Language and Evolution: Homepage Robin Allott


Reconciling motor theory with the diversity of languages

The theory is that the structures (phonology, lexicon and syntax) of language in general, and of particular languages, are directly modelled on the neural systems for controlling motor action. But languages differ widely from each other in these aspects said to be modelled on the neural motor system. There are many different phonemic systems (the distinctive speech sounds used in different languages, including phonemic use of tones), obviously very large differences in lexicons (the words used in different languages), and equally large differences in syntaxes (word order, the use of inflections, the formation of compound words, agreement between different parts of speech, gender). If all languages are said to have the same origin, that is, by modelling on neural motor systems, how is it that languages differ so much from each other?

The first step in attempting to reconcile the motor theory with observed diversity of language is to examine the nature of the diversity. What is the extent and nature of the differences between languages. What is the orthodox or most common explanation of the differences between present-day languages? How satisfactory are the existing explanations? We have to recognise that the diversity of languages must come from somewhere, either from their first origin (whatever that may have been) or if their first origin was single and uniform from processes of change in language, which led to the multiplication of different languages, to arrive at the state that we see today.

Nature of the diversity

Languages are strikingly similar as well as strikingly different. They are similar in their very general structure:

a. the speech sounds of any single language are selected from a restricted set; the speech sounds of all languages can be analysed in uniform ways in terms of the anatomy and functioning of the vocal organs; for most speech sounds (possible exceptions tones and clicks) people belonging to different language communities can produce good enough examples even of unfamiliar sounds.

b. as regards lexicon, whilst each language has a distinctive collection of words peculiar to it, there is extensive similarity of words between languages, a very great amount of borrowing of words between languages, and of course great similarity between the things, actions, etc, for which words are available in different languages. A large part of traditional historical linguistics has been concerned with the similarities in lexicon between languages

c. grammatical structure has something in common in all known languages, particularly at the deeper levels of grammar. All known languages have words or word-like elements combined in accordance with rules into sentences; all known languages distinguish in some way nounlike and verblike sentence components. As regards syntax more specifically, there is variation only within certain broad types, and languages have been classified in terms of their syntactic similarities. Languages which may be widely different in their lexicons and phonology may have syntactic systems which are similar, e.g. Chinese and English.

But although there are these considerable similarities, particularly phonological, which themselves call for explanation, the differences between present-day languages are even more considerable, particularly in lexicon, and demand an explanation, whether in traditional terms or in relation to any biological theory of language origin (such as the motor theory).

How are the differences currently explained? One line of explanation is that there was only one original language (monogenesis), and all other languages, however different they now appear, were derived from it. On this account, differences in phonology and syntax between different languages resulted from processes of language change over immense periods of time; even if lines of descent and relation are difficult to construct or hypothesise for many modern languages, language change must be the explanation, and the residual unexplained differences represent the end results of language 'evolution' (in a general sense) so extended that the links can no longer be found between different languages. Note that this line of explanation, that all languages are related by descent and change, has been applied most vigorously to phonology and syntax, and that the difficulty is greater in suggesting how the lexicons of different languages can have diverged so much. For differences between lexicons, an ancillary explanation often put forward is that the words individual languages have chosen to represent experience of the world are arbitrary, that is, they are mere chance patterns of sound which at some stage in a language's development from its original source have been adopted by the group or community speaking the particular language.

Another explanation - or perhaps it should be described as the ultimate pessimistic account of language diversity - is that everything in language is in origin arbitrary; the words found in the lexicon, the syntax - systems of inflection, gender, etc. - and the choice of phonemic sets in particular languages is totally arbitrary. Whilst some relationships between individual languages are recognised, e.g. Indo-European, apart from a few broad families like this everything is the result of chance, not the product of order. This view probably accepts that components of a language originally totally arbitrary in origin can undergo more systematic processes of language change, e.g. the transmutation of Latin CANIS into French CHIEN, but once the root form has been found in the particular protolanguage, there is no point in attempting to go any further back to explain why the particular protoform was adopted.

How satisfactory are these more orthodox explanations for language diversity? The second explanation, the arbitrary character of differences between languages, really amounts to no explanation. It says no more than that there are differences and no account can be given of the causation of the differences. Any assertion that some event or process, in linguistics or any other field of science, is arbitrary, provokes the question: How do you know that it is arbitrary; how do you know that there was no chain of causation; how have you been able to exclude all possible causative explanations? Please give some example of the arbitrary selection or creation of a particular word in a particular language, of an inflection, of a syntactic rule, e.g. that a given preposition takes a given case, of selection of a particular phoneme in a given language.

The more usual explanation, that diversity of current languages is the result of processes of language change affecting an earlier language or earlier languages, is also less satisfactory than it sounds. It asserts that languages change - as obviously they must have; it suggests that certain types of change have taken place, e.g. phonological, lexical or syntactic, but it does not explain the process of change; it simply relates present diversity to past diversification. Language and languages have changed (so far as there are records of earlier forms of languages) but why have languages changed? How have languages changed? What is the process of change which produces the diversity? What is the source of the specific change?

On the assumption of monogenesis, the current diversity of languages coupled with the diversity of languages over time (languages have fallen out of use and apparently earlier forms of existing languages have been replaced) reduces to the question of the nature, causes, direction, of language change.

The alternative to monogenesis for the origin of language is polygenesis, i.e.that language originated in human evolution at a number of distinct points. If polygenesis is assumed, then the different protolanguages would have developed independently and led over millennia to the diversity of the descendant languages found at the present day. However, the assumption of polygenesis only thrusts the problem of language diversity further back. Why should there have been different protolanguages? By what process were the different protolanguages formed? How was selection made of the various phonemic systems, lexicons and syntaxes constituting the protolanguages and distinguishing them from each other? And, of course, even with polygenesis as one's assumption, the question of language change over time still remains, since the multiple protolanguages clearly developed and changed to produce the diversity of their descendant languages now found.

The problem of language change

If we could explain what drives and directs language change, then we could start to explain language diversity, present and past. This explanation could be either in the context of orthodox linguistics or in relation to a neurological/physiological (essentially biological) account of language origin. The next step is to look at what is known about language change and what has been suggested as the causation and process of language change. To fix our thoughts, some extracts are introduced here (in order of date) from a number of linguists on the issue of language change:

What supports the necessity for change? ... How do differences that result in the most varied dialectal forms originate? What pattern does their evolution follow?... An explanation that has been favoured for several years attributes change in pronunciation to our phonetic education during childhood ... What prompts a generation to retain certain mistakes to the exclusion of others that are just as natural is not clear. From all appearances the choice of faulty pronunciation is completely arbitrary, and there is no obvious reason for it. Besides, why did the phenomenon break through at one time rather than another? ... [Alternatively, linguists have suggested a variety of general causes for language change] Climatic influence, racial predisposition, tendency towards least effort are all permanent or lasting. Why is it that they act sporadically, sometimes on one point of the phonological system and sometimes on another? A historical event must have a determining cause, yet we are not told what chances in each instance to unleash a change whose general cause has existed for a long time. This is the most difficult point to explain. The blind nature of the evolution of sounds... Time changes all things ; there is no reason why language should escape this universal law. (Saussure 1966: 77, 149-150, 199, 231)

Are we not imputing to this history a certain mystical quality? Are we not giving to language a power to change of its own accord over and above the tendency of individuals to vary the norm? .. Language exists only in so far as it is actually used.. What significant changes take place in it must exist, to begin with, as individual variations .. random phenomena, like the waves of the sea, moving backward and forward in purposeless flux. The linguistic drift has direction.. The drift of a language is constituted by the unconscious selection on the part of its speakers of those individual variations that are cumulative in some special direction. (Sapir 1921: 154-155)

Every conceivable cause [for language change] has been alleged: 'race', climate, topographic conditions ... No permanent factor can account for specific changes .. theories of this kind are confuted by the fact that when sound-change has removed some phonetic feature, later sound-change may result in the renewal of just this feature... Although many sound-changes shorten linguistic forms, simplify the phonetic system or in some other way lessen the labor of utterance, yet no student has succeeded in establishing a correlation between sound-change and any antecedent phenomenon. The causes of sound-change are unknown. (Bloomfield 1933: 385-386)

The fundamental cause of linguistic change and hence of language diversification is the minute deviations occurring in the transmission of speech from one generation to another. These changes are mostly very gradual in their operation, becoming noticeable only cumulatively over the course of several generations. But, in some areas of vocabulary, particular words closely related to rapid cultural change are subject to equally rapid and therefore noticeable changes within a generation or even within a decade. Grammatical and phonological structures are relatively stable and change noticeably over centuries rather than decades. (Robins 1979: 652, 660-662)

For centuries, men have speculated about the causes of language change. The confusion and controversies surrounding causes of language change ... some reputable linguists have regarded the whole field as a disaster area, and opted out altogether. Children are unlikely to initiate change, since change is spread in social groups, and babies do not have sufficient influence to persuade other people to imitate them. Change is likely to be triggered by social factors, such as fashion, foreign influence, and social need. However, these factors cannot take effect unless the language is 'ready' for a particular change. They simply make use of inherent tendencies which reside in the physical and mental make-up of human beings. Continual language change is natural and inevitable ... there is no evidence that language is either progressing or decaying .. [or] that languages are moving in any particular direction from the point of view of language structure ... Finally, it is always possible that language is developing in some mysterious fashion that linguists have not yet identified. Only time and further research will tell. There is much more to be discovered. (Aitchison 1981: 111-112. 168, 190, 234)

These extracts show how puzzling most linguists have found the diversity of languages and the problem of language change. Language variation is the product of change, but the variation is much more extensive than simply the distinction between different languages: each language contains within it extensive variation in the form of dialect areas and dialect features, which show an even more complicated distribution than language diversity. Once one recognises that the variation is not only between languages but within languages, the problem becomes even greater, and the explanations even more tenuous and less satisfactory.

But if in traditional linguistic terms, the problems of language diversity and language variation receive no very convincing solutions, consider what the approach to these issues might be on a biological theory of language origin and function. Let us take some specific examples of language and dialect variation and change so as to be able to judge more concretely what the relevance of a biological approach might be:

Dialectal variation - Within a single language, there can be major dialectal variation in typology, syntax and morphology, as well as differences in lexicon. The position in Serbo-Croat and Slovene is an extreme example of this structural variation: dialects are found with synthetic declensions (case endings, as in Latin) and analytic declension (use of prepositions and word order, as in English). In addition, there are among these dialects complex systems of verbal tenses contrasting with simple ones, as well as dialects with or without the dual number or the neuter gender. The dialects of Serbo-Croatian and Slovene also exhibit almost every type of prosodic structure (e.g. tone, stress, length) found in European languages. Some dialects differentiate long and short vowels or rising and falling accents, while others do not; and in some, but not all of them, stress fulfils a grammatical function. Of the several dozen vowel and diphthong sounds that occur in these dialects, only five are common to all of them; all the rest are restricted to relatively small areas. (Ivic 1979)

Similarities across language frontiers - Languages which are spoken in the same general geographical region tend to have phonetic features in common even if they are not closely related historically. Examples: some assimilation in the sound systems of Spanish and Basque (Swadesh 1972); Bushman and Zulu both have clicks, even though racially and linguistically, the two are judged to be unrelated; on the northwest coast of North America most of the American Indian languages have a glottalised series of sounds and a proliferation of sounds produced in the back of the mouth, in addition to other similarities; Indo-Aryan languages have cerebral or retroflex speech-sounds unknown otherwise in their language group - these sounds are found in Dravidian (an unrelated language family, though spoken in the same area as the Indian languages); Swabian dialects of German are found with nasalised vowels quite unlike German generally; the dialects are spoken in proximity to French, which makes abundant use of nasalised vowels. How are we to explain these and hundreds of similar phonetic convergences?(Sapir 1921). Tonal languages are not only extremely widespread through Southeast Asia and China, they also show remarkable similarities to one another ... the tonal systems of Chinese, Miao-Yao and Tai have developed in almost identical fashion. (Norman 1988)

Diversity associated with different groupings but also cutting across groupings - Similarities across language frontiers are not restricted to phonology. Isoglosses can be drawn for many linguistic features which bear no relation to language frontiers. The best known linguistic area in Europe is undoubtedly the Balkans. The languages spoken in this area, particularly Romanian, Bulgarian, Macedonian, Albanian and Greek, are not for the most part closely related but nevertheless show striking resemblances to one another in many respects. The most striking example of the similarity is the fact that four of the languages, Albanian, Bulgarian, Macedonian and Romanian, all have a postposed definite article. This feature does not occur in [non-Balkan] languages which are historically related to the Balkan languages (Chambers and Trudgill 1980: 184-185). There does not seem any adequate linguistic explanation for this sort of distribution which cuts across the accepted groupings of languages.

Australian and Papuan languages: These are of special interest because (particularly in the case of the Australian languages) they have undergone virtually no external influences for tens of thousands of years. The following account of the Australian languages is drawn from a number of sources (Dixon 1980, Kaldor 1982, Wurm 1972, Vazsolyi 1976, O'Grady et al. 1966). There are some 200 Australian languages spoken by about 600 tribes. No relationships have been shown to exist between Australian languages and any outside languages. All but 2 or 3 of the languages belong to one language family (Pama-Nyungan) which occupied 7/9ths of the entire land area. Each tribe has or had something of the order of 500 members on average. The multi-tribe Western Desert language community is the largest, with 6000 speakers spread over perhaps half a million square miles.

Phonology: Australian languages show remarkable similarity throughout the continent in the structuring of their speech sounds, a surprising phonemic uniformity, unparalleled anywhere else in the world. Most Aboriginal languages have only three vowels I A U close to Italian vowels. Diphthongs are non-existent or atypical. Most Australian languages do not distinguish consonantal sounds from each other on the basis of voicing .The effect of this (coupled with the restricted set of vowels) is that, for example, speakers of Western Desert languages have difficulty in distinguishing BIT BEET PIT PET PEET BID BED BEAD. A large number of consonants are absent. They do not have F "fish", V "van", TH "thin", TH "this", S "sun", SH "shoe", Z "zero", ZH "pleasure", CH "child", DZ "joke", H "hat". No (or very rare) consonant clusters, e.g. "kin" for "skin". Typically they have an extended range of stop consonants and nasals. A striking uniformity in the languages is that the number of linear distinctions made among oral stops in a given language is identical with the number of linear distinctions made among nasals. The few phonologically aberrant languages in the North are the result of external influences, probably from New Guinea, which also manifest themselves in racial features in the same areas. Syntax: Most of the languages are suffixing but there is a smaller group of prefixing languages. The suffixing languages show long strings of suffixes, agglutinative in type. The prefixing languages form a geographically solid block. Word order is very flexible with the generally preferred order being SOV. Adjectives often follow nouns. A typical syntactic feature of Australian languages is the presence of ergative case markers. The ergative marker expresses the exclusive function 'subject of a transitive sentence' (but not 'subject of an intransitive sentence').

Lexicon: Except for pronouns which are similar across the continent and the limitation that there are only two or three numerals, the most striking lexical feature of Australian languages is the fewness of cognates between languages which otherwise seem closely related. Two languages may share grammatical properties and similarities in speech sounds but hardly share any vocabulary items. A curious feature is that a few cognates have been found to be shared by languages at opposite ends of the continent with no intervening cognates.

Linguists comment that something special seems to be going on in Australia which presents a challenge to accepted methods in comparative linguistics. There seems to be a contradiction between the results of shared vocabulary counts and the results of comparison of other aspects of language viz: that of speech sounds and of abstract grammatical/semantic properties. This contrasts not only with other language groupings in the same general geographical region eg. Austronesian (where phonology, syntax and lexicon appear to have a more systematic relation) but also very strikingly with the situation in Papua/New Guinea. There one also finds a great diversity of languages for a small population. However, in New Guinea the languages differ phonologically, lexically and grammatically from each other.

Papua/New Guinea is said (Foley 1986) to be the linguistically most complex area in the world. There are around 750 Papuan languages (nearly 20% of the world's total). The Papuan languages are not all genetically related but belong to at least sixty different language families; no Papuan language family has been demonstrated to have any genetic affiliation outside the immediate New Guinea area. Given the relatively small area, and the small total population (approximately four million), they present an unparalleled picture of linguistic diversity. The Papuan languages provide a different but even greater challenge than does the Australian language situation, for accepted approaches to the problems of language diversity and language change.

Biological change as a model

In historical linguistics, terms drawn from biology have been freely used. Linguists speak about the genetic relationships of language, about language descent, language families and language drift. Some linguists have tended to look upon an individual language as an organism: the evolution of language is a frequently used concept;some linguists talk about the 'life' of particular languages: "the invisible and impersonal drift that is the life of language" (Sapir 1921: 171), their readiness to change in certain directions and even their fitness to survive. One of the criticisms directed against the comparative method is that it is based upon a misleading genealogical metaphor: "There is obviously no point in time at which it can be said that new languages are 'born' of a common parent language.. It is easy enough to recognise the inappropriateness of these biological expressions".(Lyons 1979: 1009)

The trouble comes from adopting a limited and wrong analogy from biology. Rightly used, biology in its modern form provides an instructive model or analogy for language and an account which can be directly explanatory of the phenomena of language. This was most fully developed by Stevick (1963): The new biological model is different from the older one that we properly reject. .. Biological models for variation in the rate of linguistic change are to be found ... in the study of population structure. .. If complexity is a criterion, evolutionary biology more than qualifies as a model for language history ... Linguistic and biological histories are closely similar particular examples of the general evolutionary model.(160-162, 168-169) He demonstrates this convincingly by paraphrasing key sections of Dobzhansky (1951) substituting language and language-related terms for Dobzhansky's references to species, races, populations etc. Both language and species continue, modify, and diversify. Hence both language and biology distinguish variation and change.

Population genetics can provide not only a model or an analogy but also an explanation for language variation and language change. To see how this is possible it is necessary to give some brief account of the main themes in the biology of populations. The issues in population genetics, as in language, are those of variation and change.

Biological variation -

Individual variation: Individuals within any population usually show variations in almost all characteristics. The differences that exist between individuals in a population are so many that any individual is likely to be distinct and possibly unique.

Group (deme) variation: Population sub-division seems to be almost as common as individual variability. There is considerable evidence that most populations can be subdivided into semi-isolated demes.

Population or community variation: A population is a collection of individuals that can freely interbreed. It is a cohesive genetic unit within which alleles are freely interchanged. It is often referred to as a collective gene pool out of which alleles are drawn at random during each reproduction. A much higher level of heterogeneity occurs in natural populations than used to be thought possible.

Race variation: The notion of race is statistical and describes the characteristics of populations. A race is a breeding population characterised by frequencies of a collection of inherited traits that differ from those of other populations of the same species. Genetically, races differ with respect to allele frequencies rather than in the possession of unique alleles. Natural races tend to be separated by intergrading zones rather than by sharp lines of demarcation.

Species variation: A species typically consists of a number of partially overlapping populations showing various intergradations of morphological or physiological characters. Recent applications of recombinant DNA techniques have uncovered much more variation than previously imagined both within and between species.

Biological change -

The process of evolution consists essentially in transforming individual variation within species into variation between species. Variations are maintained in the population by local demic differentiation. Because the demes are small, genetic drift contributes significantly to altering allele frequencies between demes. Small populations will be subject to large fluctuations. A series of small populations formed from a single large one will inevitably diverge, even if they were initially identical.

Genetic drift is the effect of the sampling process that occurs in reproduction. A special case of the sampling effect occurs when a small number of individuals from an originally large population become founders of a new population. The allele frequencies of the new population will be those of the founders, which may or may not be representative of the original population. Founder effects are often invoked as an explanation for differences in allele frequencies among different human ethnic groups.

New species may arise subsequent to the interruption of gene flow in part of an original species' range. There are many ways in which gene flow may be interrupted. The most widely recognised is that of geographic isolation due to physical barriers such as mountain ranges, rivers,the sea. Some geneticists place a great deal of emphasis on the founder effect and genetic drift in small populations as mechanisms for producing rapid genetic differentiation and hence the formation of new species. Intrinsic isolating mechanisms such as the formation of social groups in many mammals, including humans, tend to reduce genetic exchanges between groups and to encourage genetic differentiation. There is a special problem of the amount of racial variation that exists in modern human populations. Berry (1977) suggests that hunting groups must have had all the possibilities of differing widely from each other through the repeated operation of the founder effect and the greater variety of habitats to which they would have had to respond in different parts of their range.

Biological change as an explanation

The essence of the motor theory is that each main aspect of language, the phonological, the lexical and syntactic, depends upon the integration of action and perception organisation with language, that phonemes, words and syntactic rules are derived from motor organisation in specific ways. That is, language directly reflects or expresses neural organisation, the system of neuronal connections which constitute motor programs and result in articulatory movements. But the reality of neural organisation is that each brain differs from every other brain, the patterns of neural connection vary from individual to individual; not only this but the conformation of vocal organs varies from individual to individual, the musculature varies, and obviously enough the quality of the voice which results varies from individual to individual. These differences between individuals are the product of differences in the genetic control operating in development and interacting with the environment.

From population genetics, it is clear that every individual is genetically unique. And also that, on any technique of measurement, genetic variation is great within populations. Any population whilst embracing a great degree of variability also has uniformities which characterise the gene pool of the population. These uniformities exert a uniform (statistically averaged) effect over the physical and mental constitution, brain and body development of the population. They mean that certain features of brain and body development are found more frequently in one population than in another. The obvious extent of morphological differences between populations and between races makes it impossible to believe that non-observable features of sensory and cerebral organisation will not also reflect differing gene frequencies (Baker 1974). In population genetics, the propositions of individual variability within populations and variation in gene frequencies between populations and even more between races. are generally accepted. There are also well-established accounts of how, given individual variation within a population, factors can operate to increase and consolidate differences between populations and between subgroups within the populations. These factors take the form of migration, geographic separation, founder effects, environmental forces (providing the setting for expression of genes in the gene pool) and so on. Population genetics thus indicates how change in cerebral and bodily organisation, brain wiring and morphological features (particularly those affecting the organs of articulation) can occur. The whole account in population genetics of the processes of race and species formation bears a close resemblance to the accounts given by linguists of the processes by which linguistic variation within a community can develop into differentiation of dialects and later of separate languages, by very similar processes of geographic separation, migration, isolation - with reproductive isolation in the case of the development of races and species being paralleled by communicative isolation in the case of dialects and languages.

The next step in the reconciliation is that in the same way as the expression of the genome in bodily development is the product of interaction with the environment, so the development of a language is the result of interaction between the genetically controlled cerebral, anatomical and physiological features of the population (reflecting the relative gene frequencies within the population) and the environment within which the development of the language takes place. The importance of the environment derives from the reality that language cannot be an individual-centred capacity but has to be an inter-individual or group product. Such a group product will be biassed towards the aspects of language most compatible with the cerebral and bodily articulatory organisation prevailing in the group within which the language is to be used. A relation will be apparent between the phonemic system adopted in the group and the physical conformation most prevalent within the group. So far as basic and original lexicon is derived from integration of articulation with perceptual and motor organisation, the words found in the group will have their structure determined (within the limits set by the phonology preferentially adopted) by neural organisation governing perception and the planning of motor action, Similarly, in the origin of a language the structure of the motor system will have a dominant influence on the basic syntactic features of the language of the group.

Given conditions prevailing throughout most of human history (barriers to communication, small and scattered populations etc), this account of the link between gene frequencies in the group and the characteristics of language in the group leads easily on to explanation of processes of dialect development, language splitting, language change etc. Dialectal variations will tend to develop as a result of changes in controlling gene frequencies within a local area, or the transmission of gene variations by intermarriage across dialect divisions. By the same process, genetic effects on articulatory morphology, and cerebral motor organisation, can explain the tendency for phonological characteristics to be spread across language borders, such as the cerebrals found in both Dravidian and Indo-Aryan languages, and many other examples. Similarly, over time within any language community, the characteristics of the language, its sound-complement, its lexicon and syntax, will tend to reflect, though with a considerable drag, changes in the gene frequencies for factors controlling articulatory morphology and cerebral organisation for perception and movement. Gene frequencies may change within a population for a variety of reasons, most obviously as a result of inward migration from other communities with different gene frequencies, as a result of change in the relative size of different subgroups within the population, and also as a result in smaller populations of fluctuations in gene frequencies which occur as a sampling effect. This link between changing gene frequencies and changing, particularly phonological, features can explain the apparently erratic character of some sound changes, where features disappear from a language and later reappear.

But how can changing gene frequencies, affecting articulatory morphology and cerebral organisation, actually have an impact on the characteristics of a language which already exists and appears to be preserved by its function of being a uniform medium of communication within the group? The relevant changes in gene frequencies may be taking place, must be taking place, all the time, but at what point does language also change to reflect them? Part of the problem is linguists' idealised conception of an individual language as a uniform system spread over a language community. In fact, any language as actually used in a community as a spoken medium (reliance on written versions of a language can be misleading) is an immense collection of local, occupational, class, register, variations. A language changes when some variation becomes more widely used and others become less widely used. So the question reduces partly to how the range of variation is generated within the community, and how change in gene frequencies affecting articulatory and cerebral organisation is translated into language variation.

This leads one back to the development of language by children. Regardless of their genetic inheritance, children learn the language of the community in which they grow up. Nevertheless changes in gene frequencies affecting articulation and cerebral organisation for perception and motor control translate most directly into changes in the physical and neural development of children. At this stage one needs a physiologically-based theory of the process of development (not just acquisition) of a language by children. A plausible account starts from research into the acquisition of behavioural characteristics by birds and other animals, of a kind where the behaviours acquired have to be related most appropriately to the environments in which the birds or animals find themselves and will spend their adult lives. The topics here are imprinting, the acquisition once and for all of a particular cerebral set, and the somewhat different idea of the critical period, the period of plasticity in cerebral and behavioural organisation that many young animals experience. Moulding in the critical period for a human infant, a bird or a monkey, is the result of interaction between the genetic endowment of the infant and the environment to which it is exposed. If the genetic endowment changes (as a result of changes in gene frequencies) then the product of interaction between that endowment and the environment will also necessarily be different. If gene frequencies alter significantly in a population over a period, then the tendencies and preferences of the children developing in that period will also alter. This will, insofar as the genes concerned are those affecting relevant physical or cerebral organisation for language, constitute a force tending towards language change. Even though children as such have no important influence on speech patterns and preferences, the tendencies and preferences changed in them as a result of changes in gene frequencies will continue to exist when they grow up into adults and it will be at the adult stage that the force deriving from population change in gene frequencies will start to have an impact on the phonological, lexical or syntactic features of the languages.

Supporting argument and evidence

Supporting argument and evidence can be presented at various levels. The broad plausibility of the idea of a genetic basis for language (in a strict biological sense) is the very obvious parallelism between the analysis and findings of population genetics and the phenomena of language differentiation, language variation, dialectal features, language spread, etc. The second broad line of argument derives from the motor theory as such, that is from the theory that there is a specific neural basis for language derived from pre-existing cerebral organisation.

Once it is accepted that cerebral and articulatory organisation have direct significance for language, and that 'genetic' factors in a strict biological sense have a controlling influence on these within and between populations, then there is ample more detailed evidence from mainline population genetics for the circumstances in which gene frequencies may change between populations, within populations and between individuals and groups within a population. Authoritative accounts of this are given eg. by Dobzhansky (1951), Berry (1977) and many others. That is, the variation in the relevant genetic aspects both over space and over time is available from which language patterns and changes may be determined. Differences between gene frequencies relevant for language would extend not only to physical aspects dealt with above but also to neural organisation.

More specific evidence bearing directly on variation within populations and between populations which can be relevant for the character and development of languages is drawn from various sources. There is evidence for differences between individuals both in brain organisation and in articulatory organisation. Relevant evidence can be presented under three heads: evidence for differences between individuals and between populations and races in aspects of brain and body relevant for speech and language; evidence of correlation between measurements of genetic distance and language differences between human communities, evidence bearing on the transmission of changes in language via the process by which children develop their language capacity.

Physical variation relevant for language diversification

Genetic distance and language groupings

There have been some studies of the extent to which there is correlation between language and genetic (biological) groupings. Given the linguistic peculiarities in Australia and Papua/New Guinea, results of research there are of special interest. In Papua/New Guinea, Bougainville was chosen for study because of the remarkable diversity of indigenous languages (19 languages for 45000 people). The researchers were interested in physical differences between the various groups in Bougainville and how they related to language relationships and patterns of intermarriage. They found that the morphological differences encountered were genetically rather than environmentally determined; the pattern the villages made in the discriminant analysis was a 'neat reflection' of language group discreteness, with consistent clustering of villages from the same language area. The role of the language group as morphological (physical conformation) unit was apparent (Friedlaender 1976).

Similar studies with Australian aboriginals produced rather similar results. The Australian aboriginal has physical features which differentiate him clearly from all other races but at the same time there is considerable regional variation and there is little doubt that physical differences exist between populations of Aborigines. The results of the analysis using a variety of measures of genetic distance (seven genetic marker systems including blood groupings) suggested that for large parts of the Australian continent, linguistic differentiation has been proceeding in a similar manner to genetic differentiation. Four tribes were studied in depth: statistical analysis showed genetic distances between the tribes similar to the order of relationships given by linguistic studies (Sanghvi et al. 1971, Balakrishnan et al. 1975).

This kind of work is still at an early stage. There can be argument about the best measures of genetic variation and genetic distance (blood-groupings, protein complexes or polygenic characters such as observable morphology); morphological variation seems more relevant than single-gene characters. Nevertheless, the findings so far are consistent with a direct relation between genetic diversity and language diversity in the particular areas studied. Interestingly, similar studies have been carried out on the relation between genetic variation and song variation in birds. The object was to see if any genetic variations coincided with a boundary between two dialects of birdsong. It was found that the dialects of the group of birds studied, a subspecies of white-crowned sparrow, formed a kind of genetic mosaic, with enough differentiation for each song-dialect area to constitute a deme or selective unit (Bright 1984).

Transmission of language change

The next issue is the suggestion that changes in population and deme gene frequencies take effect by way of their impact on children's development of their language and act through altering the biases, tendencies and preferences which those children subsequently experience when they become adults, and which at that time start to cause language change. The evidence relevant for this is drawn from general work on neural and behavioural development in infants and animals. The most useful evidence relates to birds. This is important for two major reasons: first, neural organisation is similar across the range of vertebrates; the basic elements, neurons, synapses, neurotransmitters, are the same, and if certain processes are found in birds or other vertebrates, then there is no reason why one should not look for parallel processes in human beings. The second reason is that birds are vocal animals like human beings, and if the neural control of vocalisation is found to take particular forms in birds (despite the different anatomy of articulation) then this has a direct relevance for human vocalisation. Research findings described in Bright (1984) offer a significant pointer: whilst the period of learning or moulding of vocalisation comes in a limited period of the bird's early life, the effects are only expressed when the bird becomes adult; the moulding of neural connections made in the infant period is necessarily carried through into the adult period. The suggestion that genetic changes in a human population are carried through into language in a similar way is speculative but plausible; Meillet (1937), VendryÖs (1902) and other linguists had speculated along much the same lines long before progress in neurology and ethology could offer any evidence. If in fact this is how language change occurs, then it overcomes the valid counter-argument of Saussure that language is not transmitted discontinuously from generation to generation, and also Aitchison's (1981) comment that children, as children, can exert no powerful influence on current language.

The process of dialectal differentiation has been described for birdsong along the following lines (Purves and Lichtman 1985 summarising work of Thorpe, Marler,Nottebohm, Konishi and others). Chaffinches are born with an innate but imperfect idea of what they are to sing but the fully formed song must be learned from conspecifics during a limited (critical) period in early life. When they reach sexual maturity, they proceed to reproduce this type of song. This results in considerable variations between the songs of different flocks of the same species: each has its own self-perpetuating dialect. Dialect-like differences on this view are an accidental by-product of song learning. A young bird, within the first year of life, hears other chaffinches singing. The following spring it starts to sing itself, matching its own output to its memory of what it has previously heard, producing near-perfect copies six to eight months later. In song learning birds make mistakes; as a result, song is almost certain to change, though slowly. Swamp sparrows in the phase of plastic song generate far more song material than is needed for normal adult song production; the excess song components get winnowed out until the mature repertoire is left. Much of the work on neural plasticity has been done with cats. The development of the visual system of cats also takes place during a critical period. The final organisation of pattern detecting neurons in the visual cortex of a cat is fundamentally determined by the kitten's early visual experience; the effect appears to be that the feature-detecting system is optimally matched to the animal's visual environment.

From these ethological studies a main conclusion is that various behaviours can only be learned during a restricted (critical) period, e.g. birdsong in some species must be learned during a limited time in the first season, and socialisation in monkeys must occur during early life or be forever abnormal. Neural connections can be modified only within a certain age range. The studies of birds, cats and monkeys are in accord with a growing body of neurobiological evidence that confirms that experience can modify innate behaviour in a more or less permanent way. One reason for the intense interest in these results is the notion that the purpose of critical periods is to allow the outside world to influence cortical connectivity. The goal of neural development is appropriate behaviour. Central patterns of neural organisation are present very early in embryonic development. Many receptive field properties are established prior to any experience. Neonatal neurons are less sharply tuned than their adult counterparts, but the basic wiring that subserves these various aspects of adult responsiveness is largely present at birth. Although many aspects of connectivity depend on intrinsic mechanisms, other aspects of connectivity remain malleable at birth. As a result of further postnatal maturation, experience can influence the final pattern of connections. (Purves and Lichtman 1985, Atkinson, Barlow and Braddick 1982)

Genetic basis for changes in phonology, lexicon and syntax


The neural and anatomical basis for change as a result of genetic factors is most plausible for phonology of a language, which is least likely to be affected by passing fashion. Radical changes in phonology occur rarely and when they do affect a wide swathe of the language, e.g. the German soundshifts and the English vowel shift. Nevertheless, the most frequent argument against any physiological basis for diversity or change in language has focused on phonology. Saussure's comment has been repeated with little variation by later linguists:

Does the phonational apparatus vary from one race to the next? No, scarcely more than from one individual to the next. A newborn Negro transplanted to France speaks French as well as a native Frenchman (Saussure 1966: 147).

The powerful objection to all hypotheses which posit a mutation in the vocal organs or a modification of the cerebral centres is the simple fact that young children or refugee children learn the language of adoption with no trace of imperfection (Lord 1966: 83).

There seem to be no physiological differences between the races which are reflected in speech. Although the Chinese and other Mongoloid peoples have differently shaped incisors, a Caucasoid child brought up in China among the Chinese will speak Chinese as a native speaker, even if both the child's parents spent their whole lives in Chicago (Davis 1983: 3).

Apart from the fact that there are anatomical and physiological differences between individuals and races capable of affecting the articulatory system, such arguments are based on views of race, inheritance and development which are out of date. Brosnahan deals with the point effectively:

It is the preoccupation with the individual which on the linguistic side has led to the obfuscation of the whole problem with largely irrelevant and non-cogent arguments, such as whether a given individual can or cannot learn perfectly a language other than that of his parents. Experience seems to show that this is quite possible but far from this being an argument against genetic factors in language, it would seem indeed only possible because all languages have been developed by human beings of basically similar genetic structure, and because human beings inherit a potential plasticity of development with regard to sound production and speech. (Brosnahan 1961: 30)

Meillet and Vendryes, both eminent historical linguists, saw no difficulty in linking phonological change with physiological change. So Vendryes (following Meillet):

Every sound change has a natural cause; the regularities in the transmission of sounds results from changes in the articulatory system; it is in the passage from one generation to another that an articulatory system is ordinarily modified. [Sound changes] depend on the physiological makeup of the brain and oral cavity. Every sound change can then be considered as the result of deep-seated and hidden forces, for which the term tendency is suitable enough. It is these tendencies that continually modify the structure of language ...They are linked intimately to the physiological aspect of the speech organs and determine, consequently, in a perfectly natural way, the normal evolution of language ... Each individual .. by the very fact that he differs from his contemporaries, carries the germ of new tendencies. Within each group, a kind of mean is established and the resemblance of the whole is sufficient, so that one can abstract the tendencies of detail... If the identity of tendencies is the cause of the homogeneity of dialects, inversely it is the difference of tendencies that creates the diversity among dialects (Vendryes 1902: 112,117).

This account of the matter is close to that put forward by Darlington (1947) which was developed later in detail by Brosnahan (1961). Jakobson also emphasised the biological aspect of phonology: the universal sound patterning of speech is "the most enrooted in the psychobiological nature of humans"; the sound patterns of single languages are "varying implementation of universal invariants".(Jakobson and Waugh 1987: 64, 236) More recently, Lieberman has commented, in a phonological context, that "the presence of genetically distinct subgroups in the human population who may differ with respect to some of the neural mechanisms that determine the perception of speech offers an explanation both for the diversity of human language and for linguistic change". (Lieberman 1986: 106)


Changes in gene frequencies within a population, a language community, a dialect group, do not explain sufficiently why new words are added to the lexicon, or old words drop out, or different words replace established words as and when they do. The rate at which the lexicon changes is faster than can plausibly be ascribed to changes in population gene frequencies. But there is an important point that should be made in this regard. Given that language and dialect variation reduce to language change, nevertheless it is not the case that all language change has to be treated as directly genetic in timing as well as in origin.

In the case of phonology and syntax, the implication of the motor theory is that neural and anatomical factors restrict the range of possible systems in any language and that diverging gene frequencies within language communities over time affect the balance of preference between the different possibilities, that is, the change in gene frequencies does not alter the total set of phonemes or the total set of syntactic patterns which are possible but the system of phonemes or syntactic patterns which are actually chosen by a particular community.

In the case of the lexicon, the approach is the same. Motor and perceptual organisation limit the range of words potentially appropriate for objects or actions and the lexicon of any language represents a selection made from the set of possibly appropriate words. There are two separate questions, how the range of possible words is limited - what is the generating mechanism for the forms which words for particular objects, actions, etc. may take. Secondly, how, given a range of possible words for a particular object or action, a particular community currently makes use of one form of the possible words rather than any other. The range of possible words in a language is limited in the first instance by the phonological system of the language; for example, English cannot readily adopt words which depend on phonemic tone; Australian languages cannot adopt words which depend on the voicing distinction. But within this phonological limitation, a number of different wordforms may be possible within a language to refer to a particular object, action, etc., and there may be wordforms already existing in other languages which are phonologically possible for borrowing into another language community. What determines which wordforms will actually be available and actually be used in a given language community?

The argument in the motor theory as developed earlier was that words are neural programs, formed from the subprograms of which the phonemes consist, and that the structures of the wordforms derive from perceptual and motor organisation involved in the objects, actions, etc. for which words are required. It is argued that there is no necessarily unique wordform for a particular object or action, in the same way as there can be variation in action organisation to achieve a given end or there can be variation in modes of perception between individuals, scanning patterns for objects, and order of analysis of complex scenes. The original formation of a particular word structure for a particular object or action will have been derived from the object, action, etc., under the influence of the particular perceptual and motor organisation of the individual first to use the word; at this level, the word used is a selection made from a number (but not unlimited number) of appropriate wordforms, governed by the individual's neural and anatomical structures subserving motor and perceptual organisation, i.e. ultimately the genetic endowment for motoric and perceptual organisation as it has interacted with the individual's developmental environment. Thus all the words found in a language, and in all languages, have their origin in the physiological/neurological organisation of individuals in the different communities, and to the extent that genetic endowment and perceptual/motoric organisation differ between individuals and on average between communities, different words may be formed as appropriate in different language communities.

The separate question is why, once wordforms have been generated in this essentially genetic way, they should be adopted by a particular group, deme, population, etc., at a particular time. At this point, two sorts of influence may operate: the first is the extent to which the group has gene frequencies harmonious with those of the originating individual, i.e. an 'identity of tendencies' (to use VendryÖs' phrase) that predisposes the community to treat the word as appropriate for the particular object or action, etc. The second, assuming that there is a selection of possibly appropriate words within that community, or readily available by borrowing from another community, depends on the extent of use of the word by classes or groups or individuals within the community, questions of prestige, effective use and so on. What makes rapid change in lexical items possible is the previous existence of variant lexical forms within a community, which can extend their range of use, as a particular variant becomes associated with a larger or smaller group, or a more prominent group, or a group with more effective use of the word.


Syntax is a term which is used variously but here is taken to mean everything in language not covered by the terms phonology and lexicon. There are, however, several distinct segments of syntax: perhaps the most basic is syntax as ordering of words in the utterance; the next connects with ideas about the 'parts of sentence' - subject, object,predicate etc.; the next is concerned with 'parts of speech' - classification of words in terms of the regularities governing their use in the sentence (nouns, verbs). An important broad classification is into function and content words. Function words can be subdivided into those which are essentially syntactic and those which are concerned with other types of relation, e.g. positioning in time or space.

Diversity and change in syntax can affect any of these categories or classifications of what is covered by the term syntax. More specifically:

- variation in word order: SOV or SVO, the positioning of attributes and nominals, of prepositions or adverbials. suffixing, infixing and prefixing - or the ordering of clauses; - variation in the components of the sentence: subject, object, indirect object, verb, predicate, particles; - variation in the functional categories of words: presence or absence of prepositions, conjunctions, subject, object or ergative markers, or more generally of function words; - variation in type: analytic, agglutinative, fusional (inflective), poly-synthetic; - variation in classificatory systems: declensions and conjugations, gender, consonantal and vowel harmony, measures, concord, etc.

Given this wide range of types of diversity in syntax, the issue is how the diversity may have originated and how change in the various types of syntactic structures may have come about. The simplest way to approach this is to make a broad distinction between syntactic features (in the narrow sense of ordering regularities) and lexical aspects of syntax (that is, operations of syntax which depend upon the existence of particular words (or subwords), e.g. the availability of conjunctions, prepositions, case endings, adverbs of time or space and so on.

The specifically syntactic aspects, i.e. ordering regularities, are in the motor theory derived from ordering regularities in the neural sub-structure for motor control and perception (ordering regularities in visual perception and in the execution of bodily actions). These very basic features of any individual's neurological/physiological system are most likely to be affected by genetic factors in development but least likely to change rapidly. Given the extent of individual variation, in brain, morphology and physiology generally, it is not surprising that there should be a considerable range of possible modes of organisation to serve as the base for a large number of different syntactic processes. The general arguments about the effects of relative gene frequencies, and changes over time in gene frequencies, between populations and races are likely to apply, so that different communities may readily have developed different syntactic processes, reflecting genetically based tendencies and forces prevailing in the particular community, population or race. These very basic syntactic processes are least likely to be the subject of any conscious alteration or any change in response to fashion or prestige; the history of syntactic change in this segment then would be more likely to resemble that for changes in phonology, i.e. slow rates of change, widespread in their effects on any language. Historical linguistics shows that this is in fact the picture. Changes from SOV to SVO proceed slowly but eventually operate widely over a particular language use.

The position is somewhat different in the case of the 'lexical' aspects of syntactic diversity and change. The distinction between 'ordering' function words and other relational function words is important. One would expect processes of change, e.g. in the availability of articles, in the use of the dual, in the existence of a gender or other classification, to proceed more slowly than changes, e.g. in relational terms for time and space. An important category is that of subwords which form the basis of inflectional systems and which, in function, cover much the same ground as prepositions or conjunctions. It is difficult to avoid thinking that the emergence of a preposition, for example: BY, constitutes a conceptual advance over an inflectional system in which the idea conveyed by BY is merged in each separate content word. The preposition represents a higher level of abstraction than the inflected form, the conjunction than the use of a simple sequence of sentences. The change would have been the result of intellectual advance, in the analysis not only of the perceived world, but also of the individual's thought and language processes.

How does this bear on syntactic diversity and change? If the move to analytic forms is the result of intellectual progress, evolution of the whole syntactic system would be expected to proceed slowly but in a definite direction. The forms which prepositions and conjunctions take would certainly not be the result of any conscious or arbitrary choice by any community or group, but be the result of unconscious preferences and tendencies. The history of prepositions and conjunctions is obscure but there is evidence that many of them derived from previously existing words falling into other word categories; so in Chinese, prepositions appear to have developed from verbs, such as 'gei' (a verb meaning 'give') used as the preposition 'to'(Norman 1988). In English, prepositions, conjunctions and adverbs often have a common origin.


Writers such as Labov (1984) have attempted to research, on the ground, current language change (mostly for limited aspects of usage or pronunciation in restricted geographical areas or social groups). His work, he would claim, supports a view of the social causes of language change, meaning by this that variant forms found in particular groups are spread more widely for reasons of prestige or as part of attempts by one group to differentiate itself from another group. However, Labov's approach (as Aitchison 1981 has pointed out) deals only with a superficial aspect of language change.

This paper starts from the motor theory of language origin that language was modelled on the structures and processes of the neural motor system. A problem which any biological theory has to tackle is why, if there was a strictly biological origin for language, such a great diversity of languages developed. The proposal is that language changes and diversifies as a result of genetic change over long periods of time. Language in the human has an obvious biological base: speech depends on the anatomy of the articulatory apparatus, on the specialised organs for hearing and most importantly on the neural systems which make speaking and hearing possible. Genetic change over time can affect these physiological and neurological substrates of language.

At first sight this seems to assume that language as currently used is derived directly from the physical and neural organisation of the individuals composing a language community. This is obviously not the case. Children with differing genetic inheritance acquire the language of the local community, not the language of the community to which they are genetically related. If language change is a consequence over time of genetic change, then the process by which genetic change is translated into language change must be more subtle. Whilst variation in genetic make-up will be expressed directly in the phenotype, in the body and brain of the individual, variation in genetic factors affecting the substrates for language in a population confront an established inter-individual system of language, which depends not simply on each individual but on uniformities in language behaviour between members of the group or population.

The approach adopted to reconcile a biological theory of language origin and evolution and the diversity of languages actually found starts from recognition of the many parallels between variation and change in a strict biological sense (the physical characteristics of populations) and variation and change in language over long periods of time. Many of the phenomena analysed in modern population genetics have close parallels in the findings of historical linguistics. Both systems display great diversity and continuing change within and between populations. The next step is that the processes of biological and linguistic change and variation are not simply fortuitously similar but there may be causal links between them. There may not only be a biological model for language change but also a biological explanation. Over long periods of time, there must be changes in the average genetic composition of populations (changes in gene frequencies) which affect the substrates for language of the population and prepare the ground for changes in language forms to reflect differences of the current population from earlier populations in which a language originally developed.

Changing gene frequencies within a population will be manifested in the genomes of children born into the community. To the extent that the gene frequencies differ, then children will be born with physiological and neurological substrates for language which differ from those in earlier populations. These differences take effect, not at the stage where the child is developing its language capacity - it can only learn the local language - but at the stage when the child becomes an adult. The speech preferences of a community depend on the tendencies existing within the members of the community, which are derived from their physiological and neurological make-up. When children with genetic differences relevant for language grow up, their tendencies and preferences will go to modify the prevailing tendencies and preferences within the community. If over a period such changed gene frequencies become sufficiently widespread, then they will alter the linguistic preferences and tendencies of the adult community and exert an influence on the selection of the language forms actually used. So far as the language of a community is taken to be that form which is most widely accepted, then changing gene frequencies will have operated to bring about change in the language.

Changes in gene frequencies within a population affecting the substrates of language come about in a number of ways described in general terms earlier in the paper: the separation off, and isolation, of a particular segment of the population, e.g. the foundation of a colony, inward migration from another population, the suppression within a community of a distinct segment of the population. To make the idea more concrete, events in British history such as the regional pattern of settlement of Angles, Saxons and Danes, the dominance of French-speaking rulers after the Norman Conquest, the Black Death (when a third of the population died), the rise of London to be by far the major centre of population, the formation of colonies in the United States and Australia, the Industrial Revolution with a huge inward migration from Ireland, obviously could have major effects on gene frequencies, that is on the composition of the gene pool. Such major changes will have carried with them the possibility of altered gene frequencies relevant for the anatomic, physiological and neural substrates of language. There are obvious historical parallels elsewhere (notably throughout Europe). The major changes in gene frequencies resulting from them could have provided the motive force and explanation for the recorded changes in languages and dialects.