Assessing vowel space area metrics in the context of dialect variation

Abstract

There has been a growing interest in the development of a sensitive methodology to define a working vowel space area (VSA) as a metric in basic and clinical research applications. In this study, three approaches to the assessment of VSA were tested to evaluate their efficacy in characterizing cross-dialectal and cross-generational variation: The traditional vowel quadrilateral, the traditional convex hull, and a more liberal convex hull. In the two traditional approaches, VSA was computed as a planar convex polygon shaped by phonologically distinct vowel categories as its corners. The mean F1/F2 values at vowel midpoint were used to define the respective areas. The liberal convex hull utilized vowel dynamics and variable F1/F2 temporal locations to refine the outer boundaries and maximize the VSA. This approach used an unrestricted number of vowels and measurement locations to define the optimal VSA. All computations were based on a common speech material produced by 135 female speakers representing three American English dialects and four generations ranging from 8 to 91-year olds. The three metrics yielded inconsistent and contradictory estimates of VSA. Discussion will focus on the limited utility of polygon geometry in characterizing a working VSA in American English dialects.