VOICE IN THE WILDERNESS
How Humans Acquired the Power of Speech
by Phillip Lieberman

    THEIR NAMES were Jaguar of Sweet Laughter, Jaguar of the Night, Black Jaguar, and Mahucutah, the Not-Brushed. According to Popol Vuh, the sacred book of the Maya, they were the first creatures "to come out human." The four looked and listened, walked and worked; "they talked and they made words." Like other creation myths, the Popo Vuh attributes the origin of man, and all of his qualities, to an agent that exists above, before, or beyond nature--the Sovereign Plumed Spirit.

    Some ideas never die. Though creationism as an explanation of the origin of life has been displaced by evolutionary theory, the idea persists as an explanation of the origin of language. The terms are different, of course; we no longer speak of plumed spirits, sovereign or otherwise. But the essential elements of the story remain the same: the ability to make words appeared spontaneously, in full bloom, without ties to man's biological past.

    The architect of this view, the linguist Noam Chomsky, of the Massachusetts Institute of Technology, believes the human brain is organized in terms of modules, each operating independently of the others. Moreover, the module that governs linguistic ability--the language organ, as it were--has no counterpart among other animals, including the other primates. Chomsky argues that, like language, this organ appeared in man suddenly and in its present form. In short, the biological foundation of human language lies beyond the explanatory range of twentieth-century science.

    The principal corollary of this theory, indeed, the idea that has influenced modern linguistics most, is that the language organ is the basis of a universal syntax--an innate set of grammatical laws that enables people to generate and comprehend speech. In principle, all the world's languages are reducible to these deep structures, as Chomsky calls them, for they are largely independent of meaning, as illustrated in the semantically nonsensical but syntactically perfect sentence "Angry chairs eat the night." Like the language organ itself, so the theory goes, the rules that govern speech--the orderly manipulation of sounds--are without biological precedent.

    Chomsky's case is plausible--to a point. Human speech is unique; relying on symbols to represent abstract thoughts, language little resembles such simple signals as the songs of birds or the barks and cries of apes. And linguistic ability is associated with specific neural structures found only in human brains. Indeed, the most important one was identified as long ago as 1861. Called Broca's area, after its discoverer, the French neurologist Pierre-Paul Broca, it is located toward the rear of the left frontal lobe of the cerebral cortex. People with extensive lesions in this region exhibit Broca's aphasia: they can control the tongue, lips, and other parts of the vocal apparatus, but they cannot produce all the vowels and consonants that constitute speech. Although the idea that a specific part of the brain controls language has been modified considerably since the nineteenth century (the neural bases of all mental functions are considerably more diffuse than once thought), it is clear that human speech cannot exist in the absence of Broca's area.

    A problem arises, however, with the next step in Chomsky's argument: that, since speech and the speech-related parts of the brain are found in Homo sapiens alone, they emerged ex nihilo. This conclusion is based on a profound misunderstanding of nature's inventiveness. As far as anyone knows, only one creature on Earth bears an elephant trunk, but no scientist would seriously claim that, therefore, the elephant's trunk is the work of processes other than those that drive evolution. Then why jump to the same conclusion about the neural mechanisms underlying speech?

    Besides, much more is known about the origin of language today than was known three decades ago, when Chomsky first proposed his ideas. The neural mechanisms for speech and syntax not only are the outcome of Darwinian processes; they did not evolve alone. During the past several hundred thousand years, changes in vocal anatomy prompted alterations in the brain, and vice versa. Without this stepwise, hand-in-hand elaboration, human speech would not have developed. Contrary to Chomsky and like-minded linguists, then, modern man is what he is not despite evolution but because of it. Biologists, of course, have always taken this as an article of faith. But now, for the first time, the idea can be proved; the biological bases of speech have been identified.

    ONE OF THE SIMPLEST but long overlooked clues to the special status of human speech is the nature of the sounds that compose it. During the first half of this century linguists presumed that any set of sounds would suffice to transmit words. But that is not so, as was discovered during the first attempt to make a machine that reads books to blind people, which took place during the 1950s and 1960s at Haskins Laboratories, a privately endowed research firm then in New York City.

    A reading machine requires several components--a print reader, a page turner, and the like--but the key to the device's success is a mechanism that converts phonetic symbols into sounds. Haskins researchers believed that the first step in constructing such a mechanism was to record on tape all the phonetic elements used to make English words. They assumed that a recording of the word cat, for instance, could be sliced into three segments corresponding to the initial k sound, the final t sound, and the intermediate ae sound. But words constructed of such discrete segments were virtually unintelligible, because, the researchers found, it is impossible to separate actual speech sounds in this way! No matter how short a tape segment is cut, one will always hear the consonant k and the ae that follows. The two are merged, or encoded, to such an extent that they are transmitted in the same time that it takes to transmit a single vowel or consonant.

    To appreciate the importance of this, imagine a pencil tapping on a tabletop. Beyond a rate of about nine taps a second, most people cannot keep track of the sounds. And beyond fifteen taps a second, the individual sounds are perceived as a buzz. In contrast, a short sentence, such as this one, contains fifty or so speech sounds. It can be uttered in about two seconds and human listeners have no trouble comprehending it. If the sentence were transmitted at the nonspeech rate (seven to nine sounds a second), a listener might forget the sentence's beginning before hearing its end. The ability to encode phonetic sounds is thus an integral part of linguistic ability, because it allows a great deal of information to be transmitted within the constraint of short-term memory. The only species whose vocal apparatus is suited to making the entire range of these sounds is Homo sapiens.

    In the most general sense, the human vocal tract is the top half of the airway that connects the lungs to the atmosphere. The airway itself consists of several distinct sections. Extending from the lungs to about the level of the Adam's apple is the trachea, or windpipe, capped by the larynx, or voice box. Above the larynx, the trachea joins the esophagus, which connects the stomach to the mouth. Ordinarily referred to as the throat, this region, where the air and food tubes come together, branches into two passages--the oral and nasal cavities--to complete the airway. As far as speech is concerned, the most important parts of this system are the voice box and the vocal tract which is composed of the throat and the oral and nasal passages, including the velum (the flexible, soft part of the palate that closes the nose to the mouth), the tongue, and the lips.

    Sound is produced when air flowing from the lungs passes into the voice box and through the gap made by the vocal cords, which open and close like a valve. During speech, the vocal cords vibrate as air passes through them, releasing a rapid series of puffs into the vocal tract. The rate at which the puffs are produced, called the fundamental frequency of phonation, determines the pitch of a person's voice; the faster the rate, the higher the pitch. But speech sounds have other qualities besides pitch.

    In addition to the acoustic energy at fundamental frequencies, energy is present at their overtones--the harmonics of fundamental frequencies. By means of its shape--the width of the throat, the location of the velum and the tongue, and the position of the mouth--the vocal tract acts as a filter that suppresses some of these harmonics, letting maximum energy through at certain frequencies, called formant frequencies, and thereby producing discrete speech sounds. The vowel e of met, for example, differs from the vowel i of bit, because it is made up of dissimilar formant frequencies. The process is similar to the one in which the pipes of an organ filter sound. Whereas the musical quality of a note is determined by the shape of each pipe, the phonetic quality of a speech sound is determined by the configuration of the vocal tract.

    All apes, including gibbons and chimpanzees, as well as human infants under the age of three months, lack the physical apparatus to produce a complete range of formant frequencies--the encoded sounds of speech, such as the k and ae of cat. The main reason for this incapacity is that, in these groups, the voice box is located high in the throat, creating a vocal tract shaped like a single, curved tube, whereas, in mature Homo sapiens, the larynx is situated low in the throat, resulting in a vocal tract that is essentially two tubes--the oral passage and the throat--joined at a right angle behind the tongue. So critical is the overall configuration of the vocal tract to the generation of formant frequencies that only when the voice box has retreated into the throat is the capacity for speech virtually certain.

    Oddly enough, besides giving man the ability to talk, a low larynx greatly increases his chances of choking. In all other terrestrial mammals--dogs, cats, apes--the larynx is elevated, so it can move up in the throat, like a periscope, and lock in to the opening of the nose, permitting air to pass through it and into the lungs, while food and water pass around it and into the esophagus. Thus, every mammal but Homo sapiens can breathe and drink at the same time. In the human throat, everything swallowed passes over the opening of the trachea (a liability that infants are spared), and, each year, thousands of people die when food or drink lodges in the trachea, obstructing the pathway to the lungs.

    The vulnerability of this anatomical configuration suggests that, in the course of evolution, humans encountered conditions that favored the development of vocal communication; there would have been no other reason to risk an anatomy that makes speech sounds possible. Tracing the evolution of the larynx and the vocal tract will make this point clearer. More than that, it will help show how the speech centers of the brain also are an outcome of evolutionary events.

     NATURE IS A MASTER of bricolage: whenever it needs something new, it makes use of whatever materials are at hand. And never is nature more resourceful than during the process of preadaption, when a structure that evolved to perform a function under one set of circumstances gradually assumes a different function under other circumstances. As it happens, one of the classic examples of this opportunistic approach to invention, first described by Charles Darwin, is the evolution of the mammalian respiratory system.

    The starting point was the swim bladder of the lungfish. About four hundred million years ago, the lungfish evolved the ability to breathe air directly from the external environment (perhaps because its watery home was periodically subject to drought). It developed a simple larynxlike slit, behind the gills, that allowed air into the swim bladder when the creature was exposed to the atmosphere and that kept water out when it was submerged. As the lungfish's descendants moved onto land and became acclimated to a terrestrial existence, the swim bladder evolved into an organ whose sole function was respiration. No longer needed for flotation, this cavity became a complex network of smaller and smaller compartments that greatly enlarged the surface area through which oxygen could enter, and carbon dioxide exit, the bloodstream. The mammalian lung was born.

    In time, as vocal communication came to play an important role--in signaling potential mates, warning others of danger, and so on--the airway that connected the lungs to the mouth was modified. Some mammals began to rely on phonation more than others, and their larynges show it. In horses, for example, the vocal cords can be drawn to the sides of the voice box like a pair of lips, creating an opening larger than the trachea itself. In this way, the voice box is designed to reduce, as much as possible, resistance to airflow. Thus, the horse can deliver to its lungs a great deal of oxygen, a critical factor in an animal whose survival once depended on running long distances to escape predators. Phonation--chiefly, the neigh--is part of the horse's behavioral repertoire, but it is not nearly as important as respiration. Man, in contrast, sacrificed a certain degree of respiratory efficiency to be able to make speech sounds. Fully open, the human voice box is only one-half the diameter of the trachea. For the sake of phonation, the vocal cords partially obstruct the flow of air from the lungs, just as a reed diverts airflow in a clarinet. To a lesser extent, the same is true of all social mammals--wolves, lions, chimpanzees--so, to varying degrees, the voice box was readapted for both respiration and phonation.

    Once man's hominid ancestors began to reap the benefits of phonation--the adaptive advantages of communication--the entire vocal tract began to change. Not only did the voice box descend into the throat; the tongue also intruded there, resulting in the ability to make certain sounds, such as a and i. Such a change would have been pointless unless it served to facilitate speech; the recessed tongue was accompanied with a shorter jaw and fewer teeth, rendering the overall structure of the mouth less suitable for eating.

    As a consequence of these adaptations, the hominid throat and orofacial anatomy became capable of performing the most complex motor activities of which mammals are capable. Indeed, the vocal tract had become highly sophisticated in the primates that preceded the hominids. Computer-modeling studies of chimpanzees, for example, suggest that they should be able to produce a significant subset of human speech sounds, including nasalized versions of such words as bit, tip, dad, and pup. In fact, they should be able to produce nasalized versions of all human sounds with the exception of the vowels a, i, and u and such consonants as g and k. Yet, during the past three hundred years, all attempts to teach chimpanzees to produce approximations of words have failed, principally because apes lack the voluntary control necessary to perform the muscular maneuvers that underlie the sounds of human speech. Evidently, such control comes only with the development of Broca's area and related structures in the brain. And therein lies the key to understanding the most important step in the origin of language.

    ALTHOUGH THE BRAINS of vertebrates vary widely in size and shape, their overall organization--hindbrain, midbrain, and forebrain--has not changed during the past half-billion years. In primitive fishes, the principal neural component was the hindbrain, a bulb of nerve fibers, at the top of the spinal cord, that coordinated simple motor reflexes. The much smaller midbrain and forebrain were devoted to processing, respectively, visual and olfactory information.

    With the rise of land-dwelling vertebrates, and the diverse and demanding existence they faced, the center of neural activity gradually shifted to the forebrain, which underwent the greatest amount of growth. Even such early evolutionary arrivals as amphibians and reptiles had a well-developed cerebrum, a layer of tissue, tucked just behind the eyes, which controlled such simple behaviors as mating and foraging for food. By the time mammals evolved, the cerebrum had become the largest part of the brain, making possible the higher-order functions of the nervous system, including learning and memory. In primates, the cerebrum envelopes most of the brain.

    The cerebrums of apes and hominids are split into two hemispheres, each subdivided into four lobes: the frontal lobe, a mass just behind the forehead; the varietal lobe, a wedge behind the frontal lobe; the temporal lobe, beneath the frontal and parietal lobes; and the occipital lobe, at the rear of the brain. Much of the brain's neural activity takes place in the highly convoluted outer layer of the cerebrum, the cerebral cortex.

    The growth of the cerebrum from a simple pea-sized ball of tissue, in ancient fishes, to an elaborate grapefruit-sized globe, in Homo sapiens, was accompanied with--indeed, was largely a response to--increasingly sophisticated sensory and motor systems. Among other things, a highly elaborate cerebral cortex made it possible for such systems to function automatically--that is, without "thinking." Automatization was essential to survival, because it permitted swift responses to external stimuli, allowing an animal to focus on the higher-level aspects of a task. If, while stalking an antelope, a lion had to pay attention to the position of its own limbs, rather than to the movement of its prey, it would have starved. Or if an antelope could not have fled from a predator without first considering where and how to place its hooves, it would have perished. In both instances, numerous motor activities--the coordination of all the muscles and organs involved in running--had become more or less reflexive after a certain period of practice.

    When motor activities become automatic, a direct, reflex-like control circuit forms in the motor cortex, which is located in the frontal lobe, next to the fissure that separates the frontal lobe from the parietal lobe. This has been demonstrated in monkeys that have learned to perform simple manual tasks, such as moving a handle into a designated position. Using small electrodes inserted in the muscles of the forearm, as well as in the motor cortex and the occipital lobes (the vision center) of the animal's brain, researchers have found that, at first, monkeys rely on their eyes to find the handle and to adjust its position: the sensory centers of the brain are heavily involved in processing incoming electrical signals before the motor centers send instructions to the forearm muscles. After training, electrical activity occurs primarily in the brain's motor area, and the response time is cut in half.

    Such automatization occurs whenever behavior is routine. In a sense, the elaboration of the cerebral cortex represented the gradual enhancement of the neural bases underlying all the repet- itive aspects of animal life, including, in mammals, manipulation of the throat and orofacial anatomy. In apes, the motor area that governs these parts of the body is located in the lateral frontal cortex, on the left side of brain, approximately where Broca's area is located in human brains.

    This proximity is no accident: comparative studies have demonstrated that the ape's lateral frontal cortex is the precursor of Broca's area--that, Chomsky notwithstanding, the neural mechanisms associated with language have a counterpart in the brains of our predecessors. Just as the swim bladder, by virtue of its ability to hold air, was preadapted to becoming a lung, and the larynx, because it could control the flow of air, was preadapted to becoming a voice box, so, too, the lateral frontal cortex, by dint of its mechanisms for governing throat and orofacial structures, was preadapted to assuming control of the motor functions that permit articulation. It is not hard to see how this could have occurred.

    Once the neural mechanisms had matured enough to orchestrate the activity of the throat and orofacial region, and once some form of oral communication was in place--the combination of calls and cries that constituted the first vocal signs--environmental pressures for enhancing communication would select for additional neural refinements. Early man, being relatively weak and defenseless, especially on his own, relied entirely on his wits and his partnership with his fellows to survive. The key to both was the development of cognitive power and communicative ability, which, together, yielded language. From that point onward, the evolution of the brain and that of the vocal tract were inseparable.

    Vocal tracts, being largely muscle and cartilage, do not show up in the fossil record, but their character has been registered, indirectly, in the degree of bend in the bases--the basicraniums --of fossil skulls. Comparative studies of contemporary primates show that, because it allows room in the throat for an elevated larynx, a straighter, unflexed basicranium is the hallmark of a nonhuman vocal tract. This is true of the skull of Neanderthal man, who lived in western Europe until about forty thousand years ago, indicating that, despite other anatomical similarities with modern man, that hominid was incapable of human speech. Indeed, it is highly likely that Neanderthal died out because he could not compete with more modern human beings who were better adapted for speech and language, such as Skhul V, who lived fifty thousand years ago in the area that is now Israel. The oldest skulls with a modern configuration, from about one hundred and fifty thousand years ago, were unearthed in Zambia, at a site called Broken Hill. The basicraniums of these fossils are highly flexed, indicating that the creatures who lived at Broken Hill talked with one another. A less equivocal marker for the emergence of Homo sapiens has not been found.

    But there is more to language than making sounds. One could produce at will an extraordinary range of vowels and consonants, at a variety of tempos, and still fail to communicate a single thought or an iota of information. To mean something, sound must follow certain patterns; it must be produced according to a set of rules--a syntax.

    Contrary to Chomsky's theory about the origin of deep structures, this syntax did not appear out of nowhere; it is as deeply rooted in the development of the brain as is the motor control that underlies speech. Indeed, it evolved from, and in conjunction with, the orchestration of the orofacial and vocal tract muscles. Taken together, the rules by which these muscles move constitute a kind of syntax for motor activity.

    Even the simplest automatic acts performed by the mouth and throat involve commands too numerous to list, but a schematic version will illustrate the point. Here is an abbreviated set of commands for a familiar activity: first, open mouth to allow food to enter; second, position food with tongue; third, raise and lower teeth to chew food; fourth, push food to back of throat; fifth, swallow. Note that the commands must occur in a certain sequence. Food cannot be swallowed until it is chewed, nor chewed until it is placed between the teeth. Note, too, that each command could be broken down into subroutines--additional series of commands applicable to particular parts of the mouth or even to individual muscles. Indeed, the set is itself a subroutine that is part of a much more elaborate activity--eating. We do not think about the sequence of motor commands we use to consume food, nor do we treat every meal as a novel situation. Once learned, such rule-governed behavior takes place automatically.

    For every motor subroutine, there exists a corresponding circuit, a specific change in the connections between neurons, in the motor centers of the cortex. For the subroutines that govern the voice box and the vocal tract, the circuits are associated with Broca's area. What's more, the organization of these neural mechanisms--sequences made up of subroutines--is very much like the organization of sentences, in particular, their phrase patterns.

    Consider the sentence "The boy ate his beans." Like the motor commands for eating, it consists of discrete segments arranged in a particular sequence. The first division is between the noun phrase the boy and the verb phrase ate his beans. Each phrase can then be broken down into smaller segments: the boy into an identifying article, The, and a noun, boy, and ate his beans into a verb, ate, and another noun phrase, his beans. These distinctions are not arbitrary; they reflect the internal logic of the sentence--the syntactical rules that enable strings of words (or sounds, in the case of speech) to have meaning.

    Judging by the size and shape of fossil skulls and by the structure of ape brains, the cerebral cortex of early hominids was well equipped to automate all of the body's motor activities. The subroutines underlying this neural coordination, in turn, were suited to assume control of other rule-governed behaviors--namely, the phrase patterns of speech. In other words, the neural circuitry that evolved to orchestrate the muscles of the throat and orofacial anatomy and that was, therefore, preadapted to controlling the production of vowels and consonants, also was preadapted to producing those sounds according to the logic of syntax.

    Evidence for this view comes from studies of the aged, of victims of Parkinson's disease, and of people with severe lesions in Broca's area and the surrounding tissue. In all three groups, those with various deficits in speech production--for example, a slower than normal rate of delivery--also have difficulty comprehending sentences of moderate syntactical complexity ("Because it was raining, the girl played in the house." Or "Mother picked up the baby, who was crying.") This coupling of speech impairment and inability to understand phrase patterns would not occur if both the rules of speech-related motor activity and the logic of syntax were not an outgrowth of the same neural mechanisms--mechanisms that, in a more primitive form, governed the orofacial and vocal tract behavior of early hominids and, before that, of apes and monkeys.

    Establishing this evolutionary lineage in no way diminishes the significance of human speech. Indeed, unlike creationism, the evolutionary perspective acknowledges what is unique about speech: that it arose from things unlike itself--most recently, brain cells and throat muscles and, before that, all the preceding stages of life. If there is anything to marvel at in the emergence of language, it is that, given the right series of circumstances and enough time, a planet made of inanimate rock can, without outside intervention, give birth to creatures who talk and make words.

PHILIP LIEBERMAN, a professor of cognitive and linguistic sciences at Brown University, is the author of THE BIOLOGY AND EVOLUTION OF LANGUAGE. He is working on a book called TALKING THINKING, AND DOING GOOD.