Lisa Davidson, New York University

An investigation of hiatus (non-)resolution in American English

Most research on hiatus resolution in American English has focused on the phonological oddity of /r/-insertion in certain environments, while assuming that hiatus in all other environments is resolved through glide insertion (e.g., McCarthy 1993, Uffmann 2007, Krämer 2008). However, despite suspicions from a small number of authors that "linking glides" may be fundamentally different from lexical glides in English (e.g., Cruttenden 2001, Heselwood 2006), a systematic investigation of ostensible minimal pairs (e.g. "my ears" vs. "my years") has never been carried out. In this talk, we examine the nature of hiatus resolution by comparing three environments: (1) vowel-vowel sequences within words (VV: "kiosk") (2) vowel-vowel across word boundaries (VBV: "see otters"), and (3) vowel-glide-vowel sequences across word boundaries (VGV: "see yachts"). Our analysis is based on reading passages containing matched sets of VV, VBV, and VGV words and phrases, which were read by 14 speakers. The results show two main findings. First, the production of a glottal stop before the second vowel accounts for nearly half of the responses for VBV phrases (45% of responses), whereas glottal stops are present in less than 5% of either VV or VGV phrases. Second, an acoustic comparison of VV, VBV, and VGV phrases that were not produced with glottal stops show starkly significant differences between the vowel-glide-vowel and the vowel-vowel  sequences on all measures, including analyses of duration, intensity, and formants. These results indicate that American English speakers tend to resolve hiatus at word boundaries with glottal stop insertion, whereas there is no hiatus resolution at all within words. The talk will end with a brief phonological analysis  intended to capture the asymmetry between hiatus at word boundaries versus within words.