(20-04-2025, 08:41 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view. (20-04-2025, 07:24 PM)project963 Wrote: You are not allowed to view links. Register or Login to view.Ok, I runned some simulations and changing from newer to old (very old) is ok as far I see. Since the word is in its sub group based on my model I had to do 4 more changes, se attachment. I will upload new spreadsheet later tonight
Would be great to also check how your system would react if the meaning of daiin is replaced not with an antonym, but with a completely unrelated adjective. Would there be any contradictions or problems if daiin was set to mean, say, transparent (or opaque)?
I'm trying to understand how much freedom the system allows in selecting meanings.
Since it's base on a category system of the first glyphs in a word it must be within same word group. In my system 8 is only timeline words.
Update files:
Analysis: You are not allowed to view links.
Register or
Login to view.
Dictionary: You are not allowed to view links.
Register or
Login to view.
Dictionary is best viewed in a pc since is so cramped and may move boxes to wrong colours
This risks becoming repetitive, but 'older' has exactly the same problem as 'newer'. In any text of any length, such a word is very far down the list of most frequent words.
daiin is the most frequent word in the MS.
It is worth taking some sample texts and see what is the most frequent word. One should probably do this with langauges with different levels of inflection.
I guess the modern approach would be to just ask some AI chat bot...
(21-04-2025, 06:54 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.This risks becoming repetitive, but 'older' has exactly the same problem as 'newer'. In any text of any length, such a word is very far down the list of most frequent words.
daiin is the most frequent word in the MS.
It is worth taking some sample texts and see what is the most frequent word. One should probably do this with langauges with different levels of inflection.
I guess the modern approach would be to just ask some AI chat bot...
When I first tested old and newer I had perhaps just analysed 200-300 words. I have now analysed 970 witch makes it possible to do a better testing. But the correct word must be a timeline word and it must be possible to "bend" it because of words with similar start glyphs but different end glyphs.
The constraints are, in general, are explained in the first page of the spreadsheet called "rules of glyphs".
(21-04-2025, 11:03 AM)project963 Wrote: You are not allowed to view links. Register or Login to view.When I first tested old and newer I had perhaps just analysed 200-300 words. I have now analysed 970 witch makes it possible to do a better testing. But the correct word must be a timeline word and it must be possible to "bend" it because of words with similar start glyphs but different end glyphs.
The constraints are, in general, are explained in the first page of the spreadsheet called "rules of glyphs".
So it could be something like "later/earlier" with possible frequent usage as "later" meaning "then"? ("get two naked ladies, then add a pinch of saffron, then check if Mars is in Leo, then follow the instructions from f111r")
(21-04-2025, 11:11 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view. (21-04-2025, 11:03 AM)project963 Wrote: You are not allowed to view links. Register or Login to view.When I first tested old and newer I had perhaps just analysed 200-300 words. I have now analysed 970 witch makes it possible to do a better testing. But the correct word must be a timeline word and it must be possible to "bend" it because of words with similar start glyphs but different end glyphs.
The constraints are, in general, are explained in the first page of the spreadsheet called "rules of glyphs".
So it could be something like "later/earlier" with possible frequent usage as "later" meaning "then"? ("get two naked ladies, then add a pinch of saffron, then check if Mars is in Leo, then follow the instructions from f111r")
I tried to find a different solution. I now call this sub category "value over time" instead. It's like looking at something old/older etc. at giving it a "value". I think it could work for now
I have run simulations on my 9 pages and it looks ok. Like 1r line 2: "leaf shape greenish colour is very very valuable". I will upload updated spreadsheet later.
Could you identify grammar structures in your reconstruction? For example:
1) Do adjectives precede or follow the nouns? ("green leaf" or "leaf green")
2) Do verbs precede or follow the subject? ("plant grows" or "grows plant")
3) Do objects precede of follow the verb? ("take the root" or "the root take")
4) Sorting order of adjectives? ("small old green leaf" or "old green small leaf", etc.)
This might help in identifying the language family.
(21-04-2025, 07:36 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.Could you identify grammar structures in your reconstruction? For example:
1) Do adjectives precede or follow the nouns? ("green leaf" or "leaf green")
2) Do verbs precede or follow the subject? ("plant grows" or "grows plant")
3) Do objects precede of follow the verb? ("take the root" or "the root take")
4) Sorting order of adjectives? ("small old green leaf" or "old green small leaf", etc.)
This might help in identifying the language family.
These questions are on the border of my competence as I have limited linguistic education. Please remember that this project should be considered as a basic model setup. Grammar etc. should be somewhat secondary, at least in the initial phase. But I can keep this in mind as it relevant and I understand the value.
But notice that I have change old/newer etc. with a new group of words: valuable ISH.
The other issue I have is that the text seems to play tricks as I unfold the basic grammar; like when I discovered direction words like "middel/topp/bottom" etc. it later showed that it was not 100% as the right grammar was LOOK "middle/topp/bottom" etc. Yes its appears almost the same but not.
I have also found a trick in the left bottom part if 85v where you see a circle with a triangle inside. The first to words are "circle (form)" + "triangle". Yes it's not a real sentence but for me it looks like word + word sentence.
These thing makes me personally more and more humbled for the deeps of this language that will be discovered later
Today I will finish translating the 10th folio. I will then do more translation words tests for the next days. That should give some new knowledge and maybe new subcategories.
Sometimes I feel that the language is so bizarre that it could be a version of what the old testament mentions as the universal language of old Babylon. Yes something so easy that anyone can understand it once you learned the basics. But that's just a crazy idea with no proof I guess
I think it makes sense to spend some time on identifying the word order patterns. If they are ignored completely, given a high level of flexibility in the system, it seems likely that your project might converge on a mapping that is not compatible with any natural language.
(22-04-2025, 12:06 PM)project963 Wrote: You are not allowed to view links. Register or Login to view.But notice that I have change old/newer etc. with a new group of words: valuable ISH.
What makes you think that "valuable" is better than "older/newer"?
Quote:The other issue I have is that the text seems to play tricks as I unfold the basic grammar; like when I discovered direction words like "middel/topp/bottom" etc.
Same question for any "discovered" word: how do
you know any of them is correct?
If you blindly trust your model to deliver accurate translations, have you previously tested it on texts with a known meaning, as a control experiment? Another control experiment would be to train your model with the same data (VM pages) but in a different order: would your model output a different translation?
(22-04-2025, 01:10 PM)nablator Wrote: You are not allowed to view links. Register or Login to view. (22-04-2025, 12:06 PM)project963 Wrote: You are not allowed to view links. Register or Login to view.But notice that I have change old/newer etc. with a new group of words: valuable ISH.
What makes you think that "valuable" is better than "older/newer"?
Quote:The other issue I have is that the text seems to play tricks as I unfold the basic grammar; like when I discovered direction words like "middel/topp/bottom" etc.
Same question for any "discovered" word: how do you know any of them is correct?
If you blindly trust your model to deliver accurate translations, have you previously tested it on texts with a known meaning, as a control experiment? Another control experiment would be to input the same data (VM pages) in a different order: would your model output a different translation?
When I started my project back in September 2024 I had absolutely no clue what this language was. I was looking at 57v for two weeks and I thought that the glyphs looked so nice that it must be a real language, if u know what I mean.
So the first thing I did was to startup SketchUp/3d software and analyse 57r in 3d. That gave little mening. Next I picked up a book that I had been reading lately about machine learning. I thought maybe coding worked better? So I started to give each word a code starting with w0001. I started looking at pictures and shortly I found to pictures with same words. Then I got an idea that the meaning must be like a "description" word, not names etc. The first word was "connecting", like in the upper left corner of 67v with the picture of three faces connected with a line. After that with this system thinking it just expanded. And with my system I can also identified words in the same sub category without having to look at a picture. What I'm saying is that this system allows human intuition to replace missing pictures.
The control part of the grammar should be solved when building sentences. If sentences produce bad grammar some of this can be solved within category rules. And if category rules don't work then we will know that the model doesn't work in the future.
The point of publishing a work this in unfinished stage is also that, if everything goes well, people can see how the system unfolded in the future. Perhaps there is a academic value to play with open hands?