+- The Voynich Ninja (https://www.voynich.ninja)
+-- Forum: Voynich Research (https://www.voynich.ninja/forum-27.html)
+--- Forum: Analysis of the text (https://www.voynich.ninja/forum-41.html)
+--- Thread: Currier A and B (/thread-120.html)
RE: Currier A and B - Anton - 09-09-2016
Quote:It does not matter if you reverse lookup between A&B as a whole / or A first and then B language words and compare them, the resulting words are the same.
You seem to misnterpret what I mean. As a side comment, we don't know if a vord is an "A language vord" or a "B language vord" until we ascertain that this vord's occurrences are limited to the A-folios or to the B-folios, respectively. But I don't speak of vords limited to A or B. I speak of vords that are unique to A (or B) and ask whether, at the same time, they are used in B (or A). (It is trivial that a VMS-wide unique vord would not occur elsewhere, so such vords are explicitly omitted from consideration as indicated in my items 5 and 6).
RE: Currier A and B - Davidsch - 12-09-2016
Anton, again:
it does not matter if you look at it left or right, vords or words: the resulting top 10-words remain the same and
i am interested to hear arguments for listing all the other lower frequencies.
RE: Currier A and B - Anton - 12-09-2016
I don't know if the top-10 is the same for Currier A and B (judging by Julian's thread it is not, but his methodology was not correct and he has not posted updated results yet).
The argument I expressed above - it is an attempt to shed more light on the nature of the underlying defference between A and B.
Frequent vords are frequent vords. It is of no surprise that they appear both in A and B. Like the word "rot" appears in English, Swedish, German etc. Moreover, two different plain text words may map into the same pair of overlay words with two different procedures. Like plaintext 1 via procedure A yields ciphertext 1, plaintext 2 via procedure A yields ciphertext 2, but, at the same time, plaintext 1 via procedure B yields ciphertext 2 and plaintext 2 via procedure A yields ciphertext 1. Furthermore, if some nomenclator is in place, then frequent words would just map into the same ciphertext in both A and B.
RE: Currier A and B - Davidsch - 13-09-2016
Quote:Frequent vords are frequent vords. It is of no surprise that they appear both in A and B.
..
Furthermore, if some nomenclator is in place, then frequent words would just map into the same ciphertext in both A and B.
yes, that is what i wrote. Is there anything left to do, let me know.
RE: Currier A and B - ReneZ - 13-09-2016
Both the combinations qo and ol are very frequent in the MS, so the word qol could be expected to be frequent as well, but it is not really.
It is found with high-frequency in the Bio-B text, and much less frequently in the stars/recipes (also B) text.
There are only sporadic occurrences in A language, more as parts of other words, and even fewer in herbal B.
On the other hand, the words chedy and shedy are very frequent in all B-language texts, including herbal-B, and basically non-existent in all A-language text.
While the former could be contents-related, the second appears dialect-related, or however one wants to call it....
RE: Currier A and B - Davidsch - 13-09-2016
Yes Rene,
See my old research: You are not allowed to view links. Register or Login to view.
there I used the TT transcription without Rosette.
There the words were counted in currier A alone, currier B alone and in the text A+B.
Resulting top words:
[attachment=604]
RE: Currier A and B - Emma May Smith - 13-09-2016
(13-09-2016, 12:09 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.Both the combinations qo and ol are very frequent in the MS, so the word qol could be expected to be frequent as well, but it is not really.
It is found with high-frequency in the Bio-B text, and much less frequently in the stars/recipes (also B) text.
There are only sporadic occurrences in A language, more as parts of other words, and even fewer in herbal B.
On the other hand, the words chedy and shedy are very frequent in all B-language texts, including herbal-B, and basically non-existent in all A-language text.
While the former could be contents-related, the second appears dialect-related, or however one wants to call it....
The latter is, of course, a subset of the notorious bigram [ed]. Practically unknown in A yet occurs over 4800 times in B.
I think that if we could figure out a reason for it the outcome would be very insightful.
RE: Currier A and B - Anton - 13-09-2016
That seems to be not in the least part caused by that dy itself is heavily shifted towards Currier B. 82,6% of all dy occurences fall into Currier B.
And ed is mostly used in edy.
***
What about average word legths in A and in B? Am I right that a B word is longer on average?
RE: Currier A and B - Emma May Smith - 13-09-2016
(13-09-2016, 09:55 PM)Anton Wrote: You are not allowed to view links. Register or Login to view.That seems to be not in the least part caused by that dy itself is heavily shifted towards Currier B. 82,6% of all dy occurences fall into Currier B.
And ed is mostly used in edy.
The [dy] occurrences in A (800) are mostly the trigrams [ody] (430) and [ldy] (115), and standalone [dy] (125). Compared with [edy], these other occurrences are much more similar in A and B. So [dy] seems like a normal part of both languages, but for some reason its combination with [e] really goes crazy in B.
Now, one of my theories posits that [y] and [a] are somehow linked, and the stats for the [ed] phenomenon seem to bear that out. Out of 5,000 instances of [ed]: about 4150 (83%) are [edy], about 520 (11%) are [eda], and about 200 (4%) occur at the end of words. Also, 98% of [edy] and 93% of [eda] occur in Currier B. Whatever is causing [edy] seems to also be causing [eda]. The simplest explanation is that [y] and [a] are variants.
Also, another one of my theories is that [y] is often deleted medially when followed by [d]. Thus the bigram [ed] is caused by a word ending [ey] having [dy] added to the end: [ey + dy = eydy = edy]. The same would happen if a word ended [chy] or [shy] and had [dy] added. And this is what we see: [chdy] is 91% language B and [shdy] is 90% language B. These trigrams, though less common, as part of the same phenomenon.
My theory would predict that Currier A should see higher rate of words ending [ey], [chy], and [shy]. While it seems to be true for [chy] and [shy], it isn't for [ey]. It would seem that [e] as a whole is more common in Currier B than A, and that more than one factor is changing to make the stats what they are.
RE: Currier A and B - julian - 15-09-2016
(12-09-2016, 02:14 PM)Anton Wrote: You are not allowed to view links. Register or Login to view.I don't know if the top-10 is the same for Currier A and B (judging by Julian's thread it is not, but his methodology was not correct and he has not posted updated results yet).
The argument I expressed above - it is an attempt to shed more light on the nature of the underlying defference between A and B.
Frequent vords are frequent vords. It is of no surprise that they appear both in A and B. Like the word "rot" appears in English, Swedish, German etc. Moreover, two different plain text words may map into the same pair of overlay words with two different procedures. Like plaintext 1 via procedure A yields ciphertext 1, plaintext 2 via procedure A yields ciphertext 2, but, at the same time, plaintext 1 via procedure B yields ciphertext 2 and plaintext 2 via procedure A yields ciphertext 1. Furthermore, if some nomenclator is in place, then frequent words would just map into the same ciphertext in both A and B.
I have been converting my analysis code to use the Takeshi EVA transcription, from GC's Voyn_101, hence the delay. I think the methodology was sound in that the comparison was between text on folios in Currier A and Currier B, initially with Herbal/Recipes but then in response to advice, more recently Recipes/Recipes.
The original interest in doing this was to see how Currier A "maps" to Currier B, in the sense of how to convert an A word to a B word so that the word frequencies in the converted A words matches those of the B words. It seemed to me that this might be a good clue as to how the cipher works.
So, as explained elsewhere, if the most common word in A is "daiin" and in B is "aiin", and we say they are the same plaintext word (seems reasonable), then the glyph "d" in Language A must be a null ... but this may be flawed reasoning :-)
I do think that the language differences are very important to try to understand.