Evidence for Quantal Vowel Articulations

Abstract

The articulation of different vowels is often represented by points located in multidimensional “articulatory space,” where the dimensions represent the coordinates of tongue and lip positions. Data are presented to show that the relations between vowel articulatory positions and properties of the acoustic output (e.g., the formant frequencies) are highly nonlinear. In some regions of articulatory space, the properties of the acoustic signal are relatively uninfluenced by perturbations in articulation, and these are separated by other regions in which the acoustic output is quite sensitive to articulatory perturbations. The former group of regions represent configurations appropriate for vowels in language since the requirements on articulatory precision are minimal. It can be argued that there are six basic quantal vowel articulations, corresponding to /i e æ ɑ o u/, and that the vowels in any language are always selected from these regions. Additional vowels are actualized by superposing on these basic vowels changes in additional articulatory “dimensions” such as rounding, nasality, tongue-root widening, duration, and fundamental frequency. [Work supported in part by the Air Force Cambridge Research Laboratories and in part by the National Institutes of Health.]