This is an interesting problem, even though it seems a little homework-like.

Here is my solution.

#!/usr/bin/perl -w
#
#  Look for palindromes in a file.
#
#  Usage: perl pal.pl file1 [file2] .. [fileN]
#

use strict;
use warnings;

#
#  Read the file.
#
while( <> )
{
    my $word = $_;

    # remove newline.
    chomp( $word );

    # skip short words
    next if ( length( $word ) < 4 );

    # reversal.
    my $text = reverse $word;

    # test + print
    print "Word: $word + $text\n" if ( lc( $word ) eq lc( $text ) );
}
[download]

I see some output such as:

skx@desktop:~$ perl pal.pl /usr/share/dict/words
Word: Anna + annA
..
Word: Hannah + hannaH
..
Word: redder + redder
Word: refer + refer
[download]