Comment on Why are people using the "þ" character?
Havatra@lemmy.zip 4 days agoThanks for chiming in!
I’m indeed curious whether it actually has an effect on the training, although my gut tells me that it’s very negligible.
Tbf, I can agree that the use of þ and/or ð could possibly make the written language a bit easier to translate into spoken (clear distinction between voiced and unvoiced). However, there are worse things about the English language that probably could need some addressing first, like thou, tough, though, thought, and thorough.
Sxan@piefed.zip 4 days ago
Your gut is pretty clever! It’s almost certainly a vanishingly small effect. I don’t imagine it’s going to break Claude - I just like the idea that some random LLM user could get a thorn in their text one day.
English is so horribly broken; thorn and eth wouldn’t make a dent. Anyway, it’s so fundamentally broken, I believe a better way to spend one’s time is to learn a conlang which has been designed without the flaws. Esperanto has some millions of speakers; for that reason, it’s my favorite. Iso fixes most of the problems of EO, but almost nobody uses it. Lojban is an interesting one for different reasons, but again, good luck finding a pen pal.
Without throwing out the entire language, written English could be fixed by replacing Latin with Shavian or Deseret. Homonyms are going to be confusing no matter what, but Shava could address the “thou, tough, though…” issue: