This dissertation examines /s/-retraction, a sound change in progress by which /s/ approaches /sh/ in the context of /r/, such that street may sound more like shtreet. Previous research on this phenomenon has focused largely on the production of /str/ clusters, asking how /str/ is articulated, how it is realized acoustically and what phonological process ultimately leads to a more /sh/-like /s/. This dissertation contributes to the understanding of the phenomenon by turning to the perception of /s/-retraction. Specifically, this dissertation seeks to answer two fundamental questions: Do listeners have detailed phonological knowledge about /s/-retraction? And if so, how do they use that knowledge? The results of three different experiments are presented that examine different aspects of how listeners may use their knowledge about /s/-retraction and what that might tell us about the origins, grammatical status, and trajectory of /s/-retraction as a sound change and the transition and propagation of sound change in general. Firstly, Experiment I asks how that knowledge influences speech processing. Using a lexical identification task with eye tracking, Experiment I finds that listeners can use the cues of /s/-retraction in order to anticipate the upcoming /r/. Secondly, Experiment II asks how that knowledge influences sociolinguistic perception. Using a phoneme categorization task, Experiment II finds that only listeners who most strongly endorse traditional stereotypes of masculinity are likely to attribute a retracted /s/ to a performance of masculine toughness. Finally, Experiment III asks how the perception of /s/-retraction influences an individual's own production when they take a turn as a speaker. Using a covert shadowing task, Experiment III finds that listeners converge toward manipulated degrees of /s/-retraction, but only if the model talker exhibits a pattern sufficiently similar to their own. Taken together, the findings for /s/-retraction challenge a convergence path to sound change, by which conversational shifts persist and accumulate to lead to lasting change, but support a coarticulatory path to sound change, by which listeners gradually shift the cues of coarticulation from the source to the target.