Here's a condensed form that can correctly differentiate all cases as per sauoq's requirements elsewhere in this thread. It certainly isn't something for the beginner's class though.

my ($foo, $bar, $baz, $quux);
defined && /^([[:alpha:]]+)$/ ? $foo = $1 : taint_fail('foo') for scal
+ar param('foo');
defined && /^([[:alpha:]]+)$/ ? $bar = $1 : taint_fail('bar') for scal
+ar param('bar');
defined && /^([[:alpha:]]+)$/ ? $baz = $1 : taint_fail('baz') for scal
+ar param('baz');
defined && /^([[:alpha:]]+)$/ ? $quux = $1 : taint_fail('quux') for sc
+alar param('quux');
[download]

But any remotely hubristic programmer will of course write this:

my %check = (
    foo => qr/^([[:alpha:]]+)$/,
    bar => qr/^([[:alpha:]]+)$/,
    baz => qr/^([[:alpha:]]+)$/,
    quux => qr/^([[:alpha:]]+)$/,
);

my %f;

for my $pname (keys %check) {;
    defined && /$check{$pname}/ ? $f{$pname} = $1 : untaint_fail($pnam
+e)
        for scalar param($pname);
}
[download]

This will call untaint_fail only when the regex failed, and preserve undef vs empty string where they're valid input.

Update: Oops. The code would previously leave things tainted since it was assigning $1 back to the tainted variable. Shifting things around minorly fixed that.

Makeshifts last the longest.