Listeners compensate for prosodic position when categorizing word-final obstruents

Abstract

This study explores how perceptual sensitivities to contextual variability extend to prosodically induced variation. In English, vowel duration is reliably longer preceding a voiced obstruent, as opposed to a voiceless obstruent, and listeners use preceding length as a cue to obstruent voicing (e.g. Raphael, 1972). Segmental duration also co-varies systematically with prosodic position, being longer phrase-finally (e.g. Turk & Shattuck-Hufnagel, 2007). In this study, we tested the extent to which listeners’ categorization of word-final obstruents is altered by the prosodic position of the target. Participants heard a continuum that varied only in vowel length, and categorized stimuli as either “coat” or “code”. Prosodic position in a carrier phrase was manipulated by splicing the target word into either a phrase-final or phrase-medial context. Results show listener expectations about phrase final lengthening mediate categorization, with significantly longer vowel durations required in phrase-final position for a “code” response. The results are discussed in terms of other top-down influences on segmental categorization (e.g. lexical information), and implications for further work on the prosody-phonetics interface.