The Babel tool will generate characters by language group, or multiple language groups.
Typically, if you are seeing a � character, that usually indicates that you don't have a font capable of displaying the Unicode character (or the Unicode code point is not assigned a character).
Corruption is usually indicated by a single chars displaying as 2 different chars, or as the question mark.
Also, plain text will only support character sets for the default system locale of the operating system. Since you are dealing with web pages, the default encoding of the Internet is UTF-8 (unless the pages are encoded otherwise).
Also, I recommend trying Indic languages since Indic language groups are Unicode only languages and have no ANSI fallback. This way you can make sure your pages support Unicode 100%. Thai and Vietnamese are also interesting to test for tonal marks and combining characters.
CJK characters of course. Inputting the characters using the IME may result in different behavior as compared to copy/paste.