Language Acquisition Device (LAD)
Language Acquisition Device (LAD)
Notion that some knowledge about language is built into the brain of the human child.
Learning theorists argue that the environments of young children everywhere are supportive of learning. All young children are surrounded by speech from the beginning: parents and others address remarks to babies, respond to their vocalizations and heap praise on their earliest attempts to say words. People adjust their own speech to accommodate the young child's needs, simplifying their vocabulary, shortening their sentences, and talking about the here-and-now, usually in a special speech "register." This motherese or parentese that cues the child that this is talk meant for the child's ears.
There are some theorists who argue that this environmental support gives the child everything necessary to "figure out" the rules of language. In other words, the child is like a miniature linguist, collecting evidence to decide among hypotheses about the grammar. But mathematical linguists have determined that any set of sentences is compatible with an infinite set of possible grammars. In 1968 a startling proof showed that human language is unlearnable in a finite amount of time: there are too many potential rule systems that could have generated the set of sentences a learner has heard at any one time.
One solution is to argue that the child receives accurate corrective feedback about his hypotheses. Under this kind of condition, language would be learnable in a finite time. But evidence for reliable and consistent corrective feedback in the average child's environment is very weak. When it comes to grammar, most parents notice only the superficial mistakes made by their children, and these only when the child is "old enough to know better," e.g., saying/oofs or corned when in grade school. Parents do provide some feedback about the clarity or truth of their children's sentences, but studies reveal that adults do not provide reliable feedback on the grammaticality of children's sentences. Explicit correction seems to be too rarely and inconsistently used to "train" the child to speak grammatically.
The alternative solution is to claim that the child has some preconceived ideas, or innate knowledge. Learning language is possible in a finite amount of time if the learner already knows the range of possibilities existing in universal grammar. Linguistic approaches to language acquisition assume that some knowledge about language is built into the human child.
Linguistic theory states that languages are deeply similar in ways that have only just begun to be uncovered. All languages seem to make use of the same small inventory of categories for the construction of sentences: noun phrases, verb phrases, sentences, and the like. In addition, there are principles that seem to be universal and which constrain the forms that sentences can take. Currently many people are persuaded that these facts might be part of the assumptions that the child brings to the language acquisition task. The child may in fact "know" that sentences are built from the abstract categories, and know in advance the principles that dictate that the rules are going to have a limited range of possible forms. These central ideas were introduced by Noam Chomsky in 1965 and have motivated much work on language development. Chomsky proposed that children are born with a Language Acquisition Device (LAD) that contains hypotheses that guide their language learning.
To demonstrate the subtlety of the abstract principles on which these arguments depend, consider the following short story:
Once there was a boy who loved climbing trees in the forest. One afternoon he slipped and fell to the
Now comes the question:
a) When did the boy say he hurt himself?
Notice there are two possible answers, either to when he said it, or to when he hurt himself. That is, the "when" question could be connected to the "say" or to "hurt." Research shows that three-year-olds also allow both answers: sometimes giving one, sometimes the other. But now consider the subtle variant:
b) When did the boy say how he hurt himself?
Suddenly, the ambiguity is gone, and only one answer seems right: "that night in the bath." Three-yearolds also only give that answer to b). They seem to know already the constraint that question words may not "move" over another question word: a constraint that is embedded in universal grammar.
For a learning theory account to be viable, the two-year-old would have to have the ability to sift through evidence of this subtlety to arrive at the appropriate generalization. The problem that then arises is the rarity of such sentences. In hundreds of hours of recorded conversations between several young children and their caregivers, there are typically only a couple of dozen examples, and of course never the close contrasting pairs described above. It does not seem plausible that the child learns the contrasts for himself.
Instead, it is argued, the child is in possession of considerable pre-existing knowledge about the forms that rules can take. There is still considerable scope for learning theories. Even if the starting point is not a "blank slate" but a LAD, the child has to learn the meaning of every word in her language: no one has argued that words are innate! Furthermore, the child has to use the evidence of conversation to make a multitude of choices about the rules of her language (and their irregularities). The doctrine of innate ideas may have a relatively limited role to play except in defining the boundaries within which learning can take place.
