Not that controversial, but I think there’s a lot to be said for diachronic explanations of language, and the synchronic/diachronic distinction is somewhat artificial.
Nick Nicholas’ answer to What is functional grammar? will explain that a little bit: functional accounts are kind of diachronic to begin with (what function does this linguistic component serve in communication => how did this linguistic component develop to fit this function). And more hardcore functionalism is all about language structures as process rather than as blueprints; hence the more extreme formulation of Paul Hopper’s Emergent Grammar.
There are linguists who think that way, but it’s fair to say it’s a minority view.
I’ve ranted defending philology in Nick Nicholas’ answer to What are the differences between linguistics and philology? That looks controversial and fuddy duddy; where it’s actually going isn’t, although again it’s a minority view: language is socially embedded, and ignoring the social to focus only on linguistic structures is a simplification. It is an epistemologically necessary simplification at times, of the kind you see all the time in less mushy disciplines like Physics (frictionless plane).
But language change in particular is always socially mediated; and much more “pure” “synchronic” language stuff is socially messed up than people like to admit. In particular, language phonemes at times seem to me to be a frictionless plane.