WordPerfect is not really a BiDi wordprocessor. At least the versions released in Europe and America. If you want to include an Arabic or Hebrew text, you have to enter it litterally backwards, so that it shows correctly.
Once the WordPerfect document converted by libwpd, the characters from the special Hebrew and Arabic character sets become normal UTF-8 encoded characters. Some logic inside OOo detects that the characters are from unicode range that normally belongs to RTL languages and renders them right-to-left by default. This means that all Arabic and Hebrew text that shows correctly in WordPerfect is reversed in OOo.
I would not mind to have someone with clue explain me how to force OOo to render ALL characters left-to-right, independently from the unicode range they belong to.