The effects of prosody and production planning in external t-sandhi

The change in pronunciation of word-final /t/ in English is known as external coronal stop sandhi and is understood to be motivated by several phonological conditions. Though much of the research in this area (e.g., Tanner et al., 2017, Kilbourne-Ceron et al., 2016, Coetzee and Kawahara, 2013) has generally focused on a single t-sandhi alternation within a corpus (e.g., from /t/ to /ɾ/), the results of these studies make it clear that it is always a combination of factors that play a role in each alternation. In this paper, I propose that examining the distribution of various types of t-sandhi clause-internally would provide a more complete look of where speakers preferred to use different forms of /t/ in their natural speech: tightly controlling an area of analysis would make this otherwise wide scope more manageable. This project uses two production experiments to examine the interaction between prosody and other conditioning factors on t-sandhi: established segmental contexts and findings from Production Planning studies were used to control the environment surrounding word-final /t/ and record the distribution of its pronunciation. The results of these studies show that the more marked forms of t-sandhi (in terms of production) favor prosodically defined environments not explored in previous studies: word external, clause-internal flapping exclusively preferred environments where the following word did not carry lexical stress on the first syllable; released-t’s were never produced in environments 2 where the /t/ was preceded by a vowel and followed by a consonant, and also displayed changes in production rates that were sensitive to prosodic correlates. Production planning variables also showed some expected results which show areas of interest for further t-sandhi and production experiments focused on the production planning hypothesis (Tanner et al. 2017, Kilbourne-Ceron et al. 2016).