Recent work investigating the role of probabilistic phonotactics in spoken word recognition suggests the operation of two levels of representation, each having distinctly different consequences for processing. The lexical level is marked by competitive effects associated with similarity neighborhood activation, whereas increased probabilities of segments and sequences of segments facilitate processing at the sublexical level. I will discuss a series of studies that provide support for the hypothesis that the processing of spoken stimuli is a function of both facilitative effects associated with increased phonotactic probabilities and competitive effects associated with the activation of similarity neighborhoods. I will also describe recent extensions of this work aimed at evaluating two hypotheses regarding the segmentation of words from fluent speech, one phonotactic (the trough hypothesis) and one lexical (the lexical burst hypothesis). Finally, I will describe our attempts to account for effects of neighborhood activation and probabilistic phonotactics from the perspective of Grossberg's adaptive resonance theory (Grossberg, Boardman, and Cohen, 1997).