Which plaintext languages to try?

[oshfdk]

There is no chance that all those important double consonants like -rr-, -ff-, -nn-, -mm-, -tt-, -ss- etc., which are crucial for Latin and all it's derived languages, are contained somewhere in the VMS by a "clever encoding", "magical cipher" or such.

All tries with Latin (and there were a lot) failed, due the incompatibility of VMS structure, spelling, words with pure replacements AND with all decoding principles.
This includes all Romanic languages, Romanian, German and West-slavic languages as well.

I suspect that the Voynich manuscript is not compatible with a normal manuscript plaintext in any language at all. No matter what the language is, as long as the corpus is large enough there ought to be regularities in the form of repeated expressions with some systematic changes. The easiest way to look for them is looking for what I call the longest repeats with a gap or gaps, patterns like ABC and ADC with sufficiently long A and C and short B and D. Like those I found in a Chinese MS in this post: You are not allowed to view links. Register or Login to view.

I believe that these would be present in a large enough book written in a natural language, no matter what language, topic or script. A pattern of A*C where A and C are a few words long and the central part is short and takes multiple values are expected to repeat tens of times in a manuscript of the size of the Voynich MS. And they do in books of comparable size.

There are no patterns of this form in the Voynich MS that would be long enough or repeat more than 2-3 times. Hence, if Voynich MS has some meaningful plaintext underneath, it's likely this plaintext has been randomized in a way that removes this structure. This randomization is possible, and even expected, when using a one-to-many cipher, that has several possible encodings for one plaintext. The general idea of one-to-many ciphers was certainly known at the time when the Voynich MS was created, most (all?) of the ciphers in Francesco Tranchedino's list are of this kind.

Unlike these ciphers, the Voynich MS doesn't appear to be a simple one-to-many substitution, but my bet is still on it being a "clever encoding" and not a natural language plaintext. And if it is a clever encoding, then I suppose Latin should be the first candidate. It's possible the plaintext is actually Slavic or Tatar or Chinese or some exotic forgotten language. But it's quite unlikely that  Voynichese, as it is written, is just a simple transcription of that language.