The Voynich Ninja

Full Version: Extension to the Currier languages
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2
Interesting blog post. I want to add some remarks since the statistics for the Voynich text should be interpreted in the context of what is characteristic for the Voynich text.

It is known that the Voynich text isn't homogenous since glyph combinations or bigrams occur with different frequencies in different parts of the manuscript [see You are not allowed to view links. Register or Login to view., p. 3]. For instance, the glyph combination EVA-ed is almost nonexistent in Currier A, or the glyph combination EVA-qo is extremely rare for labels and also for the f1r, f1v, and f8v. Also the glyph combination EVA-cth is rare or missing on some folios (f39r, f39v, f40r, f40v, f105v, and f106r). In the same way also the glyph combination EVA-ckh is rare or missing on some folios (see f19v, f20r, f22v, f23r, f114r). Even the bigram EVA-ai is rare or missing on some folios (f27v and f90r1). Regardless of the glyph combination chosen, it is always possible to find folios where it is rare or missing.

This means that changes in bigram frequencies are quite normal for the Voynich text. Therefore other criteria would allow to partition the Voynich text further. For instance EVA-m could be used to distinguish between Currier-A folios with EVA-m (bifolio 3/6, bifolio 17/24 ...) and Currier-A folios without EVA-m (bifolio 2/7, bifolio 10/15, bifolio 20/21, f25r/f25v, ...)  Or to distinguish between folios with EVA-m in line final position (Currier B) and folios without a preference for a certain line position (Currier-A). 

On the other side the structure of the text remains the same, paragraphs usually start with a gallow glyph, the line works as a functional entity, etc. Another observation is that word types in the VMS belong to a single network: see Timm & Schinner 2019. The existence of a single network allows the conclusion that the whole Voynich text is the result of a single system. In other words, the system and the structure of the text is always the same, even if the glyph combinations vary. Therefore it is interesting to note that it is indeed possible to group the folios according to there illustrations since they often share similar bigram statistics (This is not true for Herbal folios since it necessary to distinguish between Herbal A and Herbal B).

A possible explanation for this observation is that the scribe wrote the text on all folios sharing the same type of illustrations after each other. If this is true it is possible to use the statistics to reveal the original order for the sections in the Voynich text: "Now, reordering the sections with respect to the frequency of token <chedy> replaces the seemingly irregular mixture of two separate languages by the gradual evolution of a single system from 'state A' to 'state B'. Since words typical for Currier A also exist in Currier B, but not the other way round, it is reasonable to assume that the order shown in Table 2 indeed represents the original sequence in which the sections of the VMS had been created." [Timm & Schinner 2019, p. 6].


It is therefore a valuable result that the outcome of the experiment described in Renés blogpost confirms the research results we describe in our paper [see Timm & Schinner 2019].
(25-04-2024, 10:37 PM)Torsten Wrote: You are not allowed to view links. Register or Login to view.Interesting blog post. I want to add some remarks since the statistics for the Voynich text should be interpreted in the context of what is characteristic for the Voynich text.

[...]

It is therefore a valuable result that the outcome of the experiment described in Renés blogpost confirms the research results we describe in our paper [see Timm & Schinner 2019].

The good thing about these statistics is that they are objective. They can be computed by different people independently, and come out the same. That works as long as we are sure that they have been computed in the same way.

In any case, I have no issue at all with the statistics produced by you (Torsten) and consider them valid and valuable.

There are variations at the lower levels (bifolios inside groups with similar illustrations) and at the higher levels (valid for all folios with similar illustrations). Other variations transcend the illustrations, meaning that they are clearly observable, but are not related to a particular secion of the MS.

There are simply too many to list, and also, many of them are too subtle to use as criterium for anything.

For me, the explanation is open. I also think that there is one single system, which allows for some variation.
But that is a hypothesis for which I do not yet have any evidence.
(25-04-2024, 06:05 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.However, what I will need to check, and cannot do now, is if this only affects the word counts, or also
the numbers in the other columns. This is because these come from different scripts.

So..... it could be resolved after all.

This analysis has gone through several iterations, and I only created the web page at the end.
Fortunately, I took notes, which are on the laptop I have with me, and it turns out that the creation of the groups was not as I remembered when I made the web page.

Indeed, these notes clearly show that group A does not include the 10 herbal pages in quires 15 and 17. These had been added to group P. I already knew that their text statistics better fit with the pharmaceutical pages, so evidently that led to this decision.

I have updated the page to reflect this.
(26-04-2024, 12:23 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.In any case, I have no issue at all with the statistics produced by you (Torsten) and consider them valid and valuable.

Where did you discuss or just give a reference to any valuable statistic of mine?

(26-04-2024, 12:23 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.There are variations at the lower levels (bifolios inside groups with similar illustrations) and at the higher levels (valid for all folios with similar illustrations). Other variations transcend the illustrations, meaning that they are clearly observable, but are not related to a particular secion of the MS.

Keep in mind that the most obvious change between Herbal A and Herbal B happens for the same type of illustration.

(26-04-2024, 12:23 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.For me, the explanation is open. I also think that there is one single system, which allows for some variation.
But that is a hypothesis for which I do not yet have any evidence.

The following table lists the four most frequent 'ch/sh'-words for different sections (see You are not allowed to view links. Register or Login to view.):
Herbal A     chol   chor   shol   sho
Pharma A     chol   cheol  chor   cheor
Astro        shey   chol   cheo   chy
Cosmo        chedy  chey   chol   chdy
Herbal B     chedy  chdy   shedy  chkchy
Stars B      chedy  chey   shedy  shey
Biological B shedy  chedy  shey   chey

The table demonstrates a shift from 'chol/chor' via 'cheol/cheor', 'cheo/sheo', 'chey/shey' to 'chedy/shedy'. This shift demonstrates a "gradual evolution of a single system from state A to state B" (Timm & Schinner 2019, p. 7). In your blog post, there is a similar graph illustrating the relationships between various 'dialects'. It seems that our viewpoints are not that different.
Since I was quite unhappy about my initial classification of the statistics of words beginning with Eva-q, I did an update yesterday (30 April), in which the majority of pages in the MS are now classified as 'neutral' and there are only the post-fixes '+' and '-' for pages with relatively many, or relatively few cases of such words.

This has also been reflected in the page descriptions.

Furthermore, I have started adding some statistics to the page: You are not allowed to view links. Register or Login to view.
but that is still on-going.
For what it's worth, I used your most recent assignments and compared them to our plant styles. I used:
  • α for the "alpha style" mostly associated with Currier A. 
  • β for the "beta style" of Currier B
  • 0 for "unmarked" drawings that have no alpha or beta properties
  • "conflict" for plants that have both alpha and beta properties

Some observations:
  • Half of the pages you gave "+" have a problematic image (either conflict or 0)
  • Half of the pages you marked "C" have a problematic image
  • None of your "Ae" pages have a problematic image
  • None of your "A-"pages have a problematic image
  • Just one "Bs" page has an "unmarked" image, the rest is fine

In other words, "Ae, A-, Bs" pages behave exceptionally well. "C" and "+" pages spell trouble.
Thanks!

Do I understand correctly that you used the individual page classificiations?
I had not yet made that comparison.
As you know, I made similar comparisons on the bifolio level, which shows quite a consistent
result, along similar lines.

Further intriguing evidence, that is still looking for a good explanation.
Yes, each individual page. I am not certain yet if bifolio level is the way to go. It really depends if it - whatever is going on - is primarily sheet based or order based. When bifolios are nested, there are many plants between the first plant of the quire and the last, even though these two plants exist on the same side of a sheet of vellum.

The best approach may be to use folio level, so the recto plus verso, since these are both consecutive and on the same physical sheet. Indeed, plant drawings on the recto and verso of the same folio are often similar in our system. So I may try averaging them out and comparing this to your folio data.
It looks similar when I use folios (r+v), though perhaps a bit clearer due to condensation.

  • Folios you assign "-" behave well
  • Both folios you assign "+" (41 and 51) are unmarked (0)
  • Ae behaves well.
  • Bs behaves well
  • most conflicts are on C pages


One thing the current list shows that I hadn't noticed before is that pages you mark "Bs" perform much better for our β properties than pages you mark B. This might tell us that "Bs" is a meaningful category, showing a further distance from Herbal A.

Code:
Folio RZ GR_style
1 A- α-
2 A α
3 A α-
4 A α
5 A α
6 A α
7 A α
8 A- α
9 A α
10 A α
11 A α
13 A α
14 A α
15 A α
16 A α
17 A α
18 A conflict
19 A α
20 A α
21 A α
22 A α
23 A α
24 A α-
25 A α
26 B 0
27 A α
28 A α
29 A α
30 A α-
31 Bs β
32 A α
33 C 0
34 B β
35 A α-
36 A α
37 A α
38 A α
39 C conflict
40 C conflict
41 B+ 0
42 A α
43 Bs β
44 A α
45 A conflict
46 Bs β
47 A- α
48 B β-
49 A α
50 Bs 0
51 A+ 0
52 A α-
53 A 0
54 A α
55 C conflict
56 A α-
57 C- α
65 C 0
66 B α
87 Ae α
90 Ae α
93 A α-
94 B conflict
95 C β-
96 Ae α
The trigram olk has 543 mathes You are not allowed to view links. Register or Login to view.
Only 12 matches in folios written by the scribe 1:
f51v     olkeeody
f57v     olkchdal
f88r     olkeedy
f89r2    cholkeedy  olkey
f89v1   olkeeycthy
f93r     olkees
f99v     olkeol
f100v   qoolkeey
f102r2  opolkod
f102v1  olkey
f102v2  otolky
Pages: 1 2