Writing monolingual dictionaries is typically something that would only become possible after the Renaissance, when people started viewing one version of the language as the true version, something to be catalogued. In the Middle Ages, the rule was generally that written language was a reflection of the way people in a certain region spoke. (Text was usually meant to be spoken as well, so writing and speaking went hand in hand).
The same goes for thinking about grammar. Latin had rules, but this was not a concern for spoken languages. The first English grammar only appears by the end of the 16th century, see here: You are not allowed to view links.
Register or
Login to view.
In a sense, a consciously ungrammatical text in medieval vernacular would be anachronistic, because they did not think about the vernacular's grammar. It was the way you spoke and that was that.
(Note: some of this may be different for Italian, which I don't know much about. They tended to be earlier than the rest in their thinking about language).
So to get to the point, this is what I think about various scenarios. Keep in mind that the VM text is large, so the phenomenon at play must be used extensively, not just for a few sentences or a single poem.
* A language experiment with faulty grammar: almost impossible, anachronistic.
* Someone who is not a native speaker struggling to write the language: possible, though it would be the kind of thing we usually see in just a few words, in marginalia etc. We also see it in medieval Latin, where authors' mastery of the dead language varies. But even the worst Latin writers do much better than Voynich translations
* Poetic liberties: this is certainly possible, authors may write weird things because of rhyme, rhythm and metre. But these oddities would be limited, and the bulk of the text would still be grammatically sound.
* A weird dialect. This is possible, but in this case the person who proposes the VM solution should explain how this particular dialect explains the phenomena observed. What we usually see is that VM translators use dictionary forms, and the result is a grammatical mess.
And that is what it usually boils down to. The
translators tend to disregard the fact that languages use grammar. This may be because in the case of English speakers, they don't feel the importance of cases in languages like Latin. Or because they focus so much on dictionary forms that they forget all inflection and conjugation. It feels like a bad excuse to then transfer this lack of concern for grammar over to the VM makers.