- Options
- Login
- Desktop Style
Options
Making a test for translations unbiased by images
- Making a test for translations unbiased by images
-
Making a test for translations unbiased by images
qoltedy > Yesterday, 03:50 AM
I'm wondering if anyone would be willing to help me with my Voynich decoding theory by making a blind test to see if I'm being biased by the images while attempting to translate using my method.
If someone could get me an EVA transcription of one of the balneological folios, separated by line and paragraph, as well as the labels simply labeled "Label 1, label 2, etc" then I could test whether my translation method is accurate even without being biased by the images.
So far I've only attempted translating f77r, so any folio that's not that I wouldn't recognize.
For example, for f77r, I'd want something like
"Label 1: dorchdar
Label 2: dotedy
...
Paragraph 1:
Line 1: poldarair ol qokol chey ...
Line 2: tolchdar shey qotaiin...
Paragraph 2:
Line 1: otedy qokor shedy..."
I would do it myself but I don't want to bias myself by associating the images with the text before I translate it -
RE: Making a test for translations unbiased by images
oshfdk > Yesterday, 08:55 AM
Hi and welcome!
How are you going to verify the translation?
If you wish, I can offer you a simpler test to check whether you can identify parts of real Voynich transcription and gibberish. I can give you 10 random lines from the transcription of the manuscript, each of them in two variants: the original and one with word and word pieces scrambled randomly, except for the very start and the very beginning of the line.
If your method can successfully identify in 8 out of 10 pairs which version is the meaningful text, and which one is random gibberish, then it would make sense to think about testing the quality of translation. Note that I won't need you to give us your answers, since in any case it's very easy to cheat on this test by searching for the lines in the transcription. It's only for your own self-evaluation of the method. So, I'm going to give you the test file and after you confirm that you have finished with the identification, I'll give you the answers. -
RE: Making a test for translations unbiased by images
nablator > Yesterday, 11:33 AM
(Yesterday, 03:50 AM)qoltedy Wrote: You are not allowed to view links. Register or Login to view.If someone could get me an EVA transcription of one of the balneological folios, separated by line and paragraph, as well as the labels simply labeled "Label 1, label 2, etc" then I could test whether my translation method is accurate even without being biased by the images.
Hi,
You can find EVA transliterations on René Zandbergen's website: You are not allowed to view links. Register or Login to view.
You are not allowed to view links. Register or Login to view.
You are not allowed to view links. Register or Login to view.
You are not allowed to view links. Register or Login to view.
etc.
The labels are type "L":
<f75v.18,@Ln> okshy
See Table 9: Definition of locus types in the IVTFF format documentation: You are not allowed to view links. Register or Login to view. -
RE: Making a test for translations unbiased by images
RadioFM > Yesterday, 02:23 PM
Hi! Ideally, these words should be translatable:
opcholdy
opcholdg
asal
dary
daldal
dydyd
chedy
c'heey
okal
okeor
Also these lines:
"otcholocThol chol chol chody kan"
"tchotchol chol cThy"
"qokeedy qokeedy qokedy qokedy qokeedy ldy" -
RE: Making a test for translations unbiased by images
qoltedy > Yesterday, 04:23 PM
(Yesterday, 08:55 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.Hi and welcome!
How are you going to verify the translation?
If you wish, I can offer you a simpler test to check whether you can identify parts of real Voynich transcription and gibberish. I can give you 10 random lines from the transcription of the manuscript, each of them in two variants: the original and one with word and word pieces scrambled randomly, except for the very start and the very beginning of the line.
If your method can successfully identify in 8 out of 10 pairs which version is the meaningful text, and which one is random gibberish, then it would make sense to think about testing the quality of translation. Note that I won't need you to give us your answers, since in any case it's very easy to cheat on this test by searching for the lines in the transcription. It's only for your own self-evaluation of the method. So, I'm going to give you the test file and after you confirm that you have finished with the identification, I'll give you the answers.
In theory, one could verify the translation by seeing if the passage actually makes sense or just outputs gibberish, and whether it can reasonably match the visual content. And word definitions would need to be reasonably consistent across the entire text as well.
I would appreciate your test with the word scramblings so I can see if my method truly holds up. -
RE: Making a test for translations unbiased by images
MarcoP > Yesterday, 05:56 PM
(Yesterday, 04:23 PM)qoltedy Wrote: You are not allowed to view links. Register or Login to view.In theory, one could verify the translation by seeing if the passage actually makes sense or just outputs gibberish, and whether it can reasonably match the visual content.
The two criteria you mention are highly subjective, so they don’t help much for validation. Many medieval texts don’t make sense from a rational point of view, and Voynich images are so obscure that we can argue endlessly about what is a “reasonable match” according to our subjective taste.
It’s much better if one focuses on grammar, a subject about which it’s easier to reach some consensus.
The passage must be grammatically correct according to the grammar of an actual language. The translation is not even interesting from my personal point of view: what interests me is that the “translator” can point out the grammar of the supposed underlying language and why the words appear in that specific order.
hell it self is weary of earth: for why? the son of darknesse cometh now to challenge his right: and seeing all things prepared and provided, desireth to establish himself a kingdom
Does the passage above make sense? We can have different opinions about that. But I think most people would agree that it is grammatical English (e.g. verbs agree with subjects “hell is weary” is OK, “hell are weary” isn’t; "and" connects two words sharing the same form: "prepared and provided" is OK - "seeing all things prepared and self" is not OK).
Similarly, most people would agree that the following passage is not English, but just a word-salad:
earth: darknesse the himself it a now to to challenge hell why? desireth all of of right: is son and and prepared for self weary things seeing cometh provided, kingdom establish his -
RE: Making a test for translations unbiased by images
oshfdk > 8 hours ago
(Yesterday, 04:23 PM)qoltedy Wrote: You are not allowed to view links. Register or Login to view.In theory, one could verify the translation by seeing if the passage actually makes sense or just outputs gibberish, and whether it can reasonably match the visual content. And word definitions would need to be reasonably consistent across the entire text as well.
"Translation" usually implies a lot of flexibility in interpreting the text. If your method is deterministic and can be reproduced following simple unambiguous rules with no or little possibility of multiple interpretations, then yes, producing meaningful passages would be very interesting.
(Yesterday, 04:23 PM)qoltedy Wrote: You are not allowed to view links. Register or Login to view.I would appreciate your test with the word scramblings so I can see if my method truly holds up.
Thank you! It is very important to test new approaches that help people self-evaluate their solutions.
I'm attaching a text file with 10 pairs of lines. In each pair one line is verbatim as it appears in ZL EVA transcription, and another is the same line with most words and word pieces shuffled.
Your mission, should you choose to accept it, is to identify for each pair which of the two lines is the original (for example, it produces some meaningful translation) and which is the scrambled version (produces word salad/gibberish). You can do this in public (post your results) or in private (just write back when you are done), it's up to you. I will then post the list of the original unscrambled lines. The main goal is for you to understand whether your method can successfully identify meaningful text. Since it's trivial to just find the correct lines in the transcription file, publishing the results doesn't prove anything and not publishing them doesn't mean anything bad either.
Just in case the following is a technical description of how this file was created. The source lines were taken from ZL EVA transcription using the following command:
Code:cat zl.txt| perl -ne 'print if(/\S{70,}/)' | grep -o -E "\S+$" | grep -v -E '[<{:?]' | shuf -n 10
Which takes all lines longer than 70 characters, removes the locus tag, discards all lines containing special formatting, uncertain readings, etc and then randomly selects 10 lines from the result.
Then these 10 lines were passed through the scrambler code, which keeps the very beginning and ending of each line and mixes word and word pieces splitting at spaces and non-pedestaled gallows characters and shuffling the pieces in a way that preserves their preceding and following separators (so, if a certain chunk was between a gallows character and a space, it will end up between a gallows character and a space).
pairstask.txt (Size: 1.44 KB / Downloads: 6)
-
RE: Making a test for translations unbiased by images
qoltedy > 8 hours ago
(Yesterday, 05:56 PM)MarcoP Wrote: You are not allowed to view links. Register or Login to view.
(Yesterday, 04:23 PM)qoltedy Wrote: You are not allowed to view links. Register or Login to view.In theory, one could verify the translation by seeing if the passage actually makes sense or just outputs gibberish, and whether it can reasonably match the visual content.
The two criteria you mention are highly subjective, so they don’t help much for validation. Many medieval texts don’t make sense from a rational point of view, and Voynich images are so obscure that we can argue endlessly about what is a “reasonable match” according to our subjective taste.
It’s much better if one focuses on grammar, a subject about which it’s easier to reach some consensus.
The passage must be grammatically correct according to the grammar of an actual language. The translation is not even interesting from my personal point of view: what interests me is that the “translator” can point out the grammar of the supposed underlying language and why the words appear in that specific order.
hell it self is weary of earth: for why? the son of darknesse cometh now to challenge his right: and seeing all things prepared and provided, desireth to establish himself a kingdom
Does the passage above make sense? We can have different opinions about that. But I think most people would agree that it is grammatical English (e.g. verbs agree with subjects “hell is weary” is OK, “hell are weary” isn’t; "and" connects two words sharing the same form: "prepared and provided" is OK - "seeing all things prepared and self" is not OK).
Similarly, most people would agree that the following passage is not English, but just a word-salad:
earth: darknesse the himself it a now to to challenge hell why? desireth all of of right: is son and and prepared for self weary things seeing cometh provided, kingdom establish his
I would argue that if we are to take the hypothesis of an a priori constructed language seriously, rather than a cipher of an existing language, even the grammar structure could again be interpreted as subjective. That passage may not make sense from a traditional SVO grammar structure, but with an arbitrarily constructed grammar structure, your above example passage COULD make sense in a constructed language. The challenge would be the consistency of that grammar structure throughout the manuscript, if it always followed SOV/VSO/SVO in some predictable scheme.
I would say that consistency across the entire corpus in terms of word meanings and grammar structure, is likely one of the few ways to verify a translation if it's a constructed language.