(10-08-2020, 08:57 PM)nickpelling Wrote: You are not allowed to view links. Register or Login to view.Me, I'm still bemused by our inability to properly account for the first glyph of each line
Yes, there is much to be learned there still.
The statistics that have been presented in the past are unfortunately 'contaminated' because they are for all lines, i.e. including the first lines of paragraphs.
The behaviour is different there.
A 'first-character' effect for all except first lines of paragraphs exists, but is much more subtle.
The average word length for these words is barely higher than that of other words.
(11-08-2020, 12:48 AM)-JKP- Wrote: You are not allowed to view links. Register or Login to view.For a long time I sensed resistance to the idea that the VMS might be broken up into components that are not specifically word-related.
Yes, I am aware of my irrational preference for ideas that give some hope of decipherment. We all know of the role of well separated words with a clear context in the decipherment of Egyptian hieroglyphs and Mycenaean Linear B.
As I said, seeing that Bakker, Mair and Sproat agree on the fact that the VMS is undecipherable makes me challenge my convictions.
Timm and Schinner's meaningless-text-algorithm seems to be the only solution to have gained some consensus among other researchers. On the other hand, Friedman came to the final hypothesis of an artificial language of the a priori type, suggesting that he believed that the creator(s) could understand the text. I doubt it will ever be possible to decide between meaningless and undecipherable.
If we look at very early texts, words were often run together. It was up to the reader to know where the word breaks should be. This tradition still existed to some extent in the early medieval period.
In contrast, separating text into syllables was not so common in narrative manuscripts, not so common in medieval ciphers before c. 1460s, but was very common in missals and other music-related manuscripts, as well as in maps and diagrams. It also occurred in mnemonic Llullian-style imagery and, to a lesser extent, in scientific manuscripts.
If the person who created the drawings is the same person who created the VMS text, then they would have been familiar with breaking up words into syllables. If the illustrator and the Voynichese creator were different people (which they frequently were in the Middle Ages and even in the modern age), then I suspect the chance of the VMS being broken up into syllables (or some kind of chunks) is less. As Nick has often pointed out, they just didn't do it that way in the earlier years.
I've often tried to discern if the illustrator's hand matches one of the scribal hands but it's very difficult to compare them. We might never know. And even if the illustrator and one of the scribes match, it still doesn't tell us who invented Voynichese. A mastermind isn't always a scribe.
But if the illustrator created the Voynichese glyphs and how they go together... I think it's possible. The concept was there. The VMS is remarkable in many ways. Perhaps it is remarkable in this way as well.
(11-08-2020, 10:19 AM)MarcoP Wrote: You are not allowed to view links. Register or Login to view.As I said, seeing that Bakker, Mair and Sproat agree on the fact that the VMS is undecipherable makes me challenge my convictions.
Fortunately, it is only an opinion that they agree on, not a fact :-)
But it could be undecipherable and even meaningless...
(05-08-2020, 08:15 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view. (04-08-2020, 07:48 PM)MarcoP Wrote: You are not allowed to view links. Register or Login to view.Hi Rene,
the calculations I posted are my own. They are certainly based on brute force and do not assume anything clever. I just wanted to point out that the search space is huge: my feeling is that one is going to fall into local minima, maybe better than those found by Yokubinas, Cheshire, Ardic &C, but still inconclusive.
Quote:Many of the statistics we already have are making both specific cases rather unlikely.
I am not sure you mean what I hope you mean 
Please be more specific.
Hi Marco,
so I was reading a bit too fast... I have almost no time these days.
However it makes no fundamental difference.
I hope to write a bit more this evening.
I still meant to write more about this.
One of the fundamental questions for me is whether it is possible to make a 1-to-1 mapping of Voynich words to plain text words in some language, and arrive at a meaningful text.
Such a mapping could follow some logic, for example in case of a simple substitution cipher, or no logic at all, in case the words are simply enumerated in a list. While these two cases are very different from the point of view of deciphering, for the present question they are the same. The enumeration option was one of your two examples.
The point is that it does not seem to be too likely to be possible, due to the strange statistics of repeating word sequences.
This does not mean that all hope is lost. It could be that there are some simple rules why this is the case.
For example, there seem to be no cases of repeating sequences that cross a line break.
(I am sorry, but I writing from memory). In some special cases (I did some substitutions) there were such cases, but then both strings had the line break in the same place.
(I am not sure if I explained that clearly).
The bigram example is more tricky. If this were done by the author, it would have been a bad idea, because the decipherer would constantly lose track.
Here we would also have a typical case (=problem) that it makes the words very short, and it means that something else must be going on as well....
It's far from obvious to me that we can, with the current state of our knowledge, map between even Herbal-B and Q20-ese, let alone between A and B.
So I remain somewhat unconvinced by the idea of trivially mapping any part of the VMs to a real-world language en masse.
Having said that, I still think there's good reason to suspect that EVA ch maps to plaintext u/v (and that strike-through gallows are merely concealing consonant followed by u, nothing more complex than that), so I remain optimistic that progress can be made, albeit slowly.
(11-08-2020, 11:39 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.One of the fundamental questions for me is whether it is possible to make a 1-to-1 mapping of Voynich words to plain text words in some language, and arrive at a meaningful text.
Such a mapping could follow some logic, for example in case of a simple substitution cipher, or no logic at all, in case the words are simply enumerated in a list. While these two cases are very different from the point of view of deciphering, for the present question they are the same. The enumeration option was one of your two examples.
The point is that it does not seem to be too likely to be possible, due to the strange statistics of repeating word sequences.
Thank you for your comments, Rene!
I agree that repetitions make a 1-to-1 mapping of words very unlikely. In particular, the simple fact that the most frequent word is reduplicated as
daiin.daiin many times all through the manuscript appears to be incompatible with European languages (e.g. I don't think I have ever seen any occurrence of "et et" in a Latin manuscript).
Something I like in Timm and Schinner's theory is that they addressed the issue of the high reduplication rate. Other solvers tend to ignore it as not deserving an explanation: in English you can have "that that", so everything is totally ordinary. Doh.
(11-08-2020, 11:39 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.This does not mean that all hope is lost. It could be that there are some simple rules why this is the case.
For example, there seem to be no cases of repeating sequences that cross a line break.
(I am sorry, but I writing from memory). In some special cases (I did some substitutions) there were such cases, but then both strings had the line break in the same place.
(I am not sure if I explained that clearly).
I am afraid I cannot follow what you wrote about repetitions and line-breaks and how this suggests the existence of simple rules.
But I agree that the impossibility of the 1-to-1 word mapping solution does not rule out the theoretical possibility of other solutions.
If one compares a manuscript text with a printed version of the same text (as I did You are not allowed to view links.
Register or
Login to view.), one finds that several manuscript-words map into the same printed-word. This is due to common things like:
- hyphenation
- inconsistent abbreviation
- inconsistent spelling
- inconsistent word-spacing
On the forum, we discussed some possible evidence of "homographs" i.e. Voynich words that appear to correspond to different plain-text words (e.g.
otol).
[
attachment=4686]
The different dialects mentioned by Nick are even more direct evidence that the same plain-text word could be expressed in more than one way (according to the different sections).
So, if the manuscript is meaningful and spaces correspond to word-spaces, the mapping likely is many-to-many and we are in a much more complex position than a 1-to-1 mapping between words.
Hi Marco,
I have been (as so often) a bit too short and succinct, in order to make my views clear.
Here are a few assorted comments:
- I don't think that the Timm (and Schinner) approach really explains the repetitions of high-frequency words. The method is basically arbitrary, so it is suggested that the person executing it would occasionally just (arbitrarily) repeat the very last word he had just written several times. This is not really a good explanation. It is only something that cannot be excluded.
- The point about line breaks is related to the various 'vague' observations that the line is a functional unit one way or another. Nobody knows exactly how. However, there appear to be special rules for first and last words in lines. This means that (or could mean that) if the source text has normal repeating sequences, these could be interrupted because there are special rules to be applied to line breaks. I consider that the lack of repeating sequences that cross line breaks is a (very) mild confirmation of that possibility. Something that really requires more study.
- With respect to:
Quote:The different dialects mentioned by Nick are even more direct evidence that the same plain-text word could be expressed in more than one way (according to the different sections).
, I hate to contradict Nick (no, just kidding :-) ) but this is a hypothesis, and I would not count it as evidence
I am getting more and more careful and skeptical with age :-)
Thank you, Rene,
I totally agree: skepticism is the most appropriate attitude! I am also grateful to Bakker, Mair and Sproat for helping me question my ideas.