09-03-2020, 11:14 PM
(This post was last modified: 10-03-2020, 09:22 AM by -JKP-.
Edit Reason: fixed typo: duplicate word
)
RenegadeHealer wrote:In summary, I think the arguments for Voynichese's incompatibility with natural human language are not as widely and well understood as they deserve to be, and I don't think that's entirely due to willful ignorance. How can this viewpoint be better worded and promulgated, so that anyone out there with the chops to dispute it (or confirm it!) understands what they're replying to?
I think this post deserves a separate answer because it's a good question.
.
Part of the problem is that most of the existing transcripts turn Voynichese into something THAT LOOKS like natural language.
Take the Takahashi transcription, for example, the most commonly used transcript for "solutions" and for computational attacks.
The Takahashi transcription uses the EVA alphabet, which was designed to be easy to remember and type. How was it made easy to remember and type? BUT CONVERTING A CONSONANT-LOOKING SHAPE INTO A VOWEL-LOOKING SHAPE. The tokens look like more like words, which are easier to scan and stick more easily in our heads. This encourages the brain to think of the glyphs as consonants and vowels.
So then what happens? Voynichese takes on the look of a language. We see "words" like oteedy. But there is no oteedy in the VMS. It's o‡cc89, which is going to look less like a word to most people.
How this influences statistical studies...
So take a vowel-consonant-balance study that uses the Takahashi transcript. Attacks are based on artificial "words" like "oteedy", and for reasons I don't understand many of the researchers assume "o" and "e" actually are vowels. Even if the VMS were a simple substitution cipher (which I doubt) YOU CAN'T MAKE THIS ASSUMPTION. A one-letter shift (from abcde to bcdef) immediately changes the "vowel-consonant" appearance of even the simplest ciphers.
As long as transcripts are designed in a way that makes Voynichese LOOK like language, new researchers will probably start out with assumptions about its structure that have more to do with the transcription alphabet than they do with actual VMS text.