Today, 10:06 AM
Last weeks I have been deeping into the statistics of the VMS. I got stuck, as always... then I thougt that I'd better look at what the scribe never does, rather than at what he does... and found a new important rule (at least, I think it is new in this form, and important).
Inside a word, many glyph pairs never occur: n is never followed by k or t, y is never followed by o, there is no oo, no ya. Everybody knows this. What I found is that these same prohibitions keep working across the space between words. And this theory explains some curiosities of the VOynich text. When the last glyph of a word and the first glyph of the next one would be one of those forbidden pairs, the scribe puts a glyph in between. When the pair is fine, he mostly doesn't.
It works like "a pear / an apple": the extra letter only shows up where two letters would clash.
An example: The word taiin. When the word before ends in n, r, d, s or m (glyphs that cannot stand before a gallows inside a word) it is written ytaiin, 23 times out of 23. When the word before ends in o, y or l (ot, yt, lt are normal inside a word) it is written taiin, bare, 21 times out of 30.
Same with kaiin (ykaiin 80% of the time after a clash, 36% without), aiin (daiin 94% against 45%), okaiin (qokaiin 73% against 34%), ar (dar 80% against 23%).
I detected four glyphs and four clashes:
The known "-y then q-" (Currier) and the "yk / yt after a drawing" (MarcoP and Emma May Smith) are two cases of this.
I wrote the tests down before running them, and they all hold: the same word in both contexts, each of the three hands separately, Currier A and B separately, and on text I hadn't looked at when I found it (rings and labels). It is graded: the rarer the pair inside a word, the more often the extra glyph appears. And if you look at the word two positions back instead of the neighbour, the effect vanishes. Only the word written right before counts, and the line break resets it.
It is not a hard rule. Of all the word pairs that would clash, two thirds get the extra glyph and one third don't. For a fifteenth century hand I think that is what a rule looks like (this is something that I suppose; I mean, ortography rules may not have been as strict as nowadays)
But the same glyphs do other things too: y and o before a gallows also mark a restart: after a drawing gap, at the start of a label or of a radial line, a gallows word carries y or o almost always (95-100%), with nothing before it to clash with. q is the opposite, it almost never starts anything (12 labels in the whole book begin with q), but inside a paragraph it is at 58% at the start of a line, whatever the line above ended in. d is at 98% at the start of a line.
So, what about the space between the words? I have the position of every glyph on the scans, so I checked whether the extra glyph depends on how far apart the two words are. It does. q is a "close" form: after a clash, 82% of o-words carry q when the two words are close, 34% when the space is wide. y and o before a gallows are there at any distance after a clash, but without a clash they appear when the space is wide, as a restart. d doesn't care about the distance. And a clash left unrepaired comes with a slightly wider space.
The space between words is not a clean cut. The constraints that hold inside a word keep acting across the space, and the scribe had four glyphs to deal with it. If there is a language behind the text, this is what sounds do when speech runs on and not when it pauses (French liaison does exactly that), and the q that only appears when the words are written close together fits that picture rather well. If there is no language, it is a procedure that looks one glyph back, and at the pen, before choosing the shape of the next word. I can't tell these two apart. What I can say is that the exceptions are not noise: they follow the position and the distance, each glyph in its own way.
Comments and constructive attacks (
) welcome, as always.
Inside a word, many glyph pairs never occur: n is never followed by k or t, y is never followed by o, there is no oo, no ya. Everybody knows this. What I found is that these same prohibitions keep working across the space between words. And this theory explains some curiosities of the VOynich text. When the last glyph of a word and the first glyph of the next one would be one of those forbidden pairs, the scribe puts a glyph in between. When the pair is fine, he mostly doesn't.
It works like "a pear / an apple": the extra letter only shows up where two letters would clash.
An example: The word taiin. When the word before ends in n, r, d, s or m (glyphs that cannot stand before a gallows inside a word) it is written ytaiin, 23 times out of 23. When the word before ends in o, y or l (ot, yt, lt are normal inside a word) it is written taiin, bare, 21 times out of 30.
Same with kaiin (ykaiin 80% of the time after a clash, 36% without), aiin (daiin 94% against 45%), okaiin (qokaiin 73% against 34%), ar (dar 80% against 23%).
I detected four glyphs and four clashes:
| word before ends in | next word starts with | glyph put between | with clash | without |
| n, r, d, s, m | a gallows (k, t, p, f) | y | 69% | 23% |
| n, r, d, s, m | a gallows (k, t, p, f) | o | 95% | 70% |
| o, y | o (okaiin, ol, okeey) | q | 63% | 23% |
| o, y | a (aiin, ar, al) | d | 89% | 34% |
The known "-y then q-" (Currier) and the "yk / yt after a drawing" (MarcoP and Emma May Smith) are two cases of this.
I wrote the tests down before running them, and they all hold: the same word in both contexts, each of the three hands separately, Currier A and B separately, and on text I hadn't looked at when I found it (rings and labels). It is graded: the rarer the pair inside a word, the more often the extra glyph appears. And if you look at the word two positions back instead of the neighbour, the effect vanishes. Only the word written right before counts, and the line break resets it.
It is not a hard rule. Of all the word pairs that would clash, two thirds get the extra glyph and one third don't. For a fifteenth century hand I think that is what a rule looks like (this is something that I suppose; I mean, ortography rules may not have been as strict as nowadays)
But the same glyphs do other things too: y and o before a gallows also mark a restart: after a drawing gap, at the start of a label or of a radial line, a gallows word carries y or o almost always (95-100%), with nothing before it to clash with. q is the opposite, it almost never starts anything (12 labels in the whole book begin with q), but inside a paragraph it is at 58% at the start of a line, whatever the line above ended in. d is at 98% at the start of a line.
So, what about the space between the words? I have the position of every glyph on the scans, so I checked whether the extra glyph depends on how far apart the two words are. It does. q is a "close" form: after a clash, 82% of o-words carry q when the two words are close, 34% when the space is wide. y and o before a gallows are there at any distance after a clash, but without a clash they appear when the space is wide, as a restart. d doesn't care about the distance. And a clash left unrepaired comes with a slightly wider space.
The space between words is not a clean cut. The constraints that hold inside a word keep acting across the space, and the scribe had four glyphs to deal with it. If there is a language behind the text, this is what sounds do when speech runs on and not when it pauses (French liaison does exactly that), and the q that only appears when the words are written close together fits that picture rather well. If there is no language, it is a procedure that looks one glyph back, and at the pen, before choosing the shape of the next word. I can't tell these two apart. What I can say is that the exceptions are not noise: they follow the position and the distance, each glyph in its own way.
Comments and constructive attacks (
) welcome, as always.