Completeness and axiomatizability in many-valued logic