V-to-V coarticulation is not consistent in the wild

This paper examines the phenomenon of anticipatory VCV coarticulation in the Buckeye Corpus of conversational English. Most results establishing these effects have been obtained from curated laboratory speech. The present study attempted to replicate these findings “in the wild” using a large corpus of natural speech. The procedure measured as many tokens of ‘schwa’ vowels as possible, and compared them when they preceded different vowels [i] vs. [ɑ] at one syllable removed. The occurrence of V-to-V coarticulation was found to be highly variable across speakers, with only 12 of the 37 speakers analyzed exhibiting a statistically significant effect.