Fifth International World Wide Web Conference

May 6-10, Paris, France

Internationalization Workshop, 6 May 1996


This one-day workshop started with seven presentations selected from the submitted position papers. These were:

François Yergeau
Status of I18N and Internationalization of URLs
John Larmouth
The Importance of Architectures, Models and Interfaces
Martin Bryan
Maintaining Links between Translations
Faith Zack
Server and CGI Extensions
Tomas Carrasco Benitez
Web Internationalization and Multilinguism
Jean Dougnac
Multilingual Software Architecture
Gregor Erbach
Multi-lingual HTML/Multi-lingual Servers

After lunch we focussed on 4 areas for discussion. These were:


Recognizing that there is already a font working group we focussed on specific aspects of fonts with respect to internationalization, such as


We noted here a conflict between the need to represent a wide range of characters in URLs and the World-wide aspect of the web, and the difficulty of typing in a URL.

We recognised the advantages of the URN proposals, and regretted the apparent lack of progress in this area.

Some discussion arose over the use of UTF7 and UTF8 encodings, and there was a strong feeling that Latin 1 should not be used.

In any case it is the Query part of a URL that is most important at this time. It is regretted that current browsers do not support making bookmarks using the POST method.

We also note a confusion over whether URLs are a sequence of octets or characters. We recommend that they be defined as a sequence of characters, to clear this up.

Language Negotiation

We regretted that despite the fact that good proposals exist for language negotiation, virtually no browsers or servers currently implement them. We feel it a priority that this should change as soon as possible.

We propose that the quality factor be interpreted as a indication of the quality of a translation (for instance machine translation would probably be low quality) rather than the current definition which interprets quality as how well the user knows the language.

We note a need for a language preference list rather than just a single choice of language, with, for instance machine-translated English being less preferred than human-translated French.

We also discussed mechanisms for discovering the various language versions of a document that are available.


It was unresolved whether there should be standards for authoring.

We discussed aspects of document delivery, multiple languages in one document, client-side extensions needed to support language choices in documents, and thesauruses to aid in multi-lingual searches.


We identified several immediate possible actions:


For more information

See the W3C i18n page.

Contact Bert Bos.