I think that Jay, Keith, and Bill agree that an explanation of what is
going on at any moment cannot be given (except in special cases) in terms
of context-independent rules. So we have a theoretical problem of how to
account for the fact that participants generally act as if they know what's
going on, even though we cannot account for this in terms of general rules.
Jay introduces the important idea that we must also consider participants'
theories about what they are doing. And in most of our cases, those
theories of speech and action tend to be rule-based. So part of our
account of what is going on in any particular context must be the
participants' own rule-based theories about what they are doing -- even
though we as analysts know that such rule-based theories are always
incomplete.
With regard to play, this raises the question of folk theories of play,
and how these enter into actual instances of play -- both as those
theories influence the children's sense of what they are doing, and the
analyst's account.
Stanton Wortham