Estimating lingual constriction location in high vowels: a comparison of EMA- and EPG-based measures

Abstract

EMA and EPG offer complementary information about tongue articulation. In this paper, comparisons are made of lingual constriction location estimates for high vowel continua, based on EMA and EPG data simultaneously recorded for one speaker. These data pertain to an experiment which correlated articulatory estimates of constriction location with formant measurements for comparison with Fant’s (1970) model-based nomograms. A circle model was used to estimate the constriction location from two dorsally located EMA sensors. The center of the constriction was deemed to be where the circle was closest to the roof of the mouth. Those estimates are compared to three types of EPG-based estimates of lingual constriction. The first is the row of narrowest constriction in a scaled vertical view of the EPG palate. The second is a “center of gravity” estimate. The third is an approximate side view of the tongue contour, drawn on the basis of the innermost activated electrode on each EPG row. Results suggest a broad agreement as well as consistent discrepancies between EMA and each EPG estimate. Likely reasons for discrepancies are discussed, and it is suggested that some of these may offer new insights into three-dimensional aspects of articulation that might be exploited in research.