Goodluck and Solan (EJ 205 641) presented alternative formulations about why errors predicted by basic operations fail to occur and suggested a refined hypothesis. Each aspect of their argument is addressed, and it is concluded that descriptive power, methodology and principles for restricting error predictions favor our original analysis. (RD)

Presents two analyses of deviant phonological systems. The one based on production evidence alone accounts for (1) differences in surface behavior of a set of phonetic segments with three possible phoneme sources, and (2) obstruent clusters across morpheme boundaries. The "substitution analysis" identifies the child's underlying representations…