The release and aspiration of word-final /t/ and /d/ are important sociolinguistic variables in American English because they have strong, contextually driven indexicality. Word-final /t d/ releases are usually coded impressionistically due to the absence of automated methods for identifying prepausal release bursts or aspiration. This paper introduces an automated method for identifying released tokens prepausally and for measuring phonetic properties of releases. We use the method to code prepausal /t d/ release versus non-release in a corpus of conversational English in Raleigh. We assess the data in relation to internal and social factors in order to validate the automated method, finding that the patterns in the automatically generated distributions match those in previous studies. We next show that among Raleigh White speakers but not Black speakers, /t d/ releases are becoming more frequent and stronger after obstruents across apparent time, a change that reflects Raleigh’s changing cultural landscape.
