Integrality of nasalization and F1 in vowels in isolation and before oral and nasal consonants: a detection-theoretic application of the Garner paradigm.

In vowel height contrasts, tongue height and soft palate height covary. A series of vowel classification experiments examined the perceptual interactions between F1 and nasalization, the principal acoustic correlates of these articulations. Listeners classified imperfectly discriminable stimuli in the set of tasks that compose the Garner paradigm. Detection-theoretic models applied to the data led to the conclusion that vowels, whether in isolation, before oral consonants, or before nasal consonants, display integrality of F1 and nasalization. The contrary conclusion reached by Krakow et al. [J. Acoust. Soc. Am. 83, 1146-1158 (1988)] on the basis of data from a trading relations experiment reflect a limitation of that design for studying perceptual interaction. A second experiment used an array "rotated" in the stimulus space to determine whether F1 and nasalization are privileged, perceptually primary dimensions. A new method for predicting classification performance for the rotated array without the assumption of primacy showed that they are not.