What if it's a "Lossy Encoder" for 15th-Century Latin?

[Pointless..]

Well, then it’s possible... But even such a system would be much easier to implement. You can simply use substitution so as not to get bogged down. In any case, even if the system is transparent and understandable to a legitimate decrypter, it would be foolish to rely on their ingenuity...
Perhaps your idea will become a starting point in the field of research into Friedman’s artificial language. It fits perfectly into the logic of philosophical language. For example, for each such word, you can add prefixes such as d-, ch-, or qo-, which will make it easier to identify missing words.

Thank you for your comment. This is exactly what I am looking for here: an exchange of ideas, discussion of different possibilities, and perspectives I may not have considered.

My current thinking is that the manuscript is not language as such. It may contain information derived from a language but encoded in a way that removes or obscures many of its linguistic features.

Of course, I may be wrong about some of the premises or methods, but I hope the idea can at least lead to a fruitful discussion and perhaps open up new lines of research.