Interestingly enough, Haskell may be flexible enough to implement this in the sense that we already have Core as the underlying language. All that would be needed is a mapping from those parsed tokens back to the same underlying representation of True and False.