The Voynich Ninja

Excellent!

(24-02-2017, 02:30 PM)Sam G Wrote: You are not allowed to view links. Register or Login to view.There's really no problem at all with a sound that can only occur at the beginning of words, and the absence in labels is easily explained if q is a prefix that serves a grammatical function used in writing sentences that is not needed in labeling individual items.

This is basically what I am saying. It is not a rendition of a plaintext letter, but something different.
As a 'sound' I proposed a stop (e.g. glottal stop). It is, however, very frequent for that.

(24-02-2017, 02:30 PM)Sam G Wrote: You are not allowed to view links. Register or Login to view.The labels obviously can't be verbosely encoded.  There's no way around this.

Here's an example of a herbal page with labels:

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

The labels, consisting of three Greek letters, are in fact numbers.
The labels we see in the Voynich MS could be verbosely encoded numbers.

(24-02-2017, 04:04 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.

(24-02-2017, 02:30 PM)Sam G Wrote: You are not allowed to view links. Register or Login to view.The labels obviously can't be verbosely encoded.  There's no way around this.

Here's an example of a herbal page with labels:

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

The labels, consisting of three Greek letters, are in fact numbers.
The labels we see in the Voynich MS could be verbosely encoded numbers.

And the rest of the text would then be... more numbers?

(24-02-2017, 02:32 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.In any case, for me y is another character that cannot be mapped to a letter.

If one glyph cannot be mapped to a letter then the same is true for all other glyphs of the VMS. Let me explain this conclusion with [y], [o] and [a] as example.

[y] can replace [o] and [a]. Moreover (deleted - see next post) 
Within the network of similar words [y] is used instead of [a] as last glyph in a word. For instance [dar] without [r] results in [dy] or [okal] without [l] results in [oky]:

aiiin( 41) daiin  ( 17) odaiiin( 4) okaiiin(  4) qokaiiin(  2) kaiiin( 3) taiiin ...
aiin (469) daiin  (863) odaiin (60) okaiin (212) qokaiin (262) kaiin (65)
ain  ( 89) dain   (211) odain  (18) okain  (144) qokain  (279) kain  (48)
an   (  7) dan    ( 20) odan   ( 2) okan   (  5) qokan   (  8) kan   ( 3)
aiir ( 23) daiir  ( 23) odaiir ( 2) okaiir (  6) qokaiir (  3) kaiir (--)          
air  ( 74) dair   (106) odair  ( 5) okair  ( 22) qokair  ( 17) kair  (14)
ar   (350) dar    (318) odar   (24) okar   (129) qokar   (152) kar   (52)
ail  (  5) dail   (  2) odail  (--) okail  (  1) qokail  (  1) kail  ( 1)
al   (260) dal    (253) odal   (13) okal   (138) qokal   (191) kal   (23)
am   ( 88) dam    ( 98) odam   ( 6) okam   ( 26) qokam   ( 25) kam   ( 9) 
os   ( 29) dos    (  1) odos   (--) okos   (  8) qokos   (  1) kos   ( 3)
or   (363) dor    ( 73) odor   ( 8) okor   ( 34) qokor   ( 36) kor   (26)        
ol   (537) dol    (117) odol   ( 2) okol   ( 82) qokol   (104) kol   (37)
y    (151) dy     (270) ody    (46) oky    (102) qoky    (147) ky    (25)

For word types having an [y] the glyph [y] is used in 3096 out of 4148 cases (74.6 %) as last glyph whereas [a] is used as last glyph in 73 out of 3760 word types (1.9 %). For this reason the word [chedy] occurs 501 times whereas the word [cheda] occurs only once.

For [o] this rule is less strict. [o] occurs in 314 out of 6920 word types (4.5 %) as last glyph for words including [o]. For instance beside [chy] (155 times) also [cho] exists [font=Trebuchet MS]68 times and [/font]the word [chedo] occurs two times within the VMS. (Deleted since both instances of [chedo] are most likely transcription errors.)

My conclusion is that [y] is used as spelling variant for [a] and [o]. This means if [y] cannot be mapped to a letter then the same is true for [a] and [o].

Torsten: it seems that you are inferring from the lack of -a or -o words but the abundance of -y words that -y is an orthographic variant where those same words should end with -a and -o.

I'd accept that word-initial y- could be a variant of word-initial o- , but I don't think you're going to convince anyone here that word-initial y- would replace word-initial a- in all but the very rarest of circumstances. So this is already somewhat asymmetric.

Of course, this proposed replacement scheme doesn't work for mid-word -a- or mid-word -o- instances at all, which is why it falls under the category of "orthographic", so y can be explained away as a scribal variant. :-)

But given that so many words end in -dy, are you really saying that -dy is a replacement for -do or -da? I'm not sure you've really thought that through to the end line, sorry. :-(

(24-02-2017, 10:16 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.Linguistically, this could represent a 'stop'. I don't know if there are any known examples where these are written. However, they would make sense in running text and might not be needed in front of single words. Also the next character would have to be a vowel. (Looking forward to Emma's comment on this).

Alternatively, they could be similar to the Greek 'spiritus'. (Vague memories of my one year of classical Greek).
These also only occur with vowels.

I don't have a comment on your q as stop suggestion, but the idea below it is very much my current thinking. Rough breathing in Greek is (well, was) a remnant of heta which represented the /h/ phoneme. That sound occurred in a very curious environment: 1) it was always before vowels (ignore rough breathing on rho, as it has a different origin) and 2) always at the beginning of words.

The non-occurrence of q at the beginning of labels could be due to the proceeding word. We know that not only does q have to appear before o, but two-thirds of the time it is after y (which is likely also to be a vowel). The sound may well depend on a very precise conditioning environment for it to be readily apparent to the writer and therefore worthy of representation.

(24-02-2017, 02:32 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.The character y is interesting in yet another way.
It is very frequent, and it seems to behave almost normally.

However, when one makes a list of the vocabulary of Voynichese words, one find that more than one third of all words ends with a y.

I just cannot see how any natural language would behave in this way.

In some languages all or almost all words end in a vowel, and the Voynich text may only have two clear vowels in o and y. I'm not convinced that vowels are fully differentiated, even though they are fully marked. The two vowel characters could simply represent a fundamental feature (such as front and back, high and low, rounded and unrounded) or even length. It could be that multisyllable words often end in short vowels due to prosodic considerations.

(24-02-2017, 04:04 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.Here's an example of a herbal page with labels:

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

The labels, consisting of three Greek letters, are in fact numbers.
The labels we see in the Voynich MS could be verbosely encoded numbers.

Greek letters were regularly used as numbers (as you know). But this is not verbose encoding, rather it is a separate use of the same characters, switching from phonemic to morphographic. I think that the use of Voynich characters as numbers can be mostly ruled out, for various reasons. The labels themselves are too well-structured to be numerals, anyway.

(24-02-2017, 06:06 PM)nickpelling Wrote: You are not allowed to view links. Register or Login to view.Torsten: it seems that you are inferring from the lack of -a or -o words but the abundance of -y words that -y is an orthographic variant where those same words should end with -a and -o.

I'd accept that word-initial y- could be a variant of word-initial o- , but I don't think you're going to convince anyone here that word-initial y- would replace word-initial a- in all but the very rarest of circumstances. So this is already somewhat asymmetric.

Indeed, my statement was not specific enough. What I mean is that it is possible to replace each instance of [y] with [o] or [a]. What I mean is that [y] is a variant of [o] and [a]. (clarified)

The usage of [y] is very strict. There are 4184 word types with [y]. There are  3096 word types ending in [y] (74.6 %)  and 625 word types starting with [y] (15 %). Some of the remaining words are still words like [dydydy] or [chedydy]. Therefore words like [kydain] or [dykain] are not very common.

Word initial [y] is interchangeable with [o] but not with [a]. If there is a word starting with [y] there is nearly always also a similar word starting with [o]:
oaiin (26) daiin  (863) odaiin (60) okaiin (212) otaiin (154)
oain  (11) dain   (211) odain  (18) okain  (144) otain  ( 96)
oaiir ( 4) daiir  ( 23) odaiir ( 2) okaiir (  6) otaiir (  4)                                
oair  ( 3) dair   (106) odair  ( 5) okair  ( 22) otair  ( 21)
oar   (16) dar    (318) odar   (24) okar   (129) otar   (141)
oal   ( 3) dal    (253) odal   (13) okal   (138) otal   (143)
oam   ( 1) dam    ( 98) odam   ( 6) okam   ( 26) otam   ( 47)                 
oor   ( 3) dor    ( 73) odor   ( 8) okor   ( 34) otor   ( 46)            
ool   (--) dol    (117) odol   ( 2) okol   ( 82) otol   ( 86)
oy    ( 6) dy     (270) ody    (46) oky    (102) oty    (115)
yaiin ( 6)              ydaiin (21) ykaiin ( 45) ytaiin ( 43)
yain  (--)              ydain  ( 5) ykain  ( 10) ytain  ( 13)
yaiir ( 2)              ydaiir (--) ykaiir (  2) ytaiir (  1)
yair  ( 2)              ydair  ( 2) ykair  (  8) ytair  (  3)
yar   ( 2)              ydar   ( 2) ykar   ( 36) ytar   ( 26)
yal   ( 1)              ydal   ( 3) ykal   ( 16) ytal   ( 19)
yam   (--)              ydam   ( 1) ykam   (  5) ytam   ( 13)                     
yor   ( 2)              ydor   ( 1) ykor   ( 10) ytor   ( 14)
yol   ( 2)              ydol   (--) ykol   ( 14) ytol   ( 12)
yy    ( 1)              ydy    ( 8) yky    ( 18) yty    ( 24)

Quote:Of course, this proposed replacement scheme doesn't work for mid-word -a- or mid-word -o- instances at all, which is why it falls under the category of "orthographic", so y can be explained away as a scribal variant. :-) 

But given that so many words end in -dy, are you really saying that -dy is a replacement for -do or -da? I'm not sure you've really thought that through to the end line, sorry. :-(

My interpretation of the network of similar words is that word final [y] is a replacement for word final [a] and [o]. Word final [a] is nearly always replaced with [y]. Therefore a word final [a] is rare. Word final [o] did occur in only 314 out of 6920 word types (4.5 %).
ar (350) dar (318) odar (24) okar (129) qokar (152) kar (52) sar ( 84) rar (21) orar (10) olkar (19) lkar (30)
al (260) dal (253) odal (13) okal (138) qokal (191) kal (23) sal ( 55) ral (17) oral (10) olkal (11) lkal ( 5)
or (363) dor ( 73) odor ( 8) okor ( 34) qokor ( 36) kor (26) sor ( 57) ror (17) oror ( 5) olkor ( 4) lkor ( 4) lkr ( -)
ol (537) dol (117) odol ( 2) okol ( 82) qokol (104) kol (37) sol ( 75) rol (20) orol (15) olkol ( 5) lkol ( 5) lkl ( 9)
y  (151) dy  (270) ody  (46) oky  (102) qoky  (147) ky  (25) sy  ( 35) ry  (13) ory  (17) olky  (22) lky  (17) lk  ( 1)
[font=Courier New]o  ( 81) do  ( 16) odo  (--) oko  (  8) qoko  (  9) ko  ( 2) so  (  5) ro  (10) oro  ( 5) olko  ( 1) lko  ( 3)[/font]
[font=Courier New][font=Courier New]a  (  3) da  (  9) oda  (--) oka  (  1) qoka  (---) ka  ( 1) sa  (  4) ra  (--) ora  (--) olka  (--) lka  (--)[/font][/font]

Torsten: sorry to have to point out the obvious, but you can't argue it in both directions at the same time - either you think that y is a variant of a and o, or you think that o and a are variants of y. You say both...

"it is possible to replace each instance of [y] with [o] or [a]."

...and...

"word final [y] is a replacement for word final [a] and [o]"

So, which direction are you actually arguing for?

(24-02-2017, 11:00 PM)nickpelling Wrote: You are not allowed to view links. Register or Login to view.Torsten: sorry to have to point out the obvious, but you can't argue it in both directions at the same time - either you think that y is a variant of a and o, or you think that o and a are variants of y. You say both...

"it is possible to replace each instance of [y] with [o] or [a]."

...and...

"word final [y] is a replacement for word final [a] and [o]"

So, which direction are you actually arguing for?

I mean that the scribe of the manuscript has used [y] as a variant of [a] and [o]. If this is the case it would be possible for us to remove all instances of [y] by replacing them with [o] at the beginning of a word and with [a] or [o] at the end of a word.

I started this thread in order to clarify why five out of (approx.) 25 Voynich characters ( q  f  p  m  y ) almost certainly cannot represent substitutions of written characters in some plain text.
The short summary is, that f and p appear almost exclusively at top lines of paragraphs, m at ends of lines (not sentences), q at starts of words, where it is followed by o, while y is at the end of every third word type.

This is a way of saying that the Voynich MS text cannot be the result of applying a simple substitution to a plain text in most languages of interest.
This was already known from entropy considerations, but there are plenty of people who either don't understand the maths, or don't trust it. This alternative explanation is something more simple and more visual. It is also complementary, and anyone claiming that the text actually is a straightforward rendition of a meaningful plain text has to explain both.

There is of course much more to be said about it.

For one, there are several underlying assumptions.
One is, that the text has to be read from left to right. This actually makes essentially no difference. (And as a matter of fact, entropy is independent of the reading direction).
Another is, that the word spaces are real. This is a more complicated question, but if one were to assume that the word spaces are not real, one runs into many other problems.

Another highly interesting question is, whether the conclusion formulated for 'written characaters' (letters) is also valid for 'spoken language' (phonemes). This is certainly much less clear. At the same time, one has to realise that several of the observations (in particular for f  p  ) are anomalies that are specifically related to the writing.

If five of the characters cannot represent plain text letters, then that puts in doubt the entire Voynichese alphabet. There are in fact other anomalies that are less obvious.
Without enumerating them all, the main one (in my opinion) is the pairing of characters. There are several pairs that can be substituted for each other almost arbitrarily:

ch / Sh
k / t
f / p
l / r
o / qo [font=Arial] (at word start)[/font]

However, the frequencies of the resulting words are not evenly distributed, so it does not look like something arbitrary.

The last pair in the above list is particularly interesting, since both o and qo appear in the running text, but only one of them appears in the labels.

That's a dead giveaway of something.
We just haven't figured out yet of what.
It does tell us that the creation of the text is not something arbitrary. There is reasoning and a plan behind it.
The same conclusion follows from the fact that label word distribution is almost flat, while the running text has a Zipf-like distribution.