https://youtube.com/watch?v=4cRlqE3D3RQ
https://youtube.com/watch?v=8nHbImkFKE4
tl;dr: it's probably real writing, likely related to Roma/Syriac
The videos are an impressive phonological reconstruction, but I predict (based on the assumption that the math isn't lying), that it would be effectively impossible to get much beyond ad-hoc phonetic correspondences with Romany, to any predictive morphology or syntax.
The solution in this article is rather plausible. If the writing is in a highly restricted vocabulary, with highly restricted syntax, and highly constrained domain, it would be possible to get the observed information density.
Comparing it to, say, Linear B or Egyptian hieroglyphs is instructive. Both of those clearly have the information density of regular human language.
In the end, the solution might be a combination. It might use some Roma/Syriac nouns, but it seems clear it doesn't use them in anything like a normal linguistic context.
Caveat: IANALinguist
Natural language can be very surprising. :)
Holy Montague! You've found it! ;)
That is an extraordinarily European viewpoint, where alphabets have a few tens of glyphs, each representing a single letter (or, in the case of capital letters, two glyphs per letter).
Imagine an alphabet similar to Arabic where each letter may have up to five glyphs, or even an alphabet in which the _position_ of the glyph changes it's letter. Or Korean, where each 'letter' is composed of two or three interchangable components. Or Han, in which the number of instrument strokes in the glyph affect it's meaning.
There are so many variations on what constitutes a letter, never mind an individual glyph or a full word/concept, that one cannot use strictly European analysis techniques on arbitrary writing systems.
But we don't have to imagine a random alphabet which Voynich might have been written in. We can look at the actual language in the manuscript. It often shows three or four Latin-like glyphs repeated. Unless you have access to some special way these symbols are differentiated or otherwise convey information, a way that all other would-be interpreters have missed, you really don't have an argument.
And as far as European thinking goes - there's no evidence I know of a non-European origin to the manuscript.
Edit: and the languages you cite have a larger, not smaller information content than European languages, see;
https://linguistics.stackexchange.com/questions/6167/most-su...
However, many medieval documents even in known real languages do contain almost undifferentiated glyphs one after the other:
https://en.wikipedia.org/wiki/Minim_(palaeography)
Not quite the same, but something similar might conceivably apply to e.g. the Voynich's c-like characters.
IE, there's no text to translate, it's not an abbreviation for any standard text.
> changes it's letter
changes its letter
> changes it's meaning
changes its meaning
The punch line of one of his stories came while he was discussing the history of attempts to interpret it. One character had a dense theory that involved, as a step in the middle, involved the text being an anagram. The problem being that anagramming destroys information and, given a large enough text, is impossible to reverse. (Interestingly, this is why some 17th-18th century scientists who were obsessed with both secrecy and precedence published their results in anagram form---if someone later came along and published the same result, they could say, "See, I was first; here's what that gibberish means.")
My point being that if you're going to posit an "out-there" theory, you have to make sure your theory can get you back here.
For a short text, you may have a point, but for a text as large as the Voynich, information content is a pretty reasonable way to examine what's going on. The number of characters (as far as I know (http://www.voynich.nu/transcr.html)) is within the range of alphabetic writing. (Alphabets have <100 characters, syllabaries typically have a few hundred (Note: English has something like 5000 syllables.), and "morpho-syllabic" languages like Chinese (Thank you, John DeFrancis!) have a few thousand.)
To have more bits per character, the document would have to have more different characters. To pack more bits in via position, for example, would be possible, but you can't go crazy with the idea because you still need to be able to decrypt it. (Hangul is a neat example; it's alphabetic, but written with the letters arranged into blocks representing syllables. Neat!)
It is not at all, you misunderstand the point.
It is the technique that codebreakers used to check whether something is linguistic or not. And it is rather consistent across all languages. The 'Alphabet size' of a string might vary (i.e. logographic languages have much higher symbolic vocabulary sizes, so different per-symbol information content), but that doesn't affect the overall information density, because those symbols also take more information to encode. I haven't done the analysis myself, but as I understand it, Voynich is way way outside the range of normal human languages, European or otherwise.
And raising the issue of logographic or syllabic systems, is a point I made (did you not understand they are your examples). Even if one argued (as some have done) that the symbols in Voynich are logographic or syllabic, that would make it even lower information density, even further from human language.
I'm sorry I wasn't clear, but this stuff is very well-established in the Voynich literature. It puts genuine constraints over what the language can contain. It doesn't mean it's gibberish, but it is very unlikely to be 'real writing' derived from any known language, extant or extinct. Crucially, if it is real writing, it is of a form so dramatically different to anything we've ever seen before (i.e. fundamentally different syntactic and semantic structures), that one suspects the kind of correspondence translation techniques in the videos will be futile.
Thank you for the links!
Edit: Btw, looking at some suggested videos I've found this https://youtu.be/PoNm65v1thU a bit intriguing.
Roma and Syriac are two radically different languages (Indo-European vs Semitic), so it's peculiar that the conclusion would be that it's one of the two!
Disclaimer, I haven't watched the video.
Watch the third video[1], it's much shorter but provides quite compelling evidence to support his theory.
I mean if tolkien had written the Lord Of The Rings in elvish the XKCD answer would still be essentially correct but it wouldn't tell us what it said.