You are probably reading the text as bytes and assuming ASCII. You would need to use (or create) an Unicode cvs reader. You should specify how the text should be delivered (UTF8, UTF16, etc...) or create some sort of detection scheme for that like some editors try to do. The major part is that the Chinese signs do not all fit into the space between 128 and 255 and therefore an suitable encoding must be used (or probably is already used by the creators of the cvs file but not by the cvs reader you use)
Unicode info:
http://en.wikipedia.org/wiki/Unicode[
^]
CVS Reader with Unicode support (version 2.0 up)
A Fast CSV Reader[
^]
Good luck!