Language Learning

A Special Case for Developmental Psychology?

Christine J. Howe

• Publisher: Routledge
• Year: 2017
• ISBN:
  9781351662581
• Edition: 1
• Topic: Psychology
• Subtopic: History & Theory in Psychology
• Index: Psychology

Originally published in 1993, the starting place for this book is the notion, current in the literature for around 30 years at that time, that children could not learn their native language without substantial innate knowledge of its grammatical structure. It is argued that the notion is as problematic for contemporary theories of development as it was for theories of the past. Accepting this, the book attempts an in-depth study of the notions credibility.

Central to the book's argument is the conclusion that the innateness hypothesis runs into two major problems. Firstly, its proponents are too ready to treat children as embryonic linguists, concerned with the representation of sentences as an end in itself. A more realistic approach would be to regard children as communication engineers, storing sentences to optimize the production and retrieval of meaning. Secondly, even when the communication analogy is adopted, it is glibly assumed that the meanings children impute will be the ones adults intend. One of the book's major contentions is that a careful reading of contemporary research suggests that the meanings may differ considerably.

Identifying such problems, the book considers how development should proceed, given learning along communication lines and a more plausible analysis of meaning. It makes detailed predictions about what would be anticipated given no innate knowledge of grammar. Focusing on English but giving full acknowledgement to cross-linguistic research, it concludes that the predictions are consistent with both the known timescale of learning and the established facts about children's knowledge. Thus the book aspires to a serious challenge to the innateness hypothesis via, as its final chapter will argue, a model which is much more reassuring to psychological theory.

CHAPTER ONE

The Case for Innate Knowledge

Until about thirty years ago, the concerns of linguists and psychologists of language differed markedly. Linguists were primarily interested in the criteria by which they could specify the sentences of given natural languages, criteria that they assumed would amount to a “grammar”. Even those whose study ended with English Language at school will have a passing acquaintance with what specification via a grammar would involve. Whatever else, it would mean the categorisation of “lexical items” (that is, words plus inflections) into various “form classes”, and the utilisation of rules that showed how form class members could be combined to produce sentences. Psychologists on the other hand, tended to concentrate on what was called “verbal learning”. They performed countless experiments on memory for words or, even, memory for nonsense syllables. Sometimes, a list would be presented and the subject asked to recall as many items as possible. At other times, the task would be “paired associate learning” where pairs of items were presented and the subject asked to recall one member in response to the other. Through such experiments, psychologists investigated the effects of varying the number of items, the time between items and the relatedness of items. The journals of the day carried the results of these experiments in endless permutations.

The lack of overlap between linguistic and psychological concerns certainly did not lead to sterile research. The work done in both disciplines has relevance today, and will be used to some extent in this essay. Nevertheless, around 1960 the work done in linguistics was felt by its authors to be overly limited. A broadening of perspective took place, which led eventually not just to a deep interest in psychological issues but also to a fresh approach to them. The broadening was precipitated by the conviction, amongst linguists, that the notion of a grammar must be psychologically real. In other words, when ordinary speakers produce sentences, they too are demonstrating an awareness of form classes and the rules that order them. Initially the significance of this conclusion was felt to lie in a new criterion for theoretical adequacy. Amongst linguists it had long been realised that there are numerous ways in which form classes and rules of ordering can be arranged to specify sentences. Thus, what is sometimes called “observational adequacy” is not sufficient to establish one grammar over others and further criteria are required. Traditionally, these criteria have related to elegance and simplicity but once the psychological reality of form classes and rules of ordering was postulated, a new possibility was raised. Perhaps only a small number of potential grammars corresponds to ordinary human knowledge. Perhaps only one does. In which case, it (or they) might reasonably be taken as the best representation.

Having recognised psychological reality as a constraint on their activities, it was inevitable that linguists would take an interest in psychology itself. Discovering that, applied to language, psychology was coextensive with verbal learning, they perceived a gap. After all, if ordinary humans really do have knowledge of a grammar, they must, somehow, acquire it. However, as the grammar will not be told to them, they will not acquire it by memorising the linguistic output of other speakers. As the verbal learning experiments were focused on the memorising of this kind of output, they clearly could not bear on the acquisition of ordinary human grammar, raising the question of precisely how it is acquired. Intrigued, a number of linguists began to reflect on the issue, and they soon came to the conclusion that it could not even be inferred from linguistic experiences alone. Obviously these experiences must be playing some part. The differences between the world’s languages are such that their speakers must acquire different grammars. These differences have to result from differing experiences. However, the linguists were convinced that linguistic experiences could not be the whole story. On the contrary, they believed that acquisition must involve the reconciliation of linguistic experiences with innate knowledge of the grammar’s basic form. This, as the linguists were quick to realise, would make grammar more than an omission from the psychological study of language. It would elevate grammar into a major challenge to psychological theory.

The challenge would arise not so much from the denial of the “blank slate” approach. By the 1960s, most psychologists concerned with language had already abandoned this idea. Indeed, they would probably have welcomed a strengthening of the case against behaviourism. The challenge would not even be from the idea of innate knowledge per se. Innate “schemas” were being advocated in the field of perceptual development with relatively little controversy. The problem would be the sheer complexity of the innate knowledge being proposed for grammar. If true, it would amount to an a priori categorial structure, and this would be quite unlike what was being suggested elsewhere. Hence, it would be hard to reconcile with the general learning theory which, in the mid-1960s, was presupposed by most psychological researchers.

Have things now changed? Yes and no. It is no longer true that a general learning theory is presupposed by most psychological researchers. The problem of grammar has become too well known. Rather, the possibility of a general theory is now hotly debated, with uncertainties over empirical research that wait on a resolution. What is still true, however, is the continued centrality of accounting for grammar, with protagonists of all sides of the argument accepting that consensus here would be a major step forward.

Thus, there can be little doubt that the idea of innate knowledge of grammar is of contemporary theoretical relevance. How, then, does the idea stand? Is there really no alternative if learning is to be explained? These questions will underpin the discussion of much of this essay. By way of introduction, the present chapter will examine the reasoning that led to the idea in the first place. It will start with the logic underlying the decision to ask how grammar is acquired. It will find nothing remotely suspect. It will then turn to the considerations that produced the controversial answer and argue that they included two assumptions that might, in principle, be challenged. Recognising this, the chapter will ask whether the force of the considerations would be weakened, should either assumption be rejected. For one of the assumptions, the chapter will give a clear and categorical answer: rejecting it would do nothing to weaken the case for innateness. For the other, the chapter will discuss it only to the extent of identifying the most promising approach to an appraisal. As it raises issues of enormous complexity, following the approach will have to wait for the chapters to come.

FIRST STEPS TO INNATE KNOWLEDGE

Moving then to the origins of the innateness hypothesis, it will have been noted that it followed from what can be construed as two discrete developments in the field of linguistics. The first was the shift of emphasis from the ability of grammars to represent sentences to the ability of humans to produce them. The second was the conviction that the ability of humans to produce sentences implies at least one of the grammars capable of representation. As a result the first task will be to scrutinise these developments closely, to ask, for example, whether people really can be said to “produce sentences” in any meaningful sense of the term, and if they can, whether their productions really do guarantee knowledge of a grammar. Both questions will be discussed in the early part of the section. Finding no reason to answer in the negative, the section will turn to the innateness hypothesis itself, and ask why it followed so straightforwardly from the acceptance of psychological reality. As will become clear, a main reason was the coincidence in time between the acceptance of psychological reality and new insights into observationally adequate grammar.

The Knowledge Behind Sentence Production

Having followed the arguments so far, those previously unacquainted with the thinking of linguists might be tempted to raise a simple objection. Surely, they might say, there is no need to invoke grammar to explain how ordinary speakers produce sentences because ordinary speakers seldom produce sentences. Sentences are rarefied commodities beloved by school teachers, but rarely found in everyday usage. In particular, sentences are divorced from the vagaries of dialect, while spontaneous productions manifest these features in abundance. The objection would have to be accepted if linguists were concerned with the notion of a “sentence” that underpins explicit judgements. There is plenty of evidence that when asked to pick out the sentences in a set of utterances, people will veer towards school standards and away from their own vernacular. Thus, they will accept "I have not done it” and reject "I ain’t done it” even when they themselves sometimes use the latter.

Interestingly, however, intuitive judgements only approximate school standards. They do not, as Hill (1961) and MacLay and Sleator (1960) found when they asked speakers to judge strings like "Send one to Harry and I”, exactly mirror them. Hence, there is some compromise with actual usage but nevertheless, the general point holds good: the concept of a sentence underpinning explicit judgements precludes much that is produced spontaneously. However, when these precluded productions occur in ordinary discourse, the covert response is seldom rejecting. Moreover, even when it is, it still acknowledges the string as part of the language and for linguists this is enough. The linguistic concept of a sentence amounts to those strings that would be implicitly accepted as exemplars of a language. Given this concept “I ain’t done it” and “I have not done it” would both pass as sentences, but “Ain’t I it done” and “Have I done not it” would fail. Given this concept also, it should be clear that the vast majority of adult utterances must be viewed as sentences and thus, from the linguistic perspective, the ability to produce sentences is not the prerogative of an educated elite. It is a virtually universal property of adult behaviour.

However, if we accept this point about sentences, do we also have to accept that it implies knowledge of a grammar? Why can’t we just say that when people produce sentences, they retrieve the ones they have already heard from a list in their memory? In the early 1960s several reasons were advanced for rejecting this possibility. One was that there are an infinite number of sentences and, for that matter, non-sentences open to mature speakers of every natural language. Hence, the sentences could never be specified in a list. The theorist most responsible for highlighting this point was Noam Chomsky (especially Chomsky, 1957). The flavour of Chomsky’s thesis can be grasped quite simply via embedded relative clauses. No doubt, the product eventually becomes hard to understand. Nevertheless, we can embed relative clauses ad infinitum with each additional clause producing a new sentence. The point is beautifully illustrated by the verses of “I know an old lady who swallowed a horse … to catch a fly”.

More concrete evidence against the listing of sentences was obtained from the speech of children. It was recognised that the notion of a list implies that sentences are treated as unanalysable wholes, meaning that children’s utterances should be full or, at worst, truncated versions of adults’. Data reported in the early 1960s showed this not to be the case. For instance, through his work into the earliest word combinations of three young boys, Braine (1963a) reported many strings that look decidedly unadultlike. The best known is “All gone sticky”, which, it must be admitted, could be an abbreviation of something like “The cake’s all gone and now your hands are sticky” but that almost certainly does not mirror a simple sentence. Working with slightly older subjects, Ervin (1964) reported even more compelling evidence against children’s speech replicating adults’. Ervin and, later, other researchers, observed systematic overgeneralisation of the inflections when, for example, the plural “—s” is applied to “foot” and “mouse” and the past “—ed” to “come” and “break”. Forms like “foots”, “mouses”, “comed”, and “breaked” most definitely do not appear in adult speech. Likewise, the juxtaposition of noun and pronoun in (1.1) could not reflect anything used by more mature speakers.¹

(1.1)

(Melanie (20.6) takes cup)

Melanie:		Milk
		Drink it. Drink it
		Drink it milk

Yet such strings, which appear frequently in data that I have collected, have also been documented by such writers as Bloom, Lightbown, and Hood (1975).

Thus, theoretical studies of adults’ speech and empirical studies of children’s pointed to the same conclusion. Our knowledge of what constitutes a sentence is not represented in the form of a list. However, besides showing this, the studies were also thought to clarify just how our knowledge is represented. It was recognised that an open-ended set of sentences could only be represented in a finite manner if experienced sentences were treated as composites, made up of units like lexical items. However, compositionality implied a mechanism whereby the components can be successfully reordered, a mechanism that in other words allows “I can run” but not “can run I” to be constructed from “I”, “can”, and “run”. It was concluded that the only mechanisms that could do this while guaranteeing the open-endedness of sentences would be ones that called upon rules. However, it was also concluded that not just any rules would do. It became clear from the likes of “All gone sticky”, “foots” and “Drink it milk” that humans can juxtapose lexical items that they have never actually experienced in adjacent positions. It was realised that they would not be able to do this if they were learning rules that order individual items. Rather, they must be categorising the lexical items in some way and learning rules that order categories. However, once this was accepted, the outcome wa...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Preface
• 1. The Case for Innate Knowledge
• 2. The Contextual Completion of Meaning
• 3. The Time-scale to Observational Adequacy
• 4. The Approximation to Psychological Reality
• 5. The Establishment of an Alternative Theory
• References
• Author Index
• Subject Index