Truth by convention

I stipulate that "It xyzzies" is true. Clearly I have failed to make "It xyzzies" meaningful. My stipulation is compatible with "It xyzzies" meaning that 2+2=4, but also with its meaning that everything is round or non-round. So stipulating a sentence to be true isn't going to be sufficient to introduce the sentence into our language. But if I say anything more about the meaning of the sentence, I risk that no true sentence might fit what I say, and we lose the point of truth by convention. Nor does it at all help to stipulate a family of sentences at once, e.g., stipulating that whenever S and T are true, so is "S*T", and when "S*T" is true, S is true, and when "S*T" is true, T is true. That still fails to introduce a connective "*", unless we say more about the meaning. The stipulation I gave is compatible with too many things. For instance, "S*T" could mean "God believes S and God believes T", or it could mean "(S or S) and (T and T)". And if I say more about which one I mean, I risk the stipulating being unsuccessful. So that's that for truth by convention. It's fun to drive nails in the coffins of dead theories.