$rd is the record delimiter and $fd the field delimiter. Both must not be found in either keys or strings. I've used \n and \t for simplicity, but special characters or character sequences are probably smarter.

my $rd = "\n";
my $fd = "\t";

my %db;
$db{worda}[0] = "stringa";
$db{worda}[1] = "stringb";
$db{worda}[2] = "stringc";
$db{wordb}[0] = "stringd";
$db{wordb}[1] = "stringe";
$db{wordc}[0] = "stringf";
$db{wordd}[0] = "stringg";
$db{worde}[1] = "stringh";
$db{worde}[2] = "stringi";
$db{worde}[3] = "stringj";

for (keys %db) {
    print join $fd, $_, @{$db{$_}};
    print $rd;
}
[download]

And to reverse the generated data:

my $rd = "\n";
my $fd = "\t";

my (%db, $p, $i);

$/ = $rd;
while (<DATA>) {
    chomp; ($_, @_) = split $fd;
    $db{$_} = [@_];
}

for (sort keys %db) {
    $p = $db{$_};
    for ($i = 0; $i <= $#$p; $i++) {
        print "\$db{$_}[$i] = \"$p->[$i]\";\n";
    }
}

__DATA__
wordc    stringf
wordd    stringg
worde        stringh    stringi    stringj
worda    stringa    stringb    stringc
wordb    stringd    stringe
[download]