Sequential word repetitions in the VMS

[Koen G]

Marco: I mean that it's hard to tell whether substitution is purely typographical (my initial impression) or whether it reflects actual phonetic changes like Emma argues. I think that there are arguments in favor of both, but problems with both as well. 

The first option is not limited to capitals but might also include flourishes or variants of the same letter, as is often the case in cursive scripts. For example, EVA a and y know complementary distribution (based on their position). Like this, one could see "y" like "a" with a flourish.

I'd also not be the first one to remark that characters added at the beginning of a line might be similar to paragraph markers, making them again related to typography rather than representing a new sound. Normal letters might double as paragraph markers. Punctuation is a similar option of course, though I'm not sure how this would work.

But as you remark, one problem is that the glyph set would become extremely small, making it almost impossible that Voynichese is the result of simple substitution.
A problem with Emma's proposal is, in my opinion, that certain sounds of the language mostly emerge as sandhi effects, making them rare within the core phoneme inventory of the language. The effect of this would be a similar limitation of the actual phonemes of the language, by making some sounds either rare or limited to sandhi effects.