Parsing time and rate normalization versus durational contrast

Abstract

An acoustic interval is perceived as shorter next to a long neighboring interval than next to a short one. This finding has been attributed to rate normalization (Miller and Liberman, Percept. Psychophys. 25, 457–465, 1979) and to durational contrast (Diehl and Walsh, J. Acous. Soc. Am. 85, 2154–2164, 1989). In the first experiment, we present listeners with a vowel varying in duration from short to long in a context where it could be interpreted as one vowel or two in sentences manipulated to sound as though they were spoken at fast, moderate, and slow rates. We predict that the vowel is more often perceived as two vowels in the fast than the slow sentence. The second experiment tests the effect of parsing the vowel into one versus two vowels on judgments of a [b-w] transition duration continuum and a [b-p] voice onset time continuum. If the vowel is parsed into two vowels, a given transition duration or VOT will sound longer and thus more [w] or [p]-like than if it is parsed into just one. Such a result would reflect durational contrast between the transition duration or VOT and the vowel’s perceived duration. It would not reflect rate normalization directly but only indirectly via its effect on parsing. [Work supported by NIH.]