ASA 125th Meeting Ottawa 1993 May

5pSP15. Integrality of nasalization and F1 in vowels in isolation and before oral and nasal consonants.

John Kingston

Linguist. Dept., South College, Univ. of Massachusetts, Amherst, MA 01003

Neil A. Macmillan

Univ. of Massachusetts, Amherst, MA 01003

Vowels vary multidimensionally, and the perceptual effects of the dimensions may combine independently or may interact. The dimensions examined in these experiments were the acoustic correlates of soft palate height and tongue height, which covary directly in natural speech. A modified Garner paradigm was used in which accuracy replaced speed as the dependent measure; a detection-theory analysis permitted inferences about perceptual representation. For isolated vowels, nasalization and F1 were integral, in the sense that corresponding perceptual dimensions were correlated, replicating Kingston [Phonetica 47, 149--179 (1991)]. The degree of integrality was yet stronger for vowels before consonants, whether the consonant was nasal or oral. These results suggest that soft palate height and tongue height are covaried by speakers so that their acoustic correlates contribute to a single perceptual property, the distribution of energy at the low end of the vowel's spectrum. The lack of a context effect is unexpected given the finding of Krakow et al. [J. Acoust. Soc. Am. 83, 1146--1158 (1988)] that nasalization in a vowel does not affect judgment of its height if there is a neighboring nasal to which that nasalization can be attributed. This discrepancy may occur because listeners in this experiment made judgments of nasalization as well as vowel quality.