Support for languages not otherwise listed, including Thai,
Vietnamese, Hebrew, Indonesian, and plenty more.  The split is
made simply on the basis of the size of the support, to keep
both collection sizes and the number of collections reasonable.