A priori or A posteriori cracking?

[Arichichi]

Which is better for cracking the Voynich Manuscript? Thinking that a certain page talks about what is displayed in the page? Or thinking that it has no specific meaning relative to the featured pictures in the page? To rephrase my question, I want to know whether the cryptologists that deal with languages have ever tried to crack natural languages without a priori knowledge, just to test how strong their methods are. For example, has anyone tried to learn Korean or a similar language by studying a random Korean book and relying only on that book just to test the effectiveness of the code breaking methods that he relies on?

[davidjackson]

Well, that's the $20,000 questions isn't it.

Thinking that a certain page talks about what is displayed in the page?

That's the "block paradigm" solution (originally proposed by Nick Pelling) and there are a few teams working on that. I don't know of any breakthrough.

To rephrase my question, I want to know whether the cryptologists that deal with languages have ever tried to crack natural languages without a priori knowledge, just to test how strong their methods are.

I don't really see how that would work. Natural languages aren't codes, and you wouldn't approach the problem in the same way.

[obelus]

In The Story of Decipherment, Pope cites the case of pragmatic Japanese medical students who deciphered a forbidden Dutch textbook in 1775.  They were motivated specifically by the anatomical illustrations, but "they none of them knew the values of the European letters or how they operated."  It was accomplished through general ingenuity, including scraps of a priori knowledge, rather than formal cryptanalysis.  In his translated memoir, Gempaku Sugita states that "we translated by conjecture."

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

[Arichichi]

To rephrase my question, I want to know whether the cryptologists that deal with languages have ever tried to crack natural languages without a priori knowledge, just to test how strong their methods are.

I don't really see how that would work. Natural languages aren't codes, and you wouldn't approach the problem in the same way.

The idea is to consider that you don't have access to translation, or grammar books, or dictionaries, or any resource that would help you learn the language, and you'd learn the language through consulting one book that narrates a story you have never read before, or something of this kind.

A perfect example is the case presented in the answer above.

[Mark Knowles]

Thinking that a certain page talks about what is displayed in the page?

That's the "block paradigm" solution (originally proposed by Nick Pelling) and there are a few teams working on that.

Nick's block paradigm or a simple crib combined of a few distinct identified specific words are basically good approaches. However as yet nobody has identified a proportion of text in the Voynich that maps onto a portion of text in another document and there has been very little agreement on a some kind of crib. This could change at any moment if someone were to discover a document which illuminates this relationship. Ultimately nobody knows what is lurking in the archives waiting to make this subject crystal clear.

Deciphering the Voynich without some kind of crib at all, whilst not impossible would be extremely difficult.

I should say from my experience deciphering a document when one has a lot more information than we do in the case of the Voynich can be far from easy on occasions.

[R. Sale]

I would add to that in respect to the potential herbal identifications and textual comparisons that such investigations have actively failed. In all categories, herbal, astronomical, religious, etc., there are no 'candidates' for investigation.

Having a text to use as a crib would be incredible. Not to mention that no one would believe it.

[Mark Knowles]

I would add to that in respect to the potential herbal identifications and textual comparisons that such investigations have actively failed. In all categories, herbal, astronomical, religious, etc., there are no 'candidates' for investigation.

But failure in the past does not mean failure in the future. A large percentage of the relevant material out there has not been studied by Voynich researchers, particularly as researchers tend only to rely on the limited amount of material that has been digitised.

[davidjackson]

It was accomplished through general ingenuity, including scraps of a priori knowledge, rather than formal cryptanalysis. 

Yes, but the Dutch were in Japan at the time, and were permitted to give lessons to Japanese students. There were official Dutch speaking Japanese translators. It wasn't an unknown language.

[Anton]

I have little belief in the "block paradigm" in the strict sense of the word (that is, as it was put forward by Nick), because there is, in my opinion, very little probability of discovering the plain text source from which the VMS borrows something. If the VMS text is unique, then no such source has ever existed. If it is not unique then the respective primary sources are most probably lost forever by now, because the best indicator would be similar drawings, however the VMS drawings are quaint even in the seemingly commonplace patterns such as the Zodiac, and have no perfect parallels to themselves.

However, I think that the prospective approach would be, so to speak, the "loose block paradigm". Meaning that one takes texts of more or less the same time period, of supposedly more or less the same place, subject and underlay language (these three of course are not certain and are subject to educated guesses) and tries to explore correlations between word-level statistics of these known texts with the same of the VMS.

My favourite idea is to switch from character-level to the word-level. This of course requires assuming no character shuffling as the starting point of research, but the overwheming majority of character-level approaches did assume the same.

[Mark Knowles]

I have little belief in the "block paradigm" in the strict sense of the word (that is, as it was put forward by Nick)

I think I have more optimism that a block paradigm is out there and will be found, though I have no certainty in the matter. Ultimately we won't know unless and until it materialises.