Prosodic Structure and Sentence Recognition

In three experiments, subjects heard alternate versions of naturally spoken sentences that allowed two prosodic realizations (They are FRYING chickens vs They are frying CHICKENS). In Experiment 1, a paraphrase choice task showed that interpretations reliably depended on the prosody used. In Experiment 2, presented sentences were tested for recognition against unfamiliar sentences. Recognition ratings were higher if the sentences were spoken with the same prosody at presentation and test than if two different prosodic versions were presented on these two occasions. This matching effect was not larger for sentences whose prosodies biased listeners toward different syntactic analyses than for sentences whose prosodies effected changes in focus. Experiment 3 showed this integration of prosody and text for nonsense sentences.