Ungrammatical repetition of the same few words

[Koen G]

I think I understand what you mean. So it would be like a nonsense generating method, but the scribes would have actual words from their language in mind while writing?

The problem there is that to get something as restricted as Voynichese, those words would have to be selected and planned very carefully. And then they would also include minor variations of these words to get things like [dain, daiin] etc. 

So if I understand correctly, what you're asking, it's basically whether the makers had real words in mind while they were generating the text. Within a hypothetical framework of a meaningless text, I think it's almost unavoidable that the writers attached some phonetic values to the words, kind of like we do with EVA. But this is impossible to prove, or even test, I think.