I think this is a good argument against the Natural Language hypothesizers. Thank you Torsten.
That's nice work!
I am not so sure about ThomasCoon's conclusion, and I suspect that such a plot
for a longer piece of text in Pinyin (either with or without the numerical tone indicators)
might turn out very similar.
Another conclusion I think I see in these graphs (especially the top one for the entire
MS) is that the selection of the different groups is fairly arbitrary. There is one big cloud,
and there is no real boundary between the different groups. They are completely contiguous
everywhere.
In fact, in the Currier B plot there is a hint of some clustering, but inside the purple/magenta cloud...
Quote:I think this is a good argument against the Natural Language hypothesizers.
There is much what can be said about this network of similar words. If you think about what it is necessary to generate even 7000 different word types where the types has to differ in only one glyph from each other! In my eyes it is obvious that you can only reach such a goal with some methodical approach. For the VMS some more observations are important.
There is the shift from Currier A to Currier B which happens since words near "
chol" are frequently used in Currier A and word types near "
chedy" are frequently used in Currier B. This is an indication for a change of the word set during writing the manuscript. With other words this indicates that the words where generated during writing the manuscript.
(Another feature pointing in the the same direction is that the words used in the text is depend in some way on there position within a page or line. Some words are for instance typical for a specific position within a line.)
Last but not least it is possible to draw such a graph for each page of the manuscript. This demonstrates that similar words do co-occur for each page. This is an indication for a local element in the method to generate the text. See for instance the pairs
daiin/
dain and
chol/
chor on page You are not allowed to view links.
Register or
Login to view. in You are not allowed to view links.
Register or
Login to view.:
[
attachment=1108]
Quote:... the selection of the different groups is fairly arbitrary.
The groups are selected by the most common words "
daiin", "
ol" and "
chedy". If you take also the frequency for each type into account it makes sense.
See for instance "
daiin" and "
ol":
aiiin( 41) daiin ( 17)
aiin (469) daiin (863)
ain ( 89) dain (211)
an ( 7) dan ( 20)
aiir ( 23) daiir ( 23)
air ( 74) dair (106)
ar (350) dar (318)
ail ( 5) dail ( 2)
al (260) dal (253)
am ( 88) dam ( 98)
os ( 29) dos ( 1)
or (363) dor ( 73)
ol (537) dol (117)
The word types near "ol" are frequent and the word types near "daiin" are frequent. It is more as if the two groups of word types do overlap.
I have uploaded the gephi project for the graph for the complete manuscript. You need to download the free tool gephi from You are not allowed to view links. Register or Login to view. to open it.
[
attachment=1106]
(03-02-2017, 03:04 AM)ThomasCoon Wrote: You are not allowed to view links. Register or Login to view.I think this is a good argument against the Natural Language hypothesizers. Thank you Torsten.
On the Russian forum, I have found brief hypothesis which I have not seen before.
Alphabet has developed for itself a deaf-mute person.
In this case, the text simultaneously is both a synthetic language and unilateral encryption.
(03-02-2017, 10:41 AM)Wladimir D Wrote: You are not allowed to view links. Register or Login to view. (03-02-2017, 03:04 AM)ThomasCoon Wrote: You are not allowed to view links. Register or Login to view.I think this is a good argument against the Natural Language hypothesizers. Thank you Torsten.
On the Russian forum, I have found brief hypothesis which I have not seen before.
Alphabet has developed for itself a deaf-mute person.
In this case, the text simultaneously is both a synthetic language and unilateral encryption.
Wladimir, it's interesting that you should mention that. I have often wondered, when looking at the text on the last page, whether it might have been written by someone who was hearing impaired (not necessarily completely deaf), because of the way some of the words could be interpreted if one assumes that some of the sounds have been dropped. For the parts that are mostly but not quite German, it's quite hard to tell if it's written by a foreigner (for whom German was an additional language) or by someone with a hearing impairment.
Quote:Torsten : The word 'chedy' is the most frequent word in Currier B…, the third most frequent word for the whole VMS did never occur for pages in Currier A.
If the word chedy is really the first of the Courrier B and the third of the manuscript, it may be the verb to be or the verb to go? In Greek, for example, in both cases, it is ειμι: εισθα - you go, εστε - you are. Personally, I prefer "you go", it sticks well with the idea of a text instruction.
Ruby
(03-02-2017, 08:04 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.Another conclusion I think I see in these graphs (especially the top one for the entire
MS) is that the selection of the different groups is fairly arbitrary. There is one big cloud,
and there is no real boundary between the different groups. They are completely contiguous
everywhere.
In fact, in the Currier B plot there is a hint of some clustering, but inside the purple/magenta cloud...
Hello Rene,
Torsten has defined the colors in the following way:
To highlight words similar to "daiin", "ol" and "chedy" graph coloring is used. All nodes for a word that contains the glyph "i" are orange. Nodes for a word ending in "d" or "y" are purple. Nodes for words containing "ol", "or", "ar", "al" or end with "am" or "os" are green. All other nodes are colored using blue.
I think that the images do present some hints of clustering, but the clusters tend to appear as vertical regions in the diagrams, while Torsten's colors tend to be distributed horizontally.
The orange 'i' region seems to be somehow separated from the others.
The region I numbered 3 seems to correspond to sh- and ch- words.
Many thanks to Torsten for sharing these very interesting graphs!
Torsten I wrote it many times before: I think you need comparative languages and large corpora of them.
But I see no other comparative languages in your research.
Perhaps a piece of text that has been formed around a specific theme by a specialist, for example on plants.
I think you are all confused by these graphs: it does not show anything conclusive.
It only shows that some words are more in other sections than others.
I agree with David: it would be interesting to see how this compares to real texts, especially some in languages which are said to be close to Voynichese in relevant aspects, and/or in text types where the author repeats certain sets of words often.
It's also not completely clear to me how the only conclusion can be that the text is artificially generated. I understand why straightforward interpretation as a normal text in natural language is problematic, but I'm not certain why it should point towards artifical construction.