I hope the research cited is better, because the article makes a complete hash of mixing interface language, script and language actually used to communicate. A

I hope the research cited is better, because the article makes a complete hash of mixing interface language, script and language actually used to communicate. All of these are thrown together as "use of language on digital stuff". It's quite common around the world to use an interface in language X (because the options are limited) but communicate in their own language. (I personally do it all the time, so much so that I prefer an English interface over one in my own language for the sake of consistency.) Lack of support for certain kinds of script may be an issue, but that never becomes clear from the article. If the data is heavily based on the formal support for languages (instead of actual usage) then it's seriously skewed. And even if it isn't, the article doesn't tell us where the actual problem is.