The production of prosodic cues and their role in the comprehension of syntactically ambiguous sentences

This thesis explores two questions about the role of prosody in language. From the perspective of speech production, it examines the regularity with which speakers provide their listeners with prosody-based cues to sentence structure. From the perspective of speech comprehension, it evaluates the psychological validity of prosodic cues in syntactic disambiguation and the time-course of the influence of these acoustic factors on the interpretation of syntactically ambiguous sentences. An Interactive Production Monitor model is introduced, in which the speech production system incrementally evaluates the likelihood that a to-be-uttered message will be correctly understood. This system monitors the availability of various sources of disambiguating information. The Contingent (Prosodic) Cueing Hypothesis claims that when alternative sources of disambiguating information (such as contextual) are present, prosodic cues are less reliable. When no alternative cues are present, speakers produce carefully controlled, prosodically informative speech. Thus, the production of prosodic cues to syntactic structure is contingent on the availability of alternative sources of disambiguating information.; Findings from an elicited-speech experiment examining the production of prepositional phrase-attachment ambiguities, such as (1), support the Contingent (Prosodic) Cueing Hypothesis.{dollar}{dollar}(1) rm The chauffeur annoyed the man with the cigar.{dollar}{dollar}The traditionally expected pitch and pause duration correlates of syntactic structure occur reliably only when prosody is the sole available disambiguating cue (i.e., in the absence of supporting context). Subsequent perception studies, using the same sentence fragments, demonstrate that naive listeners exploit the acoustic variation in speech to understand the intended message of an indeterminate utterance. However, consonant with the production study, reliable perceptual effects obtain only for items originally produced without alternative sources of disambiguating information available to the listener. These findings verify that prosody imposes psycholinguistically salient constraints on sentence comprehension. These studies also provide perceptually-based evidence that exploitable acoustic cues to syntactic structure do not occur in all speech production environments.; These results support the Interactive Production Monitor model, which states that the speech planning mechanism allocates the cognitive resources needed to produce highly constrained, acoustically informative speech only when acoustic cues are deemed necessary to guarantee that the listener will recognize the message.