The Articulate Mammal (34 page)

On the other hand it wrongly linked up sentences such as:

THE BOY WAS LOATH TO WASH.

and

THE BOY WAS DIFFICULT TO WASH.

which seemed to be quite different. So finally, he became convinced that the most satisfactory system was a transformational model of language, in which sentences felt to be similar share the same deep structure. He came to believe that all sentences had both a hidden, deep structure and an obvious surface structure which might look quite different, and he accepted that these two levels were linked by processes known as transformations.

However, the Emperor remained puzzled about how this model of an internalized grammar might tie in with the way humans produce and comprehend sentences. He felt that Noam had been quite unclear on the topic.

Several of the things discovered by the mythical Emperor of Jupiter are points made by Noam Chomsky in his early, slim, but extremely influential work,
Syntactic Structures
(1957). In this, he explains why a left-to-right or ‘finitestate’ model of language is deficient, and also why a top-to-bottom or ‘phrase
structure’ model is inadequate. He then justifies the need for a transformational grammar. He elaborated this basic model in his ‘classic’ work
Aspects of the Theory of Syntax
(1965). Within 20 years, however, his views had radically changed. Let us see how this alteration might be justified to the Emperor of Jupiter.

RETURN TO JUPITER

Many years later, after he had orbited the universe several times, and been acclaimed as one of the pioneers of his century, Noam decided to return to Jupiter. He wanted to see how the Emperor was coping with his old transformational system. More importantly, he wanted to explain his new ideas on language.

Noam found the Emperor full of complaints. After Noam’s departure from Jupiter, the Emperor had continued to work on Noam’s system. He had been helped by some of Noam’s spaceship colleagues who had stayed behind on Jupiter to do some research on the climate. But things just hadn’t worked out as he had hoped.

The Emperor had two types of grumble. There were general grumbles about the whole system, and specific grumbles about particular transformations.

His main complaint was that the system just didn’t work properly. He had hoped that by now he would have found a set of rules which could account for all the possible sentences of English, and no others. But in spite of working long hours, there were dozens of sentences which he’d heard Noam’s colleagues speak, for which he hadn’t been able to specify the full set of rules. And the very best set of rules he’d come up with still included numerous sentences which apparently weren’t English.

Furthermore, he had considerable doubts about his transformational rules. As long as he got the right outcome, it didn’t seem to matter very much how he got there. Almost anything could be transformed into anything! There seemed to be too much latitude. Surely the whole thing ought to be tightened up a bit?

Noam agreed with these general points. The Emperor had discovered for himself the same problems as Noam had noticed. It seemed almost impossible to find a definitive set of rules which could specify what was, and what was not, a permissible sentence of English. The second, and more serious problem, was the enormous ‘power’ of the system: transformations appeared to be able to do almost anything. There were not enough constraints keeping them in check. A system which can do anything, as if with the wave of a magic wand, is not very informative.

Noam explained that he had been working very hard on the question of constraints. It was far more important, he had decided, to specify the general bounds within which human language worked, than to spend hours and
hours fiddling with the exact rules which would account for any one particular language.

Encouraged by this, the Emperor started on his detailed complaints, which were mostly about transformations. First, he grumbled, some transformations were quite arbitrary, because they were linked to particular lexical items. You simply had to know which words were involved. For example, you could say:

FRED GAVE A GIRAFFE TO THE ZOO.

FRED DONATED A GIRAFFE TO THE ZOO.

Then, a transformation supposedly specified that with GIVE, you could also say:

FRED GAVE THE ZOO A GIRAFFE.

But this transformation did not work with DONATE. You could not say:

*FRED DONATED THE ZOO A GIRAFFE.

Wasn’t this odd? he asked.

Noam agreed that any transformation which was restricted to particular lexical items was not a proper transformation. Instead, it was part of the dictionary or ‘lexicon’ which existed in any speaker’s mind. In his more recent system, he had moved information about the structures which could follow GIVE and DONATE into this dictionary.

The Emperor continued grumbling. Some transformations seemed to him pointless. Why did a sentence such as:

FENELLA THOUGHT THAT SHE WAS ILL.

have a deep structure which included the word FENELLA twice, saying in effect:

FENELLA THOUGHT THAT FENELLA WAS ILL?

Wasn’t this rather pointless? Couldn’t one leave SHE in the deep structure, and add a note saying SHE referred to FENELLA?

Noam agreed that a transformation which changed FENELLA into SHE was quite unnecessary, and that the matter could be dealt with in the way the Emperor suggested. In any case, the linking up of a pronoun SHE to other words should be dealt with by the semantic component, not by a transformation.

The Emperor continued moaning. Why were there so many different transformations which all had more or less the same effect? Consider:

IT SEEMED THAT THE DUCHESS WAS DRUNK.

IT WAS DIFFICULT TO PLEASE THE DUCHESS.

These two sentences were fairly like their deep structures, compared to two others, which involved bringing THE DUCHESS to the front:

THE DUCHESS SEEMED TO BE DRUNK.

THE DUCHESS WAS DIFFICULT TO PLEASE.

Yet each of these two sentences involved a different transformation! Supposedly, they had to be different, because the deep structures were different. Wasn’t this unnecessary proliferation of transformations?

Noam agreed with this criticism. It was foolish to have different transformations which performed the same manoeuvre. In his recent system, they had been combined.

The Emperor in his moans and groans had outlined many of the problems which eventually surrounded old-style transformations. They were too powerful, there were too many of them, they were too disparate. Gradually, they were reduced in number. Some were handed over to other components of the grammar, others were combined. In the end, only one transformation survived. This moved items about, though within strict limits.

The Emperor was amazed! Fancy having a transformational grammar with hardly any transformations! How on Jupiter did such a system work?

Noam started waving his arms about in excitement as he propounded his new system. He was on the verge of specifying a genetic blueprint for language, he announced. There were a number of fixed principles, which worked for all languages. There were also others which allowed a limited amount of variation. If you specified these properly, you hardly needed any rules at all!

The Emperor looked doubtful. Perhaps Noam had contracted space-sickness, which had sent him mad. How could one do without rules?

Noam tried to explain. Suppose you were designing a human being, he suggested. You had to give him or her a head. That would be a fixed principle. But the colour of the skin could vary in certain specified ways. As for doing away with rules, one might have a general principle saying: ‘Limbs come in pairs.’ Then one need not have separate rules which said: ‘Humans have two arms’ and ‘Humans have two legs.’ This sort of a system was applicable to language, he was convinced.

The Emperor was suspicious. Surely language was much too complicated to be dealt with in this simple way?

Not at all, argued Noam. On the contrary, language possibly consisted of a number of rather simple components. Each of the components worked in accordance with some quite straightforward principles, and they only appeared complex because of the way they interacted with principles from other components.

The Emperor seemed puzzled. So Noam used another analogy. ‘Think of a human mouth,’ he suggested. ‘There’s a mobile tongue which pushes food about. There are salivary glands which moisten it. And there are fixed teeth which grind it down. Each of these components is quite simple. Yet when they are working together the interaction is quite complex, and the effect powerful!’ (Matthei and Roeper 1983).

The Emperor was partially persuaded. He begged Noam to hand over his genetic blueprint for language. But Noam stalled. He hadn’t yet worked out how many components were involved, he admitted, nor what the basic principles were. He was fairly confident only that ‘economy’ or simplicity played a major role. Matters would be clearer in a hundred or so years’ time, he predicted.

The Emperor felt quite frustrated. And he was even more puzzled as to how Noam’s new system might link up with how humans understand and produce speech.

In this fictitious account, we have outlined several of the problems which caused disillusionment with transformations as they were formulated in the ‘classic’ (1965) version of transformational grammar. And we have put forward the general aims expressed by Chomsky first of all in
Lectures on Government and Binding
(1981), but expressed most clearly in
Knowledge of Language: Its Nature, Origin and Use
(1986) and later in the
The Minimalist Program
(1995b). Chomsky became more concerned with specifying the nature of the human language system than with formulating a complete picture of any one language. He believed that a ‘principles and parameters’ approach (with
parameter
referring to a factor which can be set variably) would largely do away with rules. And he became convinced that the overall system is modular, in that it is composed of a set of modules (components) which are simple in themselves, but become complex when they interact with other modules.

But this left many, including the fictional Emperor of Jupiter, deeply disappointed, as will be outlined below.

THE EMPEROR’S DISILLUSION

The Emperor of Jupiter felt let down. But was he angry with Noam, or angry with himself? He wasn’t sure. He had, he felt, spent far too many years chasing moonbeams, exciting, glistening ideas that always just eluded him. Maybe he should have realized long ago that Noam and his earthling mates
were born with abilities which were not available to people from Jupiter, just as Noam didn’t take easily to the toe-wiggling that came so easily to the Jupiter inhabitants.

But what exactly was it that the earthlings could do? What underpinned their ability to talk to each other? Even before they started chatting, they seemed to have some hidden understanding of others. It wasn’t just that they had formulaic ways of greeting each other, and (mostly) took it in turns to talk. Astonishingly, they seemed to be able to look into the other person’s mind, and to guess (correctly) whether he or she needed to be helped or left alone. This type of mind-reading seemed truly amazing! And, judging from talking to Noam’s crew, it was an ability which earthling babies developed early on in their lives. They not only had an enviable facility for combining sounds into words, and words into longer sequences, but they were also able to find the words they wanted remarkably fast, even though there seemed to be tens of thousands of them!

Eventually, the Emperor of Jupiter decided that he was profoundly grateful to Noam, because without Noam’s inspiration he and his fellow Jupiterians might never have realized how interesting, and how important, human language was. But the time had now come to move on to other areas of interest. In particular, he wanted to think about the relationship between a grammar and the way that grammar is used in actual speech. Noam had been most unclear on this point.

To conclude, the Emperor of Jupiter felt, as eventually did earthbound linguists, that Noam Chomsky had usefully highlighted the importance of language, and drawn attention to some of its key properties. But he had not explained clearly the link between the grammar of a language, and the way that grammar could be used in actual speech.

This will be the topic of the next chapter.

9

____________________________

THE WHITE ELEPHANT PROBLEM

Do we need a grammar in order to speak?

‘I have answered three questions, and that is enough,’

Said his father; ‘don’t give yourself airs!

Do you think I can listen all day to such stuff?

Be off, or I’ll kick you downstairs!’

Lewis Carroll,
Old Father William

Chomsky, for around half a century, tried to ‘capture’ a speaker’s abstract knowledge of language. But it remained unclear how knowledge related to usage. According to Chomsky, the two were rather distant, since he denied that linguistic knowledge is directly related to the way we understand and produce utterances. This leads to a crucial and rather startling question: can a grammar actually be
irrelevant
to the problem of understanding and producing speech?

If we had put this question to a hardcore linguist, at a time when Chomsky’s views were dominant, he would probably have answered: ‘Of course language knowledge and language usage are not totally separate, they just have to be studied separately, because the relationship between them is indirect.’