Unicode allows the use of 16 bit characters. A Unicode string can contain almost any character used on earth, including Roman, Cyrilic, Hebrew, Arabic, Hindu, Chinese, Japanese, Korean, and others. The NCHAR, NTEXT, and NVARCHAR data types support Unicode data.
No collation that I know of will support traditional chinese without using Unicode.
I'm pretty sure that you mean "Accent Insensitive", meaning that accents are ignored, so that A, À, Ã, and all of their other accented variations are considered to be equivalent.
I'd store your data in Unicode columns, and only force collation on them when needed for things like sorting and comparison.