I was looking for some information about error correction in phonology, and found this interesting (albeit short) article by Janet B. Pierrehumbert:
(Academia-bashers need not apply.)
The article focuses on lexicon in general, rather than phonology, although section 2, The phonological code, does mention error correction.
A few quotes to whet your appetite:
“Many individual words can be uniquely identified even if one or more segments are missing. This is shown by phoneme restoration experiments, in which people fail to notice that a speech segment has been replaced by noise, and by gating experiments, in which people prove able to progressively narrow the set of lexical choices as more and more of the word is provided, often achieving a unique identification before the end of the word.”
“Like species, lexical innovations compete with preexisting forms to survive. Words are viable only insofar as they are successfuly replicated. For species, biological reproduction is the mechanism for replication. For words, the mechanism is imitation. Children bring to the task of language acquisi- tion fundamental drives to attend to and imitate speech patterns, and to map word forms to word meanings on the basis of phono- logical and semantical contrast.”
“A rare word may simply fail to occur by chance in the experience of a learner, and in that case it will not be learned and reproduced for future learners. In the aggregate, statistical sampling considerations mean that the frequencies of individual words are subject to random walk effects over generations, and that any word whose frequency happens to become too low will be irretrievably lost. The random walk of frequencies can create morphological gaps. It entails that the total number of distinct words in the community lexicon would decrease over time, if new words were not continually added.”