Thanks. I'll whip up a benchmark with some of the different approaches, (actually have, but don't have yours in yet. Though I'll probably switch the filehandle use.

Also, I'd probably use a hash instead of a hashref for $words as it is slightly faster.

Not sure if it's faster or not (though "feels" it), I'd probably rewrite the following to use a hash slice

# before
    if ($word =~ /$expr/) {
       for my $j (-5 .. 5) {
           $words->{$i + $j}++;
       }
    };
# after
    my $mwords = 5;
    @words{$i-$mwords..$i+$mwords} = 1
       if $word =~ /$expr/;
    # Note: Keys created, but all but first have undef values
[download]