One Wittgensteinian view of religious language (language about God and salvation) is that it shouldn't be considered incorrect or without real referent but rather a different sort of language game or usage altogether from the one which is committed to truth, real referents, consistency of meaning, clarity to reasoning players of the game, and standards of evidence. I've long thought this a plausible interpretation of religious language. The fatal flaw of this theory is that too many players of the religious-language game simply mean their talk to be truthful and to refer to real supernatural things, such as God, and to supernatural events, such as salvation. Yet there is something so compelling to me about the idea that if you dismiss religious language on the grounds that it does not refer to anything real, you miss something truthful about religious language.
I've made a little progress in figuring out what is missed. There are certain components of some religious experience that produce non-trivial truths. People who have had these experiences can discuss them with religious language. The discussions successfully convey truths. It may be that people who have not had these experiences can be caused to have them by hearing religious language. And all this can be so even if, as I think is the case, there is no God or other supernatural objects or events. I'll show you what I mean in the next few posts.
In this connection, I fondly remember Professor Ramon Lemos (1928-2006). He took every opportunity to point out that there are truths that you can "just see," self-evident truths, his favorite example being the truth that orange is more like red than it is like green.
UPDATE: By odd coincidence, Bill Vallicella posted on this topic today. Go read the whole thing.