🧵 Untitled Thread
Anonymous at Mon, 29 Jul 2024 00:28:01 UTC No. 16301175
Would changing the rules of spelling help against frequency analysis? For example since e is the most common letter we could change a bunch of spelling to replace e with 3 so that e's frequency is split into two symbols. Or use z more often in spellings. This could be done until the distribution of all letters is uniform, but I'm not sure how useful that'd be or if it's already been done before.
Anonymous at Mon, 29 Jul 2024 01:56:43 UTC No. 16301272
>the frequency distributions of letters is bad and we must change it
Why tho
Anonymous at Mon, 29 Jul 2024 04:52:43 UTC No. 16301423
This is just like using only the most commonly used word by frequency but no one has time for that
Anonymous at Mon, 29 Jul 2024 05:38:42 UTC No. 16301459
>>16301175
someone do this for ebonics please.
Anonymous at Mon, 29 Jul 2024 05:45:41 UTC No. 16301462
>>16301272
Cause LLMs. They basically just schizo a random letter/word based on whats likely to come after their prior word. With alt speilings 1 can has tunnel into different portion of thé vectordb
Anonymous at Mon, 29 Jul 2024 11:06:45 UTC No. 16301656
>>16301462
LLMs knowing about the intricacies of language is probably helpful in some way I'm too tired to think through, but if you really wanted to then there's nothing stopping you from giving each definition of a word a different underlying representation in the db and just replacing any instance of those representations with that word.
Anonymous at Mon, 29 Jul 2024 11:25:34 UTC No. 16301671
>>16301656
Id really like to see a strictly etymological llm that operates on prefix, suffix, and morpheme level and train it across all latin, greek and germanic language sets. Any paradox words are discarded and tyr deeped most consistent root wins. What ever it spits out ill define as the true esperanto
Anonymous at Mon, 29 Jul 2024 11:53:02 UTC No. 16301690
>>16301671
If you just want a good language with a familiar set of letters then Latin is objectively the way to go. I have considered the idea of an extremely barebones language that can only be used to express basic ideas with very simple grammatical rules as some universal language that everyone can easily learn to have at least basic conversations with anyone else across the world though, it could be interesting.
Anonymous at Mon, 29 Jul 2024 12:34:19 UTC No. 16301728
>>16301690
I have been trying latin our past couple years, im a complete language-let though so learning is a slow process. Ive found hieroglpyhs more interest though of late.
Sounds interesting, id use it just as a stepping stone to learning language if any
Anonymous at Mon, 29 Jul 2024 12:38:35 UTC No. 16301734
>>16301728
Hieroglyphs are definitely interesting, but from a pure practicality perspective there's definitely a reason why most of humanity transitioned to phonetic languages. I still think that the Japanese system of each symbol representing a syllable has the most potential though, I think that with the right grammatical rules it could be used to completely replace phonetic languages with no need for symbols like Kanji.
Anonymous at Mon, 29 Jul 2024 13:11:58 UTC No. 16301762
>>16301734
>>16301734
>syllable
Same, for the practicalities of trade and commerce
Though it seems the glyphs have a home for the metaphysical
Anonymous at Mon, 29 Jul 2024 14:25:57 UTC No. 16301846
>>16301762
Ancient civilizations and topics pertaining them will always be an interesting subject, regardless of true practicality.