CAVEAT:

WikiWord uses MWDumper to read Wikipedia dump files, which in turn uses Apache
Xerces to decode the XML. Xerces has a bug in the UTF-8 implementation
(persisting at least from 2.7.1 until 2.9.1) that prevents it from correctly
reading some wikipedia dumps, an example being english Wikipedia's dump from
march 2008, <enwiki-20080312-pages-articles.xml.bz2>.
See <https://issues.apache.org/jira/browse/XERCESJ-1257> for the bug report.