The Articulate Mammal (40 page)

was verified more quickly than:

Assumption 3
‘When a complex sentence is composed of a main clause and one or more subordinate clauses, the main clause usually comes first.’ That is, it is more usual to find a sentence such as:

NERO FIDDLED [WHILE ROME BURNED].

than:

[WHILE ROME BURNED] NERO FIDDLED.

Similarly:

PETRONELLA EXPECTED [THAT PERICLES WOULD SCRUB THE FLOOR].

is considerably more likely than:

*[THAT PERICLES WOULD SCRUB THE FLOOR] PETRONELLA EXPECTED.

The strategy which follows from assumption 3 seems to be ‘Interpret the first clause as the main clause unless you have clear indications to the contrary.’ The existence of this strategy accounts for the correct interpretation of:

IT WAS OBVIOUS HE WAS DRUNK FROM THE WAY HE STAGGERED ACROSS THE ROAD.

Here, the subordinate clause is not marked in any way, but the hearer automatically assumes that it comes after the main clause. This strategy also partly accounts for the difficulty of:

THE ELEPHANT SQUEEZED INTO A TELEPHONE BOOTH COLLAPSED.

Until coming across the unexpected word COLLAPSED at the end of the sentence the hearer probably assumes that THE ELEPHANT SQUEEZED … was the beginning of the main clause.

Assumption 4
‘Sentences usually make sense.’ That is, people generally say things that are sensible. They utter sequences such as:

HAVE YOU DONE THE WASHING UP?

THE TRAIN GOES AT EIGHT O’CLOCK.

rather than:

HAPPINESS SHOOTS LLAMAS.

THE HONEY SPREAD MOTHER WITH A KNIFE.

The strategy attached to this assumption is the most powerful of all – though from the linguistic point of view, it is the least satisfactory because it is so vague. It says: ‘Use your knowledge of the world to pick the most likely interpretation of the sentence you are hearing.’ In certain circumstances this can override all other strategies, and reverse well-attested aspects of language behaviour. For example, under normal circumstances people find it much easier to remember sentences that are superficially grammatical than random strings of words. It is considerably easier to learn the apparently grammatical:

THE YIGS WUR VUMLY RIXING HUM IN JEGEST MIV.

than the shorter string:

THE YIG WUR VUM RIX HUM IN JEG MIV.

(Epstein 1961)

But this well-attested result can be
reversed
if the subjects are presented with semantically strange grammatical sentences and ungrammatical strings of words which appear to make sense. Subjects remember more words from strings such as:

NEIGHBOURS SLEEPING NOISY WAKE PARTIES

DETER DRIVERS ACCIDENTS FATAL CARELESS

than they do from sentences such as:

RAPID BOUQUETS DETER SUDDEN NEIGHBOURS.

PINK ACCIDENTS CAUSE SLEEPING STORMS.

(Marks and Miller 1964)

So far, then, we have listed a number of assumptions which hearers have about English, and suggested a number of linked ‘perceptual strategies’:

1 Divide each sentence up into sentoids by looking for NP–V(–NP) sequences (‘canonical sentoid strategy’).

2 Interpret an NP–V–NP sequence as actor–action–object.

3 Interpret the first clause as the main clause.

4 Use your knowledge of the world to pick the most likely interpretation.

Even quite odd sentences seem easy to understand if they fit in with the ‘strategies’ listed above:

THE KANGAROO SQUEEZED THE ORANGE AND THE KOOKABURRA ATE THE PIPS.

But sentences which do not fulfil the hearer’s expectations are more difficult to comprehend. Each of the following goes against one of the four basic strategies. The sentences can be understood reasonably easily, but they need marginally more attention from the hearer:

AFTER RUSHING ACROSS THE FIELD THE BULL TOSSED HARRY.

THE VAN WAS HIT BY THE BUS, AND THE CAR WAS RAMMED BY A TAXI.

THE POSTMAN BIT THE DOG, AND THE BABY SCRATCHED THE CAT.

When a sentence goes against more than one strategy the effect is rather worse:

THE SHARK PUSHED THROUGH THE SEAWEED WAS ATTACKED BY A TADPOLE.

The sentence is neither ungrammatical, nor incomprehensible. It just seems clumsy and strange, and would possibly cause a hearer to say: ‘I’m sorry, I didn’t get that. Could you repeat it?’

It is an interesting fact that speakers tend to avoid sentences which go against perceptual strategies to too great an extent. People just do not
say
things such as:

THE POODLE WALKED RAPIDLY UP THE MOUNTAIN COLLAPSED.

JOAN GAVE JUNE A PRESENT ON SATURDAY AND JANE ON SUNDAY.

Strictly speaking, these sentences are not ungrammatical, just odd and unacceptable. Compare the syntactically similar sentences:

THE RAG DOLL WASHED IN THE WASHING MACHINE FELL TO PIECES.

MAX GAVE HIS DOG A BATH YESTERDAY AND HIS CAT LAST WEEK.

However, since the ‘sensible’ sentences above are interpretable only because the speaker is able to use the imprecise strategy 4 (‘Use your knowledge of the world to pick the most likely interpretation’), sentences of this type may be in the process of being eliminated from the English language – since perceptual needs can often influence linguistic rules. To quote Bever: ‘The syntax of a language is partly moulded by grammatical responses to behavioural constraints’ (1970: 321).

Obviously, the four strategies noted so far are not the only ones we use when we comprehend sentences. Bever’s paper triggered a search for others,
particularly ones which might apply to a wider range of languages than his first three (e.g. Kimball 1973; Gruber
et al.
1978). Let us therefore outline two which might have a broader application, and partially encapsulate the ‘canonical sentoid’ strategy (Frazier and Rayner 1982, 1988).

The first of these says: ‘Assume you are dealing with a simple structure, unless you have evidence to the contrary.’ This has been called the ‘Principle of Minimal Attachment’, because each word is attached to the existing structure with the minimum amount of extra elaboration. On hearing the word PARADED in a sentence such as:

THE LION PARADED THROUGH THE TOWN ESCAPED.

it is far simpler to set up a simple NP–VP structure, than one which involves the added complexity of an extra sentence inserted after THE LION.

The second says: ‘Try and associate any new item with the phrase currently being processed.’ This has been called the ‘Principle of Late Closure’, because the previous phrase is held open, waiting for new additions, until there is strong evidence that it is complete. In a sentence such as:

FIONA DISCOVERED ON MONDAY THE PENGUIN HAD HURT ITS FOOT.

it is more natural to assume that ON MONDAY goes with the previous verb DISCOVERED, even though it would be equally plausible from the meaning
point of view to assume that Monday was the day on which the penguin injured itself.

Both these strategies would explain why:

THE MAN THE GIRL THE BOY MET BELIEVED LAUGHED

was so readily interpreted as ‘The man, the boy, and the girl, all met, believed and laughed’ (p. 216). This interpretation involves a much simpler structure than the ‘correct’ centre-embedded version, and tacking each new person introduced onto the previous one fits in with Late Closure.

As we have seen, the notion of strategies works well. There is plenty of proof that we impose our expectations on to what we hear, so at first sight there is no more to be said. All we need to do, it might seem, is to continue adding to our list of strategies until we have enough to cover the whole of language, and then try to divide them up into strategies that relate only to a single language, such as English; strategies that apply to a whole group of languages, such as those which have the basic word order subject–verb–object; and third, strategies that are universal.

However, when we consider the situation in detail, the notion of strategies raises some problems. Above all, language is enormously complex. Hardly any sentences are as straightforward as:

MARY LIKES STRAWBERRIES.

or even:

SEBASTIAN DISCOVERED THAT THE GORILLA HAD ESCAPED.

Many of them are considerably more complicated. For example, anyone who listens to a serious discussion is likely to hear sequences such as the following, in which a doctor is giving his opinion on a controversial illness (ME ‘myalgic encephalomyelitis’):

I was very sceptical initially, I have to say, for a while it seemed to be y’know this decade’s thing, and went along the same sort of lines as Total AllergySyndrome, and things like that, which ultimately became pretty well discredited as diagnoses and so I think initially, I think it was seen as this year’s trendy illness to have.

(Wetherell
et al.
2001: 156)

How many strategies are involved in a sentence like this? Obviously, many more than the few we have discussed. Perhaps twenty? Or a hundred? And supposing there are a hundred, what order do people apply them in? When faced with the problem of organizing dozens of strategies into a coherent model of comprehension some psycholinguists have argued first, that the task is impossible. Second, that the whole notion of strategy is meaningless in relation to these longer sentences. Strategies become vague devices of immense power which provide very little concrete information about sentence processing. As one psycholinguist commented: ‘One wonders what couldn’t be accomplished with an armful of strategies’ (Gough 1971: 269).