(via Silver Oliver who works at the BBC): Do tag clouds work in Chinese? (Chinese translation of this article). Rex Wong talks about the alphabetical ordering problems of non-alphabetical languages, problems sizing up Chinese fonts, and the mixing of Chinese characters and English words.
At the end of the day, though, tag clouds do seem to work fine in Chinese, as they do in Japanese. The alphabetical ordering algorythm is kind of irrelevant because tag clouds guide the eye mostly by size (a tag cloud is not used for known-item searching, where alphabetical ordering matters, but for serendipitous discovery).
Also, collation algorythms that order these languages in a “good enough” manner already exist and are built into the infrastructure we use (programming languages and databases), so you can probably safely forget about this.
I’ve been told that in Japan, tag clouds are pretty popular and work well.
Chinese tag cloud:

Japanese tag cloud:

How about tag clouds in other languages? Please point me to examples of tag clouds in as many languages as you can find, I want to try to make a collection.
Another issue of course is if you aggregate lots of content and tags, how do you deal with mixed-language tag clouds? Here’s an (older) example of mixed languages in a tagcloud:

The best approach seems to be to separate out tagclouds in different languages. There are automatic ways to do that (I have to research some more what the best ways are), and then present a tagcloud to the user in their language. (Some more thoughts on this.)
A note about alphabetical ordering of none alphabetical languages (like Chinese or Japanese): it often is possible to order these languages, using an alternative collation system called radical and stroke ordering. Below is a screenshot of an “alphabetical” index in Japanese.

And a question to finish this off: for our Japanese speakers: what is Taggy?