Alias (name variant) for work titles

just a bit, at least we have a method that helps at the moment.

I wonder, why the original author should still be mentioned, as I stated in my message above, the relation is already in the database and as @AgathaCrustie mentioned, the translation was not written by the author. Every duplicate information is a potential risk of wrong information.
Couldn’t it be an option, that a work is either written or translated?

There’s been a lot of conceptual work on aggregates in the FRBR field - here’s a summary: https://doi.org/10.5860/lrts.59n3.120

A compilation of translated works would be a manifestation (seems like an edition here) containing several expressions (works on BB, since the overarching work-as-concept layer doesn’t exist here?), and with its own expression and work (separate from the contained expressions) that can contain information about the editors, etc. The issue of where to put the shared translation information looks like it arises because the work and expression layers have been merged into one here. Each translated expression should have its own title, and the work can collect the titles of the expressions for adding to its alias lists, or not as is needed.

One implementation question that occurs if you follow that model for compilations/anthologies, is whether an edition can contain multiple BB works? I’ve not done a lot of editing here to find out yet.

As an aside, I personally tend to think of expressions as “variants” or “versions” of a text, which can be published in many different forms. When people talk of a “third edition” of a textbook, they usually mean both a new manifestation (different cover, etc.) and a new expression (the revision of the contents). With fiction publication, often there are several versions of the text with multiple publications of each, so the separation of work, expression, and manifestation (with items as individual copies) becomes very useful for properly describing the metadata without duplication.