A decade+ ago, I was working on automated essay scoring for a large testing corporation and the US gov't asked if we could adapt the engine to rating sentiment and emotionality to specific writings. I told the boss, sure...yeah...it would be simple. I mean, I got into AI trying to teach computers how to read music and identify themes and otherwise, this shouldn't be too hard.
Turns out, it was actually easier than either of the other two things we had been working on. And it is pretty universal regardless of the language we used. Hell, it detects sarcasm MUCH better than a 20 year old. I realize this isn't really a feat of skill when used in this comparison.
Given the technology these days and the heuristics available? I'm surprised it isn't much further used. Then again, the work I did really wasn't used in the public, so...there is probably a lot more research and implementation than folks know.