The Voynich Ninja

I made the following statement elsewhere:

(23-02-2017, 11:35 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.Suggesting that some symbols represent vowels assumes that the symbols represent either letters or sounds.
Both are natural assumptions, but I have very severe doubts about them.

The following symbols:  q  f  p  m  y   are demonstrably not to be identified with letters.
That's five out of (say) 25. How confident can one be that the others are?
And even if they are, what to make of a mixture of letters and non-letters?

This (naturally) caused some reactions.

To discuss such things in a meaningful way is usually difficult, because, in order to be precise, one really has to write in great detail, and chances to be misunderstood or misinterpreted are always great.

(23-02-2017, 06:47 PM)Emma May Smith Wrote: You are not allowed to view links. Register or Login to view.I do not know why the characters you state are "not to be identified with letters".

I find absolutely no problem with both q and y being straightforward characters which represent phonemes. The characters f p m can be linked to other characters and have strong positional rules, suggestive of being variants.

I would like to stress that, when I wrote "letters or sounds", for me these are two different things.
Sounds = phonemes.
With letters I mean characters of an alphabet, or as in German "Buchstaben".

It is possible to show that the five I mentioned (but also the less frequent g ) are not representations of plain text characters, but the reasons (arguments) are different, and fall into three classes.
The same arguments cannot necessarily be applied to the question whether these symbols could represent phonemes, so I specifically wrote "letters". However, there are also problems with that.

Many proposed solutions of the Voynich MS text include a table that translates the Voynich symbols to plain text characters, and then present bits and pieces of plain text. (Note that "many proposed solutions" are not You are not allowed to view links. Register or Login to view.   here. I am talking about specific cases, including solutions that I have received by Email over the years).

If such a table maps the 5 or 6 characters in question to plain text characters (and especially if f p m g are not variants of other characters) there can be no doubt that this is wrong.

There is (much) more to be said about this, but not right now.

Rene: it's good that you make this thread because I also wondered what you meant precisely.

I don't fully understand yet.

I see why it is problematic if someone maps one-to-one sounds to Voynichese glyphs. But does that mean that they are not letters? Couldn't we just explain this by positing "several-to-several" mapping?

In English, for example, several-to-several also works. The sound "k" can be represented by "c, k, q, C, K, Q" and I probably missed some. And the reverse is true as well. "k, K" can also be a null for example, and "C, c" can be an s. 

The difference is that in Voynichese these varieties are very position dependent. A "k" will be represented differently in word-initial position than it is in the middle of a word, and so on. 

So if you say that Voynichese can't be the result of one-to-one substitution, I agree. But I don't see how you can make the step from there to saying that gallows aren't letters.

Koen, if you look back at some of my previous posts about the text, you'll notice that I almost always mention that some glyphs may function as "modifiers" or "markers", which are similar (sometimes overlapping) but not identical concepts. I just assumed that anyone who had really studied the text would understand what I mean, but not everyone has concentrated on the text, so perhaps it's worth mentioning what I mean when I say that.

In natural language, there are no precedents for the behavior of EVA-y. Notice that it occurs mainly in the last position, sometimes in the first position, and rarely in the middle of vords. It doesn't behave as letters do in syllabic languages or abjads. It does LOOK like a Latin abbreviation (perhaps intentionally), but its behavior is too un-natural-language-like even if it's expanded as a Latin abbreviation. It is simply too frequent and regimented. Many languages have a vowel at the ends of words, but it is too constrained even if it represents several vowels, and occurs too infrequently in the middle of vords for something that's mapped to vowels.

So... some people might make the argument that maybe EVA-y is a vowel (or several vowels) and something else is used to represent vowels in the middles of words. But as soon as one makes this kind of argument (that more than one glyph can represent a specific letter or kind of letter), then the necessary character set becomes large, and the VMS character set is not large. The same argument applies if it's a consonant (it's less likely that a specific consonant would occur at the ends of words than a set of vowels, but it's possible)—the character set would need to be larger than the existing VMS glyph set.

If one then argues that EVA-y is a null (which is quite possible), then again one is plagued by a reduced character set and unusually short words and the only way one can reconcile the increased difference from natural language it is by re-evaluating the spaces.


Also, in natural languages, there are no precedents for a letter occurring as does the VMS "o", not even in languages where "al-" or "the" or various articles or joining words occur very frequently at the beginning (note that in some languages "and" is attached to the beginning of a word rather than standing separately). It is possible that "o" functions as marker, modifier, or conjunction (or something similar), but if it does, and if one still insists (as some do), that it's a vowel, then not all of the "o" glyphs function in a similar manner and one would have to investigate the possibility of dual-purpose glyphs (which is not an unreasonable idea, give that the position of glyphs is one of the more striking aspects of the text). In other words, if "o" functions as a vowel in some parts of the manuscript (and I haven't seen a convincing argument to prove this yet but will accept it as one possibility), then I'm pretty sure it is ALSO doing something else.


Now... all these arguments are very specific in the sense that they apply only if one maintains that the VMS is natural language. If it is not (and I am leaning about 65% in the direction of "not"), then there's no need to impose natural language expectations on the text and it can be structured any way the creator wants.

Rene is specifically talking about the single-leg gallows here. There are quite different factors to consider with double-leg gallows.

I would also add that I think that the position that says o and a are necessarily vowels is badly mistaken: vowel identifying algorithms don't see them as vowels, so why do linguists persist in wanting them to be so?

A year ago, I posted this on the forum.

This is Voynichese. It's an alternate glyphset that I created a few years ago. It's an actual passage from the text—all I did was change the shapes and give a shape to the spaces. I even kept some of the more unusual glyph-shape relationships (I didn't want to, I wanted to change those too, but since this was done for explanatory reasons, I thought it might help a person's mind see the relationships between this and VMS text more easily if I provided some contextual anchors).

If a person is sincere about studying the Voynich text, one should at least temporarily divorce oneself from the assumption that VMS vowels are vowels and that spaces are spaces.

Thanks for starting this thread, as I feel this could be an interesting discussion. Let me make clear the meanings I attach to certain terminology, and then I'll discuss the characters you mention.

First, I avoid the word 'letter' because I feel it is too definitive by referring to the writing of language. I prefer that each separate mark is called a character, though glyph is also fine. We don't know fully how to parse the characters, though most would appear to be discrete and the total number is finite. Most of the text is made up of a small number of identifiable characters, and this is true should the text be linguistic, a cipher, or some kind of procedurally-generated hoax. So all characters are characters by definition.

Next, the issue of sound only occurs if we believe that the text is a plainly-written language (a cipher text would have underlying sound, but not necessarily linked to individual letters and not recoverable by the simple study of the text's surface patterns). All scripts represent linguistic sound to some degree. If we make the assumption that the text is linguistic then it follows that we must determine what kind of script we are observing. The only possible answer, due to the relatively low number of common characters, is a script which is mostly phonemic.

(For those with no linguistic background, a phoneme is a sound perceived by speakers of a language. If speakers consider two sounds to be the same, then those two sounds together constitute a phoneme, even if speakers of other languages consider them to be different sounds. For example, the /l/ sounds in English 'milk' and 'lime' are different, but English speakers generally consider them to be the same (and would be surprised to be told they're not!). In the context of Voynich research I doubt the distinction between sound (what a linguist would call a 'phone') and phoneme is terribly important.)

So, the question we get down to is, in a linguistic analysis, can all the characters be explained phonemically? That is, the pattern of the characters within the text must be explainable as a fair representation of phonemes within language. Note that we, for the moment, don't have to explain the pattern of phonemes with reference to a specific language, only that the patterns can be linguistic or language-like. This is important because, even were the text not a natural language, if the writer wanted to create something which could be spoken he would still have to abide by linguistic rules because it would be spoken by humans. Remember, Klingon and Sindarin are human languages, just not natural ones!

For many characters in the Voynich script there seems to be little doubt or disagreement that they are phonemically plausible. Even strings such as eee or iin could easily be explained as trigraphs (or poorly parsed single characters) without serious issue. Of the characters mentioned above by Rene, I would say that they fall into two groups: q and y are core characters whereas f p m g are marginal. I see no reason why q and y cannot be phonemic, and I think arguments as to their oddness betray a lack of linguistic imagination.

For f p m g there are clear environments in which they occur which do not undermine linguistic explanation but rather shift it from the characters to the environment. We know that the beginnings and ends of lines show different patterns, with the latter being the major locations of m and g. We can turn them into a single phenomenon by saying that the line break is a shared environment, and that whatever conditions the ends of lines also conditions the beginnings. We still won't know what it is, but we can show that the distribution of those two characters is not a feature just of those characters. Likewise with f and p, their environment needs to be explained and not necessarily the characters.

I'm sorry if this is a lot of words, but I am happy for the discussion.

(23-02-2017, 11:09 PM)nickpelling Wrote: You are not allowed to view links. Register or Login to view.Rene is specifically talking about the single-leg gallows here. There are quite different factors to consider with double-leg gallows.

I would also add that I think that the position that says o and a are necessarily vowels is badly mistaken: vowel identifying algorithms don't see them as vowels, so why do linguists persist in wanting them to be so?

Sukhotin's algorithm does see a and o as vowels. The outcome may not be strong, but that's a long way from what you're suggesting here.

One thing that does not support the idea of gallows p and t representing letters is the gallows coverage. I don't remember offhand if f and k exhibit gallows coverage anywhere, but p and t definitely do.

In the following, let me use 'character' for the symbols we see in the Voynich MS.

I did use the term 'letter' (the second 'letter' in the thread title) to specifically refer to letters of an alphabet in the supposed plain text. Not to sounds or phonemes.

An example may help. The following figure is from Heidelberg CPG 597, an alchemical book from 1426. It has some encrypted passages. Some people will have seen this.

[attachment=1171]

It says: "In dem namen des vaters und des suns und des heiligen geistes so fach ich schuldiger man dizes puch an".

We can reconstruct the table that the author used to encrypt this line.

This is not what happened in the Voynich MS, or at least  with the 5 or 6 characters that I highlighted.

Let's look at the case of f and p. Their special property is that they appear predominantly in first lines of paragraphs.

Here (and in general) it is important to approach it from the side of the author / encryptor.
If one looks at it from the side of the decryptor, one may simply do a substitution and end up with some plain text that has the property that some characters only appear in the top lines of paragraphs. This may not even be noticed.

However, for the encryptor, if he was working from a plain text, then this plain text would have already had the property that some characters appear in top lines only, which is not reasonable.

So, what could they be? The well-known example from Cappelli's dictionary gives one suggestion. These are embellishments. Thus f and p could also be embellishments, i.e. either elaborate forms of other characters, or the embellishments themselves, i.e. effectively nulls.
In both cases, we need to reduce the alphabet of Voynichese by two counts.

In this scenario it is possible, of course, that the author / encryptor was working from a handwritten plain text with such embellishments, and he used f and p to represent them.
(He wouldn't have been working from a printed plain text.....)

The case of these two characters is made even more interesting by the fact that they are happy to appear (in enlarged or embellished form) as first characters of paragraphs, and the number of different characters that appear in this position is so small, that they look more like pilcrows than real letters.
This may have some implications for the other characters that can appear here, notably k and t.

I do lean towards the 'embellished way to represent same sound' explanation. And of course, an embellished P for example might double as a pilcrow.

But I agree that the phoneme inventory becomes rather limited then. If I understand both JKP and Rene correctly, then problematic aspects could be explained somewhat, but all these explanations have the drawback that we'd expect a much larger glyph set if they were true. If that is indeed what is argued, then I fully agree. And anyone defending a natural language hypothesis should take this into account. 

Now there are some ways to explain this. One might be that certain glyphs, like the unusual o's JKP mentions, function like diacritics, modifying the sound of the following glyph. But to be honest I have not studied this enough to know whether it's feasable or not.