Signatures of Domain-General Categorization Mechanisms in Color Word Learning

Learning color words is a difficult problem for young children. Because color is abstract, this difficulty has been attributed to challenges in integrating over heterogeneous objects to discover color as a dimension of reference. On this account, learning that color words refer to the color dimension is slow, but subsequently mapping these words to particular shades is fast. Recent work suggests an alternative: Children may rapidly identify color as a referential dimension, but only gradually discover the precise boundaries of each color word. This alternative proposal predicts that the learning mechanisms underlying the acquisition of color words should parallel those underlying the acquisition of concrete object categories. We test this prediction, finding that children’s performance in a color naming task is modulated by three factors that have previously been studied in category learning: input frequency, category size, and perceptual salience. Because it allows for precise psychophysical measurement of category properties, color presents a unique case study for investigating language acquisition and categorization more broadly.