The Articulate Mammal (31 page)

WHAT COWBOY DOING?

WHY YOU SMILING?

The full IS … ING construction appeared only after a long delay. In both Adam’s and Sarah’s speech, the gap was longer than 12 months. For a year, it seems, they just did not fully recognize the connection between the IS and the ING. And there appears to be nothing inherently difficult about the phonetic forms AM, IS, ARE. This is shown by the fact that for all three children AM, IS, ARE occurred with nouns, as in:

HE IS A COWBOY.

some time before AM, IS, ARE, in progressive constructions.

It seems that all three children were puzzled by the discontinuity involved. They assigned -ING to the progressive early, but were baffled by the preceding IS. In other words, discontinuities seem to go against children’s natural intuitions about what language is like. This point is again illustrated by the development of relative clauses. As we noted earlier, those which do not interrupt the main clause such as:

THE FARMER WAS ANGRY WITH THE PIG [WHICH ATE THE TURNIPS].

develop before those which do:

THE PIG [WHICH ATE THE TURNIPS] ESCAPED.

Moreover, if children are asked to repeat a sentence in which a main clause is interrupted by a relative clause, they tend to alter the sentence in order to avoid this happening. A child asked to repeat the sentence.

THE OWL [WHO EATS CANDY] RUNS FAST.

repeated it as:

THE OWL EAT A CANDY AND HE RUN FAST (Slobin and Welsh 1973).

Operating principles are not the whole answer, however. The four described above cannot account for the whole of acquisition. But as soon as we start adding to them – Slobin eventually listed about forty – then they start to clash with one another. Every time we find one which doesn’t work, we can claim it is because another one is in operation, cancelling out the first. This leads to the whole idea being vacuous (Bowerman 1986). Unless we can find out which have precedence, and how children cope in cases of conflict, then we are back at square one – looking for some basic principles which guide children through the morass of possibilities.

Some people have suggested that the principles interact with particular languages. If one operating principle works well, due to the structure of the language concerned, then it is given priority over others (Bates and MacWhinney 1987; MacWhinney 1987, MacWhinney and Bates 1989). For example, English has a fairly rigid word order. Children will repeatedly come across the same order for (say) verb + adverb (WALK SLOWLY, HOLD IT CAREFULLY, SHUT IT QUIETLY). Counter-evidence such as SOFTLY FALLS THE LIGHT OF DAY is rare. Evidence of word order is thus both easily available and reliable, so English children will pay particular attention to the order of words. In another language, such as Turkish, the ends of words may get extra scrutiny. According to this view, then, the various operating principles compete with one another, and the structure of the language determines which ones will win out over others.

In short, children are enormously good at sussing out how their own language works. But perhaps we need more than the vague notion of competing strategies. Let us try to be more precise.

ADVANCING AND RETREATING

How do children advance? And how do they retreat? Being endowed with processing mechanisms which involve certain outline expectations about language does not tell us exactly
how
a child acquires any particular construction. Nor does it tell us how children manage to abandon their mistakes. Let us consider these matters.

A construction does not pop up suddenly, like a chicken out of an egg. There may be quite a gap between its
emergence
(first appearance) and its
acquisition
(reliable use). A typical profile of a developing structure was outlined earlier, when we discussed Sally’s past tenses (
Chapter 6
). Judging from Sally’s behaviour, children learn the first examples of a construction by rote, without fully analysing them. In this way, a structure gets a firm hold in a few places. The child then tentatively experiments by extending it to new examples. If she gets reinforcement for these experiments, the construction is likely to proliferate, affecting more and more vocabulary items. As an end result, a rule is acquired.

This general pattern of ‘lexical diffusion’ (
Chapter 6
) occurs in more complex constructions also, such as sentences which contain the sequence TO + verb (Bloom
et al
. 1984):

FELIX TRIED TO REACH THE APPLE.

I WANT THAT DOG TO STAY OUTSIDE.

The earliest examples of this construction occurred without any overt appearance of TO, as in:

I WANNA PEE-PEE.

I WANNA TAKE KITTY.

where the child had no realization that TO is a separate item. When a distinct TO did appear, children behaved as if it was fastened to the end of the previous verb, each of which was one of a small group of newly acquired verbs, such as TRY TO, LIKE TO, SUPPOSED TO, as in:

Gradually, they added in more and more verbs. They also re-analysed their old WANNA sequences, and produced utterances such as:

Finally, they began to acquire sentences in which a noun occurred between the first verb and TO:

I’LL HELP YOU TO FIND THE BUTTONS.

I WANT THIS DOLL TO STAY HERE.

They therefore realized that TO was more closely associated with the second verb.

As the move from WANNA to WANT TO suggests, verbs may be the key to understanding how children move forwards. Let us consider this.

VERBS AS MAYPOLES

Verbs are the maypole around which a sentence revolves. This may seem odd, since verbs vary so much. If an adult describes, say, an egg being dropped, he or she will use the word EGG, but the verb will vary from person to person: ‘Mildred
dropped
the egg’, ‘The egg
slipped
through her fingers’, ‘The egg
smashed
on the ground’, ‘The egg
broke
’, and so on. In one experiment, adults who were shown a videoclip agreed on the verb less than 10 per cent of the time (Gleitman and Gillette 1995).

But on reflection, verbs are not so strange. They describe straightforward events or linked ones. To ‘boil potatoes’ covers several different actions, but all are causally related. No verb means ‘Simultaneously, John yawned and the cat fell off the roof.’ No verb SUBNOUGATE exists for ‘To eat the bottom caramels in a candy box and carefully replace the top level, hoping no one will notice’ (Pinker 1989: 196).

Verb structure ties in with the verb’s meaning. Take the sentences PENELOPE SNEEZED and PETE KICKED THE CAT. English speakers know that only Penelope is involved in the action in the first, and that Pete is doing something to something else in the second. And children pay careful attention to the words round the verb, as shown by Kelli, a blind child (Landau and Gleitman 1985).

Kelli couldn’t see, but she learned to speak with only a marginal delay, compared with sighted children. She, like other children, focused on verbs and the words accompanying them. For example, she distinguished LOOK and SEE by paying attention to the different ways in which her mother used these words:

LOOK
, HERE’S HOW YOU WIND UP THE CLOCK.

YOU
LOOK
LIKE A KANGAROO.

SEE
IF YOU CAN PUT THE SLIPPER ON.

LET’S
SEE
IF GRANNY’S HOME.

These structural differences were important – though some of the clues also related to meaning. Kelli’s mother tended to use LOOK when an object was near at hand:

LET’S
LOOK
AT THIS (where LOOK meant ‘feel’).

and SEE when it was further away:

COME AND
SEE
THE KITTY.

As Kelli shows, syntax and meaning are intertwined in a way that is not always easy to tease out.

The realization that verbs are the key to children’s speech has led to considerable further work, and the way youngsters acquire them is slowly becoming clearer (Tomasello 1992, 2003, Tomasello and Brooks 1998). Michael Tomasello noted that his daughter Travis was young when she learned a wide array of verbs and relational words – 162 before her second birthday. Change of state verbs came early, for example, FALL-DOWN when she fell in a pool, and so did activity verbs, as with TRAVIS LICK-IT, as she licked a popsicle (iced lolly). Each of these verb uses seemed to be independent of others. Travis’s earliest three or more word sentences, produced between the ages of 18 – 21 months, were almost all structured by verbs, and they typically involved building on word combinations that were already in use, such as FALL-DOWN WEEZER which she announced as she dropped the cat, whose name was Weezer, and WEEZER LICK-IT ARMS, as Weezer licked her arms.

In these early stages, Travis showed very few signs of having broad, general grammatical rules. Mostly, she worked on a verb-by-verb basis, with specific people performing the verb’s actions, somewhat like the ‘limited scope formulae’ identified by Martin Braine (
Chapter 6
). Tomasello has labelled his verb findings the ‘Verb Island hypothesis’, which stresses the idea that the known verbs are isolated islands of knowledge, not yet linked up into broader rules.

BACKTRACKING

According to the view outlined above, children listen carefully to what people say, and add on verbs one by one. But youngsters do more than this: they sometimes generalize their knowledge to new verbs. The puzzle of why children do not go ahead and produce enormously overgeneralized grammars ‘constitutes one of the most intriguing and difficult challenges for all students of language acquisition’ (Bowerman 1988: 73). Researchers are still trying to discover how children acquire subtle verb distinctions, both in English and other languages, especially in cases where a verb behaves in an unpredictable way. A child needs to notice that you can say either:

MARION BAKED A CAKE FOR PETER.

or:

MARION BAKED PETER A CAKE.

But this double possibility isn’t always available. You can say:

DONALD OPENED THE DOOR FOR PAMELA.

but not:

*DONALD OPENED PAMELA THE DOOR.

The recipient has to be able to possess the object in order to come in front of it. The problem is clear, but the answer is not: ‘The essential challenge thus becomes that of developing a theory that allows just the appropriate degree of productivity … avoiding the opposing pitfalls of overgeneralization on the one hand and undergeneralization on the other’ (Baker 1992). Similarly, you can say:

KEITH GAVE HELEN A BUNCH OF ROSES.

or:

KEITH GAVE A BUNCH OF ROSES TO HELEN.

You could also say:

KEITH GAVE HELEN A HEADACHE.

but not:

*KEITH GAVE A HEADACHE TO HELEN.

You can only give something to someone if a change of location is involved. Apparently, children notice these subtle meaning distinctions and tie them in with the syntax (Pinker 1989) – though how they do this is still unclear.

Yet overgeneralizations are less frequent than is sometimes assumed. They appear common because people tend to notice odd sentences such as:

MOMMY, OPEN HADWEN THE DOOR (Mazurkewich and White 1984).

But how do children retreat from these erroneous forms? It’s not yet clear. Perhaps they are just experimenting, so do not have these wrong rules truly
fixed in their minds. Or perhaps at this stage children are ultra-sensitive to constructions they are working on (
Chapter 4
).

In this chapter, then, we have tried to see exactly how children extract grammar from the data they hear around them. Chomsky appears to be wrong when he suggests that children are born with detailed linguistic knowledge which is triggered by only minimal exposure to language. In place of a Content Cuthbert, a child whose mind contains chunks of information about language, we should substitute a Process Peggy, a child whose mind is set up with puzzle-solving equipment.

Process Peggy seems to be geared specifically to language. Her achievements cannot be explained solely by her daily needs, her general cognitive ability, or her parents’ speech, though these undoubtedly help her as she struggles to solve linguistic puzzles.

We now have some general idea of the kinds of linguistic expectations which Process Peggy brings to language, and how she advances as she acquires each new construction though we are less clear about how she backtracks, when she discovers she has made a mistake. The exact proportion of specific language mechanisms to other aspects of intelligence is also unclear. It may be that the two are so inextricably mixed, that perhaps we never shall succeed in fully untangling them. This is another question for the future.