Differences in production and perception of VOT in Polish and English

Abstract

English and Polish use different Voice Onset Time (VOT) categories to contrast voiced and voicelesstops: short lag and long lag (English) versus prevoiced and short lag (Polish). When VOT is measured for word‐initial stops in minimal pairs, sentences, and conversation, the VOT distributions for voiced and voiceless stops are clearly separated in Polish, but not always in English, especially in casual speech [Lisker and Abramson, Language and Speech 10, 1–28 (1967); Moslin, Brown University dissertation (1978)]. The perception of VOT in synthetic stimuli was compared with production data for Polish and American English listeners. Both groups labeled stimuli in three continua that varied in range of VOT: −20 to + 80, −100 to +50, and −100 to +20 ms VOT. The Polish, but not the American, listeners showed a mean boundary shift (from +20 to +4 ms VOT) as a function of these ranges. Thus the Polish VOT perceptual categories are somewhat unstable and do not always match the VOT production categories, although th...