The Voynich Ninja

What do you make of the fact that words beginning in q are almost never found in labels?  That suggests that words beginning in qok- and words beginning in qot- are in fact related in some way.  But then the different ok:qok and ot:qot frequency ratios suggests an important difference.

Quote:I still don't see how this really explains it.  It seems like a good theory should account not only for which words exist, but also for how common they are.

@Sam G:

First of all a theory is good if they can explain at least one thing better then all competitive theories. So far I have only given three simple rules to describe the relations between similar words in the VMS. This three simple rules are not a theory and they did not explain the VMS. They only allow to estimate how common the words are compared to other words in the VMS.

Now to your question. The important point is that there is a network of relations between words in the VMS. Even the three simple rules result in a huge number of possibilities. Let us for instance check three possible changes for the word 'qokaiin'. This three changes are:
1) the gallow glyph replaced with another gallow glyph
2) the glyph 'in' replaced with 'iin'
3) groups starting with and without 'q'

With this three changes you have already a grid with 16 possibilities:
qokain (279) | qokaiin (262) | okain (144) | okaiin (212)
qotain ( 64) | qotaiin ( 79) | otain ( 96) | otaiin (154)
qopain (  -) | qopaiin (  6) | opain (  2) | opaiin ( 13)
qofain (  1) | qofaiin (  1) | ofain (  -) | ofaiin (  5)

With this grid alone it is also not possible to tell why 'okaiin' is more frequent then 'okain'. The reason is that each word in this grid is also connected to other similar words not shown here. 'okaiin' is for instance also related to 'daiin' and since 'daiin' (836) is four times as frequent as 'dain' (211) it makes sense that 'okaiin' is more frequent then 'okain' even if 'qokaiin' is less frequent then 'qokain'.

I understand that there are lots of similar words, but that does not mean that the words are meaningless.  I also don't see how we get from the rules you have proposed to a complete text generation method or how such a hypothetical method would explain the frequencies with which the different words occur.

To me, many of these patterns look grammatical.  For instance, it seems clear that okaiin is related to otaiin, okol to otol, etc., and then for some reason relating to the purpose of these words, the k-form calls for the q-prefix more often than the t-form does.

Like I said above I think it's also important to consider the fact that q is almost non-existent in the labels.  This also seems straightforwardly grammatical: q is something used in writing complete sentences but that is not needed in labeling individual items.

If we assume that the text is gibberish that has been produced by some mechanism, then there would seem to be two basic possibilities for explaining the absence of q in labels:

1) The creator of the text had an explicit rule in his mind not to use q in the labels.

2) The creator of the text did not have such a rule, but instead the absence of q is an unintended consequence of how the gibberish-generating procedure worked.

If the former case, we would have to then ask why the creator of the text would have such a highly specific rule, and what other such rules he might have had.

If the latter case, then surely the absence of q in labels must be an important clue to how the gibberish-generating procedure would have worked, and would impose significant constraints on the possibilities.  In particular, it would seem to be impossible that a word was generated independently of the other words in the context within which it occurs, something that already seems clear based on other considerations.

(07-01-2017, 11:14 AM)Sam G Wrote: You are not allowed to view links. Register or Login to view.For instance, it seems clear that okaiin is related to otaiin, okol to otol, etc.

I agree on this point. The gallows do seem to form some sort of distinct "letter-class" with similar functions.

Sam G Wrote:Like I said above I think it's also important to consider the fact that q is almost non-existent in the labels.  This also seems straightforwardly grammatical: q is something used in writing complete sentences but that is not needed in labeling individual items.

Sam G is right that q is rare in labels; that's important. For some reason, the process of encrypting a line necessitates the use of "q-" but the process to encrypt a label does not. The answer could be grammatical, but it's good to keep in mind that there are other explanations: perhaps it's part of the Line-As-A-Functional-Unit system or some other line-encrypting mechanism which doesn't apply to a single vord of text.

Sam G & Thomas Coon,
I agree with your observations regarding q.
The rarity of q in the labels and its absence as paragraph initial (but not as page initial) is the reason why I wonder whether q might stand for an ampersand, attached to the word which follows it.

(07-01-2017, 03:16 PM)VViews Wrote: You are not allowed to view links. Register or Login to view.Sam G & Thomas Coon,
I agree with your observations regarding q.
The rarity of q in the labels and its absence as paragraph initial (but not as page initial) is the reason why I wonder whether q might stand for an ampersand, attached to the word which follows it.

Yes, it might. Some thoughts I had on the q...

• It was pretty standard to connect et/ampersand to the following word in many languages and one of the shapes for it was a "7", which is not unlike the VMS shape.
  
• It may be something like quod/the/that or whatever is used as an article in its associated language, but not quite as generic as "the" in English, possibly more definite (like "that") to create emphasis. "Quod" was often abbreviated as "q" and some scribes wrote "q" with a pointy loop that looks very much like the VMS q (I have samples).
  
• It may be a marker, perhaps one that says, "The following/attached word is a name".
  
• If the spaces are contrived and there are nulls, it may be a "q" or any other letter.
  

Quote:Sam G Wrote:
To me, many of these patterns look grammatical.  For instance, it seems clear that okaiin is related to otaiin, okol to otol, etc., and then for some reason relating to the purpose of these words, the k-form calls for the q-prefix more often than the t-form does.

@ Sam G:

There is a whole network of relations between words for the entire VMS. For instance it is even possible to focus on a part of this network connecting okaiin with okol:
aiin (469) daiin (863) okaiin (212) qokaiin (262) kaiin (65) 
ain  ( 89) dain  (211) okain  (144) qokain  (279) kain  (48) 
air  ( 74) dair  (106) okair  ( 22) qokair  ( 17) kair  (14) 
ar   (350) dar   (318) okar   (129) qokar   (152) kar   (52) 
or   (363) dor   ( 73) okor   ( 34) qokor   ( 36) kor   (26)
ol   (537) dol   (117) okol   ( 82) qokol   (104) kol   (37)


Therefore it is not possible to describe it as grammatical forms like we know from human language.

Quote:Sam G Wrote:
I understand that there are lots of similar words, but that does not mean that the words are meaningless.  I also don't see how we get from the rules you have proposed to a complete text generation method or how such a hypothetical method would explain the frequencies with which the different words occur.

It seems that you are arguing against something I didn't said in this thread. For a text full of similar words multiple explanations are possible. Again, the three simple rules are not a text generation method. They are only a way to describe the relations for the words in the VMS.

Quote:Sam G Wrote:
Like I said above I think it's also important to consider the fact that q is almost non-existent in the labels. 

This is typical for the VMS. chedy as the third most frequent word for the whole VMS did never occur for pages in Currier A. Words ending in -m are typical for the end of a line. Words at the beginning of a paragraph start typically with a gallow glyph. It is possible to describe numerous patterns which words can occur on a certain part of the manuscript or on a certain place within a page or within a line. Since this is typical for the VMS there is no need to assume a special rule that q was not allowed for labels. 

There are two pages with q-labels. This pages are f66r and f75v:

<f66r.L.9;H>     qotesy (1 time)
<f66r.L.13;H>    qokal  (191 times)
<f66r.L.14;H>    qolsa  (1)

<f75v.L1.8;H>    qokal  (191)
<f75v.L1.16;H>   qoted  (4)
<f75v.L3.29;H>   qotedy (91)

It is interesting that in both cases multiple instances of q-labels occur. Moreover on page You are not allowed to view links. Register or Login to view. two similar labels qoted and qotedy occur on the same page. Both observations could speak for a local element in the text generation mechanism. A local element in the text generation mechanism would also explain sequences like in <f95r2.P.6> 'qotar chdy chdy qokar okar qokar' or <f112v.P.8>  'ocheor okor aiiin otaiin okal okar otal'

Here's another one I think, f89r1 top left

(07-01-2017, 05:38 PM)Torsten Wrote: You are not allowed to view links. Register or Login to view.There is a whole network of relations between words for the entire VMS. For instance it is even possible to focus on a part of this network connecting okaiin with okol:
aiin (469) daiin (863) okaiin (212) qokaiin (262) kaiin (65) 
ain  ( 89) dain  (211) okain  (144) qokain  (279) kain  (48) 
air  ( 74) dair  (106) okair  ( 22) qokair  ( 17) kair  (14) 
ar   (350) dar   (318) okar   (129) qokar   (152) kar   (52) 
or   (363) dor   ( 73) okor   ( 34) qokor   ( 36) kor   (26)
ol   (537) dol   (117) okol   ( 82) qokol   (104) kol   (37)


Therefore it is not possible to describe it as grammatical forms like we know from human language.

Well, you're just completely wrong on this point: patterns like these are actually extremely common in natural languages.  Compare the following table of the "ko-so-a-do" words in Japanese, for instance:

[attachment=1064]

I could come up with similar examples by using noun declensions or verb conjugations in many languages (including German) but I would probably have to make the tables from scratch.  Anyway, the ability to recombine elements like roots and suffixes in this way is an extremely basic phenomenon in natural languages, and frankly it surprises me that anyone would dispute this.

What I'd like to know is whether these kinds of patterns ever show up in ciphertexts or in random gibberish texts (if you can even find any examples of the latter).

Quote:

Quote:Sam G Wrote:
I understand that there are lots of similar words, but that does not mean that the words are meaningless.  I also don't see how we get from the rules you have proposed to a complete text generation method or how such a hypothetical method would explain the frequencies with which the different words occur.

It seems that you are arguing against something I didn't said in this thread. For a text full of similar words multiple explanations are possible. Again, the three simple rules are not a text generation method. They are only a way to describe the relations for the words in the VMS.

Okay, if you're not actually saying that the text was created by the rules you describe then I guess there's no real disagreement.  Surely the relations between words are interesting and relevant but I don't think they show that the text is meaningless or that it was generated via a simple procedure.  Quite the opposite, in fact, as argued above.

Quote:

Quote:Sam G Wrote:
Like I said above I think it's also important to consider the fact that q is almost non-existent in the labels. 

This is typical for the VMS. chedy as the third most frequent word for the whole VMS did never occur for pages in Currier A. Words ending in -m are typical for the end of a line. Words at the beginning of a paragraph start typically with a gallow glyph. It is possible to describe numerous patterns which words can occur on a certain part of the manuscript or on a certain place within a page or within a line. Since this is typical for the VMS there is no need to assume a special rule that q was not allowed for labels. 

Yeah, but all these other patterns need to be accounted for as well!  I was focusing on just one case because it was evident in your own examples.  Certainly accounting for all of these complex patterns is going to prove a difficult task if you assume the text is gibberish produced by some simple mechanism.  Just accounting for the patterns relating to q is not going to be easy, let alone all the others.

Quote:There are two pages with q-labels. This pages are f66r and f75v:

<f66r.L.9;H>     qotesy (1 time)
<f66r.L.13;H>    qokal  (191 times)
<f66r.L.14;H>    qolsa  (1)

<f75v.L1.8;H>    qokal  (191)
<f75v.L1.16;H>   qoted  (4)
<f75v.L3.29;H>   qotedy (91)

It is interesting that in both cases multiple instances of q-labels occur. Moreover on page You are not allowed to view links. Register or Login to view. two similar labels qoted and qotedy occur on the same page. Both observations could speak for a local element in the text generation mechanism.

There's a list of all the labels, in alphabetical order, here:

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

There's 10 total beginning with q.  4 of these are from the word column on 66r which are not really "labels" at all.  There are only two on You are not allowed to view links. Register or Login to view. (your transcription seems to be in error), and I suspect that these are not labels either but complete lines of text that have been broken up by the fluid streams.  I also think the occurrence of qkol on f89r1 is probably a y connected with a ligature, but even if we want to allow 6 (or even 10) labels containing q that's obviously much rarer than in the main body of the text.  And just look at how many of the labels start with o.  Obviously that didn't occur by random chance and therefore needs to be explained somehow.

Quote:Sam G Wrote:
Well, you're just completely wrong on this point: patterns like these are actually extremely common in natural languages

Your pattern for japanese is not like the network of similar words for the VMS.  I didn't know for a grammatical network connecting all words used in a language and I didn't know of a natural language where similar words occur with similar frequencies. 
An English text with similar features would consist of words similar to the words "the", "and" and "to". Additionally, words similar to "the" like "khe", "phe", "fhe", "tha", "tho", "thy", "thee" and "theee" would occur with frequencies similar to each other.  

The point is not that some words are similar to each other. The point is that all words occurring at least four times are connected to each other. There is no problem to add some more dimensions to the grid:
aiin (469) daiin (863) okaiin (212) qokaiin (262) kaiin (65) taiin (42) otaiin (154) qotaiin (79)
ain  ( 89) dain  (211) okain  (144) qokain  (279) kain  (48) tain  (16) otain  ( 96) qotain  (64) 
air  ( 74) dair  (106) okair  ( 22) qokair  ( 17) kair  (14) tair  (13) otair  ( 21) qotair  ( 6) 
ar   (350) dar   (318) okar   (129) qokar   (152) kar   (52) tar   (43) otar   (141) qotar   (63) 
al   (260) dal   (253) okal   (138) qokal   (191) kal   (23) tal   (20) otal   (143) qotal   (59)
am   ( 88) dam   ( 98) okam   ( 26) qokam   ( 25) kam   ( 9) tam   ( -) otam   ( 47) qotam   (12)
or   (363) dor   ( 73) okor   ( 34) qokor   ( 36) kor   (26) tor   (23) otor   ( 46) qotor   (29) 
ol   (537) dol   (117) okol   ( 82) qokol   (104) kol   (37) tol   (48) otol   ( 86) qotol   (47)

This grid alone already contains 7694 words. This is 20 % of the number of words in the VMS.

Quote:Sam G Wrote:
Surely the relations between words are interesting and relevant but I don't think they show that the text is meaningless or that it was generated via a simple procedure.  Quite the opposite, in fact, as argued above.

The observed similarities alone are not enough to decide what the VMS is and if it is meaningless or not.

Quote:Sam G Wrote:
Yeah, but all these other patterns need to be accounted for as well! 

Correct, a theory for the VMS should be able to explain all this patterns.

Quote:Sam G Wrote:
...that's obviously much rarer than in the main body of the text.

Why did you assume an uniform distribution for the words? No matter which words you will check they are not uniformly distributed over the VMS or over a page or within a line!