The Voynich Ninja

I do not know if this is the place, but i would like to ask some specific questions about his research and his paper (2014?)
Can not find another site where i can do this, i could try to email, but perhaps this works as well. Let's try. Is this the correct section?

I did not know about the paper until it was mentioned here, und i've read most of it this week.

It's an excellent approach and well written piece of research. Some points i do not fully understand and 
other things i wonder if they can be improved.

I see no problem about putting a discussion about Torsten's paper in this section. Please state your questions, and I hope Torsten could answer them.

Hi Davidsch,

 I will do my best to answer your questions.

Thank you for your time!
 
At this moment i am (still) struggling through the PDF, but since the document is from 2014,
perhaps things and insights have been changed or improved on your side and i would like to ask you to explain if possible
and where I understood incorrectly let me know.
 
a* you wrote, that there are several possibilities for glyphs to originate, as part of another glyph, by adding vocal letters with others etc.
but did you examine how do glyph-groups originate in essence ?
If you look at the whole text, it must be possible to define "basic word stems".  Did you look at such?
Otherwise I am willing to try that.
Note: for the programming I think it is best to leave out the labels and the very small words. I have such a transcription text if you need any.
 
b* the glyph groups are then changed. Are there exact rules for the change in those groups?
Then you define modifications. Change of a letter. Adding of a letter in the word. Adding of start-letter. Adding of a end-letter. Removal of a letter.
Did you write a program to analyse the exact changes on the stems, you can see the exact changes. If not, I am willing to give it a go.
 
Then the word-changes-count can be seen and we could see if there is a pattern.  If there is a pattern we have the method and we see if we can make a device for that.
(You wrote ED=edit distance, but i prefer word-change-count) Anyway, i would like to see if there is a pattern.
 
c* You write there is a maximum on the change of a word-stem. That maximum seems 4.
Perhaps I missed this research, but did you write more on this?
 
d* The low frequency characters in a “glyp group” (which i call word i assume) seem to be always on or near the4 same page.
Did you run statistics on all (most common) groups on the whole text ? That would give a nice “page cluster versus glyp group” diagram,
Which can show that  your assumption is right. That would mean that the page numbering is correct. Perhaps you can also see if there are pages “missing” based on the “missing glyp group changements”.   Anyway most probable we must be able to see that some pages belong to each other, because the groups are clustered.
 
e* You made a graph with word-glyph repetitions compared with the 1 to 20- previous lines. Is that based on your own research on are you referring solely on the work of somebody else?  Where is the data that you used for the VMS?
f* Further more, i do not understand why you choose to compare that with a poem. Do you have a program for these statistics, I would like to double check this entirely.
 
There are more detailed questions, but let’s start here.

As a start I will try to answer the first three questions about word similarities.

a*)
A "basic word stems" doesn't exist for the VMS. The words in the VMS build a multidimensional grid of similar words. Nearly all the words have a close “edit distance” to one of the three most frequent words: 'daiin', 'ol', and 'chedy'.  They occur with comparable frequency, whereas types which contain less frequent glyphs or bigrams in most cases occur less frequently. With other words, words close to the three core words occur more frequently. 

Even for the three most frequent or core words 'daiin', 'ol' and 'chedy' it is possible to demonstrate a relation. An example for a path between 'daiin' and 'ol' is 'daiin' - 'dain' - 'dan' - 'dar' - 'ar' - 'al' - 'ol'. An example for a path between 'daiin' and 'chedy' is 'daiin' - 'chdaiin' - 'chedaiin' - 'chedan' - 'chedan' - 'cheda' - 'chedy'" (see You are not allowed to view links. Register or Login to view. p. 7).
Therefore you can find a word 'dyaiin' in the VMS but not a word 'alphabet'. The Voynich alphabet has all the letters for writing 'alphabet'.  But you will never find anything like that in the VMS because it happens to be too many "edit-distances" away from the three core words. 
 
b*)
There are only common patterns. For instance you can replace glyphs with similar ones and you can add prefixes ('y', 'o', 'qo', 'ch', 'sh', 'l', 'ol').
I have manually ordered the words of the VMS by similarity and frequency in a grid (see the grid You are not allowed to view links. Register or Login to view. p. 66 - 82). Here is the part of the grid with the core word 'daiin' as starting point:
'daiin' (863) | 'aiin' (469) | 'dain' (211) | 'ain' (89)
'daiir' ( 23) | 'aiir' ( 23) | 'dair' (106) | 'air' (74)
'daiim' ( 5) | 'aiim' ( 3) | 'daim' ( 11) | 'aim' ( 7)
'daiis' ( 5) | 'aiis' ( 3) | 'dais' ( 4) | 'ais' ( 1)
'daiil' ( 1) | 'aiil' ( 1) | 'dail' ( 2) | 'ail' ( 5)
I used a program to count the words and to find the words to add next to the grid. The grid tells the relations and the rules for the words of the VMS.

But this doesn't mean that it is possible to define strict rules. For each possible rule you will always find at least one exception. For instance 'q' is normally followed by 'o'. But also words like 'qaloin' exist. Instead of 'qaloin' you would expect a word 'qolain'. Therefore it is no surprise that a word 'qolain' in the next line:
You are not allowed to view links. Register or Login to view.

c*)
There is a maximum of prefixes you can add to a core word, but this doesn't mean that the number of changes is limited.
Beside prefixes you can replace a glyph with a similar one. For instance it is possible to replace 'a' with 'y' and 'o', whereas 'y' normally occurs at the beginning or end of a word. But sometimes a glyph also occurs together with its possible replacement. This way you get words like 'dyaiin' ('dy' + 'aiin'). You are not allowed to view links. Register or Login to view.

a)

Quote:Q: .. but did you examine how do glyph-groups originate in essence ? 
A:  basic word stems" doesn't exist for the VMS. The words in the VMS build a multidimensional grid of similar words. Nearly all the words have a close “edit distance” to one of the three most frequent words: 'daiin', 'ol', and 'chedy'.  

Now i am confused, because you show any word can morph into any word, there is no basis.  

But, isn't there always a basis?  If there is no basis, there is also no group.

But you call them (glyph) groups, and there actually is no grouping anymore: 
any word can become any word in one or more steps.


daiin' - 'chdaiin' - 'chedaiin' - 'chedan' - 'chedan' - 'cheda' - 'chedy'
------------2-----------1-------------2---------?-----------1----------1  = 7

i can do that also quicker: 

daiin' - 'daiiny'(f2r) - 'chey' - 'chedy'
--------------1----------5---------1  =6

Or quicker:
daiin' - chedy
----------5


Edited: 
there are 2 glyph groups, which i seem to have confused with VMS words. 
group I : o,a,e,i,y,
group II: ch, sh, n, r, s, l, d, m, q, k,t,p,f

Actually when glyp group together (always) do exist, this is a word.


b)

Quote:A: There are only common patterns. For instance you can replace glyphs with similar ones and you can add prefixes ('y', 'o', 'qo', 'ch', 'sh', 'l', 'ol'). ... But this doesn't mean that it is possible to define strict rules.

We see there are some rules . Perhaps these can be refined. You defined some of them on page 5.
Another rule is the position of some of the letters.
And there is a pool of letters to choose from. 
I am trying to see if we can use domino-blocks to build us a word. on a domino block there are at least 2 letters. Perhaps more.
We will see.


c)

Quote:Q: You write there is a maximum on the change of a word-stem. That maximum seems 4. 
A: There is a maximum of prefixes you can add to a core word, but this doesn't mean that the number of changes is limited.

Beside prefixes you can replace a glyph with a similar one. For instance it is possible to replace 'a' with 'y' and 'o', whereas 'y' normally occurs at the beginning or end of a word. But sometimes a glyph also occurs together with its possible replacement. This way you get words like 'dyaiin' ('dy' + 'aiin'). 

You stated on page 7 that a maximum of four changes/ or occurences is obvious. I was referring to that.

If we look at the word modifications there is no real maximum, except the pool of letters and the rules we found. Correct?

a)

There was a typo in my last post. I have written 'chedan' two times instead of 'chedain' - 'chedan'. 
The correct path between 'daiin' and 'chedy' is therefore:
'daiin' - 'chdaiin' - 'chedaiin' - 'chedain' - 'chedan' - 'cheda' - 'chedy'.

The point is, that in each step only one element is changed. This means, that if you see a relation  between two similar words all the words in the VMS are related since all the words in the VMS are similar to each other. This also means that it is not possible for the VMS to detect a mistake, since it is normal that everything is changing.
(Note: That the EVA-transcription is using two characters to transcribe the glyph 'ch' doesn't mean that you can count this as two glyphs (see You are not allowed to view links. Register or Login to view.).

I call them (glyph) groups only to avoid to call them "words". Today I would call them glyph sequences instead.

b)

There is no doubt that you can define a set of rules. The only problem is, that such writing rules will only work for 99 % or sometimes 80 % of the words in the VMS.

c)

On page 7 I wrote "The fact that a word is not part of the grid only means that this word
occurs less than four times." With other words, I have stopped to add words to the grid after
adding all words with four or more occurrences in the VMS. The reason for doing so was, that 
for words occurring less than four times transcription errors become important. 

There is for sure a maximum of modifications for each word. The only problem is that this number is different for each word.

For 'dain' this number is 19 since so many similar words with an edit distance of one exists:
'dan', 'daiin', 'dair', 'ydain', 'daind', 'dais', 'daing', 'odain', 'ain', 'dainl', 'lain', 'rain', 'daein', 'dainy', 'doin', 'dkain', 'tdain', 'ldain', 'daisn'.

For 'daiin' this number would be 22:
 'dain', 'odaiin', 'daiiin', 'daiiny', 'aiin', 'daiim', 'doiin', 'daiir', 'ydaiin', 'raiin',  'doaiin', 'dairin', 'tdaiin', 'deaiin', 'daiis', 'laiin', 'dariin', 'daiino', 'pdaiin', 'ldaiin', 'daisin', 'dyaiin'.

But what does this numbers say about the VMS in your eyes?

a) ...sequences. 
b) 

Quote:There is no doubt that you can define a set of rules. The only problem is, that such writing rules will only work for 99 % or sometimes 80 % of the words in the VMS.

What would be interesting is building blocks of rules.
Finally we could find a physical thing (cardan grille, domino block, alberti wheel etc...) that could show how the words have been composed basicly.
Of course there are exceptions while writing.


c)

Quote:..
There is for sure a maximum of modifications for each word. The only problem is that this number is different for each word.
...
But what does this numbers say about the VMS in your eyes?

The maximum modifications on a word is flexible. The maximum modifications per PAGE. would be interesting to see what the max is.
A program would have to measure the modification "distance" towards all words, if you take the page as entity.
That means you only start to measure on line1, first word. Based on your paper, it is my guess that there is a solid maximum.
However, you should limit the program to length 4,5 and 6 to make things easier.


The outcome could mean several things, depending on the results, just some ideas:
* there is no maximum:  a codebook on words was used perhaps
* there is a maximum and a average line: the "words" are basis of a natural language, or the "words" are ciphered
* if there is a big fluctuation (no structure): the text has no meaning, it is just creative writing
* if there is a big fluctuation no structure: the main text is filling with no meaning, only very few letters or some structure has a meaning

Quote:What would be interesting is building blocks of rules.

Don and Wladimir have been working upon certain building structures. They shared their tables and charts in the Forum. I cannot say anything in favour or against those, since I haven't time to explore them in detail.

Thank you anton, i read those threads. 
What i would like to do, is define fixed rules, which are based on a scientific (read: verifiable) approach