Rate effects on Swedish VOT: Evidence for phonological overspecification

Abstract

Previous research has found asymmetric effects of speaking rate on VOT cross-linguistically: as rate slows, long-lag VOTs and negative VOTs increase, but short-lag VOTs remain essentially unchanged. If we assume, as have many phonologists, that the two-way contrast in voicing languages (e.g. French) is [voice] vs. [Ø] and in aspirating languages (e.g. English) is [spread glottis] vs. [Ø], then it appears that at slower rates, a phonological contrast is heightened by selective increase in the phonetic cue for the specified feature. Thus, slowing down causes longer aspiration in aspirating languages and longer prevoicing in voicing languages but no change in short-lag stops. We report the results of an experiment on Central Standard Swedish stops designed to investigate the effect of speaking rate on VOT. CS Swedish uses both prevoiced and aspirated stops in utterance-initial position, hence the phonological feature(s) involved in this contrast is not clear. We found that both prevoicing and aspiration increase in slow speech in Swedish. This suggests that both [voice] and [spread glottis] are the specified features of phonological contrast in CS Swedish, and in turn raises questions about whether phonological specification more generally is economical. Moreover, the fact that speaking rate affects VOT even in situations like CS Swedish in which the phonological contrast is over-specified suggests that such modification is largely due to production dynamics, not speakers’ sensitivity to listeners’ needs.