The Voynich Ninja

-JKP-

While I have avoided treating anything outside my own field, it is impossible to properly explain any problematic imagery without going quite deeply into the historical and cultural matters which validate - or invalidate  - the sort of issues which crop up in the course of research, and the end result was that all the lines of investigation for the 'ladies' folios took me to the high, overland routes and the period when (as various historical sources inform us, including the Latins' sources and the Islamic ones) the lingua franca of that overland route during the Mongol century and to a little later was Cuman.

One final item which persuaded me to stick my neck out was that matter of Georg Baresch never asking Athanasius Kircher for a translation - only an identification of the script.   This niggled, and while I could imagine it was politesse, the inference really did seem to be that he wanted no more.   Baresch wasn't a world-traveller but as it happens we know that the last use of Cuman survived in that region, the last speaker - so far as the records tell us - dying at about the time that the manuscript finally went to Kircher.    It seems awfully sad, somehow, to think that Baresch might possibly have had the language itself and been denied his wish to read the manuscript only because Cuman was written using other scripts, and these had been forgotten first.   

I won't talk about the prevalence of non-standard dialects, the forms of othography and the way a language alters over time.  Historical linguistics is a whole other thing again.  Emma's field, I think.

Exciting, anyway.

I was excited when I saw the announcement.


Not so excited after watching the video. They will have to show how they go from the transliteration to the translation (which wasn't really covered in the video) because they did not address any of the problems inherent in the lack of diversity and positional constraints in the VMS text compared to natural languages.


They do allow for multiple interpretations (mostly pronunciation) in some of the characters, but they still interpreted many of the glyphs according to EVA designations and if you do that, such as reading EVA-i as "i", then they are already in trouble in terms of structural differences between natural language and Voynichese.

Also, if you take a very common token and translate it (as a month-name, for example), then you have to account for why it would occur with extreme frequency in other places in the manuscript where one would not expect to see that particular word (or to see it with such great frequency). They didn't address this at all.


.
I'm in wait-and-see mode but I'm going to make sure I keep busy in the meantime.

Think about it, using almost the same glyph-transliterations (mostly-EVA) and almost the same approach (substitution code), different people have "translated" the same text as Latin, French, Italian, Romani, Turkish, Sanskrit, English, Welsh, Hebrew, Arabic, Czech, Slovak, Greek, Finnish, and probably a few other languages I haven't yet read about.

In any document that's long enough, if the shapes that look like consonants and the shapes that look like vowels are balanced so one follows the other throughout the text, you are going to find dozens, sometimes hundreds, of words in almost any language you can imagine. The trick is to find an approach that looks at the text as a cohesive whole so that it confirms itself, so meaning doesn't have to be wrestled out of it or imposed upon it. I haven't see that yet.

One thing I appreciate in this work is that the main author is a native speaker of the candidate language. My impression is that several poor attempts with Latin are due to insufficient linguistic competence.
I also appreciate that they say that the ms appears to be written in verse: the relevance of poetry as a parallel has been discussed by You are not allowed to view links. Register or Login to view. and it could help explain some LAAFU (line as a functional unit) effects.

Finally, their work on f67r1 as a calendar should be easy to understand and verify, when we have all the details. I couldn't find any confirmation for Seperu / Seferi meaning "November", but things like that could be easily confirmed or rejected by someone familiar with Turkish. Since they could only match 5 of the 12 hypothetic month names, their translation of something so less constrained as a whole herbal page seems to me necessarily unreliable and premature.

I finally finished watching the video. These guys are head and shoulders above your average google-translate-Voynich-solver. They realize that language changes drastically over time and use actual dictionaries!

I know from several current researchers that they'd put Turkic languages on top in their list of possibilities, with good reasons. So this is quite exciting. 

There's really just one question I have right now: which degree of freedom did they allow themselves? This is where most proposed solutions stumble.

I find it hard to understand all this enthousiasm.

This was posted more than a year ago (Feb. 20 2017) on You are not allowed to view links. Register or Login to view.

Quote:The whom it may concern,
The Historical Voynich Manuscript book translation problem was partly solved by Ahmet Ardic at 01/01/2015. The FSG created VOYNICH alphabet and others are wrong.
The real alphabet has been created, typed, checked, matched and prove 1/1 for any of the all the book pages.
So, The historical Translation problems is fully solved at 20/02/2017 by Ahmet Ardic.
It is partly translated yet (because of my busy time table) by Ahmet Ardic. (Calgary/AB/Canada) But it can be fully translated in a one day, if I will found with the original alphabet character written book in correctable electronic version.
Thanks,
Ahmet ARDIC
For any question please visit my You are not allowed to view links. Register or Login to view. social media page.

Also, this pseudo-Voynichese text is supposed to PROVE something:
You are not allowed to view links. Register or Login to view.

This is late-15th or early-16th century Turkish from the U Penn site:

Ne habar scizum girlerden?
Hits nesle bilmezom tsaa dimege.
Gioldassum varmi tsenumle?
Ioch ialanuz geldum.
Benumle gelurmitsun?
Irachmider tsenum utahhom?
Hits nesle bilmezom tsaa dimege.


Word-count = 25.

In their video, the researchers transliterated EVA-a as “ə”, EVA-o as “o”, and EVA-i as “i”. They don’t explain whether they are using “ə” as in “uh” (short-u) or “ə” as in “father” (short-a) but it doesn’t really matter—it’s enough to know that they have used many EVA designations to identify common vowels, with the exception of EVA-e which is transliterated as consonant “c” and the one that looks like a "wide a" is not interpreted as an "a" but as a consonant ("v").

Note how in medieval Turkish, “e”, “a” and “i” occur frequently, with “o” and “u” being less common. The counts are as follows:

e- 19      i- 14      a- 12      u-10      o- 5    



Now look at the first 25 words of You are not allowed to view links. Register or Login to view. (or any folio) of the VMS:




a- 12   o- 6    i-4

It is important to note that in Turkish, the letter “i” can follow almost any character (c, g, H, b, d, G, m, m, m, H, b, d) or be at the beginning.

In the VMS (and this is true for virtually the entire manuscript), the “i” (or double-minim glyph that resembles “u”) is always preceded by “a” and only followed by certain specific glyphs.

The letter “i” is common in old Turkish. It is significantly less common in the VMS and has none of the flexibility of the Turkish “i”.


That is just one example out of many. Straight substitution codes do not result in text that mimics natural languages (not even agglutinated languages) because VMS glyphs are position-dependent.


---
When I saw the announcement I was hoping they had found a different way to process or transliterate the text from those who had gone before. I did not expect another substitution code, especially one so closely tied to shapes that do not behave in the VMS as they do in old Turkish.

I've been trying to figure out how many liberties they are taking when going from the transliteration to the translation, but there isn't enough information in the video to really determine this, other than the statement about 60+ combination letters, and the example of the bench char for which he gave several interpretations.

The little bit I was able to glean from the video was this...


Their interpretation of the rotum on You are not allowed to view links. Register or Login to view. is that it is divided into 12 segments that represent months.

Using November as an example...

The text in Voynichese is dalar y


Based on the transliteration charts in the video this becomes: seper u


Which they then turn into seper ayi or sefer ayi (there was no explanation as to how "u" becomes "ayi" and apparently they are providing leeway for "p" to become "f").

I was not able to find seper/saper/separ/sapar in the Turkish dictionaries (either modern or old), but I did find the following:

• sefer = voyage/journey/expedition
  
• safar/safer = second month of the lunar calendar (note that the word position was identified as November but one has to keep in mind that lunar months change each year and do not correspond directly and are not divided the same as solar calendars).
  

So, there were several subjective changes in the transliteration process to turn a non-phrase (seper u) into a phrase that fits the context somewhat (safer ayi - second lunar month month - I wrote month twice because "ayi" appears to me to be redundant).


To recap:  seper u --> safer ayi --> second-lunar-month month


Note that dalar occurs on a number of pages by itself and as portions of other tokens, including a rota that is divided into 4/8 segments, and one that is divided into 9.


I don't know if this example is more or less representative of how they are translating the rest of the manuscript, but it appears that subjective interpretation of glyphs and the letters associated with those glyphs does occur.

Thanks JKP, that does give a bit of insight. Now, the redundancy of month-month is not necessarily wrong, such things happen in natural languages all the time. For example in Dutch it is common to say "dennenboom" or "beukenboom" for trees, even though the first parts are themselves trees and the "boom" (tree) part is pleonastic.

I wonder if the "u to ayi" thing might be one of those instances where they read a digit out fully, kind of like you'd spell "w8" instead of "wait". Just guessing..

You guys are always in for a joke, I see.   

ps. I did not watch a microsecond of the video, because there is no basis at all for a Turkic language whatsoever