Use a hex-dump tool to examine the actual byte-by-byte content of the file. You need to know exactly what data this "? ?" sequence corresponds to. If they are in fact the correct byte-groups for these two characters, then your problem becomes that you haven't told whoever's processing the output file that these two characters (hence, the whole file) is "Chinese." Not knowing what character set to use, and having no equivalent in the character set being used, it's spitting out question-marks. But, that's not a fault of the data.