Computational Attacks on Abbreviated Text

[nickpelling]

Usage of Tironian notae noticeably declined during the 14th century, and had almost entirely withered on the vine by 1450. What took their place were local scribal abbreviations, which form a topic that has not yet received anywhere near enough attention from Voynich researchers.

[-JKP-]

[*I'm not sure why the scribe has written this as Pre perpositio as I'm fairly sure Prepositio/Praepositio was intended.]

I'm also sure that the letter "p" with a tilde (~) means "Pre"
So, I read too "Prepositio".

...

I based my guess of Prepositio/Praepositio on the context of the rest of the sentence and also the relative consistency of the abbreviations in this text, so I was hoping it was correct. I didn't look to see if there were already translations/transliterations of this passage because I get more practice if I try to do it myself.

The Visigothic/Carolingian family of scripts is so much easier on the eyes (and the brain) than 15th-century Gothic cursive. After the great famines and plagues of the 14th century, cursive handwriting went downhill for about a century and use of abbreviations was so prevalent, some late medieval texts are almost like shorthand. The book hands were a little better. They don't seem to have degraded as much.


One of the things that is interesting about the VMS main text is that the spacing is much more like older texts than Gothic cursive. The marginalia on the last page is a variant of Gothic cursive, but the VMS main text, even though it has tails and apostrophes that suggest familiarity with Gothic cursive, is spaced much more like ciphered texts or like the early medieval texts.

[-JKP-]

Usage of Tironian notae noticeably declined during the 14th century, and had almost entirely withered on the vine by 1450. What took their place were local scribal abbreviations, which form a topic that has not yet received anywhere near enough attention from Voynich researchers.

Over the last decade, I've collected more than a thousand samples of scribal text, selected from many more thousands for the ones that most closely match the main text and the marginal text. I have also collected thousands of samples of abbreviated text.

To analyze these samples, I developed a mathematical system to compare them to each other, to search and sort them, and to objectively compare them to the VMS text. I've also made them search-and-sortable according to date, location, style of text, and a number of other parameters.


The only reason I haven't written it up yet is because it's a massive undertaking to try to organize and describe it to a general audience (I'm not saying Voynich researchers are a general audience, many are clearly on the upside of the IQ curve, but not everyone has a paleographic and calligraphic background, so the backstory needs to be filled in for any of it to make any sense). I have pages and pages of description in note form and some finished pages but I suspect it will take months to finish it.


As much as I enjoy discussing the images, the text has always been my primary interest.

[-JKP-]

Here is another way text is "abbreviated" in known ciphertexts, by representing duplicate letters with a symbol that does not resemble the symbol used for the same letter when it is written alone. Even a simple system like this would skew letter-frequency results from computational attacks. If the underlying language were known, it might be noticed that the words are slightly shorter than average, but if the underlying language were not known, it would be harder to discern:

   

You may have noticed the relative lack of letters repeated next to one another in the VMS. Even the "cc" is not necessarily a repeated letter. In early medieval texts, the "cc" shape could mean "a" or "t" depending on the time and region and the VMS "cc" is frequently written as though it were a ligature, crowded together, but might, in fact, represent one sense unit rather than two.

Repeated letters are common in many languages, especially western languages and noticeably sparse in the VMS (especially if the spaces are taken to be literal). They are less common in Asian languages that are based on syllables rather than letters (ba, be, bo, bu).


Note that the example cipher system above would be easier to crack than some because only doubled consonants are written as singles. Thus, vowel-consonant patterns might still be discerned. It would be harder to decipher if the system had a mixture of double vowels and double consonants written as one symbol.

[Torsten]

Wouldn't that also imply that the scribe was 'thinking sounds' while writing? That he somehow assigned sounds to the glyphs?

The usage of a special character alone only implies that the scribe was knowing this character.
To demonstrate that a character was used because of its meaning it would be necessary to demonstrate a pattern for its usage. For instance if you can demonstrate that words with an apostrophe are used after or before words with a rare or unique feature this would demonstrate that the scribe was using the apostrophe since it belongs to this place. 
Unfortunately it is only possible to demonstrate that characters where used beside other characters or are preferred for a special place within a word, paragraph or line. For instance the [y] is often used beside [d] and at the beginning or end of words. But patterns for relations between words are hard to find for the VMS beside the fact that words where used near to similar looking words. One possible reason for this observation is that the scribe was thinking on something lese then sounds during writing.