Improved representation of variance in measures of vowel merger

Abstract

The difference between two vowels has most commonly been measured using F1, F2, or Euclidean distances (between averages or between minimal pairs). But relying on averaged data or minimal pair data only captures some of the variability; average obscure the amount of variability within a given vowel class, while calculating distance between minimal pairs is difficult with naturalistic data. Hay, Warren, and Drager (2006) introduced an alternative approach that accounts for the variability between those clusters, taking token-specific formant values, rather than averages, as input. The measure is the Pillai-Bartlett statistic, an output of a Multivariate Analysis of Variance (MANOVA), which represents the proportion of one variance that can be predicted by another variance. A higher Pillai indicates a lower degree of overlap between two vowel clusters in F1/F2 space. Since the value is derived from a MANOVA, Pillais can account for known internal factors influencing the production of merger, such as phonological environment. This paper applies the method to results from low back merger in California English (Hall-Lew 2009), and further suggests that the Pillai is useful for any vowel change, taking the fronting of the mid- and high back vowels as an example. © 2010 Acoustical Society of America.