For languages which decide to use these recommended subtypes, I think the guidelines need clarification.

On the flat page, the discussion of examples like Ludwig van Beethoven says that any internal compositional structure should be analyzed only if annotating the language in question. So in a German corpus, "van" would attach to "Beethoven" as case (it forms a PP), but in an English corpus the whole expression would simply be flat.

So how would subtypes apply here? flat:foreign flatly (ha) says that foreign names are not included. flat:name gives examples of compositional internal structure in English and Italian, without specifying the language of the corpus. Should it be revised to say that foreign syntactic constructions (prepositions, etc.) within a name should be given a flat:name analysis rather than a compositional analysis?