The Voynich Ninja

(04-05-2022, 01:42 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.it is because the "translators" themselves disregard grammar

It's the joy and excitement of discovery that makes one blind.

I am not a palaeographer, but I like Latin and I sometimes enjoy working my way through medieval manuscripts written in Latin or Italian. In my experience, grammar is vital to understanding any written text, even more so when punctuation is absent or very limited. I find it difficult to imagine any simple measure that could tell us "to what extent" a text respects a grammar. The closest I can think of would be some kind of comparison of the frequencies of the most frequent words. As discussed You are not allowed to view links. Register or Login to view., languages have a closed set of function words typically including at least half of the 10 most frequent words in any reasonably long text. A few years ago, You are not allowed to view links. Register or Login to view. that checking how the most frequent Voynichese words are interpreted by a proposed solution is an effective way to evaluate a theory: in my opinion, this test is the simplest possible way to see if a solution has any hope of making sense grammatically.

A more sophisticated approach could include looking for specific patterns: e.g. since in English plurals typically end with -s, one can expect "-s are" ("when the thousand years are expired", "the seven heads are seven mountains", "the former things are passed away") to be more frequent than "-s is" (even if 'is' is normally more frequent than 'are'). But of course defining such patterns is much more difficult than counting word frequencies. Anyway, all languages have patterns and Voynichese also shows patterns in how words are arranged: if Voynichese represents linguistic content, at least some of its patterns are probably related to the grammar of the underlying language.

(07-05-2022, 11:39 AM)MarcoP Wrote: You are not allowed to view links. Register or Login to view.I am not a palaeographer, but I like Latin and I sometimes enjoy working my way through medieval manuscripts written in Latin or Italian. In my experience, grammar is vital to understanding any written text, even more so when punctuation is absent or very limited. I find it difficult to imagine any simple measure that could tell us "to what extent" a text respects a grammar. The closest I can think of would be some kind of comparison of the frequencies of the most frequent words. As discussed You are not allowed to view links. Register or Login to view., languages have a closed set of function words typically including at least half of the 10 most frequent words in any reasonably long text. A few years ago, You are not allowed to view links. Register or Login to view. that checking how the most frequent Voynichese words are interpreted by a proposed solution is an effective way to evaluate a theory: in my opinion, this test is the simplest possible way to see if a solution has any hope of making sense grammatically.

A more sophisticated approach could include looking for specific patterns: e.g. since in English plurals typically end with -s, one can expect "-s are" ("when the thousand years are expired", "the seven heads are seven mountains", "the former things are passed away") to be more frequent than "-s is" (even if 'is' is normally more frequent than 'are'). But of course defining such patterns is much more difficult than counting word frequencies. Anyway, all languages have patterns and Voynichese also shows patterns in how words are arranged: if Voynichese represents linguistic content, at least some of its patterns are probably related to the grammar of the underlying language.

Hi, Marko,

Thank you for pointing out that the grammar will eventually lead to the solution of VM language. Because of the unique subject matter in the VM, it would be impossible to match the text word for word with any other text, such as with the prayer Our Father, which was often used for comparison. However, the grammatical pattern generally stay, even if the words change. This is the reason why I am concentrating on VM grammar. 

The idea is to find word families, to see how various roots were used in different derivates, what prefixes and suffixes are use, as well as how the words are used in phrases.
In Slovenian, DAM was not used just for many words related to DATI (to give), but also in a peculiar way with the adjectives, such as SE DOLŽAN DAM - I obliged give - I feel obligated. It was also used as the verbal suffix, like REČI DAM - I give words, I preach; or REČEM - I say. This is similar to English TAKE A LOOK  or LOOK. 
In the VM, the words DAM and DAR (in different grammatical forms) are among the most frequently used, which could be understandable.
Here are some examples how the grammar could be explained for Slovenian language:
Slovenian verbs are conjugated for number, gender, time[attachment=6492]

The grammar can be recognized from the surrounding words [attachment=6493] 
And here are some examples of the prefixes

There is a Slovenian grammar book from the 16th century, if you are interested. 

There are also two Italian Slovenian dictionaries, however, because Slovenian is highly flexible, not all the words can be found in the dictionary.

Sincerely,

Cvetka

(07-05-2022, 11:39 AM)MarcoP Wrote: You are not allowed to view links. Register or Login to view.The closest I can think of would be some kind of comparison of the frequencies of the most frequent words. As discussed You are not allowed to view links. Register or Login to view., languages have a closed set of function words typically including at least half of the 10 most frequent words in any reasonably long text. A few years ago, You are not allowed to view links. Register or Login to view. that checking how the most frequent Voynichese words are interpreted by a proposed solution is an effective way to evaluate a theory: in my opinion, this test is the simplest possible way to see if a solution has any hope of making sense grammatically. 

Hi, Marco:

Thank you for bringing up the next level of issues with Voynichese (e.g. a level up from conditional entropy) -- the word pattern statistics.  I definitely agree that word pattern is more closely related to the "grammar" topic of this string than any other measured statistic.

I know you are very well aware of this but some aspects of word patterning does look like a natural language (e.g. the text "obeys" Zipf's law, arguably only "approximately" (You are not allowed to view links. Register or Login to view.) or perhaps more tellingly, the Voynichese word lengths' adhere to a gamma distribution (You are not allowed to view links. Register or Login to view.), and "Key words and Co-occurrence" looks like natural language (You are not allowed to view links. Register or Login to view.)).  But all is not rosy in word pattern land: there are too few word pattern repeats compared to what is found in most natural languages (You are not allowed to view links. Register or Login to view. and You are not allowed to view links. Register or Login to view.) -- which is precisely what you would be looking for in order to infer some sort of grammar.  Because what is grammar but (on a high level) other than repeating patterns of words in order to communicate?

This situation (coupled with all the myriad of ways functional words can present in a language) has made it impossible to identify so-called "functional" words in the Voynich -- which has been discussed You are not allowed to view links. Register or Login to view. and in your parts of speech experiments (You are not allowed to view links. Register or Login to view.).  But these functional words (particularly "short indication words") are those that are found at the highest frequency in many natural languages.  Of course, this is complicated by the many ways that natural language can represent functional words (e.g. as separate words, as prefixes, as suffixes, etc.). But intense attempts to link even one Voynichese word (or even perhaps, some word portion) with some sort of function have been unsuccessful to date.

All this verbiage is to say that I am extremely concerned that the highest frequency words in the underlying plaintext are not consistently encoded in Voynichese (or said another way -- the highest frequency words in Voynichese (at least -- maybe, all?) represent different plaintext words.  Or the words maybe are consistently encoded but have been re-ordered in some manner that results in the strange reduplication coupled with lack of longer word repeats.  And we would need some sort of consistent pattern for how either or both (gasp! please not) encoding methods work in order to make any headway.

Maybe this is just catastrophizing - but this is one of my biggest concerns with this work and I do not know how to address whether or not this is the case with statistics.  Maybe there is some thought on experiments that could be done to try to tease this out?  Does the MATTR experiments (You are not allowed to view links. Register or Login to view., You are not allowed to view links. Register or Login to view., and You are not allowed to view links. Register or Login to view., which I just realized I haven't yet discussed and do belong in this general category of word patterns) help distinguish these possibilities in any way?  I will definitely be thinking about it . . .

(07-05-2022, 11:39 AM)MarcoP Wrote: You are not allowed to view links. Register or Login to view....grammar is vital to understanding any written text, even more so when punctuation is absent or very limited

Thank you, Marco, I hope that our author has respected the grammar in his manuscript and that we will succeed in discovering it.

(07-05-2022, 05:59 PM)MichelleL11 Wrote: You are not allowed to view links. Register or Login to view....the myriad of ways functional words can present in a language

As you say, Michelle, identifying language could help identify function words and vice versa.

(07-05-2022, 05:59 PM)MichelleL11 Wrote: You are not allowed to view links. Register or Login to view.all is not rosy in word pattern land

I totally agree and I believe that catastrophizing is perfectly justified. Some of the word patterns in Voynichese are quite different from what one sees in written natural languages, e.g.:

• high frequency of reduplication (e.g. daiin.daiin);
  
• even higher frequency of partial reduplication (e.g. f103r.4 ychedy.qokedy.okedy.qokeey.okey);
  
• very frequent character sequences that occur across words (e.g. edy_qo, see You are not allowed to view links. Register or Login to view.);
  
• many words "prefer" to appear in different positions inside lines (also You are not allowed to view links. Register or Login to view.).
  

But the fact that making sense of these patterns is horribly difficult does not imply that they can simply be ignored. If the question is "should a translator care about grammar?", I believe that the answer must be "yes". Grammar is not an optional part of language: if we conclude that there is no grammar underlying Voynichese (however buried under a very clever or maybe very clumsy writing system), the implication is that there is nothing to translate.

(07-05-2022, 11:39 AM)MarcoP Wrote: You are not allowed to view links. Register or Login to view....languages have a closed set of function words typically including at least half of the 10 most frequent words in any reasonably long text

Marco, unfortunately I do not have a list of words in the manuscript by frequency.

I am learning the rules of vowel contraction in ancient Greek, to understand why the present tense infinitive endings are different: -ein, -oun or an. I learned that contraction is not used by all authors and in all dialects.

How can we be sure that our scribes applied the same rules of grammar?

It's a bit of an odd question to answer because we can't even be sure they are writing in the same language (I think they probably are), so we can't say they applied the same rules until we have a breakthrough.

The differences between the scribes (and even between the quires) are why it's best to study individual quires separately.