Well since no one mentioned the word boundary method, I submit the following:

#!/usr/bin/perl

$sentence= $a= "I sit here and think-  
    what the-hell am  
    i doing with perl?";

@segments = split(/\b$word\b/, $sentence);
#@segments = grep { $_ ne '' } @segments; #remove undef elements 
@segments = grep { $_ ne ' ' } @segments; #remove space elements 

#print "$#segments\n"; 
print "$sentence\n";

$"="\n";
print "@segments\n";
[download]

I don't know the best way of getting rid of the newlines. Output:

I
sit
here
and
think
- 
    
what
the
-
hell
am
 
    
i
doing
with
perl
?
[download]