Effects of reliable spectral-temporal characteristics of listening context on perception of monophthongs and diphthongs

Abstract

Brief experience with reliable characteristics of listening context, such as a precursor sentence, alters perception of subsequent vowel sounds [M. Kiefte and K. Kluender, J. Acoust. Soc. Am. 112, 2248 (2002)]. For example, an appropriately filtered precursor sentence serves to perceptually cancel effects of second formants (F2) of vowel sounds, forcing listeners to rely solely on spectral tilt for identification. Auditory systems effectively absorb spectral characteristics, including formant peaks, if those characteristics are reliable (redundant) properties of the listening context. Present studies investigate whether this perceptual calibration extends to reliable spectral-temporal properties. Listeners identified a series of stimuli with incremental increases in slope of F2 that varied perceptually from /u/ to /ui/. When repeated formant transitions, exactly matching those of the target vowels, were added to precursor sentences, /i/ percepts predominated despite no stimuli being appropriate for /i/ in isolation. When repeated formant transitions were reversed in precursors (high to low frequency), the same results were obtained. Because perceptual absorption of reliable acoustic characteristics was indifferent to F2 trajectory, it appears that processes that calibrate to reliable properties of a listening context are sensitive only to spectral, not temporal, composition of listening context. [Work supported by SSHRC and NIDCD.]