+- The Voynich Ninja (https://www.voynich.ninja)
+-- Forum: Voynich Research (https://www.voynich.ninja/forum-27.html)
+--- Forum: Analysis of the text (https://www.voynich.ninja/forum-41.html)
+--- Thread: Successful translation ? (/thread-2701.html)
Successful translation ? - ReneZ - 25-03-2019
It is not my intention to go into a never-ending discussion about the recent Judaeo-Greek translation proposal by Geoffrey Caveney. However, there are a number of general points related to this that are worth bringing up, and do not specifically concern only this example.
It is important to keep in mind that any decryption or translation is not a process by itself, but it is an inverse of a process that has already taken place.
It is the act of reversing what the person did who wrote the text.
If one proposes a decryption (even when it is just simple substitution) it must be possible to describe the original encryption process reasonably accurately, and in some detail.
The reverse process may in some cases be more complicated.
It is not sufficient to:
1) Take a transcription file
2) apply some more or less arbitrary transformation of it
3) claim that this is the meaning (i.e. the plain text).
One has to describe how this plain text was converted into the text we see in the MS.
If one has the right solution, this will be possible.
If one is doing something very different, i.e. not the right solution, but just trying to squeeze meaning out of the transcription using the incorrect method, this will not be possible.
Also a detailed description of how the decryption is done may not be possible, or if it is attempted, may clearly show the weaknesses of the method.
In recent cases where some document was decrypted successfully, there was never any discussions whether it is correct or not. Take for example the Codex Copiale.
In the case of the Judaeo-Greek example, while we alternatively hear that it is consistent, or that it does not have to be consistent, this detailed description is not given.
The translation goes into two steps. The first is from Voynichese to "Greek written in Hebrew characters", and the second is from this to more standard Greek. In both steps changes are made.
In the first case, there is a many-to-many mapping of characters. In the second, parts of words are added/deleted and some more characters are modified.
What we need to see is:
- the Greek text that is proposed to be the plain text
- the method how this is represented using Hebrew characters
- the method how this new text is then mapped to the Voynich text
Both would normally involve a table.
We then need to see an explanation why the resulting text in the Voynich MS exhibits the typical word patters and the low entropy values.
This is true for any solution, of course.
One thing that this particular proposal covers is the appearance of Eva-f and Eva-p in top lines of paragraphs, which is typically overlooked in proposed solutions (because the would-be solver usually is not aware of this feature).
RE: Successful translation ? - MarcoP - 25-03-2019
Hi Rene,
thank you for starting this thread. I think it could be useful to define a set of forum rules for those who intend to share their translation proposal.
I agree with what you write about the original-language->Voynichese steps. In particular clearly presented mapping tables are necessary to make reduplication of the process easy.
I think that the original-language->English steps also are often inaccurately presented. Sometimes, a first "word salad" step is presented, and in a second step meaningful English is created by free association with the available words. I call this "word salad" because root words are listed without any analysis of their morphology. Sadly, in many cases only the final result (meaningful English) is presented.
See for instance this "translation" ("transliteration"?) by You are not allowed to view links. Register or Login to view.:
The following is an example of the kind of English translation that I find useful (from You are not allowed to view links. Register or Login to view.). Each word is annotated with specific grammatical categories and a rigorous word-by-word translation. This word-by-word translation is easy to match with the actual translation at the top of the page. I would find it useful if, in addition, also the corresponding root word was presented for each word in the text.
RE: Successful translation ? - ReneZ - 25-03-2019
Another problem of only looking at decryption, and not at the encryption step, is that one can come up with character mapping tables that are anywhere from unlikely to impossible.
In this post: You are not allowed to view links. Register or Login to view.
there is a mapping from Voynichese to plain text. I already indicated in that thread that, what we need to see is the table that the author would have used. I try to create it here, by inverting the given table (see at the bottom). This maps plain text to Voynichese.
It should be clear that this table is unusable for encryption for various reasons:
- b, c, g, h, j, k, p, v, w, x, y, z are missing. They are not allowed to appear in the plain text.
- s can only appear in the combinations 'es' and 'os'
- qu can only be followed by e
- u can only be followed by l, n, r
With the table given by the author, one can map Voynichese to a text using latin characters. One cannot map a Latin text to Voynichese.
Inverse table follows, Latin to Voynichese:
a o
ai ch
ad q
d s
ea ch, ol
em cph
eo cfh
er y
es cth
eum aiin
eus ain
ex ai
f f
i e
l k
m d
n d
o l
oe al
os ckh
que t
r k
re y
t r
ta m
te m, p
ul sh
un sh
ur sh
RE: Successful translation ? - geoffreycaveney - 25-03-2019
I think it is fair and on-topic for me to mention in this one single comment in this thread, that I have provided a very detailed response to Rene's conditions for a solution proposal, including all of the original encryption steps that I believe the author employed.
The moderators have decided that because the detailed summary explanation in my response specifically relates to my own Judaeo-Greek decryption proposal, that it belongs in that thread, rather than this one. I respect this decision of the moderators.
I only ask that I be allowed to provide, for readers of this thread, the following link to my specific response comment in the other thread, so that it is simple and convenient for those readers who wish to read my response, to be able to find it readily.
Here is the link:
You are not allowed to view links. Register or Login to view.
Thank you to everyone for the respectful and stimulating discussion.
Geoffrey
RE: Successful translation ? - Emma May Smith - 25-03-2019
Recently I've been playing with my own "four step solution" as an experiment. I tried to identify a language, assign values to glyphs, find matching words in the target language, and create some kind of meaning out of the word soup. The truth is that it hasn't been easy.
I thought the target language was a good choice: geographically viable, historically unwritten, small phoneme inventory, simple syllable structure. I read about the language's phonotactics, its historical phonemes, the frequency of different sounds, and about related languages and dialects (to give myself some leeway). I would have been happy to assign any value to any glyph so long as it worked. But nothing seems to work as well as it should. I have barely been able to look at the meaning of individual words and never once got to the point of questioning whether sentences were grammatical.
I'm still going to play with it for a while longer as I feel that I'm learning a lot. The main lesson is that the kind of abuse you need to do to a language to have it fit the Voynich text is quite severe. I'm really unwilling to mangle a language that much even for fun.
If I find a solution which is acceptably pleasing I will share it, along with the identity of the target language (which I actually think is very good).
RE: Successful translation ? - -JKP- - 26-03-2019
(25-03-2019, 09:14 PM)Emma May Smith Wrote: You are not allowed to view links. Register or Login to view.Recently I've been playing with my own "four step solution" as an experiment. I tried to identify a language, assign values to glyphs, find matching words in the target language, and create some kind of meaning out of the word soup. The truth is that it hasn't been easy.
I thought the target language was a good choice: geographically viable, historically unwritten, small phoneme inventory, simple syllable structure. I read about the language's phonotactics, its historical phonemes, the frequency of different sounds, and about related languages and dialects (to give myself some leeway). I would have been happy to assign any value to any glyph so long as it worked. But nothing seems to work as well as it should. I have barely been able to look at the meaning of individual words and never once got to the point of questioning whether sentences were grammatical.
...
I've managed to do it with four languages. Well, 3.5 languages... the 4th one didn't work quite as well. I wrote out a list of about 30 languages after looking into their structure and went through the list one at a time.
But... I could never get more than a few long phrases, or a sentence here and there (quite a bit more than some people get but not nearly enough to satisfy me), and then the approach doesn't generalize to the text around it, and I'm pretty confident that it's NOT because the VMS switches languages—the structure of the VMS tokens doesn't change.
I'm taking a wait-and-see attitude to what Geoffrey is doing. Greek was one of the languages I tried and had some success with (this was years ago) BUT it was not Judaeo-Greek. I also tried Yiddish, Ladino, Romano (and many of the more common languages), BUT these didn't work terribly well.
The reason I decided to take a different approach to understanding the VMS after that long and very arduous exercise was because I thought to myself, "If I can get this to work in three different languages, then that probably means the VMS is structured with a common vowel-consonant appearance (I say appearance because the vowel-shapes are not necessarily vowels) that occurs in a number of languages and it's probably NOT because I have a solution."
RE: Successful translation ? - davidjackson - 30-03-2019
I would suggest that there is nothing wrong with a solution that offers a many to one transliteration - some languages do this, after all.
However, the translator has to establish a grammar to go with the transliteration. You can't switch and swop as you want to, which is what all of these translations do. There have to be rules that are consistent, in the same way that natural languages do.
For example, Greek letters <η> <ι> <υ> are pronounced [i] and so modern transcriptions often render them all as <i> into Latin languages.
RE: Successful translation ? - -JKP- - 31-03-2019
There were many substitutions in the German language in Switzerland/Alsace, Bavaria/Tyrol in the 15th century.
For example, p was substituted for b and a was frequently substituted for e (depending on the word). Niet or nit was substituted for nicht.
But it was done in a reasonably consistent way.
Thus, you see ain for ein, main for mein, puch for buch, puchstabe for buchstabe, etc., throughout the manuscript.
However, here's the complication... that doesn't mean that every e was substituted by a or that every b was substituted by p. There were certain words they did that way. Since we can read German, it's easy to see which ones they were.
Unfortunately, we can't read the VMS, so we can't make a mental note of which ones might be substituted and which ones might not, which makes it harder to evaluate whether the translation substitutions are "consistent" or not within that particular system.
RE: Successful translation ? - geoffreycaveney - 31-03-2019
David and JKP,
I propose that the author of the Voynich MS first did a many to one transliteration, for example Greek mu / Judaeo-Greek mem and Greek nu / Judaeo-Greek nun were merged into the single Voynich character [l].
But then, as a following step, the author made the text more of a cipher by applying the principles of a so-called homophonic substitution cipher, such as the familiar nomenclator ciphers. That is, such a cipher creates multiple character representations of each single letter. There is a known example of a homophonic substitution cipher employed in 1401, from Mantua, so such a method is not necessarily anachronistic for the Voynich MS.
By employing a many-to-one encryption step, followed by a homophonic substitution encryption step, the end result is that any of the original set of letters may end up being represented as any of the cipher text set of homophonic substitution characters.
The entire process may be visualized as follows:
Step 1: {mu/mem, nu/nun} > [l]
Step 2: [l] > { [l], [sh] with closed loop, final [(i)in] }
I hope that my explanation of this process is clear.
=======
JKP, I would like to cite one particular example that makes me suspect that the process may have involved a deliberate homophonic substitution cipher step:
If it is not a deliberate homophonic substitution cipher, then you will not like the discrepancy between [sh2ky] as "mythous" and [lkechy] as "mythois" (pronounced "mithis") on the very same line. This does not appear to be a natural substitution in a typical manuscript text, such as the German examples that you cite. (I do note that I interpret [sh2], with a closed loop on top, as really just a ligature of [l+ch], but still I suspect you will not like this alternation on the same line.)
But if it is a deliberate homophonic substitution cipher, then it makes perfect sense for the author to encrypt the two instances of the same word on the same line with two different homophonic substitution characters for the same letter. This is precisely how such ciphers were intended to make the text more difficult for the uninitiated to decipher and read.
The initiated reader who knows the cipher, on the other hand, can easily undo the homophonic substitution step, and easily identify the two words as the same. Then the whole text is reduced to the result of a many to one transliteration, which is perfectly natural for many languages such as Greek, as David points out.
Geoffrey
RE: Successful translation ? - Ruby Novacna - 15-04-2019
@Marco,
the translation of the gospel that you quote is the result of the work of several translators and scholars, I suppose. In order for Voynich's translation to have the same level of meaning, the work of one person would never be enough, in my opinion. To hope to give a sensible text requires a teamwork: understanding images, research texts on the same topics (what you do precisely).
About the "Test of 6 words" that you have proposed, I give you the reading of one word out of six: ypolol that I read as ηπολα for επαυλα / επαυλος - sheepfold, house, dwelling, which is in the yard, in the house. And then ? I have no idea about the meaning of the image to try to argue my reading. On the other side I refuse to give up on the pretext that I have not found a relationship with the image. Besides, is it my role to explain the images?
The mistake, in my opinion, stems from the requirement that it be the same person who presents a grandiose theory, explaining the origin, the imagery and the text. This person may not have been born yet ?
Ruby