Someone's already done the hard work for you: Lingua::EN::Numericalize.

Thank you both very much! I could not find this package, but I knew I ran into something in the past. This looks to be it.

#!/usr/bin/perl                                                       
+                                                                     
+                                 
use Lingua::EN::Numericalize;                                         
+                                                                     
+                                               
print str2nbr('First, Second, Third, Twentythird');
[download]

There is a easy way to remedy this...

 for my $text (qw/first third thirteenth twentyfirst twentysecond oneh
+undredone onehundredtwelve onehundredeleven/) {

 my $number = str2nbr($text);

 my $sufix = 'th';

 if ( $number =~ /^1$/ or $number =~ /[^1]1$/ ) { $sufix = 'st'; }
 if ( $number =~ /^2$/ or $number =~ /[^1]2$/ ) { $sufix = 'nd'; }
 if ( $number =~ /^3$/ or $number =~ /[^1]3$/ ) { $sufix = 'rd'; }

 print $number . $sufix, $/;

 }
[download]

and i get 1st, 3rd, 13th, 21st, 22nd, 101st, 112th, 111th I believe this will work for all your cases, i would test it more though. ED: changed a var name, didn't like it after i looked at again

use strict;

my $hash = {
    'first' => '1st',
    'second' => '2nd',
    'third' => '3rd',

    'twentythird' => '23rd',
};

sub ordinal_t2n {
     my @ret;

     for (@_) {
         my $b = "$_";
         $b =~ s/(\w+)/$hash->{lc($1)} or $1/eg;
         push @ret, $b;
     }
     @ret;
}

print ordinal_t2n('First, Second, Third, Twentythird',
                  'First, Second, Three is not here Third, Twentythird
+'),"\n";
[download]