To what extent should the translation respect the grammar?

[Juan_Sali]

Under the hypothesis: The whole VM is written in the same natural language with the same homophonic cipher system.
Comparing texts with similar contents, the differences between scribes will be the homophonic parts of the cipher, as if every scribe enciphered what he/she wrote using the homophonics at will.

I'm not completely sure about it being predominantly homophonic, but my "headcanon" is that there is a certain amount of flexibility in the system for the scribes, and that this would go a long way to explaining the scribal differences and also variety within the same scribe.  But there are other problems that would need to be considered with it being homophonic or indeed other mechanisms that would generate that same flexibility.

My starting hypothesis is a simple one in order to understand how a flexible system is used to create the differences. More elements can be added gradually: polyphonic, braquigraphy (I am thinking on latin), abreviations, and adding null letters to the plaintext. A mix of them can explain the caracteristics of the vochinese and the difficult of the deciphering.