Language Development

The process by which children acquire their first language in early childhood.

Human infants are acutely attuned to the human voice, and prefer it above all other sounds. In fact, they prefer the higher pitch ranges characteristic of female voices. They are also attentive to the human face, particularly the eyes, which they stare at even more if the face is talking. These preferences are present at birth, and some research indicates that babies even listen to their mother's voice during the last few months of pregnancy. Babies who were read to by their mothers while in the womb showed the ability to pick out her voice from among other female voices.

Infancy

Since the early 1970s, it has been known that babies can detect very subtle differences between English phonemes (the functional units of speech sound). For example, they can detect the difference between "pa" and "ba," or between "da" and "ga." Of course, they do not attach meaning to the differences for 12 months or more. The original technique of investigating this capacity capitalized on babies' innate ability to suck on a nipple. The nipple is linked to a device that delivers sound contingent on the baby's sucking. Babies introduced to this device suck vigorously to hear the sound, even when it is a repetitive "ba ba ba ba." Because babies also get bored with repetition, they stop sucking hard after a few minutes. At that point the researcher can change the sound in subtle ways, and see if the baby shows renewed interest. For example, it might be a different example of "ba," perhaps one with a bit more breathiness. Or, it could play a sound that would fall into a new phoneme class for adults, like "pa." Babies ignore the first kind of change, just as adults would, but they suck with new vigor for the new phoneme.

Babies have finely tuned perception when it comes to speech sounds, and, more importantly, they seem to classify many sounds the same way adult speakers would, a phenomenon known as categorical perception . These sounds that they perceive as indivisible categories are generally those that form the basis for many speech systems in the world's languages, rather than those that are used only rarely, like "th." Infants come into the world already predisposed to make certain distinctions and classifications: apparently they are not driven to make them by language exposure.

Babbling

At the beginning of infancy, vegetative noises and crying predominate. Observers note that by the age of four months, the baby's repertoire has expanded in more interesting ways. By this point babies are smiling at caregivers and in doing so they engage in a cooing noise that is irresistible to most parents. When the baby is being fed or changed, she will frequently lock gazes with her caregiver and coo in a pleasant way, often making noises that sound like "hi," and gurgles. It is common for the caregiver to respond by echoing these noises, thereby creating an elaborate interchange that can last many minutes. This may not happen universally, however, as not all cultures take the baby's vocalization so seriously. The nature of the sounds made at this stage is not fully speech-like, though there are open mouth noises like vowels, and an occasional "closure" akin to a consonant, but without the full properties that normally make a syllable out of the two.

At some point between four and 10 months, the infant begins producing more speech-like syllables, with a full resonant vowel and an appropriate "closure" of the stream of sound, approaching a true consonant. This stage is called "canonical babbling."

At about six to eight months, the range of vocalizations grows dramatically, and babies can spend hours practicing the sounds they can make with their mouths. Not all of these are human phonemes, and not all of them are found in the language around them. Research has shown that Japanese and American infants sound alike at this stage, and even congenitally deaf infants babble, though less frequently. These facts suggest that the infant is "exercising" her speech organs, but is not being guided very much, if at all, by what she has heard.

By age 10 or 12 months, however, the range of sounds being produced has somewhat narrowed, and now babies' babbling in different cultures begin to take on sound characteristics of the language that surrounds them. The babbling at this stage often consists of reduplicated syllables like "bababa" or "dadada" or "mamama." It is no accident that most of the world's languages have chosen, as names for parents, some variant of "papa," "mama," "dada," "nana." These coincide with articulations that baby can make most easily at the end of the first year.

Toddlerhood

The first words make their appearance any time between nine and 15 months or so, depending on the child's precocity and the parent's enthusiasm in noticing. That is, the baby begins making sounds that occur fairly

Average vocabulary growth of children from ages 1 to 7.

reliably in some situations AND are at least a vague approximation to an adult-sounding word.

What the baby "means" by these sounds is questionable at first. But before long, the baby uses the sounds to draw a caregiver's attention, and persists until she gets it, or uses a sound to demand an object, and persists until it is given to her. At this point the first words are being used communicatively as well. There is a fairly protracted period for most babies in which their first words come and go, as if there is a "word of the week" that replaces those gone before. One of the characteristics about these first words is that they may be situation-specific, such as the case of a child who says "car" only when looking down on the roofs of cars from her balcony. But after several months of slow growth, there is an explosion of new words, often called the "word spurt." This usually coincides with an interest in what things are called, e.g., the child asking some variant of "What's that?" Vocabulary climbs precipitously from then on—an estimated nine new words a day from ages 2 to 18 years. These developments are noted in all the cultures that have been studied to date.

The nature of the child's first 50 words is quite similar across cultures: the child often names foods, pets, animals, family members, toys, vehicles and clothing that the child can manipulate. Most of what is named can either move or be moved by the child: she generally omits words for furniture, geographical features, buildings, weather and so forth. Children vary in that some develop an early vocabulary almost exclusively of "thing" words and actions, whereas others develop a social language: words for social routines, and expressions of love, and greetings. Researchers differ as to whether these are seen as different styles inherent in the child or whether their social environment encourages them in different ways. Researchers agree that the child learns most effectively from social and interactive routines with an accomplished talker (who may be an older child), and not, at least at the start, from passive observations of adults talking, or from radio or TV shows. Experiments and observations show that children pick up words at this stage most rapidly when the caregiver uses them to name or comment on what the child is already focused on.

Word meanings

The meanings of the child's first words are not necessarily the same as those of the adults around her. For instance, children may "overgeneralize" their first words to refer to items beyond their usual scope of application. A child might call all men "Daddy," or all animals "doggie," or all round objects "ball." Others have pointed out that "undergeneralization" also occurs, though it is less likely to be noticed. For instance, a child might call only her own striped ball "ball," and stay silent about all the rest, or refer to the family dog and others of the same type as "doggie" but not name any others. The child may also use a word to refer to a wide variety of objects that hold no single property in common. A child who learned "moon" for the full moon later used it for street lamps, house lights (lights in common), doorknobs and the dial on the dishwasher (shape in common), and toenail clippings on a rug (related shape). Put into a class, these objects share nothing in common except a shifting form of resemblance to the original moon. It has been argued that children's first word meanings have only a family resemblance rather than a common thread. In fact, there are philosophers who argue that such is the nature of many adult words as well.

It has long been recognized that words are inherently ambiguous even when an object is being pointed at: does the word refer to the object, or its color, shape, texture, function, shadow? Recent work on word learning has also drawn attention to the biases the child brings to word learning. One such bias is the Whole-Object assumption, that is, children assume a new word refers to the object itself rather than a property. However, a competing constraint is mutual exclusivity : if a child already knows a word for an object, a new word is assumed to mean something else; a new object if it is available; or a part, texture, or shape of a known one. Researchers are divided at present on the extent to which these biases are learned, or inherent.

Young children also frequently name objects at an intermediate level of abstraction known as the basic object level. That is, they will use the word dog, rather than the more specific collie or the more general, animal, or flower rather than dandelion or plant. This coincides with the naming practices of most parents, and seems to be the level of greatest utility for the two-year old.

Preschool years: the two-year-old

Grammar: the two-word utterance

The first sentence is the transition that separates humans from other creatures. Most toddlers produce their first spontaneous two-word sentence at 18 to 24 months, usually once they have acquired between 50 and 500 words. Before their first sentence, they often achieve the effect of complex expressions by stringing together their simple words:

Book

Mine

Read

Then their first sentence puts these words under a single intonational envelope, with no pause. Their first sentences are not profound, but they represent a major advance in the expression of meaning. The listener is also freed of some of the burden of interpretation and does not need to guess so much from context.

For children learning English, their first sentences are telegraphic, that is, content words predominate, primarily the nouns and verbs necessary in the situation. Words that have grammatical functions, but do not themselves make reference, such as articles, prepositions and auxiliary verbs, do not occur very often. The true character of this grammar is hotly debated. The fact that the function words and inflections appear variably for a protracted period of months leads some researchers to argue that the child really knows the grammar but has some kind of production limit that precludes saying extra words. On the other side, some researchers argue that the forms that do appear may be imitations, or particular learned fragments, and that the full grammar is not yet present. Tests of comprehension or judgment that might decide between these alternatives are very hard to undertake with two-year-old children, though the little work that does exist suggests children are sensitive to the items they omit in their own speech.

At the start, the child combines the single words into two-word strings that usually preserve the common order of parents' sentences in English. At the time the English-speaking child is producing many two-word utterances, comprehension tests show he can also distinguish between sentences that contrast in word order and hence meaning:

The dog licks the cat.

The cat licks the dog.

Researchers using innovative techniques with preverbal infants have claimed infants understand basic word order contrasts before they learn to produce them. Infants who saw a choice of two brief movies along with spoken sentences preferred to look at the movie of the event that was congruent with the spoken sentence, where the only contrast was in word order.

Semantic relations

Most studies on early child language conclude that the child at the two-word stage is concerned with the expression of a small set of semantic relationships. The cross-linguistic study of children includes languages as remotely related as French, Samoan, Luo (spoken in Kenya), German, Finnish, and Cakchiquel (a Mayan language spoken in Guatemala). Two-yearold children learning all these languages expressed only a narrow range of the possible meanings that the adult language could express. All over the world, children apparently talk about the same meanings—or ideas—in their first sentences, despite the variety of forms in those languages. For example, the children refer to possession (Mommy dish, my coat), action-object sequences (hit ball, drop fork), attribute of an object (big truck, wet pants) or an object's location (cup shelf, teddy bed).

Debate has raged over how significant this finding of universal semantic relations is for the study of grammatical development. On the one hand, it might mean that building a grammar based on meaningful relations is a universal first step for language learning. On the other hand, there is the larger problem of how the child builds a grammar that resembles the adult's, because for true linguistic competence, the child needs to build a theory out of the right components: subjects, objects, noun phrases, verb phrases, and the rest. These abstract categories do not translate easily into semantic relations, if at all. To succeed at analyzing or parsing adult sentences into their true grammatical parts, the child must go beyond general meaning. The alternative interpretation of the findings about the first sentences is that children all over the world are constrained by their cognitive development to talk about the same ideas and that their doing so need not mean that their grammars are based solely on semantic relations. So the semantic analysis of children's early sentences offers fascinating data on the meanings children express at that age, but it is less clear that these semantic notions are the components out of which children's grammars are constructed. A weaker hypothesis about the role of semantics in the learning of grammar is that perhaps children exploit the correlation between certain grammatical notions, like subject, and certain semantic notions, like agent, to begin parsing adult sentences. The child could then proceed to analyze sentences by knowing already:

a. the meaning of the individual words

b. the conceptual structure of the event, namely that dog is the agent; bit is the action.

Some have proposed that the child may have some further, possibly innate, "hypotheses" that guide his code-cracking:

c. actions are usually verbs

d. things are usually nouns

e. agents are usually subjects.

Semantic notions then become vital bootstraps for the learning of grammar.

Preschool years: the three-year-old

Shades of meaning

What is missing from the two-word stage are all the modulations of meaning, the fine tunings, which add immeasurably to the subtlety of what we can express. Consider the shades of meaning in the following sentences:

He played

He's playing

He was playing

He has played

He had played

He will play

He will have played

Not all languages make these distinctions explicitly, and some languages make distinctions that English does not. In the next stage of development of English, the extra little function words and inflections that modulate the meaning of the major syntactic relations make their appearance, though it is years until they are fully mastered. For English, it is common to measure the stage of language development by counting and then averaging the morphemes (words and inflections) in a child's set of utterances, and refer to that as the mean length of utterance (MLU). The inflections are surprisingly variable in children's utterances, sometimes present and sometimes absent even within the same stretch of conversation. According to psychologist R. Brown, "All these, like an intricate sort of ivy, begin to grow up between and among the major constituent blocks, the nouns and verbs, to which stage I is largely limited."

A classic error noticed in the acquisition of English inflections is the overgeneralization of plurals and past tenses. In each case, when the regular inflection begins to be mastered, it is overgeneralized to irregular forms, resulting in errors like foots, sheeps, goed and eated. In the case of the past tense, children usually begin by correctly using a few irregular forms like fell and broke, perhaps because these forms are frequent in the input and the child learns them by rote. At first they may not be fully analyzed as past tenses of the corresponding verbs fall and break. But when the child begins to produce regular past tense endings, the irregulars are sometimes also regularized (e.g. falled and breaked). Two kinds of overgeneralizations occur: one in which the -ed ending is attached to the root form of the irregular verb (e.g. singsinged) and the other in which the ending is attached to the irregular past form (e.g. broke-broked).

Cross-linguistic work

An understanding of how children acquire grammatical morphemes is now thought to require a broader perspective than that obtained from studying English alone. A large research initiative has gathered data from children acquiring other languages, especially languages very different from English. Researchers have studied children acquiring Luo, Samoan, Kaluli, Hungarian, Sesotho and many others in an effort to understand the process of language acquisition in universal terms. One finding is that the telegraphic speech style of English children is not universal—in more heavily inflected languages like Italian, even the youngest speakers do not strip their sentences to the bare stems of nouns and verbs.

One of the purposes of the cross-linguistic work is to try to disentangle some of the variables that are confounded in a single language. For example: English-speaking children acquire the hypothetical (if…then statements) rather late, around four years of age, but the hypothetical form is complex in English grammar. It requires an ability to imagine an unreal situation. Cross-linguistic studies provide a way to tease these variables apart, for Russian has a very simple hypothetical form, though its meaning is as complex as the English version. Research shows that Russian children do not use this simple form until after they are about four years of age. Most morphemes vary along multiple dimensions: phonological, semantic and grammatical. The full program of research may reach fruition only when the massive matrix of possibilities across the world's languages can be entered into a computer, complete with detailed longitudinal data from children learning those languages.

Auxiliaries

Children's first sentences lack any auxiliaries or tense markers:

Me go home

Daddy have tea

and they also lack auxiliary-inversion for questions at this stage:

I ride train?

Sit chair?

They also lack a system for assigning nominative case to the subject, that is, adult sentences mark the subject as nominative:

Adult: I want that book

but children at this stage frequently use the accusative case:

Child: Me want that book

These facts lead some to conclude that young children's sentences lack the full syntactic structures typical of adult sentences, and undergo a radical restructuring as they develop. Others argue that the limitation is not so much at the level of knowledge of grammar, but merely performance limits, so preserving the continuity of form at an abstract level between child and adult.

In addition to learning the basic word order and inflectional system of the language, a child must learn how to produce sentences of different kinds: not just simple active declarative, but also negatives, questions, imperatives, passives and so forth. In English there are word order changes and auxiliary changes for these sentence modalities.

One type of question is called a yes/no question, for the simple reason that it requires a yes or a no answer. A second kind of question is called the Wh-question, socalled because it usually begins with the sequence Wh in English (in French, they are Qu-questions). Wh-questions do not require a simple yes or no response: instead they ask for information about one of the constituents in the sentence. What, who, when, where, why, and how all stand in for possible phrases in the sentence—the subject, or object, or a prepositional phrase. Discourse permits us to respond elliptically with only the missing constituent if we choose:

What is he buying?

Coffee.

Where is she going?

To the store.

How is she getting there?

By bike.

The structure of such questions is similar to that of yes/no questions because the auxiliary and subject are inverted, so that transformation is involved in both. In addition, the Wh-word is in initial position, though it stands for constituents in varied sentence positions. Linguistic evidence suggests that the Wh-word originated at another site in the structure and was moved there by a grammatical rule, called, appropriately, Wh-movement. Children's responses to such questions reveal the sophisticated nature of their grammatical knowledge.

Negation also involves the auxiliary component in the sentence, because for simple sentence negation, the negative is attached to the first member of the auxiliary, and may be contracted:

She isn't coming home.

He won't be having any.

How do children acquire these rules of English? When auxiliaries do emerge, it seems that they come in first in declarative sentences. Before children master the placement of the auxiliary, they ask questions using rising intonation. They may also pick up a few routine forms of yes/no questions, particularly in households that demand politeness from young children, as in:

May I have one?

When auxiliaries do begin to appear in initial position, what has the child learned? One of the claims made by modern linguistic theory is that the rules of natural languages are "structure dependent," that is, they always refer to structural units, constituents such as "noun phrase" or "auxiliary verb," not to other arbitrary units such as "the fifth word" or "the first word beginning with 'f'." The case of auxiliary inversion provides a nice illustration, used by Noam Chomsky to make this point. The child could hear sentence pairings such as:

The man is here,

Is the man here?

The boy can swim.

Can the boy swim?

The dog will bite.

Will the dog bite?

and draw the conclusion that to make a question, you take the third word and move it to the front. Of course, that hypothesis would soon be disconfirmed by a pair such as:

The tall man will come.

Will the tall man come?

not: Man the tall will come?

More likely, the child might form the rule "move the first word like can, will, is, etc. up to the front," which would fit all of the above and hundreds of other such sentences. However, that is not a structure-dependent rule, because it makes no reference to the grammatical role that word plays in the sentence. The only disconfirmation would come from the occasions when a subject relative clause appears before the auxiliary:

The man who is the teacher will be coming tomorrow.

Will the man who is the teacher be coming tomorrow?

but our earlier, structure-independent rule would produce:

Is the man who the teacher will be coming tomorrow?

The child who formulated the almost-adequate rule would fail in such circumstances, but no child has been observed to make the mistake. Hence even from the inadequate data that children receive, they formulate a complex, structure-dependent rule.

Wh-questions

Wh-questions appear among the child's first utterances, often in a routine form such as "Whazzat?" The forms are routines because they are invariant in form, but more varied productions are not slow to emerge in children's grammar. The first, stereotyped forms may be tied to particular functions or contexts, but genuine interrogatives are varied not only in form but in use.

Just as in yes/no questions, the auxiliary must be in front of the subject noun phrase in a Wh-question, and children seem to have more difficulty with auxiliary-inversion in Wh-questions than in yes/no questions. At the same time children can say:

Can he come?

they might say:

Why he can come?

failing to invert the auxiliary in the Wh-question.

What else does the child have to learn in Wh-questions? One factor concerns the link between the Whword and the "missing constituent." Certain of the Whwords enter children's speech earlier than others, and there is some consistency across studies in that order: What, who, and where tend to emerge before why and how, with when coming later. Some have explained the order in terms of semantics, or rather concreteness, of the ideas contained in these words, since when and how depend upon cognitive developments of time and causality whereas what and who do not. The question why seems to be late for this reason: it is only through discourse that a child can determine the meaning of why, which may be the reason some young children ask it endlessly. It is also a question that rarely elicits a one-word answer, so it may be a way to keep the conversation going when you can't say much yourself yet!

Creativity

A feature that is markedly evident in young children is their creativity with language. Children, like adults, continually produce sentences they have not heard before, and one can more easily recognize that novelty in children because sometimes the ideas are rather strange. For example, after hearing many "tag questions" such as "That's nice, isn't it?" and "You're a good girl, aren't you?" and "You can open that, can't you?" a three-yearold figured out how to make her own tags, and used the rules to say, "Goosebumps are hairy legs, aren't they?" and "He's a punk rocker, isn't he?," which were definitely not sentences she had heard. In addition, the creative use is revealed because children overextend rules to exceptional cases. For example, a child may say "My porridge is getting middle-sizeder" as he struggles through a huge bowl of oatmeal. It can also occur because children do not yet have the vocabulary for certain subtleties of expression. But the way that children fill these "lexical gaps" uses the same principles as adults who do the same thing. For example, an adult might use an "innovative verb" such as "I weekended in New York," and a child might similarly say, "I broomed her!" after pursuing a sibling with a broom. However, a child who said "You have to scale it first" as she put a bag on a scale was creating an innovation for which there is already an existing word—namely, weigh. The creativity of children's linguistic innovations has been emphasized because it demonstrates that children do not just imitate what they hear, but extract general rules and principles that allow them to form new expressions.

Later preschool years

Joining sentences

Once the child has mastered the fundamentals of sentence construction, what is left to learn? Actually, language would be very dull to listen to or read if we could just produce simple sentences with one verb at a time. Perhaps the first response of a novice to the field of child language is that the sentences children speak are short and not very complicated for a long period. Certainly when one measures the mean length of utterance of children younger than age four, it tends not to be very impressive, ranging from 1.0 to 4.0 morphemes per utterance. Yet by age four, the MLU (mean length of utterance) loses much of its usefulness as a measure, because children's utterances, like those of an adult, fluctuate in length dramatically depending on the circumstances of the conversation. Even before age four, there are rare, but significant, occurrences of surprising complexity, showing that the child is in command of a considerable amount of grammar when needed. The first sentences involving more than one "proposition" are simple coordinations, for instance two sentences joined by and. Later other conjunctions come in, such as so, but, after, or because. But embeddings are not much later: there is evidence of embedded structures even in the primitive talk of two-year-olds.

There are different kinds of embedded structures. One kind are relative clauses, clauses that are used to further specify a noun phrase:

The man who took the job is coming to dinner.

Here is a sample sentence from a child at 2;10 (2 years, 10 months), said in reference to playground equipment:

I'm going on the one that you're sitting on. or the slightly aberrant:

Where's a hammer we nailed those nails in?

On the other hand are complement constructions, which can be considered the equivalent further specification of the verb phrase:

The doctor decided to perform the operation.

Again, a child at age 2;11 was observed to say:

I don't like Nicky share a banana.

I'm going downstairs to see what Nicky's watching.

Both kinds of embedding are means of packing information into a single sentence that would require multiple sentences (probably with lots of pointing) to convey the equivalent ideas. When children reach the stage at which they can control these and similar structures, they become capable of expressing a much wider variety of ideas and thoughts not dependent on the immediate environment for support, and an important further step is taken in being ready for literacy.

Researchers have used innovative procedures to elicit relative clause structures from children as young as two by arranging the situation to call for specification of a referent. In one procedure, for example, the child, the experimenter, and a confederate are playing with two identical toy bears. The experimenter makes one bear ride a bike. Then the confederate is blindfolded, and the child alone watches the experimenter make that same bear do another action, say jump. Then the blindfold is removed from the confederate and the child has to help him guess which bear did something. Children of two and three can say:

Pick the one that rode the bike.

If the literature on comprehension of relative clauses is considered, it appears that children below age five are in very poor control of relative clause sentences. The typical comprehension task uses an "act-out" procedure in which several small animals are provided to the child and he is asked to act out whatever the experimenter says. After a couple of simple warm-ups, e.g.,

Show me:

The lion hit the kangaroo.

The dog jumped.

the child would be asked to act out relative clause structures in which there are no clues to meaning from the words alone, i.e., the syntax carries all the meaning:

The lion that hit the dog bit the turtle.

The cat that the dog pushed licked the mouse.

When preschoolers are given such a task, their performance is usually fairly poor, suggesting that they continue to have difficulty reconstructing the speaker's meaning from complex structures: a problem perhaps in processing rather than grammar per se.

Similarly, even five-and six-year-olds continue to have trouble figuring out who did what to whom for sentences containing various kinds of complements:

Fred told Harry to wash the car.

Fred promised Harry to wash the car.

Fred told Harry that he washed the car.

Fred told Harry after he washed the car.

The various "complement-taking" verbs in English fall into several distinct patterns, as do the complements themselves, so there is room for lots of confusion.

Finally, there are aspects of the pronoun system that may take several years to get straight. Pronouns in English have to have an "antecedent" (noun which is referred to by the pronoun) outside the sentence in which the pronoun occurs: you can't say, for example:

John hit him.

and mean John hit himself. Reflexives like "himself," on the other hand, have to be in the same clause as their antecedent; you can't say:

John was wondering why Fred hit himself.

and have it mean that Fred hit John. Children's control over antecedents, particularly of pronouns, is still being acquired after age four or five when complex sentences are involved.

Later word learning

The child's vocabulary grows enormously in the age period two to five years, and vocabulary size is frequently used by researchers as an index of the child's development. In addition to learning many new nouns and verbs, the child must organize vocabulary, for example, into hierarchies: that Rover is also a dog, a corgi, an animal, a living thing and so on. The child also learns about opposites and relatedness—all necessary forms of connection among words in the "inner lexicon." The child also becomes better able to learn words from linguistic context alone, rapidly homing in on the meaning after only a few scattered exposures. This is a surprisingly effective process, though hardly fail-safe: after being told that screens were to stop flies from bringing germs into the house, one child concluded that germs were "things flies play with."

Discourse and reference

Researchers have been acutely aware that the child's language learning does not take place in a vacuum or a laboratory—it is enmeshed in the social relationships and circumstances of the child. The child uses language for communication with peers, siblings, parents, and increasingly, relative strangers. All of these individuals make special demands on the child in terms of their different status, knowledge, requirements of politeness, clarity or formality, to which the child must adjust and adapt, and the preschool child is only beginning this process of language socialization. Even four-year-olds adjust their style, pitch and sentence length when talking to younger children or infants rather than peers or older people, and in other cultures they master formal devices that acknowledge the status or group membership of different people. However, it is recognized that the three-year-old is rather poor at predicting what others know or think, and therefore will be rather egocentric in expressing himself. Especially when communicating across a barrier or over a telephone, the child of this age might be unable to supply the right kind of information to a listener. However, other researchers show that children become increasingly adept at "repairing" their own communicative breakdowns as they get older.

Narrative and literacy

The difficulty that children have with predicting what others already know or believe shows itself also in their attempts to produce narratives, that is, extended sentences that convey a story. Retelling a story is considerably easier than constructing one about witnessed events, but may need considerable "scaffolding" by a patient listener who structures it by asking leading questions. Skill in producing a coherent narrative is one of the culminating achievements of language acquisition, but it is acquired late and varies widely according to opportunity for practice and experience with stories. In part, this is because creating a narrative is a cultural event: different cultures have different rules for how stories are structured, which must be learned. At first, children tend to focus just on the actions, with little attention to the motives, or reasons, or consequences of those actions, and little overarching structure that might explain the events. Young children also fail to use the linguistic devices that maintain cohesion among referents, so they may switch from talking about one character to another and call them all "he," to the bewilderment of the listener. Reading and writing in the grade school years depend on this ability and nurture it further, and one of the best predictors of reading readiness is how much children were read to in the first few years. As children begin to read and write, there are further gains in their vocabulary (and new ways to acquire it) and new syntactic forms emerge that are relatively rare in speaking but play important roles in text, such as stage-setting and maintaining cohesion. Mastery of these devices requires a sensitivity to the reader's needs, and it is a lifelong developmental process.