The Language Instinct: How the Mind Creates Language

by Steven Pinker

even if the genes themselves specify the basic design of language, they might have to store the specifics of language in the environment, to ensure that a person’s language is synchronized with everyone else’s despite the genetic uniqueness of every individual. In this sense, language is like another quintessentially social activity.

If children can infer parents’ meanings, they do not have to be pure cryptographers, trying to crack a code from the statistical structure of the transmissions. They can be a bit more like the archeologists with the Rosetta Stone, who had both a passage from an unknown language and its translation in a known one. For the child, the unknown language is English (or Japanese or Inslekampx or Arabic); the known one is mentalese.

If a parent punished, corrected, misunderstood, or even reacted differently to a child’s ungrammatical sentence, it could in theory inform the child that something in his growing rule system needed to be improved. But parents are remarkably unconcerned about their children’s grammar; they care about truthfulness and good behavior.

As far as grammar learning goes, the child must be a naturalist, passively observing the speech of others, rather than an experimentalist, manipulating stimuli and recording the results. The implications are profound. Languages are infinite, childhoods finite. To become speakers, children cannot just memorize; they must leap into the linguistic unknown and generalize to an infinite world of as-yet-unspoken sentences.

Therefore a well-designed child, when faced with several choices in how far to generalize, should, in general, be consecutive: start with the smallest hypothesis about the language that is consistent with what parents say, then expand it outward as the evidence requires.

Furthermore, in cases where children do make errors and recover, their grammars must have some internal checks and balances, so that hearing one kind of sentence can catapult another out of the grammar. For example, if the word-building system is organized so that an irregular form listed in the mental dictionary blocks the application of the corresponding rule, hearing held enough times will eventually drive out holded.

that the child must couch rules in grammatical categories like noun, verb, and auxiliary, not in actual words. That way, fish as a noun and fish as a verb would be kept separate, and the child would not adulterate the noun rule with instances of verbs and vice versa.

The use of part-of-speech categories, X-bar phrase structure, and meaning guessed from context is amazingly powerful, but amazing power is what a real-life child needs to learn grammar so quickly, especially without parental feedback.

If children had to learn all the combinations separately, they would need to listen to about 140 million different sentences. At a rate of a sentence every ten seconds, ten hours a day, it would take over a century. But by unconsciously labeling all nouns as “N” and all noun phrases as “NP,” the child has only to hear about twenty-five different kinds of noun phrase and learn the nouns one by one, and the millions of possible combinations become available automatically.

Once a rudimentary but roughly accurate analysis of sentence structure has been set up, the rest of the language can fall into place.

Even the adults who succeed at grammar often depend on the conscious exercise of their considerable intellects, unlike children, to whom language acquisition just happens.

In sum, acquisition of a normal language is guaranteed for children up to the age of six, is steadily compromised from then until shortly after puberty, and is rare thereafter. Maturational changes in the brain, such as the decline in metabolic rate and number of neurons during the early school-age years, and the bottoming out of the number of synapses and metabolic rate around puberty, are plausible causes. We do know that the language-learning circuitry of the brain is more plastic in childhood; children learn or recover language when the left hemisphere of the brain is damaged or even surgically removed (though not quite at normal levels), but comparable damage in an adult usually leads to permanent aphasia.

The genes, shaped by natural selection, control bodies throughout the life span; designs hang around during the times of life that they are useful, not before or after. The reason that we have arms at age sixty is not because they have stuck around since birth, but because arms are as useful to a sixty-year-old as they were to a baby.

language-acquisition circuitry is not needed once it has been used; it should be dismantled if keeping it around incurs any costs. And it probably does incur costs. Metabolically, the brain is a pig. It

So genes that strengthen young organisms at the expense of old organisms have the odds in their favor and will tend to accumulate over evolutionary timespans, whatever the bodily system, and the result is overall senescence.

Thus language acquisition might be like other biological functions. The linguistic clumsiness of tourists and students might be the price we pay for the linguistic genius we displayed as babies, just as the decrepitude of age is the price we pay for the vigor of youth.

Deaf signers with damage to their left hemispheres suffer from forms of sign aphasia that are virtually identical to the aphasia of hearing victims with similar lesions.

It is a fascinating discovery. The right hemisphere is known to specialize in visuospatial abilities, so one might have expected that sign language, which depends on visuospatial abilities, would be computed in the right hemisphere. Bellugi’s findings show that language, whether by ear and mouth or by eye and hand, is controlled by the left hemisphere. The left hemisphere must be handling the abstract rules and trees underlying language, the grammar and the dictionary and the anatomy of words, and not merely the sounds and the mouthings at the surface.

Processing information that is spread out over time but not space is another function where symmetry serves no purpose. Given a certain amount of neural tissue necessary to perform such a function, it makes more sense to put it all in one place with short interconnections, rather than have half of it communicate with the other half over a slow, noisy, long-distance connection between the hemispheres.

The grammar genes would be stretches of DNA that code for proteins, or trigger the transcription of proteins, in certain times and places in the brain, that guide, attract, or glue neurons into networks that, in combination with the synaptic tuning that takes place during learning, are necessary to compute the solution to some grammatical problem (like choosing an affix or a word).

The reason is that the relationship between particular genes and particular psychological traits is doubly indirect. First, a single gene does not build a single brain module; the brain is a delicately layered soufflé in which each gene product is an ingredient with a complex effect on many properties of many circuits. Second, a single brain module does not produce a single behavioral trait. Most of the traits that capture our attention emerge out of unique combinations of kinks in many different modules.

Nonhuman communication systems are based on one of three designs: a finite repertory of calls (one for warnings of predators, one for claims to territory, and so on), a continuous analog signal that registers the magnitude of some state (the livelier the dance of the bee, the richer the food source that it is telling its hivemates about), or a series of random variations on a theme (a birdsong repeated with a new twist each time: Charlie Parker with feathers).

For all we know, language could have had a gradual fade-in, even if no extant species, not even our closest living relatives the chimpanzees, have it. There were plenty of organisms with intermediate language abilities, but they are all dead.

Language could have arisen, and probably did arise, in a similar way: by a revamping of primate brain circuits that originally had no role in vocal communication, and by the addition of some new ones.

Also unknown is when, in the lineage beginning at the chimp-human common ancestor, proto-language first evolved, or the rate at which it developed into the modern language instinct.

The positioning of the larynx deep in the throat, and the tongue far enough low and back to articulate a range of vowels, also compromised breathing and chewing. Presumably the communicative benefits outweighed the physiological costs.

Natural selection applies to any set of entities with the properties of multiplication, variation, and heredity. Multiplication means that the entities copy themselves, that the copies are also capable of copying themselves, and so on. Variation means that the copying is not perfect; errors crop up from time to time, and these errors may give an entity traits that enable it to copy itself at higher or lower rates relative to other entities. Heredity means that a variant trait produced by a copying error reappears in subsequent copies, so the trait is perpetuated in the lineage. Natural selection is the mathematically necessary outcome that any traits that foster superior replication will tend to spread through the population over many generations. As a result, the entities will come to have traits that appear to have been designed for effective replication, including traits that are means to this end, like the ability to gather energy and materials from the environment and to safeguard them from competitors. These replicating entities are what we recognize as “organisms,” and the replication-enhancing traits they accumulated by this process are called “adaptations.”

it is not so easy to show that a trait is a product of selection. The trait has to be hereditary. It has to enhance the probability of reproduction of the organism, relative to organisms without the trait, in an environment like the one its ancestors lived in. There has to have been a sufficiently long lineage of similar organisms in the past. And because natural selection has no foresight, each intermediate stage in the evolution of an organ must have conferred some reproductive advantage on its possessor. Darwin noted that his theory made strong predictions and could easily be falsified. All it would take is the discovery of a trait that showed signs of design but that appeared somewhere other than at the end of a linage of replicators that could have used it to help in their replication.

Another way of putting it is that natural selection is the only process that can steer a lineage of organisms along the path in the astronomically vast space of possible bodies leading from a body with no eye to a body with a functioning eye. The alternatives to natural selection can, in contrast, only grope randomly. The odds that the coincidences of genetic drift would result in just the right genes coming together to build a functioning eye are infinitesimally small.

And adaptive complexity, by the way, is also the reason that the evolution of complex organs tends to be slow and gradual. It is not that large mutations and rapid change violate some law of evolution. It is only that complex engineering requires precise arrangements of delicate parts, and if the engineering is accomplished by accumulating random changes, those changes had better be small. Complex organs evolve by small steps for the same reason that a watchmaker does not use a sledgehammer and a surgeon does not use a meat cleaver.

If Chomsky maintains that grammar shows signs of complex design but is skeptical that natural selection manufactured it, what alternative does he have in mind? What he repeatedly mentions is physical law. Just as the flying fish is compelled to return to the water and calcium-filled bones are compelled to be white, human brains might, for all we know, be compelled to contain circuits for Universal Grammar.

Why would evolution ever have selected for sheer bigness of brain, that bulbous, metabolically greedy organ? A large-brained creature is sentenced to a life that combines all the disadvantages of balancing a watermelon on a broomstick, running in place in a down jacket, and, for women, passing a large kidney stone every few years. Any selection on brain size itself would surely have favored the pinhead. Selection for more powerful computational abilities (language, perception, reasoning, and so on) must have given us a big brain as a by-product, not the other way around!

All the evidence suggests that it is the precise wiring of the brain’s microcircuitry that makes language happen, not gross size, shape, or neuron packing. The pitiless laws of physics are unlikely to have done us the favor of hooking up that circuitry so that we could communicate with one another in words.

If a grammar mutant is making important distinctions that can be decoded by others only with uncertainty and great mental effort, it could set up a pressure for them to evolve the matching system that allows those distinctions to be recovered reliably by an automatic, unconscious parsing process. As I mentioned in Chapter 8, natural selection can take skills that are acquired with effort and uncertainty and hardwire them into the brain. Selection could have ratcheted up language abilities by favoring the speakers in each generation that the hearers could best decode, and the hearers who could best decode the speakers.

Grammars of intermediate complexity are easy to imagine; they could have symbols with a narrower range, rules that are less reliably applied, modules with fewer rules, and so on.

The languages of children, pidgin speakers, immigrants, tourists, aphasics, telegrams, and headlines show that there is a vast continuum of viable language systems varying in efficiency and expressive power, exactly what the theory of natural selection requires.

And a cognitive arms race clearly could propel a linguistic one. In all cultures, social interactions are mediated by persuasion and argument. How a choice is framed plays a large role in determining which alternative people choose. Thus there could easily have been selection for any edge in the ability to frame an offer so that it appears to present maximal benefit and minimal cost to the negotiating partner, and in the ability to see through such attempts and to formulate attractive counterproposals.

The rules people learn (or, more likely, fail to learn) in school are called prescriptive rules, prescribing how one “ought” to talk. Scientists studying language propose descriptive rules, describing how people do talk.

if you make up a new verb based on a noun, like someone’s name, it is always regular, even if the new verb sounds the same as an existing verb that is irregular.

chances are you would never have made it to the last chapter of a book about the human hand. People are more than curious about language; they are passionate. The reason is obvious. Language is the most accessible part of the mind. People want to know about language because they hope this knowledge will lead to insight about human nature.

Modern intellectual life is suffused with a relativism that denies that there is such a thing as a universal human nature, and the existence of a language instinct in any form challenges that denial.

For we can now do justice to the complexity of the human brain, the immediate cause of all perception, learning, and behavior. Learning is not an alternative to innateness; without an innate mechanism to do the learning, it could not happen at all. The insights we have gained about the language instinct make this clear.

Tooby and Cosmides, in their important recent essay “The Psychological Foundations of Culture,” call it the Integrated Causal Model, because it seeks to explain how evolution caused the emergence of a brain, which causes psychological processes like knowing and learning, which cause the acquisition of the values and knowledge that make up a person’s culture. It thus integrates psychology and anthropology into the rest of the natural sciences, especially neuroscience and evolutionary biology. Because of this last connection, they also call it Evolutionary Psychology.

A mechanism designed to learn the right thing in one of these domains learns exactly the wrong thing in the others. This suggests that learning is accomplished not by some single general-purpose device but by different modules, each keyed to the peculiar logic and laws of one domain. People are flexible, not because the environment pounds or sculpts their minds into arbitrary shapes, but because their minds contain so many different modules, each with provisions to learn in its own way.

culture is given its due, but not as some disembodied ghostly process or fundamental force of nature. “Culture” refers to the process whereby particular kinds of learning contagiously spread from person to person in a community and minds become coordinated into shared patterns, just as “a language” or “a dialect” refers to the process whereby the different speakers in a community acquire highly similar mental grammars.

Most learning, everyone agrees, takes place outside of classroom lessons, by generalizing from examples. Children generalize from role models, or from their own behaviors that are rewarded or not rewarded. The power comes from the generalization according to similarity. A child who echoed back a parent’s sentences verbatim would be called autistic, not a powerful learner; children generalize to sentences that are similar to their parents’, not to those sentences exactly.

That is, the “similarity” guiding the child’s generalization has to be an analysis of speech into nouns and verbs and phrases, computed by the Universal Grammar built into the learning mechanisms. Without such innate computation defining which sentence is similar to which other ones, the child would have no way of correctly generalizing—any sentence is “similar,” in one sense, to nothing but a verbatim repetition of itself, and also “similar,” in another sense, to any random rearrangement of those words, and “similar,” in still other senses, to all kinds of other inappropriate word strings. This is why it is no paradox to say that flexibility in learned behavior require innate constraints on the mind.

The qualities that should classify people as potential allies, such as showing signs of affection, should not necessarily classify them as potential mates, such as showing signs of fertility and not being close blood relatives. There must be many similarity spaces, defined by different instincts or modules, allowing those modules to generalize intelligently in some domain of knowledge such as the physical world, the biological world, or the social world.

Using data from psychology and ethnography, we can test the following prediction: when children solve problems for which they have mental modules, they should look like geniuses, knowing things they have not been taught; when they solve problems that their minds are not equipped for, it should be a long hard slog.

So the language instinct suggests a mind of adapted computational modules rather than the blank slate, lump of wax, or general-purpose computer of the Standard Social Science Model.