It is important to distinguish sociolinguistic "facts", i.e. standard languages, from structural considerations. As Svein Moennesland put it in a recent paper at ICCEES V, the answer depends on the criteria.