The Voynich Ninja

Hello all,

I have just started working with the intriguing Voynich Manuscript, attempting to shed some light on its text analysis. I conducted some basic text analysis using the V101 v. 2a transliteration file.

I programmed a bidirectional pair detection algorithm. By "bidirectional pair," I mean a pair of words that appear in both directions, such as "AAA BBB" and "BBB AAA." I found 1,284 bidirectional pairs (or 642 unique pairs, each counted twice due to the direction).

I understand that this phenomenon is not commonly found to such an extent in known languages. As far as I know, no known natural language routinely produces as many bidirectional pairs as I have identified. This could suggest either a highly artificial or constructed language or perhaps some unique feature or quirk of the manuscript's text.

I apologize if this finding has been reported before; I was unable to find any previous entries on this topic. 

Looking forward to reading your thoughts.

Here are the top 20 bidirectional pairs I have found:

BIDIRECTIONAL PAIRS
1284 bidirectional pairs

TOP BIDIRECTIONAL PAIRS:
Pair 1: ('oy', 'am'), Count: 50
Pair 2: ('am', 'oy'), Count: 6
Pair 1: ('s', 'am'), Count: 47
Pair 2: ('am', 's'), Count: 3
Pair 1: ('1oe', '1oe'), Count: 21
Pair 2: ('1oe', '1oe'), Count: 21
Pair 1: ('am', 'oe'), Count: 13
Pair 2: ('oe', 'am'), Count: 20
Pair 1: ('ay', 'ae'), Count: 25
Pair 2: ('ae', 'ay'), Count: 7
Pair 1: ('1oe', '8am'), Count: 24
Pair 2: ('8am', '1oe'), Count: 5
Pair 1: ('ay', 'am'), Count: 26
Pair 2: ('am', 'ay'), Count: 3
Pair 1: ('8am', '8am'), Count: 11
Pair 2: ('8am', '8am'), Count: 11
Pair 1: ('oe', 'oe'), Count: 11
Pair 2: ('oe', 'oe'), Count: 11
Pair 1: ('oy', 'oe'), Count: 11
Pair 2: ('oe', 'oy'), Count: 11
Pair 1: ('oe', '1c89'), Count: 19
Pair 2: ('1c89', 'oe'), Count: 3
Pair 1: ('4oe', '1c89'), Count: 10
Pair 2: ('1c89', '4oe'), Count: 10
Pair 1: ('s', 'ay'), Count: 17
Pair 2: ('ay', 's'), Count: 2
Pair 1: ('8am', 'oe'), Count: 6
Pair 2: ('oe', '8am'), Count: 13
Pair 1: ('4ohan', '1c89'), Count: 8
Pair 2: ('1c89', '4ohan'), Count: 11
Pair 1: ('oe', '1c9'), Count: 13
Pair 2: ('1c9', 'oe'), Count: 5
Pair 1: ('ay', 'ay'), Count: 9
Pair 2: ('ay', 'ay'), Count: 9
Pair 1: ('ay', 'oy'), Count: 8
Pair 2: ('oy', 'ay'), Count: 9
Pair 1: ('4ohae', '1c89'), Count: 10
Pair 2: ('1c89', '4ohae'), Count: 7
Pair 1: ('4ohc89', '4ohc89'), Count: 8
Pair 2: ('4ohc89', '4ohc89'), Count: 8

Hi quimqu,

Welcome to the forum.

The number of reversible unique pairs is R(w, 1) in You are not allowed to view links. Register or Login to view..

Ignoring unclear spaces (',' removed) and unclear words (that contain '?'):
Word tokens: 38162
Unique pairs: 34338
Reversible unique pairs: 2799

Ignoring unclear words only (',' kept as word separator)::
Word tokens: 40621
Unique pairs: 35234
Reversible unique pairs: 3327

Apparently you count a repeated word as two pairs (one unique pair in my count):

Quote:Pair 1: ('1oe', '1oe'), Count: 21
Pair 2: ('1oe', '1oe'), Count: 21

so you should have more pairs than me. We must be doing something differently.


EDIT1:
For comparison, the Latin Speculum Humanae Salvationis (Lutz & Perdrizet 1909):
Word tokens: 37021
Unique pairs: 29544
Reversible unique pairs: 1717


EDIT2: another try with V101, removing all pairs across lines but keeping all words including those that contain '?':

Without unclear spaces (',' removed)
Unique pairs: 30152
Reversible unique pairs: 2449

With unclear spaces (',' kept as word separator):
Unique pairs: 31110
Reversible unique pairs: 2955

Wow, such a great analysis from Marke Fincher. I didn't know about it. But the conclusions (as far as we can suppose) are the same: VMS does not look like to be written in a "normal" language, so, wether it is invented, or it is a normal language manipulated to a point that the natural relationships between words disappear.

In fact, I did the analysis because I had the impression, trying to read the MS, that the word were repeated constantly.

Thanks for the reply. I think that having some less or more pairs is not representative. What really matters is that those kind of high relations are not "natural".

Regards,

(16-05-2024, 05:47 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.Hi quimqu,

Welcome to the forum.

The number of reversible unique pairs is R(w, 1) in You are not allowed to view links. Register or Login to view..

Ignoring unclear spaces (',' removed) and unclear words (that contain '?'):
Word tokens: 38162
Unique pairs: 34338
Reversible unique pairs: 2799

Ignoring unclear words only:
Word tokens: 40621
Unique pairs: 35234
Reversible unique pairs: 3327

Apparently you count a repeated word as two pairs (one unique pair in my count):

Quote:Pair 1: ('1oe', '1oe'), Count: 21
Pair 2: ('1oe', '1oe'), Count: 21

so you should have more pairs than me. We must be doing something differently.

The ratio reversible pairs / total pairs at distance 1 is not much different in some natural language texts of comparable length:
10.9% VM Takeshi Takahashi: 37760 tokens, 3546/32409
9.9% Geber, Summa perfectionis (Newman 1991) 33188 tokens, 2434/24599
9.4% VM V101: 40621 tokens, 3327/35234
9.3% Dante Alighieri, La divina commedia: 97206 tokens, 5970/63628
9.0% Christine de Pisan, Le Livre du chemin de long estude: 34810 tokens, 2291/25189
5.8% Speculum Humanae Salvationis (Lutz & Perdrizet 1909): 37021 tokens, 1717/29544