So far, I haven't been able to extract the xml:lang attribute using XML::Parser. It seems to ignore attributes with the xml: prefix.

My experience doesn't match yours. I can clearly see the attribute 'xml:lang' in the list of attributes.

use XML::Parser;
use Data::Dumper;

my $parser = XML::Parser->new(Handlers =>
  {
    Start => sub { print Dumper [ @_[1 .. $#_] ] }
  }
);
$parser->parse(\*DATA);

__DATA__
<FILE>
<FOO xml:lang="en">
<BAR/>
</FOO>
<FOO xml:lang="ru">
<BAR/>
</FOO>
</FILE>
[download]

Result:

$VAR1 = [
          'FILE'
        ];
$VAR1 = [
          'FOO',
          'xml:lang',
          'en'
        ];
$VAR1 = [
          'BAR'
        ];
$VAR1 = [
          'FOO',
          'xml:lang',
          'ru'
        ];
$VAR1 = [
          'BAR'
        ];
[download]

Anyway: you might be interested in using a parser that can handle XPath, as it might make filtering out just the contents you want, just a little bit easier.

Although personally, I think XML::Parser is fine... :)