brand new to hashing...i'm trying to make the script search for a partial email address

Hashing (as in perl's hashes) provides for fast lookup using exact matching only; and is entirely the wrong mechanism for partially matching anything.

If your keys are -- like your example "smith" -- actually *always* whole words, then you could index your data by whole surnames:

[0] Perl> @x = split( '[, ]', $_ ), push @{ $bySurname{ $x[0] }{ $x[1]
+ } }, [ @x[ 2, 3 ] ] for split /\n\s/, <<'END'
Smith,John (248)-555-9430 jsmith@aol.com
 Hunter,Apryl (810)-555-3029 april@showers.org
 Stewart,Pat (405)-555-8710 pats@starfleet.co.uk
 Ching,Iris (305)-555-0919 iching@zen.org
 Doe,John (212)-555-0912 jdoe@morgue.com
 Jones,Tom (312)-555-3321 tj2342@aol.com
 Smith,John (607)-555-0023 smith@pocahontas.com
 Crosby,Dave (405)-555-1516 cros@csny.org
 Johns,Pam (313)-555-6790 pj@sleepy.com
 Jeter,Linda (810)-555-8761 netless@earthlink.net
 Garland,Judy (305)-555-1231 ozgal@rainbow.com
END
;;
[0] Perl> pp %bySurname;;
(
  "Jeter",
  { Linda => [["(810)-555-8761", "netless\@earthlink.net"]] },
  "Ching",
  { Iris => [["(305)-555-0919", "iching\@zen.org"]] },
  "Smith",
  {
    John => [
          ["(248)-555-9430", "jsmith\@aol.com"],
          ["(607)-555-0023", "smith\@pocahontas.com"],
        ],
  },
  "Crosby",
  { Dave => [["(405)-555-1516", "cros\@csny.org"]] },
  "Jones",
  { Tom => [["(312)-555-3321", "tj2342\@aol.com"]] },
  "Doe",
  { John => [["(212)-555-0912", "jdoe\@morgue.com"]] },
  "Johns",
  { Pam => [["(313)-555-6790", "pj\@sleepy.com"]] },
  "Hunter",
  { Apryl => [["(810)-555-3029", "april\@showers.org"]] },
  "Garland",
  { Judy => [["(305)-555-1231", "ozgal\@rainbow.com\n"]] },
  "Stewart",
  { Pat => [["(405)-555-8710", "pats\@starfleet.co.uk"]] },
)
[download]

Which would allow you to find all those with "smith" (provided you lc the keys, which I didn't above), but won't let you find those with "jo*" in the name.

For small numbers of lines -- a few thousands or so -- I'd keep them in a single string and using a simple text search.

For a fully wild-card search of many more than that, I'd probably build a 2 or 3 consecutive characters index.