The generally accepted view of those who study language professionally is that language is an arbitrary, cultural construct; language, on this view, is learnt by listening to speakers of the language of the particular community into which an infant is born; the words used in the language as well as the particular grammar or syntax of the language have developed historically as a social product and been handed down by tradition.

At first sight, it might seem a highly academic question whether or not language is arbitrary, of interest only to linguisticians, etymologists and so on. But to say that language is arbitrary and a purely cultural product is to assert that there is no basis for relating language to other aspects of human biology, to evolution as shown in the development of brain structure and the physiological differences between men and animals. It would be a strange result if the manifestation of the major and many would say absolutely crucial human ability, the ability to speak and understand language, should on this view turn out to be something which cannot be explained and for which, in principle, no explanation can even be attempted.

'Arbitrary' means chance, unmotivated, without purpose - and those who view languages as wholly arbitrary structures are saying that they are the product of chance, guided by no objective, that the availability of words and the structures of any language are completely purposeless. Yet at the same time all would recognise that language is the fundamental instrument for human communities, the essential medium of communication, the precise and powerful tool of thought, the basis for scientific and technological progress. If such a miraculous instrument is arbitrary in origin, function and structure, then one can only fall back on a belief in myth to explain it. The ancient Egyptians believed that the word was given to mankind by the god Ptah. They at least recognised the real problem, that language must have had some origin.

One wonders why academic students of language professionally have been so attached to what at first sight would seem a disastrous foundation for any science, the belief that the form and underlying structure of the subject of study is arbitrary, irrational, chance. Perhaps a cynical view, and a partial one, would be that one important effect of arbitrariness as the starting assumption for language study is to delimit an exclusive field of research for the linguists, to post a large 'No Entry' sign to the domain of linguistics and to tell others, psychologists, physiologists, neurologists, that they will be wasting their time if they try to apply their theories and technical procedures to language. Fortunately some of these other scientists have not been deterred, for example, Karl Lashley, Eric Lenneberg, Roger Brown, and in another quite new discipline, that of artificial intelligence and computational linguistics, pioneering work is being done, untrammelled by the traditional restraints of linguistics.

The debate whether language is natural or artificial, purposeful or arbitrary, an evolutionary or a conventional product, is an extremely ancient one, indeed perhaps the most ancient dispute of philosophers centuries before the study of language became a subject for specialists. Nearly 2,400 years ago, Plato in his dialogue Cratylus had Socrates discuss the contention that a word applied to an object (the object's name) was not just whatever people agreed to call the thing but resulted from a kind of 'inherent correctness'(1), which linked the nature of the object and the speech-sounds used to name it. Centuries later, Lucretius in De Rerum Natura dealt with language and dismissed as ridiculous the idea that any individual could have been in a position to give names to things in such a way as to persuade others to accept and use the names he prescribed(2). Much later again, Wilhelm von Humboldt(3), one of the most profound and stimulating writers on language in the modern era, contended that there was a natural basis for words, that language naturally selects for particular objects speech sounds which partly independently and partly in comparison with others produce an impression on the ear similar to that which the object makes on the mind. This natural process had, in his view, exercised a prevailing, and perhaps even an exclusive, influence on primitive word formation.

However, the conventional wisdom for the last 100 years or so has been very different. The principle of the arbitrariness of language has ruled virtually without challenge. The foundation of the modern science of linguistics has been Saussure's proclamation of 'the arbitrariness of the sign' by which has been meant that one can, one should, look for no relation of any kind between the sound-structure of the word and its meaning. The idea that the word is arbitrary goes back before Saussure (Locke was perhaps the most powerful and influential adherent of the view) but Saussure(4), as the father of modern linguistics, gave his overwhelming authority to arbitrariness as the foundation assumption for the new science. Whilst some of the many linguists who follow him are prepared to recognise that in some minor respects elements of the lexicon may not be totally arbitrary, for example, admitting that there are onomatopoeic words like 'cuckoo' (though some would say that even words like these are conventionalised), linguisticians generally are no more prepared to consider that language is a natural product than pre-Darwinian zoologists could accept the natural origin of the different species. Not only do many assume without question that words are arbitrary forms but they would also argue that language as a whole is a construct, a cultural tool, and that the arbitrariness extends to every feature of the grammar and syntax of particular languages.

Because the issue is so fundamental, not only for linguists in general but also for the thesis presented in this book, it may serve to bring out very sharply the current views by quoting directly from a number of writers, demonstrating the unanimity of the chorus, not only of linguists but also of philosophers and others who have touched on the subject:

Saussure(5): Language is a convention and the nature of the sign that is agreed upon does not matter ... Because the sign is arbitrary, it follows no law other than that of tradition, and because it is based upon tradition, it is arbitrary. 'Arbitrary' ... should not imply that the choice of the signifier is left entirely to the speaker ... I mean that it is unmotivated i.e. arbitrary in that it actually has no natural connection with the signified. Only differences that make it possible to distinguish this word from all others ... carry significance .. since one vocal image is no better suited than the next for what it is commissioned to express. ... 'Arbitrary' and 'differential' are two correlative qualities ... Language is a system of interdependent terms in which the value of each term results solely from the simultaneous presence of the others. ... The arbitrary nature of the sign explains in turn why the social fact alone can create a linguistic system - by himself the individual is incapable of fixing a single value. ... A particular language-state is always the product of historical forces and these forces explain why the sign is unchangeable i.e. why it resists any arbitrary substitution. ... The community itself cannot control as much as a single word; it is bound to the existing language. ... No longer can language be identified with a contract pure and simple.

Hockett(6): Arbitrariness The relation between a meaningful element in language and its denotation is independent of any physical and geometrical resemblance between the two ... or, as we say, the semantic relation is arbitrary rather than iconic.

Sapir(7): Onomatopoeic words are just as truly creations of the human mind, flights of the human fancy, as anything else in language.

Bloomfield(8): What we call 'horse', the Germans call 'pferd', the Frenchman 'cheval', the Cree Indian 'misatim' and so on; one set of sounds is as unreasonable as any other.

Firth(9): Words are acquired habits. With the doubtful exception of certain sibilant consonants, there would appear to be no inherent phonaesthetic value in any speech sounds. It is all a matter of habit.

Yuen Ren Chao(10): Language is a conventional system of habitual vocal behaviour. Before the establishment of a convention, any word could mean anything.

Englefield(11): The fact that languages are arbitrary is sufficient evidence that they were invented. In any language there are conventional ways of combining words to express the relations between ideas. There is no systematic correspondence between the forms of language and its meanings.

Hormann(12): What meaning is conditioned to which sign is basically quite arbitrary and therefore there is an element of randomness or absence of logical necessity in the relationship of sign and object signified.

Miller & Johnson-Laird(13): The absence of direct, tangible connections between physical objects and the acoustic patterns used as names for them is a foundation assumption for studies of linguistic meaning.

Tax(14): Cultural behavior has a quality of arbitrariness. It does not flow through the genes and is therefore not anchored in the individual. This is seen most clearly in the arbitrariness of the symbols of language.

Gregory(15): Just because words differ between languages, and because languages are so recent and change so rapidly, it is quite clear that our knowledge of the names of things cannot be innate. It cannot be built into the nervous system. Words and names cannot be inherited.

Wittgenstein(16): I want you to remember that words have these meanings which we have given them and we give them meanings by explanations. A word has the meaning someone has given it.

The consensus that emerges from these extracts is very apparent. In summary, according to the authors:

A word has the meaning someone has given it. Words are given meanings by explanations.

Language is arbitrary, conventional and traditional. Words have meaning only as parts of a system, with each word deriving its meaning solely from its difference from the other words in the system.

Differences in the words used by different languages for the same things show the unreasonableness of all the words chosen. Words are no more than 'flights of fancy'.

Words are acquired habits. Any word could mean any thing. There is no logical necessity in the relation of sign and signified.

There is no geometrical or physical resemblance between word and meaning. Words are arbitrary rather than iconic.

Words cannot be innate or built into the nervous system because languages are a recent evolutionary development and differ so much from each other.

When so many diverse authorities agree so forcefully, it may seem rash to challenge what they say, but fundamental assumptions of any science ought some time to be challenged. Progress in the past in many sciences has come from challenging the unchallenged. One line of attack is to note that not all of those whose opinions are recorded have expressed themselves in exactly the same way. There are inconsistencies and incoherences in the different accounts. Whilst virtually all those who have in more recent times discussed the relation of word and meaning, of language and reality, have described the sounds of words as arbitrary, there have been variations in how they describe what they mean by 'arbitrary' and they have used other terms, along with 'arbitrary', to present their convictions.

All would agree that 'arbitrary' must exclude 'natural'. Hockett contrasts 'arbitrary' and 'iconic' as meaning physically or geometrically resembling the referent. Others contrast 'arbitrary' and 'innate' (Gregory rules out 'innate' though he does not expressly say that words are arbitrary). The ancient distinction, going back to classical times, was between language phusei (that is, the product of physical nature> and language thesei (that is, determined by decree, prescription or convention).

Different writers give different accounts of what 'arbitrary' means. Saussure and others equate 'arbitrary and 'traditional', 'arbitrary' and 'conventional', 'arbitrary' and 'social' or 'habitual'. Hormann puts 'arbitrary' as equivalent to random, lacking logical necessity. For Bloomfield, 'arbitrary' means 'unreasonable' and for Sapir 'fanciful'. Englefield treats 'arbitrary' and 'invented' as equivalent i.e. artificial, a deliberate product of human construction. Tax equates 'arbitrary' and 'cultural'

Whether the accounts given by the various writers are coherent and they are right in treating 'arbitrary' and the other terms used as synonymous is doubtful. In common usage, 'conventional' 'social' 'traditional' 'cultural 'invented' 'artificial' 'random' 'unreasonable' 'illogical' 'habitual' 'historically-derived' carry quite different connotations and it is not possible that all the implications of using them as alternative descriptions of the character of language can be correct. For example, it is by no means the case that 'conventional' and 'arbitrary' can be used interchangeably any more than 'arbitrary' and 'cultural' or arbitrary' and 'social'. Nor can one equate these words with 'unreasonable', 'lacking logical necessity', 'artificial' or 'invented' or 'random'. A system can be artificial without being random or unreasonable; see, for example, the system of the Morse code. A system can be social but constructed in accord with reason, as many important institutions have been. Saussure himself points out elsewhere that most important social institutions have at some point a natural base, so that to say that language is a social construct is not automatically to say that it is in some sense an arbitrary construct.

One suspects that most of those who discuss the nature of language and categorise it as arbitrary are confusing the current appearance of any language, as a collection of words which mostly have no obvious relation to their meaning, with the separate question of the historical origin of the individual words which go to form the present-day collection. It is not enough for Saussure to assert that language is arbitrary or non-natural because it is traditional since this leaves untackled the important question of the origin of the traditional link of word and meaning. It is not possible, without circularity, to argue that in its first origin the particular word was heard in use in speech - the tradition must have started somewhere, somehow. At one point, Saussure asks "How would a speaker take it upon himself to associate the idea with a word-image if he had not first come across the association in an act of speaking?"(17), but this is a puzzle which he has created for himself, to explain the historical first use of a particular word (we know the answer for some modern neologisms such as the origin of 'gas' 'paraffin' 'vaseline' , as discussed by Potter(18)).

Those who speak of the cultural, social, customary origin of words and language might be reminded of the very relevant comments of Konrad Lorenz: "The undeniable fact that cultures are highly complex intellectual systems, resting on a basis of symbols expressive of cultural values, causes us to forget, given as we are to thinking in terms of opposites, that they are natural structures, which have evolved along natural lines"(19). Somewhat similarly, Hewes, after quoting the remarks of Tax extracted above on the arbitrary quality of cultural behaviour and language 'because it does not flow through the genes', points out that if culture and biology have always been separate domains, we have no way to explain how man's language capacity has been perfected.(20)

It seems absolutely clear that what is 'traditional' 'social' or 'conventional' can still be wholly or partly natural, or at the minimum natural in origin. After all, there are obvious examples of what one would call 'conventions', styles of clothes, cookery recipes, matrimonial systems, methods of composing music, but for these one does not need to look very far to find natural bases or natural constraints on the forms which they take. Equally, one can easily identify obviously natural forms of behaviour upon which conventional, traditional or social elements have been superimposed. Eating is natural but there are conventions about the manner of eating; one chooses to eat in this way rather than that, even though in a sense both ways are equally natural. Even in the case of walking, which Saussure refers to as a clearly natural form of behaviour, there are different manners of walking, including some highly conventionalised ones such as marching, goose-stepping and so on. With the human being, the natural and the social are inextricably mingled in many forms of behaviour - and the essence of the convention or tradition may exactly be the adoption in a community of one particular form of behaviour out of several equally natural possible forms of behaviour. Shaking hands; and rubbing noses are equally natural as forms of contact and greeting, but our community prefers, by tradition, a hand-shake where the New Guineans may prefer a nose-rub.

Set against the range of words, often with imprecise uses, referring to the 'arbitrary' character of language are the various words used to express the opposite view. These include 'natural' 'biological' 'innate' 'genetic' 'physical' 'nativistic' 'instinctive' 'evolutionary' 'organismic' 'motivated' 'species-specific' and so on. It is fair to say that many of these are used with no more precision than the corresponding set of words grouped with 'arbitrary'. It is often difficult to decide what in a particular context is meant by 'natural'; all linguists tend to describe particular languages as 'natural' languages but this is only to distinguish them from invented universal languages (Esperanto, Interlingua and so on) or from the ideal or formal languages found in logic, philosophy or mathematics. Clearly this use of 'natural' implies no departure from the general view that all languages are arbitrary. So 'natural' tends to acquire its specific meaning in any context from the express or implied contrast with some opposed word: natural and artificial, natural and learned, natural and invented, natural and cultural and so on.

Given this, it is clearly not enough to assert that language or any aspect of language is 'natural' without specifying precisely the manner in which language is natural. 'Innate' is a more specific description than natural in so far as it implies that one can relate language in some way to the genetic determinants of human bodily structure and functioning but there is an area of uncertainty about the meaning of 'innate: something may be specified potentially by innate factors but only realised by interaction between the innate factors and the environment. For example, if one considers the line of research into animal vision by Hubel and Wiesel(21) and others, should one categorise a kitten's ability to perceive a particular vertical, horizontal or slanted pattern of lines as innate or the learned product of environmental experience? Experiments have shown that at a certain critical period, a kitten's visual apparatus is malleable and the permanent characteristics of its adult vision depend upon the structuring of the environment to which it is exposed during the critical period. A kitten brought up in an environment without vertical lines will thereafter be unable to perceive them. Is this an innate effect or a learned effect?

Other terms - biological, physiological, evolutionary, species-specific - are not necessarily much clearer in their implications than is the word 'innate'. To say that some aspect of behaviour is a biologically-based or biologically-determined aspect of a particular species (as language is an aspect of the human species) really tells us very little more than if we say the behaviour aspect is natural, unless we are able to go on to explain in preciser detail the nature of the biological basis in established facts of physiology and neurology.

In considering the well-foundedness of the view of language as arbitrary or natural, there is another kind of difficulty. Apart from imprecision in the use of the word 'arbitrary' (as well as in the use of the word 'natural') one finds on closer examination that there is a dismaying lack of clarity in the views of linguists as regards the extent to which the thesis of the arbitrariness of language goes. Language is not only words and collections of words; it is also the speech-sounds from which words are formed and the combinations and strings of words which go to form phrases and sentences, the complete stream of spoken language. Are phonetics and syntax as arbitrary, in the view of these linguists, as the lexicon of each and every language? Are the types of sound used for forming words and ranges of words selected for use by different languages equally arbitrary? In the area of grammar (morphology and syntax), are the differing word-orders selected by different languages equally arbitrary, equally part of Sapir's 'flights of human fancy'? Are inflectional systems (declensions and conjugations), the existence of concord between nouns, adjectives and verbs, vowel-harmony, agglutinative structures, the product of random forces, the result of deliberate invention or a purposeless weaving of complexity?

Most linguists who have readily, often with very little examination, come to accept the principle that language is arbitrary, because it must be so given the divergences between languages, have not been at all precise about where arbitrariness stops and order in language begins. Even the arch-priest, Saussure, is not altogether consistent. So he says: "a language constitutes a system. In this one respect ... a language is not completely arbitrary but is ruled to some extent by logic. The system is a complex mechanism that can be grasped only through reflection.. Some signs are absolutely arbitrary; in others we note not its complete absence but the presence of degrees of arbitrariness: the sign may be relatively motivated, for example, 'vingt' is relatively less motivated than 'dix-neuf' (which by its form refers to other forms) ... but even in the most favourable cases motivation is never absolute ... The mind contrives to introduce a principle of order and regularity into certain parts of the mass of signs, and this is the role of relative motivation.If the mechanism of language were entirely rational, it could be studied independently. Since the mechanism of language is but a partial correction of a system that is by nature chaotic, we adopt the point of view imposed by the very nature of language and study it as it limits arbitrariness (motivation plays a much larger role in German than in English ... with respect to Latin, French is characterised by a huge increase in arbitrariness)(22).

Another absolutist writer on the arbitrariness of language, Englefield, argues that in any language there are conventional ways of combining words in order to express the relations between ideas: "The fact that there are similarities in the conventions of unrelated languages can be readily explained by the common purpose for which languages were invented, the common features of the human physiology and the common elements in the human environment"(23).

The result is rather paradoxical. Though syntax and morphology diverge between languages almost as much as do the particular words used (and in some cases the grammatical divergences are more striking than the divergences in vocabulary), the proponents of the arbitrariness of language are much less positive about the arbitrariness of grammar. Saussure indeed comes close to the point of treating morphology and other grammatical aspects, the derivational and compositional features of language, as compensating rational forces the aim of which is to create a coherent language system precisely so as to reduce the difficulties caused by the initial, irrational arbitrariness of language. Englefield's reference to the similarities of form between unrelated languages invokes, as explanation, the common features of human physiology, human purposes, the human environment - but if these are real operating forces, they obviously constitute a natural and not an arbitrary basis for the forms which language assumes.

If one uses Saussure's questionable distinction between absolutely arbitrary and the rather less arbitrary, the suggestion seems to be that grammar and morphology are rather less arbitrary than are the individual words of languages. Certainly, if it is contended that grammar and morphology are equally arbitrary and equally the product of invention, convention or tradition, one marvels how primitive man, throughout the world, decided how to select the grammatical features of his language. Even modern linguists, approaching the subject with refined, sophisticated techniques, find great difficulty in analysing and presenting systematically the syntactic functioning of language. Must one assume that in each tribal group, each embryo language community, there was some primeval super-Chomsky, elaborating the forms which eventually became the comprehensive and subtle systematisations of languages such as Latin and Greek? Should one postulate an early inventor of language who decreed that these words should be adjectives, these should be nouns, and the two types of words should undergo parallel formal changes to show their agreement? Did some proto-Germanic genius ponder on the varying possible uses of the determiners 'the' and 'a' in the sentence and regularise the practice of the tribe? If the arbitrary origin of individual words is difficult to explain, then the arbitrary origin of grammatical forms is even more of an enigma.

But if, in the light of this argument, we take it that the origin of syntax and morphology was not arbitrary but was in some sense innate or natural even though at this point in time we are not able to explain the manner in which the natural or innate development may have taken place, some awkward questions immediately present themselves.

How, assuming that syntax and morphology are in some sense natural or innate, does one in fact explain the diversity of syntax and morphology between languages? This is a question which Chomsky has approached but not resolved. He postulates a common deep structure underlying all surface structures but he has not attempted to say or speculate what the physiological or neurological status of this deep structure may be, how in fact it could have originated as an evolutionary development or how in fact a relation can be established between a universal innate deep structure (common to all humans) and the diversity of surface syntax which one in fact finds - and to which he would deny any innate status.

How, if grammatical forms and syntax are innate or natural, evolutionary, does this biological system (presumably genetically-programmed) establish an intimate, functioning relationship with what is said to be the incoherent, arbitrary collection of individual speech-sound forms constituting the lexicon of a language? At first sight, the genetically-determined formal aspect of language and the arbitrary substance of language would seem totally incompatible, a mixing of chalk and cheese, like trying to play chess with a collection of random objects.

There is another troublesome aspect of this particular difficulty. Though syntax and lexicon are treated as distinct by traditional grammarians and modern linguists, the distinction is more apparent than real. In reality there is no sharp dividing line between the syntactic function of individual words and their semantic function; what appears as a system of inflections in one language is represented by a set of distinct, isolable words (prepositions, auxiliary verbs) in another, uninflected language. The relation between the content of traditional grammar and the traditional lexicon differs from one language to another; the seamless garment of language in its practical operation is divided up in different ways by the linguists as professional anatomists of language. The result is that language-functions classified as syntactic or grammatical in one language (and so ex hypothesi biologically determined and not arbitrary) are performed in other languages lexically, that is, using words in principle classified as arbitrary and accidental

The discussion so far has, to some extent incidentally, brought out a number of problems that arise from the standard view in linguistics that language is arbitrary The most obvious are how the origin of language should be accounted for if it is arbitrary and how the arbitrary components of language, the words, can be brought into a functioning relationship with the less-arbitrary parts, syntax and morphology, if one takes the less extreme view that language is only partially arbitrary. Beyond these, there is a whole range of other problems, not least of which is the difficulty of explaining the phenomenal speed and completeness with which children learn to speak their mother-tongue in all its complexity and with all the bewildering extent of the available vocabulary.

The point has already been made that to say that the present collection of words in a language and the meanings attached to them are arbitrary does not at all explain how historically those particular words, arbitrary or not, were formed and became attached to those particular meanings. Whilst the current link between word-form and meaning may seem obscure, or as Saussure and those following him would say 'opaque', it is a historical enquiry not to be resolved by any fiat of linguistic theory to determine how far, in the past, some link between word and meaning was apparent and how far in the origin of the word-form there was a link between sound and meaning. Etymologists can readily offer examples of words which originally were 'transparent' (the source of their meaning was apparent) and which have subsequently through sound or meaning changes become opaque. Insofar as there have been changes in the sounds and meanings of words, it is not clear whether those who treat language as arbitrary also take the view that the processes of language change are equally irrational, arbitrary or random - almost certainly they do not. The comparative philologists of the 19th century and later have shown very clearly how systematic and rule-bound changes in word-forms have been - but how linguists who believe in the essential arbitrariness of all words would reconcile this display of orderliness and rationality in the evolution of language with their wider thinking is an unanswered question.

Of course, a powerful school of linguists (the dominant school in the 19th century) evaded the problem of language origin simply by declaring that it was not worth discussing and not relevant for the science of linguistics(25). To quote Saussure again ( since he was so much the founder and leader of the mainline view of linguists): "No society in fact knows or has ever known language other than as a product inherited from preceding generations, and one to be accepted as such. That is why the question of the origin of language is not as important as it is generally assumed to be. The question is not even worth asking; the only real object of linguistics is the normal regular life of an existing idiom."(24)

Despite Saussure's dogmatic dismissal of the problem, some linguists who consider that language is an arbitrary structure have felt it necessary to attempt an explanation or a historical account of the development of language. In many ways the most interesting and one of the earliest careful explanations was that of Condillac (the summary that follows draws on the useful account given in one of the papers for the 1975 New York conference on language origins and evolution)(25):

Man's advance in knowledge has been through the deliberate use of arbitrary signs, in the form of language and speech. The crucial element in the origin of human knowledge is the origin of language. There are three types of signs: accidental signs, natural signs (cries, vocal gestures of the same order or other natural expressions of the face, arms, hands or of the whole body; the third kind of sign is 'instituted signs' or those which we have chosen ourselves, and which bear only an arbitrary relation to our ideas. The origin of language starts with natural signs which form part of the gestural 'language of action' that is natural to man. Gradually, the deliberate use of a few simple natural signs would extend the operations of the mind and the facility of reflection; the signs would in turn be improved, increase in number, aud grow more firmly familiar. The mind and the use of signs would interact to the mutual advantage and progress of both. The natural cries served early man as a model for a new language. They articulated new sounds and by repeating them several times with an accompanying gesture indicating the objects they wished to note, they accustomed themselves to give names to things.

Another serious attempt to describe the possible origin of language by a believer in the arbitrariness of language is contained in Englefield's papers (published posthumously):

An oral language must consist for the most part of arbitrary sounds. At some point deliberate invention must have occurred, yet this is conceivable only when man had already come to see how arbitrary sounds can be made effective for communication. This they might have learnt from their experience with arbitrary gestures (the motive for development of vocal rather than gestural signs being secrecy). The most striking thing that would distinguish the 'arbitrary' from the natural would be their unintelligibility to children and to strangers. Now it is one thing to explain how certain arbitrary sounds came to be used as signs and another to explain how such signs could be built up into a complete system which completely displaced all others. The great difficulty about any conventional system lies in ensuring that all should use the signs with the same meaning. At the outset such agreement in use could only be obtained among the members of a small and compact community ... the number of words that would be required for all purposes would be so large that the unpractised memory would not be able to retain them. It seems, therefore, a plausible theory that oral languages were first cultivated by small groups of men within the community for their own special purposes. Under such conditions a few intelligent individuals could experiment and built up more or less extensive vocabularies for their private use.(26)

A more recent attempt to describe the development of an arbitrary language was made by Sayre (approaching the problem of language and other aspects of human behaviour from the standpoint of cybernetics and information theory):

Spoken language might have originated from the transmutation of call-signs of prehominids (similar, as Hockett suggests, to the call-system of modern gibbons). The vocalisation of these calls was determined genetically and they enter the animal's repertoire without learning by experience. The result of their dependence on the presence of appropriate stimulus conditions in conjunction with their innate status assures that a call-system's components are incapable of being modified by tradition or experience. The transition from innate call-signs to freely and arbitrarily combinable phonemes would take place as follows: with erect posture, tool-carrying, better diet, the developing prehominid could spend more time chattering at his impressionable fellows and hence developing neural connections related to vocalisation. In the course of relaxed play situations, opportunities would arise for alert individuals to sense similarities between the elements of compound vocalisations and to repeat the sounds with exploratory variations so breaking the hold of the innate call-system and so leading on to the creation of new signal patterns. Breaking the hold of the innate through verbal play so that tradition takes the place of heredity as the major source of communication resources would free language from direct stimulus control. In a call-system, the vocalisation is merely a phase in a causal sequence; it does not assume the role of meaning, for example, food. when the signal becomes assimilated into idle chattering activity, thus displaced the vocalisation may become joined with particular perceptual circumstances by procedures of reinforcement, increasing the likelihood of association. As the conditioned individual tends to repeat that utterance, other participants may be led through imitation to form equivalent associations in their own neuronal channels.(27)

There have been a number of other similar accounts of the possible origin of language, in gesture, in song, in natural cries, though many of them do not depend directly on the assumption that language is essentially arbitrary in its present form. Of the three accounts given above, two link the origin of 'arbitrary' verbal language to the prior use of gesture and the other to the prior use of instinctive cries and call-signs. None of them seems to give a particularly plausible account of the transition from the naturally-based signalling systems to a fully arbitrary non-natural use of speech-sounds. Indeed, on the accounts given, one might wonder why it should be thought necessary to postulate a transition from a natural base to an arbitrary system. Certainly, in the case of the Sayre/Hockett approach the gradual transformation of natural calls and cries into naturally suitable words would seem more probable, and in the case of theories assuming gestural signs as preceding language, much more persuasive accounts can be offered of how gesture could gradually evolve into language, retaining a natural basis both for word-forms and syntactic structures. None of the accounts quoted seems to deal adequately with the difficulty of explaining how it would be possible to reach a consensus or agreement between individuals in a community on the selection of the particular arbitrary sounds to be used for particular arbitrary meanings. The practical difficulties involved not only in persuading a community to use a particular word for a percept but also in identifying clearly and specifically, without the aid of language, the particular percept to which reference was being made, are brought out very strikingly in Quine's discussion of the inscrutability of reference (in his lecture under that title).(29)

Nevertheless if accounts of the origin of language as an arbitrary system seem bound to be unsatisfactory and, in common with all other accounts of language origin whether natural or arbitrary, speculative and untestable, one might at least expect more attention to be paid by those who affirm the arbitrariness of language to two more immediately practical and manageable problems, how in practice the human mind can overcome the daunting load on memory and processing capacity created by the arbitrariness of language (the multitude of unmotivated words and grammatical forms) and how children so successfully tackle the even more daunting task of acquiring in a few short months the whole complexity of language structure and functions. Saussure(28) recognised the complications which basing language on the 'irrational' principle of arbitrariness of the sign would involve and he put the vital need to reduce arbitrariness as one of the major functions of 'relative motivation' of words and of the language-system as a whole - but he made no attempt to discuss what the rational or natural basis was for features of language operating to limit arbitrariness. Neither he nor most of his followers since then have taken any real interest in the psychological problem of language-use or the problems involved in children's acquisition of language competence.

So far the discussion of the thesis and of the problems involved in assuming that language is arbitrary have been in rather general terms. It may be useful at this point to attempt a more schematic presentation of the difficulties that arise for the proponents of arbitrariness and of the matters which any adequate theory of language, whether as an arbitrary or in some sense a natural product, ought to take account. This involves referring, more briefly, to some issues which have already been raised but also adding a number of other points which the study of language behaviour and language acquisition have brought out:

Problems in connection with language origin

1 If one was asked to construct an arbitrary set of symbols intended to relate to the ordered items of an area of perception (or to the total space of perception), how would one set about doing this? (This is the problem faced by the individual originator of a language)

2. Why, if someone is engaged in the construction of a language, should he carefully avoid making use of natural linkages of sound and meaning in order to make the meanings of particular words more easily memorable? (one would suppose that some one inventing a language would use every possible means to establish links between words and meaning)

3. How, without the use of language, would it be possible for a group of persons to agree to use within the group a particular wholly arbitrary set of speech-sounds (a particular word-form) to refer to a particular item of experience?

4. How, if initially words were allocated arbitrarily to particular items of experience could one overcome the fundamental difficulty to which Quine(29) drew attention, that is, the imprecision of deixis?

5. How, if all languages are invented and not natural, does it happen that every human community has available languages of very comparable degrees of complexity, regardless of the cultural level otherwise of any particular community?

6. Why if language is an invention should the position be so different from that for other human inventions, which are found in one community and not in others? If language as an invention was transmitted between communities, why should the character of languages differ as widely as it does between communities?

7. If any word can mean any thing, how are the established phenomena of cross-linguistic sound symbolism to be explained (people speaking one language can judge systematically better than chance the likely distinctions of meaning of words in another language not known to them)

8. If, as Saussure and others suggest, the meaning of an individual word depends only on its differentiation from other words (its position in a net of word-forms) and not in any way on the character of the word itself, how does an individual word in fact come to acquire that specific meaning? since on this theory we should need to know the meaning of the whole net of words before we could understand the meaning of any single word.

9. How, if the system and the elements (words) of each language are arbitrary, is translation between languages possible? There must be some underlying equivalence, isomorphism, between one language and another for translation to be possible. If, as Jakobson suggests, there is such an isomorphism, where can it come from if languages in fact are arbitrary systems?

Problems relating to language use

10. How, if the association of word and meaning is arbitrary, is it possible as a matter of practical experience that the word is deeply integrated into the structure of our thought to such an extent that one cannot see an object without the name of the object immediately and inescapably coming to mind?

11. How, if we have to deal with a multitude of arbitrary speech-sound forms, can the limitations of human short-term memory be overcome? An analogous question would be how many arbitrary telephone numbers do we find it possible to remember, when no clue is given by the number itself to the person whose telephone number a particular one is?

12. Why if language is arbitrary is deliberate innovation in lexicon and grammar so rare and difficult? Why is language so stable?

13. How if the grammatical structure of language, syntax and morphology, is arbitrary is it possible for ordinary people to be able to distinguish correct and incorrect grammatical forms, even when the complexity of the grammatical rules is such that linguists themselves are unable to give an explicit and systematic account of them?

14. How is it possible to assert that words are arbitrary (because they differ widely between individual languages) without at the same time asserting that all grammatical features are equally arbitrary (because they also differ widely between languages)? Or if one argues that there are universal underlying features of grammar regularly related to surface features of grammar (and that these deep features are innate or natural), why should one not assume that there are universal underlying features of word-formation, which would also be innate or natural and would be related systematically to the divergent surface-forms of words in different languages?

Problems in connection with language-acquisition by children

15. How, if language is totally arbitrary, are infants able very soon after birth to discriminate phonemes categorically, not only phonemes used in the parent language but also phonemes used in other languages? (They share this ability to discriminate phonemes categorically (i.e. to group speech-sounds into sharply divided categories) with a variety of animals.

16. How, if language is arbitrary, can one explain the rapidity with which children learn their first language, and the striking regularity observed in the order in which they acquire the various grammatical features of their language?

17. How, if words are completely arbitrary sounds, can children learn the name for an object almost instantly, without any normal process of reinforcement? what kind of special learning is this?

18. How, if words are completely arbitrary, do children learn the meaning and use of words like 'yet' or 'the', where there can be no possibility of learning by ostension?

The arguments against the view that language is an arbitrary structure seem strong. It is impossible to explain with any plausibility how, if language is arbitrary, it was constructed by a deliberate act or had some non-natural origin; it is difficult to explain how the human mind could possibly operate with a totally arbitrary system of signs and it is even more difficult to explain how every child learns to speak its mother tongue with such facility and in an astonishingly short period, acquiring not only an extensive lexicon but also a reliable understanding of the complex forms and rules of syntax and morphology (including for many languages inflectional systems of remarkable difficulty and completeness), a feat of learning far outstripping anything that adults could undertake.

But if an arbitrary origin for language seems implausible, and with it the whole idea of the 'arbitrariness of the sign', one is brought up against the unexplained features of language which originally led to the widespread acceptance by linguists of the assumption of arbitrariness, namely, the great divergence of lexicons between languages (Bloomfield's worry about the many differing words for 'horse') and the equally striking divergence between languages in syntactic structure and morphology.

Language is surely not arbitrary but how can it be natural? If an arbitrary origin for language seems implausible, one can hardly rest content with the situation that the foundation assumption for linguistics at present is wrong but the alternative assumption that language is natural also, at first sight, seems inexplicable. Maybe the way out of the dilemma is to recognise that the argument cannot be resolved within the constrictions of traditional linguistics. Saussure's contention, that the origin of language (and the related question of the mode of acquisition of language by children) was not a problem with which linguistics should properly concern itself, should be abandoned. For him, "the true and unique object of linguistics is language studied in and for itself"(30) but language is a vital part of total human abilities and behaviour, the most complex and the most distinctively human. Perhaps what is needed is a new point of view, the examination of language behaviour in a broader context, treating the acquisition of language by children as part of their total development and treating the functioning of language in parallel with the functioning of the closely related faculties of vision and hearing, and the organisation of the motor-coordination system.

A great deal of research has been done on human vision and hearing, much of it by way of study of the very similar processes of vision and hearing in cats, monkeys and other animals. Much of what has been learnt is suggestive for the course of human development and for human functioning. One area where this is particularly the case is that of what is known as 'imprinting', that is, the genetic preprogramming of alternative behavioural lines of development in the young animal. The following account of what has been learnt about imprinting is largely derived from the work of Konrad Lorenz(31) and other ethologists who have followed his lead.

The phenomenon of imprinting, a highly specialised type of behavioural development, has been most closely investigated in relation to the development of vision and hearing in certain birds and animals. Similar phenomena suggesting imprinting have been observed in fish, arthropods and certain other mammals, so that the evidence is of a very widely used genetic developmental process. In the 1930's Konrad Lorenz observed that newly hatched goslings would follow him rather than the mother goose if the goslings saw him before they saw her. Since geese naturally-reared manifest a strong attachment to their parent, Lorenz concluded that for some animals at any rate they have the capacity to learn rapidly and permanently in a very specific way at a very early age, and in particular to learn the characteristics of the parent. The sensitive period for this special kind of learning, imprintability, often belongs to an exceptionally early stage in the development of the individual, sometimes a mere matter of hours, but the period is always relatively well-defined. Once the determination of the imprinting has been fixed, the fixing of the object imprinted cannot be reversed. One of the most interesting and puzzling features of imprinting is that the imprinted response is not wholly specific; where the imprinting is of the characteristics of a parent or pseudo-parent, the imprinted response does not become fixed on an individual but on the species to which the individual belongs - there is a kind of generalising quality about the process of imprinting, the permanent fixing of a characteristic set of. features.

Lorenz recorded the phenomenon of imprinting as being radically different from the usual kind of learning precisely because of its rapidity and its permanence. Not surprisingly, the discovery of imprinting in the rather specific circumstances of Lorenz's research led to widespread search for comparable phenomena in other fields. Indeed, some child psychologists and some psychiatrists soon came to perceive a similarity between the evidence of imprinting in animals and the early development and behaviour of the human infant. It has been striking that the majority of known imprinting processes (now examined for a wide range of animals, birds and insects) concern social behaviour, the adaptation of the individual to the community of the species into which it finds itself born. This has been particularly noticed for birds and animals where there is a strong social component in their normal lives. It has also been observed that imprinting is not an isolated feature of development but integrated with more general developmental processes.

It now seems obvious that for a wide range of organisms imprinting has a vital role to play in the development of the individual creature's brain and behavioural maturation and that the process of imprinting is designed to modify the individual's behaviour in a way that directly serves survival, the adaptation of the individual to the environment in which it finds itself. Whenever in this way environmental circumstances regularly produce modifications in function and behaviour which result in an adaptation to these very circumstances, there is an overwhelming probability that we are dealing with what has been called an 'open programme' specified genetically. A genetic programme of this kind contains several individual programmes for the construction of various mechanisms and the open programme has the unique capacity to absorb further information from the external world in such a way as to allow this information to determine which of the available possibilities, the available mechanisms, shall be made operational. When this has been done, a new adaptation, serving the survival of the individual, is made permanent.

Vision and language are very closely related in man. Since Lorenz's first observations on imprinting in geese, much more evidence has been found very similar to imprinting in the development of animal visual processes (and, of course, there are close parallels between visual perception in humans and visual perception in animals such as cats and monkeys). In the animal and in the human, the basic structure of the visual system is not produced by learning after birth, since the system is fully structured at birth, but on the other hand one cannot say that vision is wholly innate since there is clear evidence that it can be, and is, restructured by actual visual experience provided the experience occurs soon after birth within narrowly defined critical periods (similar to the critical periods identified by Lorenz for imprinting in geese). The relation between the genetically-determined structure at birth and the malleable aspect of the visual organisation which can be moulded after birth is a complex one.

Much of the mental equipment necessary for pattern recognition appears to be present at birth, but the way in which this innate equipment interacts with experience has been most closely studied in experiments using newly-born kittens(32). Kittens are born blind, with their eyes closed, but by 14 days they show evidence of vision and thereafter begin to recognise patterns and objects. For the kitten the critical period, the period of greatest susceptibility when the visual system can be modified by visual experience, occurs at about four to six weeks of age. How the adult cat will see depends critically on the kind of visual environment to which the kitten is exposed during this period. For a cat developing normally, this is a period which allows it to acquaint itself with the common features of the environment in which it must live, and allows it to integrate its processes of perception with the development of its motor capabilities. If, however, in the critical period a kitten is exposed to a non-normal visual environment, for example, if particular types of visual experience are totally eliminated either by removing either all vertical or all horizontal patterning, then the kitten's visual system permanently adjusts to the distorted environment so that it will no longer be able to see vertical lines or horizontal lines and edges, as the case may be. A kitten exposed in the critical period to an environment without verticals can no longer perceive verticals so that it cannot avoid vertical obstacles. The perceptible environment for the kitten has thus been permanently changed as a result of a process closely analogous to the imprinting identified by Lorenz.

Very similar experiments have been performed to assess the significance of a process analogous to imprinting in monkeys, though for them the critical period is more extensive, beginning at birth and extending over the next six to eight weeks followed by a period of lower sensitivity lasting for about one year. It seems probable from these experiments that, apart from vision, other sensory systems and higher functions may also have critical periods during which behavioural mechanisms and performance can be sharpened, made more appropriate to the animal's actual environment (or of course distorted when the environment is systematically distorted).

The balance between the innate and the imprinted in behaviour varies from species to species and from function to function. Even without imprinting, the specificity of wholly innate programming in many animals can be truly astounding. Many animals, birds and insects, know apparently in fine detail much about the world of objects before they can have had any experience. Migrating birds use the pattern of the stars to guide them even when they have never before seen the sky. Insects and birds can respond appropriately to particular objects upon first encounter. A foal just born develops the complex behavioural organisation involved in walking within minutes. What is learned by an individual certainly cannot be directly inherited by its descendants but it is clear that the genetic coding underlying behaviour can become modified (by natural selection) to give the capacity to respond appropriately to objects or situations encountered for the first time by the newly-born individual. Animals very low down on the evolutionary scale rely almost entirely upon unlearned (innate) recognition of objects, and it is hardly surprising that animals higher up the scale should have developed similar innate capabilities for behaviour and perception, culminating in the ingenious and vitally important technique of imprinting, the multi-potentiality of genetic structuring.

Perhaps even more interesting and relevant for the subject of this chapter has been research related to imprinting in the development of birdsong.(33) The course of vocal learning in bird-song resembles that already described for imprinting in that the young bird is born with an inherited responsiveness to a broad pattern of external auditory stimulation. In the course of its experience of the range of sounds in its environment, it acquires more selective responsiveness to a particular subset of specific attributes found in the environment.

This process can be illustrated by describing more specific experimental results. A male white-crowned sparrow usually begins its full song at between 100 and 200 days of age (a rather lengthy period of learning in terms of the bird's total life-cycle). The song is highly species-specific in certain properties, that is, shared with all other members of its species, but the song also exhibits well-marked local dialects. If a young bird becomes deaf before it starts to sing, it subsequently develops a highly abnormal song which is in complete contrast with the highly controlled, tonal morphology of normal song, that is, the pattern of song it develops is distorted because of its inability to hear samples of 'normal' bird-song for its species. Perhaps even more relevantly and interesting, if a young bird with its hearing completely undamaged is raised in a special restricted environment where it cannot hear any song by others of its species, it also develops an idiosyncratic song which is abnormal but diverges less from normal song than does that of the bird deaf from birth. The song of the isolated bird has a definite patterned morphology, made up of relatively pure and sustained tones, which however shows a progressive loss of species-specificity, that is, tends to diverge more and more from the norm for its species. To put the matter anthropomorphically, a songbird, like a child, must learn from other birds, if it is to vocalise correctly.

A vital further point, in which the process also resembles imprinting, is that the ability to learn the 'right' song is manifest only during a specially sensitive period of the young bird's life and at this stage it is highly selective. If a sparrow has played back to it both the song appropriate to its own species and that of another species, it will learn only the song appropriate to its own species. If the sparrow is allowed to hear only the song of a foreign species, it will ignore it and tend to develop a crude song, like that of a totally untrained sparrow, that is, the bird is tuned to learn certain sounds and not others. This is an indication of the precision of the genetic priming, since there is every reason to think that simply in terms of the anatomic mechanism required, the song patterns of close relatives should be within the vocal compass of the sparrow. Thus motor constraints on sound production do not provide an adequate explanation for the selectivity and instead we have to look to sensory processes involving the neural pathways for auditory processing, sensitising the bird only to certain patterns of sound stimulation (in rather the same way as the human infant is sensitised virtually from birth to discriminate speech-sounds from non-speech sounds).

There is obviously room for debate and much research on how the remarkable phenomena of bird-song development should be explained, neurologically and physiologically. Presumably there must be structures in the auditory system (including the neurological structuring serving the system) which embody information about the structure of appropriate vocal sounds, appropriate patterns of sound, which have a capacity to guide motor development. These 'auditory templates', as Marler describes them, are genetically specified only to an extent adequate to produce an approximation of the normal song for the species, though they are still sufficiently specific to focus the young bird's attention preferentially on the song of others of its species (if there is the opportunity to hear it) and thus to provide an explanation of the selectivity of this very special learning process. It is not irrelevant to note at this point that the selectivity of the learning process in bird-song extends not only to the basic sounds going to form the song but also to the groupings of sounds in phrases and the more extensive structuring of song in the complete melody.

It seems a natural and obvious transition in the light of the striking potential of the process of imprinting in the development of adaptive behaviour in birds and animals to consider how far the phenomena of first language learning in humans show features resembling, for example, those found in developing bird-song. Karl Lashley's firm conviction was that the rudiments of every human behavioural mechanism would be found far down the evolutionary scale and also represented in primitive activities of the nervous system. It is not a very daring venture down the evolutionary scale to look for similar mechanisms, similar physiological and neurological organisation in young humans and young birds or monkeys. If imprinting or some analogous process operates in language-learning, then one could conceive of language as a part-innate, part-environmentally determined system in its development. Readiness to learn a language would be a manifestation of a genetically-primed ability to select, as the mother tongue, one specific language from a range of equally possible languages.

Language learning by children would then be very much akin to learning of their characteristic song by birds, in that the actual learning of the song is developed by exposure to the song sung by others of the species, a potential which leaves room for a good deal of plasticity which can, on occasion,lead to birds learning songs inappropriate to their species. The 'imprinting' approach to language would also take as a parallel the visual learning of kittens as just described, which allows the environment to which kittens are exposed in the critical period to alter the permanent shape and functioning of the cortical visual apparatus. In all these cases,Lorenz's geese attaching themselves to humans rather than to parent geese, kittens learning to interpret visual environments in one way rather than another, birds learning one song rather than another, we have a demonstration of a basic capacity in the nervous system for multi-potentiality, malleability, contingent pre-tuning of the bird or animal to adapt itself to the particular character of the environment into which it is born and to adjust its behaviour (during a sensitive period) to improve its chances of survival in that particular environment.

To provide a somewhat more specific basis for considering whether some process similar to imprinting appears to operate in the developmental process of children's first language learning, it may be useful to bring together, from the work of Roger Brown and others who have particularly studied children's language. some of the more interesting evidence relating to infants' learning of speech-sounds, children's learning of particular words and the development of children's grasp of grammar, and particularly of syntax.

At the 1975 New York conference on language origins, Marler specifically put forward the hypothesis that some sort of imprinting process might be responsible for children's acquisition of speech sounds (and I have already drawn on much of what he said in the immediately preceding account of imprinting). In the light of the research on the innate and environmental aspects of the development of birdsong, he pointed out that the capacity to modify vocal production as a result of auditory experience and to produce new sounds by the imitation of external models is fundamental to normal employment of speech in language: "The categorical feature of the perception of certain speech-sounds is what one might expect if special physiological mechanisms were involved. [In the study of phoneme discrimination] some of the critical boundary features, such as voice onset times, seem to be universals in all languages studied so far, so that one may think of them as species-specific characteristics". He noted, as others have done, that practically every speech contrast investigated to date has been shown to be discriminable by infants at extraordinarily early ages, 3 months or considerably less, with the place of articulation, voicing and liquids being discriminated phonetically, that is, categorically, vowels being continuously discriminated, and other speech contrasts (such as fricatives and intonation) being at least auditorily discriminated. Other research has shown that American infants of as young as one month can distinguish between synthetically produced phonemes /b/ and /p/, where the discrimination is dependent on very fine differences in the timing of voicing onset.

It seems appropriate to infer from these results the existence of special detectors for certain speech sounds. Cross-language studies have suggested that infants are even able to discriminate alien voicing contrasts, that is, contrasts that do not occur in the languages of their parents or community and of which, as a consequence, they could have had no previous auditory experience. The ability of human infants to recognise speech-sounds as a class well before the development of speaking is clearly established. These capacities of the human infant raise some thought-provoking questions. The interesting developmental issue is exactly how these phonetic (i.e. categorical) discriminative abilities become functional by 1-3 months of age or earlier. Marler(34) suggests that if one could invoke innate but modifiable components in the auditory and neurological systems of the infant (he would describe them as 'templates') for certain speech-sounds, then these could serve as an orderly frame of reference for the infant's developing responsiveness to speech patterns of the culture in which it is born. Beyond this there is the possibility that the imprinting components have a range of modifiability extending beyond the speech-sounds of the parent language to effectively the sounds of all and any human language, whether or not genetic factors tend to give a higher degree of probability to one or other subset of the total range of possibilities e.g. for reasons deriving from the anatomical differences between races.

The evidence so far collected on the development of children's language ability, in terms of their learning of particular words and of grammatical, syntactical forms, is almost equally striking. One of the most comprehensive and careful surveys of the observational material remains that contained in Roger Brown's book A First Language(35), though study of children's language has been one of the most rapidly growing areas of psychological investigation in recent years. No attempt will be made here to give any complete survey but simply to list a few of the more salient points. In the learning of particular words, children have a very marked, and not very surprising, bias towards the concrete, the graspable, the simple. Concrete objects and verbs that describe movement are the first learned and most frequently used words by children, though research evidence shows that the range of words understood by children is far wider than the range of words they will themselves actively produce in their speech. Not only are verbs of motion primary but their meanings have a strongly perceptual basis. In contrast, of the adult list of word-frequencies 16% of nouns (in one scheme of research) named categories having a characteristic visual contour whilst no less than 67% of the children's nouns were of this kind. In adult frequencies for verbs, 33% named movement whilst again 67% of the children's verbs named actions.

As regards syntax and morphology, many writers have commented on the speed with which children learn these aspects of their mother tongue; Minsky(36) (incidentally to discussing the problems of natural language understanding in the artificial intelligence field) comments on the phenomenal ability with which many a child acquires substantial grammatical ability around his second year.

Roger Brown(37) concentrated particularly in his studies of several young American children on the order in which the children developed and comprehended the use of grammatical forms (comparing his results with those of parallel research by a number of other child psychologists). His technique allowed him to treat the time and order of appearance of specific grammatical forms and morphemes in a statistically analyzable way, since for each of the children he and his helpers noted the order in which and the time at which each of fourteen particular grammatical forms appeared. To illustrate the results of his approach, the table below shows the data from his observations for three children:

     I                        II                  III

Present possessive       Plural              Present progressive
IN                       IN  ON              ON
OR Plural                Pres. prog.         IN
Copula (uncontr.)        Past irreg.         Plural, Possessive
                         Possessive          Past reg.
                         Copula (uncontr.)   Copula (uncontr.)
Past irreg.              Articles
Articles                 3rd person (reg.)
3rd person (irreg.)      Past reg.           Past irreg.
Possessive               Auxiliary(uncontr.) Articles
3rd person (reg.)        Copula (contr.)     3rd person (reg.)
Past reg.                                    3rd person (irreg.)
Auxiliary (uncontr.)     Auxiliary (contr.)  Auxiliary (uncontr.)
Copula (contr.)          3rd person (irreg.) Copula (contr.)
Auxiliary (contr.)                           Auxiliary (contr.)

Note: Contr. = Contractible Uncontr. = Uncontractible

Roger Brown commented that the developmental order of the fourteen morphemes as shown by these unacquainted American children was quite amazingly constant. Some factor or some set of factors caused these grammatical morphemes to evolve in an approximately consistent order in these children (statistical tests confirmed the high degree of correlation exhibited). What made the pace at which the forms were acquired and the degree of uniformity in the ordering even more surprising was that in general the parents seem to pay no attention to bad syntax nor did they even seem to be aware of their children's deviation from correct syntax. For some complex features, such as the correct use of the determiners 'the' and 'a', whilst on analysis one might conclude that listeners to speech almost never really need them to disambiguate sentences, nevertheless child speakers learned to operate correctly the exceedingly intricate rules governing the employment of 'the' and 'a' in correct speech. Brown felt forced in the light of these results to speculate that the order of progression in knowledge of a child's first language was likely to prove to be approximately invariant between children learning the same language and this invariance might go even further and at a higher level of abstraction extend to children learning any language, regardless of differences in grammatical structure. In his view the primary determinants of the order of acquisition of different grammatical features of a language might well prove to be cumulative complexity, both semantic and grammatical.

In the light of the above, one can attempt to present a broad picture of the way in which a child might acquire its first language:

1 Infants learn to distinguish speech-sound as such from other types of sound very soon after birth (their ability to do this may well be genuinely innate). They also learn to distinguish categorically (that is, between different phonemes) between differing speech-sounds at an extraordinarily early age, certainly as early as one or two months and possibly even sooner. This ability shows two remarkable features: first that they can discriminate between speech-sounds not in the phoneme-set of the parent language and, second, that the infant's ability to discriminate phonemically appears to be shared to a greater or less degree with other animals, which can also make similar phonemic discriminations (the categorical discrimination of speech-sounds may have its foundation in the pre-human perceptuo-motor system).

2 At the age of about 18 months, the infant begins to utter two-word sentences and after a further 18 months he can already construct nearly all the sentence-types which are possible in the particular language. All the basic structures of adult grammar are already present in the language of the pre-school child.

3 The child tends to learn concrete words with a clear perceptual reference first and to learn verbs relating to action before other verbs. There seems (as Roger Brown remarked) to be some perceptual predisposition to bear possible morphemes and words.

It is not out of place at this point to include some reference to the interesting work of Piaget over many years on child-language and child-thought. Some aspects of the answers he got from children when he and his helpers sought to find out what understanding young children have of words and of the naming of objects are thought-provoking. The young children studied by Piaget believed that they were doing much more than learning the name of an object; Piaget found that the child typically thinks that it is reaching the essence of the thing and discovering a real explanation of it. Children of 5 or 6, according to Piaget, can only conceive of the name as coming from the thing itself: 'One has only to look at a thing to 'see' its name'. Piaget comments: "This inability to dissociate names from things ... is very curious. The name 'sun' implies a yellow ball that shines and has rays. How did we know the name of the sun? God put it into men's heads. A star was called 'star' because people thought that name would go best. The sun was called 'sun' because people thought it was a good name and a bright one. Until the age of 6 or 7 children say that names come from the things themselves. They were discovered by looking at the things. Only as children grew older were they brought to accept that words and names are conventional and have in fact no intrinsic relation to what they mean".(38)

Most writers on language are content to note the infant's amazing ability to acquire these language skills within two years or so. Few attempt to explain why this should be so. What impels the child to 'improve' his speech at all remains something of a mystery - there is no evidence whatever of a difference in communicative efficacy and no selection pressure. The observable facts are the absence of any need for the teaching of language as well as the relative ineffectiveness of programmed training upon the rate of language acquisition. The regularity of language-onset as a milestone that fits into an ordered and fairly constant sequence of other maturational milestones is another observable fact and so is the apparent similarity in language acquisition strategies, the universal similarity of primitive stages, and differences in outer form between primitive stages and adult languages The picture is of plasticity during childhood with subsequent consolidation for the rest of the individual's life.

Roger Brown(39) comments that it is difficult to imagine how children could acquire language as rapidly as they do from parents who understand it so poorly unless the children were already tuned by evolution to select just those aspects that are universally significant. "There is in short a large biological component that shapes our human languages" and Brown went on to propose a radically different possibility (from any usual language-learning theory) that children work out rules for the speech they hear, passing from levels of lesser to greater complexity, simply because the human species is programmed at a certain period in its life to operate in this fashion on linguistic output. Linguistic output would be defined by the universal properties of language and the period of progressive rule extraction would correspond to Lenneberg's proposed 'critical period'. In much the same way McNeill(40) has argued that whatever grammatical subcategories are universal in the world's languages must be innate and a child uses the distributional evidence of the language to which he is exposed simply to discover which categories are used locally and what grammatical features they govern.

Richard Gregory, in discussing the extent of innate behaviour in animals and birds, commented that what would be truly surprising on an empiricist view of nature would be to find immediate recognition by young animals or humans of artificial or unimportant shapes. For example, he says, if a child was found to recognise a language without having been taught it, that would be startling for the knowledge could not have become genetically coded - but, he added, "there is no good evidence of this kind of innate immediate knowledge"(41). Nevertheless the evidence is that an infant is able, very soon after birth, to discriminate the phonemes used in the language of its community as well as other phonemes not so used, that there seems to be some predisposition for children to recognise morphemes and words and to acquire specific grammatical structures in a surprisingly uniform order. It could be achieved, one might suppose that there could be nothing more beneficial for a child born into a particular human community than to acquire as rapidly as possible knowledge of the community language, the ability to communicate with other members of the community. Such an ability would have much the same advantage for survival of the child as the young bird's innate ability to navigate across previously unvisited territory. The mechanisms of imprinting, which operate effectively to allow geese to attach themselves to their parents, birds to acquire the song of their species and other creatures to achieve remarkable behavioural feats, would seem in principle to be useful also for human beings if this meant that they could acquire rapidly 'cultural' aspects of their human environment in much the same way as other creatures are pretuned to learn especially rapidly the vital physical features of the particular environment in which they find themselves at birth.

In his book Biological Foundations of Language, Lenneberg advances a hypothesis rather on these lines in terms of what describes as 'resonance'. The following summarises the main line he takes:

There are many reasons to believe that the processes by which the realised outer structure of a natural language comes about are deeply-rooted species-specific, innate properties of man's biological nature The unfolding of language is a process of actualization in which latent structure is transformed into realised structure. Maturation brings cognitive processes to a state that we may call language-readiness. The raw material for the individual's language synthesis is the language spoken by the adults surrounding the child - which seems to function as a releaser for the developmental language-synthesising process. Because latent structure is replicated in every child and because all languages must have an inner form of identical type (though an infinity of variations is possible) every child may learn any language with equal ease. The realised structure or outer form of the language that surrounds the growing child serves as a mould upon which the form of the child's own realised structure is modelled. This manoeuvre is possible only because all languages are so constructed as to conform to the stringent requirements imposed upon them by cerebral language-data processing mechanisms. Exposure to adult language has an excitatory effect upon the actualization process much the same way a certain frequency may have an excitatory effect upon a specific resonator ... each natural language being a selected frequency band from the limited possible frequency range that is capable of eliciting resonance ... The inference we may draw from this material is that many animal forms traverse periods of peculiar sensitivities, response-propensities or learning potentials. Insofar as we have made such a claim for language acquisition, we have postulated nothing that would be extraordinary in the realm of animal behaviour.(42)

Others have taken very much the same line, though not necessarily expressing it by use of the same metaphor of 'resonance' employed by Lenneberg. Teuber(43), speaking as a neurologist and invited to sum up discussion at a conference on brain mechanisms underlying speech and language, observed what a staggering task it was to understand how we understand language and went on to say: "I would like to venture a guess. It may well turn out that certain universal features of human language such as the patterning of phonemes in terms of distinctive features are innate ... We are forced to turn, I believe, from psychologists to ethologists with ... the innate releasing mechanisms which represent in a sense innate ideas, residing in a given nervous system as a product of evolutionary selection" and he postulated "a central neural apparatus preset for acoustical analysis of phonemes and morphemes and strings ... We might derive clues for its mode of operation from a curiously analogy recently pointed out between the hierarchical structuring of the visual system towards which the research of Hubel and Wiesel on the relation between the visual cortex and stimulation of the retina seemed to point and the approach .. to syntax in terms of a hypothesis of an underlying 'deep structure' as a universal feature of language, which, together with Lenneberg's work, supports a 'radically nativistic' view of language".

Marler's interesting discussion of the relevance of imprinting for the child's discrimination of speech-sounds has already been referred to and his approach lends itself to assuming a wider role for imprinting in language acquisition generally: "each normal child comes equipped with a special apparatus for apprehending phonemes, for compounding them into morphemes, and for segmenting the stream of speech according to some syntax. The corresponding apparatus in his nervous system is plastic enough to acquire any language, during the critical period, and to acquire any of the possible languages with essentially equal ease. The vocal learning process is subject to genetic constraints that tend to favour development in certain directions. The template-matching hypothesis of verbal learning suggests how a compromise may be struck between the need for a degree of common rule-sharing in members of a species and the advantages of learned diversification of signal structure and use. A word that is recognised means an auditory pattern stored in the child's memory ... Descriptive studies of speech development reveal changes with age that are consistent with gradually improved skill in using the motor equipment and matching the specifications of auditory templates".(44)

Other research reported at the 1975 New York conference from study of aphasia suggested that there were distinct developmental patterns for the acquisition of lexicon and syntax, with words being learnt before grammatical forms, and that these differences in timing may reflect different rates of maturation of two neurologically distinct systems. This is of especial importance as suggesting that, beyond any imprinting of a potential for grammar, there may also be imprinting in some sense for the acquisition of words, of the lexicon of an individual language.

To sum up in the light of the views quoted above. Starting from the thesis that language is not an arbitrary system, the proposition is that there must be a genetic predisposition in the child to learn the language of some community and beyond this some pretuning to allow it to learn rapidly the language of a particular community. It would be difficult to explain the nature of this genetic predisposition if in fact in any real way the lexicon and syntax of language in general or of specific languages were wholly arbitrary. What kind of neurological or physiological preparation could there be for acquiring a wholly arbitrary system of symbols? If, for example, a genuinely arbitrary set of symbols was constructed (a random system constructed by a computer) and a child was required to match this set of symbols to the equally unfamiliar features of the world to which the symbols were intended to relate, who can doubt that the task would be impossible, as it would be impossible also for any adult. The equivalent question for the imprinting of bird-song would be whether a bird could be genetically primed to learn any type of sound and structuring of sound for its song. There must be some definition of the substance from which the possible song is framed and some limitation on the type of sequence of sound to which a bird could be genetically prepared to respond.

But if in some way the infant is pretuned not only to the range of human speech-sound (as it very clearly is in the displayed ability to discriminate phonemes virtually at birth) but also to respond preferentially to the sounds, voices and structures of actual human language (so that every language should be equally easy to learn), what can there be in the nature of existing human languages which makes it possible for them to be related in some way to genetically-determined physiological and neurological structuring of the infant? What are the non-arbitrary features of any human language on which the assumed imprinting mechanisms can operate? A very similar question of course can be asked about the neurological and physiological basis of imprinting in birds for bird song.

Perhaps we should not be too hesitant about the potentialities of the imprinting process and of fully instinctive behaviour. One has only to think about the action-language of the bees, the flying ability of young birds, the myriad behavioural abilities of extremely young animals, available almost on emergence from the egg or the womb, the remarkable navigating abilities of birds and salmon, and so on, to realise that nature achieves many things that one might at first think totally impossible. The hypothesis presented in this book is that language derives its non-arbitrary character, and therefore its ability to be pre-tuned neurologically and physiologically in the infant, from a direct relationship between the structures serving language in the human being and other major aspects of human functioning, notably the neural organisations subserving vision and action. No one would deny that the major aspects of organisation for visual perception and for action are innate, wired-in to the system, even though some aspects of vision may develop through interaction between the visual apparatus and the environment. What this means is that there is a predisposition to select from the stream of visual information playing on the retina those structures which reliably represent the real features of the external world and which must be apprehended if the human, or the animal, is to be able to move and act effectively and to live safely in the world. In a sense, no animal or human can be programmed fully and innately for what it will in fact find in the environment into which it is born. Each child and each animal is born into a environment peculiar to the community to which it belongs, and has to be able to perceive as quickly as possible, to comprehend and to act on information about the particular environment, or the particular language.

The contention in this chapter has been that language is not arbitrary, that the infant is pretuned to speak and understand any one of a multitude of languages which satisfy certain characteristics, notably the use of a certain range of sounds, speech-sound, and a certain set of speech-sounds, the phonemes. The infant is genetically prepared to extract from the stream of speech-sound the particularly meaningful speech-sounds, to perceive the formation of these phonemes into a particular class of word-forms relatable to items found in perceptual experience, and to comprehend and to form words n an ordered way into strings or sentences. The fundamental guarantee of the non-arbitrariness of language, which makes imprinting possible, is its relatability to the structuring of perception (particularly visual perception) and to the structuring of the human capacity for action, a theme developed more fully in the next chapter. Perhaps the early Wittgenstein was wiser than he realised when he said. "The word falls, one is tempted to explain, into a mould of my mind prepared to receive it"(45).