Participants at Athens summit demand more internationalization, saying English language enjoys an unfair Net monopoly.
A November 1983 specification proposed that domain names would have "only letters, digits and hyphen"--which meant that Cyrillic, Arabic, kanji or Chinese letters and characters could not be used in domains. Not even diacritical marks employed in German, French and Spanish were permitted.
On Wednesday, delegates to a United Nations summit here complained that the ASCII-only choice was representative of an Internet culture that is far too English-centric and that fails to respect other languages.
"This new society leaves people isolated, marginalized," said Adama Samassékou, a former Mali government official who is the president of the African Academy of Languages. "I think the digital divide is not as important as the linguistic divide. And that's the one we should be bridging in order to guarantee the democratic governance of the Internet."
Divina Frau-Meigs, who teaches at the University of Paris III (Universite de la Sorbonne Nouvelle), called for "setting up a culture whereby you can use your own language, and that will be considered part of your human rights."
After a decade of painstaking work and negotiations, however, Internet engineering groups have solved the problem of internationalized domain names. A widely accepted standard is in place, and Web browsers such as Mozilla's Firefox, Apple Computer's Safari, and Microsoft's Internet Explorer 7 (released last month) support non-English-language characters in domains.
"Regarding the technical implementation for the World Wide Web, we are done, except for maybe some corner cases," said Patrik Fältström, a senior engineer with Cisco Systems who is the co-author of an internationalized domain name specification inside the Internet Engineering Task Force.
Internationalized domain names work something like this: Nonstandard characters are translated into ASCII through algorithms called "Nameprep" and "Punycode," with a special "xn--" prefix attached that signals that it's an encoded domain name.
There are some security risks, such as a spoofing attack demonstrated last year. That sent unwary visitors to a site that looked like Paypal.com--except it was spelled with an "a" written in Cyrillic. (In response, subsequent versions of Firefox were configured to display only the Punycode version of the domain name by default.)
Some of the participants during Wednesday's discussion--not one was from North America--called internationalized domain names a singular achievement that would resolve many concerns about supporting other languages.
Andrzej Bartosiewicz, who runs Poland's .pl domain name registry, said "if we are talking about diversity, from my perspective, internationalized domain names are the key issue. (They allow) non-English-speaking people to create their addresses, names--especially domain names--in the future."
But other participants quickly lodged additional complaints: That U.S. software makers were not moving quickly enough to support other languages, and that English was still far too prevalent on the Web.
Hamid Shahriari, part of Iran's delegation to the U.N.'s Internet Governance Forum, blamed Microsoft for having software that does "not work hardly on my own languages."
Linguistic diversity should be "the key principle of Internet governance," said Elizabeth Longworth, a former New Zealand government official who's the executive director of the office of the director-general at Unesco.
"Without diversity on the Internet, you cannot have access, you cannot have participation," Longworth said.