- From: Al Gilman <asgilman@iamdigex.net>
- Date: Sun, 08 Dec 2002 10:33:06 -0500
- To: Jonathan Chetwynd <j.chetwynd@btinternet.com>, Tim Roberts <tim@wiseguysonly.com>
- Cc: WAI list <w3c-wai-ig@w3.org>
At 08:38 AM 2002-12-08, Jonathan Chetwynd wrote: >well there certainly is a case for screen readers (and browsers) to >optionally highlight the word being read. Yes this is a well-known assistive mode. >not sure any do though Here we come to the fact that aftermarket assistive technology is special stuff for special needs and is need-specific; not ususally universal. What you are getting at is requirements for a screen reader targeted to SLD users. On the 'highlighting the region' issue: yes there is a use for this. In fact, you want it to be able to ascend the tree (retrace the breadcrumb trail) of part-whole relationships. Highlight the smallest recognized element you clicked in or have moused over. Query for its parent and the enclosing element scope is indicated, etc. This would be very useful in auditing the svg:g tree in a scene. But also for teaching topic sentence, topic paragraph, section title, etc. To do that sort of thing, a) the rhetorical notions you want to teach have to be machinably mapped to the syntax in use (XAG Guidelines 2, 4. http://www.w3.org/TR/xag) b) some sort of [preferably distinguished] screen highlight must be available for programmatic control from the tutorware add-on (alias assistive technology). c) it still takes a modular extension to interpret the syntax tree (the DOM tree) in the light of the rhetorical graph-grammar and follow XPaths through that structure to identify the elements that fill rhetorical relationship roles of interest in the current lesson. The long-awaited 'views' module in the DOM would address this. Note that you are not here talking about what the UAAG covers, which deals with the selection and focus comprehended by the application. You may need an added layer of inspection selection just as the screen readers all have an added layer of inspection cursor in addition to the cursor that the OS and Ap know about. Then the tutorware AT would find the related element by XPath following and indicate it to the user by DOM directed state manipulations reflected in highlight effects on the screen. I think that there may be some mass market utility for context highlighting on user event, if linked to (triggered by) the context menu query. Particularly for an extended context query which gives a breadcrumb trail (part/whole ancestry trace) by way of orientation before it lists action opportunities. It would make all kinds of sense to indicate in the graphic presentation the scope that corresponds to the least scope mentioned in the breadcrumb trail, which give the bounds of 'here' in the "you are here" description. I know plenty of people who could stand to have this sort of a query to assist them in decoding nested windows. The application to SLD, say in teaching the words 'brother,' 'cousin' and the like with the aid of a pictorial family tree (of the learner's own family) is obvious. For the non-SLD, teaching the rules for what courts hear what cases in what stages of the appeals process would use a similar relationally-annotated graph. A research topic with regard to this sort of structural navigation is to investigate whether it saves time for those with high input symbol cost (single switch, voice command, etc.) over the existing mouseGrid search. What if they had the option of saying 'up' and seeing the enclosing structure highlighted, and 'next' when they had reached the level at which they intended to cruise? Coupled with the annotated context menu proposal this could be a general function of the voice command UI. 'Menu' gets you the context menu. Configuration says whether the breadcrumb trail is displayed at this time or only after saying 'trail'. But then one can speak items in the active menu. The trick is to get the techies pushing this technology to partner with your team as an evaluation site. Is there anyone here who could read over the recently published Multimodal Interaction Use Cases http://www.w3.org/TR/2002/NOTE-mmi-use-cases-20021204/ and see if their collection of cases covers these needs or not? Al Recurring principles exposed here (40X points): In the try-harder escalation sequence[1] to make an interface usable under adverse conditions, the first set of functions to add to a browser are those common in an editor; if there is nothing there that does the trick, the next place to look is the functions common in a debugger. [1] For the "try-harder escalation sequence" see for example http://trace.wisc.edu/docs/modality_translation_on_demand/modality_translation.htm >Jonathan
Received on Sunday, 8 December 2002 10:28:57 UTC