Perception of vowel duration

Abstract

Lehiste and Pisoni have independently shown that the perception of vowel duration is influenced by the F0 pattern that accompanies the vowel. Here we report some related results. Sixteen subjects are asked to judge the relative duration of pairs of synthesized stimuli. The stimulus pairs are made up of the vowels /i/, /a/, /ai/, and nonspeech, where nonspeech is a single formant at 1500 Hz. The F0 patterns are: level = 120 Hz, rising = 105–135 Hz, and falling = 135 to 105 Hz. The durations are 220, 240, 260, and 280 msec. The orthogonal combination of these factors makes up a total of 512 stimulus pairs. In comparing duration judgements of stimuli of rising (R), falling (F), and level (L) F0's, we found that the following relation obtains: R > F > L, where “>” denotes “perceived as having greater duration.” This relation holds to the same extent for both speech and nonspeech stimuli. There is no signficant difference in the perceived duration between /i/ and /ai/. The most consistent effect, found also earlier in a pilot experiment we performed with 58 subjects, is that /i/ is heard as longer than /a/, in a ratio of 60% : 40%. Since /i/ is normally shorter than /a/ in natural speech, we posit a compensation mechanism mediating between perception and production that is responsible for this negative correlation between vowel height and duration. It is interesting to note that although the processes of duration judgment and pitch judgement (cf. the abstract by Chuang and Wang) are different, they seem to share a common compensation mechanism between production and perception. [Work supported by NSF Grant No. BNS 76-00017-Wang-2/78J34.2.]