Rather than index, which will not accept a pattern, use a character class in a capturing regex, match globally in a while loop and use the @- array, see perlvar. Note that a hyphen (-) in a character class has to be the first or last character or it will be interpreted as a range constructor, e.g. [a-z].

$ perl -E '
> $line =
>    q{The line, as specified, has: no digits (0 to 9); letters!};
> say qq{>$1< found at $-[ 0 ]} while
>    $line =~ m{([ (),.;:?!-])}g;'
> < found at 3
>,< found at 8
> < found at 9
> < found at 12
>,< found at 22
> < found at 23
>:< found at 27
> < found at 28
> < found at 31
> < found at 38
>(< found at 39
> < found at 41
> < found at 44
>)< found at 46
>;< found at 47
> < found at 48
>!< found at 56
$
[download]

I hope this is helpful.

If 5.10+ is available, use of the  //p regex modifier allows  ${^MATCH} (and friends) to be used eliminating the need for an explicit capture and perhaps slightly speeding the matching:

>perl -wMstrict -lE
"my $line = q{as specified, line has: no digits (0-9); letters!};
 ;;
 say qq{'${^MATCH}' found at $-[0]}
   while $line =~ m{ [ (),.;:?!-] }xmsgp;
"
' ' found at 2
',' found at 12
' ' found at 13
' ' found at 18
':' found at 22
' ' found at 23
' ' found at 26
' ' found at 33
'(' found at 34
'-' found at 36
')' found at 38
';' found at 39
' ' found at 40
'!' found at 48
[download]

use Modern::Perl;

my $string = 'Search, find and tell me! where, oh where are those funn
+y characters???';

my @positions;
push @positions, [$1, pos($string)] while ($string =~ m/([-\s,.!?;])/g
+);
}
[download]

see pos to find out the position of the last match.