Wow! That's a lot of regular expressions going on there...

Personally I'd do something more akin to:

my @stopwords = qw/ i'd add all my stop words in here /;
my %stop = map { lc $_ => 1 } @stopwords;

sub findwords {
  my $string = shift;
  my (@ok, %seen);
  while ($string =~ /((\w|')+)/g) {
    push @ok, $1 unless $stop{lc $1} or $seen{lc $1}++;
  }
  return @ok;
}}
[download]

My tests show this as coming out about 2 orders of magnitude faster, and it also copes better with apostrophized words that aren't in the stop list.

Tony