The general enterprise
I will comment first on the general enterprise that Patricia Keating outlines in her paper and then on her discussion of coarticulation, including specifically her proposals of target “windows” and interpolation mechanisms for generating articulatory contours.
Here and elsewhere, Keating suggests that much of phonetics, previously considered outside the domain of a grammar, in fact, deserves incorporation into the grammar as a set of rules: “More and more, the phonetics is being viewed as largely the same sort of creature as the phonology, that is, a formal set of rules.” (Keating 1985a: 19) The rules take the nonphysical, symbolic, phonological representation and derive “a more physical representation” (This volume, p. 451).
One motivation for this elevation of phonetics has been the failure of universal phonetics. Most phonetic generalizations are found to have exceptions, and therefore, it seems, they are of linguistic interest. Further, Keating suggests (1985a) that study of phonetic regularities reveals some patterns evident in certain languages that resemble phonological patternings in other languages (For example, whereas many languages show 20-30 ms of acoustic vowel shortening before a voiceless consonant, in some languages including English, the duration difference is much larger and, apparently, has been phonologized as a length difference.) The finding that languages may exhibit more-or-less the same patterning, but at different levels of representation in the language suggests to Keating that phonetic regularities are “not something divorced from the rest of the grammar, but something controlled as part of the grammar” (1985a: 19).