(15-03-2019, 08:59 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.I'm sorry to say, but this proposal shows all the classical signs of an incorrect solution:
- the translation table from Voynich text to plain text is vague and inconsistent
- the resulting plain text still has to be adapted in order to obtain real words
- the then resulting text is not grammatically sensible.
- the word patterns / low entropy are not explained.
One of the typical problems is that the 'translation' concentrates on the decryption part rather than on the encryption part. It is completely natural to map certain Voynich characters to different plain text characters in order to get some meaningful words out of it. However, this means that the person doing the encryption would have mapped several different plain text characters onto the same code character, which is not at all a natural thing to do.
I would appreciate it if someone could point out specifically and concretely, using actual examples of words and suffixes I have mentioned, particular examples of each of the general claims that Rene makes here. To say in general that my proposal is "vague", "inconsistent", "not grammatically sensible", without citing specific examples, is a vague statement itself. (I already addressed in detail the one specific and concrete question that Koen raised about "pan".)
Yes, the person doing the encryption would have mapped *two* different plain text characters onto the same code character. The only exception with more than that in my proposal is [d], which I have discussed and explained at some length and in some detail. Other than that, the general pattern is two plain text characters > one code character: "i" and "e" > [e], "k" and "g" > [t], "t" and "d" > [k], "p" and "b" > [p], "r" and "l" > [r], "h" and "j"/"i" > [ch], "u" and "v" > [d], followed by the additional step of [p] > [d] and [f] > [d] in many ms text contexts. [sh] is a special case because it has multiple forms, one of which I interpret as "t", and the other in fact very different character which I interpret as [m].
I do not see this system, which I just summarized concisely, as vague or inconsistent. It is just a two plain text > one cipher text system, that is all.
As for adapting the plain text to obtain real words, beyond the specific strict rules I listed above, the only adaptation so far has been slight changes in the Greek vowels, which again I believe is natural for 15th century late medieval Byzantine Greek, which would have been different from Ancient Greek, different from Koine Greek, different from Modern Greek, and in a language which is famous for its notable and distinctive vowel quality changes over time. Writing "skiais" as "skiois", "eipan" as "epan", and "oun" as "oan" are simply not that big a deal in the context of all the other historical changes in Greek vowel qualities occurring during this same period.
Regarding grammatically sensible text, I already in response to Koen raised the possibility of deliberate inversion of syllables in "geio-pan", which is a word play hypothesis that will need to be investigated systematically of course. After that we have "tis ipeirous otan skiais tis eipan oun". Is this not grammatical Greek? "When the continents are in the shadows, as they said." It is true that we have a medieval author employing archaic forms such as the Ancient Greek dative as a locative. It would be natural for such an author in such a situation to use a more modern Greek article "tis" in place of the Ancient Greek dative article "tais". I would be happy to have an in-depth discussion of the Greek grammar here, but if the only intention is to use every slight discrepancy of a 15th century text from the pure forms of Ancient Greek, Koine Greek, or Modern Greek as a reason to dismiss the entire interpretation as ungrammatical, that's not a good intention behind the discussion.
About the low entropy: I believe I mentioned in my original post that I think it would be interesting to take some simple Latin, Greek, French, and Italian plain texts, merge the voiced stops together with the voiceless stops, merge "r" and "l", merge "u" and "v" (as Classical Latin did!), merge "i" and "e" (as Modern Greek has done in pronunciation!), and then see what the entropy stats and bigram distribution plots look like.
About the word patterns: all those Latin and Greek suffixes appearing as the most common "words" will also need to be investigated systematically of course. This may bring us back to the word play hypothesis about deliberate inversion of syllables. At least this is a more precise and plausible procedure than the hopeless anagrams. As I mentioned in response to Koen, inversion of syllables has become the basis of an entire well-known slang argot of French, "Verlan".
Yes, the basic method here is two plain text letters > one cipher text character. Classical Latin did that with "u" and "v". The "betacism" sound change has done it with "b" and "v" in many languages. The "iotacism" sound change did it in the pronunciation of "i" and "e" in Greek. Thus I reject the claim that it is "not at all a natural thing to do". Yes, the Voynich author would have extended it to more pairs of letters in a form of cipher. I have proposed such a system here in a specific and concrete form which can be tested. Of course it can be adapted and adjusted and amended if the data proves it to be necessary.
During the investigation of Linear B, many researchers claimed that it was impossible for Greek to have been written in a syllabary script such as Linear A or B. It didn't fit the structure of the language! Nevertheless, it turns out that the Mycenaeans had indeed forced their language into the script that they had inherited.
Likewise, many Voynich researchers may claim that it is implausible for a cipher to merge each pair of plain text letters into a single cipher text character. Nevertheless, that is what the extremely restricted and limited Voynich character inventory and character sequence restrictions are pointing to. That is why the statistics keep pointing to Hawaiian as the most similar language: because Hawaiian has only 8 consonants! Three stops, two nasals, one liquid, v~w, and h. The most logical method to encrypt a European language to make its statistics look like Hawaiian, is to merge each pair of consonants into one consonant. My proposal here makes the resulting consonant inventory appear quite similar to that of Hawaiian.