Collaborative works


How should we deal with collaborative works? Variations on a Russian Folk Song was written by 10 different composers, each writing one variation. It’s obvious each variation should be credited to the specific composer who wrote it, but what should be used for the main work? (this, while still fairly rare, is a more common case than you might think).

Should we add all the artists as composers of the parent work? Leave no artist as composer? Or maybe link Various Artists as the composer?

Maybe a “partial” attribute for writing credits is in order?

Artist A wrote parts of Work B. Work B is partially written by artist A.

Could something like this apply as well for example to Mozart’s Requiem, or all these compositions that have been completed/finished by another?

For those, it’s normally two or three people at most — not a big deal to just list them all (often one as “additional”).

I think Various Artists is a workaround for things not being able to handle too many artists in a database field — that doesn’t seem to really apply here. Having 10 different relationships on a work isn’t a problem.

But maybe no artist makes sense. Arguably, there really isn’t any composition/authorship/creativity/etc. involved in the main work that isn’t in the sub-works. So then there really is no composer; there is merely a publisher (and maybe an editor).

Arguing against that is that relationships don’t really propagate along the hierarchy as expected. If you linked a recording to the main work, it’d show up as having no composers. That is not good.

So I guess in an ideal world, I’d suggest no composer, and that the composers of the linked parts would propagate up automatically and show up as 10 different composers of parts of the work. (Also, ideally, a composer could propagate down…) But with the software we have, I think it’d be best to link them all, as annoying as that is to enter.

I think completion (Mozart Requiem) & reconstruction (Stenhammar’s Piano Concerto 1 (score bombed lost in WW2, reconstructed by Atterberg from piano reduction and own memory, efter which a copy of original version was found)) of a Work not originally so intended should not qualify as joint composition, even in cases where original composer agrees to be joined at a later stage of the Work (i.e. Rosenberg Str4 0).

For reference, 2 more joint compositions intended as such: the F—A—E Sonata and the Requiem der Versöhnung.

I think that the top work should have no composer, unless there was some creative synthesis involved. Only the individual works were composed and these will get credited accordingly. To add all the composers or “various artists” just is just obfuscation. So I think F-A-E is wrongly credited, for example.

As much as I just said the same thing a few posts ago (but then concluded against it due to technical limitations)… I wonder what’d happen if we really applied this principal. Some things that come to mind:

  1. There are probably collections of works where a bunch of compositions are thrown together because, well, the publisher wants X songs for a solo vocal and piano because that sells. I sort of wonder if some of Brahms’s many “X Lieder, op. whatever" which have 5–12 individual songs in them may be from that (but haven’t researched it). If so, presumably those works should also have the composer struck (and Brahms left only on the sub-works). That makes entering the works correctly harder, as I now have to answer a very non-trivial question of how much this work is merely a random collection of independent songs vs. something that’s truly an integrated whole, like a symphony.

  2. Does it also apply to things like famous poets being credited as lyricist/librettist? In many cases its clear that poet provided no creative input (beyond the poem, which could potentially be linked as a based-on work) as, for example, the musical work was written well after the poet’s death.

If the site were able to propagate relationships through other relationships — e.g., composer up or down parts/part-of relationships — a lot of this wouldn’t be a real problem. You’d mark Brahms to propagate up from the individual songs, and if someone wants to do the research to determine if it’s actually a composition instead of a mere collection, they can add him to the parent work too. (I guess it’d be nice to have a way to mark the opposite as well).

(Honestly, the more useful part of propagating relationships would be the other way — it’d be nice to propagate a lot of information down from “Symphony no 1” to “Symphony no. 1: I. Allegro”. Currently we duplicate it, or not, seems to depend on the info and who is entering it.)

Concerning the poets as librettists: Many works take the theme of some other songs (for example the bach chorales) so should we make a base of folksongs to link all their instrumentations/adaptations to, or is it not worth it? I did it for a few lutherian chorales that were already present of MB, but should we do this on a larger scale?

FWIW, my Picard plugin (Classical Extras) does exactly this. It will get any composers up the tree (optionally including collection works). So having composers at the top of the tree whose only involvement was in another branch is not helpful. Lyricists are dealt with slightly differently - as soon as a lyricist is found, none are looked for at higher levels. The reason for the difference is to allow composers of original works to be included where the current work is an arrangement.


That’s one problem with it not being a feature in MB itself, there is no way to mark which of the relationships should propagate up/down/both/neither. So you have to guess.