Please notice that

printem([@list1], [@list2]);
[download]

actually creates two anonymous arrays and copy data in memory, while

printem(\@list1, \@list2)
[download]

will pass references to the original arrays.
Copying data will prevent accidental changes of the original arrays but may be quite long and memory-hungry if the copied data is big...
Here's a small sample to picture the difference :

#!/usr/bin/perl


use strict;
use warnings;


use Data::Dumper;

my @list1 = qw(a b c);
my @list2 = qw(d e f);

sub sem {
  my $list1 = shift or die "no list1";
  my $list2 = shift or die "no list2";
  $list1->[0]='x';
  @{$list2}=qw(D E F);
}

# print original data
print "list1: " . Dumper(@list1);
print "list2: " . Dumper(@list2);

# call with copy 
sem([@list1], [@list2]);

print "list1: " . Dumper(@list1);
print "list2: " . Dumper(@list2);

# call by reference
sem(\@list1, \@list2);

print "list1: " . Dumper(@list1);
print "list2: " . Dumper(@list2);
[download]