Josef Fruehwald (2012)
I argue that some phonetic changes must be phonologically conditioned from their outset. I support this argument based on two conditioned sound changes in Philadelphia: /ay/ raising and /ey/ raising. /ay/ raising has always interacted opaquely with /t/ and /d/ flapping, suggesting it has always been sensitive to the underlying voicing of the following segment, rather than the phonetic realization of that segment. For /ey/ raising, a following /l/ appeared to favor the direction of the change before the change began, but /eyl/ actually never underwent it. In my analysis, this is because /l/ doesn't meet the phonological definition of the conditioning context, even though it favors the change phonetically.
Presented at NWAV 41