The Voynich Ninja

[I know: it's speculative, just an experiment, not a theory, I'm not buying it myself.]

Omitting spaces in a ciphered or coded message only makes sense if the original string of symbols or code words can be reconstructed.

The problem with using MZ's slot sequence directly on space-less Voynichese is that there are several ways to insert spaces: the same glyphs appear several times in the slot sequence. So I tried a different slot sequence as an experiment to see what happens: this is an attempt (un-optimized, just an example) to re-parse vords of space-less Voynichese unambiguously.

In this example, I removed all spaces from the first paragraph of You are not allowed to view links. Register or Login to view..

polcheyqokedySholopchedyolpchedyofShdyoly
dcheylShlolchedqokolcholotarchedyoky
qoteylcheesolkalolchedyokarShedy
dchedyqokainololchcThyykeedyal
qolcheolokeeyolololaiinolorain
sarolesesoteeyShorqokeeyol
dSheesokainchcThyoteyokain
pchedyqokeeyotyqoteyoteeyoly
qoteesyqotedyqokeedychcPhey
choldainotedycheeyqotainly
olSheololkedyShecKhedyoltedy
ychedytedyolSheedyqokeeyloly
dcholShedyqotedyqolchedychetyry
qokechedyqolSheedyorainaroloeeedy
sainolcheedyqokeedyotedy

Then vords were re-constructed using the following unambiguous 10-slot sequence defined as a regular expression:

Replace:
"((qo|)(k|t|p|f|)(ch|Sh|)(c(K|T|P|F)h|)(eee|ee|e|)(o|a|)(iii|ii|i|)(d|l|r|s|)(m|n|)(y|))"
with:
"\1\."

Spaces at the end of lines were removed:

Replace:
"\.$"
with:
""

Result:

pol.chey.qokedy.Shol.o.pchedy.ol.pchedy.o.fShdy.oly
d.chey.l.Shl.ol.ched.qokol.chol.o.tar.chedy.o.ky
qotey.l.chees.ol.kal.ol.chedy.o.kar.Shedy
d.chedy.qokain.ol.ol.chcThy.y.keedy.al
qol.cheol.o.keey.ol.ol.ol.aiin.ol.or.ain
s.ar.ol.es.es.o.teey.Shor.qokeey.ol
d.Shees.o.kain.chcThy.o.tey.o.kain
pchedy.qokeey.o.ty.qotey.o.teey.oly
qoteesy.qotedy.qokeedy.chcPhey
chol.d.ain.o.tedy.cheey.qotain.ly
ol.Sheol.ol.kedy.She.cKhedy.ol.tedy
y.chedy.tedy.ol.Sheedy.qokeey.l.oly
d.chol.Shedy.qotedy.qol.chedy.che.ty.ry
qoke.chedy.qol.Sheedy.or.ain.ar.ol.o.eeedy
s.ain.ol.cheedy.qokeedy.o.tedy

Not bad, but all prefixing o, d, l, r, s, and "Janus pairs" are separated, these are the main differences with a typical transliteration.

Spaces that should probably be omitted, in any consecutive vord pair of the previous paragraph, according to MZ's slot sequence, are marked with ",":

pol.chey.qokedy.Shol.o,pchedy.ol,pchedy.o,fShdy.oly
d,chey.l,Shl.ol,ched.qokol.chol.o,tar.chedy.o,ky
qotey.l,chees.ol,kal.ol,chedy.o,kar.Shedy
d,chedy.qokain.ol,ol.chcThy.y,keedy.al
qol,cheol.o,keey.ol,ol.ol,aiin.ol,or.ain
s,ar.ol,es.es,o,teey.Shor.qokeey.ol
d,Shees,o,kain.chcThy.o,tey.o,kain
pchedy.qokeey.o,ty.qotey.o,teey.oly
qoteesy.qotedy.qokeedy.chcPhey
chol.d,ain.o,tedy.cheey.qotain.ly
ol,Sheol.ol,kedy.She.cKhedy.ol,tedy
y,chedy.tedy.ol,Sheedy.qokeey.l,oly
d,chol.Shedy.qotedy.qol,chedy.che.ty.ry
qoke.chedy.qol,Sheedy.or,ain.ar.ol,o,eeedy
s,ain.ol,cheedy.qokeedy.o,tedy

The two slot sequences complement each other well, using the 10-slot sequence to get the shortest possible vords and MZ's 12-slot sequence to figure out which concatenations are probable.

If most of these "," are removed, and a few "." as well the end result is close to You are not allowed to view links. Register or Login to view..

The hypothesis is that Voynichese was constructed from components in which glyphs were strictly ordered in a slot sequence or similar (it could be a partial order relation or what would be best described by a tree or a finite state machine, because all combinations allowed by the 10-slot sequence don't occur) then spaces were omitted randomly. This randomness was non-uniform, there were preferences, these preferences are (imperfectly) modeled by MZ's slot sequence.

Of course 15th century scribes knew nothing about finite state machines and regular expressions, but they could have a practical implementation that was easy to use. For example the rule that glyphs must be strictly ordered can be enforced by a simple rule in a zigzag path on a square table using glyphs as coordinates: each linear segment of the zigzag path must cross the diagonal and, if we want more constraints, vertices could be forbidden on some areas of the table. I haven't abandoned my theory/delusion about zigzag paths and board games.

Somehow Ramon Llull`s ars combinatoria comes to mind here ( I don`t know why myself )

(05-04-2024, 01:53 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.In this example, I removed all spaces from the first paragraph of You are not allowed to view links. Register or Login to view..

I tried a set of tentative rules* I came up with a while back on the same passage -- here's what I got, for comparison.



pol chey qokedy Shol opchedy olpchedy ofShdy oly
dchey l Shl ol ched qokol chol otar chedy oky
qotey l cheesolkal ol chedy okar Shedy
dchedy qokain ol ol chcThy ykeedy al
qol cheol okeey ol ol olaiin ol or ain
sar olesesoteey Shor qokeey ol
dSheesokain chcThy otey okain
pchedy qokeey oty qotey oteey oly
qoteesy qotedy qokeedy chcPhey
chol dain otedy cheey qotain ly
ol Sheol olkedy ShecKhedy oltedy
y chedytedy ol Sheedy qokeey l oly
dchol Shedy qotedy qol chedy chety ry
qokechedy qol Sheedy or ain ar ol oeeedy
sain ol cheedy qokeedy otedy

* Insert space:

• Before q
  
• After g, m, n
  
• After r except before y, i
  
• After s before ch (with or without inserted gallows), Sh, d, l 
  
• After y except before t, k
  
• After l before r, o, d, ch (with or without inserted gallows), Sh
  
• Between repetitions of o, s, l
  

Updated thoughts on Massimiliano Zattera's "slot alphabet" of the Voynich manuscript; and how to expand it to a full "Voynich glyph alphabet".

You are not allowed to view links. Register or Login to view.

[attachment=8509]
A hypothetical wall chart for the Voynich scribes. The inscriptions are in Latin. Data from Zattera (2022) except {m} and {n} by author; graphic by author.

To improve the unambiguous slot sequence, I propose q = ch, allowing two ch in the pattern (the first one is equivalent to q):
qochy = chochy
qokchey = chokchey
qokedy = chokedy
qokeor = chokeor
qokeody = chokeody
qol = chol
qedy = chedy
olqo  = olcho
qko = chko

Exceptions:
qokechy but no chokechy
qokeechy but no chokeechy

Statistically, as a replacement for q, ch is the best choice:
[attachment=8546]

The complication of gallows dancing differently around q and ch can be resolved by removing gallows from the slot sequence.

(12-05-2024, 01:27 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.Statistically, as a replacement for q, ch is the best choice

In the A -> B evolution, q seems to have gradually replaced some instances of ch, especially vords starting with cho+gallows (there are still many in Q20 but only 2 in Q13).

qol makes sense as a replacement for chol: Q20 has both, Q13 appears to have a shortage of chol and an abundance of qol. But Herbal B has few chol (10) and, surprisingly, only one qol.

Quote:The complication of gallows dancing differently around q and ch can be resolved by removing gallows from the slot sequence.

kchol, ckhol, chkol, qkol, qokl, qolk* are attested.

Hidden in chedy territory, the only qedy:

[attachment=8563]

(14-05-2024, 03:59 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.Hidden in chedy territory, the only qedy:

...  also in qol territory.

(12-05-2024, 01:27 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.To improve the unambiguous slot sequence, I propose q = ch, allowing two ch in the pattern (the first one is equivalent to q):

I agree that this is a sensible improvement of the slot table. It was already considered in the earliest days of the old mailing list, by member Mike Roe. His system is shown on this page: You are not allowed to view links. Register or Login to view.

I imagine that another simplification could be achieved by not trying to fit compound words (or what look like compound words) in the slot, but to allow a recycle back into an earlier point of the slot.

An example of such a compound word is Hermes' choldaiin.