Session Prosody IV:
Prosody IV: Stress and Rhythm
Phonetic Cues Identifying English Compounds
Tuuli Adams, New York University
|This study investigates the acoustic correlates of stress in English compounds by measuring the interaction of stress cues with different intonational environments. Effects on vowel duration, intensity, and pitch changes are compared in contrasting compounds and phrases. The results of an experiment in which participants pronounced compounds and phrases in controlled prosodic and intonational environments provide new evidence that the phonetic indicators of stress interact with these environments in a systematic way.
What is Compound Stress?
Gero Kunter, Universität Siegen
Ingo Plag, Universität Siegen
|This paper investigates the implementation of stress in English noun-noun compounds. First, a perception experiment examines how listeners perceive prominence in compounds. After that, significant acoustic correlates of prominence are established. Finally, a cluster analysis is described that classifies compounds on the basis of their phonetic features and which is capable of separating different stress categories. The results demonstrate how gradient acoustic measurements and discrete phonological contrasts can be mapped onto each other.
Rhythm metrics predict rhythmic discrimination
Laurence White, University of Reading
Sven L. Mattys, University of Bristol
Lucy Series, University of Bristol
Suzi Gage, University of Bristol
|Metrics such as VarcoV and %V provide empirical support for long-held notions about rhythmic distinctions between languages. Furthermore, listeners can discriminate languages with distinct rhythm metric scores purely on the basis of the durational information available in resynthesized monotone sasasa speech. However, some factors contributing to this durational variation, such as stress distribution and prosodic timing, are not directly reflected in rhythm scores. To test more precisely the predictive power of rhythm metrics, we used tightly controlled sasasa stimuli, eliminating stress distribution and prosodic timing cues to focus on the information directly quantified by rhythm metrics. We show that VarcoV and %V scores are predictive of listeners’ discrimination within and between languages, even with these highly constrained stimuli.
Back to Conference Schedule