Today, March 10, John Kingston and Shigeto Kawahara present "You got to be discriminating to get contrast" at the New England Sequencing and Timing (NEST) meeting, held at UMass. An abstract of their talk follows.
At NEST in 2007, we presented a study of Italian, Norwegian, Japanese and English speakers' categorization and discrimination of a silent interval that varied in duration (see also Kingston, Kawahara, Chambless, Mash, & Brenner-‐Alsop, 2009). This interval has flanked by vowels in the speech condition and by filtered square waves in the non-‐speech condition – in the speech condition, the silent interval was perceived as a voiceless stop consonant. The preceding vowel's or filtered square wave's duration was varied orthogonally from the silence's duration. Except when a consonant's and a preceding vowel's durations vary inversely in their native language (Italian and Norwegian listeners) and the stimuli were speech, listeners judged the silence to be longer when the preceding sound was longer. We described the general finding as a product of listeners' adding the durations of the two intervals together in all other conditions. These findings accord with those reported by Fowler (1992) but not those reported by Kluender, Diehl, & Wright (1988), whose listeners categorized the silent interval as longer when the preceding vowel or square wave was shorter. Fraisse (1963, 1984) and more recently Nakajima, Hoopen, Hilkhuysen, & Sasaki (1992) showed that so long as the duration ratio between successive intervals is close to 1 and doesn’t exceed 2, listeners judge the second interval as long after a long first interval, but once the ratio greatly exceeds 2, they judge the second as long after a short first interval. This observation may explain our earlier results because most of the duration ratios in our stimuli were in the 1-‐2 range.
At this NEST, we will report new results using more extreme ratios, up to 3, which show that listeners discriminate silences better when their durations vary inversely with the durations of the preceding vowels or non-‐speech sounds. Contrast still did not arise in categorization, even with the largest duration ratios. Kato, Tsuzaki, & Sagisaka (2003) report that listeners treat variation in the onset times of successive vowels but not their offset times as evidence of rate variation. Because manipulating the duration of the silence varied the onset time between successive vowels or square waves in our earlier stimuli, our listeners may have been covertly judging rate rather than the silence's duration relative to the preceding vowel's or square wave's. We may also report the results of an experiment examining this explanation of our earlier results.