The Progression to Adult Language
Typically, children’s speech progresses very rapidly by the beginning of the third year of life. Utterances now become longer and the children can say short sentences. Function words begin to appear. By 5 years the average child sounds much like an adult in the forms of her speech. How do children reach this level of sophistication? It is certainly not by memorizing all the sentences that are said to them. For one thing, we know that people, even little children, can understand sentences that are quite novel and even bizarre, such as There is a unicorn hiding behind your left ear, the very first time they hear them. More gen-erally, a good estimate of the number of English sentences less than 20 words in length is 230 (a bit more than a billion). A child memorizing a new sentence every 5 seconds, start-ing at birth, working at this 24 hours a day, would have mastered only 3% of this set by his fifth birthday, making the learning-by-memorizing strategy hopelessly implausible. But what is the alternative? We can get some hints by looking at the relatively simple case of word building. As so often, “errors” that the youngsters make along the way reveal some-thing of the procedures involved in acquiring the adult system.
An example is the learning of the past tense. When children start using past tense verbs, often they say them correctly at first. They use the suffix -ed as an adult would, applying it to regular verbs, but not to irregular ones. Thus, the child says “walked” and “talked” and also (correctly) uses irregular forms such as ran, came, and ate. By the age of 4 or 5, however, the same children sometimes say “runned,” “bringed,” and “holded” (G. F. Marcus et al., 1992; Prasada & Pinker, 1993a) as in the following conversation:
CHILD:My teacher holded the baby rabbits and we patted them.
MOTHER:Did you say your teacher held the baby rabbits?
MOTHER:What did you say she did?
CHILD:She holded the baby rabbits and we patted them.
MOTHER:Did you say she held them tightly?
CHILD:No, she holded them loosely. (Bellugi, 1971)
This kind of error offers evidence that children do not learn language solely, or even mostly, by imitation. Few adults would say “holded” or “eated,” and the mother in the quoted exchange repeatedly offers the correct form of the verb for imitation. In fact, parents are often aghast at these errors. A half-year earlier, their child was speaking cor-rectly but now is making errors. Apparently, he is regressing! So parents often try to correct these errors, but to no avail: The child stands firm, despite the correction.
But if not the result of imitation, then what is it that leads children to produce “errors” even though there was conformity to the adult model’s speech at an earlier developmental moment? Many investigators argue that the young child starts out by memorizing the past tense of each new verb, learning that the past tense of want is wanted, the past tense of climb is climbed, and so on. But this is a highly inefficientstrategy. It is far more efficient to detect the pattern: Simply add the -ed suffix to a verb every time you are speaking of the past. But once the child detects this pattern, it is apparently quite easy to get carried away—to believe that what is very often true must be true in every single case—and so overregularization errors are produced. These errors will drop out later, when the child takes the further step of realizing that, while there is a pattern, there are also exceptions.