In spoken discourse, speakers strategically deploy prosodic cues to shape listeners' perceptions and expectations of sentence completion in structually ambiguous sentences. By asking 64 listeners to judge the degree of perceived sentence completion or to anticipate the ending point of the sentence in two experiments, this perceptual-acoustics study explored three aspects: (1) Acoustic distinctions at various levels of prosodic hierarchy between different ambiguity types; (2) The influence of prosodic cues at different levels on the perception of sentence completion and the classification of ambiguity types; (3) Which level of prosodic cues was employed by the listener to perceive and anticipate sentence completion. The findings revealed: (1) Structural ambiguity types impacted f0, intensity, and duration of boundary nouns; (2) The accuracy to classify structural types was higher in models with multi-level cues than those with single-level cues, and the highest accuracy was achieved with local cues of boundary nouns; (3) Local prosodic cues increased accuracy of perception and reduced RT for anticipation, while distal cues show no significant impact. These outcomes suggest that listeners rely on local parameters to determine spoken sentence completion, whereas speakers in oral communication are influenced by a "global" encoding mechanism.