[Adapted from: Allott, R. 1989. Chapter 2 in The Motor Theory of Language Origin. Lewes: Book Guild.]
1. MOTOR CONTROL RESEARCH
2. MOTOR CONTROL AND LANGUAGE
3. SPEECH SOUNDS
4. WORDS
5. SYNTAX
6. CONCLUSION
REFERENCES
Sir Andrew Huxley in his Sherrington Lecture 'Reflections on Muscle'(1980) after speaking about the uncertainty and incompleteness of our knowledge of muscle said: "There is of course a special temptation to think well of one's own ideas, but provided that no one else is misled, the damage done by one person convincing himself of a wrong notion is not great - it may even do good by stimulating others to make fresh experiments in the hope of proving him wrong". It is with this same caution that the theory of the motor origin of language is presented.
There have been other motor theories of human function. Most notoriously, there was J.B. Watson's motor theory of thought and there was also the important Haskins Laboratories motor theory of speech perception. The theory presented here is a more general one than either of these and is based on recent progress in research into the organisation of action at the neural level. The theory is that language originated as a transfer from or translation of the elements and system of combination of elements of the neural motor system, with the expression of motor programs which originally developed for the co-ordination of vertebrate movement being redirected from the skeletal muscles to the muscles of the mouth, throat, chest etc. with the side-product that this expression of the motor programs was accompanied by the sound produced by modulated streams of air which we recognise as speech-sound. The theory is thus one of a change in the connectability of the neural system, the opening up of new channels for the external expression of motor programs. "In the brain, new functions, phylogenetically and ontogenetically, had to be grafted on to old ones in whatever manner proved to be feasible and consistent with the normal processes of evolution. At every stage, evolution had to improvise with the anatomical structures and inherent plasticities that happened to be available"(Sommerhoff 1974: 13). In so far as it is assumed that the redirection of the motor programs, the opening up of new channels, took place many hundreds of thousands of years ago (and could in any case never have been directly examined) the evidence for the motor theory cannot be direct but must be probabilistic or circumstantial. The attempt is to establish the plausibility of this account of the origin of language in much the same way as the effort has been to establish the plausibility of the theory of evolution itself.
The questions one might ask are: What sort of elements and processes must the motor system have to be able to carry out its functions? What features are there of the motor system that might be appropriate for take over by language? What aspects of language are there that could have been derived from the features of the motor system reflected in the semantic, syntactic and phonetic systems and structures of present-day language?
Two preliminary comments: first, we have many familiar examples of the expressive function of the motor system, quite apart from its involvement in language. Most obviously, the functioning of the motor system is displayed in facial expression. We do not need to be taught to understand the facial expressions of others. The smile and the frown are an immediate expression, concomitant, of a brain-state, an internal state of organisation which produces, as a by-product, not a deliberate smile but a natural smile, which must be the transfer of internal neural patterning to the motoneurons of the musculature of the face and mouth, a translation of one pattern of nervous activity into another neuronal system. Our understanding of the motor pattern of another, expressed in a smile, must follow a very similar pattern to our understanding of the motor pattern of another, expressed in a varied stream of sound, in speech. The vital question is how perception in one modality, seeing a smiling face or hearing a speaking voice, is reconverted into an internal neural patterning, in us the receiving organism, which is the same as or closely similar to that in the other person who smiles or speaks. Facial expression has its elementary units, its phonemes, the minimal elements of facial muscle change, its semantics, the meaningfulness of particular combinations of muscle movements, and its syntax, the way semantic elements can be combined simultaneously or sequentially, to convey more elaborated indications of the internal state of the individual. Darwin of course analysed in considerable detail the systematic patterns of facial expression in man and animals. That we can interpret facial expression or language, convert them into internal neural patterning structurally resembling the neuronal patterning in the person originating the smile or the spoken words, points to an intimate and reliable relation between visual perception, motor action and internal neuronal organisation. This power to interpret bodily action in another goes far beyond facial expression. Watching tennis, we often find ourselves translating the swing of the player's racket into incipient movements or tensions in our own arms; we feel the tension of the high jumper or the runner; we find our limbs moving with music or dance; feel directly with someone climbing or trying to hold on. Most obviously, we often yawn with someone yawning, smile with someone smiling, weep with someone weeping, laugh with someone laughing. Similarly, language could only have been effective from the start if the sounds made by human beings, breathing and at the same time moving their mouth, tongue and jaws, had been tied closely in some way to the practical life of human beings, and been immediately interpretable by other human beings.
The second preliminary observation, which follows quite directly from what has just been said, is that language cannot in origin have been arbitrary, either in the muscle-patternings used to produce the elementary speech-sounds, in the structures of the words formed from these elements or in the syntactic ordering of the available words. I agree with the view of Soviet linguists on this: "The common features of the world's languages are primarily attributable to the physiological homogeneity of all people and conditioned by the uniform structure of the human brain and the organs of perception., rendering them capable of reflecting the world around them in a basically identical manner. Similarities in sound reproduction can be put down to the similar construction of people's vocal organs ... The content of the external world which surrounds man is composed of objects and the connections between them. The reflection of the outer world in the human head consists in the cognition of these objects and of regular connections.For all the variety of the world's languages, they have an identical substratum in the reality which surrounds us and an identical goal or practical orientation to provide a means of communication".(Serebrennikov 1980: 3)
There is no compelling evidence that word-forms are arbitrary and much suggestive evidence that the words that emerge and survive are expressive or appropriate in one way or another to their meaning. The dogma that word-forms are arbitrary is a remnant of Creationism, very similar to the pre-Darwinian view that species were as they were for no reason other than that God had created them so. The only difference in the case of the origin of words is that Man, in his unfettered free will, is said to be the Creator. The slack argument, used to support the assertion that words are arbitrary, that words for the same objects or actions differ between languages, applies equally effectively or ineffectively to many other differences between human groups, not only syntax and phoneme-complements but also prevailing hair and eye colours, skin colour, food preferences etc. etc. There is no more reason to believe that there is only one uniquely appropriate word for a particular object or action than that there should be one uniquely appropriate way of performing any action, cooking food, playing the piano or one uniquely appropriate way of constructing a motorcar engine.
There is considerable experimental evidence, and considerable theoretical coherence, for the view that there is a fundamental relation between the syntax of language and physiological syntax, the syntaxes of action and perception, that the syntax of language is biologically based. There is also considerable positive experimental evidence that the array of phonemes is ultimately biologically, physiologically, determined. One might reasonably ask, if syntax and phonemic systems are arbitrary, what is the meaning of the search for universals in syntax or phonology? But if they are not arbitrary, given the interchangeability within and between languages of lexical and syntactic processes, how could an arbitrary lexicon interact with and on occasion substitute for a non-arbitrary syntax?
How best to approach the task of establishing the plausibility of the motor theory? One way of regarding the motor theory of language origin is that language was just a trick that one kind of animal by chance acquired and that has turned out to have substantial survival advantage. Another broader view is that the theory is not only one of language origin but more radically one of language function ie. it is a motor theory of language in general. That is, language currently in use is analogous to skilled motor action and the kind of research at present being undertaken into motor organisation at the neural level in many different organisms has a direct relevance for language theory. This seems the most promising approach since it allows one to draw on what is now an impressive body of experimental work. This research demonstrates that the idea of motor programs and motor sub-routines is not a merely fashionable computer analogy but an approach that can be explored in terms of identified neurons and neuron groups in a considerable number of living organisms, controlling specific patterns of activity in comprehensible ways. It would be useful evidence in support of a motor theory of language origin and function if one could relate the results of current research in neural motor control systematically to a model of the organisation of speech and language.
A rapid survey of some of the relevant research into motor control. Very simple organisms have no nervous systems; the sensory input impinges directly upon motile cells, the precursors of muscle cells. The history of animal development is essentially that of improvement in environmental control of the motor cells; basically, all observable animal behaviours are muscle movements, simply sequences of muscle contractions and relaxations. For a long time, the neural control of movement was identified with the investigation of reflexes, stereotyped patterns of action triggered by sensory input at a low level in the nervous system. The reflex mechanisms undoubtedly exist and can be readily demonstrated but the behaviour of the normal animal cannot be explained or determined entirely in terms of its reflexes. In rapid action, control through reflexes triggered by sensory inputs seems to be ruled out. Complex and rapid sequences of action require the postulation of some central nervous mechanism which fires with predetermined intensity and duration or activates different muscles in a predetermined order.
Lashley was one of the first neurologists to attempt an explanation in terms of what now are called motor programs. "The problems of the syntax of action are far removed from anything we can study by direct physiological methods today, yet in attempting to formulate a physiology of the cerebral cortex we cannot ignore them"(1951: ). What he foreshadowed 30 years ago was what would now be described as the modular organisation of the cortex, with the organisation of subroutines into more extended programs. Not only does the speed of required action frequently much exceed the response capability of any sort of chained sequences of reflexes (a Markov chain) with each sub-action triggered by the previous action but every complex action has to be repeated in circumstances which may differ considerably. We never perform an act like opening a door in exactly the same way twice. The input instructions have to determine the result independently of conditions at the specific muscles involved; a generalised program avoids the need for neural storage at a high level of an impossibly large number of alternative sets of instructions.
The brain accumulates high-level motor routines in terms of sequences of variously directed motions and in terms of relative position but without any specific reference to the absolute position in which, or the part of the body with which, these motions are to be effected. An important task in brain theory then is to isolate the substructures of motor behaviour and to investigate ways in which they are combined to produce a co-ordinated action(Szentagothai & Arbib 1975: 32-33). Each subprogram is generalised so that it can be realised in any limb with a sufficient diversity of musculature. The emphasis is on what has been described as the building-bricks, the basic mechanisms available to the body from which more complex co-ordinated responses can be constructed(Bizzi 1983: 3). An internalised program must be made from generalised sub-programs linked together. There is a principle of neural economy at work; it would be highly uneconomic to duplicate many times over simple instructions which have to be directed to many different points of the musculo-skeletal system. The result is what one might call a repertoire of detached motor programs and sub-programs.
These ideas on the organisation of action are soundly based on research eg. by Evarts and Bizzi. Evarts' (1973) work has showed conclusively that cells in the cortex, cerebellum and basal ganglia all become active prior to movement (the role of preplanning): "a consequence of the internal stimulus taking a routine course and using the required movement from the available repertoire"(Granit 1977: 183). "The program underlying final arm position ... indicates that a number of parallel processes underlie arm movement and that motor control may be thought to be organised in a modular fashion"(Bizzi 1980: 141). Movements involving one or more joints are strikingly similar from person to person which indicates that the elementary motor programs may well be innate, part of standard human (and even vertebrate) neural structures. ... Some of the basic ideas were presented very clearly by Bernstein: "each of the variations of a movement, eg. drawing a circle of varying size, on different surfaces, in different directions, demands a quite different muscular formula and even more than this involves a completely different set of muscles in one action. The almost equal facility and accuracy with which all these variations can be performed is evidence for the fact that they are ultimately determined by one and the same higher directional engram in relation to which dimensions and position play a secondary role ... extremely geometrical, representing a very abstract motor image of space"(Bernstein 1967: 49).
As another example, the act of grasping an object in the field of vision starts from many different postures and may reach it by as many different trajectories as there are occasions to grasp. This is made possible by the idea of a basic repertoire of instructions to which any set of muscles in the body can respond. If, for a complex set of levers, formed from muscle, tendon and bone, particular quantitative changes in length or tension, duration, speed, etc are specified, then invariably some action characteristic of the instruction set will result. If this same set of instructions is fed in to a lever complex with a different conformation, a different type of action will result but necessarily having an underlying isomorphism with the first action.
Lashley suggested that a response might be built up from its individual movements in such a way that each response involved two distinct processes, one consisting of neural activity continuing for the duration of the response and facilitating all its component movements, the other being a series of neural activities organised by internal connections to regulate the timing and sequence of the individual movements ... Many examples can be cited - from such innate patterns as swallowing to the highest form of learned response like uttering grammatical sentences - to fit the general scheme(Milner 1970: 92). In recent research, the idea of the motor program and sub-programs which go to form it has been widely used. The concept of a motor program has been defined as 'a physical system of nerve cells whose organisation constitutes a coded store of instructions, compounded both of heredity and learning.'(Young 1978: ), 'an abstract memory structure prepared before the movement which, when executed, results in movement without the involvement of feedback requiring a correction for error; a generalised motor program being one that can be carried out in a variety of ways depending on the parameters specified.'(Kelso & Clark 1982) A recent description by Gallistel of the organisation of action is that "the system that generates animal action is hierarchically structured ... At the bottom of the hierarchy are a relatively few kinds of elementary units of action. Assuming that the mnemonic code for skilled action represents complex actions as combinations of simple motions oriented with respect to body space, there must be a stage that translates this code into appropriate neuromuscular activity ... Studies have repeatedly demonstrated that skilled action is hierarchically organised (forming 'chunks', larger units of action)"(Gallistel 1980: 390, 367).
The analogy with computer programming is close but the idea of motor programs is not constrained by the computer analogy. A controlling motor program is thought of as having many activable links to other neural complexes, chains and circles etc. Rather like procedures using a computer, values can be read from the controlling program into the pre-established elementary sub-programs. The operation of activating a controlling program would then be a matter of selecting the linked sub-programs to be initiated, prescribing the order in which they are to be initiated, reading in the values to determine the manner in which the sub-programs would be executed. The essential point is that the central motor program is an organising program, calling upon other already available elements which go to form the more extended main program.
With his usual prescience, Lashley expressed his conviction that the rudiments of every human behavioural mechanism will be found far down in the evolutionary scale and also represented even in the primitive activities of the nervous system. Most recent work on neural motor programs has in fact been concentrated on organisms far down the evolutionary scale but it has nevertheless yielded valuable results with a wide application. Remarkable progress has been made using invertebrate model systems, whose behaviour is the result of interactions of identifiable neurons within networks. The initial observation, verified by many experiments, has been that the essential patterning of a movement is retained after exclusion of peripheral afferent information. The experiments on motor potentials in the absence of peripheral information confirm the existence of cortical programs(Granit 1977: 166). The existence of multipotential motor programs has been demonstrated in a wide variety of creatures, locusts, leeches, molluscs, planarian worms. The principles underlying the co-ordination of locomotor behaviour in invertebrates have been defined; locomotory sequences are controlled by centrally determined motor programs.
One aim of studies on an invertebrate like the leech is to analyse how complex behavioural acts are built up from simple, elementary reflexes ... The leech resembles other invertebrates such as the crayfish, the locust and the cricket, in which central motor programs involving a small number of individual cells have been shown to control complex patterns of co-ordinated movements. There are also similarities to the co-ordinated movements of vertebrates(Kuffler & Nicholls 1976: 369-370). In crustaceans, each of a few identifiable internuncial neurons is capable of starting a rhythmic movement of the swimming foot of the lobster ... The essential point is that the act is organised as a fixed process run by a few push buttons. .. A decisive factor must be the degree of adaptability and complexity of the repertoire of movements of an organism. The cyclic movement of the swimming foot of a crustacean is simple compared with what the human hand has to accomplish. However, the existence of command neurons underlines the validity of a principle of economy, an ideal minimum of mobilisers of effort(Granit 1977: 163). A similar organisation has been found in studying the neuronal networks controlling locomotion in locusts. The motor program for the locust jump consists of three phases .. The neuronal circuitry controlling these three phases of the jump is centred around two pairs of large thoracic interneurons(Robertson & Pearson 1985: 22). A semi-intact locust can produce flight motor patterns even though proprioceptive inputs have been removed(Hoyle 1983: 589).
More dramatically, the completely isolated brain of the mollusc TRITONIA produces the same neural program appropriate to swimming that is seen in the same cells in the intact animal. ... Both the neural program and its timing and the pattern of impulses (in the 6 specific neurons involved) were virtually identical, when triggered by a non-specific electrical impulse applied to an anterior sense nerve(Hoyle 1983: 584). The premotor pool of interneurons that underlies the generation of the swimming pattern of the mollusc has been almost completely worked out(Selverston 1985: xii). The central pattern generator is envisioned as a group of central neurons that generates a sequence of temporally or spatially co-ordinated activity ... not pre-wired to produce only one activity but multi-functional(Getting & Dekin 1985:3-4). The group of premotor neurons can participate in more than one behaviour and constitute what is described as a polymorphic network. A polymorphic network is one that can be organised into multiple states or configurations called circuits. Each circuit may involve the entire set of neurons within the network or some subset of them; each circuit can be transformed into the others so that the network can adopt any one of its different states. Pattern generation emerges as a property of the network as a whole. The ability of the network to generate patterned activity depends upon the interaction of both the synaptic connectivity and the intrinsic cellular properties of each neuron. The command function is viewed not only as initiating action but also as instructional, serving to organise the network into an appropriate configuration to generate a particular motor pattern, as well as selecting and activating the motor system. The individual expression of the subcircuits in the network is dependent upon both the type of initial stimulus and the internal state of the animal; this points to a new concept of network plasticity(Getting & Dekin 1985: 15-16). Multifunctionality is found in other creatures; a set of arthropod legs can make many different motor programs through using exactly the same muscles: two or three forward walking gaits, backward walking, sideway walking, running, swimming, righting, turning, tilting, climbing and so forth(Hoyle 1983: 588). The similarities between observations on bird and insect behaviour were striking and suggested that common general principles have evolved in neural control of movement(Hoyle 1983: 584).
The conviction that there is some special relation between speech and the organisation of action is one that a considerable number of researchers have expressed in terms such as: 'Language is a form of action'(Swinney 1981: 388); 'Praxis and language are clearly related and appear to share the same neural structure'(Kertesz & Hooper 1982: 276); 'Language as a function is not purely emergent but evolved in relation to central programs for motor behaviour'(Kinsbourne 1978: 583); 'The development of language seems to have incorporated brain mechanisms originally developed for motor learning'(Ojemann & Mateer 1979: 205); 'Control over articulation is fundamentally similar to control over other kinds of action'(Fowler, Remez & Turvey 1980: 407); 'The complex and highly co-ordinated movements used in speech depend on processes of muscular organisation which have some features in common with locomotor activity'(Roberts 1978: 366); 'Comparing speech and non-speech movements we find many similarities'(Lindblom 1983: 233); 'Languages tend to evolve sound patterns that can be seen as adaptations to the motor mechanisms of speech production' ... The phonological regularities ... arise out of biological processes that are not unique to speech but are characteristic of motor behaviour in general.'(Sommerhoff 1974: ) Others have suggested a particularly close relation between the neural programming underlying speech and that underlying hand and arm movements. So Kimura concludes that "the relation between certain limb movements and speaking is compelling ... speech should be considered primarily as another, albeit special, complex motor function ... If one considers the relevance of these findings to the question of how language or communication evolved in man's history, we are forced to conclude that there has been a close association between the movements of the upper limbs and the movements of the speech musculature. This is suggested not only by the fact that vocal utterances and certain arm movements occur together, in both deaf and hearing persons, but also by the overlap in the neural system subserving both"(Kimura 1976: 152-153). In the light of Bizzi's finding that arm and head movements depend on neural pattemns that are programmed prior to movement initiation, one would look for parallel pre-programming of the comparable speech musculature movements.
The relation between motor programming and speech programming can be examined at each level, the phonemic, the lexical and the syntactic. At the phonemic level, the link is readily established. Each phoneme in turn may be thought of as an organised set of instructions to the muscles involved in speech production; Lindblom suggests that each phoneme has an invariant 'program' that is unaffected by changes in syllable stress or speaking rate (tempo), with co-articulation resulting from the temporal overlap of successive programs (Lindblom 1963). This of course fits well with the ideas of 'auditory targeting' for phonemes proposed by MacNeilage and also with the conception that phoneme programs are 'shingled together' suggested by Liberman and his associates in their 1967 paper. These ideas lead one to the conception of a motor-alphabet underlying speech, related in some way to the elementary motor-patterning underlying other forms of action. In this connection, one has to consider what Lieberman describes as the 'very odd' phenomena of categorical speech perception, the fact that we distinguish readily between /ba/ and /pa/ even though both sounds may merge imperceptibly into one another in terms of the physical parameters of the phoneme-production eg. when speech is synthesised. Lieberman proposes that this categorical perception "reflects the presence of neural mechanisms that represent a 'match' with the constraints of the human supra-laryngeal vocal tract"(Lieberman 1984). There is a good deal of experimental evidence that this cannot be the explanation of categorical perception of speech sounds.
Much of this evidence was reviewed in the papers for the New York Conference in 1975. Dewson et al. had shown that rhesus monkeys could be taught to distinguish between a variety of speech-sounds in a categorical way. They could distinguish /i/ and /u/ and were able to transfer this discrimination from a male voice with a fundamental frequency of 136 Hz to a female voice with a fundamental frequency of 212 Hz.(Dewson, Pribram & Lynch 1969). Burdick & Miller(1975) found that the chinchilla, a member of the rodent family, could learn to distinguish between /a/ and /u/ not only for different vocal productions by the same talker, but could generalise this discrimination to vowel statements by different talkers, changes in pitch level and changes in intensity. Kuhl & Miller(1975) using synthetic speech had shown that chinchillas can distinguish between /ta/ and /da/, between /ka/ and /ga/ and between /pa/ and /ba/. In another study using natural speech, they had found that, following training, syllables containing either /t/'s or /d/'s can be discriminated by chinchillas despite variations in talkers, in the vowels following the plosives and in intensities. Work by Sinnott(1974) with monkeys had shown that they are able to distinguish between acoustic correlates of the place of human articulation with /ba/ and /da/. Morse's(1976) results using rhesus monkeys were also strikingly in agreement. Using the discriminations /b/ /d/ /g/, rhesus monkeys despite their inability to produce the full range of human speech-sounds, nevertheless discriminate between-category change in place of articulation better than a within-category change - an unexpected finding. In this connection Morse commented that it was of interest that chinchillas classified the speech stimuli so as to put the boundary in much the same place that human listeners do.
These results with rhesus monkeys and chinchillas are indeed very surprising. Why and how should the animals be able to make distinctions between sounds that are characteristic of human speech? The picture becomes even more intriguing when one takes into account parallel research into the speech-sound discriminations made by human infants, before they are in any way able to speak or produce speech-sounds. As Morse reported, with few exceptions, practically every speech contrast investigated to date had been shown to be discriminable by infants within the first two or three months of age. Their ability to do this could not depend on extensive experience with the articulation of these contrasts. Furthermore, cross-language studies suggested that infants are able to discriminate voicing contrasts that do not occur in the languages of their parents or their community. Both Warren and Morse reached similar conclusions in the light of this evidence. "Since extended speech sounds can be differentiated by animals that are themselves incapable of producing such sounds, it appears probable that our ancestors had the potential for discriminating speech sounds we now use before they could produce them; there is evidence that human speech perception employs prelinguistic abilities shared with other animals to distinguish between phonemic groupings"(Warren 1976: ). "Auditory feature analysers may conceivably be a property of the mammalian auditory system rather than unique to a species that has evolved the ability to produce the range of human speech"(Morse 1976: 704) and Liberman commented similarly that "in the development of language, nature took advantage of a categorical distinction characteristic of some mammalian auditory systems"(Liberman 1976: 721). But why, one is impelled to ask, should a wide range of mammals have this very specific categorical capacity, without any apparent relation to the auditory needs of the animals concerned?
The issues that arise are closely related to those dealt with in the important work of the Haskins Laboratories (Liberman et al.) on the categorical perception by adult humans of speech contrasts. Stimuli varying continuously along a continuum are not perceived continuously but in abrupt steps. The two phenomena - lack of acoustic invariance (in that there is no unique acoustic representation for each phoneme) and categorical perception of speech - were brought together to make the same point, that what matters for speech perception is not the constitution of the acoustic signal but the manner in which we speak so as to produce this signal. The motor theory of speech perception as then presented proposed that the brain translates the speech signal into information representing the motor commands which would be issued to the articulatory apparatus if we were to attempt to utter the signal and it is this motor command that we process in order to perceive speech. Liberman and his collaborators concluded that the motor commands (as indicated by EMG activity in the various muscles of articulation) were more nearly invariant with phonemes than were the acoustic signals. "The most general form of the view that speech is perceived by reference to production [involves] the assumption .. that at some level or levels of the production process there exist neural signals standing in one-to-one correspondence with the various segments of the language - phoneme, word, phrase etc. ... Perception consists in somehow running the process backward, the neural signals corresponding to the various segments being found at their respective levels. In phoneme perception ... the invariant is found far down in the neuromotor system, at the level of the commands to the muscles"(Liberman et al. 1967: 450-452). Liberman emphasised the essential fact that we do perceive phonemes, that phonemes are psychologically real. He added the rider that whilst a motor theory of this kind must be concerned with the nature of the mechanism, it need not necessarily deal with the question of how the mechanism was acquired in the history of the individual or the race. This was an interesting, but separate, question.
The motor theory of speech perception was from the start subjected to criticism from a number of directions. Much of the criticism could be met by supposing that auditory feedback in speech production by the individual could establish categorical distinctions subsequently decoded from the acoustic stream when the individual listened to other speakers. But in the light of the experimental evidence on the ability of infants and various animals to make categorical distinctions between speech-sounds, a much more serious difficulty arises for the motor theory of speech perception. Neither the infants nor the chinchillas or rhesus monkeys can be decoding speech-sound on the basis of a link established between the neural commands for speech production and the individual's analysis of his own speech. For some unknown reason, these animals have an auditory analysing device relevant for human speech categories but, so far as one can judge, in no way relevant for animal sound production. Presumably, it means that in mammals, there is some natural categorical system, by no means necessarily an auditory one, which has served as the basis for the construction of human phonemic speech production and speech perception. In the monkey and in the chinchilla, as well as in the infant, the sounds heard are referred back to a system which is organised so as to distinguish between certain categories of sound, essentially to some neuronal assembly which analyses in a uniform way. What could the nature of this analysing device be? The proposal in this paper is that the common element is generalised motor patterns, motor programs. The motor programs for producing phonemic sounds are derived from the primitive motor programs for producing bodily movement generally, diverted to producing movement of the organs of articulation. What the rhesus monkey shares with the human infant and with the chinchilla is very similar skeletal and muscular organisation, very similar means to control the length or tension of muscles and so the position or movement of the limbs, of the skeleton generally or in fact of all body parts subject to voluntary control.
This conclusion is in harmony with Turvey's comment: "Speech perception and speech production are related by abstract structures that are common to both but indigenous to neither"(Turvey 1977: 259). What are the implications of this for the next level of organisation of speech above the phoneme, namely the structure of the individual word? What is the extent of the biologically-determined phoneme-set available for formation into words and what processes govern the selection and combination of phonemes into words related to objects or actions?
If the existence of distinct phonemes as is here suggested is a by-product of the organisation of the motor system, then one would expect that parallel to the variety of distinct elementary movements which vertebrates and humans can perform, there would be a set of distinct phonemes. Looked at from the opposite direction, assuming that the phoneme-sets for many languages are in the range of 40 to 60, then this would imply the existence of at least 40-60 distinct elementary units of motor action, 40-60 partial movements or positions of the limbs which can be matched with the array of phonemes. The hierarchical structure of the motor system would be built on the basis of a limited set of motor elements, which could be combined in an unlimited number of different ways (motor-words), just as phonemes can form an unlimited number of spoken words. It becomes relevant to seek to observe and analyse into their elementary components the range of bodily actions that the human, or indeed the rhesus monkey or the chinchilla, can perform. If, as Kimura has suggested, there is in fact a special neural relation between speech and movements of the upper limb, one can concentrate first on analysis of human, or monkey, arm movements. The essence of a motor element, a motor sub-routine, for the arm is that it can be referred to a particular muscle/joint complex operating in a specific co-ordinate system formed from the vertical and horizontal co-ordinates of the body. For elementary movements, the human or monkey arm can move forward and up, sideways across the body or it can be bent and moved up in front of the body. In these movements there is a succession of positions which could be correlated with the set of articulatory patternings required to produce the range of phonemes, using the musculature of the vocal apparatus. The specificity of the phoneme is the accidental result of the application of the different elementary motor subprograms to the muscles which went to form the articulatory system. Every phoneme is correlated with a particular position or movement of the arm; each phoneme is the result of a distinct motor subprogram which specifies the degree of contraction in a complex of muscles, the tension and the balancing forces determining the position assumed by the limb or the trajectory of the limb movement. The subprogram may operate to determine the timing of contraction of muscles in a group controlling arm position or set the biases between agonist and antagonist muscles which determine position. What we are concerned with is the existence of a motor-alphabet underlying all complex bodily action which has been given a particular application, interpretation and external expression by being redirected to the special set of muscles involved in articulation.
Gesture is the most obvious demonstration of the parallel expression of motor programs in speech and bodily action. As McNeill has repeatedly pointed out, "the omnipresence of concrete sensorimotor ideas in speech is a phenomenon that has received almost no attention from linguists or psychologists"(McNeill 1979: 3). The translation of speech motor-programs into skeletal motor programs has been demonstrated in human beings (gesture and synchronised movement). But we are faced with the question: Where do the specific forms of words come from, if, as is suggested in the motor theory advanced here, the components of words, phonemes, are derived from the elementary motor programs for bodily action?
Words are a read-out of neural structures in much the same way as actions or facial expressions are a read-out of internal states, thought and feeling processes. Logically prior to the meaning of a word is its physical manifestation as a unit of neuromuscular action in the speaker and as an event in the listener(Swinney 1981: 303). A set of signals in the optic nerve represents the presence of a cup in the visual field. In each case certain groups of cortical nerve cells will be activated. This is J.Z. Young's(1978: 73) account of perception as related to individual words. Before we speak a word we have the impression from introspection that some process of 'mental' organisation occurs before the movements of a response begin and the preparations are by no means confined to the first movement of the contemplated response. (To type the word 'contemplated' we do not have the pure idea of pressing the 'c' and then waiting to see what movement will appear next, as the Markov-chaining theory would predict). Somewhere in the nervous system there is an activity corresponding to the whole word and beyond that in all probability a still higher order of activity corresponding to the meaning that is to be conveyed by the sentence in which the word appears(Milner 1970: 92).
If one accepts that the distinct phonemes found in the world's different languages were an accidental product of the set of distinct elementary motor programs for regulating human and other vertebrate limb movements, where did the structure of the words formed from these phoneme-motor programs come from? Warren in his paper for the New York Conference drew an analogy between the formation of syllables or words on the one hand and the formation of chemical compounds on the other. "The holistic perceptual groupings recognised by listeners may be considered as temporal compounds, that is, an aggregation of auditory items into groupings having special holistic characteristics ... When elements are combined in certain lawful manners, a compound is formed having specific properties that do not reflect in an additive fashion those of the constituent elements. A particular compound may be distinguished from others but it may not be possible (or generally desirable) for an observer to analyse these compounds into their elements directly"(Warren 1976: 712). Warren goes on to argue that phonemes are not perceived as such during speech but are derived following recognition of syllables and words and so should be considered to be useful constructs but without direct perceptual basis. This may or may not be so; our inability to identify the phonemes in heard speech directly no more indicates that phonemes are physiologically unreal than our inability to perceive sodium and chlorine separately in salt proves that sodium and chlorine are fictitious constructs. Nevertheless, Warren's chemical analogy, which others have also used, is a useful one. It suggests that words, as neural structures, can be formed from the co-activation of the motor subprograms for phonemes which are then melded together to form a distinct neural program for the whole word with, as has been suggested, higher level neural structures for strings of words forming a sentence with a distinct sentence-meaning.
In some such way one can envisage elaborated motor programs as the neural correlates of words (and sentences) but there remains the further crucial question how particular word-programs acquire their characteristic shape, and become associated with distinct perceptual objects or actions, that is, with distinct neural patterns in perception associated with external things or actions. Earlier this paper emphasised how implausible it is to assert that word-programs are arbitrary, but if they are not arbitrary, what is the source of the word-structures we find in any given language? There have been a number of profitable experimental approaches to the issue of the processes determining word-structures. Some of the most interesting were discussed by Roger Brown(1957: 115-118), those of Usnadze who concentrated on the selection of words appropriate for visual patterns and those of Wissemann who examined the formation of names for unfamiliar sound-sequences. Wissemann's experiments might be classified as concerned with traditional ideas of onomatopoeia, though Usnadze's (and Kôhler's famous TAKETE/MALUMA experiment) could not.
Wissemann provided noises which were to be assigned suitable names and asked his subjects to invent the names in whatever way they thought fit. The names invented by the subjects showed many remarkable regularities. The number of syllables in the invented words corresponded not to the duration of the sounds but to the number of divisions heard in the noise, the differentiation of the noise; a noise that had an abrupt onset was usually given a name beginning with a voiceless stop consonant ( /p/ /t/ /k/ ); a noise having a gradual onset was usually named by a word having an initial spirant consonant ( /s/ /z/ ). The initial sounds of the names thus reproduced the stimulus gradients of the referent noises, that is, the motor characteristics of the articulation of the consonants used. Wissemann's subjects also took a uniform view of the role of vowels in the invention of appropriate words; most agreed that the vowels should be used to express the pitch (high, middle, low) and tone colour (bright, colourless, dark) of the referent noise. These two dimensions appeared in perfect correlation, with high pitch corresponding to bright tone colour; the vowels /i/ /ô/ and /õ/ are clear and high while /u/ and /o/ are dark and low.
Analogous results appeared in Usnadze's experiments in which the subjects were presented with various visual shapes and asked to choose suitable words for them. The number of syllables in the words chosen would be related to the number of perceptual subwholes in a drawing; the form of the visual gradients would be reproduced appropriately in sound; the sound spectrogram pictures the plosive /p/ as a dark smudge that does not shade off at the edges while /s/ is a gradually darkening area. The spectrogram shows how recurrence, relative strength and rhythm in sound can be represented in a picture. In Kôhler's famous experiment, hitherto unknown nonsense words (TAKETE and MALUMA) were matched with extraordinary consistency by subjects from many cultures with unfamiliar 'nonsense' shapes. These experiments strongly suggest that word-structures can have lawful relationships to the structures of auditory or visual percepts, and that there is a system of visual as well as auditory onomatopoeia (name-making) which could have contributed substantially throughout the development of language to the formation of word-structures appropriate to their meaning. What is involved is an isomorphism at the motor level between speech and the contents of perception.
The very extensive research on 'sound-symbolism' fits extremely well into this perspective. There are many interesting examples of words in different languages where the meaning and sound appear closely related eg. the range of words beginning with SL in English quoted by Firth(1964). Or the Spanish word for the bird in English called wagtail 'pimpim' which a Spanish linguist says gives not an acoustic representation of the song of the bird but a representation of the optical impression made by the rapidity and abruptness of the movements characteristic of the bird(de Diego 1973: 67). This is not the occasion for reviewing all the evidence which convinces one of the reality of 'sound-symbolism' as a major force in word-formation in any single language and operating across languages. However it is interesting to recall Piaget's observation that before they are taught to accept a conventional view of language, children of 5 or 6 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 1973: 79-89), and Darwin's strongly held views that at earliest times there must have been an intimate connection between sound and language with the sounds singularly adapted to subject; he assumed that language must have commenced in some necessary connection between things and voice(Darwin 1971).
The relation between what we see, static or in movement, and our own bodily and brain state, our posture which is the foundation of motor programs, and neural patterning, is of direct relevance. Many years ago I remember seeing an Italian film, La Strada, with an actress Giulietta Masina. At one point, standing alone in the Italian countryside, she sees a dead tree, its branches black and angular. Looking at it she seems to become the tree, her arms and body adopting the pattern of the tree. Similarly the poet Keats observed that watching a bird outside the window he felt himself become the bird pecking about in the gravel. We see things and often understand them by temporarily patterning some part of ourselves on the thing seen. There is a muscular, motor patterning element in our visual perception of things and this may provide the explanation for the manner in which words are formed so that their structures are in some way appropriate to the thing seen. The object seen produces a motor-pattern which is readily transferable as a motor-program to the articulatory system and so becomes the associated word for the thing. This is a very similar process to that described earlier in the paper by which we transfer into our own neural organisation the motor-programs underlying the facial expression of others, smiling, yawning, frowning.
The form which the extension of the motor theory of the origin of language takes as applied to the lexicon, to the formation of words, is that the neuromuscular sequences which are the immediate motor programs underlying words are derived from the integration of the neural structures underlying perception in all its forms (visual, auditory, tactile etc.) and motor organisation. Granit comments that because our movements are scaled to an invariant world, it should be possible to detect in the cortex cells that somehow have the property of co-ordinating the motor and sensory spheres(Granit 1977: 182). Young suggests that the programs for seeing each object or type of object probably consists of a routine that dictates a series of operations, guided by subroutines as expectations are examined. The neural 'maps' formed are not literal topographical plans but schemes of action(Young 1978: 129). Similarly other researchers have stressed the intimate relation between the neural organisation of perception and action. "Visual perception and the plans for voluntary action are so intimately bound together that they might be considered products of one cerebral function"(Trevarthen 1968: ); perception is essentially action-orientated and the internal models essentially predictive; the organisation of behaviour is the inverse of the construction of percepts(Pribram 1971); the "co-ordinated motor activity which constitutes the behavioural response to significant patterns of sensory information, can be regarded as the result of the activity of a pattern-generating mechanism which is the counterpart of the sensory mechanisms for pattern recognition"(Roberts 1978: 291). A number of researchers have stressed the very important and helpful consequences flowing from these views of the relation between perception and motor organisation. Gregory has emphasised that the brain picks out features of the pattern (in visual perception) that combine with its internal hypotheses to provide programs of action(Gregory cited in Young 1978: 123). "The assumption that the last stage of the perceptual process and the first stage of the motor process are one and the same is attractive because it solves the problem of imitation ... the process of perceiving a movement culminates in the creation of precisely those neural signals that are needed to direct a corresponding voluntary movement"(Gallistel 1980:369). "The subparts of the perception and action systems are thought of as pieces of a jigsaw puzzle that are made to fit each other ... The notion of perceiving and action as dual representations of common neural events may be a reasonable alternative to the sensory and motor views of mind ... If perception may be organised in a way that complements the organisation of the motor apparatus in the performance of acts 'such complementation is a further step in the direction of pruning the duties of the homunculus'(Fitch, Tuller & Turvey 1982: 279-280).
The final issue in this paper on the motor theory of the origin of language (and the motor theory of language function) is where, in terms of the theory, does syntax come from? What was the origin, in simple terms, of the way in which distinct words are fitted together to form a more complex meaning, the sentence-meaning? At one extreme, Bickerton(1981) would argue that the whole of grammar is genetically encoded and that we gradually acquire the grammar of our language by a process of development, the unfolding of the genetic endowment, in much the same way as an organ such as the heart is elaborated in embryogenesis. At the other extreme, there is the view that the grammars of different languages are purely cultural artefacts, that there is no inherent necessity in syntactic processes. The question where syntax comes from, or more specifically where sentences come from, is open, to some extent, to experimental investigation. Osgood conducted a series of experiments on precisely this point in 1968 by letting his subjects observe a variety of simple scenarios (eg. using different coloured balls rolling on a flat surface) and then asking the subjects to describe what they saw. The structure of the reporting sentences differed substantially from subject to subject (suggesting that the syntax was subsidiary to the differing sequences of observation of the subjects and depended on how they focused their attention) but Osgood concluded, in the light of his analysis of the relation between observation and sentence-structure, that the syntax was related very directly to the structure of pre-linguistic perceptuo-motor behaviour. "Pre-linguistic perceptuo-motor behavior has many (even if not all) of the characteristics of linguistic behavior ... perceptual and linguistic signs and sequences must, at some level, share a common set of organisational (syntactic) rules" most notably because of the demonstrated universal ability to paraphrase perceptuo-motor events in language. He concluded that what is shared by both sign-systems is not linguistic but cognitive in nature. This non-linguistic cognitive system is 'where sentences come from'"(Osgood 1971: 499).
Lieberman has, going beyond Osgood's approach, argued that "the rules of syntax derive from a generalisation of neural mechanisms that gradually evolved in the motor cortex to facilitate the automation of motor activity ... a generalisation of the automated schemas that first evolved in animals for motor control in tasks like respiration and walking"(Liberman 35). He refers to this as a grammar for motor activity. If, as is suggested in this paper, the elements of speech, phonemes, and the structures into which they are formed, words, both have a motor origin and indeed function as motor programs in current speech, then the close relation between the organisation of motor activity, motor syntax, and the organisation of language, speech syntax, would seem inevitable, though of course one still needs to consider how it is that the syntax used may vary not only between languages but also between individuals.
Lashley discussed the close relation between motor syntax and speech syntax and, after dismissing as untenable the chain-reaction theory of motor action (associational Markov-chaining) he observed: "the remaining alternative is that the mechanism which determines the serial activation of the motor units is relatively independent, both of the motor units and of the thought structure... Syntax is not inherent in the words employed or ideas to be expressed. It is a generalised pattern imposed upon the specific acts as they occur ... There are indications that, prior to the internal or overt enunciation of the sentence, an aggregate of word units is partially activated or readied ... Not only speech but all skilled acts seem to involve the same problems of serial ordering, even down to the temporal co-ordination of muscular contractions in such a movement as reaching and grasping. It is possible to designate, that is to point to, specific examples of the phenomena of the syntax of movement ... the syntax of the act, which can be described as the habitual order or mode of relating the expressive elements: a generalised pattern or schema of integration which may be imposed on a wide range and a wide variety of specific acts. This is the essential problem of serial order: the existence of generalised schemata of action which determine the sequence of specific acts, which in themselves or their associations seem to have no temporal valence"(Lashley 1951).
The motor theory of the origin and development of language presented in this paper is also in substance a motor theory of the current functioning of language. A theory of this kind fits well with the current trend of research into neural motor control and the neural basis of perception. It also has points in common with what Pribram described as his central motor theory of the origins of human language, based on a close relation between the imaging of action, perception and speech(Pribram 1971: 369). It is built fairly directly on Karl Lashley's ideas on the underlying uniformity of the neural organisation of action and language, though it inverts his approach; where he started from the then current analysis of the hierarchical structure of language, the present approach takes motor programming as primary and derives the structure of language, as a motor phenomenon, from the necessary processes in the organisation of action, as demonstrated in research into neural control of behaviour patterns in a range of experimental animals. Curiously, this theory could be presented as a return, at a deeper level, to earlier ideas on the essentially motor basis of brain processes, though Watson of course had an over-simplified and incorrect view of the real complexities involved in motor organisation. Two final reflections: "language is the immediate actuality of thought" (Marx and Engels) becomes true if thought is the interweaving of neural processes underlying perception and the formation of motor programs. "Language is action" if beside the speech-elements (phonemes), the speech-element compounds (words) and speech sequences (syntax) one can set a motor-alphabet (of elementary motor programs for bodily action), motor-words (actions formed from motor-elements) and motor-sentences (formed from sequences of motor-words).
Allott, R.M. 1973. The Physical Foundation of Language. Seaford: ELB.
Bernstein, N. 1967. The Co-ordination and Regulation of Movements. Oxford: Pergamon
Bickerton, B. 1981. Roots of Language. Ann Arbor: Karoma.
Bizzi, E. 1980. Central and peripheral mechanisms in motor control. In Tutorials in Motor Behavior. ed. J.E. Stelmach & J. Requin, 131-143.Amsterdam: North Holland
Bizzi, E. 1983. Central processes involved in arm movement control. In The production of speech. ed. P, MacNeilage, 3-10. New York:Springer-Verlag
Brown, R. 1958. Words and Things. New York: The Free Press.
Burdick, C.K. & J.B. Miller. 1975. Speech perception by the chinchilla. Journal of the Acoustic Society of America. 58: 415-427
Darwin, C. 1859. The Origin of Species. with Introduction by L.H. Matthews (1971). London: J.M. Dent.
Darwin, C. 1871. The Descent of Man Reprint with introduction by J.T. Bonner & R.M. May. Princeton, New Jersey: Princeton University Press.
Darwin, C. 1965. The Expression of the Emotions in Man and Animals. With a Preface by Konrad Lorenz. Chicago: University of Chicago Press.
De Diego, V.G. 1973. Lecciones de Linguistica Espanola. Madrid: Editorial Gredos.
Delafresnaye, J.F. ed. 1954. Brain Mechanisms and Consciousness. Oxford: Blackwell
Dewson, J.H., K.H. Pribram & J.C. Lynch. 1969. Effects of ablations of the temporal cortex upon speech sound discrimination in the monkey. Experimental Neurology. 24: 579-591
Eimas, P.D. et al. 1971. Speech perception in infants. Science. 171: 303-306
Evarts, E.V. 1973. Brain Mechanisms in Movement. Scientific American.229: 96-103
Evarts, E.V. 1979. Brain Mechanisms of Movement. Scientific American 241:146-157
Firth, J.R. 1964. The Tongues of Men and Speech. London: Oxford University Press.
Fitch, H.L., B. Tuller & M.T. Turvey. 1982. The Bernstein Perspective. In Human Motor Behavior ed. by J.A.S. Kelso, 271-282. Hillsdale, N.J.: Erlbaum Associates
Fowler, C.A., P.Rubin, R.E. Remez & M.T. Turvey. 1980. Implications for speech production of a general theory of action. In Language Production Vol. I: Speech and talk. ed. by B. Butterworth, 373-420. London: Academic Press
Gallistel, C.R. 1980. The organisation of action: A new synthesis.Hillsdale, N.J.: Erlbaum Associates
Getting, P.A. & M.A. Dekin. 1985. TRITONIA swimming. A model system for integration within rhythmic motor systems. In Model Networksand Behavior. ed. A.I. Selverston, 3-20
Granit, R. 1977. The Purposive Brain. Cambridge, Mass.: MIT Press
Greenough, G. 1984. Synapse formation as a component of mammalian long-term memory. Brain and Behavior 1984: 229-233
Gregory, R.L. 1972. Eye and Brain. London: Weidenfeld & Nicolson.
Gregory, R.L. 1976. Concepts and mechanisms of perception. London: Duckworth.
Hoyle, G. 1983. Muscles and their neural control. New York: John Wiley. Huxley, A. 1980. Reflections on Muscle. Sherrington Lecture. Liverpool:Liverpool University Press
Kelso, J.A.S. & J.E. Clark eds. 1982. The development of movement control and coordination. New York: Wiley
Kertesz, A. & P. Hooper. 1982. Praxis and language. Neuropsychologia.20: 275-286
Kimura, D. 1973. Manual activity during speaking. Neuropsychologia 11: 45-55
Kimura, D. 1976. The neural basis of language qua gesture. In Studies in Neurolinguistics. ed. H. Whitaker, 145-156. New York:Academic Press
Kimura, M. 1983. The Neutral Theory of Molecular Evolution. Cambridge: Cambridge University Press
Kinsbourne, M. ed. 1978. Assymetrical function of the brain.Cambridge: Cambridge University Press
Kohler, W. 1964. Psychologie de la forme. Paris: Gallimard.
Kuffler, S.W. & J.G. Nicholls. 1976. From Neuron to Brain. Sunderland, Mass.: Sinauer Associates
Kuhl, P.K. & J.D. Miller. 1975. Speech perception by the chinchilla.Science 190: 69-72
Lashley, K.S. 1951. The problem of serial order in behavior. In Cerebral Mechanisms in Behavior. ed. L.A. Jeffress 112-135. New York: Hafner
Lashley, K.S. 1954. Dynamic processes in perception. In Brain mechanisms and Consciousness. ed. J.F. Delafresnaye, 422-443. Oxford: Blackwell
Lee, D.N. 1980. Visuo-motor co-ordination in space-time. In Tutorials in Motor Behavior ed. by G.E. Stelmach and J. Requin, 281-296. Amsterdam: North Holland
Liberman, A.M. 1976. Discussion paper. In Annals of New YorkAcademy of Sciences. Vol 280: 718-724
Liberman, A.M. et al. 1967. Perception of the speech code.Psychological Review 74: 431-461
Lieberman, P. 1984. The biology and evolution of language. Cambridge, Mass.: Harvard University Press. Comment on review in J. Acoustic Society of America 80: 1521-1522
Lindblom, B.F. 1963. Journal of the Acoustic Society of America.35: 1773-1781.
Lindblom, B.F. 1983. Economy of speech gestures. In The Production of Speech ed. P. MacNeilage, 217-245
MacNeilage, P. ed. 1983. The Production of Speech. New York: Springer-Verlag.
McNeill, D. 1979. The Conceptual Basis of Language. Hillsdale, N.J.: Erlbaum Associates.
McNeill, D. 1981. Action, thought and language. Cognition 10: 201-208. Millikan, C.H. & F.L. Darley eds. 1967. Brain Mechanisms underlying Speech and Language. New York: Grune & Stratton
Milner, P.M. 1970. Physiological Psychology. London: Holt, Rinehart & Winston
Morse, P.A. 1976. Speech perception in the human infant and the Rhesus monkey. In Annals of the New York Academy of Sciences Vol. 280: 694-707
Ojemann, G. & C. Mateer. 1979. "Human language cortex: Identification of common sites for sequencing motor activity and speech". In Hearing
Mechanisms and Speech. ed. U. Creutzfeldt, H. Schreich and C. Schreiner, pp. 192-220. Berlin: Springer=Verlag.
Osgood, C.E. 1971. "Where do sentences come from?". In Semantics: An inter-disciplinary reader. ed. D.D. Steinberg & L.A. Jakobovits, pp. 497-529. Cambridge: Cambridge University Press.
Piaget, J. 1973. The Child's Conception of the World. Trans. by J. & A. Tomlinson. St. Albans: Paladin.
Pribram, K.H. 1971. Languages of the brain. Englewood Cliffs, New Jersey: Prentice-Hall
Roberts, T.D.M. 1978. Neurophysiology of Postural Mechanisms. London:Butterworth
Robertson, R.M. & K.G. Pearson. 1985. Neural networks controlling locomotion in locusts. In Model networks and behavior. ed.A.I. Selverston, 21-45
Selverston, A.I. ed. 1985. Model networks and behavior. New York: Plenum
Serebrennikov, B.A. 1980. Reducibility of languages of the world. In Essays on Linguistics, Language Systems and Structures, 1-56.Moscow: Progress Publishers
Sinnott, J,M, 1974. A comparison of speech sound discrimination in humans and monkeys. Doctoral dissertation. University of Michigan,Ann Arbor, Michigan
Sommerhoff, G. 1974. Logic of the Living Brain. London: John Wiley
Swinney, D. 1981. The process of language comprehension. Cognition.10: 307-312
Szentagothai, J. & M.A. Arbib. 1975. Conceptual models of neural organisation. Cambridge, Mass.: M.I.T
Turvey, M.T. 1977. Preliminaries to a theory of action with reference to vision.In Perceiving, acting and knowing. ed. R. Shaw & J. Bransford, 211-265. Hillsdale, N.J.: Erlbaum Associates
Usnadze, D. 1924. Ein experimentelle Beitrag zum Probleme des psychologischen Grundlagen der Namengebung. Psychol. Forschungen, 5: 24-43.
Warren. R.M. 1976. Auditory perception and speech evolution. In Annals of New York Academy of Sciences Vol. 280: 708-717
Watson J.B. 1925. Behaviorism. London: Kegan Paul.
Wissemann, H. 1954. Untersuchungen zur Onomatopoiie. Heidelberg: Carl Winter Universitatsverlag.
Young, J.Z. 1978. Programs of the brain. London: Oxford
University Press.