Author of the Kalamang paper here. We’ve thought about this a good amount (e.g. there are interesting Mesoamerican scripts), but ultimately we decided to work on low-resource languages, as they’re much more useful. It’s also possible to use them as an evaluation benchmark, which isn’t really possible if nobody speaks the language. We’d like to expand the scope beyond Kalamang, though, and maybe at some point we’ll investigate an undeciphered language as well.