How Children Learn Novel Word Extensions Using Active Learning

Abstract

Consider a child who hears “dog” for the first time in reference to a dalmatian. How do they come to understand that “dog” refers to any dog but not any animal or just dalmatians? This example highlights a challenge all language learners face: what is the range of objects that a novel word can refer to (i.e., its extension)? Despite this challenge, children can quickly learn to properly extend novel words. We suggest that active learning plays an important role in this process, allowing children the opportunity to test their hypotheses about word extensions and clarify ambiguous cases. In this present study, we tasked children (5 to 8 years of age) and adults with learning novel words and provided them a chance to actively learn through sampling, where participants could select an object and learn if the novel word could be extended to it. We assessed participants’ sampling behaviors and found that children, similar to adults, favored conservative sampling choices that confirmed a novel word’s known extension, as opposed to exploring broader possible extensions. We also found that, with age, children more frequently explored word extension ambiguity by sampling objects that could potentially provide them with new information about the breadth or specificity of a novel word’s extension. However, overall, children were still less exploratory than prior studies might suggest. Taken together, these findings advance our understanding of how children learn word extensions by introducing active learning as a facilitator of this learning process.