Unicode type acceptable to WordSmith is actually utf-16.
The unrecognized characters might not be Chinese characters, and they are likely to be full-mode punctuations or other invisible characters.
Can you upload the original text and the screenshot of the concordances lines shown in the WS window?