Altered Auditory Feedback (AAF) is a powerful instrument to reduce the frequency of the stuttering episodes, although the rationales for this effect are largely unknown (Bloodstein & Bernstein-Ratner, 2008). It is also known that stutterers' fluent speech is affected by the intrinsic characteristics of the speech units, as the frequency of syllable's occurrence (Smith et al., 2010, among others) and the articulatory complexity of the syllables (Howell et al., 2006, among others). The present study describes the influence of the AAF on the production of different types of syllables varying for frequency and complexity in four adult females who use to stutter (AWS). The subjects differ in stuttering severity according to SSI-3 (Riley, 1994). The research topics were: o the influence of feedback condition, articulatory complexity and frequency of occurrence on target syllables, by quantitative analyses of disfluencies and errors; o the influence of feedback condition, articulatory complexity and frequency of occurrence on the fluent target syllables, through the analyses of acoustic indexes: - duration of target CV and CCV syllables (with V=/a/); -intrasyllabic coarticulation of CV syllables (C=voiced plosives, V=[i];[u];[a]), according to Locus of Equation method (Sussman et al., 2010). Subjects repeated nine times each target syllable after a pre-recorded voice, under normal auditory feedback (NAF) and AAF. Target syllables were always embedded within the phrase "Say CVt, then CVt, then CVt". The alteration of the auditory feedback was a combination of delay auditory feedback (DAF) whit a temporal delay of 60 ms and frequency shifted auditory feedback (FAF) with a reduction of 40% of the fundamental frequency (F?). This reduction is the most efficient one to keep fluency in adults who stutter (Antipova et al., 2008). Results showed that AAF was efficient to facilitate fluency in almost two subject. In the two subjects who produced disfluencies and errors under NAF, both higher articulatory complexity and lower frequency of occurrence of syllables increased the number of stuttering episodes and errors. Within each feedback condition, higher articulatory complexity and lower frequency of occurrence systematically affected the duration only for the most severe subject, but when feedback condition was considered as a within-subject variable in a repeated measure model of ANOVA, the articulatory complexity and frequency of occurrence, and their interaction with feedback condition, did not influence the length of target syllables for none of the subjects. Instead, only the AAF condition significantly increased the length of target syllables (for three out of four subjects). As regard the intrasyllabic coarticulation, subjects varied their answers in relation to the AAF, but, interestingly, the only subject which did not increase durations under AAF, reduced significantly the extent of CV coarticulation. We supposed that it might exist two general strategies to promote fluency: the articulatory slowing down (without changing coarticulation) and the reduction of CV coarticulation's degree. Every subject, according to its stuttering severity, could use one of them (or a combination of two).
