Odd conditions

Abstract

Abstract Since the 19th century linguists have expected to find conditioned sound changes in environments that make phonetic sense: consonants palatalize adjacent to front vowels, back vowels front if a front vowel occurs in the next syllable, stops voice between voiced segments, and so forth. Most conditioned sound changes conform to this expectation, but a surprising number do not. Some of these are well known, as the palatalization of *s before most word-initial consonants in High German. Since there is no obvious explanation for them, such changes are generally ignored in discussions of historical phonology. The result of this practice has been to give the false impression that what appear to be phonetically unmotivated sound changes are rare abnormalities that probably would conform to expectation if we had more information about them. This paper draws attention to examples of conditioning in Austronesian languages in which the phonetic properties of the context appear unrelated to those of the change, and it questions why such changes should occur. Although finding a completely satisfactory explanation has proven difficult, one general conclusion suggested by the data is that native speakers have an intuitive recognition of natural classes that is independent of phonetic motivation.