by Noam Chomsky
As for the creative aspect of language use observations, they are that human language use appears to have no causal antecedents (one can think or say anything, without regard to circumstances outside the body or in the head), that the use of language yields an indefinitely large number of structured complexes of concepts (those expressed by sentences) with regard to any context, and yet that the sentences produced on an occasion are almost always appropriate to the circumstances (discourse or other) to which they ‘speak.’ Traditional rationalists generally assumed that these observations indicated that humans must be seen as free in their thought, deliberations, and actions. Current rationalists agree, but because they believe that languages are best understood as computational systems embodied as biologically based organs in the mind/brain, they must also try to make sense of how a deterministic system in the head can play a role in yielding such obviously creative, novel, and coherent use by people. They do so by assuming that the core computational language system is “modular” and operates more or less autonomously and yet can “generate” an indefinitely large number of structured ways of speaking, thinking, and understanding. Plausibly, this is the root of the flexibility of human cognitive powers: human minds can put together in structured forms any number of structured conceptual materials, each of which is discernibly distinct from all others. But along with these assumptions about the capacity of the computational system and what it affords its users, contemporary rationalists believe that the ways that these uniquely human resources are used are not determined by the computational system. The consequences of these assumptions are important. Because of them, it is reasonable to hold that humans really are free in the ways in which they use language, and also to hold that if one wants a natural science of language, the only way to get it is to focus entirely on the nature, development, and operations of a person's “language organ,” not on the uses to which its resources are put as they appear in a person's linguistic actions and behaviors. Because of this, the science of language is the science of an internal system: rationalists are internalists. In sum, then, rationalists now like those of the past are – because they take seriously the poverty and creativity observations – scientists of mind and language who are both nativist and internalist in their assumptions about how to proceed.
Rationalism contrasts with empiricism, which commits itself to minimizing commitments to innateness – or at least, language-specific innateness – in the study of language and related “higher cognitive capacities,” holding instead that much of the cognitive structure and “content” of the human mind results primarily from ‘experience’ and some kind of generalized “learning” mechanism. Empiricists are anti-nativist and committed to including the world and relations to it in the study of mind. They, unlike rationalists, believe that a science of language should in some measure be a science of linguistic behavior and how the mind relates to the world outside the head. Their assumptions and the methodology for the study of mind that they adopt dominate current research in psychology, philosophy, and related ‘cognitive sciences.’ For them, language tends to be seen as a human invention, an institution to which the young are inducted by subjecting them to training procedures so that they adopt the ‘rules of use’ practiced by a community of speakers. They assume that the study of language must proceed on anti-nativist and externalist assumptions.
Obviously, the different assumptions of rationalists and empiricists lead them to devote research to very different subject matters, and they adopt very different views about the prospects of treating language as a “natural object,” to use Chomsky's terminology. I discuss rationalism and empiricism in a bit of detail in Appendix III. For more detail, see Chomsky's Cartesian Linguistics (particularly the 2009 Cambridge third edition, which has an introduction on the topic written by me) and Norbert Hornstein's chapter in The Cambridge Companion to Chomsky (2005). Readers are warned that “empiricism” by no means entails “empirical,” any more than “rationalist” entails “non-empirical.” Rationalists aim toward being no less empirical scientists than do chemists or biologists. Indeed, Chomsky has long insisted that the only methodology appropriate for developing a theory of language is that which is also employed in chemistry and physics. The study of language differs only in subject matter and experimental techniques from other naturalistic scientific endeavors. That this is the appropriate approach is demonstrated by its success – by progress in explanatory and descriptive adequacy, formal and explicit statement, simplicity, objectivity, and accommodation to other sciences (here, primarily biology). For some discussion, see McGilvray (forthcoming).
Part I The science of language and mind
1 Language, function, communication: language and the use of language
JM: I'll begin with a question that concerns the nature of language and the functions that language serves.
It is clear that language is central to human nature: it is likely to be the one thing that makes us distinctive. You think of language as biologically based, and so innate – built into our genomes in such a way that it appears automatically during a normal child's growth. And you acknowledge that language is a very useful cognitive tool that can serve many roles and that has given humans extraordinary cognitive advantages, compared to other creatures. But you resist the idea that language evolved because it improved the human capacity to communicate. Further, you're against the idea that language is some kind of social invention, an institution that was put together by us to help us serve our needs, and transmitted to the young by some kind of training or social inculcation. Could you explain why you hold these views?
NC: First of all, let's begin with the notion of function. That's not a clear biological notion or a psychological notion. So, for example, if I ask you what the function of the skeleton is, and you say: “the skeleton is to keep you straight and keep you from falling on the ground,” that is not false. But that also applies to its function to store calcium or to produce blood cells, or to do any of the other things it does. In fact, why the skeleton? Why do you even pick out the skeleton? We try to look at the organism from a certain point of view in order to build up a full understanding of it from the understanding of its components. But those components do all kinds of things; and what their function is depends on what you happen to be interested in. The usual sort of offhand way in which people identify the particular function of some system is the way in which it is ordinarily used, or its ‘most fundamental’ use, so that for the skeleton example, somehow, something else could store the calcium, and the skeleton would still be needed to keep the body together; so that's its function.[C]
Now let's take language. What is its characteristic use? Well, probably 99.9 percent of its use is internal to the mind. You can't go a minute without talking to yourself. It takes an incredible act of will not to talk to yourself. We don't often talk to ourselves in sentences. There's obviously language going on in our heads, but in patches, in parallel, in fragmentary pieces, and so on. So if you look at language in the way biologists look at other organs of the body and their subsystems – so you take into account all its functions in talking to yourself – what do you get? What are you doing when you talk to yourself? Most of the time you're torturing yourself [laughter]. So you might think you're being conned, or asking why does this person treat me that way? Or whatever. So you could say that the function of language is to torture yourself. Now, obviously, that's not serious.
It's perfectly true that language is used for communication. But everything you do is used for communication – your hairstyle, your mannerisms, your walk, and so on and so forth. So sure, language is also used for communication.
In fact, a very tiny part of language is externalized – what comes out of your mouth, or from your hands if you're using sign. But even that part is often not used for communication in any independently meaningful sense of the term “communication.” If by “communication” you mean any form of interaction, ok, it's used for communication. However, if you w
ant the notion of communication to mean something, let's say conveying information or something like that, a very small part of the externalized aspects of language are for communication. So if you're at a party there's a lot of talk going on. But the amount of communication that's going on is minuscule; people are just having fun, or talking to their friends, or whatever. So the overwhelming mass of language is internal; what's external is a tiny fraction of that [and what's used in communication in some serious sense is a smaller fraction still]. As functions are usually informally defined, then, it doesn't make much sense to say that the function of language is communication.
An interesting topic that should be addressed some day is that our internal speech is very likely fragments of re-internalized external speech, and the real “inner speech” is very likely inaccessible to introspection. But these are questions that open many doors, barely ajar.
Well, let's take a look at language from an evolutionary point of view. There are animal communication systems. Every animal down to ants has a communication system, and there are interesting comparative studies of them. Take, for example, Marc Hauser's book on the evolution of communication. It doesn't really have much to do with evolution; it's a comparative study of different kinds of systems of interaction among animals. And they really do appear to be communication systems. Every animal has a small number of modes of indicating something to others. Some of them we interpret as meaning, “Eagles coming; so run away!” If you look into it, it's just: those leaves are flickering, so some noise comes out of the creature's mouth. Some of them are self-identification: “Here I am”; and some of them are mating calls. But there are not very many.
There's a kind of taxonomy of animal cries, and human language doesn't even fit into the taxonomy, I think, in any of the senses. Whatever those things are, they're apparently a code; there's no relationship to human language. That's not so surprising: apparently, our nearest surviving relatives are about 10 million years apart in evolutionary time; so you wouldn't expect to find anything like human language. So animals do have communication systems, but they don't seem to have anything like a language. Take human language. Where does it come from? Well, so far as we can tell from the fossil record, hominids with higher physiological apparatus were around in a small part of Africa for hundreds of thousands of years. We know by now that human language does not postdate about sixty thousand years ago. And the way you know that is that's when the trek from Africa started. By now you can trace it very closely by genetic marking, and so on and so forth; there's pretty good consensus on it. The trek from Africa starts roughly then and went very quickly in evolutionary time. One of the first places they went to is the Pacific – the southern part of Eurasia. They end up in New Guinea, Australia, and so on, where there are [now] what we call “primitive people” who to all intents and purposes are identical to us. There's no cognitively significant genetic difference anyone can tell. If they happened to be here, they would become one of us, and they would speak English; if we were there, we would speak their languages. So far as anyone knows, there is virtually no detectable genetic difference across the species that is language-related – and in fact, in most other properties. Genetic differences within humans are extremely small, as compared with other species. We pay a lot of attention to them; that's no surprise. So, some time, maybe sixty thousand years ago, language was there, in its modern form, without further changes. Well, how long before that? From here, we can look at the fossil record, and there's not really an indication that it was there. In fact, the effects of having a complex symbolic system are barely there before 60,000–100,000 years ago. Nothing much seems to have changed for hundreds of thousands of years, and then, all of a sudden, there was a huge explosion. Around seventy, sixty thousand years ago, maybe as early as a hundred thousand, you start getting symbolic art, notations reflecting astronomical and meteorological events, complex social structures . . . , just an outburst of creative energy that somehow takes place in an instant of evolutionary time – maybe ten thousand years or so, which is nothing. So there doesn't seem to be any indication that it was there before, and it all seems to be the same after. So it looks as if – given the time involved – there was a sudden “great leap forward.” Some small genetic modification somehow that rewired the brain slightly. We know so little about neurology; but I can't imagine how else it could be. So some small genetic change led to the rewiring of the brain that made this human capacity available. And with it came the entire range of creative options [C] that are available to humans within a theory of mind – a second-order theory of mind, so you know that somebody is trying to make you think what somebody else wants you to think. It's very hard to imagine how any of this could go on without language; at least, we can't think of any way of doing it without a language. And most of it is thinking and planning and interpreting, and so on; it's internal.
Well, mutations take place in a person, not in a group. We know, incidentally, that this was a very small breeding group – some little group of hominids in some corner of Africa, apparently. Somewhere in that group, some small mutation took place, leading to the great leap forward. It had to have happened in a single person. Something happened in a person that that person transmitted to its offspring. And apparently in a very short time, it [that modification] dominated the group; so it must have had some selectional advantage. But it could have been a very short time in a small [breeding] group. Well, what was it? The simplest assumption – we have no reason to doubt it – is that what happened is that we got Merge. You got an operation that enables you to take mental objects [or concepts of some sort], already constructed, and make bigger mental objects out of them. That's Merge. As soon as you have that, you have an infinite variety of hierarchically structured expressions [and thoughts] available to you.
We already had sensory-motor systems [when Merge was introduced], which were probably marginally employed. In fact, the idea of externalizing them might very well have come along later. And we had thought systems of some kind. However, they were rudimentary – maybe we pictured things in a certain way, or whatever. Whatever they are, they don't seem to be like animal systems, for reasons that were well discussed in the seventeenth and eighteenth centuries. But they were apparently there. Once you had this technique of construction and an infinite variety of hierarchically structured expressions to make use of these things (these thought systems or [what Chomsky calls] “conceptual-intentional systems”), then you could suddenly think, plan, interpret, in a manner that no one else could. And if your offspring had that capacity too, they would have a selectional advantage. And if, somewhere along the line, the idea came of trying to externalize it [thought] somehow, it would give even further advantages. So it's conceivable that that's it so far as the evolution of language is concerned. And the reason we continue to primarily use language to think [within] ourselves is that that's the way it got started. And, after all, sixty or seventy thousand years [and maybe up to a hundred thousand] isn't a lot of time from an evolutionary point of view; it's [virtually] an instant. So we're still pretty much the same as we were back in Africa whenever this sudden change took place. [C] That's about it, so far as we know.
Now, there are a lot of more complicated theories, but there's no justification for any of them. So, for example, a common theory is that somehow, some mutation made it possible to construct two-word sentences; and that gave a memory advantage because then you could eliminate this big number of lexical items from memory. So that had selectional advantages. And then something came along and we had three-word sentences and then a series of mutations led to five . . .; finally, you get Merge, because it goes to infinity [and this gives our minds a way to put together a limited number of lexical items in an infinite array of combinations]. Well, Merge could very well have been the first step, and it might have had nothing to do with externalization; in fact, it's hard to imagine how it could have, since it would have had to have happened in a person, not in a group or tribe. So there
has to be something that gives that person an advantage that leads to advantages for his or her offspring.
JM: They get thought. In that vein, you've sometimes speculated, suggested – I'm not sure what the right word is – that along with Merge come the natural numbers, comes some successor function. Merge in the limiting case where you simply join one element to itself might effectively lead to the successor function.
NC: This is an old problem. Alfred Russell Wallace was worried about it. He recognized that mathematical capacities could not have developed by natural selection; it's impossible, because everybody's got them, and nobody's ever used them, except for some very tiny fringe of people in very recent times. Plainly, they developed some other way. Well, a natural expectation is that they're an offshoot of something. They're an offshoot of – probably like most of the rest of what's called “the human intellectual capacity” [or reason] – something like language.
