Word-internal /t,d/ deletion in spontaneous speech: Modeling the effects of extra-linguistic, lexical, and phonological factors

TLDR

The deletion of word‑internal alveolar stops in spontaneous English speech is a variation phenomenon that has not previously been investigated. This study quantifies internal deletion statistically using a range of linguistic and extra‑linguistic variables. It interprets the results within a model of speech production and compares them to prior work on word‑final deletion, laboratory internal deletion, and flapping. Effects were found for speech rate and fluency, word form and predictability, prominence, and local phonological context; deletion in codas arises from cluster simplification during segment planning, while deletion in onsets results from gradient lenition during articulation, indicating two distinct processes. The research was supported by NIDCD grant DC004330, and the authors thank collaborators and participants for their contributions.

Abstract

The deletion of word-internal alveolar stops in spontaneous English speech is a variation phenomenon that has not previously been investigated. This study quantifies internal deletion statistically using a range of linguistic and extra-linguistic variables, and interprets the results within a model of speech production. Effects were found for speech rate and fluency, word form and word predictability, prominence, and aspects of the local phonological context. Results of the study are compared to results from the numerous studies of word-final alveolar stop deletion, internal deletion in laboratory speech, and also to another internal alveolar stop process, flapping. Our findings suggest that word-internal alveolar stop deletion is not a unitary phenomenon, but two different processes that arise at different points during speech production. In syllable codas, deletion results from cluster simplification to achieve gestural economy and is introduced during segment planning. In syllable onsets, deletion is one outcome of gradient lenition that results from gestural reduction during articulation.The research reported here was supported by NIDCD grant DC004330 to Mark Pitt, Keith Johnson, and Elizabeth Hume. Our thanks to Mark Pitt, Keith Johnson, William Labov, David Sankoff, two anonymous reviewers, and numerous participants of NWAV02 for useful comments and constructive feedback that helped us develop the article.