The Voynich Ninja

Quote:I think that the images do present some hints of clustering, but the clusters tend to appear as vertical regions in the diagrams, while Torsten's colors tend to be distributed horizontally.
The orange 'i' region seems to be somehow separated from the others.
The region I numbered 3 seems to correspond to sh- and ch- words.

Hello MarcoP, 

the gap between 2 and 3 in your picture only exists since EVA is using two signs for writing "ch". If "ch" would be transcribed with "2" and "sh" with "3" such a gap would not exist.

The word types similar to "daiin" are indeed separated.

The colors are chosen for the suffixes of the word types. If you would use the prefixes "d", "ch", "qo" for coloring the graphs would be distributed vertically.  
In the following graph word types starting with "c" or "sh" words starting with "q" or "o" are purple, words starting with "d" are orange and words starting with "s" are red:

[attachment=1111]

Quote:I think you need comparative languages and large corpora of them. 

But I see no other comparative languages in your research.

Dear Davidsch,

 first of all, if you think that the VMS should be compared to a large corpora of languages you are free to do it.

In my eyes it is important to analyze the most characteristic features before you start to compare something new with something else. Otherwise you will probably find many matches for none characteristic features.

In my paper from 2016 You are not allowed to view links. Register or Login to view. I have compared the VMS with language. In this paper I come to the conclusion that: "context dependency alone is not enough to allow the conclusion that the patterns found must be the result of a genuine linguistic structure." ... "The words in the VMS build a network of words similar to each other." ... "In natural languages a word normally (cf. poems) is used because of its meaning and not because it is similar to a previously written one. The result that the words are arranged such that they co-occur with similar ones is therefore not compatible with a linguistic system."

Quote:Dear Davidsch,

 first of all, if you think that the VMS should be compared to a large corpora of languages you are free to do it.

thank you for your permission, now i finally can try that 

I agree with Torsten that the voynich is not a natural language.  Also many different word meanings can be constructed through the use of Gematria where a word shares the same number.  You guys should fear the use of Gematria here because an exact interpretation will always be speculative unless the paragraph that was translated by Gematria contains logic and follows a historical theme.  I point you to this which most of you are afraid to view at my dismay.

In this 1st paragraph of You are not allowed to view links. Register or Login to view. it follows history to the letter and I did find logic within the prefix's of the first two letters of voynich vords.  Please at-least look at it Torsten an all.

You are not allowed to view links. Register or Login to view.

Quote:It's also not completely clear to me how the only conclusion can be that the text is artificially generated. I understand why straightforward interpretation as a normal text in natural language is problematic, but I'm not certain why it should point towards artifical construction.

Hello Koen,

 in my eyes there are multiple observations pointing in this direction.

1) The words in the VMS build a network of words similar to each other. If you want to generate a network with at least 6837 similar words I wouldn't describe that as an easy task. Especially if you live in medieval times.

2) Even if all pages are slightly different they all have in common that they contain similar words. For instance page You are not allowed to view links. Register or Login to view. in You are not allowed to view links. Register or Login to view. contains many words with "on" instead of "an" and "oiin" instead of "aiin" even if such words are normally rare. In this way each page of the VMS contains a smaller sample of the graph you will get for the full manuscript.

3) As more frequent a word is as more similar words exists for it. This is true for each page and for the VMS as a whole.

4) Within a page similar words occur above each other and side by side (see for instance the words "oky", "oty", "qoky" and "qoty" on page You are not allowed to view links. Register or Login to view. in You are not allowed to view links. Register or Login to view.).

5) There is a shift from Currier A to Currier B  within the manuscript (see You are not allowed to view links. Register or Login to view.).

6) The text is influenced by its container, the page of the manuscript. There are features typical for a certain position in a line or within a page  (see You are not allowed to view links. Register or Login to view. p. 18ff and 26f).

7) Words not similar to other words like "okeokeokeody" and "okeeolkcheey" occur only once. Moreover even for them it is usually possible to split them into two or three more common words like "okeo keo keody" and "okeeol kcheey" (see You are not allowed to view links. Register or Login to view.).

8) Since the similar words in the VMS build a network the lack of corrections for the VMS is even more surprising.

It is necessary to explain all this features. My explanation for this features is that the connection between similar words exists because the text is a copy of itself. According to my auto copying hypothesis, the whole manuscript consists of permutations of the most commonly used words. By copying and modifying words already written it is possibly to generate a text with features similar to that of the Voynich Manuscript. To change glyph sequences which have already been written requires no additional tools. For someone in the early 15th century it would be possible to use such a method to generate the text of the Voynich Manuscript while writing. His motivation to use this method was that it is far more efficient to copy text then to thing of something new all the time.

---

Quote:In this 1st paragraph of You are not allowed to view links. Register or Login to view. it follows history to the letter and I did find logic within the prefix's of the first two letters of voynich vords.  Please at-least look at it Torsten an all.

Dear Stellar,

sorry, but you interpret the same words differently all the time. This way you can read everything you like into the VMS.

Torsten,

Which graphing method are you using in gephi?  I have a .cvs file for all Chaucer Middle English words from Canterbury Tales and I want to produce a graph like yours to see if there is any similarity.

@Torsten

I believe that you deny the use of words as numbers, because it is possible and in fact that indeed a language from the Voynich can be written this way.  It goes against your study which you say the VMS is meaningless.


I have yet to do proper statistics as you point out for the same vord number representing the same actual word.  However here lasse is used twice for the same vord and number.

Hi Torsten, the key problem with your approach is, still, that you've failed to definite 'similarity'. You say 'edit distance' with the meaning that one character is added, deleted, or changed. But not all edits are equal as many result in invalid words.

There is, as I pointed out in reply to your 2014 paper, an underlying logic of how words are constructed. Edits are typically valid only within this existing structure. So, for example, if you wanted to add [e] to the word [chol], the only valid result would be [cheol]. The words [echol], [choel], and [chole] simply don't exist.

You must explain where this structure comes from and why. You cannot, as you have sought to do in the past, wave it away as the creator's whim.

Torsten: thanks for your very clear and to the point answer. If I understand correctly, your hypothesis is that the text is meaningless, and was generated in a somewhat free-flowing manner? But the reason why we see certain patterns is that the maker imposed on himself a set of rules as guidelines while writing? Like procedural generation? There is likely no historical parallel, although that doesn't mean much - unique artefacts might require unique explanations.  

I have spent much more time studying the imagery than the text, so I cannot assess these statements adequately. I'd love to know what someone like Emma May Smith thinks of this, since she has studied similar aspects of Voynichese from a linguistics standpoint. Edit: while I was writing this monster post she replied already, yay.

I think much of the imagery would have been self explanatory to the intended audience (as discussed for example in the mnemonics threads which got wiped out in the forum blackout), and some of the imagery is extremely dense in content. This implies that it doesn't necessarily need a meaningful text to accompany it, many of the images can stand on their own.

That said, I still think the text contains meaning, and at least partially reflects spoken language. This seems especially true for labels. Full paragraphs are more problematic, in part due to some of the properties in your list.

To go over your points:

Quote:1) The words in the VMS build a network of words similar to each other. If you want to generate a network with at least 6837 similar words I wouldn't describe that as an easy task. Especially if you live in medieval times.
2) Even if all pages are slightly different they all have in common that they contain similar words. For instance page You are not allowed to view links. Register or Login to view. in You are not allowed to view links. Register or Login to view. contains many words with "on" instead of "an" and "oiin" instead of "aiin" even if such words are normally rare. In this way each page of the VMS contains a smaller sample of the graph you will get for the full manuscript.

I'm not sure if this can be used as an argument against meaningfulness, one might even say that it argues for it since there appear to be themes to certain sections and pages. My knowledge of different language falls short here, but there are languages other than Indo-European ones which approach the relevant statistics of Voynichese to a much greater extent.

There are also the more "floaty" (  ) aspects of natural language to take into account. Might there be a form of rhyme or verse to each page, as was often the case to aid memorization?

Or on the other end of the explanation spectrum, might we be looking at for example a system of numerals, a transcription of what was originally a large set of tables? This would explain high repetition on the one hand, and a possible difference from page to page on the other.

Quote:3) As more frequent a word is as more similar words exists for it. This is true for each page and for the VMS as a whole.

Isn't this the same in natural language? In English, "be" are both extremely frequent and short words. The number of words that are similar is quire large. He, we, bet... If English were written as phonetically as for example Spanish is, the amount of similar words would be much greater, including words like see, fee, bee, lee, she, bell... And many of these similar words are very frequent as well, like he, she, we.

Quote:4) Within a page similar words occur above each other and side by side (see for instance the words "oky", "oty", "qoky" and "qoty" on page You are not allowed to view links. Register or Login to view. in You are not allowed to view links. Register or Login to view.).

This might be explained in many ways. They may be lists of dialectal variations of plant names, for example, or names in related languages. Or similar numerals...

Quote:5) There is a shift from Currier A to Currier B  within the manuscript (see You are not allowed to view links. Register or Login to view.).

This is definitely something to take into account, but it seems equally problematic for all explanations. Someone defending the natural language hypothesis might say that they are literally different languages or different dialects of the same language.

Quote:6) The text is influenced by its container, the page of the manuscript. There are features typical for a certain position in a line or within a page  (see You are not allowed to view links. Register or Login to view. p. 18ff and 26f).

This is again a problem in and of itself. It must be taken into account by all theories. I like Emma May Smith's recent posts about LAAFU, but I understand that she is also still struggling to fully grasp the meaning of these phenomena. It's just hard to explain from any angle. If the aim was to generate text quickly, then surely they wouldn't bother with things like this.

Quote:7) Words not similar to other words like "okeokeokeody" and "okeeolkcheey" occur only once. Moreover even for them it is usually possible to split them into two or three more common words like "okeo keo keody" and "okeeolkcheey" (see You are not allowed to view links. Register or Login to view.).

My pet theory, for which I have no proof, is that Voynichese is some kind of pidgin, a "language" that emerges when people without a common language have to interact with each other on a somewhat regular basis, usually trade. The vocabulary will be limited, generally a simplified form of the dominant language with relevant vocabulary from others. The amount of sounds is simplified. And there's often repetition, for example to express plurals. This can still be seen in a language like Afrikaans today. I'd even go as far as to argue that the script may be a "pidgin script" as well, but that's outside if the scope of this thread.

Alternatively of course, if it's all numbers like coordinates, dates... occurrences like these can also be explained.

Quote:8) Since the similar words in the VMS build a network the lack of corrections for the VMS is even more surprising.

We're not as certain about the lack of corrections as we once were. People like Wladimir have shown that there's been quite some messing around

Quote:There is, as I pointed out in reply to your 2014 paper, an underlying logic of how words are constructed. Edits are typically valid only within this existing structure. So, for example, if you wanted to add [e] to the word [chol], the only valid result would be [cheol]. The words [echol], [choel], and [chole] simply don't exist.

You must explain where this structure comes from and why. You cannot, as you have sought to do in the past, wave it away as the creator's whim.

Hello Emma, 

New characters where usually added as prefixes. Therefore the chance for a word "echol" is much higher then for "chole". Words starting with "e" exists for the VMS. There are for instance the words "echkolal" and "echedy". The point is that both occur only once. This happens as a result of the copying process. If a word with a new feature is copied this results in another word using the same feature and more important also this word could be copied. Moreover if a feature was copied multiple times the scribe would come familiar with this feature. That the scribe was able to learn is the reason that in Currier B "edy" is used instead of "ody" and the explanation for the shift from Currier A to Currier B. 
It simply depends on how many times a feature was copied if this feature is unique, occurs only on one page or looks like a rule to us.

For me the text generation method was coming obvious while looking on rare words. See for instance the word "lkl". This word exists only 9 times for the VMS. But this doesn't mean that this word is equally distributed. In fact there is one page with three instances of "lkl". This page is You are not allowed to view links. Register or Login to view. and since three instances of "lkl" can be found on this page it was three times available as source for generating other words. This is the reason that on page You are not allowed to view links. Register or Login to view. also the words "lklor", "kl" and "lkol" exists. Beside "lkl" only 6 other words contain the sequence "lkl". One of this six words is the word "lklor" on page f105v. This is not a coincidence! See "lkl" / "talkl" on page You are not allowed to view links. Register or Login to view. and "lklcheol" / "lkl" on page f115v.