How Did They Do It?
Most children learn language with remarkable ease, but how are we to account for this extraordinary fact? The problem plaguing our understanding of language and language acquisition can be described as. How can one learn anything genuinely new and become linguistically creative and how this learning is possible at all, unless one already has some path into language, for example, a suitable framework in which language learning takes place? It is this framework that interests us here.
One possible picture is provided by St. Augustine, who likens the child's learning of language to a stranger coming into a foreign land, unable to understand what is said, yet already in possession of some language, only not the one spoken 'here.' To Wittgenstein, the picture painted in St. Augustine's Confessions is not representative of the scenery encountered by the first-time language learner, for this stranger who slowly decodes the puzzle of the strange surrounding sounds already has a framework. St. Augustine alters the character of the learner so that the issue of a suitable framework does not arise. Philosophers like Chomsky or Fodor, although historically distant from Augustine, try to provide a new answer to the same question. Their solution differs only in the sense that it shifts the problem onto a 'universal grammar' or a 'language acquisition device', which thus provides the entry point into language.
In his effort to dissolve philosophical issues, Wittgenstein makes a great many methodological suggestions in his later work, criss-crossing language, meaning, thought, and so forth. We also find a number of remarks about the early stages of language learning, where, according to his suggestion, at first simple forms of performative training occur between teacher and pupil, void of explanation on the part of the teacher, and void of understanding on the child's. Only later in the child's development, when common judgements have been absorbed, does an understanding of language displace merely conditioned behavioural responses. When the linguistic performances of the child indicate that there is not only imitation but also understanding, perhaps best demonstrated by the ability to teach the use of linguistic conventions, the child can be said to properly follow rules in their usage. What is troubling with Wittgenstein's suggestions, both to him and to us, is that at least some of his remarks are clearly hypothetical in the sense he wishes to avoid.
In Child's Talk, Jerome Bruner's points out that the lexico-grammatical training of quasi-Wittgensteinian language-games does play a role in the child's linguistic development, but that it is preceded by systematic and abstract pre-linguistic forms of communication of equal or even greater importance, which, together with certain cognitive endowments and suitable social encouragement, first provide the support needed for the child to develop linguistic skills. Pace Wittgenstein, Bruner suggests that children typically need not be drilled as suggested, but that they have a natural tendency towards both mutual attention and agreeing on reference ("referential intersubjectivity", Bruner 1983, 27, 122), which is augmented by a device to acquire language. Children are treated by their caretakers as proper partners in a hermeneutic process of communication as if the infant understood. However, Bruner cannot make these claims without also introducing a number of cognitive endowments, which act as guarantors for proper language learning, roughly equivalent in function to the performative training Wittgenstein's children undergo.
To Wittgenstein, we do not use language according to strict rules, nor is language taught to us by such rules outside of some highly specialized applications (Wittgenstein 1958, 25). Bruner agrees with this anti-formalist attitude. Language acquisition, to him, is quite literally the entrance into Wittgensteinian forms of life (Bruner 1983, 62). Antithetical to this, Bruner hypothesizes extensively about children's ability to learn language in what he openly calls "the a priori side of it" (Bruner 1983, 122). In Child's Talk, we thus encounter a mixture of cognitive endowments children are said to have, biological competence, along with various forms of social encouragement, cultural context, both of which go hand in hand with a child's cognitive unfolding. To him, infants have a natural disposition towards systematicity and means-end abstraction, which is augmented by the constrained, familiar situations of action and interaction which support a high degree of order. Infants are extraordinarily social and communicative, showing a Bühlerian Funktionslust: they love to play (Bruner 1983, 47). Children eagerly respond to others' actions, spending many hours playfully doing a limited number of things with their caretakers, repeatedly going through the child's ever-growing repertoire of games, responding to prompts, exchanging smiles, and so forth. Underlying a child's approach to the world is a nave realism: children do not question the existence of mental phenomena, other beings or physical objects. At first they cannot distinguish between thoughts and things, or between others' and their own mental activity. But even extremely young children have communicative intentions that they try to make clear through constant and repeated negotiations. Over time, such negotiations lead to ever more complex communicative and then to recognizable linguistic procedures, to language-games. Language learning consists not only of learning grammar, but "of realizing one's intentions in the appropriate use of that grammar" (Bruner 1983, 38). To account for the extraordinary fact that children learn language, Bruner posits some Language Acquisition Device (LAD), in mutual response to which the social environment acts as a Language Acquisition Support System (LASS). It is with this a priori side that I think we must take issue. Call LAD and LASS what you will, according to Bruner the specific nomenclature does not matter. He has little difficulty in asserting such functional mechanisms, perhaps because they replace the idea that language 'grows out of' proto-phonological, -syntactic, -semantic, or -pragmatic knowledge with the idea that, before any LAD can 'generate linguistic hypotheses,' the primitive procedures found in pre-linguistic communication must function adequately. This entails a sensitivity towards, for example, "a patterned sound system, to grammatical constraints, to referential requirements, to communicative intentions, etc." (Bruner 1983, 31). It is this sensitivity or disposition that 'grows' into place, not the finished product.
Wittgenstein would not approve of Bruner. He falls victim to his dictum that what should be done is not to attempt to define or to explain, as Bruner has done with his hypothesis about the underlying mechanism or structure of language learning, which will generate further instances of metaphysics in the narrow conception Wittgenstein has of it. The assertion of a modified Chomskian view in Bruner's Child Talk, apart from the fact that its explanatory benefit is circular, may create more problems than it solves. The introduction of a device that helps us acquire language seemingly serves the purpose of providing an explanation where none more intelligible can be found deus (lingua) ex machina. Although a response to the Augustinian picture of language learning, this is not much more helpful.
To Wittgenstein, linguistic practice is like game-playing in that it cannot be put into precise terms: "imagine people amusing themselves in a field by playing with a ball so as to start various existing games, but playing many without finishing them and in between throwing the ball aimlessly into the air ..." (Wittgenstein 1953, sect;83). Would it make much sense for someone to say that when these people are playing a ball-game that they are following definite rules at every step? No. There is a family resemblance among these (language-)games, but no single common element. But language-game playing also requires more than arbitrary actions. It is not possible for one person to obey a rule only once, since rules are customs and not singular events. As pupils, to Wittgenstein, we don't learn our language calculus any more arbitrarily than we accept our fear of fire. In linguistic convention, in playing language-games, there is agreement not of opinion, but, more fundamentally, on a shared form of life (Wittgenstein 1953, sect;241), which is rule-governed.
Without language, to Wittgenstein, we could still communicate, but we could not influence one another in such-and-such ways; we could not build roads and machines, for these activities all require a sophisticated grasp of language. Would we even know what it is to exist without language? This implies a certain discontinuity between those who have mastered language and those who seemingly have not (Wittgenstein 1953, sect;207). Regrettably, those 'who seemingly have not,' are largely ignored by him. They do not fit into his methodology, because they fail to partake in the corollary to his later work, that is, in his overarching linguistic framework within the confines of striking rock-bottom. Hence the sceptical denial that the seemingly ordered but indecipherable utterances of a group of people should be called language. Language is the vehicle of thought (Wittgenstein 1953, §329) in at least the crucial sense in which we ground our communities qua communities: if there is no discernible understanding of language, we can never know whether we share in the same meanings, expectations, agreements, and so forth. Thus, an understanding of language has become a measuring stick for what it means to be another human being, a person.
How do children enter our world, learn a language, according to Wittgenstein? An important part of early language acquisition is the learning of the correlation between objects and names, for they are propaedeutic to understanding language, preparatory to the use of a word. In the primitive language (2) of Investigations, for example, one party calls out the words, the other repeats them. A command like "Slab!" in the practiced use of the language elicits the response that a slab is brought. Still simpler, the pupil merely repeats the words after the teacher in context. To Wittgenstein, these linguistic drills are those forms of language through which a child begins to make use of words, not yet understanding them. More sophisticated concepts like "here" or "five" are learnt later, when the child already has mastered some basics. Learning to name is not a move in a language-game any more than putting a chess piece in its place is a move in chess (Wittgenstein 1953, sect;49). But in order to be able to play chess, the board must have been set up, ostensive training towards a proto-understanding of words must have taken place. It is not yet enough to have been trained to utter certain sounds on command to understand language, but it is needed to establish the practice which points at the meaning of words, where meaning is use, where the application of a word is a criterion for understanding. Behavioural training alone will not suffice: the training is a necessary cause, but not itself sufficient to bring about an understanding of language. Wittgenstein would agree with Bruner's depiction of social encouragement: the child learns by believing the adult uncritically (Wittgenstein 1969, sect;160), by being encouraged first to absorb a particular picture and then to make use of this new picture without further training (Wittgenstein 1958, 89-90). This is possible only if the pupil responds to this further encouragement in a certain way and uses the rules he or she has been taught (Wittgenstein 1953, sect;29). In order to be able to say that a child has a command over language, it must know the meaning of the words it must be able to call things by their names without a trace of doubt. It is only then that one can say, to Wittgenstein, that there is an understanding of words, sentences, grammar, and so forth.
To Bruner, it is perhaps the idea of our young as game players that is crucial to his book. In this sense, "play is the culture of childhood" (Bruner 1983, 121), although that children play games which are always like language-games remains part of the practice of game-playing, since qua protoconversation games always have a somewhat linguistic character. Bruner also draws an analogy of 'speech acts' between what he calls prelinguistic communicative acts and lexico-grammatical speech acts (Bruner 1983, 38), which fits into his broader life-like notion of language. In game-playing, stressful situations and the pressure of expectations are removed: the child can experiment without having to dread the consequences of its actions. Choices can be made which would not be acceptable under the functional pressure to achieve. Play serves "as a preparation for the technical-social life that constitutes human culture" (Bruner 1983, 45). It is within these games that more complex procedures are first tried and tested, and later become accepted practice.
Wittgenstein's suggestions for the early stages of language learning leave behind an uncomfortable riddle that may have prompted Wittgenstein to put forth his description of the ostensive teaching of words in the first place. According to Bruner, Wittgenstein grapples with the question what features of a referent a label refers to, that is how words come to mean and refer (Bruner 1983, 81). But Bruner is extremely short in his reference to the problem and we do not have an adequate explanation on what grounds this issue arises in the first place. There is more to it. We have to answer not only the question how one can establish a reliable ostensive reference onto which to erect linguistic understanding, for which Wittgenstein provides a suggestion, but also address the issue why this ought to be done in the first place. We should also bear in mind the troubling lack of alternatives: if linguistic convention is not in some form established within the teachings of a social group, then language learning would seemingly not require teaching, which seems to point at innate characteristics or some sophisticated capacity for language learning. This riddle might be called the paradox of intersubjective reference or the framework problem. There are some things Wittgenstein's young pupil of language does not have to learn by ostensive definition, however. For example, in speaking of sensations, we repeat an expression we have learned within some language-game (Wittgenstein 1953, §290). These expressions have replaced the cry of primitive pain behaviour, for example. The referent is less problematic here, for primitive pain behaviour is said to be literally present in primo, first.
For Bruner, the solution is simple: the child's caretaker side-steps this dilemma by using mostly unambiguous references to whole objects (Bruner 1983, 81), made possible through natural intersubjective harmony. As described earlier, this position rests on a combination of two elements: on the one hand there are a priori assumptions about certain inborn abilities, on the other hand we find a complementary social context, that the child's surroundings act as a support system to aid in this learning process. One side hermetically plays into the hand of the other, filling the troublesome hypothetical gap that otherwise would exist in the mind of the philosopher or the psychologist. Yet all of the serious problems concerning language learning have been shifted onto some a priori assumptions, without a supporting argument.
Presumably Wittgenstein would not concede that we have a natural grasp of certain concepts, or that we have a sensitivity or capacity for natural intersubjective referentiality. The ostensive definitions to be used in Wittgenstein's suggestion for the entry into language are somewhat ambiguous by themselves, not only as their content is concerned, but also in form, since learning and speaking always occur without explicit rules: language is multi-faceted and full of tacit assumptions, some of which may be intuitive. What we do in language-games may rest on tacit presuppositions, for example, that there is no abuse of convention in the case in front of us. The expression "understanding" (a word, our language) has various senses, yet understanding is in some way always closely bound up with language. Language-games, like those played by children, according to Wittgenstein, need no justification and attempts at justification ought to be rejected. And yet why should we engage in ostensive definition at all, having accepted the larger picture of language as self-referential? More seriously, why should it be needed as a preparation for understanding? Epistemological-metaphysical doubt in this setting is misplaced, as both Bruner and Wittgenstein agree. It certainly is a simple truism that children learn language, though we may not be able to account for this learning to our satisfaction. Why then must both writers posit either the necessity of certain practices, linguistic drill in Wittgenstein's case, or the presence of certain natural dispositions, cognitive endowments, in Bruner's in order to rid themselves of a certain feeling of epistemological uncertainty?
Wittgenstein's suggestions on language learning do not make a theory. In contrast, Bruner clearly has a theoretical framework in Child's Talk. Part of Wittgenstein's method breaks down in the attempt to deal with children's entry into language and he is uneasily aware of it. The riddle is not how words come to mean and refer, but how we can sensibly speak about the practice of language learning. Philosophical Grammar suggests that the manner in which language is learned is not contained in its use. There is something in the early stages of language learning that cannot be shown in the sense in which we could show the post at which a word is stationed in our language.
Although critical of it, Wittgenstein has not entirely shaken off the Augustinian picture of language learning. He is well aware of his own biases later in On Certainty: "I myself wrote in my book that children learn to understand a word in such and such a way. Do I know that, or do I believe it?" (Wittgenstein 1969, sect;290). Perhaps this is the cause of the behaviourist misinterpretation of Wittgenstein's depiction of language learning and the cause of his dilemma. Wittgenstein walks a line that borders on an empiricist theory of learning imposing behavioural training on the one side and on a form of conventionalism on the other by suggesting that the connection between words and objects is learned ostensively through training, but that the grammar of words, the rules used for the application of words, is yet to be uncovered by the child, not within its own mind, but within the accepted usages amongst adults in a linguistic community. It is this uncovering or discovering of tacit rules which represents an understanding of language and renders language meaningful to the child. In either case, there is a linguistic scaffold to be known, a web to be woven. That is, I suggest, Wittgenstein cannot shake off a perception of language that is still larger than life: "it is a hypothesis that the process of teaching should be needed in order to bring about these effects. It is conceivable, in this sense, that all the processes of understanding, obeying, etc., should have happened without the person ever having been taught the language. (This, just now, seems extremely paradoxical.)" (Wittgenstein 1958, 12). On the one hand, we find a hypothesis, a metaphor that helps us in some way to understand language learning: ostensive teaching is like installing an electrical connection between a switch and a bulb. But when we then attempt to apply his method of philosophy to dissolve this hypothesis, we seemingly end up with only one alternative, namely that crucial aspects of language are endowed to the child, that it could have learned language by itself, which makes for a troubling state of affairs, particularly so in light of the role language plays in Wittgenstein's work. Exit innate characteristics, re-enter the hypothesis: it seems like it could not have been otherwise. But what does "could" mean here? Wittgenstein cannot or does not say.
The infant is caught in that area that remains grey under Wittgenstein's normative claims, a puzzle that has not yet been dissolved: there is no linguistic practice to examine in children before the entry into language in turn they must examine us, by being drilled, so that language may take hold. Language as a form of life is not hermetic and thus of necessity retains certain aspects of behavioural observation and training: for the trainee in the linguistic drill it is a matter of taking in the surroundings and making sense of the seemingly ordered but strange noises. This is quite unlike the more closed process suggested by Bruner. Wittgenstein reverts to a formal prescription for how children learn language, but when this hypothesis concerning the ostensive teaching of words itself ought to fail under the method of philosophy, we strike a methodological rock-bottom and reach a standstill.
The two emerging pictures of language in Bruner and Wittgenstein are brought to a head with the direct question what parts or characteristics of language are arbitrary and which ones are not. Could the formation of concepts in part be explained by facts of nature, its constraints and allowances? Should we then maybe study children's natural characteristics, as Bruner has done in part with his observation rather than settling for grammar? To Wittgenstein, the primitive behavioural expressions of human nature, for example, in so far as displaying pain is concerned, always hold true and make it into language. There is something prelinguistic which carries meaning and which merits study. But these natural primitives must be contrasted with the much more sophisticated study of the grammar of words. The manner of Bruner's quasi-scientific observation of child behaviour would likely at best yield general facts, say, about the existence of certain behaviours. Our interest, to Wittgenstein, does not fall back upon these possible causes of the formation of concepts. Our language exists quite independently of these fields and is learned by children independently of our tinkering. What is astounding is that they learn, not how they do it. A Wittgensteinian common-sense philosophical account of language, cannot sensibly speak of crucial aspects of language itself, and we must make the choice whether amongst these we will count the very entry into language. If we try to study the grammar of words as Wittgenstein does, there is a point where we must strike rock-bottom: this is as far as we can go. Language must speak for itself. Following Wittgenstein, grammar simply is there, a barrier we cannot sensibly overcome, though it can be explored; thus, in our normative descriptions of children we must [sic] treat them as somehow grammatical beings. He himself is here subject to striking rock-bottom; his own pragmatic grammatical study is often only the showing of a particular grammatical position, that is, a position on linguistic usage. His claims on early language learning must be considered in light of this methodological dictum. What Wittgenstein is looking for is the grammatical difference between words, which contrasts them and which can be shown to the child. But grammar only describes and in no way explains the use of signs. In the context of language learning, this just will not do and our picture of grammar must either be broken up and broadened, or Wittgenstein's remarks on language learning discarded.
Bruner's position corresponds more closely to the discernible events in young children's lives, which leaves us with the uneasy conclusion that Wittgenstein perhaps failed to dismiss his riddle as just another philosophical question, that is, as mistaken. I do not wish to accept Bruner's side, because of its a priori assumptions, to which standard objections can be raised, some of which have been raised here. Most seriously, whilst Bruner has come up with a thesis for early language learning, he glosses over the consequences of having posited various innate characteristics. What is under critique in Bruner's Child's Talk is not that he has pointed out how one of Wittgenstein's hypothetical dilemmas is simply side-stepped, but that in turn he fails to shed the mantle of the philosopher-scientist. Even if he transports some of Wittgenstein's valuable insights into the character of language to child psychology, he does not go far enough.
We are working with concepts here, just as the young child is working with concepts. We show each other instances of the language-game with the word "game" as we explore the idea of game-playing in children, until the coherence of our actions reaches the mirror of our audience. That mirror is not a tabula rasa which we fill, but one full of shared convention, expectations, and common history, as both Bruner and Wittgenstein point out. Yet if we strip all hypotheses in Bruner and Wittgenstein and all conventionalism, they still share a picture of children that may perhaps be contra-Augustinian, but that itself still is caught up in the same web. The scaffold which first delineates the space to be filled in language learning remains, along with our ability for learning. These certainly are philosophical concepts and not ones embedded in our practices of child rearing.
Bruner, Jerome (1983). Child's Talk: Learning to Use Language. New York: W. W. Norton & Company.
Wittgenstein, Ludwig (1958). The Blue and Brown Books: Preliminary Studies for the 'Philosophical Investigations'. Oxford: Blackwell.
----- (1969). On Certainty. London: Harper Torchbooks.
----- (1974). Philosophical Grammar. Ed. Rush Rhees. Trans. Anthony Kenny. Berkeley: University of California Press.
----- (1953). Philosophical Investigations. Oxford: Basil Blackwell.