Rapid learning of word meanings from distributional and morpho-syntactic cues

AbstractWhat does it take to learn a new word? Many of the words we learn, we have learned from language itself -- by encountering them in various informative contexts. Here, we investigate the limits of learning from context by studying how people learn new words from very sparse contexts, at the extreme, a context in which all content words are replaced by nonsense words. We find that participants exposed to even such extremely _sparse_ contexts nevertheless learn something about the meaning of words embedded in those contexts. Performance tended to be better when knowledge was assessed by first directing people’s attention to the part of speech of the target words.

Return to previous page