A hash is the right answer, but I believe that he is looking for a hash of hashes....

#!/usr/bin/perl

open(IN,"/some/file") || die "Cant open file\nReason: $!\n";

while (<IN>) {
   chomp($line = $_);

   ($first,$second) = (split(/:/, $line))[1,2];

   $fruit{$first}{count}++;
   $fruit{$first}{$second}++;
}
close(IN);

foreach $k (sort(keys(%fruit))) {
   print "$k $fruit{$k}{count}\n";
}
[download]

The reason I used count as well, is so A) I dont have to loop to figure out what my total count for $first is, and B) I also can test for $fruit{$k}{blah} and determine if there were duplicates. You could add the test within the while loop.. I.e test for $fruit{$first}{$second} and if it exists warn or something, else increment it :)... Have fun /* * And the Creator, against his better judgement, wrote man.c */