Language and Evolution: Homepage Robin Allott


The motor theory is both a theory of the origin of language and also a theory of current language functioning. This paper explores in greater depth the possible relation between the motor theory and syntax, and how far the structures of syntax can be treated as in some sense iconic. It reflects thought in progress rather than a finished and complete account of syntax in motor terms.

First, a comment on iconicity. Iconicity is interpreted essentially as mapping or isomorphism from one structure to another. I interpret iconicity as meaning that there is picturing in language (and particularly in syntax) of something that is extra-linguistic; picturing is taken to mean not only a visual picturing but a systematic similarity between the structure of language (and particularly of syntax) and extra-linguistic structure (which may be motor or perceptual organisation). This approach to iconicity of syntax is accordingly substantially different from those quite recently developed by Haiman(1985) and Givon(1979) and has links with Osgood's ideas on 'Where sentences come from'(1971).

The second important introductory comment is that whilst an account has been given of the motor theory, there still remains a very large question mark over the definition or boundaries of syntax. There is no consensus on what is to be treated as syntactic as contrasted with what is to be thought of as lexical or semantic. There have been many competing definitions of syntax since the ancient Greeks first used the term (Householder). There have been even more competing 'grammars' which implicitly assert their own definition of syntax. Recently I saw a book with some such title as "30 Million Theories of Grammars", which someone calculated was the number of possible permutations and combinations. I take the simplest, most obvious, definition of 'syntax' (returning to the Greek origin of syntax) , that it is the system by which words are put together in any language to convey meaning.(Dik. 2) I do not then accept the currently more fashionable view that syntax is to be defined as the formulation of rules to generate 'acceptable' (ie. grammatically well-formed) sentences in any language - this undervalues the importance of meaning in language and overvalues grammatical form, detaching the study of syntax from the use of language in the real world and from any consideration of its psychological or biological base. One further clarification: the approach does not rely on the now rather out-moded Chomskyan distinction between formal deep and surface structure; syntax is treated as monostratal (to use Gazdar's term - Gazdar et al.) which at the same time of course does not exclude consideration of the neural structures underlying the utterance.


There are three major distinguishable components of syntax on the interpretation of 'syntax' adopted in this paper:

1. The principal categories of words (Nouns and Verbs, with the dependent categories of Adjectives and Adverbs). These together form the open class of content words.

2. Ordering of words, including sub-ordering, that is, the clustering of words within a larger order.

3. Function words (including subwords eg. morphemes such as terminations of abstract nouns, verb inflections etc).

On the view I am adopting, the syntax of a language results from the co-operation and interaction of these three components.

Some brief justification for identifying these three components. Little justification is needed for treating the open classes of Noun, Verb, Adjective and Adverb as a major component. It would be difficult to find any modern syntactic theory which denies or dispenses with these categories (Gazdar et al., Chomsky, Wasow) - though of course there have been assertions that some languages lack one or more of the categories.

There is equally little difficulty about the second main component, ordering and sub-ordering. Transformational and phrase structure grammars (new and old variants) are concerned very largely with ordering. It is inescapable that spoken language is a serial activity, where position of words in the string makes a difference to what is communicated.

The third major component (the function words and subwords) is of special importance for syntax. The justification for treating this class as a distinct major component takes several forms:

a. The class is closed. One can enumerate the function words and subwords in any language in a way that is not possible for nouns, verbs etc. Even more strikingly the number of function words and subwords is quite small;

b. There is evidence of a neurological distinction between function words and other words. The evidence comes from clinical treatment of the aphasias over a long period; many aphasiologists have noted that in certain types of aphasia, content and function words are differentially affected. "In Luria's 'efferent aphasias'.. the selection of content words is unimpaired, but the combination of words and their serial order may be severely disturbed. Speech may degenerate to a succession of telegraph- style utterances. Grammatical words such as conjunctions, prepositions, pronouns and articles disappear. roots of words are preserved better than grammatical endings, tense or gender. The nominative may be the only case that survives". (Sommerhoff, 362). One of the more remarkable observations reported by Marin (Marin et al., 878) was where patients were asked to read a list of homophone-pairs where one member of the pair was a content word and the other a function word. "H.T. was able to read 'four' but not 'for'; V.S. read 'sum' but not 'some'; and J.D. read 'for' and 'some' but not 'four' or 'sum'. "The difficulty in specifying such features as number, tense and aspect is not found with irregular forms. Both irregular plural nouns and irregular past tense forms are read several orders of magnitude better than their regular counterparts" (880). "There is evidence that patients who have difficulty reading function words also have difficulty with abstract nouns" (882). Howard in a paper on Agrammatism says(Howard, 1) "In languages with well-developed inflectional systems, agrammatism is particularly striking. Nouns tend to appear only in the nominative case, verbs in the infinitive, auxiliary verbs and other words from closed classes are omitted". Zurif(1982, 313) after discussing cases patients with Broca's aphasia who cannot manage syntax with the open/closed word class distinction comments "It may be hypothesised that the closed class access route normally plays a special role in the assignment of structural analysis. The notion here is that this route serves as input to a parser, permitting the on-line construction of a structural representation". "There was one exception to this general finding [inability to use function words], however; this involved prepositions , and then only when they are clearly and non-redundantly 'functional' - containing, for example, important locative information clustered with the relevant nouns"(1983, 190). The fact that there are many very different types of aphasia with patients showing different clusters of deficits in no way weakens the force of the observation that brain damage can produce specifically grammatical, function word, deficits.

Equally, evidence of the neurological distinctiveness of function words comes from stimulation experiments, in the clinical treatment of epilepsy. Following the technique of direct stimulation of the exposed cortex used originally by Penfield(1959), Ojemann and Mateer have reported, amongst many striking observations on the relation between speech and motor function, "There are also sites where electrically induced changes are confined to closed class words". (Ojemann 1983(b), 207) Ojemann (1983(a), 71) "found the identification of [cortical sites] with changes in closed but not open class words that we have related to syntax". "At a few sites, only conjunctions, prepositions and verb endings were altered during stimulation"...These sites are interpreted as specific to syntax". It is a remarkable finding that the brain appears to recognise traditional parts of speech as different classes (almost as remarkable as animals making categorical distinction of speech-sounds in a way similar to that used by humans in distinguishing phonemes).

c. There is also evidence of the distinctiveness of function words and subwords from a totally different direction, that is, procedures for parsing natural language by computer. In some well-established computer parsing systems, function words have a distinct structural role eg. in ATN systems. (See Arbib et al. 1987, 67). More specifically, it is possible to construct a computer parsing program relying solely on a context-free listing of function-words to parse sentences of any length. A system of this kind requires no lexical or grammatical information other than operationally determined groupings of function words and subwords (not strictly in terms of traditional parts of speech classes). (See also Tomita's context-free parsing algorithm(1986) , parsing from left to right and handling unknown words without special mechanisms; this and similar computer parsing programs avoid the defects of many other parsing systems which mingle semantic and syntactic elements so heavily that it is difficult to see any basic distinction between grammar and total language interpretation; if a large enough semantic data base is provided for a parsing program or a restricted enough field defined (expert systems), then successful computer parsing is not surprising or illuminating for language function).


Given this very rapid account of the main components of syntax, one can return in a more specific way to the question of how the motor theory is related to syntax. The question now takes the form of how each of the main components of syntax should be considered in the light of the motor theory, and how then the total functioning of syntax, as a result of cooperation and interaction between the three components, should be related to the motor theory.

The motor theory asserts that motor programs and the principles for combining motor programs underly the structure of language. At the same time there is a close link between motor control (action organisation) and perception (the organisation of vision). For each of the three components in syntax, the relation to the motor theory may take the form of:

a relation directly with the organisation of action (referred to by one writer as 'the grammar of action')

a relation directly with the organisation of perception (referred to by Gregory some years ago as 'the grammar of vision'(Gregory,622). Vision of course is motor-based, the eye sees by the combination of saccades and fixations plus a constant (structural) tremor which appears to play an essential role in maintaining vision.

3.1 The open categories (Noun Verb Adjective Adverb)

Each of the open categories contains an unlimited number of words. The explanation of the particular forms which these words take in relation to their meaning is a matter of the organisation of the lexicon, and not one specifically of syntax. What is necessary in this paper is to attempt to explain how these categories as such, that is, Noun, Verb, Adjective and Adverb, derive from the motor system. On the theory, the existence of Nouns, Verbs, Adjectives and Adverbs as broad categories must be explicable in terms of the neuromuscular organisation of vision or the neuromuscular organisation of bodily action.

In vision the eye scans the visual scene by a succession of rapid eye-movements interspersed with longer periods when the eye is foveated on the endpoint of the movement (saccades and fixations) coupled with the rather slower movements for accommodation and vergence.(Carpenter, Fuchs, Henn & Hepp, Zingale & Kowler) The experience of the eye, in motor terms, is the experience of the eye muscles; perception sorts out the visual scene in terms of persisting or static configurations (visual objects) such as trees, cows, houses, of the particular relations between the objects, and of changes in the relative positions of the objects or of parts of the objects themselves. The most primitive division of what is perceived is into the static and the changing, which equate reliably with Nouns and Verbs. Adjective emerge as static sub-patterns of static objects and adverbs as sub-categorisation of movement patterns. Words referring to static objects are generated initially by the motor program (composed of saccades and fixations) responsible for scanning the object; this motor program would be available for conversion into an articulatory pattern derived from the distinctive features of the visual motor pattern. This is discussed more fully in the paper presented to the 11th ICAES in Vancouver on "Structure and Development of the Lexicon in Relation to the Origin of Language"(Allott 1983). As that paper showed, the categories of Noun and Verb will also include articulatory patterns (words) derived from motor programs for bodily action eg. the word 'HIT' from the program involved in the action of hitting.

3.2 Ordering (and sub-ordering

Scanning of a visual scene by the eye is a serial process(Aslin,3), just as spoken language is a serial process. The iconicity of syntax in terms of word-order derives from this. In vision, there are particular aspects of salience and emphasis (in terms of fixation duration) similar to those involved in the expression of salience and emphasis in word-order. There are also aspects of salience and emphasis (taking the form of relative force) in the motor programs underlying bodily action and related to the content of an ordered speech utterance. Bodily action is also serial; we stretch out our arm before we pick up a glass before we bring the glass towards us and then drink from it. There is less freedom of ordering in utterances relating to action then in utterances relating to visual perception. "I am going to get in the car" "The man got in the car" as against "There was a car outside the house" "The car was outside the house" "Outside the house there was a car".

The serial ordering of vision and of action provides, in the motor theory, the primitive foundation for the patterns of ordering in speech utterances; of course, as language has become more complex, there has been great elaboration of ordering in language but one can still see a sharp distinction between ordering in the class of utterances patterned on static perception and in the class of utterances involving the representation of action.

As regards sub-ordering, that is, for example, the order of words in a cluster, such as a Noun phrase with a number of adjectives, this also can be analysed in terms of the organisation of visual perception initially. So in a phrase such as "A large white Siamese cat" , the ordering of the adjectives can be related to the ordering of perception, and the ordering of the analysis of perception. Different languages may have different practices as regards pre-position or post-position of adjectives, but this does not in any way nullify the original source of the ordering in the ordering of perception.

3.3 Function words

The relation of the motor theory to the third major component of syntax, the closed class of function words, is in some ways the most important, and at the same time requires the breaking of the most new ground. The function word component of syntax is where the most considerable differences are observed between one language and another in terms of how far they depend on separate function words (as largely in English) and how far they depend on function sub-words (inflections, morphemes for forming eg. abstract word-classes etc).

One other fundamental point where there is divergence from fashionable current schools of syntax theory. Since what we should be concerned with is how meaning is conveyed by the word- string or utterance and not with whether 'sentences are grammatically well-formed', account has to be taken of the extent to which meaning can be conveyed without much use of function words eg. in telegraphese, in Broca's aphasia speech and most of all in current colloquial speech and writing. A striking example of the possibility of dispensing with all or most function words and still succeeding in conveying meaning using only word-order and the major word-categories can be seen every day in newspaper headlines (examples can readily be provided).

To relate the class of function-words to the motor theory, the first step is to prepare a list of words considered to be function words. This cannot be based on any simple treatment of traditional parts of speech, prepositions, conjunctions etc. as function words. A function word or function subword is a speech- form which has no definable external reference and which acts in association with other function words to determine the role of non-function words in the word string or utterance (in English often to decide whether open-class words are acting as Nouns, Verbs or Adjectives and whether the non-function words are operating as Subject, Object, Complement etc). Appended to this is a list of words and subwords treated as functors as part of a successful context- free, non-lexical, computer parsing program. The classification of the function words does not follow any strict traditional parts-of speech approach; the classes and the words included in them were determined operationally, that is, as the classes needed in practice to achieve successful parsing of long stretches of text.

If, as the theory being presented proposes, the syntactic components are derived from the motor control system (either from the organisation of action or from the motor control organisation of perception), then the next stage is to consider how in detail the closed-class of function words and subwords have analogues in the motor control system or could be derived from aspects of motor programming required for Gregory's(622) 'Grammar of vision' or what might correspondingly be termed 'the grammar of action'. To do this will require a detailed account of what is known about the functioning of visual perception (reasonably well-known in its earliest stages at the level of control of movements of the eye, the characteristics of the extraocular muscles, the functioning of accommodation and so on) and the motor programming of bodily action (less well understood). To present this material would require a paper much longer than is possible on this occasion. But some light can be thrown on what is needed in due course by looking at the characteristics of function words and function subwords (as listed) to see how far they might be related to action or vision motor programming.

Features of function words which can be compared with operational aspects of 'action grammar' or 'vision grammar' are:

Timing Words such as : After before while when since until then now still already. There must be equivalents to these in the organisation of action and vision.

Direction and Relative Position Words such as: from at with by between within towards up out among here. The eye's saccade and fixation programs are very much concerned with direction of movement from one point of fixation to another and the relative positions of salient features in the visual scene. There also the deictic function words which are closely related to bodily gesture.

Hesitation Choice Change of direction Links Words such as: but whether and or either nor perhaps, and the interrogative words. In programming of bodily action there are analogues of the functions performed by these words. A line of action can be halted temporarily or changed; a new partial action added to the first action.

Salience, emphasis Words such as: very quite rather somewhat. These might be correlated with aspects of relative force in the control of action and focus or duration of fixation in visual perception.

Sequence Words such as : for as that than.

This first very rough classification is meant only as an indication of the lines along which study of the relation between motor programming and the class of function words might proceed. What it might lead to is a clearer idea of how the segments of motor programs might be fitted together, using our knowledge of syntax (in the sense proposed in this paper) to throw light on motor control just as much as using motor control and vision research to help us to tackle syntax in a new and biologically more relevant way.


It has been argued that by language there can be a transfer of neural patterning from one brain to another. It is also proposed that the syntax of language (as well as phonology and lexical organisation) has been derived from and modelled on pre- existing neural organisation for action and perception. Clinical evidence from the treatment of aphasia and experimental evidence of the effects of direct electrical stimulation of the cortex suggest that, in neural terms, syntax and lexicon are separable components in language production and comprehension. In practical terms, there can be fairly effective communication by speech or in writing without much syntactic structure. How in the light of this ought one to envisage the role of syntax in the total process by which the individual translates his experience into words and uses them to transmit the content of his experience to another individual? Rather speculatively, one might formulate an account on the following lines.

The starting point for any particular use of language, an utterance or a written sentence, the situation from which the word-string derives, is a perception or an action of the individual. At its simplest the content of the perception, what is perceived, is the relationship of elements in the visual scene. The simplest perception consists of a number of elements in the perceived scene together with their relation to one another e.g. a cow is standing near a tree. These meaningful elements constitute the minimum semantic elements which will be taken into the utterance (or sentence) which describes the scene. The relation existing between the elements in the scene is transferred to constitute the relation between the elements in the word string which describe the visual scene. In terms of neural patterning, the precursor of the utterance is a compact form of the semantic elements derived from the perceived scene; this compact non-syntactic form (with some analogy to telegraphese or newspaper headlines) has to be converted from a simultaneous patterning into the serial form required for the normal use of language. Words normally cannot be uttered all together, in one burst. The compact semantic form has to be unrolled into a serial syntactic form. This is done by the addition of function words, inflections, punctuation (pauses) and grouping of the meaningful (semantic) elements.. In this expanded syntactic form, the content of the perception can be transmitted by speech or writing.

The hearer of the utterance, or the reader of the written sentence, has to perform a process which is the reverse of that performed by the originator of the word-string. The expanded syntactic form has to be stripped of function-words, inflections etc. However, as they disappear, these syntactic elements guide the manner in which the content words, the semantic elements, are to be related in neural patterning. The hearer re-creates for himself the compact semantic (neural) form from which the speaker's utterance originally started.The compact form is then interpreted by the receiver who, if the transmission is successful, will have a structured neural patterning corresponding to, isomorphic with, the neural patterning from which the word-string was constructed in the first place. The transmitted patterning is interpreted by the receiver in much the same way as would be a perception originating within the receiver himself. This view of the functioning of language seems to be related to the concept of 'inner speech' developed by Vygotsky and discussed by Luria (1977, 102). It makes it possible to understand how 'newspaper speak', telegraphese, the speech of Broca's aphasics, can dispense to a great extent with syntax by relying simply on the conjunction of the semantically weighty content words.