Vivian Cook

Writing Home

What in language terms is present in the child’s mind at birth? Everything or nothing? New linguistic theories upset the old views of learning to talk

 The revolution in linguistics that Noam Chomsky started had an immediate effect on other disciplines. His influence has been particularly marked on the study of how children learn language. In fact, a large part language had seemed almost unlearnable within the conventional theories of how language is acquired.

We now have a whole new approach towards the acquisition of language. Some of the new beliefs and conclusions are likely to have a strong impact on child psychologists, educationalists —and parents.

Chomsky recognised that the description of language has to take into account features of a sentence that are not revealed in its "surface." Take, for instance, the two sentences: "£500 was paid for that car” and "The garage was paid for that car." Superficially, these sentences appear to have the same syntactic structure. They are both passive, they both include prepositional phrases, and, they both have noun phrases as their subject. The only difference between them seems to be one of vocabulary - the difference between "£500" and "the garage”.

However the speaker of English knows that the first sentence means that £500 has changed hands, and second sentence does not mean that the garage has changed hands. To explain this difference it is necessary to say that the sentences differ syntactically in some way that is not immediately apparent. They have the same surface structure but they have a different underlying structure.

Chomsky uses a by now famous pair of sentences to illustrate the same point: "John is eager to please," and “John is easy to please." These two sentences, also, would be described as having an identical Surface structure, but the English speaker knows that in first case "John" is the subject of "eager to please” i.e. John pleases other people; and that in the second case "John" is the object of "easy to please” i.e. other people please John. Chomsky's “transformational generative grammar" showed that the analysis of language would not be complete unless it provided an account of these underlying structural difference. Unfortunately, the question of exactly how to describe them has split this school of linguistics down the middle. Some adherents insist that a single definite level of deep structure is required. Others insist that there are ever-receding depths of deeper and deeper structure.

Nevertheless a theory of language acquisition that aims at completeness cannot simply shelve the problem of deep structure. The psychologist, Martin Braine, for example, has produced a theory he calls “contextual generalisation." In this process, the child learns syntactic structure by first perceiving the location of words and phrases within the sentence. He hears “naughty boy" and "good girl" and realises that the phrase contains two positions: a first position, occupied by "naughty" and "good," and a second position filled by "boy" and "girl." The child generalises from this that any first-position word can precede any second-position word. This enables him to produce the "new" phrase "naughty girl". Thus, by location and generalisation he can build up the syntax of the language. Contextual generalisation applies also to the learning of sentences. The child hears “Train's coming" and "The car's going," and this allows him to say "The car's coming" by adding a first-position phrase to a second-position phrase.

But however much this is extended, contextual generalisation cannot explain the learning of deep structure that is not somehow manifested on the surface. There is no difference in the order of words or phrases between "£500 was paid for the car" and "The garage was paid for the car." So the child could not learn the difference simply by locating words and phrases within the sentence. Contextual generalisation can at best provide a partial account of language acquisition.

If you look at how adults and children interact, you can find clues about ways in which deep structures might be demonstrated to the child without depending on word order. One type of interaction is the develop­ment of the "wh" question in English ("wh" questions are those containing a question word like "who," "what," or "how"). To master the "wh" question, the child has to become aware of the deep structure of the sentence so that he knows that certain syntactic elements in it can be "questioned" in different ways. "Who likes gin?", "What does Peter like?", and "When does Peter like it? " differ by; which element in the deep structure is questioned. All of them are related to the structure underlying "Peter likes gin in the summer."

Roger Brown and his team at Harvard University have suggested that the child can, in fact, come to see this underlying structure by starting with a form of question found quite commonly in adult speech and known as the "occasional question." Examples of occasional questions are: "Peter likes what?", "You're going where?", and "You broke it how?" In the occasional question, the child hears the question element without the complications, of the changes of order found in the usual "wh" question. The occa­sional question frequently figures in exchanges between mother and child: Child: "I want milk." Mother: "You want what?" Or—Mother: "What do you want?" Child: (Silence). Mother: "You want what?"

So the child can see the structure of the sentence by hearing the declarative sentence and the occasional question in close juxtaposition. From the declarative sentence he can proceed to the occasional question, .and he can then learn the changes, of order necessary for the "wh" question. This postulated order of acquisition predicts that he will make certain types of "error" like "What this is?" and "Where you go?" And these errors are indeed found.

Yet, despite the amount of attention that has been paid to these processes of interaction, it is still far from clear how essential their role is. Take the example of two types of interaction, also described by Brown and his colleagues, and called "expansion" and "modelling".

Expansion is the process at work in the following exchanges: Child: "Sat wall." Mother: "He sat on the wall." Or—Child: ''Pick glove." Mother: "Pick the glove up." The mother fills out the child's sentence so that it is grammatical in terms of adult grammar and is appropriate to the situation at the moment of speaking.

The other process, modelling, consists simply of the child observing appropriate grammatical sentences said by an adult that are not expansions of his sentence. Originally, it seemed likely that expansion was of crucial importance in acquiring syntax, because the child could relate his own sentence to the equivalent adult form and could see the relationship between the syntax and the situation. However, an experiment with groups of three year old children showed that, on the contrary, the group which had its sentences modelled consistently—i.e., answered appropriately without expansion—did better than the group whose sentences were continually expanded. Either modelling has hidden virtues, or else expansion is less important than would appear at first sight.

The problem of how deep structure is acquired can also be solved very neatly by saying that it is not ac­quired. In other words, it is already present in the child's mind from the outset. If all children had innate deep structure, this would mean that all human languages would share the same deep structure. The great differences between them would be due to the different ways in which they convert their common deep structure into surface structure. All children would then start with the same linguistic heritage. What they have to acquire would be the way in which this can be converted into the surface structure of the particular language they hear about them.

This solution succeeds in taking nearly everyone's breath away not only by its insistence on the innateness of some ideas but also by its effrontery in ac­tually specifying what these ideas are. But at the moment, there is little concrete evidence either way. A general feeling is that, while part of a child's knowledge of language is innate, it is too early to say which part.

In fact, the child's contribution to language acquisition may be a knowledge of certain abstract gram­matical relations such as subject and predicate. It may, on the other hand, be a capacity for organising lan­guage material in a certain way - i.e., a limitation on the form that grammatical rules may take.

Apart from their emphasis on deep structure, those linguists and psychologists influenced by Chomsky differ from previous ways of thinking in the angle from which they approach child language. Formerly, the typical starting point was adult grammar. The child's sentences were described in terms of their differences from the adult's. If the child said, "Where horse go?" this was treated as a mutilated version of the adult sentence, "Where did the horse go? "The person describing the language behaved exactly like the mother expanding her child's sentences.

Now, the emphasis is on the child's language at a given moment as a complete self-contained system. As the child does not know where he is going, it seems wrong to describe his progress in terms of his final destination, unless we are postulating that the entire adult grammar is present in the child together with a " mechanism that interferes with its proper functioning. It seems better to look at the child's development stage by stage. Only in retrospect, after all the stages have been described, can the progression towards adult grammar be seen. Thus, "Where horse go?" would be described in terms of the child's grammar at that particular stage, which has a rule that puts the question word at the beginning of the sentence. Later, this would be compared with the rule for question-formation in the adult’s grammar, which also changes the order of elements.

This change of emphasis has made description of child language even more difficult than it was before. It is no longer possible to use something already known—adult grammar—as a yardstick. Instead, the child's own grammar has to be described solely from evidence provided by the child.

The work of the linguists, E.S. Klima and Ursula Bellugi, provides a classic example of this. They describe a child's development of negation over three periods. Each period has its own distinct grammar. In the first period, the child produces sentences like "More . . . no," "No play that”, and “Drink all tea," The rule for negation in the child’s grammar here is that "no" or "not” can occur at the beginning or end of the sentence but never in the middle. It is likely that the child does not even understand sentences with negation in the middle. The second period has sentences like "I can't catch you”, “Don’t leave me," and "He no bite you." The child’s grammar now permits "no" and "not" to occur medially and also uses two negative auxiliaries “don’t” and "cant."

These auxiliaries are only found in negative sentences. The third period is characterised by sentences like "I didn't did it," "You don't want some supper” and "I am not a doctor." Other auxiliaries occur and they are found in questions, and declaratives as well. The child still has no rule that links negation with "any," and consequently he produces only sentences like "I didn't see something".

The child's grammar approximates closer and closer to adult grammar but is nevertheless a self consistent system at each stage . "No sit there” may be deviant to the adult but it is perfectly regular in terms of the child's first-period grammar. An analogy is often made between the progress of the child and the development of a scientific theory,. The child observes certain linguistic data and constructs a hypothesis to explain them. The hypothesis takes the form of a grammar. He tests his hypothesis by using his grammar to produce new sentences and sees whether these are acceptable or not. The new data he perceives and the outcome of this testing, cause him to improve or abandon his original hypothesis.

To highlight radical differences between this way of' thinking and others, one might choose four different quotations. A behaviourist might say: “We teach both children and adults to speak correctly by saying “That’s right” when any appropriate behaviour occurs”. A Chomskyan transformationalist would reply that "There is not even a shred of evidence that approval and disapproval are usually dependent on correct syntax” A Piagetian psychologist might say that, in the speech of the six year old, "Causal relations are rarely expressed but are generally indicated by a simple juxtaposition of the related terms”. A transformationalists would retort that, in the speech of the three-year-old "There is a 'because' stage in which many replies take the form of logical motivational, and causal links”.

Not all the differences between these quotations can be dismissed solely as differences in terminology. For, within a matter of ten years, the study of language acquisition within the framework of transformational generative grammar - by basing itself on large-scale observations of children in action and on a coherent theory of language - has produced results that other schools cannot ignore.

Probably its greatest contribution comes from concentrating on language as the central feature in child development. This has shown the complexity of what the child learns and something of the nature of how he learns. At the moment, the emphasis on language and the extensive reliance on innate deep structure are the chief points on which this approach differs from Piaget. To the Piagetian, language is only one aspect of development. The growth of language presupposes a stage in which the child acquires a general ability to use symbols which manifests in other forms as well as language. Yet, despite this, there is much more in common between these two positions than between either of them and behaviourism.

Historical note: about a year after this was published in 1971 I received a highly indignant letter from B.F. Skinner, now lost. From memory, this started promisingly: 'Dear Miss (?) Cook (I can never get these British names' and then demanded to know why, rather than say what a behaviourist might say, I did not quote what one actually did say - the pretend quotation above “We teach both children and adults to speak correctly by saying “That’s right” when any appropriate behaviour occurs” actually came from a certain B.F. Skinner, with a minor change I think. He then accused me of, like Chomsky, falling into the Jacobsonian fallacy, perhaps the greatest compliment I have ever received. He concluded that he had never bothered to defend himself against Chomsky but that MacQuorcodale had done it for him; I dutifully read MacQuorcodale but could not decide if this was true as I couldn't understand him.