Hi Torsten,
thank you for your comments.
Here are a few personal opinions, for what they are worth.
(22-06-2019, 07:38 PM)Torsten Wrote: You are not allowed to view links. Register or Login to view. (22-06-2019, 05:36 PM)Emma May Smith Wrote: You are not allowed to view links. Register or Login to view.I would be happy to listen to arguments why a different explanation is better. But we must explain word break combinations. They are real phenomena and any theory which doesn't account for them must be wrong. Even a tentative explanation is better than complete ignorance.
My answer is that it is just another example of local repetition. If the the self-citation method was used to generate the text it was only possible to copy an element of a word already written. Therefore it [is] expected that the second glyph in a combination depends on the first. There is no contradiction between your paper and the self-citation method.
I am quite sure that a procedural method can reproduce any statistical feature of Voynichese (or of any other text). Also, I believe that only a complete and satisfactory decoding / translation can disprove the idea that the text is meaningless: hence, I think it is possible that the text was procedurally generated with something like the method you describe.
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. Also, please correct me if I am wrong, but I think the generated text You are not allowed to view links.
Register or
Login to view. doesn't feature any clear dependency between the last glyph of a word and the first glyph of the next word? For instance, the most frequent combination -l.c- occurs 705 times, nearly identical to the expected 693.
Even if this is the case, I guess that, if you wanted to, you could soon push to github a new version of your software that exhibits this behaviour.
But the fact that biased word-break combinations can be observed in written texts in several languages seems significant to me. It is suggestive of the fact that Voynichese is phonetic (i.e. glyphs correspond to sounds). From the few examples we discuss in the paper, when biased combinations are less marked, they can be due to frequent word combinations, but when they are extensive, they are the result of words being altered by an immediately adjacent word.
Another feature that points to a phonetic explanation is that occurrences of the same character across word boundaries (e.g. -
d.d- -
l.l-) are avoided (this is mentioned in the paper, but see also You are not allowed to view links.
Register or
Login to view.). From what I understand, languages tend to avoid the consecutive repetition of the same sound: actually they tend to promote the highest possible difference between consecutive sounds. Again, the observation is compatible with basic phonetic principles.
Finally (also discussed in our paper) most of the combinations that are frequent across words are also frequent "inside" words. The two sets of data largely confirm the tentative classification in vowels-consonants discussed by Jacques Guy (as well as Rene's HMM analysis). VC / CV alternation is also preferred across word boundaries in written natural languages.
Even if all these points mean what Emma and I think they mean, and Voynichese is phonetic, it could still be meaningless, something like glossolalia. In your paper you mention aesthetic choices by the scribe as visual choices, but they could be phonetic choices instead: meaningless words that sound good, rather than meaningless words that look good.
But I prefer to think that they are meaningful words that sound good (i.e. that follow euphonic preferences, like vowel-sound alternation).
(22-06-2019, 07:38 PM)Torsten Wrote: You are not allowed to view links. Register or Login to view.We argue that "the shape of a glyph must be compatible with the shape of the previous one and is also influenced by its position within a word or a line." (Timm & Schinner 2019, p. 10). Did you note that one outcome of your Charts 4.1. and 4.2. is that similar glyphs are related to each other? How do you explain that <ch> behaves like <sh>, <a> like <y>, <r> like <s>, <m> like <n>, and <o> like <y>?
As you probably know, many European medieval scripts shared the two features you mention:
- "the shape of a glyph must be compatible with the shape of the previous one": 2-shaped 'r' is used after "round-letters" (e.g. "cornibus" "procellis"); v-shaped 'r' is used after straight letters (e.g. "terre" "aquarum")
- "the shape of a glyph is influenced by its position within a word or a line": round-s appears at the end of words ("procellis" "bellis"); long-s appears inside words ("Bestie" "pestilentia")
![[Image: bMS3b9M.png]](https://i.imgur.com/bMS3b9M.png)
(examples from You are not allowed to view links.
Register or
Login to view.)
In your paper you write that (in the VMS) "the scribe also used aesthetically motivated design rules for glyph selection". The examples above illustrate that most medieval scribes did that. Clearly, in the VMS these things could be much more pronounced, but here we are dealing with what seems to be a newly invented alphabet, so more freedom was possible.
You are not allowed to view links.
Register or
Login to view. mentioned above "the theory that [the VMS] was written by a European in a foreign language who decided to create their own phonetic alphabet." I believe there is plenty of evidence that the alphabet was invented by a European scribe, who used a repertoire of shapes that he was familiar with. In the context of a phonetic interpretation, the fact that similar glyphs behave similarly could be explained as a conscious choice by the scribe, who assigned similar glyphs to similar sounds.
Something similar happened in Giovanni Fontana's cipher alphabet (dating to the same period as the VMS): he deliberately used a uniform scheme to represent the five vowels. This fits particularly well with the similar behaviour and similar look of Voynich 'o','a','y'.
Emma has also discussed the possible You are not allowed to view links.
Register or
Login to view.: this could be a case of positional variants similar to the round-s / long-s discussed above.
Of course, my reply to your question can only be tentative, but the possibility that the Voynich alphabet was created on the basis of aesthetic and phonetic principles seems worth considering. I am sure that the idea is not new.
(22-06-2019, 07:38 PM)Torsten Wrote: You are not allowed to view links. Register or Login to view.See also Timm & Schinner 2019, p. 10: "It even happens that the first glyph of a word is suggested by the last glyph of the previous word (Currier 1976): “‘Words’ ending in the <y> sort of symbol, which is very frequent, are followed about four times as often by ‘words’ beginning with <qo>.” This feature explains why the prefix <qo> becomes more frequent at the same time as words similar to <chedy> are used more frequently."
I agree that the correlation between
qo- and
chedy looks significant. It can well be that the presence of
qo- is caused by that of
chedy. The nature of this relationship remains to be defined: it could be that it is a random, meaningless "suggestion" as you say, or it could be that there is an underlying linguistic phenomenon, or it could be an artefact of some kind of cipher which I am totally unable to imagine. The only way to understand more of what is happening is collecting more data.
Some other slightly OT thoughts:
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. Of course, the amount of reduplication is a major obstacle for all attempts to find meaning in the text, cipher systems included. Koen's experiments with Token-Type-Ratio (TTR) measures, as well as Nablator's work inspired by your line-distance / string-distance graphs, are suggestive of how greatly texts can vary also in the context of a single language: this gives me hope that reduplication rates similar to those in the VMS can exist somewhere, maybe in some weird text in some weird language. But this is just wishful thinking, for the time being. I don't think I have seen any other take on reduplication as well developed as what you did.
But my preference goes to Emma's efforts to look for more and more patterns. The cross and delight of your approach is that your research is more or less over: you have come to a model that you are satisfied with, you are "convinced that the hoax hypothesis can be the only correct interpretation of the VMS". My feeling is that there is much more to be discovered about Voynichese, so I am happy to keep searching, even if it is a slow and difficult process.
The kind of language variability I mentioned above also makes it difficult to search for close parallels for specific phenomena. Even word-break correlations can be represented differently in different sources written in one language: it is one of those areas in which properties of the writing system are entangled with properties of the language. Checking Arabic or Welsh is more difficult than running a piece of software on a single text: one risks to waste a lot of time in "trial and error" attempts that don't lead anywhere. I am confident that Emma's approach of accurately examining each small feature, trying to add more and more details, can only add to our knowledge of the manuscript. In the end, if all goes well, it should allow to identify candidate languages, and then it should be relatively easy to see if some meaning can be found or not.
