I don't know necessarily about all Pennington's assumptions in this review of April McMahon's Change, Chance, and Optimality--I've learned a bit in the year since I first encountered the article, but it's still unclear why, e.g.underlying lexical representations have to be learnable rather than a description or theoretical construct--they still have explanatory power, and I suspect the whole project of treating e.g. optimality theory as learnable is foolish(--which I guess makes me an exemplar-dynamic man--we hear, internalize, imitate)--but this was the first piece of work to posit social constraints on phonology, which opens up huge vistas for those who want to explain stuff in OT terms. [ /-MASCULINE], anyone? ( )