Origin of Language

The original version of this booklet was given as the 1997 lecture in the series ‘David Oppenheimer Memorial Lectures’, organized by Professor Margaret Esiri of the Department of Clinical Neurology, Radcliffe Infirmary, Oxford.

Ü Ü Ü

Works mentioned or quoted (listed under their authors’ names at the end of this booklet) are referenced in my text and notes simply by the relevant page numbers, prefaced by a date of publication when it is necessary to distinguish between publications by the same author.

1. The CONTRIBUTION of
the ENLIGHTENMENT

y interest in the origin of language began when the late Dr. David Oppenheimer and I adapted the posthumous papers of one of our teachers, Ronald Englefield, into a book on the origin and nature of language that was published in London in 1977. Englefield's argument was strong enough to attract the attention of the new York publisher Scribner, who brought out an American edition of the book in that same year. One of my objects in this booklet is to show that his book deserves more attention than it has received.

Englefield's point of departure was the neglected work of eighteenth-century Enlightenment thinkers, best represented by Condillac, who stressed that man's first efforts at communication must have involved signs that are self-explanatory. In language as it exists today, sound and sense are linked purely by convention, and so one has to know the meanings that have been fixed by convention if one is to understand the words. But this state of affairs could not have obtained at the beginning, because conventions could have been agreed only as a result of communication that was already possible without them. The obvious signs that are intelligible without prior convention are gestures, which can mimic actions.

It may be objected that there is such a thing as onomatopoeia – words which imitate natural sounds and so are universally intelligible. But a vocabulary large enough to be of much use cannot be obtained on this basis. We recognize most objects and actions by their visible aspects, not by their sounds. Admittedly, words in many languages may have (or seem to have) some resemblance to the objects they denote; but this has little to do with their origin, for they can often be traced back to earlier forms, where any such resemblance is lacking. Because the burden of language on the memory is very great, words of which the meaning is reinforced by any feature of mnemonic value have a survival advantage in competition with synonyms which lack any such feature, and this is why we may find words of which the sound bears some recognizable relation to the meaning.

Condillac, then, posited what he called a “langage d’action” as the starting point, and as we shall see, it involved not only gestures, but also pantomiming imitations of all manner of actions. But for the moment I wish to stress another point he made, because it has so often been overlooked, namely that the very first signs were not made intentionally as signs, but were the normal reactions to particular situations. He envisaged someone trying to reach some object, or to move something heavy and making movements with his arms in his efforts, and he noted that an observer would feel an impulse to come and help him:

The first one did not say to himself ‘I must gesticulate like this to let him [the observer] know what I want and to get him to help me’; nor the second ‘I see from his gestures that he wants something. I will give it to him’; but each acted in accordance with his immediate need, simply by instinct, for reflection could not yet play any part (pp.261f).

Even today, if someone is seen struggling to retrieve a page of his papers from the floor, a nearby spectator will feel an urge to help by picking it up. Having once observed this helpful reaction, he might on a future occasion do little more than initiate his own movements of retrieval in order to draw attention to his plight and elicit the spectator’s help. Hence Condillac argued that, on the basis of the kind of experiences described in the above quotation, movements and gestures would come to be made purposefully. In this way, movements, which originally had not been made in order to alert other persons, became deliberate signals to them, because a secondary effect of these original movements was first observed and then exploited. At this stage, the movements will have been abbreviated. Stretching out the hand towards the object desired (instead of making a serious attempt to reach it) will have sufficed to elicit the observer’s co-operation, so that full-scale action became reduced to a gesture.

Condillac went on to note that the use of signs led to a development of man’s mental powers, and that this in turn led to an improvement in the signs (pp.263ff). Language and reason, then, developed together, just as we say today that every function of an organism develops progressively with the organ on which it depends.

It has often been argued that language cannot have originated as an invention, since no one could have envisaged the advantages which would accrue from its use before it existed. But this impasse can be avoided if, with Condillac, we allow that deliberate communication was not the starting point, but resulted in time from actions which had served some non-communicative purpose. As a generalization one may say that all purposive behaviour that is learnt must be based either on behaviour that was at first spontaneous, or which served some other previously learnt purpose. If man has come to learn the advantage of working together with his fellows, of co-operating with them, he must initially – without purposeful intent – have got into a situation where someone helped him in what he was doing, and where he could in consequence become aware of the advantages of such help.

Condillac’s “langage d’action” included all possible means by which one person may suggest to another an idea of a thing or of an action. For instance, to make someone think of a boat, I could imitate the movements of rowing or hoisting sail or grasping the tiller, and also behave as though I were being rocked about in a boat. To suggest the idea ‘man’ or ‘fish’ I could draw an outline in the air with a finger or make a rough drawing or model. One can represent actions such as chopping or digging most intelligibly if one is holding some object that vaguely resembles the tool. On this basis we may expect not only pantomime, but also rough models and constructions out of all kinds of handy materials. The point is that every possible method would have been resorted to. Pantomime was very expressive, but unsuitable for rapid communication, as it takes too long to perform. Drawing and modelling were also inconvenient, but for a long time every possible means must have been employed, each supplementing the other. Where a simple gesture did not suffice, these other means will have been added to it, until this multiplicity of signs finally conveyed the desired idea.

This view of the origin of language today commands sizable assent. Westcott (p.128) noted “a growing tendency” in favour of it, and Kendon (1991, p.215) agrees that “language began as a sort of pantomime”. However, many scholars, from the late eighteenth century until and including Chomsky and Pinker today, have ignored gesture and made speech the starting point. One reason for this is that the notion of a gesture language was formed from study of the languages of the deaf and dumb, which are highly conventionalized and quite meaningless to the ordinary spectator. Gestures of this kind do not supply any simpler starting point than speech, so it was felt that one may as well begin with this latter, particularly as in either case it has to be explained why sounds came to be made at all. In the case of a gestural origin, re-adaptation is involved to make natural gestures into a gesture-language. Similarly man must have made sounds in the first place before he could think of adapting them so that they became means of communication. Jespersen, who discussed the origin of language without even mentioning gestures, surmised that man will originally have made sounds as a form of play, as activity which “had no other purpose than that of exercising the muscles of the mouth and throat and of amusing oneself and others by the production of pleasant or possibly only strange sounds” (p.437). This would of course not exclude a gestural origin for language. It could still be argued that the fact that noises could be used on a large scale as signs was a discovery, and could not have been made unless the two components – use of signs and vocal acts – already existed. The use of signs will have been already there in the form of gestures and visible representations, and the vocal acts were used in play.

It must, however, be admitted that recent work has established what earlier investigators could not readily have envisaged, namely that some non-human primates employ a limited number of arbitrary sounds to convey meaning. Robin Dunbar instances the work of Dorothy Cheney and Robert Seyfarth, who have analysed on a sound spectrograph the grunts (indistinguishable to the human ear) made by vervet monkeys, and have shown that calls given when approaching an animal dominant in the social group differ from those used when approaching a subordinate, and again from those emitted when sighting another group in the distance, or when about to move into open country from the safety of the trees. The animals hearing these different calls respond in ways appropriate to each of them. These monkeys also use different sounds to warn against different types of predators. The cry which signals danger from an eagle differs from the warning against a leopard, and again from that which indicates the presence of a snake. The behaviour which these cries elicit is again correspondingly different. When warned against eagles the troupe descends from the trees. The warning against leopards prompts the very opposite reaction, while the snake call makes the animals stand on their hind legs and investigate the ground around them. Dunbar infers from all this that “there is no need for a gestural phase” in the origin of language. “It can all be done by voice” (pp.140f). Much of it can indeed, as among ourselves today; but whether this evidence of very limited non-human use of the voice allows the inference that it always was done in this way is questionable.

2. TRANSITION from
GESTURE to SPEECH

erhaps the most compelling reason for denying a gestural origin to language is the difficulty of explaining any transition from something like Condillac’s “langage d’action” of self-explanatory signs to the sound languages of today. The gesture can bear a visible relationship to the action or object for which it stands. If I want to convey the idea that yonder hill is very high, I can point to it and express the idea ‘high’ by jumping up high myself. But a sound can have such a close relation to its meaning only in an extremely small number of cases, and the connection between the two must nearly always be arbitrary and fixed by convention.

Condillac’s view as to how sounds came into language was that cries of various kinds were among the natural reactions of the first period and came to be used as signs in the same way as did gestures. But in becoming converted to signs they gradually lost their natural emphasis and became controlled or “articulate” sounds. A groan was originally a spontaneous expression of emotion; but when it was deliberately made in order to summon help, it became a simulated groan. More sounds, he says, could be added and made intelligible by being accompanied with self-explanatory gestures; i.e. arbitrary sounds would be accompanied with signs of known meaning. And, most important of all, these sounds would be spoken in situations where both parties – speaker and listener – were engaged on some specific task and were perfectly clear about what they were trying to achieve. One can see the force of this important point if one thinks of a situation in which someone might make the sign for ‘water’. He might mean that he wants a drink or a swim, or that a house is flooded or that the crops need watering. What he actually means will depend on the situation confronting both partners to the conversation – a situation of which they are both aware and where the one knows what can be expected of the other if they are to exploit or remedy it. If each knows that they are both worried about a lost sheep, making the sign for ‘sheep’ and following it with a joyful expression will convey the information that the animal has been found. In a different situation, the same sign could mean something quite different. Condillac put the matter as follows:

In order to understand how men agreed together about the meaning of the first words which were introduced, it is sufficient to observe that they would be uttered in circumstances where everybody would be led to refer them to the same perceptions. In this way they would fix their meaning more precisely according as the circumstances, being repeated, accustomed the mind more and more to attach the same ideas to the same signs. The gesture-language would help to remove ambiguities and uncertainties which, to begin with, would be frequent (p.362).

In like vein, Gordon Hewes has recently observed that “primordial language, gestural or spoken, would have been far more dependent than formal modern speech or written language on the immediate social and environmental context for its decoding” (1976, p.497). Hence it was not necessary that someone or some committee invented words and promulgated an official list. New words would be introduced, or at any rate adopted, only if the circumstances of their first use were sufficiently public. Many would have but a brief existence in the private intercourse of two companions, or perhaps of a family; but some, because they were specially useful or specially easy to remember, would have their use extended beyond the original circle.

In the nineteenth century this kind of explanation was generally rejected as crude and psychologically inconceivable. Max Müller, for instance, complained that “no one has yet explained how, without language, a discussion, however imperfect, on the merits of each word, such as must needs have preceded a mutual agreement, could have been carried on” (p.34). Rousseau’s famous paradox had been formulated with the same objection in mind: “Words would seem to have been necessary to establish the use of words” (“La parole parait avoir été fort nécessaire pour établir l’usage de la parole”, pp.148f); that is to say, the invention of so elaborate an instrument as language, and its adoption by all members of the community, require an intellectual capacity and a degree of co-operation that could have been reached only in a community already in possession of language.

The answer to this paradox is that a spoken language of words never has been the one and only way of conveying ideas, and still is not the only means today. A threatening gesture, a disapproving look, and other self-explanatory signs, such as drawings or models can convey a great deal – just how much has been well brought out by Macdonald Critchley in his Silent Language. Englefield too shows that a great deal can be conveyed without words. In a series of experiments which he carried out on schoolchildren, he found that even simple abstract ideas could be expressed with a combination of gestures. The following are among his examples:

Hot: boy puts hand near fire and pretends to find it hot, snaps fingers as if burnt, wipes forehead, eases his collar.

Love: sits beside another boy, kisses him, puts imaginary ring on his finger.

Empty: takes off shoe and looks inside; opens empty tin and shows it (pp.175ff).

This is not to say, as did Sayce, that gesture language is “instinctive” (p.93), but rather that man is physically and mentally so constituted as to be able to communicate without convention.

Condillac still leaves us far from understanding how language came to consist almost entirely of sounds. He and other eighteenth-century writers saw clearly enough that, if only sounds can be understood, they have great advantages over all other forms of communication, in that they are easy to make, and to make rapidly, easy to combine, can be heard when gestures would be invisible – at a distance (as when hunting), in mist or darkness (as in the forest), and through obstacles which impede the view. They also radiate in all directions, unlike gestures, which require that the performer has his audience facing him. They employ an organ required for alternative purposes only at meal times, whereas the hands have innumerable other functions to perform. And – unlike gestures – they actually attract attention.

Some scholars write as though these advantages in themselves suffice to explain why arbitrary sounds took over from self-explanatory gestures, as if the transition posed no problem. Thus Corballis lists some of the advantages of sounds, and comments: “These factors may have therefore created a selective pressure that led, eventually, to gestures being superseded by vocal communication” (p.158). Others who allow that language began as a sort of pantomime offer mechanisms for explaining the transition to the vocal mode that are little short of ridiculous. Sir Richard Paget, for instance, held that hand gestures themselves prompt the gesticulators to make sounds. In the late nineteenth century, Lazarus Geiger and Wilhelm Wundt supposed that gestures were, for some reason, imitated by movements of the lips, tongue and mouth – an idea recently taken up by Gordon Hewes, who, however realizes that it is in fact “very weak in accounting for the transformation of gestural language into speech” (1973, pp.10, 19f).

Robin Dunbar is so impressed by the obvious disadvantages of gestural, compared with vocal communication that he finds it “difficult to see how a gestural language of any complexity could get going” (p.135). This objection to a gestural origin has been answered by Englefield, and, more recently, by Kendon. Because the self-explanatory signs of which language will have originally consisted would gradually be abridged through familiarity, they would thereby lose some or all of their original intelligibility except to those who had been making constant use of them and who had thus experienced the gradual process of abridgement. To outsiders they would appear as signs whose meaning depended on conventions. As already noted, the language of the deaf and dumb, as practised today, is conventionalized in this sense, and cannot be understood without knowledge of the relevant conventions. But this abridgement, which cut out the need for laborious pantomime, led not only to loss of universal intelligibility but also to a great increase in efficiency in communication between the initiated. Kendon (1993, pp.57f) illustrates the process with a modern example (from a 1981 paper by Caroline Scroggs) concerning the behaviour of a deaf boy who had no training in sign language: “When ‘motorcycle’ was first introduced into his discourse he gave an elaborate pantomime of mounting the cycle, starting it and revving it up, using hand motions to indicate the twisting of the throttle on the handlebar. In subsequent references to the motorcycle, however, just this hand motion was used. Thus a hand action derived from a pantomime of twisting the throttle came to serve as a symbol for the concept of ‘motorcycle’.” In due course, as a result of economy of action, this single element from the original elaborate pantomime “becomes simplified to such a degree that its image-like or iconic character is no longer apparent. It turns into an arbitrary form”, and, in ceasing to be a ‘picture’ of something, it “becomes available for recombination with other forms and so may come to participate in compound signs or sentences.”

That this can be presumed to be what happened in the beginnings of language can to some extent be confirmed by what is known of the origin of writing once it had become available. Picture-writing became conventionalized and gained in efficiency, while it ceased to be understood by all except the scribes. Englefield added to this argument the significant point that, once the originally self-explanatory signs used for communication had become so abridged and disguised as to be unintelligible to outsiders, the communicating parties will, sooner or later, have become aware that the signs they were using were conventional, that communication does not necessarily require self-explanatory signs, that – given the will to communicate and given that this will is understood by the two parties to the communication – no special relationship between sign and meaning is necessary, and that quite arbitrary conventionalized signs can be understood by those who have learned to use them.

Let me put this important point as follows: A sign cannot be associated with a thing or event or action unless the thing is first indicated in some way. Every act of naming must therefore be preceded by some indication or description of the thing to be named. This part of the process may consist merely in pointing to or presenting the object, or executing the action, or imitating the event. But if this part of the process has been successfully achieved, if the other party has been brought to understand what thing or event is being referred to, then the assignment of a sign to denote it may be perfectly arbitrary. In the case of proper names, it is almost necessarily arbitrary. Dogs in general may be known as bow-wows, but it would hardly be possible that names should be given on a similar onomatopoeic principle to every member of a community. The communicating parties would then in time come to think that, if the signs are to be conventional, they may as well be sounds, because, as we saw, sounds have such obvious advantages over other forms. On this basis, names could be consciously invented, for the will to communicate and the realization of the means were there. Of course, the need to remember a large number of names leads to the practice of choosing, if possible, a name that the object itself suggests, or which suggests the object. Any kind of association will do, and there is no need for any resemblance. Many objects are named from the place from which they come, or from the person who invented or discovered them, or from the use to which they are put.

How difficult scholars find it to account for any transition from gesture to speech is illustrated by Merlin Donald’s book. He favours the view that gesture preceded speech (“mimesis was ahead of language”, p.199) and is aware that, in time, gestures would be standardized and so would develop into “arbitrary symbols” (p.220), with the result that “fairly elaborate systems of gestural symbolism” may well have preceded speech (p.225). He thinks, however, that some additional mental development was required to effect the transition from the one to the other: “Mimetic representation laid the groundwork for language and symbolic thought, but lacked some critical element; the mental modelling apparatus was still incomplete” (p.233). In fact, however, Englefield’s suggestion seems to supply all that was needed to effect the transition – namely an awareness that the gesture language had come to consist of arbitrary symbols, and a realization that, if the symbols can be arbitrary, they may with advantage be sounds.

This assignment of arbitrary sounds as names will have been a slow and protracted business. The advantages of an exclusively oral language would not have become obvious until a considerable vocabulary had been accumulated, and achieving this will have taxed unpractised memories. Moreover, if all were to use the same words with the same meanings, they must have been in fairly close contact; so the whole process could have begun only in a small and compact community. Here again, Englefield has a helpful suggestion, namely that the need of small, tightly-knit groups (perhaps within a larger community) to develop a private or secret means of communication, with a vocabulary restricted to the special objects of their interest, could have given a significant start to vocabulary-building: “Under such conditions a few intelligent individuals could experiment and build up more or less extensive vocabularies for their private use. By initiating fresh members into their group they could gradually enlarge the usefulness of the new language.” He adds: “There is plenty of evidence of the use of secret languages among primitive peoples….[Such] cases show that one advantage of oral language has often been exploited, namely that by means of an easy substitution of terms one can form a system of communication which is intelligible only to those for whom it is intended. At a later time writing was found to offer the same advantages [with codes and cryptograms]….Gesture [on the other hand] does not so readily lend itself to disguise….And it seems a plausible suggestion that, at a time when communication depended still in the main on the language of movement and visible signs, the special advantages of using conventional sounds as the names of certain things would be exploited” (pp.87-88). In sum, conventional sounds could have been used for the sake of secrecy in restricted groups, “not as a complete language for all purposes, but as a private supplement to the normal language of gesture… But once such a language had come into existence it might be learnt by an ever increasing number of young people, and in time could be degraded into the common language; a new one being invented, if required, for secret communication” (p.94).

3. The ROLE of DELIBERATE
INVENTION in MAN
and APE

hat Englefield stresses most is that, while a few names may have resulted from the abridgement of natural or imitative sounds, this could not have produced a vocabulary of any size, and certainly not the enormous number of words found in every known language, so that at some point deliberate invention must have entered into the process. We see here, then, as in other forms of human activity, the increasing influence of reflection: acts which were first performed because experience showed their effectiveness became the subject of scrutiny as to what made them effective, and this led to conscious adaptations and improvements.

Many nineteenth-century anthropologists, misled by the great differences between their own beliefs and the irrational and magical beliefs of “savages”, supposed that early man must have been incapable of such rational reflection – an assessment which overlooks a distinction between beliefs of two different kinds. There are, first, our beliefs about our immediate environment; when these are very incomplete or erroneous, our behaviour is likely to be ill-adapted to our needs, so that we expose ourselves to some immediate unpleasantness. But in this way, attention is called to our mistake, and we may be led to rectify it. If, for instance, we act on the belief that ether is a good fire extinguisher, we shall be in for a rude shock, and if we survive the experience, the belief will not survive with us. On the other hand, any ideas we have formed about the nature of the universe, or about the distant future or past, are unlikely to lead to any noticeably inappropriate reactions on our part, and so we may well persist in erroneous beliefs of these kinds all our lives without ever experiencing the smallest surprise or disappointment. This difference is important, for it explains why even the most primitive peoples sometimes appear to have made considerable progress in the practical arts, while continuing to hold quite groundless beliefs in matters that do not lend themselves to experimental control.

Since chimpanzees can rationally adapt means to ends in solving problems which interest them, there is surely no difficulty in supposing the same to have been true of our human ancestors. De Waal has pointed to "increasing evidence....for cognitive continuity between humans and great apes" (1994, p.264). In fact very substantial evidence for this was marshalled by Wolfgang Köhler more than seventy years ago. The significance of his work is today still widely disputed, underestimated or ignored (as when the fashioning and use of tools by chimpanzees is hailed as a discovery from recent study of these animals in the wild, although Köhler long ago documented it in their behaviour in captivity). Let us review his masterly observations.

When Köhler put food out of reach outside the bars of the cage of his chimpanzees, they realized that they could drag it in if only they had a pole, and so went in search of one. That thinking was involved was clear from their behaviour. They did not try out various possibilities at random on the ground, in the presence of the initial situation, until they happened to hit on some action that would help them. The focus of the situation, the food, was ignored, and the animal turned away from it under the guidance of a new goal, the pole. We assume therefore that he was able to supply in his imagination this feature that was missing in the initial situation, and could see himself, in his mind's eye, converting this initial situation into a favourable one with the aid of this missing feature. In one experiment the animal, during his search, came across an old box in which he recognized a potential pole, and accordingly broke off a slat, with which he returned to the bars of the cage and drew in the food. Other substitutes for poles which these animals sought out included branches from trees, bars broken from a shoe-scraper and screwed up bundles of straw. It is as if the animal had formed for himself an abstract definition of the tool required - not of course in any kind of language, but in terms of the task to be accomplished. He knows, from comparison of visual aspects, that it must be of a certain length; and from his muscular and tactile experience he knows that it must have a certain weight and rigidity, so that a rope or blanket will not do. The animal of course does not always have in mind all the relevant features of what he needs, and so sometimes chooses inappropriate tools - returning, for instance, in triumph with a stick only to find that it is not quite long enough.

There are still scholars who treat Köhler’s findings with considerable reserve. Stephen Walker will allow only that he showed that the chimpanzees can remember spatial features and can manipulate objects (p.348). But surely Köhler showed, as we shall see, that they can manipulate not merely objects but also ideas of objects and can think, for instance, not only of stick and fruit, but also of bringing the one into contact with the other. Against this, Graham Davey objects that “previous experience with some of the objects in a problem-solving situation aided solution of the problem" (p.305). But of course. What Köhler called “insightful” behaviour - that is, first thinking out how a present situation can he converted into a more desirable one, and then putting this plan into action - does not, as Davey seems to suppose, imply a mind devoid of previous experience. Even we humans do not imagine processes or events which we have not, at least in their elements, experienced. Köhler’s observations convinced him that experience which these animals had gained elsewhere, whether in play or business, was deliberately exploited.^{note 1}

Of great significance is the fact that the bright idea for solving the problem could sometimes come to the animal when neither the problem situation nor the means of solving it were still visible to him. Thus on one occasion fruit had been attached to the roof of the cage, and could easily be reached by standing on a box, but the only available box was out of sight in an adjacent corridor. The ape looked round for something that could serve him as a tool, and lighted upon a long bolt fastened to a door. While trying to detach it, he hung upon the door in such a way that he could not see the fruit; and the box in the corridor was completely out of sight round a corner. This is Köhler’s account of what then happened:

Abruptly and without external occasion [the ape] stops working at the bolt, remains for an instant motionless, then jumps to the ground, gallops along the corridor and returns immediately with the box (p.38).

Clearly, at the moment when the animal stopped motionless, although both box and fruit were invisible to him, they were represented in his brain, so that, in his mind's eye, he saw himself moving the box to under the fruit. In other words, he performed, in imagination, a manipulation which he then proceeded to carry out in fact. This power to form ideas and to manipulate them internally is the very essence of thinking.

Because Köhler's findings are still imperfectly appreciated, Derek Bickerton can declare that “only humans can work on problems that do not immediately confront them”; and he believes that it is language that makes it possible for man thus to attend to problems not immediately before his eyes.^note
2Now of course the inventions of the ape are closely related to his immediate situation. But we have just seen that a chimpanzee who had previously seen the fruit in one place and the box in another could imagine the box moved to under the fruit at a moment when neither was visible to him. The insight which involves imagining the initial problem situation, as well as the desired one and the means of attaining it, is different only in degree from that which can draw on perception for the initial situation. And language is not necessarily involved in either case.