No, I don’t. I do think it might be conceptually possible and intellectually funny to come with flags that are generated from some pattern that some corpus of

No, I don’t. I do think it might be conceptually possible and intellectually funny to come with flags that are generated from some pattern that some corpus of languages exhibit. Be it sui generis through some LLM, or relying more on scholar lexical/syntactic categorizations, or even something really more trivial like using a language-code[1] as entry. On the other hand, with such an approach I’m afraid that chances are low that we can gain any adherence and a large recognition of the resulting flags from their speakers. To my mind the best approaches as of today where already exposed in this thread: something best effort to propose the most likely user recognizable language and a clearly visible option to open a locale selector. If iconography seems a must have in the UI, showing something like a A/あ icon is rather usual by now. [1] https://en.wikipedia.org/wiki/Language_code https://en.wikipedia.org/wiki/Language_code