Have this fixed in 7.4. When language is set to UTF-8 the wordwrap function is used after utf8_decoding the string, so it splits and the proper location and then utf8_encodes the remaining string.