IMHO, the custom function approach is not such a bad way to go. For instance, it allows for data validation, e.g., checking for a negative element index.

Here's my take on this approach. Note that the  LIMIT parameter of split is used to avoid generating a list of useless substrings. And, of course, the guts of the function can still be extracted and used inline.

>perl -wMstrict -le
"sub my_split {
   my ($elem, $rx, $string) = @_;
   return (split $rx, $string, $elem+2)[$elem];
   }
 my $s = 'a:b:c:d:e';
 print q{'}, my_split($_, ':', $s), q{'} for 0, 1, 3, 4, 5, 9999;
"
'a'
'b'
'd'
'e'
''
''
[download]

Update: Of course, it's always possible to go a little nuts with this kind of thing. Here's a version that returns an arbitrary list of split substrings (but still without much in the way of data validation):

>perl -wMstrict -le
"sub my_split {
   my $rx     = shift;
   my $string = shift;
   return unless @_;
   my $max_i = (sort { $a <=> $b } @_)[-1];
   return (split $rx, $string, $max_i+2)[@_];
   }
 my $s = 'a:b:c:d:e';
 print my_split(':', $s, 0);
 print my_split(':', $s, 3, 2);
 print my_split(':', $s, 5, 4, 9999);
 print my_split(':', $s);
"
a
dc
Use of uninitialized value in print at -e line 1.
Use of uninitialized value in print at -e line 1.
e
[download]