There is a problem when adding non-english languages into zimbra spellchecker. Although process basically works, when using special chars results are bad displayed. Also, if special chars are checked, word suggestions are really strange.
The problem is due to word splitting (for example with words with accents or tilde, or any kind of special char) and aspell encoding (iso-8859-1 by default).
To solve this problem, you have to modify aspell.php located at /opt/zimbra/httpd/htdocs/aspell.php
1- Replace this line:
$words = preg_split('/[^\w\'-]+/', $text);
with this other:
$words = preg_split('/[^\w\'-áéíóúäëïöüàèùìòñçâêîôû]+/', $text);
2- Add an utf-8 conversion for output text after this line:
$suggestions = implode(",", pspell_suggest($dictionary, $word));
Tested with spanish version. This modification should also work for french, german, portuguese, italian and others western europe languages. Nordic languages perhaps have to add extra chars at preg_split line.
For detailed info, I have updated New dictionaries :