Funded Grants

Learning to Learn: How Polysemy Scaffolds Development

At a time when they still struggle to tie their shoes, most preschoolers already know thousands of words. Children’s rapid vocabulary development is striking given the learning challenges they face. Discerning a new word’s meaning is a daunting task, since words are typically uttered in complex scenes in which they could have many possible interpretations. Compounding this problem, children often have to learn multiple meanings for each word, because most words in language are polysemous: they express a family of related meanings. For example, the English word chicken labels an animal, a kind of meat, a game, and a cowardly person.

Is it difficult for children to learn a new word when it carries multiple meanings? According to almost all current theories, it should be. This is because polysemous words violate a core assumption that is thought to guide learning, namely, that a single word should map onto a single category of meaning. In contrast, my research program advances the opposite view. I propose that polysemy is widespread in language because it facilitates learning. Considering how children learn polysemous words transforms our understanding of the basic mechanisms that drive learning, and provides insight into a major question in developmental science: how children learn so much within such a brief period of time.

My research proposes that children not only have the capacity to learn multiple meanings for a word, but can also leverage their knowledge of one word meaning to make inferences about— and even anticipate—its other meanings. Thus, rather than impeding learning, polysemy allows children to learn-to-learn, building on their initial knowledge of a word to guess its other meanings. From this perspective, polysemy arises in the lexicon in response to the pressure to make language maximally learnable. Polysemy thus provides an example of how language—as a culturally-transmitted system—has been adapted to fit how even the youngest minds make sense of the world.

I test these hypotheses through several lines of work, using experimental methods with children and computational analyses of large-scale linguistic corpora. First, I explore whether children have difficulty learning polysemous words, as predicted by current theories. Second, I examine ways in which polysemy might instead facilitate word learning. Finally, I explore the origins of polysemy, by tracing how polysemy has developed over history and across languages.

Because most words are polysemous, children will often be learning a new meaning of a word for which they have already learned a prior meaning. The acquisition of polysemous words thus provides an understudied example of how development builds on itself, as prior learning supports and constrains future learning. With support from the James S. McDonnell Foundation, I will extend my research to elucidate these developmental processes. Specifically, I will explore (1) how children make sense of polysemy early in life and develop knowledge of polysemy from their language environments, and (2) whether learning polysemous words might scaffold not only language development but also cognitive development, by prompting children to learn more about the world around them.