26-04-2022, 08:32 PM
(26-04-2022, 05:40 PM)davidjackson Wrote: You are not allowed to view links. Register or Login to view.I would suggest, that in your tea example, the Voynich is drinking coffee
I'm more of a coffee guy anyways.
Searcher: this is a good question. If the same words are repeated over and over, this could cause certain glyph combinations (which are common in those words) to become prevalent. One thing I can say is that over large windows of text, for example the whole of Herbal A or Q20, Voynichese's type-token ratio (TTR) is normal. This means that the variation in its vocabulary is well within expected bounds.
We can actually test the impact of duplicate words: take a VM text, remove all duplicate words so that one token per type remains, and compare entropy values before and after. (This would give it better odds than normal, a TTR of 1 is unusual).
I did this quickly with unmodified EVA, that's all I have time for right now. The results are stronger than I expected - though do keep in mind that this is an unusual situation where each token is a new type. And it is still below the "magical limit" of h2=3.
[attachment=6461]