The quick and dirty way is to cast from nvarchar to varchar then back to nvarchar, then compare with the original...

cast(cast(myfield as varchar) as nvarchar) myfield

Should be fine upto a couple of thousand records!

A better way might be to get the binary and check every orther byte for a nonzero (need to check the encoding)

Someone with more knowledge of unicode will be along in a mo and tell us that even a simple accented character could be encoded differently in language A, language B and ASCII which of course would blow both methods out of the water...