the language of thought
What can’t be said, can’t be said, and it can’t be whistled either.
Frank Ramsey
If it is true that words have meanings, why don’t we throw away words and keep just the meanings?
Ludwig Wittgenstein
If thoughts depended on words, how could a new word ever be coined? How could a child learn a new word to begin with? How could translation to a new language ever be possible?
Steven Pinker
And the whole Earth was of one language, and of one speech. And it came to pass, as they journeyed from the east, that they found a plain in the land of Shinar; and they dwelt there ... And they said, Go to, let us build a city and a tower, whose top may reach unto heaven; and let us make a name, lest we be scattered abroad upon the face of the whole earth. . . And the Lord said, Behold, the people is one, and they have all one language; and this they begin to do: and now nothing will be restrained from them, which they have imagined to do. Go to, let us go down, and there confound their language, that they may not understand one another’s speech. So the Lord scattered them abroad from thence upon the face of all the earth: and they left off to build the city. Therefore is the name of it called Babel. (Genesis 11: 1–9)
Philosophers have lately asked whether the mythical common language spoken before Babel in fact lives on unbeknownst to us in the mind of every man and woman – a primordial code, or ‘mentalese’, into which all natural language is translated for processing by the brain. We may think in English, but there is reason to suspect that English is not the first language of English people. We know that information-processing goes on in the brains of babies before they have learned to talk, and that it occurs when people are actively thinking of different things and also when they are asleep. The faithful at prayer often have the experience of their call being completed before they have articulated its words. With a shrug and a nod, an intention is communicated. Or one can read a sentence not only without moving one’s lips, but also without speaking internally in thought. We all have the experience of words being on the tip of our tongues, or occasions when we know what we mean but can’t remember the right word.
We need to ask what this pre-verbal information-processing consists in. Sometimes we might think in pictures rather than words, but this would not account for all the instances mentioned. A picture, by itself, can be ambiguous, whereas our thoughts, it seems, cannot. The famous optical illusion of the duck-rabbit opposite can be seen as either of the two animals, but we do not make this mistake when thinking of rabbits and ducks.
Words in a natural language can also be ambiguous. The headlines ‘Four Held Over Bomb Are Freed’ (The Times, London, 7 June 1980) and ‘Judge Deals Blow to Bryant Defense’ (<www.CNN.com> Thursday, 22 April 2004) could be understood in either of two ways. But the journalists who wrote those words presumably knew very well what they meant to say and could have understood their own thoughts in only one way. As the American cognitive psychologist Steven Pinker writes, ‘If there can be two thoughts corresponding to one word, thoughts can’t be words.’1
Pinker talks of the factor that resolves such ambiguity as an extra element in the head. Philosophers, on the other hand, spent the better part of the twentieth century trying to show that it is context that makes the difference. The meaning of a phrase was to be determined by its setting, which might encompass both other words and the manner in which it is said. The Austrian philosopher Ludwig Wittgenstein argued that the limits of one’s language denoted the limits of one’s thoughts. He believed that all terms must have publicly available criteria for their use, or else there could be no discernible difference between employing them correctly or incorrectly. By contrast, expressions in a ‘private’ language – used only in an individual’s internal dialogue – would be correct if they ‘felt’ appropriate, which would undermine the necessity that there be a difference between getting things right and getting them wrong. Without criteria enforced by a wider linguistic community, no term could ever retain a fixed meaning. However, we might develop the same criteria for meaning via a different route if certain features of one’s language are common to all individuals as human beings. As Noam Chomsky has established, the basis of language is biological and laid down in the history of our species. If certain meanings were synchronized in the community of the dead, in the form of our genetic make-up, there would be no need for public engagement to underwrite all the grammar of the living, though it may be necessary for some features of language such as vocabulary. Shared human characteristics were essential in Wittgenstein’s thought also, but it was our ways of experiencing the world that he had in mind – such as the fact that we share the same kind of perceptual machinery. If the model of the brain as an information-processor can show that we share rather more than that, then the so-called ‘Linguistic Turn’ effected in Western philosophy at the beginning of the twentieth century, largely at Wittgenstein’s behest, was overrated.
The notion of a mentalese, or ‘The Language of Thought Hypothesis’ as it is known, was introduced by the American philosopher Jerry Fodor in 1975.2 Fodor was a long-time colleague of Chomsky’s at MIT before he moved to Manhattan in 1986 to teach at Rutgers University. Fodor and his wife now live in a corner apartment overlooking the Hudson just a short walk from the Lincoln Center, where he can indulge his passion for opera. I was astonished at his intellect and speed of thought, although the spell was almost broken at the end of our meeting when his wife ticked him off for forgetting to book their concert tickets again. His appointment to Rutgers began a gathering of top minds that has made New York City the hub of the philosophical world today. In part this is a testament to Fodor’s status as one of philosophy’s few living guru figures. An avuncular, full-cheeked Ernest Borgnine lookalike, Fodor seems too jovial to be a guru, but he has been described as ‘the sun around which all the planets gravitated’.3 These planets now include such philosophers as Colin McGinn, Alvin Goldman and Thomas Nagel.
Fodor likes to use the analogy of computer languages to explain his position. The codes that programmers use look fiendishly complex to the layman, but they are not the ‘real’ language of the computer. Programming languages such as C++ and Java are relatively userfriendly simplifications of what is required to get desktop machines to do what we ask of them. The real language ‘understood’ by our PCs is machine code – a series of ones and zeros into which all programming languages have to be translated for software to work. In the same way, it is highly unlikely that the languages in which human beings converse, such as English and French, speak directly to the brain at the level upon which thinking takes place. We certainly seem to think in English, but then computers seem to work in C++ or Java if one does not know better. Fodor also uses the example of animals, many of which clearly employ thought in planning a hunt and solving other problems, as well as in learning, although they do not possess language. Crows, for example, have long been seen dropping clams from a great height on to beaches to crack them open. Eagles in Greece employ the same process for breaking open tortoise shells, as proved fatal for the ancient playwright Aeschylus, who died after an eagle mistook his shiny, bald pate for a rock. Birds did not solve the problem of how to break into walnuts until Japanese crows made the ‘discovery’ around 1990. A BBC television crew captured several crows waiting patiently at a pedestrian crossing with walnuts in their beaks. When the lights turned to red, they would leave the nuts in the middle of the road and hop back on to the pavement. After the shells had been crushed under the wheels of cars, the birds crossed over with the other pedestrians to collect their meal.4 Obviously, animals do not possess human language, but neither do most species seem to possess any sophisticated system of communication. Yet their behaviour shows them to be capable of complex thought and planning. Presumably, our own ancestors were able to effect similar intelligent behaviour before we possessed sophisticated natural languages.
It might be thought that a richly developed natural language would enable a species to achieve more complex planning than it could using mentalese alone – otherwise we could ask what the point of English and French would be. But Fodor’s contention is that such languages were a technological advance in expression and communication rather than in thought. What can be achieved in a piece of PC software written in C++ or Java is dependent upon the limits of machine code rather than the other way round. It would make a difference if natural language learning could change the coding of mentalese. Mentalese would then become more than intertranslatable – it would form partially merged languages. Natural language in that case might as well be the medium of thought. If we agree with Fodor that it is not, then there are ramifications for the popular view that it has the power to shape our outlook. According to the thesis of ‘linguistic determinism’ developed by the American researchers Edward Sapir and Benjamin Lee Whorf in the 1920s and 1930s, what we take to be the real world is in fact a construction foisted on the individual by the dominant linguistic conventions of society. Most philosophers have never taken very seriously the threat of linguistic determinism to shape sense perceptions. For instance, in the Navajo language there is only one word for both green and blue, but, as Pinker writes, ‘No matter how influential language might be, it would seem preposterous to a physiologist that it could reach down into the retina and rewire the ganglion cells.’5 However, other varieties of mental content might not be so immune.
Modern political correctness is a direct descendant of the Sapir–Whorf thesis. It maintains, for example, that sexist language leads to sexist thoughts, and that the latter can be eliminated by abolishing the former. Big Brother had the same idea in George Orwell’s novel, 1984:
The purpose of Newspeak was not only to provide a medium of expression for the world-view and mental habits proper to the devotees of Ingsoc [English Socialism], but to make all other modes of thought impossible. It was intended that when Newspeak had been adopted once and for all and Oldspeak forgotten, a heretical thought – that is, a thought diverging from the principles of Ingsoc – should be literally unthinkable, at least as far as thought is dependent on words. Its vocabulary was so constructed as to give exact and often very subtle expression to every meaning that a Party member could properly wish to express, while excluding all other meanings and also the possibility of arriving at them by indirect method. This was done partly by the invention of new words and by stripping such words as remained of unorthodox meanings, and so far as possible of all secondary meanings whatever…A person growing up with Newspeak as his sole language would no more know that ‘equal’ had once had the secondary meaning of ‘politically equal’, or that ‘free’ had once meant ‘intellectually free’, than, for instance, a person who had never heard of chess would be aware of the secondary meanings attaching to ‘queen’ or ‘rook’. There would be many crimes and errors which it would be beyond his power to commit, simply because they were nameless and therefore unimaginable.
If Fodor is correct, then such a policy – along with that of politically correct ideologues – would be doomed to failure since concepts, however undesirable they may be, are innate and do not depend upon natural language.
The innateness of concepts is the most startling consequence of the Language of Thought Hypothesis. Unlike English or French, we do not learn mentalese, and if this is true, then it becomes a mystery how we can manage to acquire new concepts, since any we come across will be rendered in English – with no pre-existing key to translate them into the language of thought. So if we are to be capable of acquiring new concepts, we must possess the code for them already, waiting inside us to be awakened. Given our evolutionary history, it is just possible that concepts such as ‘tiger’ and ‘man-eating’ might have been hardwired into us at some point. But it is less likely that others such as ‘satellite’, ‘Internet’ and ‘X-ray’ could have got in in that way. Fodor’s best hope is that because mentalese is a combinatorial language composed of various elements, complicated concepts can be formed out of simpler ones – in the way that ‘bachelor’ can be formed out of ‘man’ and ‘unmarried’. Such words could be eliminated from the dictionary without harming our powers of expression. It would not be too much trouble to find a different way of saying what we mean. However, this is not the case with words such as ‘green’ and ‘blue’, or even ‘hot’ and ‘cold’. It is difficult, if not impossible, to define them without using the concept in its definition – hence the perennial conundrum of how to explain colours to someone blind from birth. This is why philosophers call them ‘primitive’ concepts. The question then is how many of our concepts are primitive, and the answer, it seems, is quite a lot of them.
Fodor himself believes that almost all concepts are primitive. Because this particular nativist position sounds so absurd, philosophers commonly refer to it as ‘Mad Dog’ nativism. Whoever coined the phrase could not have had its softly spoken champion’s persona in mind, but his ideas, however, are another matter. As he patiently explained to me:
My own nativism is very extreme. I’m sceptical about whether there is such a process as learning. I don’t think learning is a very clear idea, and I think as you try to apply it the concept gets less and less clear. I think there must be an enormous amount of innate cognitive structure and information, but this is an eccentric view. My guess is that if you want to know which concepts are innate, ask yourself which concepts get coded by single words. Everyone agrees that primitive concepts are innate. Even Locke and Hume thought that sensory concepts were innate. They cannot be learned. So the only concepts that can be learned are those that are reducible to the basic concepts, whatever they are. What I’m suggesting is that all concepts are primitive. I don’t think ‘brown cow’ is a concept – you can see how that’s constructed. But the primitive vocabulary has to be big enough to encompass everything that is irreducible, and as far as we know, damn near everything is irreducible.
Unfortunately, attempts to perform such reductions to basic concepts have been uniform failures. That we can entertain an infinite number of thoughts, understand an infinite number of sentences – even those we have never entertained before – is, for Fodor, explained by the fact that mentalese has a compositional semantics, meaning that beliefs are composed of various elements and have a combinatorial structure. Unfortunately, this was what philosophers used to say about natural languages, and there is no reason to assume that we will have any more success laying down the rules for their elusive language of thought than we ever had with, say, English and French. However, this can just as easily work in Fodor’s favour. Neuroscience has yet to come across any symbols in mentalese among our neurons, but neither has it come across the words of poetry that people can recite or the figures in algebraic formulae that mathematicians know, even though conceptual processing must occur somewhere in the brain. Fodor explains:
Suppose that just as you can look in a magazine and find inscriptions of sentences, there are inscriptions of mental representations in your head. They are involved in causal relations just like inscriptions in magazines are. That has to be true if the theory is right. What we don’t know – but it really hasn’t mattered so far – is what the neurological or physical or chemical realization of these things is. We know that inscriptions in books are made of ink, but we don’t know the corresponding fact about mental representations. My attitude to this subject is perfectly realistic with a capital R. If the theory is true it is true because the brain or the nervous system or the soul or whatever it is acts in the way the theory describes.
The eighteenth-century empiricists who denied the existence of innate ideas faced a problem concerning how mental representations were combined to create articulated expressions and complex concepts and experiences, and settled on rules of association as the answer. The Scottish thinker David Hume’s suggestions were resemblance and contiguity (or cause and effect). As Fodor points out, ‘They couldn’t distinguish between thinking the cow is red or thinking red cow and thinking red and then thinking cow because the two are associated.’ Fodor wishes to replace this failed model with one of computation. Hume’s associations lacked an associator. That is, they lacked an ‘inner self’ that would manipulate the ideas and impressions to give the desired effect. The hope is that computational values will be able to think themselves into shape without the need for such an entity. As the otherwise unsympathetic Daniel Dennett writes: ‘perhaps the prima facie absurd notion of self-understanding representations is an idea whose time has come, for what are the “data structures” of computer science if not just that: representations that understand themselves? In a computer, a command to dig goes straight to the shovel, as it were, eliminating the comprehending and obeying middleman.’6 As Dennett adds, one may prefer to judge that this lack of an intelligent middleman is precisely why computers do not contain representations, but to think in this way is to reject one of the most promising advances in all of philosophy. Fodor himself describes it as ‘the best research strategy that’s become visible so far. If it doesn’t work, we will learn a lot from its not working. It’s not going to not work for trivial reasons.’
Fodor’s theory is the kind of speculation that all Western philosophers traded in before the Linguistic Turn of the early twentieth century persuaded them to stick to conceptual analysis and leave the understanding of nature to scientists. It is philosophy as a protoscience that, if successful, might spawn a new empirical discipline for the first time since Noam Chomsky invented linguistics. If the long-lost ‘language of Babel’ is proven to exist, then it would raise the possibility that, just as the brain processes the evidence of our senses to make the outside world intelligible, it also distorts our thoughts when it sends them into the world as speech.
