My takeaway is that turkish "I" and English "I" as well as English "æ" and Icelandic "æ" are crammed into the same character even though that causes all sorts of problems, but there are about 20 different characters that represent "x"...
The solution to the problem, by the way, is to determine what the user expects and give it to them. That is, you can't just sort words without defining what sort you want to do (according to German phone book sort? Portuguese dictionary sort? etc.).