Informative prosodic boundaries.

In principle, a prosodic boundary in an utterance might affect its interpretation in a local, context-independent fashion. In a right-branching language like English, the presence of a large prosodic boundary might signal the end of the current constituent, requiring the following constituent to be attached high in the syntactic tree. We present three listening experiments that test an alternative position suggested in Carlson, Clifton, and Frazier (2001) as the "informative boundary" hypothesis. This hypothesis claims that the interpretation of a prosodic boundary is determined not by its absolute size but by its size relative to relevant certain other boundaries. Experiment 1 confirmed the predictions of this hypothesis in phrases like the old men and women with very large houses, manipulating the boundaries before and and with. Experiment 2 investigated the effect in a variety of diverse syntactic structures, varying syntactic category and status (head vs. nonhead) of the ambiguous constituent. It confirmed the predictions of the informative boundary hypothesis in every structure tested except for '-ly' adverbs that are ambiguous between a manner interpretation and a speaker-evaluation interpretation. Experiment 3 demonstrated that sentence interpretation was affected by the size of the late boundary relative to a relevant early boundary, but not relative to an early boundary that was predicted to be irrelevant.