Comment on "A geometric representation of spectral and temporal vowel features: Quantification of vowel overlap in three linguistic varieties" [J. Acoust. Soc. Am. 119, 2334-2350 (2006)] (L)

In a recent paper by Wassink [J. Acoust Soc. Am. 119, 2334-2350 (2006)] the spectral overlap assessment metric (SOAM) was proposed for quantifying the degree of acoustic overlap between vowels. The SOAM does not fully take account of probability densities. An alternative metric is proposed which is based on quadratic discriminant analysis and takes account of probability densities in the form of a posteriori probabilities. Unlike the SOAM, the a posteriori probability-based metric allows for a direct comparison of vowel overlaps calculated using different numbers of dimensions, e.g., three dimensions (Fl, F2, and duration) versus two dimensions (Fl and F2).