Velar–vowel coarticulation in a virtual target model of stop production

Abstract

Velar–vowel coarticulation in English, resulting in so-called velar fronting in front vowel contexts, was studied using ultrasound imaging of the tongue during /k/ onsets of monosyllabic words with no coda or a labial coda. Ten native English speakers were recorded and analyzed. A variety of coarticulation patterns that often appear to contain small differences in typical closure location for similar vowels was found. An account of the coarticulation pattern is provided using a virtual target model of stop consonant production where there are two /k/ allophones in English, one for front vowels and one for non-front vowels. Small differences in closure location along the palate between productions within each context are the result of the trajectory of movement of the tongue from the vowel to vowel through the virtual target beyond the limit of the palate. The overall pattern is thus seen as a combination of a large planned allophonic difference between consonant closure targets and smaller phonetic differences for each particular vowel quality that are the result of coarticulation.