Micro$oft Word uses non-standard characters and they create problems in using simplexml_load_file.
Many systems include non-standard Word character in their implementation of ISO-8859-1. So an XML document containing that characters can appear well-formed (i.e.) to many browsers. But if you try to load this kind of documents with simplexml_load_file you'll have a little bunch of troubles..
I believe that this is exactly the same question discussed in htmlentites. Following notes to htmlentitles are interesting here too (given in the reverse order, to grant the history):
http://it.php.net/manual/en/function.htmlentities.php#26379
http://it.php.net/manual/en/function.htmlentities.php#41152
http://it.php.net/manual/en/function.htmlentities.php#42126
http://it.php.net/manual/en/function.htmlentities.php#42511