Posting to open-bibliography
Considering these variant preferred labels as "language forms" is not exactly the best way of looking at it. In AACR2, the form of, e.g. a name of a corporate body, is almost always based on the language of the body itself. Therefore, the heading for the Budapest Museum of Fine Arts is set up in Hungarian. We can see this at work very clearly in VIAF, which attempts to bring varying national forms together: http://viaf.org/viaf/130785279/ We also see that not everyone follows such a rule.
But when it comes to international bodies, variants really go crazy, since the AACR2 rule is to set it up in English when available. Naturally, this applies for each cultural group and each will set up names in its own languages. So, for the World Health Organization, we see this: http://viaf.org/viaf/135618969/ where there are all kinds of preferred forms.
It gets even crazier with other international corporate bodies, e.g. The Commonwealth of Independent States, http://viaf.org/viaf/131775197/ which is an international body with different official languages, none of which is English, but because of our rules, the form is set up in English anyway. Other cultures use different forms.
Concerning personal names, the rules become still more complex, but forms are remain based mainly on the form of the name of the author on the item(s) they publish. So, we have Tolstoi, Nikita Ilich http://viaf.org/viaf/9962566/ (Russian form), but for Tolstoy, Leo, graf, 1828-1910 http://viaf.org/viaf/96987389/ he is set up in an anglicized form because he is more famous, so other rules kick in.
In any case, this should show that "language form" is a bit dubious when discussing the preferred forms. Also, it shows how vital are the cross-references for a U.S. art student trying to find exhibition catalogs from Budapest. Still, just these few examples demonstrate how the VIAF has already become such a wonderful and incredibly useful tool. I hope that it may point the way forward, somehow.