An EPG and ultrasound study of lingual coarticulation in vowel-consonant sequences

Abstract

While EPG registers the location and amount of tongue-palate contact, ultrasound can capture most of the tongue contour. Previous studies have not systematically quantified lingual coarticulation using EPG and ultrasound simultaneously. This study used both techniques for analysing vowel-consonant coarticulatory effects. Four speakers of Scottish English produced /VC/ sequences with the consonants /p, f, t, s, l, r, k/ and the vowels /a, i/. The difference between each consonant in the two vowel contexts was computed using an EPG measure and an ultrasound measure. Additionally, temporal coarticulation was analysed, using EPG data. A significant positive correlation was observed between the two measures, with labial consonants, followed by /r/, having the highest values. The two techniques also provided complementary data on lingual coarticulation. The velar stop was more coarticulated on the EPG measure than on the ultrasound measure, because EPG registered a shift in closure location across vowel contexts, while ultrasound captured the close proximity of the tongue root across the vowel contexts. The sibilant was more coarticulated on the ultrasound measure than on the EPG measure, because ultrasound, unlike EPG, registered vowel-dependent difference in the tongue root. Combined EPG and ultrasound data would be useful in future studies of coarticulation.