Recoverability refers to the ease of recovering the underlying form--stored mental representations--given a surface form--actual, variable output signals s (e.g., [Daet^, Daet[superscript h] ] ? /Daet/ "that"). Recovery can be achieved from phonetic cues explicitly present in the acoustic signal or through prediction from the context. However, recoverability can be compromised when the information in the signal is insufficient or the predictability in a given context is not high enough. The purpose of this dissertation is to investigate through experiments and computational modeling how phonotactic predictability in a given context affects (i) the amount of information present in the signal during production, (ii) the attention paid to the information during perception, and (iii) how reliance on phonotactics and phonetic cues might be learned. The language in focus is Japanese, a language well-known for its CV phonotactic preference, and the process of high vowel reduction, which often results in consonant clusters that violate this phonotactic restriction. The results suggest that language users prioritize predictability during speech processing, with phonetic-cue interpretation applying when predictability is not reliable enough. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://bibliotheek.ehb.be:2222/en-US/products/dissertations/individuals.shtml.]