ABSTRACT
Auditory-word recognition is a remarkably robust aspect of language processing. The acoustic signal may be distorted and/or disrupted in a wide variety of ways, yet spoken-word recognition appears to proceed with little difficulty. Among those potentially problematic aspects of spoken language are (a) misarticulations in the form of speech errors, (b) dynamic processes among successive phonemes, and (c) masking of speech sounds due to environmental noise. Beyond the inherent variability in the acoustic signal, one specific observation that has been emphasized in early models of spoken-word recognition is that the speech signal is distributed over time. Unlike visual language, in which information may be obtained in parallel, information in the speech signal is available to the listener in a time-dependent manner. Simply stated, the beginnings of words reach a listener’s ear prior to the ends of words. These observations concerning the physical characteristics of spoken language have influenced researchers concerned with the recognition of phonemes and with how listeners recognize spoken words. The implicit assumption in much of the research has been that the problems of ambiguity and distortion will have catastrophic consequences for word recognition. The focus of the current chapter is whether this assumption is warranted. I report data from a number of experiments that suggest perturbations in the speech signal are not major impediments to word recognition.
