One theory of propositions is that they're "useful fictions." However, this faces the formidable objection that fictionalism makes it hard to explain how something fictional can tell us about the real world.
Platonism is implausible, and so the next most obvious realist choice is to say that propositions only exist to the extent that people express them (whether via linguistic or mental representations). But propositions seem too numerous to square with that conception of realism. For example, there is probably a very large number n such that p = "the nth digit of pi is 7" is true, but nobody will ever express p.
One alternative that seems natural is that some propositions are fictional and others aren't. This "partial" fictionalism is probably modest enough to avoid some of the usual objections against full-blown fictionalism or full-blown realism. Surely it has been defended before. Does anyone have helpful references for such a view?