Effects of semantic predictability and dialect variation on vowel production in clear and plain lab speech

Abstract

Speech addressed to a non-native or hearing impaired listener features longer, more peripheral vowels. In addition, more extreme dialect-specific forms are produced in semantically predictable contexts, and less extreme forms (more standard forms) in unpredictable contexts. This study investigated the interactions between predictability and speaking style on Southern American English monophthongization of the vowel /aj/. The Midland dialect of American English served as the comparison. Participants read a set of sentences with monosyllabic target words in sentence-final position. Target words varied in semantic predictability based on the preceding sentential context. Each set of sentences was produced twice by each participant—first as if talking to a friend (“plain” speech) and again as if talking to a non-native or hearing impaired listener (“clear” speech). The duration, dispersion, and trajectory length of the vowel in each target word were measured. Preliminary results suggest that, as expected, Southern /aj/ has a shorter trajectory length than Midland /aj/, and in both dialects, /aj/ has a shorter trajectory length in clear speech than plain speech. However, these processes do not interact with each other or with semantic predictability, suggesting that style and predictability effects are independent of the realization of some dialect variants.