| |
Learnability theory, in linguistics, is relevant to explanations of how children acquire their first language. By means of mathematical modelling, learnability theory takes into account the conditions faced by the language-learning child and examines different classes of languages (both natural human languages and other possibilities), in order to determine which ones are logically possible to learn. Typically, a mathematical model, known as a learning procedure, is set up which substitutes for the language learning capacity of the child. This learning procedure is presented with sentences from a language, one after another, and is faced with the task of identifying the relevant language, in addition to formulating an appropriate grammar. If the procedure is unsuccessful, further guesses are made after the presentation of each new string until success is achieved. In essence, then, the learning procedure is mimicking the eventual achievement of the language-learning child. In fact, the level of mimicry is arguably quite low, since learning procedures often operate on the basis of highly idealized learning conditions, which bear little relation to those actually experienced by the child. For example, the procedure typically has an infinite amount of time available, it is never exposed to ungrammatical sentences, it never makes any errors itself and after the presentation of each new sentence, an entire grammatical system is projected.
Sometimes, the strictures of the learning situation, in conjunction with the characteristics of the particular class of language involved, will result in inevitable failure, since no amount of guessing will allow the language to be identified. Thus, certain types of language are said to be logically impossible to learn. Given the kind of exposure to language experienced by children (the input conditions), learnability models tend to reach the paradoxical conclusion that even natural human languages are unlearnable. The fact that all normal children do in fact learn a language is then normally explained by appealing to the concept of innate linguistic knowledge. It is reasoned that the child must be able to compensate for the paucity of the linguistic input by relying on a genetically determined knowledge of crucial aspects of language. Being innate, the need to learn these features of language, in the sense of deducing what they are by analysing appropriate sentences, is obviated. MS
See also innateness; universal grammar. |
|
