When a novel word refers to something in the world, how do learners decide whether that word have a more specific meaning (e.g., dog) or more general meaning (e.g., animal)? Here we focus on the role of semantic contrast between referential alternatives. We do this in the context of learning novel words cross-situationally, asking when learners adopt more specific meanings (resulting in homophonic words: e.g., ‘fami' means both dog and butterfly) or adopt a single superordinate meaning (e.g., ‘fami' means animal). We hypothesize that learners will be more likely to establish homophonous meanings when contrasting referents are from a neighboring category of the target, and more likely to establish a superordinate meaning when contrasting referents are from more distant categories. We also expect homophone learning to be more difficult because of its additional demands on learning and memory. Our predictions were borne out in a series of experiments and modeling.

A central challenge of cross-situational word learning is retaining word-referent mappings across exposures. We evaluate Memory-Bound Pursuit (MBP), a hypothesis-testing model of cross-situational word-learning which aims to account for learners’ memory constraints via a single parameter targeting the number of words that can be learned concurrently. Here, we show that by varying this parameter with age, MBP can capture both children’s and adults’ cross-situational word-learning success under varying levels of ambiguity. We also present new experimental findings supporting novel predictions made by MBP about the retention of word-referent mappings across intervening exposures. These findings suggest that MBP provides a strong baseline model of cross-situational word learning, capturing both developmental trends and experimental evidence of memory limitations for word learning.