Second draft of a paper submitted to Evolution of Communication

The role of gesture in the establishment of symbolic abilities: continuities and discontinuities in early language development.

Andrew Lock

Department of Psychology
Massey University
Palmerston North
New Zealand

Introduction

The notion that gestures have played a role in the phylogenetic elaboration of language goes back to the eighteenth century (Hewes, 1996). That they might play a role in the ontogenetic elaboration of language is a more recent claim (for example, de Laguna, 1927). The scenarios put forward in both of these domains bring the issue of continuity in the elaboration of linguistic communication systems to the fore.

At one extreme there are claims for continuity (e.g. Lock, 1980), that 'words develop as direct transformations of gestures' (Johnson, Davis and Macken, 1996: p. 713). Such claims often appeal to a theoretical framework such as that put forward by Macmurray (e.g., 1961, p. 60):

Long before the child learns to speak he is able to communicate, meaningfully and intentionally, with his mother. In learning a language, he is acquiring a more effective and elaborate means of doing something which he can already do in a crude and more primitive fashion. If this were not so, not merely the child's acquiring of speech, but the very existence of language would be an inexplicable mystery.

The emphasis here is on communication as a continuous domain of development, and little is actually claimed about where a developing language might get its grammatical structure from.

At the other end of the continuum are claims of discontinuity (e.g., Petitto, 1987). The theoretical framework within which such claims are located tends to be the 'autonomous linguistic' one associated with Chomsky (e.g., 1968, p. 59):

human language appears to be a unique phenomenon, without significant analogue in the animal world. If this is so, it is quite senseless to raise the problem of explaining the evolution of human language from more primitive systems of communication that appear at lower levels of intellectual capacity.

Ontogenetically, the same argument holds, though expressed differently. Gesture, as a 'primitive system' plays no role in the acquisition of language. Hence, the 'autonomous' approach to language - conceiving it as based on a domain specific module - needs to be invoked. This module is theoretically invoked to provide the 'unique' prerequisites that allow a solution to the 'learnability paradox' of mastering the 'unique' principles that underwrite grammatical competence when the 'data' underspecifies those principles. Grammar is thus based on discontinuous processes, and little is said about how a developing language might have a communicative use.

This paper explores a middle ground between these two poles. It represents a revisiting of two earlier pieces (Lock, 1980; 1985) after a long excursion into the area of the evolution of symbolic abilities (e.g., Lock and Peters, 1996). A number of lines of thinking that run through this revisiting are directly parallel to a related path followed over a similar period of time by Elizabeth Bates and her colleagues (e.g., 1979, 1988; Elman et al, 1996; see also Smith and Thelen, 1993).

First, traditional conceptions of the contributions of 'nature' and 'nurture' to development are overly simple. The developmental trajectories mapped out in ontogeny result from highly interactive systems revisiting each others 'products' over the course of time, so as to constitute a self-constraining system that constructs its mature forms out of reworkings of earlier stages as they unfold. Second, it is thus highly improbable that any discrete or localised systems that may be elucidated as end points of development - cognitive modules of any sort, for example - result from pre-specified plans for their eventual hard wiring: 'modules are made, not born' (Bates, Bretherton, & Snyder, 1988: p. 284). Third, the argument from adult form to innate specification is illegitimate. The fact that 'language' can be conceived of as an object, and a set of rules elaborated to account for the structural characteristics of that object, does not mean those rules have to be given as prerequisites to the construction of that object, any more than the fact that the natural numbers fall into exclusive categories of odd and even requires our 'knowing' that as a precondition for being able to count.

These three points have obvious implications as to what 'continuity' might mean when applied to any developing system. This is particularly clear in the case of trying to establish possible relations between gestures and language. For example, the motoric organisation of gestures is effected differently from that of the organs of speech. Again, gestures are generally thought of as existing in the visual modality, and the pathways by which they are interpreted as meaningful will be different from those involved in getting sense out of an auditory input. How could there possibly be any continuity across such differences?

The approach here is to look for principles of continuity at an abstract level. How these principles are realised in development is a separate issue. For example, suppose we think at a simplistic level about the issue of temporal organisation in manual gesture versus speech. Two manual gestures can be produced simultaneously, one by each hand. This provides opportunities for the way in which combinations might occur in this medium. By contrast, two words cannot be said at once, and this provides constraints on how meanings are expressed, in that words can only be combined sequentially. The organisation of the delivery systems of meaning has, therefore, to be discontinuous between the two modalities. But at an abstract level, there can still be continuities between how 'elementary particles' of meaning are combined, irrespective of the communicative modality being used. To get at these continuities requires looking at things in a new way. I will outline this way briefly before using it as a framework in which to consider language development.

Time, environment, organism and meaning

Neither evolutionary nor developmental theories have been very good at providing a set of temporal concepts for characterising processes of change. If they had been, then all the controversies generated concerning teleology in evolution, for example, would have been non-starters. One of the contributing factors here is that the 'Modern Synthesis' which reconciled Darwinism with Mendelian genetics was characterised by a rigid separation of organisms from their environment (Patten, 1982). The selection pressures which lead to the differential survival of particular genes over time are the forces responsible for evolutionary change. These forces are located in the environment as 'autonomous events'.

Subsequently, Levins and Lewontin (1985; see also Lewontin, 1978, 1982, 1983), in developing a critique of the concept of adaptation as it is employed in evolutionary theory, have brought a more interactive conception into focus, arguing that 'the environment and the organism actively codetermine each other' (1985: p. 99). This view was also developed in the works of George Herbert Mead and Jacob von Uexkull, and these are outlined here as providing a grounding for the approach developed here.

To start with Mead: for him, 'the organism ... is in a sense responsible for its environment' (1934: p. 130).

Take the case of food: If an animal that can digest grass, such as an ox, comes into the world, then grass becomes food. That object did not exist before, that is, grass as food (1934: 129);

and similarly:

When there is [a] relation between form and environment, then objects can appear which would not have been there otherwise; but the animal does not create ... food in the sense that he makes an object out of nothing. Rather, when the form is put into the environment, then there arises such a thing as food. Wheat becomes food: just as water arises in the relation of hydrogen and oxygen (ibid: 333).

Mead provides here the basis for a temporal and interactive vocabulary to be developed, and one which can encompass mental as well as somatic phenomena. He makes the point that the environment, in this case 'grass-as-food', maps out the logical structure of the act of the 'ox-as-grass-eater':

the structure of the environment is a mapping out of organic responses to nature: any environment, whether social or individual, is a mapping out of the logical structure of the act to which it answers, an act seeking overt expression (ibid: 129).

At the same time, the 'reverse' view can be formulated or extrapolated: that the existence of the ox and its digestive system - its 'bacteriological laboratory' (ibid: 131) - has been mapped out from the logical structure implied by the existence of grass, along with the constraints provided by the starting point of the proto-ox's actual biology.

We now have the basis for a conceptualisation of change through time that has a logical structure to it provided by a set of implications. And a beautiful thing about implications has been captured economically by Smedslund (1969: 8):

A crucial difference between a cause-effect relationship (A causes B) and an implication (A implies B) is that the former involves logically independent phenomena, which must be shown empirically to succeed each other, whereas the latter involves logically dependent phenomena. Physical phenomena may be linked by a theory from which it follows that A leads to B. On the other hand, a mental phenomenon A in itself implies B, without any theory (emphasis added).

And the beautiful thing about Mead's argument is that it extends Smedslund's point beyond mental phenomena to encompass biological ones as well. Biological materials can engender implications because, in this conceptualisation, they provide the 'value perspective' that transforms 'logically independent phenomena' into logically dependent ones. That is, the empirical questions about an evolutionary process are 'how is it done?', not 'what is being done?'.

For example, consider the question: 'which came (= could come) first, animals or plants'? The answer is obvious, not because we learned it in school biology, but because the logical requirements of 'perspective' have been articulated. Organisms which sustain their organisation over time on the basis of an extra-terrestrial energy source must come before organisms that sustain themselves on the basis of a terrestrial one. But once that energy source from outside of the terrestrial system has been captured and conserved in the form of 'plants', then at the same time, 'plants' have constituted themselves as a newly-emergent possible energy source which conserves, transduces, re-presents or mediates (take your pick) the original one. Plants imply animals: alternatively, plants establish the environmental niche that 'animals' could exploit - all other 'things' being equal. Next question: 'given 'plants', which came (= could come) first (= next), carnivores or herbivores (leaving out 'parasites' as an unwanted complicating sub-category of carnivores at this point)'? Again, the answer is obvious. But the next question does not have such an obvious answer: 'which came (= could come) first (= next), carnivores or lots of different variants of plants'? Clearly, once patterns of functioning organisation are attained, both synchronic and diachronic sets of implications are brought up as possibilities for future explication.

These possibilities exist, and act as constraints on the future elaboration of evolving organisms (and conversely, the structure of evolving organisms act as a set of constraints as to how possible these possibilities are). As James (1956) and Shotter (1981) put it:

[Indeterminism] admits that possibilities may be in excess of actualities, and that things not yet revealed to our knowledge may in themselves really be ambiguous. ... actualities seem to float in a sea of possibilities from out of which they are chosen: and, somewhere, indeterminism says, such possibilities exist, and form a part of truth (James, 1956: pp. 150-1).

and;

... an action in progress can, while having so far produced a certain degree of specification into its content, leave that content open to yet further specification - but only specification of an already specified kind (Shotter, 1981: p. 276).

That is, possible futures (co-)exist in the present as the implications of present states of affairs: they are part of the 'value perspective' constituted by an organism.

Turning to von Uexkull yields the important point that these perspectives and possibilities are not things seen 'from outside of' the evolving system, but are constituted within it in the organisms perception of the environment, in what he terms an organism's Umwelt:

[The Umwelt] is the world around the animal as the animal sees it, the subjective world as contrasted with the environment. The effects of stimulation appear in this Umwelt as elementary sensations, Merkzeichen, which, organised and projected into the object, become meaningful perceptions, conceived by the [animal] as the properties of that object, Merkmal (von Uexkull, 1957: xiii).

His point is that animals provide 'value-perspectives' that determine the way in which 'the world' is presented to them. Sensory systems are not passive transducers of information, but constitutive transducers. Some of these value-perspectives are given to animals as a result of their evolutionary histories; and some may be 'imported' into the constitutive perceptual system by learning. Either way, the world is presented to an animal as possessing objects that are meaningful to it through the values that arise as a result of its being in a particular relation to its environment. It is in this sense that von Uexkull's Umwelt contains meaningful items:

Every action ... that consists of perception and operation imprints its meaning on the meaningless object and thereby makes it into a subject-related meaning-carrier in the respective Umwelt (subjective universe) (von Uexkull, 1982 [1940]: 31).

A more recent but similar formulation is given by Thompson (e.g., Thompson et al., 1992; Thompson, 1995a; 1995b), who argues that

colour perception does not represent something that is already present in the world apart from perceivers; rather it presents the world in a manner that satisfies the perceiver's adaptive ecological needs (Thompson, 1995b: 25).

(See also Maturana and Varela, 1980; Hendriks-Jansen, 1996; and Clark, 1997, for related discussions).

This brief tour of notions of the co-construction and co-constitution of an organism and its environment provides a set of conceptual tools that are useful in describing the process of language acquisition so that issues of continuity and discontinuity are made more apparent. These tools enable us to see the early stages of language acquisition as a constructive process in which the child builds a meaningful world from two sources: their given 'value perspectives', and their immersion in joint actions with adults who thereby transact their version of the world to their offspring. In addition, we can start to look at the problem confronting the child in reinventing the particular language system of its cultural milieu as one of being guided to make explicit the particular implications of its Umwelt at any particular point in time.

The ontogenetic explication of behavioural implications

1. Perspectives and the elaboration of explicit, functional communication.

When we experience a need we do not at first know what it is we need. We must search to discover what allays our discomfort. This is not found by comparing various objects and activities with some objective, determinate criterion, but through ... our sense of gratification. This gratification is experienced as the discovery of what we needed all along, but it is a retrospective understanding and covers up the fact that we were unable to make our need determinate without first receiving that gratification. The original fulfilment of any need is, therefore, ... a creative discovery. Dreyfus, 1967: 25-6.

Dreyfus' account of 'needs determination' has exactly the form of temporality needed to characterise ontogenetic change in early infancy. A 'retrospective understanding' results from coming to a realisation of the implications of one's situation as an infant. That is, having bodily sensations provides a value perspective on the world that development provides the means for discovering and controlling. This can be clarified by running a thought experiment.

Sometime after having been born, an infant is going to feel, for the first time, what he or she will come to know as 'hunger'. Let us assume that infants have no way of telling what this discomfort is, nor how to get rid of it, nor even an inkling that they could get rid of it: but it makes them fret or cry yell. Let us assign the following anthropomorphic meaning to this early infant crying: 'whatever this feeling is, I do not want it'. This carries with it the implication that infants want something else, but this is not something they know about yet. Somehow, the feeling is removed. So they begin to know that things can get better, and the anthropomorphic meaning of their crying shifts to encompass this: they now cry with a grasp of the meaning of their crying which can be rendered as 'I want something else'.

With time that 'something else' is made more explicit as 'I want whatever it was that worked last time', thus focussing the infant's attention onto events in the outside world rather than solely on bodily feelings. They were, naturally yet implicitly, already in particular relations of value with objects and events in their external world. These relations are just beginning to become explicitly grasped, and, to use von Uexkull's phrasing (above, 1982: 40), action stamps meaning onto the perceived object. That this happens can be seen in the way infants begin to control their actions toward objects. Piaget (1951: 58-60) provides some examples of this at around the beginning of the third month of age:

Laurent only tries to nurse when he is in his mother's arms and no longer when on the dressing table...

Lucienne at 0;3(12) stops crying when she sees her mother unfastening her dress for the meal.

Here, both children give evidence of having 'retrospectively discovered' the implied conditions of their needs, and their actions are now informed by these discoveries. As a result, infants come to live in a 'new' perceptual world in which objects and events have a known significance for them that they previously lacked an appreciation of: they are in a constitutive relation with a changing, developing Umwelt of now explicitly meaningful objects.

The 'sting in the tail' at this point is that the above discoveries are mediated by the actions of another. An infant might have come to an explicit grasp of the implications of bodily discomfort as it is constituted in his or her Umwelt - by acting with a value rendered here as 'I want that' towards particular circumstances at particular times - but the implication of this, that 'you must do something so as I get what I want' still exists beyond the infant's explicit control. But when an infant gets to be able to bring this implication under control, at around 10 months of age, he or she is confronted with the pragmatic need to make explicit what that 'something' is as well. That is, the infant could do with being able to indicate to the adult who mediates his or her needs both that he or she wants something and what that something is. And this is something that 12 month old infants are capable of doing: they can attract and direct an adult's attention vocally and gesturally.

By the age of a year, then, the previously implicit structuring of the infant' s communicative abilities - that is, the functional exigencies of communication based on 'needs' as placing the infant in a relation to objects in the world - has become explicitly rendered and marked. Thus, for example, natal distress conveys holistically to the perceiving, 'default' adult the entire set of subject-relation-world elements that the adult distinguishes by separate signifiers. From the infant's point of view, crying has the implicit value (I DO NOT WANT THIS), and from the adult's perspective it has the value (I WANT THAT). What the infant has to do is use the clues given in interaction to shift from the default meaning to the adult's one. This happens by the structuring of the infant's perception to encompass now-explicitly meaningful objects.

Subsequent to this shift, the holistic THAT has to be further differentiated so as to distinguish its components of 'the object or state that is wanted' from 'the agency that provides it'. And at 10 months, the infant's communication has become explicitly organised by this realised implicational structure (bar the subject component which is still unarticulated): attention-getting and attention-directing indicate a differentiation of agent from action; and objects are specified deictically. Rather than a meaning that might be rendered as something like (I WANT (YOU DO) X) being implicitly conveyed by one action (e.g., crying) which has an implicit, unarticulated holistic 'structure', we find an activity explicitly structured along the lines of ((I) WANT (YOU DO) X), in which the need, the object and the other agent are all explicitly assignable to components of the communicative act. This structuring has been revealed to the infant, and comes to be explicitly reflected in the production of his or her communications, through it being an embedded property of the infant's 'zone of proximal development' (Vygotsky, 1978): a zone of both 'sense-making' and 'sense-structuring' in which structural specification is occurring, but only 'specification of an already specified kind'.

As a concrete example, consider the following:

Paul; age 14m 23d: Mother enters the room holding a cup of tea. Paul turns from his play in her direction and obviously sees it. (i) He cries vestigially and so attracts mother's attention; immediately he points toward her and smacks his lips concurrently.

Mother: No, you can't have this one, it's Andy's.

Mother gives me the cup of tea, and I put it on the mantelpiece to cool. Paul crawls across to me and grasps my knees. (ii) I turn to look at him; he looks towards the mantelpiece and points, turns back to me, continues to point, and smacks his lips (Lock, 1980: 95-6; italics in original).

Notice that the child is able to 'code' separate portions of the intended meaning, and can do so creatively, taking the situation into account (in that on the second occasion crying so as to establish mutual attention is not required, and it is not used). This productive ability is graphically represented in Figure 1 as an illustration of how different actions and gestures can be creatively combined in different situations to explicitly code meaningful portions of the earlier implicitly rendered holistic messages.

Insert Figure 1 about here

This form of structuring is a pragmatically motivated and derived one. It is in no sense particularly linguistic, and nor is it exclusive to humans:

Bobby; age 18m: Bobby's mother has friends in and is pouring a cup of tea. Bobby, who has been sat next to her, gets up and goes out of the room. He returns with his drinking vessel and puts it in mother's lap. He sits down and vocalises. He gets some tea poured for him.

Bobby is my dog. Primatologists can doubtless provide similar examples. Seen this way, a discontinuity view, such as expressed by Bruner (1990: 75-6) -

Once the child masters through interaction the appropriate prelinguistic forms for managing ostensive reference, he or she can move beyond them to operate, as it were, within the confines of language proper. This is not to say that the linguistic form 'grows out of' the prelinguistic practices. It is, I think, impossible in principle to establish any formal continuity between an earlier 'preverbal' and a later functionally 'equivalent' linguistic form.

- is a valid candidate for an explanatory framework in which to locate subsequent developments. Dogs and apes do not go on to talk in ways that are linguistically structured because they lack the ability to.

But irrespective of whether there is any formal continuity between these early verbal combinations and later grammatical combinations, this does not preclude, as Bruner implies, there being a developmental continuity between the underlying systems that operate at the different times. The argument developed here is that there is a continuity from preverbal to early verbal combinations. The transition is largely one of modality of expression. Physical gestures become accompanied by vocal gestures that have a more-or-less conventional form through the child's attempts at productively imitating the tokens they hear and incorporate into their repertoires. Thus, there is nothing inherently linguistic about the structuring of early word combinations either. If that is the case, then there is a continuity between the structuring of preverbal and early verbal combinations. This pushes the issue of formal continuity further down the track.

It also raises the question as to what is responsible for the eventual 'break point' that produces this formal discontinuity. On the one hand, this could arise because a 'grammatical module' comes into play downstream to provide a new way of handling combinations. On the other hand, this could occur because of the reworking of the combinatorial principles that is necessitated by the need to 'keep track' of the increasing catalogue of explicitly-realisable elements in the emerging communication system. That is, 'grammar' grows out of an increasing lexicon, and its formal characteristics are emergents of this growth, rather than principles imported from some entirely separate source. While I would tend to favour the latter option, this issue goes beyond the scope of this paper. Here, I am going to focus on issues of continuity and discontinuity in the early stages of verbal combinations.

There are both continuities and discontinuities in this early period. On the one hand, continuity seems to be the situation with respect to principles of combination. On the other hand, there is a major transformation occurring at the level of the elements that can be combined. This transformation is concerned with developing a propositional frame through the attainment of symbolic reference. It is in understanding this transition that von Uexkull's (op. cit.) Umwelt paradigm is useful, for it takes our attention away from the purely abstract stance towards what is going on, and relocates it in the embodied perceptual world in which these changes are occurring. Here, I just want to take von Uexkull's approach heuristically, particularly his formulation noted earlier that 'every action ... that consists of perception and operation imprints its meaning on the meaningless object' (1982: 41).

The focus, then, is on how the infant's 'sense-making' processes are active in structuring the infant's directly-accessible perceptual world. That is, how objects and agents come to be seen as objects and agents of particular kinds. Experiential transformations are being hypothesised here. Rather than taking a 'cognitive' stance, which could be portrayed as saying what the infant 'sees' remains the same, and the interpretive procedures for making sense of what is seen are the things that are being elaborated developmentally - behind the scenes, so to speak - the suggestion is that it is useful to 'see' what is happening in a different light. These hypothesised experiential transformations of infant perception are not fortuitous. On the one hand, the given 'value perspective' of being an infant provides an already nascent structural implicature; and on the other, already encultured others transact and thereby socially-mediate the infant's perception. Mother might say, 'Oh look...he's got boots on just like Daddy's', and, at a certain stage in development when an infant can handle this kind of input, then this is what the infant comes to see (in a similar way we come to see 'tables' while chimps come to see 'something to walk on'). Objects become perceptually polysemic: they come to present a number of implications for continuing an action 'all-at-once', because those meanings have been 'imprinted into them'. This is a variant of Gibson's direct perception paradigm.

2. Into the symbolic by the bootstraps

meaning can be described, accounted for, or stated in terms of symbols or language at its highest and most complex stage of development (the stage it reaches in human experience), but language simply lifts out of the social process a situation which is logically or implicitly there already (Mead, 1934: 79)
.

The steps taken at the beginning of the second year of life in this process of developmentally explicating meanings are both profound and complex. What occurs is both the construction of the symbolic domain from the previously established functional communicative one ; and the construction of a propositional frame within which these symbols are handled. These constructions feed into each other, but follow separate courses. Mead's notion of the 'significant symbol' is a useful one here:

significant symbols carried back to their origins prove to be gestures, i.e., parts of social acts through which individuals adjust their conduct to that of others. They become symbols when the act which they preface is aroused as an attitude in the other individual. They become significant symbols when the individual that uses the gesture which calls out such an attitude in another calls out the same attitude in himself (Mead, 1938: 221-2).

Note that the distinguishing characteristic of the significant symbol is it has the same meaning for both participants in an act.

The point Mead strives for in his distinguishing of 'significant' (self-conscious) from 'non-significant' (non self-conscious) gestures is that the former gain their meaning in relation to a perspective or stance that is located outside that given 'naturally' to the organism that uses them. This perspective is not that of the first person (the 'gesturer') nor second person (the 'gesturee') participants in the symbolically-communicative act, but that of the third person (either an objective observer, or one intersubjectively constituted as 'both of them' (first- and second-person interactants: i.e., the 'we' who agree)).

For example, consider an act such as 'giving'. If A gives something to B, the object goes away from A towards B. Similarly, if B gives something to A the object goes from B to A. But if a third person, C, triangulates on this act, the transferred object need have no motion towards or away from him or her. As a neutral observer, C has in addition no attitude viz-a-viz possession or loss: C can conceive of give, and not count the cost. A and B can only do this once they have developed a deal of detachment, via symbolic forms that mediate that development.

However, the attainment of 'significance' is a slow process, best captured by the Piagetian conception of 'décalage'. For even once the child is into the symbolic, not all symbols have this status of significance. For example, neither my 4 and 6 year old son and daughter - who have the benefit of being able to use the verb 'give', apparently appropriately - can interactively 'give' in this significant way, for giving still involves a 'loss of possession' of an object that is in their current project space. They are still on the road to significance in this interactive domain, even at the age of 6. But in other areas, they clearly have attained a mastery of significant symbols. They and I can talk about the colour of something, our words gaining their meanings for us from the perspective of our act of agreement, our shared perspective rather than my perspective or their perspectives. That shared perspective is constituted by the move into the referential domain during the child's second year, when they begin to construct propositions. This propositional frame or ability has to be developmentally constructed.

3. Constructing the referential-propositional mode

At the end of the first year, a human child has an elaborated way of making demands. They have controlled ways of expressively communicating relationships that hold between themselves and objects in the world. They can engage another's agency in bringing about certain changes in those relations. They are able to deictically delineate objects in the environment, particularly by pointing towards them. But, these abilities are not propositional. That is, they do not explicitly 'state' anything about the relationship between objects out there in the world. Rather, they convey relationships between objects and the implied point of perspective of the infant's perceptual world. Structurally, these early communications convey the relation between an implied self and a real object.

In addition, most infants can use the deictic gesture of pointing within a shared indicative frame. Both the nature of this ability and how it is accomplished are still quite mysterious: but the ability to mark and co-ordinate shared attention upon a feature in a shared intersubjective world from a significant perspective appears to be a hallmark of being human. Structurally, pointing conveys something like 'look at that'. But it is soon co-opted into the 'naming game' where it constitutes reference and conveys identity. In its early stages naming would appear to have two characteristics First, it 'asserts' identity (That thing I am pointing at/looking at is an X). Second, it does so neutrally with respect to any relationship between the infant and the object (for example, 'that' object is not in any marked relationship - such as +/- want - with respect to the infant). Once naming games are started, they expand into new territory, because some of these neutral relations are not equally 'neutral'

A 'fire', for example, is not apprehended by the mythical, detached, reflective and disembodied subject of the Cartesian tradition, but by a real embodied infant who 'knows' the object viscerally as well as visually. Put an infant in the 'original word game' with a picture book, and a 'word' such as 'dog' might identify a picture. Put an infant in front of a fire and while the word 'hot' might identify the object, it is also a quality - hot - which is felt when in front of a fire. The same could be said with respect to other apparently 'neutral' relations: if a child says 'on' when pointing at a light, is he or she 'naming' the object or potentially predicating a state to it?

A similar situation holds in the case of early demands. The 12 month old infant clearly has an implicit grasp of another's agency. As such, a demand is a pivotal point between the infant's non-propositional ability to 'state' a relationship between his or her implied self and an object or event - (I) want that - and predicating the object into a relationship with another person - where 'that' is implicitly qualified by (you do), i.e., (you do that).

A third such case is when infants 'say things' about what they do, such as 'up' and 'down' when they get up and/or down from something. In the context of their other capabilities there is nothing particularly difficult to comprehend in their being able to do this, for the word is being used in relation to their implied self and what they are actually doing. And, in common with other 'words', 'up' and 'down' come to be used in demands. This can happen because the child already has a structural demand frame via which to relate the world to the implied self: words are co-opted into demands.

In the same way, objects that are acted on come to be announced: 'truck', perhaps, when picking a truck up: the object can be talked about in relation to the implied self because the frame is available. And, conversely, 'motion' words are used in cases where the infant is the actor and object of the change being 'announced' is the recipient of the infant's actions: 'down' may be said as the truck is put down. Here, words have come to be used in a situation where the child had already established some articulable conception of changes in position (from their commenting on their own changes in position); some articulable grasp of the object's identity; and the situation bears a relation to the implied self through its being the locus of the action.

What are being articulated here are very small bootstrapping steps from one ability to another: an explication of previously implied possibilities. And the point which has been reached in the above description is, I suggest, a momentous one, for given the explicit establishment of these abilities, then all the components are in place for a new implication to be made explicit and thereby constitute the propositional frame. That is, these particular early word uses are transitional between egocentrically-rooted predication and allocentrically-rooted predication.

Once this transitional point has been reached, then these developments are sufficient in concert to open up the new possibility of action being anchored around objects per se - independently of the child's involvement with them: sufficient for changes in his or her framing of action to occur. That possibility enables the perception of both the changes in state and position they experience, and those they cause an object to undergo as instances of an equivalent event. In other words, the original ability implies a propositional one - the child's doing something to an object (non-propositional) implies the object is having something done to it (propositional). In Piagetian terminology, the child demonstrates a very early form of transitive inference: if I am putting A down, then A is going down. The child may now be expected to begin naming objects that are being acted on irrespective of the actor, and similarly also the actions of agents (or implicit selves) other than themselves.

4. Explicating further implications

Having arrived at the propositional frame, infants can co-opt it to give explicit form to further features that were originally implicit in previous productions. That is, they may begin to talk about non-self agents, whose actions upon objects they can already handle. Through coming to do this they will explicate the possibility of 'referring' to their implied selves, and enable some reflexive and objective existence of their position within their perceptualization and conceptualisation of the world. The developments just outlined are summarised in Figure 2.

Insert Figure 2 about here.

Evidence for this developmental sequence actually is given by Greenfield and Smith's (1976) data on the emergence of first words for one child called Nicky:

(1) Prior to their observation session II, 18 (27), Nicky is able to

    (a) label objects;

    (b) label some of his own actions

    (c) use these words in the pursuit of his own intentions.

(2) At the time of session II he evidences three new abilities, being able to use utterances labelling:

        (a) certain actions he performs on inanimate objects, or actions he performs when an object other than himself is implicated in that performance - for example, down while pulling his train down;

        (b) the state of certain objects; for example, on while pointing at a light which is switched on;

        (c) the inanimate objects of his own direct actions; For example, milk when he picks up a glass of milk; bap when he picks up his diaper.

(3) After session II he labels:

        (a) certain actions that others perform ; whether upon an object or not; for example bang when another child bangs his head (session VI, 22, (22) );

        (b) again in session VI, 22 (22), the inanimate objects of other people's direct actions; for example, bed when mother is making the bed; potatoes when she is putting them in a dish.

The claim here is that this sequence of development in Nicky's 'language' is not arbitrary, but fixed: that it could not occur in any other order, and that each new ability has definite antecedents in the child's previous abilities, which determine the sequence by the hierarchy of implications they engender.

5. The status of these early 'words' viz-a-viz preverbal gestures

The child develops a pre-linguistic gestural system that, with contextual support, can function for deictic ends. These 'non-verbal' gestures can become more explicitly encapsulating of their 'meaning' and thus rely less on contextual support for their communicative efficiency (Acredolo and Goodwin, 1985; Zinober and Martlew, 1985). What would be the difference between the child's request in Figure 3 if she added the sounds 'drink' or 'beaker' to her action? Would these be 'words' (a category different from gestures), or just specific 'gestures' in a different modality (a category the same as gestures)?

Insert Figure 3 about here

Again, consider two examples from my daughter at 16 months. First, she used a 'gesture' of raising her hands above her head and pulling them down on hearing her mother drive the car up to our front door, when she herself was with me in the kitchen, and so could only hear the event. She had previously used this 'gesture' in the context of wanting boxes and jars opened or to look in cupboards that she couldn't open. Why she should have adopted this gestural form in these contexts was a mystery at this point. Thus, if there were any relation between her using it now and what she had heard - and since she had looked to the direction of the sound of the car as she produced her gesture; and attention checked back to me; and raised her arms so as to be lifted down to the floor from high chair (which is different from the previous gesture in that here while the arms are raised they are not pulled down, but left up so as to be picked up); and then toddled off to the front door as soon as she was on the ground; then there appeared to be a relation in her view of things - even if it was not a relation that was immediately apparent. On getting outside she greeted her mother, and then followed me toward the garage to come and stand next to me after the car had been driven in there. She looked at the overhead garage door, then at me, then stood on tiptoe, reached her arms over head and pulled them down. I then reached up on tiptoe, grasped the door and swung it down to close it. I had previously done this exaggeratedly with her as a spectator. This would appear to be the hitherto unrecognised source of her idiosyncratic gesture related to changes of enclosure, and the link that relates it as appropriately used when 'telling me' that mommy had just arrived home in the car.

Second, and later that same day, we heard a dog barking in the distance when she was in her bedroom. She pointed in the general direction and said something like 'doggie', a word previously used in the context of pictures of dogs in books. The link between the two contextual supports was more readily apparent because in this case it is established and mediated by a standard as opposed to idiosyncratic 'gesture'. Now, are the abilities underlying these two productions the same? Is there a continuity between her gestural and her spoken communication? Such flimsy anecdotal evidence would suggest that there is. Given a more solid data base as their corpus, Bates et al (1979: 177) come to a similar view, that in these early stages of elaborating a symbolic communicative ability, the domains of the vocal and gestural are fundamentally equivalent:

At this point in development, the only difference between the two domains is in the modality of expression. Indeed, we see no evidence to suggest that a 13-month-old is in any way biased toward the development of vocal language as opposed to gestural language.

6. Combining 'words'

Putting 'words' together has been seen as the point of entry into language proper, since words put together have order, and ordering principles are the stuff of grammar. The field presently seems happier to view the early combinations that children produce as not being informed by grammar (that is, not being linguistic?), but perhaps being an ability that can be re-worked developmentally so as to provide a starting point for the construction of grammar. At the same time, however, there are a number of 'discontinuity' views argued for, with the point of 'discontinuity' being variously assigned. For example, Pettito (1987) denies continuity between gesture and 'word' communicative systems in ontogeny; others see a discontinuity between early word combinations and later ones that occur 'after grammar has kicked in'.

Let me preview the argument to be made here. It is based in the view that any developmental system is multi-determined in the course of its elaboration. This means that it is possible for a development 'downstream' to be both continuous with earlier achievements and discontinuous with them simultaneously. Thus, there is a sense in which the ability to combine words is continuous with the earlier ability to combine gestures. There is a sense it which it is discontinuous, since gestures can be combined simultaneously while words can only be combined sequentially. There is a sense in which words are continuous with earlier gestures, the shift in modality being unimportant. And there is a sense in which these later verbal gestures are discontinuous with the earlier ones, in that the latter have a realised propositionality that was only implicit in the former. Clear-cut distinctions thus exist in the rhetoric of the theorist rather than the processes at hand.

The child develops a pre-linguistic gestural system that, with contextual support, can function for deictic ends. On occasion, these gestures are supplemented with others. Lock (e.g., 1980) has hypothesised that gestures first occur in sequences before being contemporaneously combined. Later, gestures accompany words, which are now being considered as 'verbal gestures'. Developmental changes have been noted with respect to the relation between verbal and non-verbal gestures. For example, Zinober and Martlew (1985) found that prior to 21 months their subjects used gestures in a supportive role to words, in that word and gesture have the same apparent meaning. After 21 months, gestures complemented words, in that a child might say 'book' while placing her mother's hand on the book:

the utterance 'book' gives the name of the object being negotiated. The gesture indicates who shall perform the action (the mother) and what action is required (picking up the book). Together, [word] and gesture communicate 'you pick up the book, mummy' (Zinober and Martlew, 1985: 200).

Again, 'words' themselves come to be combined without accompanying gestures. The question is whether these developments of gestural combination, gesture-word combination, and word-word combination are related to each other?

7. Some relevant data

The kind of progression seen in early gestures - from sequential to combinatorial - has been noted in other studies with respect to the early combining of words. For example, Greenfield and Smith (1976) note for the two children they studied that the transition from one- to two-word speech did not occur abruptly, but that there was a transitional period in which the children were producing two one-word utterances sequentially, with a gap of 1.1-1.4 seconds separating each word. They note that

a number of earlier investigators of child speech (e.g., Guillaume, 1927; Stern, 1930; Leopold, 1949; Cohen, 1952) have noticed the occurrence of sequences of single-word utterances as a later stage within one-word speech (1976:41).

Scollon (1976) introduced the notion of 'vertical constructions' within dialogue as being linked with these sequential developments. Thus the child may say 'dog', and mother replies with the question 'What's the dog doing?', leading the child to say 'woof'. Elide the mother's part, and here is another source of the linking of one word with another, in this case within a propositional frame. Other than this frame, however, there is little difference between such a vertical construction and the one captured in Figure 4, where the child obtains the mother's attention non-verbally, mother asks 'What do you want', and the child replies by deictically specifying 'that'.

Insert Figure 4 about here

Table 1 contains utterances from Greenfield and Smith's (1976) corpus for one child, Matthew. Utterances 1-5 occurred in their observation V when Matthew was 17(13) (months/days) old; and utterance 6 during observation session VII when he was 19(21). They are all demands, an ability which is elaborated in the earlier, pre-linguistic phase. The continuity between these abilities has been pointed to by de Laguna (1927) in her discussion of what she termed 'predicative proclamation' :

What is proclaimed in (predicative proclamation) always has reference to some object or event or general state of affairs whose existence is presumed. This presumption may take the form of pointing to the thing in question, or perhaps of intently regarding it. The predicative proclamation does not announce its presence or existence, but calls attention to some specific property having a bearing on the given situation. In such a case there is virtual or implicit predication; but the language form is rudimentary. The verbal utterance must be supplemented by some other form of bodily response, like pointing, which serves to indicate the object to which the verbal specification applies . . .. What is needed to transform this rudimentary predication into the full-fledged sentence, is that the act of pointing or otherwise indicating the object be replaced by an act of speech, the utterance of a word (De Laguna, 1927: 98-9).

Insert Table 1 about here

A real continuity is apparent in the examples given for Matthew: and as Bloom has observed, there is nothing intrinsically linguistic or grammatical about such early combinations:

Linguistically, the potential juxtaposition of . . . "certain words" . . . with other "words" could generally be predicted from the context in which they occurred. ... It was concluded that Allison's early sporadic two-word utterances were not manifestations of any underlying rule-system but, rather, were a manifestation of the essentially transitive nature of the notions coded by the particular forms "more';, "away", "stop", etc., to the nonlinguistic states of affairs to which they referred. SOMETHING had to STOP; there had to be SOMETHING TO RECUR; SOMETHING DISAPPEARED, etc. (1973: 114).

There is, however, another facet of this combinatorial process that is being masked here. That facet is that not all two-word utterances are the same. Werner and Kaplan (1963) distinguished two types of early word combinations, summarised by Leonard (1976: 43) as follows:

one where the expression of the word combination could be likened to a single word expression since the individual element overlaps too extensively to form a true grammatical relationship, perhaps resembling functional relations, and the other representing word combinations where the two elements of the utterance emphasise different features of the presented event and approach a sentential relationship.
Werner and Kaplan pointed out that the former type of utterances seem to appear before the latter, and this claim holds up in data reported by Bloom (1973), Brown (1973), Greenfield and Smith (1976), and Leonard (1976). This difference in timing of emergence would appear to be based in the need to construct the propositional frame discussed in the previous section. If we use De Laguna's terminology, combining words in predicates occurs before combining them across propositions, for in the former it is not necessary to have constructed an ability to relate something other than the self as a topic or subject about which something can be said. But while there is an underlying discontinuity, in that there has been the construction of a new ability, the outputs from this developmental strand appear to be immediately recruited into the combinatorial ability. Thus Greenfield and Smith (1972: 113) note that two of the categories that are explicated at a later time compared to others - what they term 'Agent of Action' and 'Action of non-self Agent' - immediately appear in two-word combinations:

two-word Agent-Action utterances occur [for both Nicky and Matthew] as soon as the Action of other Agents becomes a productive relation. Agent-Action sequences are non-existent in Nicky's corpus, rare in Matthew's.

This suggests that the combinatorial ability is a general one that has a developmental continuity and can be applied to combine content items that emerge from a different strand of development where the endpoint of the strand, through an incremental process of small changes, appears as a developmental discontinuity.

What happens next goes beyond the scope of this paper to deal with in detail. Is it the case that innate knowledge or a language module comes into play at this developmental point? There appears to be no clear answer at the moment, but there are glimmerings of a framework for one. First, there is neurological evidence that around the time 'grammar' comes in - that is, when previously equivalent combinatorial abilities underwriting word, gesture and object combinations begin to show subsequently separable paths of development - there is a simultaneous 'rewiring' of relevant parts of the developing brain, as noted by Greenfield (1990/1991):

Broca's area ... provide[s] a common neural underpinning for early programs of action in speech and tool use. These programs differentiate from age two on, when Broca's area establishes differentiated circuits with the anterior frontal cortex.

Second, the child is developing in a context in which the previously elaborated products of previous ontogenetic histories are culturally conserved in the child's environment. Bechtel (1993: 140) puts it this way:

Languages, as cultural products, have evolved devices providing compositional syntax and semantics; humans, in learning these languages, have developed capacities for decoding information encoded using the compositional syntax and semantics of these languages.

In this way, it becomes possible to see how a coherent account of 'acquiring a language' might be generated that does not invoke pre-given syntactic knowledge. Bechtel (op.cit.: 135-8) refers to St. John and McClelland's (1990) demonstration of competent sentence processing by a connectionist model, commenting that this 'network knows how to extract information from grammatically structured sentences, but in order to do this it does not have to have an internal representation of the sentence upon which computations can be performed' (op.cit.: 138). Thus, the key proposal is to:

view the human cognitive system as having adapted to a linguistic environment and to find the source of productivity and systematicity in the external symbols of the language (op.cit.: 140).

Third, Bamberg (1996), drawing on Capps and Ochs (1995a, 1995b) takes this tack further in his study of emotional socialisation with respect to children's discourses about anger and sadness. What he terms the 'grammars' of being angry and sad, in talking from both the first-person and third-person perspectives, are quite complex in English, and work so as to disentangle the different discursive requirements of handling empathy and blame. Bamberg sets his particular findings into a general framework that sees the development of a linguistic ability to handle 'emotion talk' as 'a process of appropriating the tools necessary to talk meaningfully about the social relationships in which emotions are embedded'. Thus, grammar is set up as 'a set of linguistic constructions is directly tied to the discursive purpose to which the particular constructions are put to use':

grammar, if understood correctly, i.e. not as abstract principles of a universalist nature, but as social know-how relevant for the construction of social meaning, plays an integral role in coming to grips with what emotions do and what they are used for in social communicative practices. As such, learning to use the linguistic construction procedures for socially appropriate purposes is deeply embedded in cultural practices (Bamberg, op.cit.).

Thus, putting these points together, it may be that, as Bickerton, for example,(1990: 180-1) has pointed out, 'syntax is, to a large extent, a projection of the lexicon, wherein all the sub-categorisation frames of verbs and all the grammatical items (and much else) are stored.' But, rather than attributing the existence of these sub-categorisation frames and the like to pregiven syntactic principles, as Bickerton and other 'autonomous linguists' do, development might be occurring from different bases. It could well be that grammar emerges out of the child's struggle to keep track of the increasing number of implications of earlier communications that come to be given an explicit form as development progresses. It could well be that the form of grammar results from learning mechanisms - constrained by the working mode of a differentiating Broca's area - finding ways to deal with what is constituted in the Umwelt that concurrently 'feeds' them their 'data'. That Umwelt is the child's socially-mediated grasp of a world in which the products of previous ontogenies are conserved as if they were the feedforward products of previous efforts by the same learning mechanisms, now feeding back as the input to those mechanisms. Thus, the emphasis can be shifted from the cognitive perspective of there being some abstract object ('Language') that is analysed by some 'Language Acquisition Device', to one in which the learner is the generator of, and participator in, shared acts of 'making sense'.

Conclusions

The above analysis distinguishes between the attainment of productive symbolic abilities and the attainment of productive propositional abilities. The developmental move from pre-propositional to propositional may be the one that other species find so difficult. At this level of analysis, both continuity and apparent discontinuity emerge as characteristic of the early phases of communicative development (that is, approximately the first 2 years of life). The continuities are that:

        1. the early appearing 'words' are in essence more specific 'gestures' which tend, in hearing children, to develop in the vocal domain rather than any other;

        2. the ability to combine these 'gestures' appears to be a general one, and as such is continuous across the period from its emergence in the combination of pre-symbolic gestures into the two-'word' 'stage'; and

        3. the ability to 'bootstrap' implications into explicit forms underlies the elaboration of both pre- and post-verbal 'gestures', irrespective of their production modality.

The discontinuities are that:

        1. as a result of small incremental steps, later abilities have a propositional character which sets them off from the earlier ones (while recognising that it is the appearance and use of the earlier ones that has enabled the construction of the later ones). This discontinuity is one between the nature of the start and end points of an otherwise continuous chain; and

        2. 'referential' vs 'demand' modalities appear to have ontogenetically distinct roots.

Further, there is little in the nature of these combinatorial abilities that would require one to invoke a notion of '(syntactic) grammar' in explaining them (though psycholinguists might like to use such a notion in describing them). Order is based on functional considerations, such as what can or cannot be intersubjectively presupposed; what is the focus of the child's current perceptual interest; what is pragmatically emphasised, etc.. (At a more fine-grained level of analysis Bates, Bretherton and Snyder, (1988) articulate a level of complexity in continuities and discontinuities that is at least compatible with the above.) At the same time, that there are 'elements' that can be combined rests on a constitutive process of coming to be able to 'grasp' possible distinctions that are available to the child through the elaboration of the child's Umwelt at any particular point in time. However, these implications exist in a particular sequence of logical relations that are rooted in the 'value perspective' entailed by being an infant. The same sequence extends out from the value perspective of even chimpanzee infants. Where human infants differ is in being able to explicate and give form to these implications. This redefines the task of developmental psycholinguistics as one of specifying the 'mechanisms' that enable this to be done.


References

        Acredolo, L. and Goodwin, S. (1985) Symbolic gesturing in language development. Human Development 28: 40-9.

        Bamberg, M. (1996) Language, concepts, and emotions: the role of language in the construction of emotions. Visting Scholar Series 6. Department of Psychology, Massey University, Palmerston North, New Zealand.

        Bates, E., Benigni, L., Bretherton, I., Camaioni, L. and Volterra, V. (1979) The emergence of symbols: Cognition and communication in infancy. New York: Academic Press.

        Bates, E., Bretherton, I and Snyder, L. (1988) From first words to grammar. Cambridge: Cambridge University Press.

        Bechtel, W. (1993) The case for connectionism. Philosophical Studies 71: 119-54.

        Bickerton, D. (1990) Language and Species. Chicago: University of Chicago Press.

        Bloom, L. (1973) One word at a time. The Hague: Mouton

        Brown, R. (1973) A fist language: The early stages. London: George Allen and Unwin.

                Bruner, J. S. (1990) Acts of meaning. Cambridge, Mass.: Harvard University Press.

        Capps, L. and Ochs, E. (1995a) Constructing panic: The discourse of agoraphobia. Cambridge, Mass.: Harvard University Press.

        Capps, L. and Ochs, E. (1995b) Out of place: Narrative insights into agoraphobia. Discourse Processes 19: 407-439.

        Chomsky, N. (1968) Language and mind. New York: Harcourt.

        Clark, A. (1997) Being there: Putting brain, body and world together again. Cambridge, Mass.: Harvard University Press.

        Cohen, M. (1952) Sur l'étude du langage enfantin. Enfance 5: 181-249

        de Laguna, G. A. (1927) Speech: It's function and development. New Haven: Yale University Press

        Elman, J. L., Bates, E. A., Johnson, M. H., Karmiloff-Smith, A., Rarisi, D. and         Plunkett, K. (1996) Rethinking innateness: A connectionist perspective on development. Cambridge, Mass.: M. I. T. Press.

        Greenfield, P. M. (1991) Language, tools and brain: The ontogeny and phylogeny of of hierarchically organized sequential behavior. Behavioral and Brain Sciences 14: 531-95.

        Greenfield, P. M. and Smith J. (1976) The structure of communication in early language development. New York: Academic Press.

        Guillaume, P. (1927) Les débuts de la phrase dans la langage de l'enfant. Journal de Psychologie 24: 203-229.

        Hendriks-Jansen, H. (1996) Catching ourselves in the act: Situated activity, interactive emergence, evolution, and human thought. Cambridge, Mass.: Harvard University Press.

        Hewes, G. W. (1996) A history of the study of language origins and the gestural primacy hypothesis. In A. J. Lock and C. R. Peters (eds.) Handbook of human symbolic evolution. Oxford: Clarendon Press. Pp.: 571-95

        Ingram, D. (1981) The transition from early symbols to syntax. In R. Schiefelbusch and D. Bricker (Eds.) Early language: Acquisition and intervention. Baltimore: University Park Press.

        James, W. (1956) The will to believe. New York: Dover Books.

        Johnson, C., Davis, H. and Macken, M. (1996) Symbols and structures in language acquisition. In A. J. Lock and C. R. Peters (eds.) Handbook of human symbolic evolution. Oxford: Clarendon Press. Pp.: 686-746.

        Leonard, L. B. (1976) Meaning in child language. New York: Grune and Stratton.

        Levins, R. and Lewontin, R. (1985) The organism as the subject and object of evolution, In R. Levins and R. Lewontin (Eds.) The Dialectical Biologist. Cambridge, Mass.: Harvard University Press.

        Lewontin, R. C. (1978) Adaptation. Scientific American 239: 156-69.

        Lewontin, R. C. (1982) Organism and environment. In H. C. Plotkin (Ed.) Learning, Development and Culture. Chichester: Wiley.

        Lewontin, R. C. (1983) Gene, organism, and environment. In D. S. Bendall (Ed.) Evolution from molecules to men. Cambridge: Cambridge University Press.

        Lock, A. J. (1980) The Guided Reinvention of Language. London: Academic Press.

        Lock, A. J. 1985) Processes of change and the elaboration of language. In G. Gollin (Ed.) The comparative development of adaptive skills: evolutionary implications. Hillsdale: Erlbaum. Pp.: 239-70.

        Lock, A. J. and Peters, C. R. (1996) Ontogeny: symbolic development and symbolic evolution. In A. J. Lock and C. R. Peters (eds.) Handbook of human symbolic evolution. Oxford: Clarendon Press. Pp.: 371-99.

        Macmurray, J. (1961) Persons in relation. London: Faber and Faber.

        MacNamara, J. (1972) Cognitive basis of language learning in infants. Psychological Review 79: 1-13.

        Maturana, H. and Varela, F. (1980) Autopoesis and cognition: The realization of the living. The Hague: D. Reidel Publishing Co.

        Mead, G. H. (1934) Mind, self and society. Chicago: Chicago University Press

        Mead, G. H. (1938) The philosophy of the act. Chicago: Chicago University Press.

        Nelson (1973) Structure and strategy in learning to talk. Monographs of the Society for Research in Child Development 38 Serial No. 1249.

        Patten, B. C. (1982) Environs: Relativistic elementary particles for ecology. American Naturalist 119: 179-219.

        Pettito, L. (1987) 'Language' in the prelinguistic child. In F. Kessel (Ed.) The development of language and language researchers. Hillsdale, N.J.: Erlbaum

        Piaget, J. (1951) Play, dreams and imitation in children. London: Routledge and Kegan Paul.

        Ramer, A. (1975) The merging of the communicative and categorial functions of language. Paper presented to the American Speech and Hearing Association, Washington, D.C.

        Scollon, R. (1976) Conversations with a one-year-old. Honolulu: University of Hawaii Press

        Shotter, J. (1981) Vico, moral worlds, accountability and personhood. In A. J. Lock and P. L. F. Heelas (Eds.) Indigenous psychologies: The anthropology of the self. London: Academic Press. Pp.: 265-84

        Sinclair-de-Zwart, H. (1973) Language acquisition and cognitive development. In T. Moore (Ed.) Cognitive development and the acquisition of language. New York: Academic Press. Pp.: 9-26.

        Slobin, D. I. (1970) Universals of grammatical development in children. In G. B. Flores d'Arcais and W. M. Levelt (Eds) Advances in psycholinguistics. Amsterdam: North Holland. Pp.: 174-86.

        

        Smedslund, J. (1969) Meanings, implications and universals: Towards a psychology of man. Scandinavian Journal of Psychology 10: 1-15.

        Smith, L. and Thelen, E. (1993) A dynamic systems approach to development: Applications. Cambridge, Mass.: Harvard University Press.

                St. John, M. F. and McClelland, J. L. (1990) Learning and applying contextual constraints in sentence processing. Artificial Intelligence 46: 217-57.

        Stern, W. (1930) The chief periods of further speech development. Reprinted in A. Bar-Adon and W. Leopold (Eds) (1976) Child language: A book of readings. Englewood Cliffs, N.J.: Prentice-Hall. Pp.: 45-51.

        Thompson, E., Palacios, A. and Varela, F. J. (1992) Ways of coloring: Comparative color vision as a case study for cognitive science. Behavioral abd Brain Sciences 15: 1-74.

        Thompson, E. (1995a) Colour vision, evolution, and perceptual content. Synthese 104: 1-32.

        Thompson, E. (1995b) Colour vision: A study in cognitive science and the philosophy of perception. London: Routledge.

                von Uexkull, J. (1957) A stroll through the world of animals and men. In C. H. Schiller and K. S. Lashley (trans. and eds) Instinctive behavior: The development of a modern concept. New York: International Universities Press

        von Uexkull, T. (1982) Introduction: Meaning and science in Jacob von Uexkull's concept of biology. Semiotica 42: 1-24.

        Vygotsky, L. S. (1978) Mind in society. Cambridge, Mass.: Harvard University Press.

        Werner, H. and Kaplan, B. (1963) Symbol formation. New York: Wiley.

        Zinober, B. and Martlew, M. (1985) The development of communicative gestures. In M. Barret (Ed.) Children's single word speech. Chichester: Wiley. Pp.: 183-215.