Grammaticalized syntactically disambiguating prosody.

Abstract

Suprasegmental temporal patterns can be used to reliably disambiguate syntactically ambiguous sentences in English [e.g., Price et al. (1991)], but the extent to which speakers reliably produce these patterns appears to vary with training [Albritton et al. (1996)]. This study investigated whether the production and perception of disambiguating temporal patterns also vary with the type of structure being disambiguated. Five speakers produced five different types of ambiguous sentences with disambiguating prosody after having been explicitly instructed in the different meanings that the sentences could convey. Despite equivalent instruction, acoustic analyzes revealed that only some types of sentences were consistently disambiguated by temporal patterns. Same‐different judgments by 14 listeners revealed that speakers employed additional prosodic cues to distinguish meaning; however, a question‐answer task indicated that the accuracy with which listeners were able to pair a sentence rendition with its intended meaning varied extensively with type. Consistency in production mostly overlapped with accuracy in perception, but in one case a consistent prosodic distinction failed to yield accurate meaning judgments. Overall, the results suggest that disambiguating prosody is only grammaticalized for certain types of syntactically ambiguous sentences. [Work supported by NIH Grant 1R01HD061458.]