| Abstract: | Though language manifests linearly, one word at a time, children must learn that words are embedded in constituents, which are in turn embedded in larger constituents. That is, they must learn that syntax is hierarchically structured. Prosody—speech melody and rhythm—is likewise hierarchically organized, with smaller prosodic constituents embedded in larger ones. This paper presents results showing that 20‐month‐olds (N = 40) can use modifier + clause prosody to learn constituents at multiple levels in a hierarchically structured artificial grammar, but that they fail to learn from clause + modifier prosody. We consider several explanations for this asymmetry, and suggest that it may be due to the frequency of the two constructions in the input. Two parallel experiments with adults suggest that older learners may not be able to learn the prosodically marked constituents. Instead, adult responses are likely based on perceptually prominent boundary words. |
