A biological time-clock ordains the sequence in which the language web is woven, though not the exact dates. But nobody is quite sure when the clock starts ticking, and when it stops. According to Lenneberg, humans are scheduled to acquire language within a critical period between the ages of two and thirteen, a time preordained by human nature. After that, the acquisition of language was difficult, he claimed, maybe impossible.

Lenneberg turns out to be partially wrong and partially right. He is wrong about the starting-point. Language acquisition begins well before the age of two. Babies only a few days old can pick out their own language, according to some research by Jacques Mehler and his colleagues in Paris. The infants responded to French with increased sucking movements, a standard reaction to sounds which interested them. But they did not display the same reaction to other languages. So infants still in the womb may become accustomed to the rhythms of the language spoken around them.

And language development does not come to a shuddering halt at adolescence, as Lenneberg assumed: vocabulary even undergoes a spurt at this time. So the idea of a fixed critical period is now disputed.

Yet most people find it easier to learn languages when they are young--so a sensitive period may exist, a time early in life when acquiring language is easiest, and which tails off gradually, though never entirely.

A 'natural sieve' hypothesis is one idea put forward to explain this. Very young children may extract only certain limited features from what they hear, and may automatically filter out many complexities. Later learners may have lost this inbuilt filter, and be less able to cope as everything pounds in on them simultaneously. A 'tuning-in' hypothesis is another possibility. At each age, a child is naturally attuned to some particular aspect of language. Infants may be tuned in to the sounds, older children to the syntax, and from around ten onwards, the vocabulary becomes a major concern. Selective attention of this type fits in well with what we know about biologically programmed behaviour.

Jean Aitchison

Figure 3. Chomsky's switch-setting.

The outlines of the language web are therefore preordained. Acquiring language involves weaving in the network details of one's own native tongue, with particular portions scheduled to be filled in at particular ages.

Japanese, Welsh or Samoan--children handle all languages with equal efficiency. The American linguist Noam Chomsky has suggested that children might be innately endowed with advance information on the main ways in which languages can vary. So children may have to discover whether they are dealing with an English-type language, which puts verbs in front of its objects, or a Turkish-type one which does the reverse. Once a decision is made, the child metaphorically 'sets a switch', with multiple repercussions. If, as in English, a language has verbs before its objects, as in climb the tree, then it will also probably have prepositions before nouns as in up the tree. A language such as Hindi or Turkish would have the reverse, and say, as it were, the tree climb, and the tree up. It is as if the child was sitting in a linguistic bath, and watching which way water swirled down the plughole, clockwise or anti-clockwise. Once the youngster had found this out, then it would automatically know the linguistic equivalent of whether it was in the northern or southern hemisphere, and whether days got warmer to the north or to the south. In technical terminology, children 'set parameters', a mathematical term for a fixed property whose values vary (see figure 3).

Chomsky makes acquiring language sound like turning on a light, more instantaneous than it really is. But his theory rightly emphasizes that any language holds together in a network of implications. If a language has one type of construction, others are predictable from it.

But natural web-spinning can be both helped and sometimes hindered by the speech of those around. Early research talked of motherese, mother's speech. This left out fathers and friends, so caretaker speech became the fashionable term, later amended to caregiver speech, and in academic publications, to CDS 'child directed speech'. I'll leave it at caregivers. Another term 'baby-talk' is best avoided, because it usually refers to gee-gee, puff-puff, moo-cow-type words, so puzzlingly widespread in England when talking to babies or sending Valentines.

Caregiver speech can be odd. Some parents are more concerned with truth than with language. The ill-formed 'Daddy hat on' might meet with approval, 'Yes, that's right', if daddy was wearing a hat. But the well-formed 'Daddy's got a hat on' might meet with disapproval, 'No, that's wrong', if daddy wasn't wearing a hat. You might expect children to grow up telling the truth, but speaking ungrammatically, as some early researchers pointed out. In fact, the opposite happens.

Parents also reportedly care about etiquette: 'Say please', or they pick on swear words: 'Don't let me hear you say that word again', or they notice occasional pronunciation problems: 'Say Trisha, not Twisha.' If they do pick on language formation, it's often verb endings: this may be useful, if the child is tuned in at that time to learning these. If not, the correction is likely to be ignored. One much-quoted conversation was about baby rabbits:

'My teacher holded the baby rabbits, and we patted them', said the child.
'Would you say she held them tightly?' asked mother.
'Oh no, she holded them loosely', replied her daughter.

At best, a sensitive parent provides support, by being aware of structures to which the child is attuned. Mostly, parents muddle along, sometimes getting it right, sometimes wrong. At worst, a grumbling tone of voice can sap confidence: a child may realize that something is wrong, but not always know what. Only talk directly addressed to the youngster has an effect. Vincent, a hearing child born to deaf parents, learned to communicate with sign language. He himself could hear, and he used to sit in front of the television, and watch the pictures with fascination. But apparently, he did not pay any attention to the sounds. He did not start to speak until he went to school, where people talked to him. And a recent survey in Manchester found that television can delay speech development even in some normal children: they are riveted by the colours and flashing lights, and tune out the sounds.