The Implications of Polysemy for Theories of Word Learnin

Most common words in English and in other languages are polysemous, expressing a family of distinct but related meanings (e.g., “chicken” can refer to a kind of animal, meat, game, or cowardly person). Yet within developmental science, word learning is typically studied as a problem in which children need to learn one meaning for each word. I will argue that simplifying the object of study in this way has led researchers to formulate theories which incorrectly predict that children should struggle at learning polysemous words. Instead, I will present research showing not only that children can learn multiple meanings for polysemous words from early in language acquisition, but can also use their knowledge of one meaning of a word to make inferences about—and even anticipate—its other, related meanings. Thus, rather than impeding learning, I will suggest that polysemy facilitates lexical development. Indeed, I will review my research on historical semantic change and cross-linguistic typology which suggests that polysemy may evolve in language in response to the pressure to make language maximally learnable. Taken together, this body of work disconfirms the predictions of both classic and contemporary theories of word learning, motivating a re-thinking of how word learning proceeds.