Friday, November 11, 2016
Statistical language learning: Computational, maturational, and linguistic constraints
Newport, E. L. (2016). Statistical language learning: Computational, maturational, and linguistic constraints. Language and Cognition, 8, (447-461).
Is the ability to use language innate, or is it learned? Do children and adults differ in how they learn language? Why do certain patterns consistently occur in many languages across the world? For example, number words (‘two’) usually come before the noun they’re counting as in “two cars”, in English, and “deux voitures”, in French. To answer these questions, we can look to the ways in which humans learn. Our system of language is essentially a series of patterns; from what makes a word, to how sentences are formed. Humans are quite good at learning these patterns, and the ability to learn these patterns in language is known as statistical language learning. In this paper, Newport described research over the past 20 years showing that statistical language learning can explain many of the patterns we observe when humans start to learn language, even across different languages of the world.
Language learners show an extraordinary ability to learn language through statistical language learning. However, when the input is variable or inconsistent, language learning across children and adults look quite different. Researchers have examined this phenomenon in the lab by comparing how children and adults learn artificial languages. These artificial languages contain a consistent grammatical rule that applies to most cases, but this rule is violated some of the time. Adults learn the language exactly as they heard it, learning both the rules and the violations. Young children, on the other hand, learn the rule but not the exceptions, and generalize that rule across the language. This difference in how adults and children learn a novel language is important for understanding language acquisition.
The finding of children learning and generalizing the consistent patterns in a language has been shown in a number of experimental studies and real-world settings. But why does children’s language learning look different from adults? One explanation is that children are biased towards learning the consistent patterns because of how they learn. Children are able to process less information than adults; because of this, it is easier for children to learn and use a consistent pattern that they can apply to many cases than it is to learn inconsistent patterns, or many exceptions to a rule. This tendency to adopt the easy-to-learn patterns may explain why certain patterns exist in many languages across the world. Examining how young children learn language tells us a lot about language acquisition, and why certain patterns exist across many languages of the world.
Blogger: Nicolette Noonan; Nicolette is a Psychology Ph.D. student, supervised by Drs. Lisa Archibald and Marc Joanisse