[Article] The Voynich Manuscript: A Non-Stochastic Glossolalia

[MarcoP]

the h2 value of voynichese is entirely dependent on the transcription used, and some transcriptions even have a higher h2 than european languages. So the statistics of the text are only similar to glossolalia if you use Eva, which is unlikely to be an accurate transcription of the text's phonemes.

Of course it's true that conditional entropy depends on the transcription (or transliteration) system, but Rene's results (You are not allowed to view links. Register or Login to view.) do not confirm this statement. He found lower than 2.2 conditional entropy using 4 different transliteration methods (Currier-D'Imperio, First Study Group, EVA, CUVA); the value for European languages typically is >3.

Lindemann and Bowern (You are not allowed to view links. Register or Login to view., 2021) used 3 different transliterations (one of which is EVA) and found results in the 2.1-2.5 range.

---

About Fasano's paper, I agree with Michelle's observations. Also, the absence of an in-depth discussion of Gaskell and Bowern You are not allowed to view links. Register or Login to view. (2022) is a serious fault (actually, their paper is not even mentioned!).

Much of the argument is based on statistical figures (such as Zipf's law or conditional entropy) but no actual measures are discussed. Only short samples of glossolalia are presented, accompanied by notes like "the sample size is too small to be conclusive."

I find the subject very interesting and I hope to see more research about it in the future.