Independence of temporal contrast effects across different timescales in speech perception

Abstract

Perception of speech sounds is shaped by the context of the earlier sounds, including speaking rate. Often, acoustic differences between earlier and later sounds are perceptually magnified. In temporal contrast effects (TCEs), a fast-rate precursor sentence can cause the following target word to sound longer (e.g., longer-VOT “tier”), and a slow-rate precursor sentence can cause the target word to sound shorter (e.g., shorter-VOT “deer”). The novel contribution of this study is the exploration of TCEs across different context durations to examine their consistency on a granular level. On each trial, listeners heard one of three contexts (“a”, “the word”, “this time I want you to click on the word”) spoken at a fast or slow rate, then a target word to be identified as “deer” or “tier”. TCEs were observed at all context durations but were surprisingly not consistent in magnitude. While speaking rate forms an important context for speech sound recognition, the degree of its influence may not be reliable across different timescales. This reveals important bounds on how context effects relate to one another. [Work supported by NIDCD.]