Acoustic parameters for place of articulation in nasals and stops

Abstract

A metric that identified labial and coronal place of articulation in English nasal consonants with over 80% accuracy had been developed by correlating distinctive patterns of rapid energy change at the nasal release with each place of articulation [K. M. Kurowski and S. E. Blumstein, J. Acoust. Soc. Am. 81, 1917–1927 (1987); K. M. Kurowski, unpublished dissertation, Brown Univ. (1990)]. This study attempted to address two questions. (1) Could the same metric be successfully applied across manner to classify place of articulation in homorganic voiced stops? (2) In the event that the metric failed here, to what extent did the burst onset alter the patterns of rapid energy change predicted by the metric? The same three speakers used in the nasal study (1990) produced CV syllables, consisting of [b,d] followed by [i,e,a,o,u]. The results showed that the classification scores across speakers for the voiced stops (labial = 64%; coronal = 83%) were lower than those for nasals (labial = 79%; coronal = 87%). Still, the general patterning of the results was the same across manner: Labial scores remained lower than those of the coronals due to the intractable problem of classifying labials in the environment of [i]. Two reanalyses using different window sizes demonstrated that the aperiodicity of the initial portion of the burst inflated some targeted areas of energy change, adversely affecting the performance of the metric. Various approaches to the question of how to reconcile formal aspects of the present metric with the added complexity introduced by the voiced stops are discussed in the light of their ramifications for the theory of acoustic invariance. [Work supported by NIH]