Patterns of vowel-to-vowel coarticulation in Greek and English

Abstract

Many previous studies have indicated that coarticulatory behavior differs systematically across languages. One specific hypothesis, made by Manuel and Krakow (1984), is that languages with larger vowel inventories show reduced vowel-to-vowel coarticulation because their crowded vowel spaces impose more stringent requirements for maintaining phonological distinctiveness. The present study is designed (a) to test this prediction in Greek versus English, which have small and large vowel inventories, respectively; and (b) to investigate which parameters of the vowel system are most responsible for shaping language-specific coarticulatory constraints. Adult monolingual speakers of each language were recorded, producing VCV utterances embedded in phonologically similar carrier phrases. Vowels included all five vowels of Greek, /i,e,a,o,u/, and their closest counterparts in American English, and the medial consonant varied between labial and alveolar places of articulation. To provide an independent measure of production variability for each vowel, speakers also produced the vowels in real words of the form /pVt/ or /bVt/, embedded in the same carrier phrases. Preliminary data are presented for two speakers of each language, and discussed in terms of vowel type, distribution within the vowel space, and the size of vowel areas within F1−F2 space.