my @phrase = split ' ' => $phrase; 
my @random; 
 
while (@phrase) { 
    my $index = rand scalar @phrase; 
    push @random, splice( @phrase, $index, 1 ); 
} 
 
$phrase = "@random";

my $phrase = "My name is Paul"; 
my @phrase = split /\s/, $phrase; 
my $rand; 
my %rand; 
 
for (@phrase) { 
       { 
       $rand = int rand @phrase; 
       exists $rand{$rand} ? redo : $rand{$rand}++; 
       } 
       print "$phrase[$rand] "; 
 }

use Benchmark; 
 
my $phrase = 'May the force be with you.'; 
 
timethese( 500_000, { 
 
    'Kanji' =>  sub { 
                    my @phrase = split ' ' => $phrase; 
                    my @random; 
                    while (@phrase) { 
                        my $index = rand scalar @phrase; 
                        push @random, splice( @phrase, $index, 1 ); 
                    } 
 
                }, 
    'RedRum' => sub { 
                    my @phrase = split /\s/, $phrase; 
                    my $rand; 
                    my %rand; 
                    for (@phrase) { 
                        { 
                            $rand = int rand @phrase; 
                            exists $rand{$rand} ? redo : $rand{$rand}++; 
                        } 
                    } 
                } 
    } 
); 
 
 
Benchmark: timing 500000 iterations of Kanji, RedRum... 
     Kanji: 15 wallclock secs (15.76 usr +  0.00 sys = 15.76 CPU) @ 31725.89/s (n=500000) 
    RedRum: 36 wallclock secs (36.47 usr +  0.00 sys = 36.47 CPU) @ 13709.90/s (n=500000)

As a special case, specifying a PATTERN of space 
(`' '') will split on white space just as `split' 
with no arguments does. Thus, `split(' ')' can be 
used to emulate awk's default behavior, whereas 
`split(/ /)' will give you as many null initial 
fields as there are leading spaces. A `split' on 
`/\s+/' is like a `split(' ')' except that any 
leading whitespace produces a null first field. A 
`split' with no arguments really does a 
`split(' ', $_)' internally.