O wise Perl monks,

I have written a script that parses a file and applies a number of text substitutions. I have a dictionary (let's call it %patdic) of search/substitute pairs and, for each input text line, I run something like

foreach my $search (keys %patdic) {
    s/$search/$patdic{$search}/g;
}
[download]

This works well as long as $patdic{$search} does not contain the $1 construct. Consider the following example:

#!/usr/bin/env perl

use strict;
use warnings;

$_=q{nell'amaca};
my $search=q{nell'([[:alpha:]]+a)};
my $subst='na $1';

s/$search/$subst/g;
print;
print "\n";
[download]

This will return "na $1", while I would like it to return "na amaca". Any pointers, please?