Discussion of "A possible generating algorithm of the Voynich manuscript"

[Torsten]

I am quite sure that a procedural method can reproduce any statistical feature of Voynichese (or of any other text).

The self-citation method can only reproduce statistical features similar to Voynichese (see Timm & Schinner 2019, p. 15f).

I don't understand why self-citation should cause the initial glyph of one word to depend on the final glyph of the previous word.

The previous word is the last written word. To copy an element of the previous word is therefore an obvious choice. The only problem is that this choice is indeed obvious. The shape of <q> is quite similar to the shape of <y>. See for instance the difference between <y> and <q> in <yoky> and <qoty> on page You are not allowed to view links. Register or Login to view..

It is possible to demonstrate that indeed an element of the previous word was copied. Consecutive tokens start more often than expected with the same initial glyphs or end with the same final glyphs. See table 1. "Initial-Initial dependency" and table 3 "Final-Final dependency" in the Analysis of word dependency by Vladimir Sazonov: "As it is clear from the tables 1 and 3, the words very often reproduce the same form (the same initial or the same final) in the next word. For example, -aiin or qo- or -dy can repeat many times with good density" (You are not allowed to view links. Register or Login to view.).

Also, please correct me if I am wrong, but I think the generated text here doesn't feature any clear dependency between the last glyph of a word and the first glyph of the next word?

"Keep in mind that the VMS was created by a human writer who had complete freedom to vary some details of the generating algorithm on the spur of a moment. ..." (Timm & Schinner 2019, p. 16). With other words we didn't argue that the text was created by a computer program and we didn't argue that our program is able to simulate the complexity of a human mind.

Beside line breaks numerous similar rules for describing some kind of local repetition exists. Even if I would try to add each rule it would still be possible to point to an another similar rule. At the same time each additional rule would increase the complexity of the program and it would be argued that the program is too sophisticated. Therefore our goal was to keep the algorithm as simple as possible.

Something I particularly like about your approach is that you take reduplication and quasi-reduplication seriously. You have created a system that nicely models them.

I didn't create the self-citation method. In my paper from 2015 I try to reverse engineer the way the text of the VMS was written. See chapter 6 "Evidence" (Timm 2015, p. 12f) and chapter 7 "The text generation method" (Timm 2015, p. 14f). This means the self-citation method is a way to describe Voynichese (see also Timm & Schinner 2019, p. 9: "We start with focusing on the set of rules most likely used in the VMS. ...").

With other words the self-citation method is the most simple way to explain the observed statistical features. Since every word is at the same time the result of the copying process and also source for generating new words it is necessary that every Voynich-word is n-th copy of any other Voynich-word. The only possible outcome is therefore a network of similar words. 

One result of the self-citation method is that a more frequently generated word is copied more frequently. This means that a frequently generated word results in many similar words whereas each similar token also increases the chance that the original word is generated again.

This recursive process results in the observation that "high-frequency tokens also tend to have high numbers of similar words" (Timm & Schinner 2019, p. 6) and explains that similar tokens occur with similar frequencies. It is therefore necessary that a text generated by the self-citation method fulfills both of Zipf’s laws. 

Since similar words have similar length it is also necessary that the network of similar words results in a "binomial-like" word length distribution.

Because it is handy to copy a word from the same position some lines above (see Timm 2015, p. 18) the self-citation method results in long-range correlations like the correlations found for the Voynich-text (see Timm & Schinner 2019, p. 14).