- From: Benjamin Hawkes-Lewis <bhawkeslewis@googlemail.com>
- Date: Fri, 13 Apr 2007 00:12:37 +0100
- To: public-html@w3.org
Thanks to Gregory J. Rosmaita for raising this issue and stressing the difference between quotations and mere direct speech and scare quotes; the fate of q is minor obsession of mine: http://www.benjaminhawkeslewis.com/www/accessibility/q-element/ I strongly agree that we need to maintain and improve a machine-discoverable way of demarcating quotations from external sources and following citations. This feature was pretty fundamental to the vision of "hypertext" originally articulated by Ted Nelson in 1965, so the desire expressed by some to remove its vestiges from HyperText Markup Language is somewhat depressing. Separating content from presentation facilitates restyling, but more importantly: 1) Identifying quotations is crucial for screen reader and voice browser accessibility. Such user agents must either present quotations in a different voice or announce them. Users often wish to minimize non-essential punctuation. If it is hard for browser developers to dictate language-sensitive quotation punctuation, how much harder must it be for screen reader developers to cope with the entire variety of punctuation in the absence of markup! 2) Excerpting sources is an key component of human discourse. The easier we make it to review original sources, the sooner errors and misrepresentations will be discovered and the more effective communication will be. Solving the minor technical problems associated with q and blockquote will therefore have tangible benefits to society. Pace Anne van Kesteren, forcing authors to insert quotation punctuation in raw text would /not/ free browser developers from the need to implement complex CSS for styling quotations since raw text /cannot/ be used to express the full variety of quotation punctuation actually in use. See Problem C analyzed in my earlier message to to www-style: http://lists.w3.org/Archives/Public/www-style/2006Sep/0141.htm Ian Hickson's suggestion that complex regular-expression-based CSS replace author-specified quotation punctuation sounds feasible, but to achieve the accessibility benefits discussed above the spec would have to mandate a particular set of punctuation, or assistive technology would once again be reduced to playing a guessing game. Placing complex human-readable bibliographical information into an attribute such as cite, as suggested by Gregory, or title, as suggested by Olivier Gendrin, would be a mistake, because within an attribute: 1) You cannot identify changes in language within the text in machine-discoverable manner, which is an important accessibility requirement (e.g. so screen readers can switch to the correct pronunciation for that language). 2) You cannot include links (e.g. to alternate editions). 3) You cannot express other machine-readable semantics (e.g. hCite). for/id attributes connecting quotation to citation elements would not suffer from the same issues, although I think a "from" attribute would be more logical than a "for". How would this system cope with repeated references to different "fragments" (e.g. deadtree pages, film times, fragment identifiers) from the same resource? Would there be a different <cite> element for every page cited from a book? Would the full citation information need to be restated for each mention? Or would there be a way to chain <cite> elements together in a machine discoverable way? An alternative which would seemingly avoid all these problems would be to adopt a data-rich, machine-readable citation URI format (perhaps patterned on OpenURL) which could be included in the quotation element's cite attribute and parsed, displayed, and followed by user agents in ways that suit users, rather than the transient, time-wasting, and error-prone citation formatting guidelines to which authors are enslaved. By "data-rich" I mean that such a URI format would contain within itself essential metadata (e.g. author, title, date, ISBN, page number for books) without requiring further lookup to discover these. Thus in the event of catastrophic failure, the relationship between texts, and even the original purport of lost texts, could still be reconstructed, much as with ancient scholia: http://en.wikipedia.org/wiki/Scholia I do not believe localization of quotation punctuation by user agents to be a key feature; so far that requirement has largely proved a regrettable roadblock to implementation. If it is specified, the spec itself should enumerate what punctuation should be used for what languages. But browsers should be obligated to expose quotations and provide easy access to their sources. One way to do the later is to provide link through to sources via the context menu. -- Benjamin Hawkes-Lewis
Received on Thursday, 12 April 2007 23:17:56 UTC