In what way do all these example convert the Emphasis tag wrong?

Update: Now that you've provided some examples, I highly doubt that any of the generic ways to parse your XML will know that "inner tags" are not really tags. The easiest way is likely to reconstruct your title tag by rolling up the children tags with it. Most modules provide a ->as_XML method which you can use to turn inner tags back to their string representation.

All the above scripts are NOT properly converting the child "Emphasis"

I couldn't find a way to get the inner content of a node without getting the node's tags as well, so needed to use a regular expression to remove them. Hopefully this will get you on your way:

use Data::Dumper;
use XML::LibXML;

my $xml = q|
<Publisher>
    <UniqueDOI>978-3-642-123456</UniqueDOI>
    <ChapterInfo ChapterType="OriginalPaper">
        <Title Language="En">Is Light Blue (<Emphasis Type="Italic">az
+zurro</Emphasis>) Color Name Universal in the Italian Language?</Titl
+e>
    </ChapterInfo>
</Publisher>
|;

my $doc = XML::LibXML->load_xml(string => $xml);
my @Publishers = $doc->findnodes('//Publisher');

for my $Publisher ( @Publishers ) {
    my ($ChapterInfo) = $Publisher->findnodes('ChapterInfo');
    my ($Title)       = $ChapterInfo->findnodes('Title');
    
    # get the Title node as literal XML
    my $content = $Title->toString();
    print "Title content:\n$content\n";
    
    # remove first and last XML tags
    $content =~ s/^<[^>]*>(.*)<[^>]*>$/$1/;
    
    # construct the hash reference
    my $hash = {
        UniqueDOI    => $Publisher->findvalue('UniqueDOI'),
        ChapterInfo  => {
            ChapterType  => $ChapterInfo->getAttribute('ChapterType'),
            Title        => {
                Language    => $Title->getAttribute('Language'),
                content     => $content,
            },
        },
    };
    print Dumper($hash);
}
[download]

Output:

Title content:
<Title Language="En">Is Light Blue (<Emphasis Type="Italic">azzurro</E
+mphasis>) Color Name Universal in the Italian Language?</Title>

$VAR1 = {
    'UniqueDOI' => '978-3-642-123456',
    'ChapterInfo' => {
        'ChapterType' => 'OriginalPaper'
        'Title' => {
            'Language' => 'En',
            'content' => 'Is Light Blue (<Emphasis Type="Italic">azzur
+ro</Emphasis>) Color Name Universal in the Italian Language?'
        },
    },
};
[download]

I accept your answer. What I have provided in my OP was just a sample and my XML is big and I manipulated the XML before converting it to a HashRef.

for $TmpNode ($dom->findnodes('//Emphasis')) {
    $tStr = $TmpNode->toString(1);
    $new_node = $dom->createTextNode( "$tStr" );
    $TmpNode->replaceNode($new_node);
}
[download]

In order to isolate the problem please show us a dump of the hashref too.°

I suppose it's a question of specifying the XML format (DTD, Schema,...)

But there are still edge cases where JSON is the wrong format though ...

° never mind, read JSON in the title and oversaw the fat commas => :)

Update: I have provided the error part of the HashRef. Please note.

You can use XML::Rules and specify what tags do you want to keep as text. Something like this:

use strict;
use XML::Rules;
use Data::Dumper;

my $parser = XML::Rules->new(
    stripspaces => 2,
    rules => {
        '_default' => 'as array',
        'Publisher' => 'pass',
        'Emphasis' => sub {
            my ($tag,$attr,$parser) = @_[0,1,4];
            return $parser->ToXML($tag, $attr);
        }
    }
);

print Dumper($parser->parse(\*DATA));

__DATA__
<Publisher>
  <UniqueDOI>978-3-642-123456</UniqueDOI>
  <ChapterInfo ChapterType="OriginalPaper">
    <Title Language="En">Is Light Blue (<Emphasis Type="Italic">azzurr
+o</Emphasis> o bianco) Color Name Universal in the Italian Language?<
+/Title>
  </ChapterInfo>
</Publisher>
[download]

You can specify a comma separated list of tags in place of the 'Emphasis' and if some other tags are not allowed to repeat, you may include 'their,names' => 'as is' in the rules hash.