The Voynich Ninja

I have been wondering lately if we can, as a group, improve upon the transcriptions currently available online for people to use. The main one would obviously be Takahashi.
I've been reading about several theories regarding the text lately and I noticed a theme within, they are all take bad data on face value and the data they extrapolate from it is flawed. An example is the "Curve/Line theory" - many exceptions are not exceptions, just incorrect transcription. A lot of the conforming examples could be exceptions also, but how will we know without accurate data? 

For example if we look at Takahashi transcriptions. 

"EN"
16 matches

I would argue (and I believe most would agree) only a couple of the 16 matches are "en".

Most of the problem (with this example specifically) is that the transcription does not take into account that not every "n" is "n" (below), I know some will disagree, but honestly I think if you consider these to be the same glyph you must also consider "s" and "r" the same, in the same way I suppose "d" and "m" must be the same.. and there's probably more examples. They all have the same features. r, m and n start with "\", s, d and g start with "c", but there are many more obvious errors.



I'm not savvy in web design, but would it not be fairly simple to set up a version of the voynichese website where people can submit amendments, these could them be reviewed and approved or rejected and then changed or retained? I for one would be willing to help in whatever way I can, and I'm guessing many others would also. 





Just to disclaimer this a bit, I am very thankful for the work put in thus far to provide us with what we have, but I just don't think we shouldn't try to improve upon it if we can.

Perhaps you may find this useful.
voynich.nu >> Text Analysis - Transliteration of the Text
You are not allowed to view links. Register or Login to view.

It doesn't help with theories I've seen tbh, everyone uses Voynichese data. It's top of google searches and visual.
I guess the point is, can WE not provide people with the best data we have? If that's Rene's attempts to transcript or others, but (not to discredit Rene's efforts) also, no attempt is going to be flawless, a live model would be much more valuable imo

(16-10-2024, 11:04 PM)Bluetoes101 Wrote: You are not allowed to view links. Register or Login to view.Most of the problem (with this example specifically) is that the transcription does not take into account that not every "n" is "n" (below),

In the You are not allowed to view links. Register or Login to view., these two forms are pretty reliably distinguished as [b] and [n].  My feeling is that it's the distinction between [s] and [r] that tends to cause the most trouble among curve/line pairs.

Didn't Lisa once say she was planning to work on this?  I'm not sure if she did, and if so, how far the project progressed, but it might be of interest.

Apart from that, I wonder if there is a practical way for us to work on this. Maybe take x pages a day, start from Takahashi (Voynichese.com) and discuss all the problems with those pages?

A disadvantage of this approach would be that you're not looking at phenomena like "EN" throughout the MS, but only focusing on one page at a time. But of course, one can still look at cases on other pages for comparison.

(17-10-2024, 07:59 AM)Koen G Wrote: You are not allowed to view links. Register or Login to view.Apart from that, I wonder if there is a practical way for us to work on this. Maybe take x pages a day, start from Takahashi (Voynichese.com) and discuss all the problems with those pages?

With all the disagreements (just look at your thread about d/g variants) no collaborative endeavor can converge to a consistent transliteration. Questions like these (with many glyph variants not as easily recognizable as these) can't be decided by discussion:
Is it o or a? s or r? g or m?
Is cs the same as Sh?
Is Ih the same as ch?
Is ei the same as a?
Is j the same as d?
Is there a space or not?
How many @ are needed?

The Takeshi transliteration is not the best, not the most complete and not the most accurate that is available, and not everybody is using it.

There are three (publicly available) that are more accurate and two of these are complete.
The link to this information was already given by @RobGea above.

Improving on this in general is something that I still consider useful, though it is not something that should be undertaken without a considerable amount of planning (and agreement, if a team effort).

What alphabet to use?
Make a new one?

@nablator already pointed out a number of issues that one could vote on, but will not really be resolved.

The market share of the flawed TT transcription will remain disproportionate as long as voynichese.com uses it though. This website is just very convenient and intuitive, and many people will use it because they lack understanding of other methods.

If the idea is "simply" to fix obvious errors in the old Takeshi Takahashi transliteration, it's been done several times already:

- Alexander Max Bauer's A Transcription Based on a Revised Version of Takeshi Takahashi’s Transliteration You are not allowed to view links. Register or Login to view. (though it doesn't fix the b that are clearly not n, the u that are clearly not an, and other unfortunate choices of the TT transliteration)
- Konstantin Hamidullin (@farmerjohn) did a good job on You are not allowed to view links. Register or Login to view. fixing most of the errors and adding the missing parts. The browser is not running on Firefox anymore it seems.

OK, voynichese.com is certainly the most popular source of transliteration data, but it is primarily for interactive/visual use. People doing data processing / number crunching will use batch data. I am aware of numerous people doing this and writing papers based on this.

My point is that, also in order to create something 'better' than voynichese.com, the starting point should not be the TT transliteration. There are much more complete and accurate files to start from.

Indeed, @famerjohn's web site is already a big step forward, but I am not aware that his data is also available in batch form. (If would be great if it were).

I have a complete set of files that could be used to set up a next generation voynichese.com, but I don't know how to do it.