By Andrea Moro

MIT Press ISBN 9 780262 034890

reviewed by Alan Dent


            Impossible animals are easy to imagine; one which generates more energy than it consumes, for example. Is the same true of languages ? That is, do languages obey laws which limit them in the way physical laws limit animals ? “It is hard to think of a physical law..that would make language impossible” writes Moro. The same is not true, however, of formal, internal laws. It is easy to find those. Chomsky’s Universal Grammar reveals the strict limits which make language possible. Merge for example, defined by Chomsky as “an operation that takes two objects already constructed and forms a new one”. A language that doesn’t do that would be impossible.

            Defining impossible languages is a back handed way of illustrating what makes language possible. Eric Lenneberg argued in 1967 that though the rules of language superficially resemble those of games, in fact : “The former are biologically determined; the latter are arbitrary.” This is the crucial distinction. That different languages have particular grammars has misled many, including Rabelais, to conclude that they are arbitrary, cultural constructs. Look beneath what is “visible and complicated” to discover what is “simple and invisible” as the physicist Jean-Baptiste Perrin put it and you find that the differing grammars emerge from a limited, universal set of rules. It is the limits which make the flexibility of language possible.

            It has been accepted for a hundred years, however, that language embraces something arbitrary: the “association between sounds and meanings” as Moro puts it. This is the Saussurean assumption according to which there is no necessary reason why, for example, the sound “love” should be associated with a positive feeling and “hate” with a negative one. It is simply a matter of agreement. The linguistic community “agrees” to the association and therefore it prevails. This leaves unanswered the vital question of how the agreement comes into being, and the equally vital one of what prevents it breaking down. More seriously, however, it is mistaken. It ought to be called  the Saussurean Fallacy. Say the word “love”. As you do so your lips will purse slightly, as if in a kiss. Say the word “hate”. Your mouth will stretch, showing your teeth in an incipient snarl. This is not accidental. The association between the sounds of these words and their meanings comes from what the neuroscientist Ramachandran calls “ a pre-existing cross-translation” between different areas of the brain. His bouba-kiki effect is an amusing but decisive illustration of this.

            Of course, there is a degree of arbitrariness between sound and sense which permits “small”, “petit”, “klein”, “pequeno” etc to have the same meaning. Yet this does not negate the fact that within each language the cross-translation is at work. What works for “love” and “hate” in English works in most, if not all languages.

            Accounting for language requires, of course, an explanation of the multiplicity of languages. Moro suggests that the Babel effect may have helped prevent overpopulation in areas where resources would have been inadequate if language differences hadn’t prevented too many people living together. He relates this to Jerne’s ideas about the immune system: that we are born with the capacity to fight many more antigens than we will encounter. Biology has provided for multiple languages. Given the flexibility of our organs of speech (which didn’t evolve for language) this looks inevitable. Universal grammar’s rules do not include that the only sound to be associated with the tall thing with leaves must be “tree”.

            Chomsky has written that language is more like a snowflake than a giraffe’s neck. What this means is that “minimal components..combined with simple rules..recursively applied” can result in great complexity. What goes along with this is that there can be no proto-syntax. Every sentence employs the whole structure of syntax, in the same sense that adding 2+2 employs or at least implies all the rules of arithmetic. This is uncontroversial. However, we must bear in mind that language can function when syntax breaks down. Just as Chomsky’s famous “Colorless green ideas sleep furiously” is syntactically perfect but semantic nonsense, so “Er…throat…dry…water” while not a sentence (and in the strict sense therefore not language) works semantically. Chomsky refines his idea about snowflakes and giraffe’s necks thus: “Its (language’s) specific properties are determined by laws of nature; they have not developed through the accumulation of historical accidents.”

            Does this mean that historical accidents are not a law of nature ? A snowflake is merely physical, a giraffe’s neck biological. A snowflake’s specific properties are determined by the laws of Physics.  If Lenneberg is right, and surely he is,  language is determined by biology, something Chomsky proposes. Isn’t historical accident a biological law ? Couldn’t the “specific properties” of language be determined by laws which include historical accidents ? Proto-syntax is impossible in the sense mentioned above, but is proto-language  ? Might syntax have emerged from proto-language ?

            Syntax is a sieve. It filters out what prevents a sentence working and it does so without us being aware of the filtering. It’s intuitive, like walking. Syntax is like a rainbow: our eyes have evolved to perceive a narrow range of light. Without this we would be overwhelmed by signals from light. Syntax works in the same way. It’s a limitation. From the limitation comes infinite structures. Our brains are set-up to converge on the language or languages which prevail in our environment. We can assimilate any of the thousands of languages in the world, but there is a limiting mechanism, like a speed regulator on an engine, which forces our attention onto the prevailing language or languages.

            Syntax is “autonomously represented in the brain”. The pars triangularis, part of the Broca’s area, is implicated in syntactic errors. This doesn’t provide conclusive evidence that this part of the brain is exclusively responsible for particular syntactic functions. The brain is more complex. Activity in one area doesn’t necessarily mean that other areas are not crucially involved. Tracking particular language functions down to specific neurons is a long way away and may never be achieved.

            Moro sees Giacomo Rizzolatti’s theory of mirror neurons as reductionist and believes it would be better if it could be shown to be wrong. Otherwise, he contends, it will impossible to establish language as exclusively human as gestures and action planning are shared by all primates. Moro points out that recursive, hierarchical structures are not visible and therefore the mechanism suggested by mirror neuron theory doesn’t apply. Hierarchical structures are obviously not visible but spoken language is audible and written language visible. Perhaps when we hear a sentence spoken, the neurons which had to fire in the speaker’s brain are activated in ours. Maybe when we read a sentence, the neurons which fired in the author’s brain are activated in ours. Might it be worth devising a few ingenious experiments to test this ? Just what this might suggest is not clear but it might mean that syntax is possible only because of the action of mirror neurons. Language capacity is what is biologically wired-in. Might it be mirror neurons which allow the art to be acquired by instinct ?

            The old notion that analogy may be sufficient to account for syntax is easily dismissed by Moro. That analogy permits the assimilation of vocabulary or the repetition of more or less redundant sentences is simple enough; but syntax permits infinite generation, something which analogy can’t account for.

            Language is a wave (an idea which might have appealed to Einstein). Whether externalised in speech when it becomes the movement of air waves or internal when it is waves of electrical impulses firing neurons. It’s known that the sound waves and the electrical waves which convert what is received by the ear into signals the brain can decipher, imitate one another. The same is true of endophasic activity (reading to yourself i.e. not actually sounding the words): the electrical waves in the Broca’s or related areas preserve the form of the physical waves which would have been produced had the words read been sounded. This doesn’t mean, of course, that sound waves alone contain all the information needed to process language, otherwise ambiguity would be impossible (I saw Galileo with a telescope, for example). What it does mean is that acoustic information is embedded in language capacity, it is part of language competence, that is, of our biological inheritance.

            It is virtually impossible to stop talking to yourself. Some people do it out loud (professors and nerds who read difficult books about language for example) but whether they do or they don’t, the brain generates language as endlessly and as easily as heat pours from the sun. Most language use is internal. When we speak what we communicate is mostly non-verbal. When we read, logically, the communication must be entirely verbal; but try these:

                        Don’t  you speak to me like that.
                        Don’t you speak Spanish ?

There is not much difference between the two sentences, yet aren’t you aware of a great distance in tone ? Doesn’t that come from attribution ? Don’t you attribute to the words an emotional tone which is imported from experience ? Couldn’t it be argued, therefore, that even in written language much of what is communicated is non-verbal in that to understand fully what we read we must attribute an emotional tone drawn from previous occasions; (isn’t this what makes scientific papers so hard to read?: they lack emotional tone and rely on pure intellectual functioning).

            No human can exist without language, Moro asserts. It may be more accurate to say no human can exist without language capacity but even then we have to be careful: a brain-damaged new-born who cannot assimilate language is human. We must not set the limits too narrowly or play god in deciding who is to be deemed human. Moro sensibly quotes Peter Medawar’s caveat that if a trait is genetically programmed there must be people who lack it.  Yet the point is essentially sound. We are, as a species, genetically linguistic. We are remote from a genetic theory of language capacity, but it may be that language capacity is provided by genes which are crucial for human life.

            The evidence that language capacity didn’t evolve for communication is strong. Communication doesn’t require recursive embedding:

                        The train arrives at twelve  does the job. So, as a matter of fact, does train here twelve.

                        The train that arrives at twelve bringing my wife who has the money the man gave her to buy the car my brother told me about when I met him by chance on the street on my way to see the solicitor about our house purchase, is an express is not the kind of structure you need if all you want to do is communicate. This recursive embedding serves thought. It is perfectly possible that the primary function of language is internal and external communication, secondary.

            Language, Moro says, is like light. We know it’s there because of the effects it creates. Light itself is invisible. We can’t see language capacity either but know it must be there because of its effects. Trying to understand language capacity is a bit like trying to see the back of your own head without a mirror. What we have is the evidence of words and sentences and the way we use them without knowing how we do it. Form this paucity of data we have to try to piece together a theory.

            Language capacity permits us to combine a restricted set of components in a recursive way to produce an infinite number of sentences. No other animal can do this. There is a similarity between language and a snowflake, but Chomsky’s assertion that a snowflake is a product of the “laws of nature” and that language must be like it i.e. not the result of a series of historical accidents is not conclusive. Language capacity is biological and biology is subject to evolution and natural selection. The latter is, of course, not the sole mover. Yet physicist can explain how snowflakes came into exist, but not how language capacity arrived.

            The Saussurean Fallacy may point, not to a proto-syntax but to a proto-language, internally employed to facilitate  metaphorical understanding and externally for communication, which gave rise to syntax. Might this have happened quickly enough to look like emergence ?

            Moro’s book is an excellent place to begin if these questions interest you.