You will need input from various specializations, naturally, but I agree with Anton that methodology is more important.
To me it's more about how quickly people are apparently willing to call something a solution. Propose a conversion scheme that makes sense for a couple of words, and you've made a Voynich solution. While we know that it doesn't work like that, because dozens of people have done it in a different way, all with the same feeling of success.
The real solution will be in discovering how to deal with the various statistical oddities of the text and how those can be linked to language somehow. But people who understand this and try to tackle these problems aren't very likely to call "solution" soon because they better understand the challenges we face.
I think there are several reasons why the same mistakes are so often made...
On the surface, the VMS looks deceptively simple (which is probably why so many "solutions" are substitution systems). It has shapes that look like Latin vowels and shapes that look like Latin consonants and abbreviations, and a few that look almost like familiar glyphs (gallows) but not quite (they are almost like Latin ligatures or Greek numerals... but not quite).
So the brain is probably seduced by the familiar shapes into thinking the less familiar ones can be easily unraveled if they just work on the ones that look normal and then fill in the rest. But the structure of VMS does not follow natural-language patterns in terms of entropy, letter frequency (for substitution systems), or letter-positions.
As soon as they begin to intuitively sense this... even before they consciously notice it, people start compensating for the oddities by assigning multiple vowel sounds to the too-frequent vowel-like glyphs, and multiple abbreviations and endings to the too-frequent final-letters, and that's when the subjective interpretation takes hold and many eager solvers go into "fill in the blanks" mode. If they ignore all the tokens that aren't translatable using their system, the result is a deep conviction that they are on the right track.
Unfortunately, all of those who believe they are on the right track are on different tracks. So we have "solutions" in Latin (numerous Latin "solutions", each one slightly different), Greek, Czech, Italian, Spanish, Portuguese, "proto-Romance" (and numerous other polyglot assertions), Pavlavi, Sanskrit, Turkish, English, Welsh, Nahuatl, et al, all because the balance of consonant-to-vowel-shapes is somewhat similar to natural language if you ignore the repetitious and positional nature of the text. Hebrew is a little different, because it is an abjad (it will depend on whether the vowels are read as vowels or if vowels are added in) but if the vowel shapes are interpreted as vowels (as a Romanized Hebrew), then it falls in with the other languages.
There may be grains of truth in some of the solutions (small grains), but the challenge is to PROVE that it works or, at least, to explain the methodology in a way that others can replicate it and come up with similar results, ones that have some grammatical consistency if the solution is claimed to be linguistic.
I was thinking about the many "solutions" offered in the last couple of years. They tend to be from different parts of the manuscript, which makes it difficult to compare them. Maybe it would help to have a "test page" section and to designate one illustrated folio and one non-illustrated folio for anyone who offers a solution to translate and upload to one area so that different systems can be directly compared.
(13-07-2019, 04:27 PM)-JKP- Wrote: You are not allowed to view links. Register or Login to view.I was thinking about the many "solutions" offered in the last couple of years. They tend to be from different parts of the manuscript, which makes it difficult to compare them. Maybe it would help to have a "test page" section and to designate one illustrated folio and one non-illustrated folio for anyone who offers a solution to translate and upload to one area so that different systems can be directly compared.
I like that idea, but maybe there should be more than two, given the differences of herbal a, herbal b, bio b, etc. Plus this might force the issue to explain the differences, which is one of the things often ignored.
But yes i have thought more than once it would be nice to compare translations more easily to see if there are commonalities or things that can be learned from the attempts. Some might be similar in terms of building on others' work, but then that could be seen too.
I'd put effort into helping design and refine a test for solutions, or some kind of ranking system that indicates the problems the solution does and does not successfully address. But I predict they would come with a set of excuses, for example in this case "we don't master the underlying language enough to translate an entire section by ourselves".
That could be part of the test ranking... with a recommendation that those with knowledge of X language engage in review for possible rank increase.
(13-07-2019, 05:59 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.I'd put effort into helping design and refine a test for solutions, or some kind of ranking system that indicates the problems the solution does and does not successfully address. But I predict they would come with a set of excuses, for example in this case "we don't master the underlying language enough to translate an entire section by ourselves".
There is no way to force common sense, let alone a prescriptive epistemology (a solution has to pass some tests) on researchers. Working toward the remote prospect of understanding the peculiarities of Voynichese has little appeal compared to the ego-boosting quality of being THE solver (among hundreds, but all the others are wrong of course) by relying on a simple substitution, a generous dose of imagination and subjectivity and ignoring all the (apparently not so well-known) problems.
True, it couldn't be forced on researchers but such a framework could be used to quickly review a solution and if it were used consistently as others are submitted, then people concerned with checking their work before submitting might use it as a guideline.