The Voynich Ninja

Stolfi was right from what I have seen regarding "m" seeming to need an "a" or "o" precursor and "g" not so much, I believe this is because "m" acts like a fancy "r" and "g" acts like a fancy "d", or I suppose you could imply the same about "r" and "s" as more direct matches (alterations of the same glyph shape using \ or C as a base). [edit: Just read Renes response, yeah "g" could as easily be more like "y", I guess the overall point is just that is does not act like a glyph built up from "\" but one from "C"] 

I don't see why they both couldn't represent something similar, but I have found one requires a/o more than the other. I guess you could say the same of English "ing" and "t(/s)ion" so whether they are in-clusters or just suffixes I don't know, I shall bow (plead? ) to the linguists

(21-10-2024, 03:37 PM)MarcoP Wrote: You are not allowed to view links. Register or Login to view.The top result for 'g' (and 6th for 'm') is the empty string, where removing -g/-m results in a word that occurs in the page (e.g. You are not allowed to view links. Register or Login to view.).

This is interesting and unexpected. 

For the case of 'm', the sixth place is quite low, given that:
- 'm' is most frequently preceded by 'a'
- words ending 'a' are rare
- the 1st to 5th place in this list are all occupied by characters or short strings that tend to follow 'a' and are equally at the end of words.

For the case of 'g', the empty string is a serious possibility, but the 2nd and 3rd case are just about as frequent.

From my point of view, first of all, the suggestion that m and g should be abbreviations is very unlikely. Nothing in the MS suggests that abbreviation was on the author's mind. It is almost as unlikely as the suggestion that they should be independent characters by themselves.

Secondly, because of their very specific preferable use at line ends, seeing them as 'beautifications'  remains my top favourite. 
For 'm', my preferred l/r are 1st and 3rd in Marco's list and for 'g' my preferred y is shared 2nd.

Patrick: I see what you mean. I agree that it's prudent to keep all options open, even though at the moment I'm more inclined to explore what happens when certain glyphs are made equivalent to one another.

It probably makes sense to consider the behavior of [p] and [f] in this context as well. The existence of Grove words seems to point to at least one instance where the unexpected glyph is "extra", which could be seen as an argument in favor of your view. However, [p] and [f] are not extra in other cases, where they seem to replace something else. Moreover, there's a pretty good parallel in paragraph or "item" markers in the behavior of Grove words, so they don't need any special pleading. It may then be that the esthetic of this paragraph marker is maintained throughout the top line.

Also we obviously don't need simple substitution for positional variants or abbreviations to be relevant. They could be examined independent of the system. The structure of Voynichese words so strongly suggest that [m] is a part of the word, it's needed to "finish" the word in an appropriate manner. So it's probably a line-final variant of something else. The question remains why this variant is only triggered in some cases, and whether there is anything more to it.

Thank you, Rene. Indeed one of my errors caused an overestimation of the empty string. I fixed that and here are the new figures (there likely are more errors).

==> _results.m.freq <==
_r 281 11.87157
_iin 238 10.05492
_l 233 9.84368
_in 164 6.92860
_ir 89 3.76003
_ly 41 1.73215
_ 30 1.26743
_ldy 30 1.26743
_n 28 1.18293
_ry 24 1.01394

==> _results.g.freq <==
_l 13 4.12698
_ 13 4.12698
_y 12 3.80952
_dy 12 3.80952
_r 11 3.49206
_in 11 3.49206
_iin 10 3.17460
_s 8 2.53968
_ol 6 1.90476
_or 6 1.90476

Thanks Marco, that's really interesting!

Given the fact that several options are of about the same magnitude, would this indicate that m could indeed stand in for multiple glyphs?

Also, do I understand correctly that IN-groups would come out on top in both cases if they are considered as a whole?
(In no way do I have any personal preference here, this is one of those things I haven't given much thought before and I gladly accept others' opinions).

(22-10-2024, 08:50 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.Also, do I understand correctly that IN-groups would come out on top in both cases if they are considered as a whole?

That's what I understand as well. About 20% of the occurrences of final -g are of the forms -ag -aig -aiig, resulting in matches with -ain -aiin words.

There's more to Eva-g than I first thought...

Even though it appears only about 160 times in the text, which is almost exactly 1 out of every 1000 characters, it may still tell us something.

The ZL and GC transliterations seem largely consistent about this character.
ZL (v.3a) has 170 and GC (v.2a) has 164. I counted these by searching for the STA code B4.
The RF file (v.1a) which is an automated merger of these two has 166.

Using the RF file in the following, and counting uncertain spaces as spaces, g is followed by a space (or end of line) in 95% of all cases.
What precedes it is more diverse. I remembered wrong when I thought that this is mostly Eva-d.
Most frequent are (count / %):
a   41 / 24.7%
o   27 / 16.3%
e   27 / 16.3%
l   18 / 10.8%
d   14 / 8.4%
(spc)  13 / 7.8%

This last case is interesting. Most of these are g at line end preceded by a space.

Having seen some (but not all) of the evidence, I am indeed inclined to believe that the g can have different functions or roles. Sometimes it just appears like a meaningless addition or null. In other cases it appears to be a beautified y, but I won't exclude other substitutions as well.

What's really important is that, in spite of the similarity between the independent transliterations, a visual inspection of the actual handwriting shows that there are different forms of g. One clearly has a c shape to start with, while the other is more similar to m. The same situation as with Eva-r and Eva-s really.

To be continued...

While the following was an argument why the characters m and g are not likely to be abbreviations, to which I tend to agree, it also presents a far wider issue.

(20-10-2024, 07:41 PM)pfeaster Wrote: You are not allowed to view links. Register or Login to view.(2) Many glyphs have stronger or weaker "positional preferences"; [m] and [g] are just a bit more conspicuous about it than some others.  Explaining these two cases as abbreviations wouldn't resolve the broader "positional preference" problem.

Indeed, the most obvious positional preferences (that are not on a word level) are those of Eva-m and g, and of Eva-f and p. It is not too difficult to come up with various possible equivalences that might explain these. This does not mean that we *know*, just that we can think of possible explanations.

However, the more subtle feature that has been described under the heading 'rightward and downward' is far more difficult. None of the proposed translation solutions (ever!) goes anywhere to explain this.


Some options that I can think of are:
- The entire text is meaningless.
- Only (say) the leftmost part of the text is meaningful and the rest is filler
- The encoding scheme allows for some degree of freedom by the encoder

In all three cases, the gradual changes would just be the result of 'natural human drift', for lack of a better term.

What these three options have in common is that they would also explain another major problem with the text: the lack of repeated sequences (of words).
I think in general that any explanation for 'rightward and downward' will also explain this lack of repeating sequences.

The other direction is not necessarily true.
Some amount of errors or spelling freedom will destroy the repeated sequences, but would not easily cause the 'rightward and downward'  problem.

Note that more possibilities than the above three undoubtedly exist.