CircleID

The concept of language specific heirchy is not a bad one, yet attempting this initially with multiple language TLDs is something that may not scale very well, as is discussed both in John's IETF posting, and in the comments to Danny Younger's recent article [Languages in the Root: A TLD Launch Strategy Based on ISO 639].

What might make most sense is to create one TLD as a proof of concept, like .lang or .utf8, delegate it to the Unicode Consortium or other language focused organization, and let them dole out subdomains or stubs underneath it to the appropriate bodies (IE. The Fronch could operate .fr.utf8, China could potentially operate .zh.utf8 through their information ministry and CNNIC). 

The party that is entrusted with this delegation could subsequently identify some application development guidelines similar to the great work that the IDNA working group of the IETF produced, to allow application level support of these language based destinations.

Jothan, your comment seems a little wide of the mark, given the article to which it is attached. The primary issue addressed by this article is that of allowing non-English-speakers the possibility of addressing *existing* TLDs in their own language. The suggestion here is that the localisation may be carried out purely by the user interface, without touching DNS infrastructure at all.

The nicest thing about this idea is that anyone with sufficient interest could go right ahead and implement a proof of concept by extending an existing open source browser and/or email client. It sounds like the path of least resistance to me.

I can see where my comment seems obtuse to the discussion, but I did want to bring attention to Danny's Language TLD proposal, because it found its way to ICANN's TLD comment forum.