How Children Learn the Meaning of Words and How LSA Does It (Too)

For long psycholinguistics has tried to answer the question “how children learn the meaning of words?” Paul Bloom answers the question in his book with the same title. He argues that the mind does not have a module for language acquisition. Instead, children learn the meaning of words by a set of general cognitive abilities, including the ability to infer intentions, to perceive the world in terms of objects and events, and the ability to understand syntactic structures. As discourse psychologists we have to emphasize that children likely do not learn the meaning of words through learning words. Instead, it seems more likely that the meaning of words is acquired through discourse. Discourse can take the form of conversations by oneself, with others or by others, situated in time and space. In fact, one could argue that children learn—at least in part—words solely by understanding their relation in context. This is how some computational techniques approximate the meaning of words, sentences, paragraphs and texts. One such common technique is Latent Semantic Analysis (LSA). LSA is a statistical, corpus-based, technique for representing world knowledge. It computes similarity comparisons for terms and documents by taking advantage of the fact that particular words occur in particular documents. LSA takes quantitative information about co-occurrences of words in documents (paragraphs and sentences) and translates this into an multidimensional space. In short, the input of LSA is a large coTHE JOURNAL OF THE LEARNING SCIENCES, 14(2), 301–309 Copyright © 2005, Lawrence Erlbaum Associates, Inc.