Two studies investigated the influence of the independently varied durations of preceding and following signal portions on the amount of closure silence needed to perceive ‘splash’ rather than ‘slash’. Increases (or decreases) in the durations of the [s] and [l] acoustic segments had opposite effects which cancelled when the silent intervals were short (experiment 1), but yielded a net effect due to [s] duration when the silent intervals were long (experiment 2). These findings, which resolve a conflict between earlier results in the literature, are interpreted as reflecting a perceptual compensation for coarticulatory shortening of [s] before stop consonants, in conjunction with (possibly psychoacoustic) contrastive interactions between the perceived durations of adjacent acoustic segments. The results suggest that local temporal signal properties, as distinct from global perceived speaking rate, are an important factor in phonetic per