The Voynich Ninja

56 examples in 31 pages, nearly two to one, and that suggests a hint of concentration, but a lot depends on other, unknown factors. Like the folio designations.  Which examples are close together? What are the individual vord examples and where are they located? What are the various examples for vord-2, and how do they compare?

Graphing some of these phenomena to the individual folios might be informative. Are there differences between scribes or is there no apparent distinction regarding this prominent aspect of linguistic content?

Good morning from germany!
If the individual "words" in the VMS actually represent only encrypted individual letters, then the word repetitions could also simply be Roman numeral characters...!?

The problem remains that there are many more different Voynichese words than there are letters in the alphabet. Like two orders of magnitude more.

I still think Roman numerals are a good possibility. But imagine we find a solid way to convert Voynichese to Roman numerals. This would basically mean we now have a long series of numeric values. What do we do with those?

No No, I'm thinking of an encryption similar to Trithemius "Hale Mary-Cypher"! 
The letters of the alphabet generate different words (perhaps combinations of two words?) during encryption.
This means that a "Voynich-word" could result in the individual letter "C" when decrypted. Three times then just "CCC"...!?

(02-04-2023, 09:30 AM)DrAntagon Wrote: You are not allowed to view links. Register or Login to view.No No, I'm thinking of an encryption similar to Trithemius "Hale Mary-Cypher"! 
The letters of the alphabet generate different words (perhaps combinations of two words?) during encryption.
This means that a "Voynich-word" could result in the individual letter "C" when decrypted. Three times then just "CCC"...!?

Yes, in a cipher like that CCC could result in something like 'daiin.daiin.daiin", thought I'd expect the repetition would instead be obscured by selecting different words for the three occurrences of C.
Another interesting feature of Voynichese is that consecutive words with an edit-distance of 1 are even more frequent than consecutive identical words (about 2% vs 1% IIRC).
What do we make of sequences like these?

qokeedy.qokeedy.qokedy.qokedy.qokeedy
qokeedy.qotedy.qokeedy.qokeedy.qokeey
otedy.okedy.okedy.qokedy
qokeedy.qokeeey.qokeeody
okeedy.qokedy.qokeedy

These patterns are considerably more frequent than in a randomly scrambled version of the text: there must be something that causes similar words to prefer occurring consecutively.

In case you have not seen it yet, You are not allowed to view links. Register or Login to view. presented a paper about the Hail-Mary cipher at the Malta Voynich conference last year.

(02-04-2023, 10:23 AM)MarcoP Wrote: You are not allowed to view links. Register or Login to view.These patterns are considerably more frequent than in a randomly scrambled version of the text: there must be something that causes similar words to prefer occurring consecutively.

In my opinion that also suggests, that the writer read the words (or word-combinations) from a table in which one variation was always under the next.

So are we talking of mapping each plaintext letter to one of 50 or so Voynichese words?

I imagine a table in which in the columns under each individual letter of the alphabet there was a long row of letter combinations of different lengths, which could then be combined to form a word of any length....

In that case I wonder if you could even find 20 or so discrete categories of "words". Or can some elements, like "dy" return in various tables?

Under the idea of a Hail-Mary cipher, similar Voynichese types are unlikely to appear under the same column (i.e. encode the same character). Long sequences of the same character repeated are rare in medieval manuscripts, basically only found in Roman numerals, and in that context they are limited to a maximum of four consecutive occurrences.

Fragments like these are hard to explain in that way:

<f75r.37,+P0>    qokedy.dy.sheety.qokedy.qokeedy.qokechdy.lol
<f75r.38,+P0>    qokeedy.qokeedy.qokedy.qokedy.qokeedy.ldy
<f75r.39,+P0>    yshedy.qokeedy.qokeedy.olkeedy.otey.koldy

A B C A A A D
A A A A A E
F A A A G H

<f42r.20,+P0>    shol.chol.shoky.okol.sho.chol.shol.chal
<f42r.21,+P0>    shol.chol.chol.shol.ctoiin.c'os.odan

A A B C A A A A
A A A A D E F

Another problem with anything based on a table of words (nomenclator) is the great uncertainty in Voynichese word separation. Entries in a nomenclator must be clearly identifiable. This could be achieved with a system similar to that discussed by Patrick Feaster, where written spaces are irrelevant and tokens are separated by a set of rules (e.g. 'q' always marks the start of a new token, or 'n' always marks the end of a token etc).

EDIT: in You are not allowed to view links. Register or Login to view., Jürgen Hermes shows part of Trithemius' tables. Here it can be seen that similar words are spread through different letters (which in You are not allowed to view links. Register or Login to view. appeared in rows). As discussed in the paper, this arrangement can result in a high number of consecutive similar words.
Illustration from the paper:

EDIT2: with the arrangement discussed by Hermes, the sequence qokeedy.qokeedy.qokedy.qokedy.qokeedy could be part of a word like GREENNESS

EDIT3: f42r.20 corrected as suggested by Nablator