I would think it is simply impossible to count any such thing ... esp in speech -
there are some big studies of "corpora" of language use - but they still dont really show what you suggest, as far as I know but you might get some idea from following this link