmarried men. In the case of the terms ‘‘Indian nickel’’ and ‘‘Buffalo nickel’’ things are different; we learn these terms through ostension. If we learn that ‘‘Indian nickel’’ and ‘‘Buffalo nickel’’ pick out the same object, we again do this inductively. Quine argues that some technical terms of science do not have a socially fixed stimulus meaning to govern how they are used. In this sense, Quine argues that they are like terms such as ‘‘bachelors’’ and ‘‘unmarried men’’. There is, however, one difference between the technical terms of science and terms like ‘‘bachelor’’; ‘‘bachelor’’ gets its meaning by being connected to ‘‘unmarried man’’, while in the case of technical terms their meaning is governed by their relation to a whole scientific theory.
Quine’s discussion of analyticity above is largely consistent with the story told by Putnam. Quine argues that a sentence like ‘‘All bachelors are unmarried men’’, appears to be analytic because we learn the meaning of ‘‘bachelor’’ through associating it with ‘‘unmarried men’’. It is not actually analytic in the sense of being immune to revision, nor does it provide us with any kind of a priori justification. The sentence merely has the appearance of being analytic because of its mode of being learned. As language evolves the sentence ‘All bachelors are unmarried men’ may no longer be considered analytic though in terms of present usage it certainly gives the appearance of being analytic. Likewise Quine’s web of belief story explains our grasp of less trivial sentences which have the appearance of being analytic, though as in the ‘bachelor’ case, he again accepts that such sentences are subject to revision as our language and theories evolve.
Like Fodor, and unlike Putnam, Quine does not beg any questions because he is not trying to explain the existence of analytic truths by using one-criterion concepts. Quine is merely arguing that since our manner of learning some concepts will be through one-criterion concepts this explains why some people have analyticity intuitions. Ultimately, Quine argues that there is no sensible way of explicating an analytic/synthetic distinction. For this reason he argues that we have no reason to believe that such a distinction exists in language. He acknowledges that people have an intuition of analyticity but he explains away this intuition through his notion of one-criterion concepts.
Overall then Chomsky and Quine are agreed on certain points. Neither thinker argues that any sense can be made of the notion of analyticity as a defence of a priori justification, or of our supposed grasp of necessary truths. Both thinkers, however, do manage to offer at least some explanation of why people have an intuition that a distinction between analytic and synthetic truths exists. Chomsky’s explanation of analyticity is sketchy and vague and has little evidential support. He claims that people are born with innate concepts which as a matter of fact yield analytic connections; however, he does not offer empirical evidence to support this claim. Furthermore, his claims about analyticity commit him to the view that analytic connections exist in different forms in different faculties. So his purported explanation of analyticity relies on prior commitments to faculty psychology, and while this commitment may turn out to be justified, it is very much an open question. More importantly, his views on the analyticity of mathematics yield a type of scepticism which cannot be taken seriously. Quine’s arguments against the notion of an analytic/synthetic distinction were not refuted by Chomsky. His explanation of why people have an intuition of analyticity relies on a conception of language acquisition which he has not shown to be correct. The question whether Quine’s conception of people’s intuitions of analyticity is correct or not is parasitic on whether his conception of language is correct or not. At this moment in time the evidence does not indicate whether Quine is correct on this point or not. What is clear is that Chomsky’s positive arguments for the existence of an analytic/synthetic distinction have little evidence in support of them.