2008/04/09

Carlson, R., Hirschberg, J., & Swerts, M. (2005). Cues to upcoming Swedish prosodic boundaries: Subjective judgment studies and acoustic correlates. Speech Communication, 46, 326-333.

Presentation: Shelly

Summary: Sarah

Past studies on the perception of prosodic boundary showed that listeners can readily tell the presence or absence of the prosodic boundary, as well as the degrees of boundary strength. Among the salient cues associated with prosodic boundaries, it was found that the presence of silence played a key role. As a result, in this study, the authors aimed to explore whether listeners, both native and nonnative, were able to predict the occurrence and strength of upcoming boundaries, when the pausal information is absent. And if they did, which potential cues might account for their predictions? To answer these questions, a perception experiment was performed. Stimuli were utterances extracted from corpus data. Subjects were Swedish speakers, constituting the native group, and English speakers, the nonnative group. They were asked to judge the strength of each boundary they heard. Results showed that the judgments of both native and nonnative listeners corresponded well with the actual labeled boundary levels in the corpus. In addition, some acoustic cues were found in high correlation with the perceived boundary strength, including creak on the final syllable and lower fundamental frequency. Final lengthening, however, was not shown to be significantly correlated. The authors thus concluded that, listeners primarily based on the acoustic cues to perceive boundaries, without reliance on syntactic information. Also, it seemed that it was the preceding word or phrase that influenced the judgments on prosodic boundaries, since longer preceding context did not lead to greater accuracy.