The Voynich Ninja

OK, with noone else eager to prepare raw samples from this new Latin "solution", I went on and did that myself.

The sample includes You are not allowed to view links. Register or Login to view. to You are not allowed to view links. Register or Login to view. (~11000 characters), asterisk is counted as a single character, dot is counted as well (I don't understand whence the dots come from, but anyway there are only a couple of them in the sample).

Results:

h1 = 3.55
h2 = 2.07

So we can call this variant of Latin "LE-Latin" (low-entropy Latin)

If one considers that there are almost no Latin words at all in this "solution" even though the syllables were expanded with valid Latin endings, I'd be hesitant to call it Latin.


It would be like calling this English:

blofely wening mifer sanly interaty to mying froning maling

ah, I hate to repeat myself, but:

SITA VSVI LATE IN ISTA PER CANES

The initial spacing is misleading, but if you pronounce it, it will _sound like german:

Sieht aus wie Latein, ist aber kanes ("kanes" being austrian dialect for "keines")

It translates to:

Looks like Latin, but isn't


Spare your computing power (and efforts).

I'm sorry the german culinariæ are so much more fun than the english one's..

So I need to quote Thomas Jefferson:

"Fifty-two volumes in folio, of the acta sanctorum, in dog-latin, would be a formidable enterprise to the most laborious German."

Have a nice day

P.s.: If you replace sense with scythe in the greek example.. but hey. If my efforts towards the importance of being unearnest are too advanced, let me know.

There's also a common Flemish joke along those lines, it's an old one cause my grampa used to tell it..

DATIS NEPIS POTENTIS NEGRO TE

--> Dat is ne pispot en 't is ne grote.
"That's a chamber pot and it's a big one."

With respect to such a solution, this chart helped provide me some perspective (below). I found this chart from a link on ReneZ's site. I saved it because "the Latin solution" has probably been a very common solution tried but I have been gravitating to "the French solution." French, of course, is based on the root of Latin, which could help explain the latin roots, but incomplete of examples of transcriptions/translations. Somewhere, David Jackson called this the n-gram solution? Dog/Vulgar Latin and Dog/Vulgar French would seem to come up with very similar responses, only different. ; )

If you want to try it for yourself, pick the longest vords you can find and go to a dictionary and use alternative symbols for non-abbreviations. For example, I have been playing with the benched gallows to represent "effe, elle, esse, enne, ette, etc" in French, specifically (EVA-cFh). French happens to be a very wordy, vowely, elaborate language with many beautiful threads inside the weave of language...

The problems I am having are with the dictionary search engines themselves. For example, a vord like
ocfhorokear -- f57v -- has three common o's and two common r's. If we assume it is not cipher, it could still be a different combination of characters, but still very similar. So a common search like o?o?o???? needs to understand that the question marks can represent more than one character per glyph. I have been getting around this thru brute-force. Changing the length of the words I'm searching for. With all the possible solutions to o?o?o????, I move to ?????r???r, etc. Often times, a combination of * and ? are the only tools available. Under this method, to consider a paragraph is exponential. But the "labels" stand a chance...and the long vords provide better odds.

[attachment=1570]

In this case, ocfhorokear became s'effectuer (to make). Again, no proof. Theory only.

Why French? The Mad King, Charles VI. Valentina Visconti was French by her mother and raised by her grandparents after her mother died. Isabeau, the Queen, her cousin was Bavarian. Her Dad, Italian. Suddenly you have a very interesting triangle of knowledge which has all in one way or another been linked to the VMS.

Bottom line, the search tool doesn't yet exist to attack Middle French in this manner. Or really, any language, especially if you consider anagrams or mirrored text or glyphs, both valid and used at the time.

Rien ne m'est plus, plus ne m'est rien.
-Valentina Visconti

It appears that my gut feeling that PL might be using Google Translate to "unzip" the supposed VMS "Latin" may have been correct (see reply number 9):

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

Unfortunately, this method can generate many valid-looking words (that are not valid) because Google Translate's software makes a best guess even if you supply bad data. The more phrases you add, the harder it tries.

Also, there's a PL response to a comment on that link in which he says:

"It is not gobbledegook - it is a beta version of a computer transcription of the Voynich glyphs into Latin.

If it was gobbledegook it would show random words on random topics.  In fact, the beta transcription used with the images and a little judgement shows many words connected with the art of the apothecary."

My reaction?

Gobbledegook does not mean random. One definition refers to text that is so jargonistic, it is unintelligible. A second definition is text that is gibberish, which is the definition that I believe applies here. Text can be gibberish without being random. Many natural-language-generating software programs create gibberish using real words. There are whole websites created this way. They look like sentences until you start reading them. Then you realize it's nothing more than an algorithm churning out nonsense.

As far as I can tell, PL believes he is seeing "valid" Latin words because

• He does not have a basic working knowledge of Latin.
  
• A transcription of thousands of words is statistically likely to create a few that appear valid.
  
• He's using valid Latin prefixes and suffixes to expand some of the glyphs, which naturally will result in a few valid words because half the word is supplied by the part that was assumed and expanded.
  
• He is mis-translating (for example, he translated "banis" as bath—I know of no Latin dictionary in which 'banis" means bath, croque is not a Latin word either, it does not mean crocus or saffron in Latin, it's a French word for a coarse/crunchy texture).
  

His transcription is galimatias. There's only one seemingly-valid Latin word for every few lines and there is zero correct Latin grammar, or anything even close to it, even if one accepts the possibility of the text being lists, rough notes, etc., rather than sentences.


I wasn't even planning to comment on this after the second installment—I thought it was a closed subject—but the blogger is still actively defending the transcription, and people without knowledge of Latin are being led on, in my opinion, so I felt compelled to repeat my prior opinion—the transcription does not come close to being Latin and does not further our knowledge of the VMS in any way that I can discern.

I grabbed three lines at random from PL's "Latin" translation.

Nonsense words (in any language) have a strikethrough, words from other languages are underlined. Genuine Latin words are in bold (notice that none are in bold):

umlio estoque este iloesesteum feretqueum esaesaus esaus ileumeses
esetarus estlietnus oilanrus eteum rusetum ferolioque lietquees esqueota
estoil etorus eteum esaus ileeum olioque esanus


You might be able to argue that "rusetum" is russet, but I saw no evidence that this word was used in the 15th century.

If you break up lietquees into three words, you might be able to argue that it's Spanish/Portuguese.

If you remove a letter, you might argue that ileeum could be ileum (intestinal viscera) but as it stands, it doesn't mean anything.


So, out of three lines of text, there's only one word that is arguably Latin (and it's rare and probably didn't exist in the 15th century), there are three words that are valid in other languages, one of which is Esaus, which could be taken for a Biblical name (Esau) but otherwise wouldn't mean anything. The other 18 words are nonsense words in any language.

(By the way, you can't depend on Google translate to specify the correct language for a word. It makes a best guess, based on the letter combinations as to what language it MIGHT be, but that doesn't mean the letters actually mean anything in that language. For example, if you put "-um" at the end of a word, it will often guess Latin even if it's not a Latin word. If you put "-os" at the end, it will often guess Greek, and so on.)

(11-08-2017, 05:09 PM)-JKP- Wrote: You are not allowed to view links. Register or Login to view.(By the way, you can't depend on Google translate to specify the correct language for a word. It makes a best guess, based on the letter combinations as to what language it MIGHT be, but that doesn't mean the letters actually mean anything in that language. For example, if you put "-um" at the end of a word, it will often guess Latin even if it's not a Latin word. If you put "-os" at the end, it will often guess Greek, and so on.)

Google translate considers Voynichese Irelandian 

(11-08-2017, 08:01 PM)Searcher Wrote: You are not allowed to view links. Register or Login to view.Google translate considers Voynichese Irelandian 

LOL. Perhaps the blarney stone is appropriate.  

Haha the Blarney stone. I visited the castle many years ago but haven't seen it mentioned since. 

You've been wonderfully thorough in showing how problematic this really is. Still, it was interesting to see someone with the balls to unleash their system on the entire text.