- From: Kelly Ford <kelly@kellford.com>
- Date: Wed, 29 Aug 2001 09:59:37 -0400 (EDT)
- To: <wai-xtech@w3.org>
Hello Gregory, I want to digest all you wrote for a bit. But if my comments made it sound like coding should be done to one particular AT product, that wasn't my intent. Rather I was attempting to illustrate how a couple of different products interpreted this example. Again I want to fully digest your comments. But I have stated publicly in the past that I believe JAWS should handle certain aspects of how it displays web pages differently. Kelly On Wed, 29 Aug 2001, gregory j. rosmaita wrote: > aloha, kelly! > > while your observations are on-point, i think that your conclusions miss the > point... what you reported -- the difference between what was intended and > what was experienced -- is the type of feedback that AT developers need to > hear: that their support for the markup is insufficient, and that their > extant implementations need tweaking. > > yes, there are sometimes problems in what users of JFW 3.x hear on a > "properly" marked-up page, but faulty implementation on the part of specific > ATs shouldn't be the force that drives our discussions... and, while the > onus lies squarely on the shoulders of any AT developer and mainstream UA > developers, it's up to us, as individuals and members of larger (presumably > more influential) entities, to get them to implement to spec, so as to break > the vicious circle in which AT developers blame mainstream developers, > mainstream developers ask "where's the demand?", and authors respond to our > efforts with a quote yes, that's all fine and good, and your efforts are > noble, but when something implements this particular feature or supports > such-and-such attribute, _then_ we'll code to it - otherwise, why bother? > what's the point - political correctness? unquote > > if, to get technology specific for a moment, JFW can't get the requisite > information from the proprietary DOM implemented by microsoft in MSIE, then > microsoft should be called to task... and, it should be pointed out to all > involved (to get provincial for a moment) that in order to comply with > section 508, someone bloody well better start supporting the "accessibility > features of the markup languages" which federally-purchased user agents and > authoring tools support. hence, the path to be traveled is this: the > company formerly known as henter-joyce should be asked why they do not > support a toggle which would allow the user to switch between "speak > hyperlink text", "speak alt text" and "speak title text" - something which > i've already done on a number of occasions - if > the-company-formerly-known-as-henter-joyce claims they can't get the > information from the IE-DOM, then microsoft must be approached and asked if > the requisite information is available from the IE-DOM - and, if so, how it > can be addressed/called, and if not, why not? > > we can't just sit back and wait for this to happen -- we, the users of the > technology need to bring pressure to bear on the responsible parties, which > is why adherence to standards, particularly by the W3C, is crucial -- if W3C > doesn't implement the accessibility features of the markup language who > will? and, when said accessibility features are implemented, they need to > be, and i can't stress this strongly enough, implemented AS INTENDED BY THE > MARKUP LANGUAGE AND NOT DICTATED BY THE LIMITATIONS OF CURRENT TECHNOLOGY, > ESPECIALLY WHEN THE DEVELOPERS OF SAID TECHNOLOGY HAVE BEEN TOLD OF THE > SHORT-COMINGS OF THEIR IMPLEMENTATIONS > > sorry for shouting, but i'm sick and tired of exchanging advice on how to > tailor one's markup to cater to or make up for the short-comings of a > particular AT, no matter how much of a stranglehold that particular AT may > have on a "niche" market - THAT is what i'd call proprietary markup... > what's the difference between coding to JFW's aural output, and tailoring > one's coding to browser X? that's the path that leads down the road to the > evolutionary cul-de-sac where the OBJECT element dwells... > > the "title" attribute isn't "broken" simply because JFW doesn't offer users > the ability to switch between "speak hyperlink text", "speak title", and > "speak alt (if graphic)"... moreover, the user's ability to toggle between > these three states must not be limited to a configuration option -- it must > be available both on-the-fly and as a "display" option in a list-of-links, > as there are many scenarios in which a user might want to query a link to > discern all three bits of information (the actual hyperlink text, text > contained in "title", and the "alt" text associated with a graphic which is > used as a link) -- one such scenario is thumbnailed further on in this > missive. > > why should "title" appear before "alt" in the cascade? simply because the > purpose of "title" is to convey information about the element to a user, > while the SOLE purpose of "alt" is to provide a brief description/textual > equivalent of the image with which it is associated -- in the case of a > magnifying glass used to iconically symbolize a link to a search interface, > the correct alt text would be "a magnifying glass" and NOT "Search" - that > is the appropriate "title" text for the link, as it is the link (to which > the "title" attribute should be appended) that provides the link to the > search functionality, not the image of the magnifying glass -- that is > merely quote graphically defined hyperlink text unquote - no matter what it > links to, it is an image of a magnifying glass, so it should be alt-texted > "magnifying glass" or (less informative, but equally accurate, provided the > same exact icon is used whenever a link within the site that points to the > search facility is defined as a graphical hyperlink) "search icon". the > "thing" that "does the linking" is the link (the A element in combination > with the "href" attribute), so when a graphic is used as hyperlink text, the > "title" defined for the link should contain orientational information about > the link, which in this case would be "Search" > > take the following "real life" example, drawn from the navigational sidebar > of: > <http://www.whitehouse.gov/news/freedominitiative/freedominitiative.html> > > <A HREF="/"><IMG SRC="/images/whlogo-big-stripe.gif" > WIDTH="124" HEIGHT="85" > ALT="White House Logo" BORDER="0"></A> > > now, there ain't nothing wrong with that "alt" text -- i showed someone the > graphic, and, indeed, it contains the white house logo... fine and good, > but does the link lead to the white house logo? does it lead to a > description of the white house logo? no, it leads to the white house's > home/front page, which is why the link should have a "title" defined for it, > such as: "Welcome to the White House" (the TITLE of the referenced page); > "The White House" (common sensical); or, "Official White House Web Site" (a > bit stuffy, but authoritative) > > as regards "regular plain old text hyperlinks", the "title" attribute allows > the author a means of describing the function/purpose of the link, usually > with more detail than that contained in the hyperlink text defined for the > link, since that allows authors more flexibility when attempting to > gracefully and naturally integrate hyperlink text into the surrounding > prose, whilst still providing text that can, in isolation, sufficiently > communicate the function/purpose/destination/special features of the link. > note, too, that "in isolation" doesn't just mean "in a list of links" -- > when one considers a single word which serves as a hyperlink and which is > embedded in a paragraph of text, that for many, constitutes "viewing the > hyperlink text in isolation" -- an isolation which can be broken (as > jonathan, i'm sure would attest) via a windows-type tooltip which reveals > the contents of the "title" defined for the hyperlink. > > so, until it is possible, via either the UA's or the AT's interface, to > query an element and obtain from it any (or all) of the conditional content > defined for it by an author, such functionality remains both a blessing and > a curse... a blessing in that the ability to describe or annotate a link > using the "title" attribute allows the author to more gracefully and easily > incorporate hyperlink text into the surrounding prose, while providing a > means to convey orientational/contextual information when the hyperlink is > included in a list of links or is read/spoken/felt in isolation... this is > obviously a disadvantage to anyone who's access to a list of links is > limited to a single mode of presenting link information, as it can lead to a > serious disconnect between what is heard/felt when the page is aurally or > tactilely rendered, and what information is available to a user via a list > of links... this is the design flaw in JFW's list of links, as there is no > way to toggle between listing links by hyperlink (screen) text and "title" > or "alt", nor of choosing, in the case of a graphic which is being used as a > hyperlink, between the "title" defined for the encasing link and the "alt" > text defined for the graphic, but again, i emphasize, it is a shortcoming in > a specific implementation and not with the spec... > > why is such flexibility necessary? to meet different needs and conditions, > of course... when i am moving through a page autonomously, i need to know > the nature and function of the link; the former my AT can (and does) > identify; the latter, the author needs to provide in the form of a > "title" -- unless i trust my UA to "get" the contents of the TITLE defined > for the linked resource, and depending upon the type of resource to which > the hyperlink points, there may not be a practical [read: non-RDF] way to > natively associate a title with the resource... in order to find and/or > actually navigate to a graphically defined hyperlink, however, i may need > some knowledge of the contents of the graphic... why? > > it's not uncommon to be told to "find the button that says download Adobe > Acrobat" or "find the button that says 'Opera 5 Free Browser'" when what one > is actually searching for is a graphic used as a hyperlink that contains an > image of those words, not the structural element BUTTON, nor an INPUT > element whose type is "submit"... if (as with JFW or Opera) it is easiest > to find such structural elements by entering a special "forms mode", i may > never find what i've been told to look for -- in fact, the only guaranteed > result is mutual frustration... if, therefore, the "alt" text defined for > an image which is used as a hyperlink describes the function of the link, > and not the image, the problem of finding the link (not to mention the > user's frustration level) will be compounded, and that user's desire to > visit that site -- or to even utilize the resource to which the user was > being guided -- has either been substantially diminished or entirely > extinguished... > > if, on the other hand, the hyperlink contains a "title" such as "download > Adobe Acrobat Reader 5 NOW!" or "use this link to download Adobe's free > Acrobat Reader", and the "alt" text defined for the graphic contains the > same text contained (in a binary, pixelized form) in the graphic itself, > then it would not only be possible for a visitor operating in a > graphics-free environment to either independently discover the link which > allows one to obtain acrobat, or to be guided to it by both casual reference > (in conversation or in a commercial) and direct reference, such as that > provided by a friend, family member, assistant, or customer service / > technical support person... it doesn't help matters that the term "button" > has entered the vernacular as a generic term for a graphical > hyperlink -especially those which lead (directly or indirectly) to the > initialization of a download routing - nor that sites which disseminate > graphics for use as hyperlinks routinely refer to them as "buttons", either. > > the bottom line is, use "title" to describe the function of a graphical > hyperlink; use "alt" to describe the graphic itself. > > the purpose of a graphic as a graphic is separate from its function as a > link -- that it is used as a link does change the function of the graphic in > that it's not just a pretty picture any more, but it doesn't change the > content of the graphic itself; hence, the fact that a graphic is used as > part of the definition of a hyperlink shouldn't impact the "alt" string > defined for the graphic... it should, however, prompt an author to provide > a description of the function or destination of the link using the "title" > attribute for the A element which defines the link... > > "alt" text should describe the graphic and nothing but the graphic; the > "title" text defined for the hyperlink itself should describe the function > and/or the expected/intended (re)action... > > let's not bend extant markup to our will, but, rather, raise holy hell until > the responsible parties implement/support/make available all of the > "accessibility features" for the markup languages that they support, > natively or through a programmatic interface... > > gregory. > > --- original post --- > Date: Mon, 27 Aug 2001 03:23:22 -0700 > To: wai-xtech@w3.org > From: Kelly Ford <Kelly@kellford.com> > Subject: Re: archive: example message page with long titles > > Hello Al, > > At 09:47 PM 8/26/01 -0400, you wrote: > >What does the list of links view in Jaws show? > > > >This is controlled by the user setting I mentioned in my first > >message. The specific setting is called "Text Links Verbosity" and both > >the list of links and Virtual PC view of a web page in JAWS use this > >setting to control what is displayed. > > >Can anyone tell us what the list of links view in Window-Eyes uses? > > > >It never uses the title text. Window-Eyes always uses the general link > >name in all navigation modes. > > >If one does not use the list-of-links view, how hard is it to read all the > >text, including what comes in each list item before the link text, and > follow > >the link that you wish when you get to it? > > Here's the heart of the matter. At present I'm confining my comments to > JAWS and Window-Eyes, although I know that HPR behaves somewhat similarly. > > Both JAWS and Window-Eyes offer three basic navigation methods of a web > page. First is what I'll call the buffered or default view. In this > navigation mode the text of a web page is loaded into a buffer and the user > navigates around with the cursor keys or a few acceleration options, such > as focus to first input control, focus to first non-linked text and such. > > In this mode tables are decolumnized and linked text is displayed on a line > by itself. Assume I have the sentence: > > I want to win loads of cash from Powerball but must buy a ticket first. > > The hyperlink in the sentence is the word Powerball. I've also defined > title text on the link as: > > A Multi-state lottery game. > > This will show up to JAWS and Window-Eyes as follows. The user would be > pressing down arrow or perhaps in a continuous reading mode. > > ***Window-Eyes > > I want to win loads of cash from > *L Powerball > but must buy a ticket first. > > ***JAWS set to read title text, the default mode > I want to win loads of cash from > *L A Multi-state lottery game. > but must buy a ticket first. > > ***JAWS set to read screen text, requires user action to change > I want to win loads of cash from > *L Powerball > but must buy a ticket first. > > The second navigation option offered by both JAWS and Window-Eyes is the > list of links feature. This simply generates a list of all links found on > a web page and places them into a standard Windows list box. While the > user has several options from this point, what's important here is how the > user can find the context of the links from such a mode. Both JAWS and > Window-Eyes offer a command, no matter the name, which is the functional > equivalent of moving the user to the link of interest on the web page. The > user activates this feature from the list of links and is returned to the > buffered view of the web page on the line containing the link. The user > can then use any reading command to read what's before or after the linked > text. > > Finally the user has the option to tab through the web page while in this > buffered mode. In this navigation mode, the user is stopped at anything > reachable via tab, links, input controls and such. I should mention that > in this navigation mode, JAWS ignores the "text links verbosity" option and > will always read title text for a link if such has been defined. > > In your example page, I noticed that items like reply, next and previous > were present as both text and part of the link titles. In some navigation > modes for JAWS in particular, this will cause text duplication, i.e. if the > user is cursoring through the buffer with the default of reading title or > alt text. Here's an example: > > Next message: > *L Next: Kynn Bartlett: "RE: guideline 7.1 about screen flickering (fwd)" > Previous message: > *L Previous: Anne Pemberton: "RE: guideline 7.1 about screen flickering > (fwd)" > > Kelly >
Received on Wednesday, 29 August 2001 10:00:08 UTC