Bruce Lawson’s personal site

Standards and accessibility – sitting in a tree?

(Last Updated on )

The excellent Jim Thatcher – who developed one of the first screen readers – has redesigned his site, and in his accessibility statement he writes,

This site uses CSS for layout, not tables. I do not think that that is essential for accessibility at all. Recently I believe I have seen more questionable situations with reading order stemming from CSS based layouts than ever from table based layouts.

This chimes in with some questions I’ve been asking myself lately and, as it’s Philosophy Friday today, I’ll pose the question: how inextricably linked are standards and accessibility?

Orthodoxy has it that seperation of structure and presentation via css and valid (x)html de facto enhances accessibility. Almost any article you read on why css is the way to style sites claims enhanced accessibility as a by-product. And this has been a useful orthodoxy; designers have been encouraged to move to css partly through the righteous side-effect that designs will become more accessible. That’s given the cause of accessibility a tremendous boost, if only because people will have heard about it.

Conversely, accessibility types have taken a cue from the design world and started to make their pages less ugly and clunky. Gawds may not be beautiful, but it’s better than useit, for example.

One thing that is certain is that accessibility is enhanced by structure. For example, one of the simplest JAWS keyboard commands is the H key, which jumps to the next heading in a document (be it <h1>, <h2>, whatever). Insert-F6 calls up a dialogue box listing all headers, so a blind user can quickly “see” what’s on the page. So, if you look at a well marked-up page (I’m choosing zeldman.com, as he’s the man who showed me The Light, but could just as easily cite Matt Machell‘s ongoing redesigns for the University of Central England), you’ll see that a Jaws user could quickly and easily identify the story headers, jump between them until they found something they wanted to read.

A badly marked-up document, conversely, offers no such structural clues as it simply has no real structure, only visual indications of what’s a “heading”. But Zeldman or Matt could certainly lay out their sites using nested tables, and still maintain the structured headings, if they so chose.

So does Accessibility require xhtml/ css web “standards”? The wierd and not terribly wonderful WAI/ WCAG guidelines say, “Use markup and style sheets and do so properly .. Using markup improperly — not according to specification — hinders accessibility”. That might be orthodoxy, but is that true?

Personally, I try to use them properly (what does “properly” mean?). Joe Clark has established from screen reader manufacturers that screenreaders like valid code. You can still follow the diktat to “Use W3C technologies .. and use the latest versions”: by using xhtml 1.0 transitional you can lay everything out in tables, and as long as they linerarise OK, that’s accessible.

Certainly, a lot of organisations that agitate for accessibility use tables for layout – see The International Center for Disability Resources on the Internet, the Royal National Institute of the Blind and the UK’s Disability Rights Commission. They’re all good, campaigning organisations who were advancing the cause of web accessibility while I was still in short trousers. But are they wrong?

This isn’t a rhetorical question. Is their accessibility hindered by their not using “web standards”? Presumably not, or they’d have big trouble with their core audience.

So if presentational layouts are not an impediment to accessibility, why are they prohibited in the WAI/ WCAG guidelines?

My preference is for valid xhtml/ css for layout as a starting point for accessibility. Many other people say the same (as a follower rather than a leader, that’s where I’ve got my information from). But is that just my preference or do standards and accessibility really sit in a tree, k.i.double-s.i.n.g.?

5 Responses to “ Standards and accessibility – sitting in a tree? ”

Comment by Isofarro

I wouldn’t go as far to say that ICDRI, RNIB and the DRC are wrong when it comes to using tables for layout. From a practical sense, a table-layout based website can achieve an RNIB See-it-right accreditation – its accessible to the RNIB’s core audience.

The best accessibility level a website can achieve is WAI A+, which essentially means the barriers that make it impossible to access a website, although some barriers that make it difficult for a visitor to access can remain.

The main problem area I see for tables layout is where tabular data is being presented. How easy is it for an assistive technology to recognise where a table is used for layout, or for tabular data? Its quite straightforward if the tabular data is nothing more than a number or short word, but where the data contains multiple paragraphs, its going to be tricky to spot the difference without accessible tabular data markup.

Comment by Ian Lloyd

There has definitely been a shift away from using evil, EVIL tables in the last couple of years. If you use tables, you only have to turn your back for one second and before you know it they’re trashing your house and trying to jump on your teenage daughter. Or something like that.

Seriously though, I agree that using standards does (or can) promote better accessibility, but using standards does not guarantee an accessible site. I could build a site that conforms to all the web standards (and by these I mean the W3C HTML and CSS specs) yet fails dismally on that other set of standards that the W3C are responsible for, namely WAI. Hell, you’ve proven this yourself, Bruce – your CSS Zen Garden entry may have validated but it nearly made my eyes bleed and probably exploded a few other people’s heads or caused a few people to start fitting right there in front of their computer screen. But hey, it validated, right?

And so on to tables – the strive for CSS layouts precludes use of tables for layout, assuming you want to make site-wide changes at a dash from your CSS file (perhaps a style switcher of some kind is used for instant changes?). A table layout hinders this, but a table layout does not completely prohibit this. Also, a table layout is, by its nature, far less likely to work on a handheld device that does not support CSS. I know that I would always opt for a table-free layout. But, a table-based layout can be used in a site and for it still to pass automatic WAI conformance checks and also manual checks. Like I said though, it limits flexibility for user change, even if a linearized table’s information makes sense to a screen reader (it’s not just blind people we need to consider!)

I’m with Isofarro that the bigger issue with tables is data tables – the markup and effort required to produce an accessible table, particular complex ones that have merged cells (rowsapan and colspan) is a real pain with most tools. Very few do a good job and then, once you’ve put in all that effort, none of the screen readers make a good job of interpreting it. All very frustrating. At least with layout tables they can be understood!

Comment by bruce

I completely agree that removal of presentational tables enhances upgradability, browsing with a handheld etc – but does it *necessarily* enhance accessibility for people with assistive devices? I’ve been running some tests to see if Web Standards mean JAWS users get a better experience and, from preliminary finidings, I’m delighted to say that, yes, semantic markup does make screenreaders easier to comprehend. More later on that.

I’d still love to know, if anyone from ICDRI, RNIB or the DRC are reading, why they chose the tables-for-layout route, as I’m assuming they feel that it enhances accessibility.

Comment by Bob Regan

I am actually not sure that tables are evil. Jim is right in that you can do CSS badly just the same as tables badly. This issue not just the technology but more importantly how it is used. As layouts in CSS get more complex, we will start to miss the bad old days of tables. The advantage today is that divs just kind of work out better. This may be changing as the Zen Garden illustrates new techniques for complex layouts.

I guess the bottom line is that we need guys like you to keep writing for designers about what the issues are and how to deal with them. It is designers’ job to keep ignoring us and figuring out new ways to f*** things up. In the end, standards don’t make accessibility automatic. They just structure the conversation so we are all speaking the same language.

Comment by Matt

Hey, cheers for the shout-out!

As to tables, like any tool they can be used for good or ill. If the content still linearises well, and they aren’t horribly nested, then you’re doing a better job than somebody using divs but in a messed up order.

Leave a Reply

HTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong> . To display code, manually escape it.