Two wrongs make a Nielsen

Set aside for a moment the principle that all media should be accessible. (Indeed, they must be accessible for people with disabilities by law in many parts of the world even if those laws are obscure or rarely enforced. Certain language groups also have legal protections.) Concentrate instead on the vaguely tokenistic argument that media about accessibility have to be accessible themselves – even to groups not officially “targeted.”

Cultural exchange

In traditional Web sites, we can draw the example of Typetalk and Australian Communications Exchange (ACE), the U.K. and Australian relay services.

• Typetalk, “a link to the hearing world for people with communication difficulties [sic],” offers a splash screen and, later, a three-frame page with about 17 images lacking text equivalents. Typetalk also would love for you to run Flash, which, after extensive neglect, is only now accessible to certain screen-reader users. (There is no indication that Typetalk’s Flash content actually uses the current accessible features.)
• ACE promises and delivers a greater accessibility level (claiming valid HTML 4 and CSS and the increasingly dodgy Bobby certification), but does not claim even Level A conformance with the Web Content Accessibility Guidelines and insists on JavaScript for simple functions like navigation menus.

The sites work just fine for the “targeted” audience of deaf people, but really, is that enough?

• There seems to be a hierarchy of value at work: If a site concerns deaf people, then the thinking is “We’ve already handled the accessibility problem, thank you very much.”
• Multiple disabilities are common. (We cannot find actual numbers of online computer users with multiple disabilities from the only two credible sources, UCSF [PDF] and NTIA [PDF].) In fact, one group that uses relay services extensively is the deaf-blind population (few in number at 11,000 to 40,000 in the U.S. alone, but relay calls are usually their only way to use a phone).
• Knowledge of other disabilities can be as poor among people with a certain disability as knowledge of disabilities at all would be among nondisabled people. Isn’t a Web site a reasonable way for blind people to learn about relay services? (And don’t people whose only disability is visual impairment also place and receive relay calls?)

All right, then. Managers of these relay-service Web sites really should know better, but in any event they don’t have a lot of cash and their sins are merely venial. So we’ll stop picking on them.

So what about Jakob Nielsen?

Blowing it big-time

For years, NUblog has derided usability blowhard Jakob Nielsen. (“In the future, first of all, Web sites will be designed by my guidelines... for the simple reason that if they don’t, they are dead.”) For this and other reasons, Jakob Nielsen writes that we have “achieved permanently-blacklisted status... and any E-mail we get from [them] gets deleted unread.” (Juicy!)

But the mighty Nielsen Norman Group enjoys the privileged position of having dumptrucks of cash in the bank with which to carry out usability studies. To its unquestioned credit, the report entitled “Beyond alt Text: Making the Web Easy to Use for Users With Disabilities” surveyed actual “users” with visual or motor impairments in the U.S. and Japan. There’s a companion methodology report available on running your own usability studies with disabled subjects, also to NNGroup’s unquestioned credit.

The reports cost US$190 and $82, respectively. No problem there: You can charge for information.

So what is the problem?

Initially, HRH Nielsen published the report in PDF format only. Finally! An E-book somebody wants to read!

But what is the stated Nielsen policy on PDF publication?

Ensure that your PDF document format is at least one version behind the latest offering.... PDF Version 5 was released recently, but I recommend sticking to Version 3 until 2002 (at which time you can use Version 4; Version 5 should not be used until 2004). I got many complaints when I made... usability guidelines for E-commerce available for download in PDF 4 format recently: even among Internet professionals, there are still many people who have Acrobat Version 3 installed on their machine. I am now wiser and will be keep using PDF 3 for any printable reports I publish in 2001.

But Acrobat and PDF are two separate things. (Did you know that Acrobat is not the only program that can create a PDF?) The only Acrobat version guaranteed to work passably with screen readers is Acrobat 5. (The older Access plug-in was not a bad first start by any means, but it required Acrobat 4.05, subsequent to the “3.0” that Nielsen errantly demands.)

There are still many provisos:

• If you want your document to be read reliably and completely by a screen reader, you need to use the logical structure introduced in PDF 1.3 or the tagged-PDF features introduced in PDF 1.4. The full Acrobat 5 program can create a PDF 1.4 file. What Nielsen describes as “PDF 3” amounts to PDF 1.2 – not good enough for screen-reader accessibility. Nielsen’s advice guarantees inaccessible PDF usage.
• Screen-reader support works only on Windows. (That’s Apple’s fault for failing to produce an accessibility infrastructure similar to Microsoft Active Accessibility. On Linux and related platforms, Gnome is giving it the old college try, but it’s not equivalent to MSAA yet by a longshot.)
• Only Window-Eyes and Jaws are known to support Acrobat 5 access, the former more strongly than the latter.
• Acrobat 5 will muddle through even older documents passably. You can open an Acrobat 3 file in Acrobat 5 and give it a whirl in your screen reader. Single-column documents will generally read fine, but this report is described as “richly illustrated,” which implies multicolumn layout, which complicates so-called logical reading order.
• To make a fully-accessible PDF requires the use of tags (cognate with HTML markup), and only a few authoring tools (Word 2000, PageMaker 7, InDesign 2) can include tags automatically. Adding tags post facto (using utilities for Acrobat itself or Acrobat Capture) is so onerous a task as to be unreasonable for nearly anyone under nearly any conditions.

So: Jakob Nielsen has released an accessibility report in an inaccessible format.

The correct course of action would have involved authoring an Acrobat 5 version (with proper tags) and, if necessary to maintain policy consistency, a separate Acrobat 3 version. Nielsen Norman Group has quite enough money to manage that task.

But we’re not done yet!

NNGroup trumped itself by releasing an “audiobook” version on “standard” CD and cassette. The mighty Group betrays massive ignorance of talking-book protocols.

• “Full text of the report read by Kara Pernice Coyne, one of the authors. Great while commuting or exercising, or if you are looking for a human voice alternative to listening to a screen reader.” Nielsen implies, with no factual basis, that screen readers can already easily voice the original report.
• It is hubris incarnate for anyone but trained actors or readers to deliver extended texts in voice.
  • We still cringe at sitting through Faith Popcorn’s whiny, nasal New York Jewish accent (sic – it’s a stereotype on each of those levels) during an audio reading of one of her own books. We know she can’t do much about her voice – much, that is, apart from inflicting it on anyone else in a sound recording.
  • True, a tiny few authors have quite a gift for reading literature. Usability authors are not obviously in that cohort.
  • Looking at another access technique, audio description of television and film, recognizable celebrity voices should not be used, nor should actors from the actual production being described (a true case – Disney’s Dinosaur has descriptions voiced by one of its actors).
• Genuine audiobooks intended for blind or “print-disabled” readers are recorded in specialized formats (as required by various copyright laws); “standard” CDs and tapes are not those formats. True audiobooks include metadata like indications of page breaks; they can be accelerated to superhuman but intelligible speeds. The claimed utility of the CD or tape “while commuting or exercising” reveals NNGroup’s insincerity and lack of commitment to accessibility.
• The separate methodology report is not available on CD or tape (even those incorrect CD and tape formats), further emphasizing NNGroup’s apathy toward accessibility.

Inaccessible PDFs and amateur, inaccessible CDs and tapes are no way to distribute a report on accessibility. NNGroup has the money to do it right – and it isn’t too late, either. Remaster the PDF in Acrobat 5, and send the report out for recording by an experienced reader using the correct format. Simple, really – and do the same for all future reports regardless of topic. (Tokenism, remember?) Any accessibility expert could have provided that advice for free – and nobody’d end up “dead,” either.

Is all this “screen reader” crapola confusing? Then watch the online video “Introduction to the Screen Reader.” It is slightly too technical and spends too much time on non-Internet uses, but it’s fun to watch. Oh, and ACCESSIBILITY ECUMENISM ALERT – that video, though concerning the blind, is actually captioned.

Posted on 2002-03-13

(NUblog: Joe Clark)