#!/bin/bash
#
# Copyright (C) 2004 Jörg Tiedemann  <joerg@stp.ling.uu.se>
#
# $Id: xces2dic,v 1.4 2005/11/21 17:03:44 joerg72 Exp $

grep '<wordLink ' |\
perl -e '#if ($]>=5.008){binmode(STDIN, ":encoding(utf-8)");};
         while (<>){if (/lexPair=\"([^\"]*)\"/){print "$1\n";}}' |\
sed 's/^ *//;s/ *$//;s/ *; */;/;' |\
tr ';' "\t" |\
sort |\
uniq -c |\
sort -rn