In contrast to Hans Kurath and Raven McDavid's traditional analysis of vowel variation presented in their 1961 Pronunciation of English in the Atlantic States, which was a largely intuitive, qualitative treatment of the evidence that focused on patterns of regional contrast, this study focuses on the picture of quantitative variation in American English vowels along the North Midland and South Midland border as represented in data from the Linguistic Atlas of Middle and South Atlantic States. For each of the 12 vowels studied, its various pronunciations within a single word (e.g., /i/ in three, /I/ in six, etc.) were tallied for the overall sample of 1,162 speakers and for nine subsamples, representative of three domains from the altas data: sex, education/social adjustment (three types), and occupation (four types). The results show that for each vowel, at each level of analysis, the pronunciation type/frequency rankings form a nonlinear distribution, represented graphically by an asymptotic hyperbolic curve (or A-curve), with only a few of the most frequent vowel realizations accounting for the vast majority of tokens. While there is a large amount of pronunciation overlap between regional or social groups, comparison of the type/frequency rankings shows that different groups of speakers each have different orders of the frequent variant realizations of the vowels, which helps explain how one can distinguish one group's speech from another's but does not support the traditional notion of contrast as absolute, binary, and categorical.

The text of this article is only available as a PDF.
You do not currently have access to this content.