Individual differences in the production of prosodic boundaries in American English

This study investigates individual differences in the weighting of phonetic properties in the production of prosodic boundaries in American English. The motivation of the study is to inform understanding of individual speaker variation and its accommodation in the representation of prosodic structure. In an acoustic study, 32 speakers produced eight sentence pairs differing in type of boundary (Intonational Phrase (IP) boundary versus Word boundary). Pause duration, phrase-final lengthening (three syllables before the boundary), phrase-initial lengthening (one syllable after the boundary), and pitch reset were examined. The results of a series of statistical analyses showed substantial individual differences in (1) which segmental and suprasegmental properties speakers phonetically modulated to produce IP boundaries and (2) the scope and the degree of such modulations. In addition, the results suggest that there seems to be no apparent relationship between how speakers modulated boundary-adjacent syllable durations and whether and how they used two other acoustic correlates for IP boundary, namely, pause duration and pitch reset. The current models of prosodic structure need to accommodate the fact that individuals may vary significantly while systematically modulating the acoustic correlates relevant for encoding a prosodic contrast.