Have you ever wondered how children learn to talk? Many people, when asked that question, respond that they do it by imitating. This is at least partially true. Without imitation, we couldn't account for the fact that children in Texas usually learn Texan English, children in Paris usually learn Parisian French, and not vice versa. But imitation as an answer doesn't take us very far. For one thing, children routinely say things they've never heard: "Mommy, come quick—Waldo swallowed a frog!" That is a novel statement for a novel situation. When you think about it, it is inconceivable that children could learn in advance by imitation all of the sentences they will ever have to say.

At this point some would amend their position to say that children don't imitate others sentence by sentence. Instead, they imitate the nouns and verbs and sentence structures of others around them; they can fit their own words into these imitated structures to create novel sentences. But the facts of children's speech do not fit this explanation either. Children produce many sorts of grammatical constructions that they have not heard before. A two-year-old says, "Allgone milk" and "Daddy bye-bye" and for a time, rarely utters sentences of more than two words. A three-year-old says, "I seed two gooses" and "I have small foots"—two particular plural forms that nobody else in the family uses.

At any given point in development, a child's speech more closely resembles the speech of other children at the same stage of development than it does the speech of adults in the child's environment—even if there are not other children around. Any explanation of children's speech that depends on strict imitation cannot stand up to these facts.

What do children do as they learn to talk? Children seek from their early days to make sense of the communication around them. As their minds mature, they attempt—through a sort of gradual trial-and-error process—to construct a system of rules that will allow them to produce sentences like those they hear others use. "Rules" is used here in a loose sense. They are not consciously saying to themselves, "Hmm ... whenever I mean more than one, I must put an S on the end of the noun." Yet some sort of unspoken assumption close to this must have been made or else why would the three-year-old say "gooses" and "foots"?

There is much evidence that children's early sentences result from the use of some sort of rules—and not simply from the haphazard imitation of adult sentences.

Imagine that you are in a kitchen with a two-year-old and his mother. The child is seated in his highchair eating. Suddenly he bangs his cup on the highchair tray and says, "Mommy milk, Mommy milk." We assume from the context—his gesture with the cup and so forth—that he means something like, "Mommy, get me some more milk." If we have spent much time around this child, this may seem like one of his typical sentences—typical for one thing in that, for the past few weeks, at least, we have rarely heard him utter sentences with more than two words in them

On reflection, we may be struck by what a good sentence it is for having only two words! If we had to pick two words to convey the idea in "Mommy, get me some more milk," we could not improve on "Mommy milk." A lot of young children's sentences are like this; that is, they are of a uniform shortness, starting out as one-word sentences. Later, as children mature a bit, they begin to use two-word sentences and then move up to three-word sentences and so on.

Most early sentences are like this sample sentence, too, in that children show a knack for picking the most important words to convey their meanings. "Mommy milk" packs a lot of information; "get more" conveys less. Early sentences use informative words and leave out in-between words such as "and," "to," "with," "should," "have," "will," "the," "very," and the like. We assume that the limits to the number of words children can put in their early sentences have to do with biology and maturity. But the nature of their choice of words and the order they put them in reveals some deliberation, some rules.

Another piece of evidence for the operation of rules in early speech is seen when a child is asked to imitate adult sentences. Normally, young children cannot correctly imitate a sentence that is more complicated than one they could produce on their own. The following exchange between a psychologist and his young daughter illustrates this point:

Child: Want other one spoon, Daddy.
Father: You mean, you want the other spoon.
Child: Yes, I want other one spoon, please Daddy.
Father: Can you say, "The other spoon"?
Child: Other ... one ... spoon.
Father: Say,"other."
Child: Other.
Father: "Spoon."
Child: Spoon.
Father: "Other spoon."
Child: Other ... spoon. Now give me other one spoon?3

Similar difficulty was encountered by a researcher who attempted to lead another child away from an incorrect use of a past tense of the verb "hold":

Child: My teacher holded the baby rabbits and we patted them.
Adult: Did you say your teacher held the baby rabbits?
Child: Yes.
Adult: What did you say?
Child: She holded the baby rabbits and we patted them.
Adult: Did you say she held them tightly?
Child: No, she holded them loosely.4

This child apparently is not going to say "held" until she changes the rule in her head that produces that form. And language rules, like other habits, take some time to change!

Children do not trade in their immature speech for mature speech all at once. They always go through a sequence of stages of language use, moving from simple to complex. Thus, we hear a child at two years of age ask, "Why you singing?" and we note that all of her questions are of the same form. At two years, four months, we hear her ask, "Why you are singing?" and other questions of this more complex form. Just before the age of three, she arrives at the standard form for the English question, "Why are you singing?"5

It is obvious that this child is not learning to talk simply by memorizing sentences or sentence types. Rather, she is formulating her own rules to help her understand sentences she hears around her to produce sentences like them. Once she formulates a rule, she uses it confidently until she begins to notice differences between her sentences and the sentences adults use. Then she will gradually add to and amend her rules so that she is able to produce sentences more like adults'. She doesn't junk her old rules altogether; this would be too disruptive. Feature by feature, she makes her rules more and more like the rules adults use to produce mature sentences.6

Remarkably, children usually go through the same sequence of rule learning as they mature in speech production. A study by Brown7 showed that three separate children started using major grammatical features in roughly the same order:

present progressive I riding
plural nouns two skates
linking verbs I am big
articles before nouns the birds
past tense markers we skated

Child language researchers are not sure why children tend to acquire language rules in the same order, although one theorist has suggested that it may be because children are born "prewired" to learn language in a certain way.8

The uniformity of order surely has nothing to do with what we teach. The language children hear around them cannot be much different from age one to age two, or from age two to age three—though their own language changes dramatically during that time. Whatever the explanation turns out to be, it is bound to be related to language-learning processes going on inside the child.

Not all of children's early speech is different from adult speech. Sometimes we do hear two- and three-year-olds repeating phrases—learned by imitation—that seem more advanced than normal speech for that age. We sometimes hear "Why dincha tell me?" at two and a half, but later, oddly enough, the child reverts to a less mature form: "Why you didn't tell me?" Eventually he will come to use the correct form: "Why didn't you tell me?"

The implication is that some imitated but unassimilated forms may be used for a time as formulas—that is, as whole structures that the child hasn't analyzed and for which rules have not been found that will generate them. But as language development advances, the rules invade the formulas; the utterances produced by formula disappear, and they may not be heard again until the rules have been developed to produce them.

If children construct their own rules to use and understand language, how is it that everyone winds up speaking English instead of her or his own' private language?

We sometimes do hear of sets of twins who—being raised in isolation from others or in other unusual circumstances—make up a private idiomatic language that makes no sense to anyone but themselves.9 But that doesn't happen very often. Every year, millions and millions of children learn to speak English (and, in their respective settings, hundreds of other languages) through their own efforts, without being taught. That is the normal pattern of things.

Clearly, when children construct language rules, they are attempting to find rules or patterns that account for the language used by others in their presence. It is as if they were carefully feeling and probing the language to find its joints and seams, its outer shape and its inner workings.

Children's early hunches about the way spoken language works can be wrong, of course. An area of language where this is sometimes seen is in naming things. We have an example in our young friend, Will, who produced voluminous speech throughout his second and third years. Except for a few words, most of Will's speech was unintelligible to his parents or other adults. One of Will's recognizable words was "bupmum," used to refer to his favorite vehicle, the family's Land Rover (a British-made jeep). According to Will's father, "bupmum" was a pretty fair rendering of the sound made by the exhaust popping out of the Rover's rusted tailpipe. When the family sold it and bought a Volkswagen, Will reflected the change in his name for the new car: "mummum" (a smoother-sounding name for a better-running engine). Later, he used "mummum" to refer to all cars and trucks. Still later, an element of the name showed up in his name for motorboat: "boatmum." At four, Will was speaking standard English. But in those early years, it seemed to those who knew him that he was seeking names for things in the sounds that emanated from them—a perfectly sensible strategy, really, but not one around which English is organized.

So far in this discussion of children's language learning, we have emphasized the child's own efforts to make sense of and construct rules for the language she hears around her. But what do the adults contribute? Have they only to keep up a patter of talk, from which the child can abstract rules of grammar? Such may have been the drift of earlier descriptions of language acquisition, but now it is widely recognized that adults—parents or primary caregivers—are much more actively involved in children's language learning.

First of all, adults do provide the raw material of language from which children construct their own ideas of the way language works. In those fortunately rare cases in which children have been kept isolated from human contact, the children have been found not to have developed language—to no one's surprise. But secondly, it seems clear that when adults are speaking to children, they modify their speech considerably, into a form of speech that is sometimes called "motherese": they use fewer words per utterance and simpler syntax; they speak more slowly and in a higher range (babies have been shown to prefer high-pitched voices to low-pitched ones); and they exaggerate the stress and intonation of their speech. One researcher has compared all this exaggeration to the way an instructor demonstrates a golf swing. It is as if the mother were saying, "Here, pay attention to upness and downness and stress and words—these are the important things."10

But there's more. Most parents in English-speaking countries read to their children. The practice of reading to children has long been believed to help those children learn to read. However, recent assessments of its benefits are more specific. Some argue that reading to children leads them to associate pleasure with written language and enables them to formulate schemata for stories and other forms of written discourse. Other researchers go further and suggest that children who are read to learn a written form of language from the very beginning. They learn that language can be elaborated to explain things that are not in the context of the speech. This decontextualized language is just the sort of language that is used in reading and writing.11

So the picture that emerges from more recent studies of language learning shows that (1) parents are actively involved in their children's language learning, that they tend to direct a form of language toward their children that is easier to learn from than the speech they use with older people; and (2) written language—complete with the word choices and structures of stories, and the use of language to create a world of understanding on its own, a world removed from the context in which it is read—is often part of children's language experience from the very beginning.


3. Martin Braine, "The Acquisition of Language in Infant and Child," in C. E. Reed, ed., The Learning of Language (New York: Appleton-Century-Crofts, 1971). Quoted in Jean Aitchison, The Articulate Mammal (London: Hutchinson, 1973), p. 74.

4. Courtney Cazden, Child Language and Education (New York: Holt, Rinehart & Winston, 1972). Quoted in Aitchison, p. 72.

5. Aitchison.

6. Jill De Villiers and Peter De Villiers, Language Acquisition (Cambridge: Harvard University Press, 1979).

7. Roger Brown, A First Language (Cambridge: M.I.T. Press, 1973).

8. Chomsky.

9. Catherine Gorney, "'Gibberish' Language of Identical Twins Still Baffles the Experts," The Houston Chronicle, 29 July 1979, see. 10, p. 4.

10. Daniel Stern, The First Relationship: Infant and Mother (Cambridge: Harvard University Press, 1977).

11. David Olson, "'See! Jumping!' Some Oral Antecedents of Literacy," in Hillel Goelman, Antoinette Oberg, and Frank Smith, eds., Awakening to Literacy (Portsmouth, NH: Heinemann, 1984).