Characteristics of stop releases in American English spontaneous speech

This study examines the factors affecting the production of stop releases in American English spontaneous speech. Previous research has shown that releases are conditioned by phonetic and social factors. However, previous studies either rely exclusively on read speech, or for sociolinguistic studies, focus on phrase-final stops. In this study, spontaneous speech is collected from two sources: interviews from the non-profit StoryCorps project and from sentences spontaneously generated in a picture description task. Stop releases were examined before obstruents and nasals in word-medial position (e.g. rugby), word-final, phrase-medial position (e.g. They crack nuts), and pre-pausally (e.g. I look up). Phonetic factors taken into account include identity of the stop, directionality of place of articulation in the consonant cluster (front-to-back vs. back-to-front) and manner of C2. For the StoryCorps data, race of the speaker was also found to be an important predictor. Results showed that approximately a quarter of the stops followed by a consonant were released, but release was strongly affected by the place of the stop and the manner of the following consonant. Release of pre-pausal stops differed between black and white speakers; the latter had double the amount of final release. Other realizations of the stops, such as deletion, lenition, and glottalization are also analyzed.

► Stop releases in pre-obstruent and pre-pausal positions in American English spontaneous speech are investigated.
► Approximately a quarter of the stops followed by a consonant are released.
► However, amount of release is heavily influenced by the place of the stop and the manner of the following consonant.
► Pre-pausal release depends on the speakers’ race; white speakers have twice the amount of release as black speakers.
► Other possible realizations of the stops, such as deletion, lenition, and glottalization are also analyzed.