Voynichese is a numeric cipher?

[ololololo]

Well, let's conduct a small experiment. If Voynichese is a numerical substitution, it means that letters or parts of words are hidden under numbers. In general, we can take words with a more or less obvious translation and try to "decrypt" them by matching them as closely as possible with the original word.
If you don't understand, I'll try to translate it. 
For example, we will take words whose translations are well-known and obvious: otol (fire, f77r) and okalo (intestine, f77v).
Let's see if we can decipher and match them in three languages: Latin, Middle High German, and Bavarian (the most common variations of the language used in the manuscript, the idea of Bavarian I borrowed from JoJo_Jost).
1). Let's start with otol. As we have already learned, this means fire. To begin deciphering, we need to break this word into "syllables." In fact, there are several ways to divide it:
ot-ol 
o-t-o-l

o-t-ol
In Latin, fire is Ignis, in MHG it's viur/fieur, and in Bavarian it's faia. 
If the author has encrypted Ignis, it is likely that he has encrypted it as ot-ol = i-g-nis (or i-gn-is).  Of course, it looks contrived, but as it was written in the post, the author could encrypt not only letters, but also abbreviations (for example, he did not encrypt the ending -rum with three letters, but with one symbol).
But what if the author has encrypted Viur/Fieur? Well, for one thing, there are no matches between the words - word "viur" has two different vowels. We can assume that this stands for o-t-o-l = v-i-u-r or o-t-ol = v-i-ur. But then it turns out that o is ambiguous - at the beginning it means v, and at the end it means u. In fact, this is possible because u was often written as v in the past, and this is a fairly common encryption technique. Then it turns out that o-t-o-l = v-i-v-r or o-t-ol = v-i-vr. It more or less fits... I couldn't find a corresponding word for "fieur," and it doesn't seem to fit either, as it has five letters, while otol has a maximum of four.
What if the author has encrypted the Bavarian Faia? Well, first of all, I want to point out that although this word corresponds to otol (two identical vowels, two different consonants), it does not match the letter arrangement. But if we write the letters in reverse - aiaf, then we get o-t-o-l = a-i-a-f. Of course, it fits, but we had to rewrite the word.
2). The same applies to okalo. Here are the syllable breakdown options:
o-k-a-l-o
ok-al-o
o-k-al-o
In Latin, the intestines are called intestinum, in MHG it's darm/darme, and in Bavarian it's doam.
So, intestinum. Let's start by comparing the words - they have three vowels, two of which are i (which is encoded as o in the previous example). But we see a big difference in the number of letters. By trying to match, we get: o-k-al-o = i-n-testi-num. So, is the final o = num? However, this is not so straightforward. Since this is Latin, it may be a shortened version of the word - ītēstī̄ū, ītēstī̄m or intestī'. The last example is suitable for Latin, and then it turns out that o-k-al-o = i-n-test-i. The only questionable detail is the al digram. But perhaps the digrams could mean not only letters, but also abbreviations. 
What about darm/darme? There is a problem - in the previous example, o is u/v, but there are no such letters here. Also, the number of letters in darm is less than in okalo. Still, if try to decipher, we'll get ok-al-o = d-ar-m or ok-al-o = da-rm-e. As we can see, no form of the word "intestine" is suitable for MHG.
Finally, doam. The number of letters in doam is less than in okalo. The word contains the letter o, which in the previous example represented a. But in doam, she's the third, and in okalo, she's the first. We can try to match the words, and okalo will become kaol or laok. Bringing it into line, we get k-a-o-l = d-o-a-m or k-a-ol = d-o-am. Considering my assumption that the order of the numbers in the word could have been intentionally changed, and that extra digits could have been added, this decryption option makes sense. 
So, we were able to decipher these words somehow. The most likely candidates for the original language of the manuscript in this case were Latin (which behaves quite appropriately in our examples) and Bavarian (with some nuances, but they are perfectly explained by my assumptions about the secondary complexity of the cipher). Unfortunately, MHG is not suitable.

P.S. This is not a decryption, but an attempt to apply substitution logic to Voynichese. In general, this worked more or less well with these simple examples, but that doesn't mean you can continue translating the text - it won't work.