OK. In choral/vocal works of the later Middle Ages and early Renaissance, such as motets, very often the person who wrote the work down does not provide a textual underlay for all the parts. I've looked through just about every reference material I've been able to lay my hands on, and these possibilities sort themselves out of the material:
1) Untexted parts were intended to be performed instrumentally (Bukofzer)
2) Untexted parts should not be vocalised to (Bukofzer)
3) Untexted parts may be freely vocalised to, even to the point of
supplying text, since that would have been current practice (Bent, Bent and Trowell)
4) Free vocalisation should consist of the prevailing vowel from the texted part(s) (Oxford History of Music)
5) Free vocalisation should consist of the texted parts where an obvious metrical correspondence lies between a part with underlay and one without (Various refs)
6) As 5), but with instrumentation supplied for the noncorresponding notes as in 1)
7) As 6), but with free vocalisation instead of instrumentation.
I'm resisting 1), since there's plenty of evidence that instruments were generally not allowed in church performance, which is what you'd expect for motets, but I'm happy with it for, say, the untexted first tenor in Dunstable's O rosa bella, which seems just fine as a chitarrone line. 3) sounds entirely reasonable, but I'm concerned that if I followed it strictly, I'd need a damn sight more musicological and liturgological knowledge (and material) than I have. I'm working with 4) at the moment, but it's not always easy to determine which is the "prevailing vowel".
Any ideas? know anyone who'd have any ideas, or have some idea what best current practice is?