- From: Joseph Scheuhammer <clown@alum.mit.edu>
- Date: Mon, 24 Feb 2014 10:08:09 -0500
- To: Sailesh Panchang <sailesh.panchang@deque.com>
- CC: James Teh <jamie@nvaccess.org>, Bryan Garaventa <bryan.garaventa@whatsock.com>, James Craig <jcraig@apple.com>, mick@nvaccess.org, "Schnabel, Stefan" <stefan.schnabel@sap.com>, WAI XTech <wai-xtech@w3.org>
On 2014-02-14 6:11 PM, Sailesh Panchang wrote: > My interpretation of what I expect as a user for the 'I am a > listitem' SPAN tag is based on the above distinction of aria-label on > a container and on a UI control. > >>> >>Here's a use case: a screen-reader/magnifier user. The magnifier presents an enhanced version of what the browser has rendered on screen, namely "What is my nom de >>plume?". If, at the same time the screen reader speaks something different, say, "What is my name?", that is going to confuse the user. > Does this not apply to your example too! > i.e. > <span role="listitem">What is my <a href="#" aria-label="name">nom de > plume</a>?</span> Not at all. You previously asked if adding an aria-label would change the content. I replied that the content in the a11y tree is the same as that rendered on screen. The AT *can* present the same content as rendered by the browser, but that is up to the AT. My point was that if the content of the accessible was different that that rendered on screen then the AT could not avoid presenting something different. In the listitem example, the author provided a label for the link that is different from the content. Nonetheless, both the label and the content are available from the accessibility API, as well as other information, such as the role, possibly a description, and a variety of states and properties. Part of my point was that each of these pieces of information are explicitly individuated in the accessible object. It is up to the AT to determine how to orchestrate the presentation of these pieces. I think what you are getting at is that in the case of a list item or a link, having the name different from the content is an issue. I agree. However, my example was exaggerated to clarify what is going on with respect to the ARIA name calculation. There is no practical reason for an author to supply a name of "name" when the content of the link, "nom de plume", is sufficient. Still, there have been other examples on this thread that point to where aria-label might be used to be more informative. Bryan gave this one, involving a link to a footnote: <a href="#" aria-label="footnote1"><sup>1</sup></a> The visually rendered content is a small, superscript numeral one. Given that visual convention, a sighted use will perceive this as "footnote 1". But, a superscript is not always indicative of a footnote. Sometimes a numerical superscript is an exponent. The <sup> element does not reliably mean "footnote", and a screenreader can't tell what is intended from the <sup> markup alone. The screen reader might be able to make that inference from the context that the superscript appears within; here, it is a link, and no other mathematical symbols are involved. That heuristic may work. However, the author can do something to avoid the heuristic entirely by adding an aria-label to give a better idea of what this use of <sup> is for. Furthermore, the magnifier/screen-reader user gets more or the less the same information from sight and sound in this example. They see the visual indicator for footnote 1 (a superscript '1'), and they hear "footnote 1" (depending on how the screen reader presents the accessible). The visual and audio renderings are consistent and reinforce each other. Conclusions: 1. In some cases, the point of allowing the author to provide a label is to give a more meaningful name than what would be provided by the content alone. If the content is sufficient, authors need not provide any additional name. 2. The accessible objects in the a11y API list various aspects of the object as separate pieces of information -- the role is X, the name is Y, the content is Z, the checked state is W, the description is V, and so on. An AT can leverage this individuation and provide a meaningful, useful presentation. A screen reader could, for example, say: "List item with name 'What is my name?', and content 'What is my link: (nom de plume )?'", although I can imagine that, for power users, that level of detail could get tedious. I would hope, though, that screen readers allow the user to query an object specifically for its name, contents, etc. -- ;;;;joseph. 'A: After all, it isn't rocket science.' 'K: Right. It's merely computer science.' - J. D. Klaun -
Received on Monday, 24 February 2014 15:08:44 UTC