- From: Misha Wolf <misha.wolf@reuters.com>
- Date: Sat, 22 Feb 1997 19:37:43 +0000 (GMT)
- To: Tomas Carrasco Benitez <carrasco@innet.lu>, www-international <www-international@w3.org>, meta2 <meta2@mrrl.lut.ac.uk>, Unicode <unicode@unicode.org>
Tomas, You raised the question of whether to do language labelling via <HTML LANG=xx> or via the <MEAT HTTP-EQUIV="Content-Language" ...> mechanism. In the Language pages for the Tenth International Unicode Conference (go to <http://www.reuters.com/unicode/iuc10> and click on "Language"), we put the language labels on HTML containers. Each page has <HTML LANG=xx>. The twenty nine pages in individual languages each have additional language tags on the containers holding text which is in a language other than the main language of the page. For example, the Simplified Chinese page starts with <HTML LANG=zh-cn> and also has "LANG=en" where appropriate, to mark portions of English text or images containing English text. The Unicode pages, which contain text in all twenty nine languages, follow the same model. These pages start with <HTML LANG=en>, as there is some English introductory text, and then use <TR LANG=xx> to create table rows with text in other languages. The <META HTTP-EQUIV ...> mechanism is too coarse, as it applies to the entire document. If an EC page contained text in all the Community languages, you would surely want the indexing engines to associate each piece of text with the appropriate language, rather than to treat the whole page as being in a single language. Misha
Received on Saturday, 22 February 1997 14:37:56 UTC