Language: A Dialogue
Phil: Hi Lyn. I was hoping I would bump into you. I’ve heard that you have some quite bizarre ideas about language.
Lyn: Erm, that’s rich. Are you still worrying about how you know you’re not a brain in a vat?
P: Touché. Seriously, though, I’ve heard that you think that languages are mental states, properties of human brains - or ‘mind/brains’, as you say. Well, the thing is, Lyn, while we philosophers are notorious for our disagreements, if we agree about anything, it is that languages are not internal states.
L: Yes, it is indeed my view that ‘languages’ are states of the human mind/brain. I know that this is not the received view in philosophy, but it is none the worse for that. First off, instead of bothering about what the word ‘language’ means, let us consider some facts. The human brain is equipped with some resources - whatever they might be - which enable us to acquire any language - English, French, Latin, etc. This strikes me as obvious, just a fact. Well, if this is so, then the human brain has some property P that distinguishes us from chimpanzees, rabbits and rocks. P is innate, unlearnt - it comes with being human, just as flying comes with being a bird, or swimming comes with being a shark. Secondly,…
P: Hang on, hang on. I don’t know what this ‘P’ is supposed to be, still less why it has to be linguistic in some sense. P might just be ‘intelligence’ or ‘pattern recognition’ or something at least which doesn’t presuppose anything specifically linguistic.
L: You are perfectly right: P is just a stand-in for whatever it is about the human brain that distinguishes it from the brains of chimpanzees, rabbits, and everything else in the known universe. I didn’t assume that it was ‘linguistic’. Let us see if we can agree on something. Take a Japanese baby just born in Tokyo. Move on five years, what language will it be speaking?
P: Well, it depends, but I suppose you want me to say ‘Japanese’.
L: You’re right; we can’t simply assume that Japanese babies will end up speaking Japanese; it depends on what language their parents and peers speak. That’s my point! The typical Japanese child ends up speaking Japanese, but this particularity is not determined by the make-up of its brain. I mean, move the Japanese child to any other country and it will pick up the language of its new environment as easily as it would have picked up Japanese. A chimpanzee won’t, nor will a rock.
P: I see what you are getting at. A child can acquire any language of any environment with equal ease, so whatever the child has innately - your P - can’t correspond to any particular language; if it did, then acquiring one language might be harder than acquiring another, or even impossible, but that looks not to be the case. On the other hand, P just can’t be any old property, for only humans have it; it must be able to cover the apparently indefinite variation in the world’s languages, because any child can acquire any language. P is this capacity to target any language. Is this what you are saying?
L: Yes, exactly.
P: But now I’m confused. I thought that you want to argue that languages are internal mental states. But such a thesis doesn’t follow from the thesis that there is some P. P corresponds to no language; we just showed that. P is merely the capacity to acquire a language.
L: You’re jumping ahead. All we have established so far is that there is something about the human brain that enables the child to acquire any given language. This property we call P. Now comes a separate question: What is P? Is it something specifically linguistic, or is it some general capacity, which might be involved in doing mathematics or playing chess, or whatever? Well, my thought is that P is specifically linguistic without corresponding to any particular language; it expresses, we might say, the general or universal form of languages, what they all share in virtue of being human languages.
P: Okay, let me see if I’ve got the idea. To acquire a given language we depend on some innate capacity, but, whatever language we do acquire, the same capacity would have worked equally well for any other language. Now, you also want to say that this capacity is specific to language, it’s not some general purpose intelligence we employ outside of the realm of language. In a sense, then, this capacity is kind of like the essence of language - that which is invariant over all languages, that which underlies our capacity to acquire any given language.
L: Spot on. I call this fecund state of mind, Universal Grammar, or UG for short. You can think of it as the initial state of the language faculty, where later states of the faculty correspond to a speaker’s acquired language.
P: But I still don’t see what argument you have for this. I mean, why should there be any essence to language? Perhaps this P - the capacity to acquire any language - just does correspond to ‘intelligence’ or some sophisticated pattern recognition capacity. In other words, your UG is not a grammar at all; it’s just the general capacity which makes us humans smarter than everything else in the known universe. 
L: Well, there might not be UG in the sense I mean; I have no mathematical proof that there must be such a thing, but I do have a very good argument.
The child’s linguistic experience - the sentences she hears, etc. - is not rich enough for her to figure out how the language as a whole works. In other words, there’s just a real gulf between what the typical child experiences and what she ends up knowing: this complex thing we call her language. She requires additional information, over and above the mere capacity to organise and segment her experiences, in order to start creatively using language herself, rather than merely repeating what she has heard. Well, if the required information doesn’t come from outside, it must come from ‘inside’, namely, UG. UG expresses what the child brings to language as part of its biological make-up.
P: Erm, this is style of argument goes back to Plato’s dialogue Meno, where Socrates shows that Meno’s slave boy understands Pythagoras’ theorem, even though no-one has taught him any maths. Plato thought that this meant that the slave boy’s soul brought the mathematical knowledge with it from whence it came - the realm of the Forms. You’ve just replaced Forms and souls for biology. More generally, you are siding with the Rationalists - Descartes and Leibniz - against the empiricists, like Locke and Hume. You think that we have innate linguistic ideas, concepts which don’t derive from our senses. We are not ‘blank slates’.
L: That’s right, I’m a Rationalist in modern dress. It’s worth pointing out, though, that no-one has ever really believed that the mind was entirely blank at birth. I mean, it just makes no sense. If it were blank, how could it be that every human ends up speaking a language but our pets don’t? Everyone thinks that something is innate - unlearnt - I differ in thinking that something specifically linguistic is innate.
P: Okay, I take your point: if my mind was entirely blank at birth, then I suppose, initially, there wouldn’t be any difference between me and my dog, but we end up differing greatly. Your point is: We only end up so different because we were very different to begin with. I can see this. But I still don’t see why what I start with, as it were, has to be linguistic in some sense. Why can’t I just be a lot smarter than the dog?
L: As I’ve said, because the experiences you have just aren’t rich enough for you to arrive at, say, English on the basis of just being really smart, whatever that might mean.
P: Surely the child is inundated with language. Anyhow, this is an empirical issue for psychology. What basis do you have for assuming that the child lacks this or that information about language from what it hears around it?
L: You misunderstand my point. I haven’t made any assumptions about the linguistic data the child is likely to meet. You are quite right. I really don’t know how rich the data are. But my argument is that no matter how rich they are, they aren’t going to be rich enough. In linguistics, we have as much data as we could hope for, but we still can’t figure out the true generalisations for English. Linguists are very bright, highly educated people, working together with a rich tradition of research behind them. Even so, they can’t figure our how English, or any other language, works. We’re trying, for sure, but without clear success yet. So, give the child - minus UG - as much data as you want. No, give the child the acquired wisdom of 2000 years of thought about language, and it still won’t be able to figure out what language it is supposed to be speaking. There is a paradox here. The linguist relies on data - lots and lots of data, as much data as she needs - but she can’t figure out how English works. On the other hand, the child picks things up by about the age of five, making remarkably few mistakes along the way, and independent of other variables, such as intelligence, social background, disability, etc. This is miraculous, yes? Well, no. It’s only miraculous if we make the assumption that the child picks the language up from the environment without something like UG. The child’s task is easy precisely because it doesn’t rely on data; most of what it needs to know is already built into its brain; the child just waits for it to be triggered. The environment plays a part here, but only to select options already available. It doesn’t instruct the process. This is widely picturesque, but it’s as if the child already knows English, French, etc., and after hearing certain kinds of very simple patterns that uniquely cluster with French, the child’s brain selects French, rather than Swahili, say, which exhibits different kinds of simple patterns. So, the linguist’s task is really hard precisely because she must rely on data, whereas, in effect, what she is trying to understand is what the child has innately without relying on data, namely UG.
P: I’m not sure I understand what you’re saying. The idea that we have every possible language in our heads at birth is just silly.
L: I don’t know, perhaps we do: these matters are empirical. Anyhow, we don’t have to take the picture seriously. A common idea is that a speaker has a menu of fixed options and some broad constraints about what can go with what. It’s as if there’s a restaurant in your head, and you can have the veal or the trout, but if you have the veal, you can’t have the salad, you must have the soup. This gives you two meals. Add to the menu, and you have a whole lot of other possible meals. Each language, or future state of the language faculty, corresponds to some selection of options, a meal, as it were. Experience selects the options - orders the meal -, but it doesn’t create the options, nor dictates what options can go together. The chef is in your head; experience is the customer. So, French need not be sitting there in the child’s brain twiddling its thumbs waiting to be called; the language faculty doesn’t realise all of its possible states simultaneously, just as not every meal is waiting in the kitchen of a restaurant - we’re not talking about McDonalds here.
P: Okay, I think I see what you mean, but this idea is very strange. You will admit that it is highly counter-intuitive.
L: Most science and philosophy is counter-intuitive. Perhaps the problem for intuition in the present case is that it is an example of a general phenomenon: we know an awful lot we don’t know we know! Most of our knowledge about language is implicit, unconscious. What we have found is that there is a whole range of significant generalisations about the structure and meaning of English sentences - let’s just stick with English, but the point applies generally - that we clearly understand and follow, but which we don’t know consciously. Further, these generalisations can’t be stated in terms of the visible properties of sentences, as it were. These are properties of order and sound. In short, we know things that we couldn’t have acquired just from the data. Linguists have only just noticed many of them. In some sense, they come from UG.
P: Could you give me some examples?
L: Sure. Everywhere you look, you find the same thing. Consider verbs to do with liquid, like ‘spill’, ‘squirt’, ‘pour’, etc. Just to take ‘spill’, although the example generalises, we can say both (2) and (3):
(2) Bill spilt water on the floor
(3) Water spilt on the floor.
(2) implies (3). We can all recognise this. But now think about verbs to do with substances of a certain viscosity, such as ‘smear’, ‘daub’, ‘plaster’, etc. Again, just to pick one, we find a different pattern from that above:
(4) Bill smeared paint on the wall.
(5) Paint smeared on the wall.
It is not so much that (4) doesn’t imply (5); rather, (5) is just ill-formed; it is not an English sentence. Well, why can’t paint smear on the wall in the way that water can spill on the floor? It’s no good thinking about the difference between paint and water. I mean, no chemist is going to do research on this question. That’s just daft. More to the point, no-one teaches this to children acquiring English. Indeed, how could it be taught to children when adults don’t even know it explicitly? Nevertheless, it’s part of being a competent English speaker that one finds no problem with (2)-(4) but finds (5) distinctly odd. Well, no-one ever utters (5); it’s just not part of English.
P: Erm, I had never noticed the difference. But I don’t see the big mystery. Let me think… the difference is just that verbs like ‘spill’, ‘squirt’, etc. don’t require someone to do the squirting, spilling, etc., whereas verbs like ‘smear’, ‘daub’, etc. require someone to do the smearing, daubing, etc. That seems simple enough. We work out the difference via knowing what the verbs mean.
L: Remember, we are thinking about the child acquiring the language. It is perfectly correct to say that ‘spill’, in some sense, doesn’t imply an agent who does the spilling, whereas ‘smear’ does imply an agent who does the smearing. But where does this difference come from? It seems to me that this difference is the very same difference I initially pointed out - you haven’t explained anything. Besides, how does the child learn what the words mean? Do you imagine that we sit our children down and tell them that ‘spill’ doesn’t involve a notion of agency whereas ‘smear’ does? This obviously doesn’t happen. Nor is it remotely plausible to think that the child can pick the difference up from its observation of liquids. I’ve never seen water spill itself on the floor - someone does the spilling -, just as I’ve never seen butter smear itself on my toast in the morning.
P: Okay, okay. I’ll have to think about this a lot more. I’m sure you have lots of other examples…
L: Consider double-object constructions…
P: Hold on! What I was going to say is that I still don’t see why you think that languages are internal mental states. It seems to me that all this talk about English presupposes a shared public language. Have I missed something?
L: Well, English is a ‘convenient fiction’.
L: Look, you’ve granted me UG; or at least you have granted me that there is some property of the human brain that distinguishes us from chimpanzees and the rest of the universe as regards language. Well, is this property just some general capacity or is it specific to language? It seems to be quite specific. We are not taught the subtle difference between ‘spill’ and ‘smear’, but we all acquire the words fine. This knowledge must come from UG, in some sense, for it doesn’t come from outside. But UG is just a property of the brain. In other words, the states of our language faculties are just variations on the menu provided by UG. If the brain develops in this way, you get ‘English’; if in that way, ‘French’; if in some other way, ‘Latin’. I use the ‘scare quotes’ because ‘English’, ‘French’, etc. are just picking out a set of language faculties that are sufficiently similar to support various generalisations (e.g., English sentences require subjects, Italian ones don’t), although each will be quite distinct in all sorts of ways. For instance, we tend to speak of ‘Chinese’, but there is much more variation between Chinese dialects then there is between French, Italian, Spanish and Romanian, say. Languages are a bit like races: Europeans, Africans, etc. Races aren’t real as far as biology is concerned; that is, no interesting generalisation in biology concerns Africans or Europeans. Biology is concerned with interbreeding populations which cross-classify what we think of as races.
P: So, you’re saying that really we each have our peculiar way of speaking and understanding, our individual dialects, if you will, which can be more or less similar within the bounds set by UG. Talk of English is a crude classification, much like our talk of Africans.
L: Yes, that’s right. There’s a joke in linguistics that a language is something with an army and a navy. For all kinds of reasons to do with politics, culture, history, whatever, we are interested in racial differences, much to the detriment of human development, I might add. Just so, for similar reasons we are interested in classifying people as English speakers or French speakers, but this is interest-relative, as they say: there is no thing - English - which all and only those we want to call English speakers know.
P: Oh, look. Just because the notion of English is a bit vague around the edges, it doesn’t mean that it doesn’t exist. It’s a social object, like our constitution, or the welfare state. Our dialects are not internal states!
L: You misunderstand me. I’m not saying there is no such thing as English. I’m not even sure what that means. All I’m saying is that English is not a notion which enters into serious investigation of language, apart from as a useful term to talk about a particular population of speakers: people of the British mainland (and scattered former dominions), or something like that. The linguist is interested in the mental states of the speakers, not some object - English - which they all relate to in some mysterious way. We have granted, I hope, that the speaker must have some very complicated mental structures to be able to speak and understand in the normal way. The character and development of these structures is what interests me. Anything else you want to say about language will presuppose such structures. Of course, dialects are not internal states, but they depend on specific internal states, states which distinguish us from one another, but still mark us out as language users in distinction to chimpanzees or rocks. What’s the problem?
P: Well, look, we have a shared store of thoughts. If I manage to communicate with you, then we have both grasped the same thought; I don’t have to infer what you mean - I can just understand your words, because they express the same thought which I would express were I to utter those words. Any worthwhile conception of language must make sense of its prime function - communication. Your conception makes communication a mystery.
L: This talk of grasping thoughts is just a metaphor, which, to be frank, I don’t understand. Besides, I really don’t think that language is for communication. I can communicate with language, but birds and bees can also communicate without language. If you want to say that language just is communication, then there is nothing much else to say. The concept of ‘communication’ is a bit like ‘swimming’ in this regard. Lots of things swim: humans, sharks, otters, hippos,… they propel themselves through water. There’s really nothing much else to say about swimming in general; although a particular account of how a shark swims will be of interest to a marine biologist, it won’t tell you anything in particular about how humans swim.
P: I’m sorry, you’re being a bit oblique.
L: Look, classifying creatures as swimmers is just not interesting, it’s merely descriptive. Just so for classifying creatures as communicators. There are a multitude of ways to swim and communicate which our familiar notions just don’t discriminate. So, sure, language allows for communication, but so does having legs, or a big tail, or scent pouches, or a changeable skin pigment. None of this has much to do with human language in particular.
P: Okay, communication might not be an interesting notion in itself, but your position appears to make it impossible.
L: Let’s back-track. We both agree that the child, and so the adult, requires quite sophisticated mental structures to use language, including, presumably, communication. Y’know, our heads aren’t just full of baked beans; there is a lot of dedicated, very complex machinery in there. No-one tells us what ‘spill’ means, but it turns out to be quite complex; that comes from inside. Okay, well, the question is: Do we need more than these mental structures to understand communication? As far as I can see, you simply assume that we do, but I just don’t see it. Imagine Smith and Jones having a conversation. On my view, this is possible because of the way things are with Smith and Jones, period. Neither has to grasp any independent thing. If their internal states are sufficiently alike (e.g., ‘spill’ doesn’t involve agency, but ‘smear’ does), then they will respond to each other in a coherent way. This is not an all or nothing thing. Often, understanding just breaks down - a conversation is a precarious affair. On your view, on the other hand, to make sense of Smith and Jones we require not only a description of the respective mental states, but also a specification of some third thing: a shared language which expresses these strange things we are supposed to grasp. The burden is on you to say why.
P: Erm, it’s getting late and I have to write a paper about brains in vats. I’ll have to get back to you on this. It is all quite intriguing, but it’s, er,… how should I put it?… unattractive. A bit like someone not so long ago, you seem to be saying, ‘There’s no such thing as society. There’s just you and your brain’.
L: Politics has really nothing to do with it, at least not explicitly: no political position - either reactionary or progressive - is deducible from what I’ve argued. That said, political and moral views do presuppose, albeit often inchoately, some conception of human nature, much as, say, any theory of education presupposes some conception of the child’s natural capacities and motivations - we are not dealing with chimpanzees or rocks. Well, my view of our nature is that we are not blank, malleable beings to be shaped and molded according to whatever structures of education and socialisation are current. We each bring with us an individual creativity with language, on a biological theme, from which we spin our very distinct perspectives on the world and each other. If anything follows from this, it is that we should treat each other with respect and dignity, and seek to have such attitudes reflected in social organisation, for we are not mere products of society. Otherwise put, a rich conception of human nature, in contrast to the poor one of the standard empiricist position, places the burden of justification on those who would seek to wield power over us - socially, educationally, politically, personally, whatever - in the name of some higher goal or authority, such as the state, the nation, the party, the race... Mind, we shouldn’t need philosophy or science to tell us this.
P: Well, thanks for the chat, Lyn - every time I spill my drink, I’ll think of you.
 Lyn expresses the views of Noam Chomsky. Chomsky’s work on language is voluminous and sometimes highly technical, but always clear. A good starting point is Rules and Representations (New York: Columbia University Press, 1980) or the first two chapters of Knowledge of Language: Its Nature, Origin, and Use (Westport: Praeger, 1986). He is often at his best in interview; see Language and Politics, ed., Carlos Otero (London: Black Rose Books, 1988). For a general introduction to Chomsky’s work, see Neil Smith, Chomsky: Ideas and Ideals (Cambridge: Cambridge University Press, 1999). For a wonderfully accessible introduction to the current ‘Chomskyan’ framework in linguistics, see David Adger, Core Syntax: A Minimalist Approach (Oxford: Oxford University Press, 2003).
 Phil’s complaint here has been often expressed; for the classic example, see Hilary Putnam, ‘The ‘Innateness Hypothesis’ and Explanatory Models in Linguistics’, in his Mind, Language and Reality (Cambridge: Cambridge University Press, 1975), 107-116. For a recent example, see Fiona Cowie, What’s Within: Nativism Reconsidered (Oxford: Oxford University Press, 1999).
 For an overview of the empirical research in this area, see Steven Pinker, The Language Instinct (London: Penguin Books, 1997).
 For a model on which, in a certain sense, all possible language are ‘in our heads’, see Charles Yang, Knowledge and Learning in Natural Language (Oxford: Oxford University Press, 2002). The book is not for beginners, although the opening chapters are accessible.
 (5) can be part of a sentence, as in ‘Bill saw paint smeared on the wall’, but ‘paint smeared on the wall’ is here a noun phrase, with ‘smeared’ as participial, it is not a sentence in its own right.
 Lyn was going to point out that whereas ‘Bill told Mary the message’ is well formed, ‘Bill reported Mary the message’ is not. This is so, even though ‘tell’ and ‘report’ are near synonyms, and both admit the non-double-object construction: ‘Bill told/reported the message to Mary’. Again, no-one tells us this.
 For a survey of systems of animal communication, without serious mention of human language, see Marc Hauser, The Evolution of Communication (Cambridge, MA: MIT Press, 1996). For a speculative proposal on the phylogenic relation between language and animal systems of communication, see Marc Huaser, Noam Chomsky and W. Tecumseh Fitch, ‘The Faculty of Language: What is it, Who has it, and How did it Evolve?’ Science, 298, November 2002, 1569-1579.
 Chomsky is perhaps better known as a political/human rights campaigner than as a linguist/philosopher. He is, however, quite reticent to speak about the connections between these two strands of his thought. For the connections that do exist, see Noam Chomsky, For Reasons of State (London: Fontana, 1973) and Chomsky on Education and Democracy, ed. Carlos Otero (London: RoutledgeFalmer, 2003).
 My thanks go to Noam Chomsky for inspiration and helpful comments.