Linguistic identification can in principle be a bit smarter than that. Aside from the obvious "if it has kana it is Japanese" metric, there are also differences in code point choice which differentiate Simplified, Traditional, and Japanese. Fully accurate identification is certainly expensive, but quick early-out heuristics could be developed.