I wish to update my definition of meaning again. Here goes.
If P has meaning, then we can find P1 and P2 such that
P => P1, where P1 is distinct from P
P => ~P2, where P2 is distinct from both ~P and ~P1.
I was shown by my colleage, leibniz, that my previous definition was plagued by the possibility that P1 was a function of P, and that did very bad things.
This revised definition says that for P to have meaning, P must be verifiable and falsifiable. My previous arguments show that, if P is meaningful, then P must be either mathematical or empirical.