@hongminhee even in European languages: it can impact sorting order or capitalisation to start with. Variations like the ones in East Asian languages are also present. The only reason why Western languages can mostly ignore that problem is because Unicode has a large number of glyph variations for the Latin alphabet but that creates other problems such as canonicalisation.

Anyway, all this to say I agree and anyone who says the lang attribute is useless has some learning to do.