The Voynich Ninja

(13-04-2020, 09:48 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.Interesting... I had mentally pictured You are not allowed to view links. Register or Login to view. instead of f115r. It also has a break with change of ink colour.

At the point of change on f105, there is rather a mess in the alignment - one of many places in the MS where the order of writing is rather mysterious/suspicious.

You are not allowed to view links. Register or Login to view. is another great example. The writing seems to literally end mid-word. The last word before the ink changes is [ot], which is very unusual.  I hope that Lisa's work can tell us whether this is a change between hands or just ink and pen.

I think that 105r is a change of ink and pen rather than of scribe...although it's also possible that the top part is Scribe 2 and the rest Scribe 3 (as on f. 115r). I went back and forth about 105r, but for my forthcoming piece in Manuscript Studies (coming out in early May) I committed to Scribe 3 for the full page. It is certainly possible that any further dialectal distinctions that are made based on my scribal corpora will help refine my work. For example, if someone discerns orthographic "rules" that are specific to my Scribe 3, and the darker-ink portion of f. 105r doesn't obey those rules but instead reflects Scribe 2 usage, that would then suggest that the top portion should perhaps be re-attributed to Scribe 2. That's exactly the kind of analysis I hope my work will spur.

(13-04-2020, 01:09 PM)LisaFaginDavis Wrote: You are not allowed to view links. Register or Login to view.I think that 105r is a change of ink and pen rather than of scribe...although it's also possible that the top part is Scribe 2 and the rest Scribe 3 (as on f. 115r). I went back and forth about 105r, but for my forthcoming piece in Manuscript Studies (coming out in early May) I committed to Scribe 3 for the full page. It is certainly possible that any further dialectal distinctions that are made based on my scribal corpora will help refine my work. For example, if someone discerns orthographic "rules" that are specific to my Scribe 3, and the darker-ink portion of f. 105r doesn't obey those rules but instead reflects Scribe 2 usage, that would then suggest that the top portion should perhaps be re-attributed to Scribe 2. That's exactly the kind of analysis I hope my work will spur.

I also hope that we can define scribes by their "language" now that your work better defines them by their writing. It really is a new outlook on the text which will spur a lot of research. The distinction between Currier A and B has already proven so useful that this further level of distinction (and the "why" behind it) can only be greater.

I feel that, in prospect of scribes being found to have distinctive dialects, we're beginning to answer some key questions about the text.

1. It's much less likely that the text is cryptographic. It wouldn't make sense for each scribe to have their own way of encoding the text. This is even more true if single pages are written by more than one scribe. Anybody reading the text would need to be able to identify the scribe and know their method of encoding.

2. The script is not a simple substitution for an existing script writing a well-established language. We more or less already knew this through statistical analysis. But the language shifting with the scribe reinforces the idea that there isn't a single ideal version of the language which is being transliterated (or at least they can't recreate it). The scribes are perhaps writing according to their understanding of the spoken language which differed between them rather than having a single standard. We might be looking at a language early in its history of literacy, which would also explain the invention of a new script.

3. The size of variation could indicate that the underlying language is natural. Were this to be a text written in a constructed language then the differences between individuals would be small or non-existent. A spoken language could contain a significant level of variation.

4. The variations themselves could indicate relatedness between glyph values. If we observe different scribes writing similar words systematically with one glyph rather than another then it indicates a sound change between their dialects. The different values of a sound change should be phonologically related in ways which obey known laws. /k/ might shift to /g/ but not to /m/.

I feel like I've wandered far from the original point so I'll end here. But it is enough to say that I'm very excited about the new research paths.

A strong relation between hand and language has of course been claimed by Currier's 1976 presentation.
He proposed two language (A and B) and two main hands (1 and 2) with several minor hands, and a complete 1-1 mapping of A with 1 an of B with 2.

One of the first outcomes of the new work could well be to 'officially' overturn this hypothesis.

I have long had doubts about both classifications. They seem to be strong and valid for the part that has been transliterated by Currier, i.e. about 40% of the text, but much less so for the remaining part of the MS.

Starting with the hands, I am no expert, but the slanted, slightly cramped handwriting that I see in the Herbal-B pages does not appear to me in the other main B-language pages of quires 13 and 20.
I am greatly interested to see the new classification.

With respect to the languages, the A vs. B split seems largely valid, but there are intermediate forms, and there are internal variations, especially in the B language.

Already many years ago I looked at this using a form of Principal Component Analysis on vectors containing bigram counts per page, with the result that people should have seen by now (  ).

[attachment=4208]

I recently found the software again and was able to reproduce this result.
It would be relatively easy to repeat this exercise, while not colour-coding the illustration type, but the 'hand'.
This would quickly show if there is a correlation, at least based on bigram counts.

Of course, there are other text statistics that might turn out to be representative, so this wouldn't be the final word.

Given that there may be more than one hand per page, I already prepared an update to the IVTFF format and my reading tool to cope with that.