[split] Verbose cipher?

[kckluge]

[...]Rene suggested that the hypothetical verbose cipher should then be homophonic (or, I add, include nulls). Maybe it is possible that the scribe, having different options to encode the most frequent characters and/or being able to add nulls, could make his choice in such a way that repeated characters are avoided, solving the problem pointed out by kckluge. This scenario seems to imply that the encoder had a conscious or unconscious idea of how Voynich words should look like and tried to stick to it (something comparable with Timm and Schinner's theory?).

Another issue with Geoffrey's list is that decoding is not deterministic: e.g. [qok] can encode a single character, or two characters as [qo][k]. Same for [qot] vs [qo][t].
Also, treating '[e]' as encoding a plain-text character is not compatible with the many occurrences of 'eee'.

Yes, in principle homophones and nulls could address this.

Not sure we should assume the creator(s) of the system were as averse to needing to do limited backtracking when deciphering as we might be. The lack of an underlying unique parse might be part of our problems if this is a verbose cipher.

As for EVA 'e'...looking at contexts where you have an 'e' not followed by 'e', 'd' or 'y' is an interesting exercise. 'k', 't', 'ch', and 'sh' is the preceding glyph ~90% of the time. Which also helps highlight the prominent role 'k', 't', 'ch', and 'sh' play in multiple phenomena relating to the structure of "words":

* as a big chunk by frequency of the small number of glyphs that can follow 'y' inside a word

* as an optional separator between qo[ktlr] prefixes and e[edy]-based suffixes