The VOT Category Boundary in Word-Initial Stops: Counter-Evidence Against Rate Normalization in English Spontaneous Speech

Abstract

Some languages, such as many varieties of English, use short-lag and long-lag VOT to distinguish word-and syllable-initial voiced vs. voiceless stop phonemes. According to a popular view, the optimal VOT category boundary between the two types of stops moves towards larger values as articulation rate becomes slower (and speech segments longer), and listeners accordingly shift the perceptual VOT category boundary. According to an alternative view, listeners do not shift the VOT category boundary with a change in articulation rate, because the same category boundary remains optimal across different rates of articulation in normal speech, although a shift in the optimal boundary location can be induced in the laboratory by instructing speakers to use artificially extreme articulation rates. In this study we compared the effectiveness of rateindependent VOT category boundaries applied to word-initial stop phonemes in spontaneous English speech, against the effectiveness of Miller et al.'s (1986) rate-dependent VOT category boundary applied to laboratory speech. The effectiveness of the two types of category boundaries were comparable, when spontaneous speech data were controlled for factors other than articulation rate. Our results suggest that perceptual VOT category boundaries need not shift with a change in articulation rate under normal circumstances.

The singleton-geminate distinction can be rate dependent: Evidence from Maltese

“…The results indicate that in contrast to VOT in English (Nakai & Scobbie, 2016) and consonant duration for quantity in Cypriot Greek (Arvaniti, 1999), the optimal boundary between singleton and geminates in Maltese varies with speaking rate and does so considerably, given the large range (80-130 ms) of estimated boundaries. Moreover, the data also indicate that both categories are similarly affected by speech rate.…”

“…As Nakai and Scobbie (2016) noted, the utility of speech-rate normalization may differ per contrast (see also Port, 1979). For the English voiced/voiceless distinction, the data indicate that there is little need for normalization of VOT, as voiced stops do not have such strongly extended VOTs at slow rates so that they become like aspirated stops as produced in fast rates.…”

“…Optimal boundaries were estimated following the procedure of Nakai and Scobbie (2016), who followed the procedure proposed by Miller, Green, and Reeves (1986). The same algorithm was used here to estimate the optimal boundary based on measured consonant duration.…”

“…However, there are two reasons why the singleton-geminate distinction might not be rate-dependent. First, Nakai and Scobbie (2016) argued that rate-normalization might not be necessary for some contrasts because the category boundary does not necessarily shift with rate. For McQueen, 2011).…”

“…To answer this, we analyzed a corpus of utterances containing singleton-geminate minimal pairs produced in a sentence context. We used the same methods as used by Nakai and Scobbie (2016) for VOT duration in English stops, by testing whether categorization accuracy improves when speech rate is taken into account. Second, do listeners consider (distal) speech rate when categorizing sounds as singleton versus geminate?…”