Thanks for that! If I’ve understood correctly, Campbell suggests that the formula is this (I’m over-simplifying, I’m sure): verse structures such as the virelai give a general, somewhat loose arrangement (“blank cheque” was the term quoted from another source) and would still necessarily have to be known to the singer/reader, but text really determines all else, and the combination of the two makes the result. So even though the results seem almost random to a modern ear, even to the point that one virelai could have a completely different rhythm, meter, and scansion from another, the formula itself is reliable. This makes sense to me.
Again, though, that’s if I understood correctly on the basis of a quick read.