Auditory‐perceptual interpretation of the vowel

Abstract

An account of the simple vowels of American English will be given in terms of the author's auditory‐perceptual theory. Data from many sources will be used to illustrate and support this view. In addition to theoretical suggestions concerning an auditory‐perceptual transformation and a segmentation rule, the perceptualized spectral patterns associated with the nonretroflex vowels will be shown to fall into regions within a slab in a three‐dimensional, auditory‐perceptual space. These regions form a “vowel map” that by simple rotations can be related to various “vowel charts” such as those of Jones, Pike, or Fant. While the presentheory is an elaboration of traditional formant‐ratio theory, it will be shown that a concept of a perceptual reference is needed not only for talker normalization but also to disambiguate vowels not distinguished by simple formant‐ratio theory. Additionally, loci for the retroflex and nasalized vowels illustrate the utility of describing vowel spectra by five numbers combined into three dimensions. Simple F1 by F2 descriptions are not adequate except for enigmatophiles. [Work supported by NINCDS and AFOSR.]