See HTML::TokeParser::Simple:

#!/usr/bin/perl

use strict;
use warnings;

use HTML::TokeParser::Simple;

my $open = 0;
my $match = 'Globalization';
my $uri = 'http://example.com/glossary?globalization';

my $p = HTML::TokeParser::Simple->new(*DATA);

while (my $t = $p->get_token) {

  if ($t->is_start_tag('a')) {
    $open++;
    print $t->as_is;
  }

  elsif ($t->is_end_tag('a')) {
    $open--;
    print $t->as_is;
  }

  elsif ($t->is_text) {
    my $text = $t->as_is;
    if ($text =~ /$match/) {
      if (not $open) {
        my $href = qq{<a href="$uri">$match</a>};
        $text =~ s/$match/$href/;
        print $text;
      } else {
        print $text;
      }
    } else {
      print  $text;
    }

  } else {
    print $t->as_is;
  }
}

__DATA__
Oh no, <a href="/encyclopedia/Anti-Globalization/index.html">Anti-glob
+alization</a> activists are coming! Globalization is rejected by...
[download]

Oh no, <a href="/encyclopedia/Anti-Globalization/index.html">Anti-glob
+alization</a> activists are coming!
<a href="http://example.com/glossary?globalization">Globalization</a> 
+is rejected by...
[download]