The Voynich Ninja

An idea recently occured to me which remains undeveloped, but I want to share it at an early stage.

A while ago Wladimir provided some cross-section page comparisons which suggested that plant names are not mentioned at least in one of the sections - either botanical or pharma (and, possibly, in both). This is, in principle, corroborated by several very promising initial results of plant identification by way of image mnemonics. Quite naturally, if a plant's name is revealed through mnemonics, there is technically no need to mention it in the nearby text. It may or may not be mentioned in other sections of the MS, for the need of reference, but there will be no need to label the plant.

So far, so good. But there's the question: why so heavily rely upon mnemonics at all? Even if the author does need these mnemonics in general (for whatever purpose), why not add textual aid to that - which would be only simple and natural?

One answer on the surface is that the author was not confident in the robustness of his cipher. So he decided to not add textual clues, in order to strengthen concealment. On one hand, this assumption shows the author as an exceptionally smart and forward-looking guy, on the other hand it shows him somewhat stupid, because he writes in a cipher while being sure that the cipher is not difficult to crack.

But suddenly an upside-down explanation came to my mind. If he does not mention plant names, then this might just mean that... they cannot be expressed in Voynichese at all! Voynichese just does not possess means of naming plants.

Likewise, the "aror sheey paradox" can be explained. Why is aror sheey in Voynichese in You are not allowed to view links. Register or Login to view. while the rest of the line is in plain text? Because only what's hidden behind aror sheey can be translated into Voynichese, and valden ubren and all that gasmich - just can not!

This suggests that the expressive power of Voynichese is quite restricted, with probably no grammar, which makes Voynichese not even a synthetic language (what does not have grammar cannot be called a language), but a kind of synthetic "notation" - the "Voynich-synth" - something not far from what Don made of it, with his proposed solution (now unfortunately offline, AFAIK) speckling with drams and ounces.

Any messages and concepts which do not fit into the framework of this notation are expressed by way of exceptionally elaborate mnemonics and visual codes (like those nymphs with their legs, papellonny, plant mnemonics, stars, whatever...).

Offhand there are ready counter-arguments to this. For example, consider the chain: plant (botanical section) -> mixture/medicine (pharma section) -> cure (recipe section). Suppose the name of the plant in the botanical section is conveyed by mnemonics. Next, plant names in the pharma section are conveyed by mnemonics as well, since they visually match plants from the botanical section, and the names can be thus reconstructed. Suppose, further, that the mixture is referenced in the recipe section not by the mixture's name, but in some roundabout fashion (dunno, say, folio number/row number). But how would the name of the respective disease be encoded? And if it can be encoded, then how it is that names of plants can not?

(28-07-2017, 11:42 PM)Anton Wrote: You are not allowed to view links. Register or Login to view....
Suppose, further, that the mixture is referenced in the recipe section not by the mixture's name, but in some roundabout fashion (dunno, say, folio number/row number). But how would the name of the respective disease be encoded? And if it can be encoded, then how it is that names of plants can not?

I've spent years trying to discover whether this is so.


A few years ago, I created a concordance of the entire manuscript to try to discern whether there were code-links between sections (like between plants that might go together in a mixture, for example).

What I can say is I have a very long and complete record of the connections between the illustrated pages and the starred-text pages at the end and it appears that there MIGHT be links between individual pages and the starred-text pages (it took months and months of hard work to map this), but I cannot find discernible CONSISTENT patterns between related plants or plants and pool pages, for example. Certainly there's overlap and even some interesting patterns, but... there's no strong indication of specific commonalities.


But, here's the problem. My original concordance respected the spaces between Vwords (I then started one that doesn't) and if the spaces in the VMS are flexible or if they are fake spaces, then it will take MUCH more work to discern whether there are code-words to link things up.


It's also possible that each section of the manuscript was intended to be semi-standalone (in other words, the starred text might not be recipes, they might be something like proverbs and thus unrelated to plant or pool pages).


Even if individual sections of the manuscript have no obvious code-links, I was disappointed that I did not find any obvious code-links between plants with the same uses or with the same morphology. I really thought I would. I found links, many of them, but not ones that you could say, ah, good, these are linked AND they belong together in such-and-such a way.

In linguistics, the distinction between synthetic and analytic languages is that the former stick more "bits" of meaning in one single word. This generally results in long words and a complex, hard to learn grammar. Finnish is more synthetic than German. An interesting development in Western Indo-European languages is that they have been evolving from synthetic to more analytic. Latin has much more "grammar" than its offspring. You need to "unpack", i.e. analyze the grammar from Latin into short words when translating to French or Spanish. For example rosarum = "de las rosas".

It's true what you say, that of course this terminology wouldn't apply completely because Voynichese wouldn't be a language in the traditional sense. It would be both synthetic (because you put a lot of stuff into one word) and artificial.

But then, even if we add "artificial", there's still the problem that there would be an apparent lack of diverse lexical items. You could combine your available bits of meaning in a variety of ways to express a near infinite number of patterns, but in my opinion it's hard to write about stuff without actually using its name. So yeah, maybe it all is in the drawings.


/offtopic I was reminded of this in an article I read about AI's which were developed to learn how to negotiate/barter with one another. The programmers forgot to tell them to stick to proper English, which resulted in them coming up with their own language system.
You are not allowed to view links. Register or Login to view.



Kind of reminds me of Voynichese 

@JKP

What you write is about another question - it's rather about whether the text is meaningful or meaningless.

I'm writing about whether the text is a flow of "language" (be that plain text or enciphered, does not matter) or not. In both cases there may or may not be cross-connections between sections.

Another counter-argument agaist the "synthetic notation" is the number of unique vords ("word types") which is 8114 according to Reddy & Knight (quoted per voynich.nu). Which is not that very low; the whole Shakespeare's vocabulary, for example, consists of only ~29000 word forms (You are not allowed to view links. Register or Login to view.).

(29-07-2017, 02:21 PM)Anton Wrote: You are not allowed to view links. Register or Login to view.@JKP

What you write is about another question - it's rather about whether the text is meaningful or meaningless.

I'm writing about whether the text is a flow of "language" (be that plain text or enciphered, does not matter) or not. In both cases there may or may not be cross-connections between sections.

Another counter-argument agaist the "synthetic notation" is the number of unique vords ("word types") which is 8114 according to Reddy & Knight (quoted per voynich.nu). Which is not that very low; the whole Shakespeare's vocabulary, for example, consists of only ~29000 word forms (You are not allowed to view links. Register or Login to view.).

No, I wasn't writing about whether the text is meaningful or meaningless.

I created a concordance to SPECIFICALLY map connections between tokens (basically a spider-web of tokens), to see if there were code-tokens to link together different sections or folios, rather than natural-language words. I can glance at this document (which is more than 2,000 pages long) and, in moments, tell you which words link to which ones in terms of form, subject matter (which section they're in), and in glyph-shape glyph/similarity.


For example...

I thought I had discovered a pattern to how big plants link to starred pages, but on careful re-evaluation, it's not consistent enough to draw any conclusions (or to yield useful information).

I also thought I had discovered a way to interpret the pages that have vertical columns and there MIGHT be, based on linking words, but it too is not clear enough to be sure.


When I did this, I was not worrying about the meaning of the tokens, or even assuming there was any. I was searching for patterns and connections.

Quote:I created a concordance to SPECIFICALLY map connections between tokens (basically a spider-web of tokens), to see if there were code-tokens to link together different sections or folios, rather than natural-language words.

Ah OK, now I understand.