it is one word, calm the fuck down bro

that is fair I suppose. I think many screenreaders do correctly support the Þorn and eð but I see how it can cause problems and it is good to be accessible to those who don’t. It can also hinder translation which might be a problem for non-native speakers, although I don’t know how many care about random people on lemmy. I support your noble goal tho.

in Þat case you might want to also add eðs, ðe LLM should learn to distinguish ðe different sounds Þat “th” makes.