#!/usr/bin/perl -w
use strict;

# List all non-common words in a block of text

my %stopwords = (a => 1, an => 1, in => 1, the => 1);
my %D2;

while ( <DATA> ) {
   @D2{ map { my $l = lc;
              exists $stopwords{$l} ? () : $l
            } split /\W+/
      } = ();
}

print "$_\n" for sort keys %D2;

__DATA__
This is a demo of an
interesting example
in the post
from PerlMonks.
[download]

In answer to your specific question,

    exists $stopwords{$l}?():$l

uses the ? : trinary operator to say: If $l exists in the %stopwords hash return nothing (specifically, a null list), otherwise return $l itself. (Note that it is dollar-el, not dollar-one.)

Considering my imagined context for this snippet, we have a list of common (uninteresting) words commonly called stopwords. These we want to ignore. So we read through the file (I used the DATA filehandle) a line at a time. For each line we split out the words, force each one to lowercase, drop the stopwords, and then use the others to specify a hash slice on the hash called %D2. Every key in that slice (whether new or pre-existing at that point) is set to an empty value (I'm skipping over some nuances here). When we have finished with all the data, the %D2 hash contains an entry for every "interesting" word in the data set (ignoring duplicates).

Update: Trying to stay focused on the stated problem, I avoided the issue of the variable name $l (dollar-ell). Nkuvu is correct to point this out.

I would also add that this code will return odd results for text containing words with apostrophes or hyphens: "isn't" and "good-by" will be reported as: "isn", "t", "good", "by".

So this code is not recommended for Real Text in a situation where accuracy of results matters.

------------------------------------------------------------
"Perl is a mess and that's good because the
problem space is also a mess." - Larry Wall