I don't disagree with your conclusion at all, but if I err, I err on the side of correspondence, because in my perception language tends to decohere: hence (I think) paradox, or meaninglessness.

I have been of the opinion that the observable and measurable set the boundaries of any coherence we can deduce: excepting perhaps in systems which are completely abstract, and which bear little or no relation to the measurable. In the world which is all that is the case, I think the 'why' must agree with 'best possible evidence' rather than making the evidence conform to the 'why'.