The Articulate Mammal (37 page)

And RUN OVER is not an isolated example. Another verb from which a person could also extract a considerable amount of information if necessary is GORE (Watt 1970). In the sentence:

CHARLIE WAS GORED.

the hearer can assume that the gorer was male and bovine. In other words, a bull. This information is
potentially
recoverable, though it
need
not be recovered.

In short, someone who knows a language has an enormous amount of knowledge which she
could
use when she understands or produces a sentence, but she does not have to. Or rather, she usually has to use some of it, but often only a rather small proportion.

Another example is the sentence:

DROWNING HEFFALUMPS CAN CAUSE A TERRIBLE COMMOTION.

Here, the hearer might well subconsciously have to query, ‘Are we dealing with heffalumps which are drowning, or someone who is drowning heffalumps?’ And he might even put his query into words.

This again suggests that a ‘classic’ transformational grammar represented a linguistic archive whose contents are available for use when a person processes a sentence. In principle, someone could, if he wanted to, recover all the knowledge stored in connection with a sentence when he understands or produces it. This may be what Chomsky meant when he wrote that:

the generative grammar represents the information concerning sentence structure that is available, in principle, to one who has acquired a language. It indicates how, ideally … he would understand a sentence.

(Chomsky 1963: 326)

The word ‘ideally’ may mean perfect understanding of the sentence as far as is possible within the limits of grammatical and lexical knowledge. Of course, in practice, no one has the time or the need to unravel every sentence in this way. Most people make a quick decision about the sentences they hear, and do not consider all the ramifications. In Bever’s words, they rely on ‘perceptual strategies’ or short cuts, rather than on full utilization of ‘epistemological structures’ or beliefs about language structure (Bever 1970: 281).

But it would be a mistake to assume that ‘epistemological structures’ are an optional extra. A person who could not detect ambiguities, who could not make judgements of grammaticality and who could not link up related sentences would only ‘know’ their language in a very limited way. There is no clear-cut line between knowing
how
to utter and comprehend sentences, and knowing
that
these sentences are grammatical and how they are related to other sentences. Humans do not behave like spiders, who can weave webs without any conscious knowledge about their skill. Humans need knowledge about their language in order to function properly as articulate
mammals. As far as language is concerned, the distinction between knowing
how
(as in knowing how to walk) and knowing
that
(as in knowing that the world is round) is a fuzzy one, because the two types of knowledge overlap.

Let us summarize what we have just said. We have concluded that a transformational grammar, in its classic form, incorporates mainly ‘archival knowledge’ or ‘epistemological structures’ – a set of beliefs or intuitions about one’s language which may not necessarily be recoverable at a conscious level. These beliefs are not merely optional extras, they are an essential part of anyone’s ability to speak and understand a language.

How are these ‘epistemological structures’ acquired? Do children learn how to use language, then later build up full knowledge about it? Or do the two learning processes go on simultaneously? This question was studied by three psychologists from the University of Pennsylvania (Gleitman
et al
. 1972). They concluded that the process of learning how to speak was intertwined with that of acquiring beliefs about one’s language. Both types of knowledge progress simultaneously, though the latter develops rather more slowly. Even 2-year-olds have
some
notion of grammaticality, though this is rather shaky. And children’s judgements about their language remain shaky even when they can speak fluently. But between the ages of 5 and 8 children start to have intuitions about their language that parallel those of an adult. Let us illustrate these points.

Even quite young children have
some
beliefs about their language, as shown by 2-year-old Allison, who judged the sequence:

*BALL ME THE THROW.

to be ‘silly’, and corrected it to:

THROW ME THE BALL.

Similarly, 2-year-old Sarah amended:

*SONG ME A SING.

to:

SING ME A SONG.

However, Sarah’s judgements were not consistently reliable, since she found:

WASH THE DISHES.

an odd sentence, and corrected it to:

WASH THE DISHES(!)

It was easier to elicit responses from the older children, and the results were more clear-cut. For example, when seven children between the ages of 5 and 8 were asked whether the sentence:

I AM KNOWING YOUR SISTER.

sounded ‘sensible’ or ‘silly’, the 5- and 6-year-olds found nothing wrong with it, but the 7- and 8-year-olds disapproved of it, though they could not always say why it was odd. The following is the response given by 7-year-old Claire:

But for other sentences, Claire not only gave adult judgements concerning grammaticality, she also gave an adult-type reason:

The researchers noted:

The ability to reflect upon language dramatically increases with age. The older children were better not only in noting deviance but also in explaining where the deviance lies.

(Gleitman
et al.
1972: 160)

Spontaneous repairs – cases in which a child corrects himself or herself without prompting – provide another way of looking at children’s awareness of language structure (Karmiloff-Smith 1986). Younger children often provide unnecessary repairs, as in:

YOU PUT THE CHURCH – THE TINY LITTLE CHURCH INTO THE TIN.

Since only one church was involved, this alteration by a 4-year-old was not essential. Slightly older children tend to correct themselves if their original speech could lead to misunderstanding:

LEND ME THE BALL – THE GREEN BALL.

This repair by a 5-year-old was important, because there were several different coloured balls around. However, the most sophisticated repairs are those which are inessential for getting across the right message, yet show a deep understanding of the linguistic system:

AND THEN FORTUNATELY THE GIRL – A GIRL OFFERS THE DOG A BONE.

Since there was only one girl in the story, this 6-year-old need not have changed
the
to
a
. But since no girl had yet appeared in the story, the change reflected a realization that it would have been correct to use
the
only if a girl had previously been mentioned.

Repairs, then, reflect a deepening awareness of the linguistic system. First, children can correctly use constructions, but cannot repair their mistakes. At a later stage, they begin to use repairs, sometimes sophisticated ones. Finally, they are able to explain various linguistic points.

PSYCHOLINGUISTICS WITHOUT LINGUISTICS?

We can now summarize the conclusions reached in this chapter. We have been examining the relationship between language knowledge (as ‘captured’ by the classic version of transformational grammar) and the way in which we handle sentences. In the first section we noted that transformations were irrelevant to the way in which we produce and comprehend language. In the second section, we saw that the hypothesis that we recover a classic Chomsky-like deep structure when we comprehend a sentence was unlikely. In the final section we concluded that a 1965 transformational grammar represented a person’s linguistic archive – a store of knowledge about language that is only partially utilized in the course of conversations. This archive develops simultaneously with, though rather more slowly than, the ability to speak and comprehend sentences.

Chomsky’s latest ideas, however, have not been discussed. This is largely because they have not been tested by psychologists, for two reasons. First, many of them turned away from transformational grammar in disappointment. As one noted: ‘By the mid 1970s there remained no unequivocal evidence that transformational grammars provided a model of either the
rules or representations that listeners and speakers use during comprehension. As a result, psycholinguistics largely severed its ties with linguistics’ (Tanenhaus 1988: 11). Second, Chomsky’s more recent ideas are mostly too imprecise and abstract to test.

Does this mean that psycholinguists can safely ignore abstract linguists such as Chomsky in the future? Should they turn to other linguistic theories, or give up linguistics altogether?

In order to answer this point, let us briefly consider the relationship of language knowledge to usage in Chomsky’s changing systems. Perhaps a cooking analogy might make this clearer, since it can illustrate the various levels of knowledge that might be involved.

Take chocolate mousse. Even a child might make a perfect mousse by following instructions, without any understanding of what she was doing, or what a mousse should look like. In language, this is equivalent to a young child who can talk, but has as yet little awareness of linguistic structure. Chomsky, as we have seen, has never regarded this ‘how to do it’ level as being his concern.

The next level involves a notion of what a proper mousse should look like, as well as a growing understanding of how a chocolate mousse relates in its composition to other mousses, such as salmon mousse or lemon mousse. In language, this is the level at which a person can reliably judge what is, and what is not a well-formed sentence. It is also the level at which sentences such as IT IS HARD TO HANDLE FLAMINGOS and FLAMINGOS ARE HARD TO HANDLE can be judged to be paraphrases of one another. This is the type of knowledge encapsulated in the classic (1965) version of a transformational grammar.

But there is a final, more basic level. This is the discovery that only a few fundamental principles underlie the whole of cooking. For example, trapping gas particles within food is the way to make it large in volume and light, a process found in mousse and buns. Or heating causes liquid to evaporate, and the remaining molecules to cling together, as in fried eggs or fudge. These principles are so general, and apply to such a wide range of foods, that they are unlikely to become apparent to the average cook, although she may underlyingly ‘know’ them, as she prepares food. In relation to language, this basic level is the one Chomsky and his followers finally aimed at: a few simple principles which would show how the whole thing works

Such a grandiose aim, if achieved, would be of great importance to psycholinguists. At the moment, it is unclear how one might test Chomsky’s fluctuating and somewhat vague proposals (as we have already noted). He himself has likened linguistic theories to those put forward by physicists to account for why the sun’s light gets converted into heat (Chomsky 1978: 202). Scientists cannot send a probe into the sun, so they have to make the
best guess they can from the light emitted at the sun’s outermost layers. Linguists cannot as yet reliably identify grammar in the brain, so similar guesses have to be made.

Such theory-building is a valid enterprise. Speculation has led to great steps forward in some areas of science. It would therefore be foolish of psycholinguists to ignore it. Meanwhile, it makes sense for them to push ahead, finding out independently how humans understand and produce speech. These are the topics of the next two chapters.

10

____________________________

THE CASE OF THE MISSING FINGERPRINT

How do we understand speech?

‘It seems very pretty,’ Alice said, ‘but it’s rather hard to understand.’ You see, she didn’t like to confess, even to herself, that she couldn’t make it out at all. ‘Somehow it seems to fill my head with ideas – only I don’t know exactly what they are!’

Lewis Carroll,
Through the Looking Glass

‘Sentence comprehension is like riding a bicycle – a feat far easier performed than described’ (Cutler 1976: 134). This feat is the topic of this chapter.

First, however, we should consider whether there might be a link between speech comprehension and production. It would be simpler for psycholinguists if they were directly related. However, there is no reason to assume this is so, any more than we should presume that the same muscles are used in sucking and blowing. We must therefore allow for four possibilities: