Perceptual Assesst"lent of F'ricative~stop Coarticulation*

The perceptual dependence of stop consonants on preceding fricatives (Mann and Repp, in press) was further investigated in two experiments employing both natural and synthetic speech. These experiments consistently replicated our original finding that listeners report more velar stops following [s]. In addition, our data confirmed earlier reports that natural fricative noises (excerpted from utterances of [sto..] , [sko..] 9 [S to..] , and [5ko..]) contain cues to the following stop consonants; this was revealed in sUbjects' identifications of stops from isolated fricative noises and from stimuli consisting of these noises followed by synthetic CV portions drawn from a [to..] [ko.] continuum. However, these cues in the noise portion could not account for the contextual effect of fricative identi ty ([ S] vs. [sJ) on stop perception (more "k" responses following [s]). Rather, this effect seems to be related to a coarticulatory infl uence of a preceding fricative on stop production: Subjects' responses to excised natural CV portions (with bursts and aspiration removed) were biased towards a relatively more forward place of stop articulation when the CVs had originally been preceded by [s]; and the identification of a preceding fricati ve was biased in the direction of the fricative context in which a given CV portion had been produced These s support an articulatory explanation for the effect of ing fricatives on stop consonant perception. In a recent paper (Mann & Repp, in press), we described a perceptual dependency of consonants on preceding fricatives: a stop ambiguous between [t] and [k] was more likely to be labeled "k" when preceded by [s] than when preceded by ] or by no fricative at all This perceptual context effect was demonstrated in a series of experiments with synthetic speech. The present experiments employed both natural and synthetic speech to investigate further the possible of this effect.