Perceptual basis for the compact‐diffuse distinction for consonants

Abstract

A distinguishing acoustic property of velar obstruents is a prominent midfrequency “compact” spectral peak (roughly 800–2000 Hz). Is there evidence for a qualitatively different auditory response for sounds with compact as opposed to diffuse spectra? We generated two ten‐item series of bursts by exciting a group of resonators with white noise and systematically varying the relative amplitudes of the spectral peaks corresponding to formants 2 to 5. Stimuli at one end of each continuum had a compact, k‐like spectrum and at the other end a diffuse p‐like or t‐like spectrum. The bursts were appended to short, transitionless vowels with or without intervening aspiration. Identification functions were obtained for the syllable‐like stimuli (with response alternatives p t k b d g) and for the isolated bursts (using an ABX paradigm). The results suggest that a sound with a compact spectral peak can cause a distinctive perceptual response when that peak is sufficiently narrow and projects well above adjacent spectral minima. Responses are also influenced by the amplitude of the burst in relation to the following vowel and by the burst duration. Possible mechanisms underlying this distinctive response are discussed. [Work supported in part by NINCDS.]