Devoicing of word-final stops

A process devoicing stops in word-final position is common among languages and is evidenced in Appalachian and Black English. Attempts to explain this phenomenon, however, have been unsuccessful. For example, generative phonology, which accounts for change by rule-addition, rule-loss, and rule-reordering, is purely descriptive. Likewise, natural phonology, which claims that change occurs when a child fails to master a phonetic opposition, never explains under what conditions “failure” occurs. Instead, devoicing may be explained in terms of the voicing cues for post-vocalic stops. (1) Voicing is signalled by the manner in which the preceding vowel is terminated. (Termination is defined in terms of the period of vocal cord vibration during the transition from vowel to stop.) [F. Parker, J. Phonetics 2, 211–221 (1974).] (2) The cue for a post-vocalic voiced stop is perceptually weaker than that for its voiceless counterpart. (Evidence comes from error detection studies and phonological change.) Thus, devoicing represents a strategy by which speakers maintain the integrity of the speech signal. This makes sense, because the processes of production and perception are at odds: encoding processes reduce acoustic differences between consonants and vowels, while decoding requirements force consonants and vowels to be kept distinct.