WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: external links & new windows

for

From: Terrill Bennett
Date: Nov 8, 2010 5:06AM

Nathalie,

The following is a *normal* function of the *browser*, with or
without a screen reader. It works with Firefox, IE, Google Chrome,
etc. After TABBING to a link:

* Pressing Ctrl+Enter opens a new tab in the background.
* Pressing Ctrl+Shit+Enter opens a new tab in the foreground.

Any user can the control tabs in a number of ways...

* Ctrl+Tab to move to the next tab
* Ctrl+Shit+Tab to move to the previous tab
* Ctrl+F4 to close the current tab, etc.
* Ctrl+N where N is between 1 and 5 (?) takes you to that tab number
in the window
* Ctrl+9 takes you to the last tab in the window

Again, no screen reader involved in any of the above. Just your
browser. But if you're using a reader, they work, too.

Hidden text can be added to links, which will only be seen by users
with styles turned off, users who override the styles on a site with
their own, and screen readers. Some will say Google will dock you
points for hidden text, but Google has stated otherwise - Google
isn't stupid, and they know the difference between padding your site
with text to raise your score, and text that's just text (even Google
has hidden text on it's own news site).

I have an example, here:
http://bennett1.org/demoCode/linkWithHiddenTxt.html

And a recording and transcript of NVDA reading the entire page, and
after I press Tab, NVDA reads just the link:
http://bennett1.org/demoCode/linkWithHiddenTxt.mp3
http://bennett1.org/demoCode/linkWithHiddenTxt.txt

Transcript, for your convenience:

Link with Hidden Text
visited link The Google search engine needs your hits!
More stuff goes here.
tab
The Google search engine needs your hits! visited link


-- terrill --


At 05:20 AM 11/8/2010, you wrote:
>Hello list,
>
>I've been pondering this topic for a while now and have gotten nowhere
>definite, which is why I'd like to ask for your opinions on the matter:
>
>In the past, I had established a homogenous practice for links that went
>thus:
>- links to pages on the same website open in the same window/ tab
>- links to pages on external websites open in new window/tab, the title
>attribute containing the forewarning "opens in new window" along with
>supplemental infos on the content awaiting the user behind the link,
>when appropriate.
>
>The "sighted" rationale behind this was, that users often spend some
>time on linked external sites, surfing around there, but then may want
>to return to the "main branch" of their reading.
>For sighted users, this is achieved simply by closing the separate new
>window - as compared to the nuisance of hitting back X times or
>searching for the original page in the (possibly long!) back-dropdown
>list. This was also appreciated as useful by the user base.
>My assumption had been that similar would be true for blind users, and
>all was well --
>*until I was confronted with the fact that screen readers do not read
>title-attributes consistently/reliably.*
>Ooops!
>
>To contain the "new window" warning in visible text may be OK in a list
>of links, but it gets rather cumbersome with links in a text. And I
>haven't seen any sites doing that in ways that would be perceivable to
>screen reader users (css stylings for sighted people, yes: little arrows
>or web-icons added in, different colors, etc.).
>
>The alternative, not to link externally at all, does not feel right
>either however - due to the aforementioned rationale concerning ease of
>use.
>Leaving the choice to the user would be viable IF people were widely
>aware of the fact that e.g. mouse-users can choose to open links in new
>tabs or not (is there a similar mechanism for screen reader users?), but
>are they?
>
>So what to do?
>What practices do you adopt in dealing with external links? Do you use
>new windows/tabs at all?
>And if you do open in new windows/tabs, how do you go about forewarning
>the user perceivably but at the same time unobrtrusively?
>
>Thanks so much for any light you can cast upon this :)
>Nathalie
>
>