Vords are too short.

[Koen G]

I took EVA to have a working basis for measurement in the first place. Of course, this results in the imponderables described by you. However, one would have to find a consensus in the interpretation of the glyphs, which I imagine to be quite difficult.

Yeah, this is really more of a worry about our inability to cope with this problem rather than a direct criticism. EVA offers one option (even though it wasn't meant as one) so we use it because there is no consensus for anything else.

When it comes to statistics like character entropy and word length though, I think EVA sits on the "low entropy" and "lengthy" side of the spectrum. One solution would be to create two "master files": one in EVA (which parses as separate strokes in case of doubt) and one which maximally combines strokes into glyphs. One could then say "average word length in Voynichese is between x and y, depending on how you like to parse".


Edit: just to be clear, if we combine some of EVA's characters into glyphs, Voynichese still has really low entropy for a language. But EVA makes it considerably lower. I don't know the impact on word length and how it compares to real languages.