(11-03-2025, 07:03 PM)Rafal Wrote: You are not allowed to view links. Register or Login to view.By, the way, I just noticed at the page You are not allowed to view links. Register or Login to view. some hybrid of "a" and "n" 
You are not allowed to view links. Register or Login to view.
That's a
u ('u') but there's only a few of them. I hate cursive.
I'm usually of the opinion that exceptions in a handwritten text of this size should be treated with caution. Otherwise I'd say this is some evidence that both the a-curve and the n-swoop are minim modifiers. And since those are positionally determined, they don't mean anything, lending credibility to the Roman numeral-like idea.
But these are exceptional, and we have no way of telling where the scribes were intentional vs sloppy.
Ah I see. There's a couple of fringe cases.
(just using Voynichese site so.. pinch of salt)
o - 21895
on - 5
oin - 26
oiin - 196
oiiin - 38
oiiiin - 1
a - 13614
an - 126
ain - 1717
aiin - 3861
aiiin - 111
aiiiin - 0
So you could make a pretty good guess until your most unsure stage "Xiiin", if that implies anything or not I don't know.
Yeah, it's pretty clear when you stay with iin-clusters.
But, to return to the original post, what about something like dal?
dal 253
dol 117
If "a" is indeed a ligature, does that mean that dal is actually deil?
I spent some time a few years ago studying the longest sequences that do not contain certain common symbols or symbol combinations in MS, this below is the longest sequence without 'a', according to
#=IVTFF Eva- 1.7
# ZL transliteration file, updated from EVMT project
# Version 1r of 11/04/2020
[
attachment=10156]
There seems to be slight overabundance of 'o's there, even one 'oo', which doesn't happen often in the MS. Some of these 'o's look sus, but they are midway between 'o' and 'a' at best, there is not a single clear 'a' there, I think.
The idea that [a] = [ei] is fairly widespread, I think. One point in its favor is that there are so many ambiguous cases that can't be resolved clearly as either [a] or [ei], but instead fall somewhere in between. If there were a meaningful distinction between [a] and [ei], these cases should have caused confusion (I'd argue that the absence of any radically stylized forms for common vords suggests that such distinctions couldn't generally be worked out from context). But apparently these cases didn't cause a problem (presuming Voynichese "worked").
For what it's worth, I suspect that [oi] and [ai] might also be interchangeable, on similar grounds, with [oi] being more deliberate and [ai] hastier or more cursive. If so, this could point to a dynamic in which the beginning of a line was slower and more laborious to encode (so that [oi] is more common there), while the later part of a line "fell into place" as the ending was reached (so that [ai] is more common there). I mean this to include cases with flourishes attached, e.g. [ar]/[or], [al]/[ol], etc. But I don't mean to suggest that [a] can be substituted for any [o] -- just an [o] followed by [i], with the anticipated [i] influencing the form of the second stroke of the [o].
Meanwhile, word-final [o] and [b] seem likely to be interchangeable as well, with [b] = [e*] and [n] = [i*] where [*] is a particular flourish. The variation between [o] and [b] would then be analogous to the variation between "closed" and "open" forms of [n]. But I'm not sure how to reconcile that with [oi] = [ai].
(12-03-2025, 09:35 PM)pfeaster Wrote: You are not allowed to view links. Register or Login to view.The idea that [a] = [ei] is fairly widespread, I think. One point in its favor is that there are so many ambiguous cases that can't be resolved clearly as either [a] or [ei], but instead fall somewhere in between. If there were a meaningful distinction between [a] and [ei], these cases should have caused confusion (I'd argue that the absence of any radically stylized forms for common vords suggests that such distinctions couldn't generally be worked out from context). But apparently these cases didn't cause a problem (presuming Voynichese "worked").
For what it's worth, I suspect that [oi] and [ai] might also be interchangeable, on similar grounds, with [oi] being more deliberate and [ai] hastier or more cursive. If so, this could point to a dynamic in which the beginning of a line was slower and more laborious to encode (so that [oi] is more common there), while the later part of a line "fell into place" as the ending was reached (so that [ai] is more common there). I mean this to include cases with flourishes attached, e.g. [ar]/[or], [al]/[ol], etc. But I don't mean to suggest that [a] can be substituted for any [o] -- just an [o] followed by [i], with the anticipated [i] influencing the form of the second stroke of the [o].
Meanwhile, word-final [o] and seem likely to be interchangeable as well, with [b] = [e*] and [n] = [i*] where
[/b][*][b]is a particular flourish. The variation between [o] and [b] would then be analogous to the variation between "closed" and "open" forms of [n]. But I'm not sure how to reconcile that with [oi] = [ai].
Yes, [b]
a could be a ligature of ei. In fact, that is how the scribes wrote the text for a in the VMS, using two written strokes. I would agree that the observations you make are valid; however, there is a question of when it is best to decide which option to choose. Presumably, any language written manually in a script form of the alphabet would intuitively display examples of this dilemma, since no two hands would be exactly the same. I would say that it is much easier to make a decision as to which letter(s) are the correct choice one the language is learned and clearly understood. When the basic language is unknown, and if presented with a choice of character(s) in manual handwritter, I would probably make a best guess initially, and then update if necessary once the language is well understood. Otherwise, trying to "fix" a decision before the language is clearly understood may take a great deal of effort. Better to finalize choices at the latter end of the analysis, I believe.
[/b][/b][/b]
(10-03-2025, 11:00 PM)Rafal Wrote: You are not allowed to view links. Register or Login to view.Quote:I have no idea, but as far as I remember, splitting 'a' into 'ei' (I think it's 'ei' rather than 'ci', since there is no bar)
You are of course right that it would be "ei" and not "ci" in EVA transcription. I must always remember that what looks like "c" is in EVA "e" 
Quote:If you replace all instances of "a" with "ci" the already struggling character entropy drops even further.
Yes, but it will become even more similar to Roman numerals. Maybe it is the right trail???
Take the most common word "daiin". If we assume that:
- "a" is "ei"
- "n" is a last, ornate "i" in the word
Then "daiin" become "8CIIII". Would it be some mix of Arabic and Latin numbers? "8 14"? Some field in a table marked by row and collumn?
And if "a" is "ei" then what is "o"? I suppose many of us would agree that these two signs behave very similarly.
Lots of questions, not much answers...
why is CIIII 14? I don't see it... I would say CIIII would mean 104. 8 * 104 makes 832, which neatly divides by 26 (832 / 26 = 32). Could it be that each letter is included 32 times?
(17-04-2025, 10:49 AM)ErinaBee Wrote: You are not allowed to view links. Register or Login to view.why is CIIII 14? I don't see it... I would say CIIII would mean 104. 8 * 104 makes 832, which neatly divides by 26 (832 / 26 = 32). Could it be that each letter is included 32 times?
[aiii] = CIIII would be 104 if the characters were read using their usual Roman numeral values. But since there aren't any equally common and recognizable tokens of V, X, L, and so on in Voynichese, there are a lot of numbers under 104 that couldn't be straightforwardly encoded in that same way.
That said, in a system like the one being proposed, Voynichese [c] could just as well represent 10, and a [d]-loop 5, and so on. Or [c] could even represent 12, and a [d]-loop 6, or something like that. There are still some problems with this -- what Voynichese vord would represent the number
two, for example? But a system could be Roman-numeral-
like without exactly following Roman numerals.
but then, what about the fact that 832 neatly divides by 26?