An extension shall be assigned to each language ( according to the three-letter code ISO 639-2 ) so that sites or sites' versions written in specific languages can be easily found and identified. It would facilitate greatly the task of search engines and would foster linguistic diversity, which is reaching its greatest wealth in Africa. These set of extensions shann be managed by the LSWgTLDs authority which is proposed to be the World Language Diversity Network ( WLDN ) recently formed within the framework of the international public law. The main initiator of the REDILI has been of the Académie Africaine des Langues which as been recognized recently, not long after the WSIS in Tunis, as an organ of the African Union ( previously the Organization of African Unity ).

First, a choice shall be made swiftly by the LSWgTLD authority, at the start of the implementation of the LSWgTLDs among the various technical ways, without waiting hopelessly for an ICANN decision, to make use of multilingual URLs and character sets and writing systems. The choice of the LSWgTLD authority concerning multilingual URLs shall be followed by all other SWgTLDs extension. The LSWgTLD authority shall be the authority concerning all SWgTLDs concerning multilingual URLs and writing systems that could be implemented in a progressive fashion in other SWgTLDs, not necessarily at their inception, in order to allow maximum flexibility and lower burden.

The LSWgTLD proposal is not made for the sake of selling yet another batch of domain names, as ICANN is perceived to do by many detractors, but to empower technology advances. that the world is entitled to benefit from. In order that linguistic diversity be effectively preserved and enhanced, there is a need that people can access on the Web, important documents written not only in the language they understand, but also in the language they cherish. Automatic machine translation tools are giving sometimes surprisingly good outputs, but most often they are producing contents of a very unequal quality, where good translations are sitting besides barely understandable babbling, sometimes laughable, sometimes dangerous because it could means the reverse of what it is intended, to the point of being even insulting. This is because automatic translation tools have difficulties in catching semantics. Automatic translation would be much improved if machine translation tools could work with the help of several human certified translations in various languages. For example, if the same document has been made available in English and in French by the authors on the same site, and translated by human users in Russian and Korean on other sites, it would be tremendous advantage for automatic translation tools to have access and make use of all existing versions in different language of the same document. For example "Soci´té Civile" would not be translated in yet other languages such as Italian as "Civil Company" with the help the english version. Of course, it is required that the translation tools could retrieve and identify the various versions at different locations, therefore the need for the ODRI, as well as standardized metadata or ontologies. It is hoped that if a document is available in three or four different languages, the automatic translation making use of the combination of those different langauges would be rather good. Of course, scientific research to develop software translation tools that could make use of a combination of existing translations of the same document should be strongly encouraged. Furthermore, it would seem judicious to foster the availability of human-certified translations in as many languages as possible to further enhance the efficiency of automatic translation tools and to promote linguistic diversity. Therefore, each time an automatic translation tool is providing on request a translation, the translated text will be displayed on a WIKI so that a good speaker in the target language could correct the mistakes that have been made in the automatic translation. WIKIs ( eg WikiPedia ) have been in general quite successful with few defacements. Various certification strategies of a human-corrected translation could be studied and tested. One could be that tentative human-corrected translations can be made available to readers along with the automated translation, so that readers could rate the various human-corrected translations. When one human-corrected translation has been best rated for a certain duration, then this translation could be retained as human-certified. Therefore it is proposed a set of measures to improve the efficiency of automatic translation tools. and to hopefully achieve multilingualisms of the most important ressources on the Web, at least within the LSWgTLs extensions.

In addition to the general measures enforced for all SWgTLDs, the following specific measures shall be implemented and enforced within LSWgTLs :

• Metadata should indicate the topic fields ( eg Sciences, Arts, Medieval Litterature - Europe, Science Fiction - Russia, etc.. ) with standardized XML or RDF schemas that automatic translation tool could access, read and make use.
• Metadata and and ODRI shall fill the need to identify that a specific document is an original document or is the translation of an original document and in which language.
• A single set of free Automatic translation tools shall be made available by the LSWgTLs authority to all registrants that would take advantages of the existence of several human certified translation of the same document.
• Automatic translation tools output shall be displayed on a general WIKI site, managed by the LSWgTLs authority, where users could contribute to correct and improve outputs. These human-corrected translations shall be then validated as human-certified translation, following a process determined by the LSWgTLs authority.

Interaction with other SWgTLds could be studied and implementated in order to allow interoperability between all SWgTLDs extensions and facilitate mutual cross-fertilization and empowerment. For example ethicalcompany.equi.fran, ethicalcompany.equi.bamb or ethicalcompany.fran.equi, ethicalcompany.bamb.equi could point directly to versions of the site of the "ethicalcompany" in various languages ( eg French, Bambara ), those combinations of SWgTLDs extensions should have to satisfied both set of conditions concerning the .equi SWgTLDs and the LSWgTLDs. This demonstrates the need of a central SWgTLD authority to manage and coordinate the governances of the various SWgTLDs.

• ITU and UNESCO Global Symposium on Promoting the Multilingual Internet ( Geneva, 9-11 May 2006 )
• Consultation on WSIS Action Line C8 : Cultural Diversity and Identity, Linguistic Diversity and Local Content at Palais des Nations, Room XX, 10h00 - 18h00. ( 12 May 2006, Geneva, Switzerland )
  
• Linguistic Semantic Web gTLDs Mailing list : Linguistic-Domains googlegroups.com
  
• First meeting of the IGF : Presentation of the LSWgTLD proposal at the Workshop Linguistic diversity 100 ( Wed. 1 November 2006, Athens, Greece )