There's a lot of helpful material in here on the relationship between illocutionary acts (which are generally thought to require speaker communicative intentions to be felicitious) and neighboring types of speech acts—in particular what McGowan calls "conversational exercitives", which are changes to the normative structure of conversations brought about by speech, and which have a much lower threshold of intentional-ness than standard exercitive illocutions.
I *think* there's a way of applying some of McGowan's apparatus to the phenomenon that I care most about, namely explicit meaning claims (of the kind the ordinary language philosophers make), which usually take the form of explicit assertions, but I think mainly function to reinforce norms of use (this is something in the neighborhood of some things Cavell says). Roughly, someone can intend to make an assertion about the meaning of an expression while simultaneously having all kinds of not-necessarily-intended normative effects on what's permissible in the conversation and related norm-governed non-conversational activities. (Now that I write that out it seems kind of obvious.)