W3C home > Mailing lists > Public > public-wai-evaltf@w3.org > April 2012

Re: Using AT for evaluation

From: Shadi Abou-Zahra <shadi@w3.org>
Date: Wed, 04 Apr 2012 20:05:18 +0200
Message-ID: <4F7C8D5E.60208@w3.org>
To: RichardWarren <richard.warren@userite.com>
CC: Elle <nethermind@gmail.com>, Detlev Fischer <detlev.fischer@testkreis.de>, public-wai-evaltf@w3.org
Hi Richard,

While WCAG 2.0 may not specifically say that content must be tested with 
assistive technology, it clearly requires content to work with assistive 

 From <http://www.w3.org/TR/WCAG20/#new-terms> [[
Using a technology in a way that is accessibility supported means that 
it works with assistive technologies (AT) and the accessibility features 
of operating systems, browsers, and other user agents. Technology 
features can only be relied upon to conform to WCAG 2.0 success criteria 
if they are used in a way that is "accessibility supported".]]

At the same time, "assistive technologies (AT) and the accessibility 
features of operating systems, browsers, and other user agents" have 
bugs and other peculiarities that a developer cannot always be held 
accountable to have to work around.

I think Elle's previous suggestion of using scenario-like approaches 
seems promising but the question remains: how to demonstrate that a 
website works with assistive technology?


On 4.4.2012 18:50, Elle wrote:
> All:
> I can definitely see Detlev's point, and I concur about "highly
> recommended" as an addition.  However, I still request that we not limit
> ourselves or our audience to simply a screen reader.
> Thanks,
> Elle
> On Wed, Apr 4, 2012 at 12:46 PM, RichardWarren
> <richard.warren@userite.com>wrote:
>> Aah,
>> Detlev has a point. Our methodology is to check for compliance with the
>> W3C guidelines, it is not, in and of itself, to test for universal
>> accessibility. The guidelines do not require, as an essential component,
>> that the site be tested with a screen reader. On that basis we do not have
>> the right to insist on using a screen reader (of whatever flavour). As
>> Detlev says it is possible to check for conformance without using a screen
>> reader, and not all Screen readers are compliant themselves.
>> HOWEVER I feel that making the use of a screen reader "optional" is not
>> sufficient. I sense the feeling of the group is that screen reader testing
>> is important - so I wonder if we can call it's use  "recommended", or even
>> "highly recommended" rather than "optional".
>> Richard
>> -----Original Message----- From: Detlev Fischer
>> Sent: Wednesday, April 04, 2012 4:40 PM
>> To: public-wai-evaltf@w3.org
>> Cc: Eval TF
>> Subject: Re: AW: Using AT for evaluation
>> Hi list,
>> I disagree with the proposed mandatory requirement to use a screen
>> reader as part of WCAG-EM, for several reasons:
>> 1. Inconclusive result from any one SR use
>> 2. Difficulty of separating conformant behaviour from SR repair
>> behaviour
>> 3. Much higher professional prerequisite for conducting evaluations
>> Let me quickly elaborate these three points.
>> 1) Inconclusive result from any one SR use. Even for well known
>> semantic constructs such as forms with legends or tables, SR
>> implementation varies so widely that a well constructed form or table
>> might be OK in one SR and can pose problems in another. What must
>> surely count is whether content has been implemented correctly, not if
>> some AT is able to use it as intended. For most cases, checks exist if
>> semantics have been implemented according to spec or technique that do
>> not require the use of a SR.
>> To take the example of correct pronunciation of words mentioned by
>> several: If a word is pronounced correctly will depend on the
>> inclusion of the word in the SR dictionary and also, the options set
>> by individual users. It will vary a lot across SR makes and versions.
>> (I have met more than one SR user who willingly turned off other
>> languages because he/he opted for speedy delivery over correct choice
>> of language).
>> Finally, leaving it open which SR to use makes results rather
>> inconclusive. I think we cannot and should not mandate, say, the use
>> of NVDA simply because it is free of charge. If anything, it should be
>> the most frequently used SR (often JAWS) which costs quite a bit (see
>> point 3).
>> 2) Difficulty of separating conformant behaviour from SR repair
>> behaviour. SR show a wide range of behaviour towards conformant and
>> badly formatted content, which makes it difficult to rely on any one
>> outcome in a conformance evaluation. Since the web is such a mess on
>> the whole, SR are good in coping with that mess, but to varying
>> degrees. Take a text rendered using Cufon. Jaws will read through it
>> OK, other SRs will stutter, voiceover just reads image - image -
>> image. So how does any one of these results help if we do not intend
>> to cover a range (which I believe is impossible due to time and budget
>> constraints)?
>> 3) Much higher professional prerequisites for conducting evaluations.
>> Requiring the use of a SR without making sure that it is used
>> proficiently will lead to dubious results. SR have shortcuts and modes
>> of operation that help expert users make sense of content. So if SR
>> use becomes a mandatory part of the evaluation there should be a
>> requirement that evaluators know their SR well enough to have
>> meaningful results. This puts the bar a lot higher for folks who look
>> for evaluation guidance but are not prepared for the time and effort
>> to learn how to use a SR adequately.
>> And adequate use will be difficult to define, especially if we do not
>> mandate a SR.
>> Having said all that, there are some points which are tricky to
>> evaluate without a SR - for example, checking that when inserting
>> content via DOM and setting the focus via a script, the SR focus will
>> indeed move to the content inserted. But this touches on the question
>> whether we should rely strongly on any observed actual AT
>> implementation behaviour at any given point in time. If AT does not
>> yet manage to deal adequately with correct code, the alternative would
>> be to check whether implementation is done according to spec and put
>> the burden on AT to catch up. Of course, checking for 'correct code'
>> is difficult without being s scripting expert, it also raises the bar.
>> This is a tricky one, and I am not really sure what's best here.
>> Regards,
>> Detlev
>> On 4 Apr 2012, at 15:57, Michael S Elledge wrote:
>>   Hi Everyone--
>>> My preference would be to require the use of a screen reader in
>>> conducting an evaluation, either with or without a person with
>>> disabilities. The option, in my mind, would be having it done by a  person
>>> with disabilities.
>>> With the availability of free screen readers like NVDA, testers  ought to
>>> be able to incorporate it in testing without incurring  unreasonable costs.
>>> I realize this falls short of the ideal, which  is evaluation by a variety
>>> of people with different disabilities,  but we've found it to be critical
>>> in our testing for discovering  issues (such as pronunciation or
>>> functionality) that would otherwise  be missed.
>>> I think another AT frequently used by persons with disabilities,  screen
>>> enlargers like ZoomText, may be unnecessary in the testing  methodology, so
>>> long as other methods are used to review an enlarged  screen. I bring that
>>> up for discussion, because others may not agree  with me.
>>> Best regards,
>>> Mike
>>> On 4/2/2012 8:17 PM, Vivienne CONWAY wrote:
>>>> HI all
>>>> I always use screen readers and am wary when clients say that they
>>>>   don't need testing with AT because they don't have any disabled  users.
>>>>   We never know the extent of employees' limitations - they  don't have to
>>>> disclose everything.  Also, it often happens that  someone suffers an
>>>> injury or illness while in employment only to  find that they can't use the
>>>> system now that worked for them  previously.  The old 1 in 5 thought
>>>> regarding disabilities applies  to those in employment as to the general
>>>> public in their use of a  website.
>>>> So... I would suggest that AT (at least the screen reader) be  strongly
>>>> encouraged for all website/application testing.
>>>> Regards
>>>> Vivienne L. Conway, B.IT(Hons), MACS CT
>>>> PhD Candidate&  Sessional Lecturer, Edith Cowan University, Perth,  W.A.
>>>> Director, Web Key IT Pty Ltd.
>>>> v.conway@ecu.edu.au
>>>> v.conway@webkeyit.com
>>>> Mob: 0415 383 673
>>>> This email is confidential and intended only for the use of the
>>>> individual or entity named above. If you are not the intended  recipient,
>>>> you are notified that any dissemination, distribution or  copying of this
>>>> email is strictly prohibited. If you have received  this email in error,
>>>> please notify me immediately by return email  or telephone and destroy the
>>>> original message.
>>>> ______________________________**__________
>>>> From: RichardWarren [richard.warren@userite.com]
>>>> Sent: Tuesday, 3 April 2012 2:28 AM
>>>> To: Kerstin Probiesch
>>>> Cc: 'Eval TF'
>>>> Subject: Re: AW: Using AT for evaluation
>>>> Dear Kerstin,
>>>> I don't object too much if a "real user" (ie blind person) doesn't  use a
>>>> screen reader to test the site - the most important thing is that a
>>>> screen
>>>> reader is used. Only a screen reader's audio output would demonstrate
>>>> misspelled words and phone numbers (thanks Denis). You would not test
>>>> without using a visual browser so you should also use an audio  browser.
>>>> For an enclosed environment, such as an Intranet, it could be  possible
>>>> to
>>>> exclude testing with certain AT - only IF you know that the  technology
>>>> will
>>>> not be required. However this would then mean that the Intranet  could
>>>> never
>>>> be used by such a disabled person and could breach employment
>>>> legislation.
>>>> Richard
>>>> -----Original Message-----
>>>> From: Kerstin Probiesch
>>>> Sent: Monday, April 02, 2012 3:51 PM
>>>> To: 'Denis Boudreau' ; 'RichardWarren'
>>>> Cc: 'Eval TF'
>>>> Subject: AW: Using AT for evaluation
>>>> Hi Richard, Denis, all,
>>>> I also think that test with "real" (at least screen reader) users are
>>>> important and that we should strongly recommend it but leave it
>>>> optional. As
>>>> I remember the discussion on our last telco there are two aspects:
>>>> - testing with AT and
>>>> - accessibility supported
>>>> I think we have an intersection but also other aspects like: are
>>>> technologies like PDF and Flash accessibility supported? Depending  upon
>>>> the
>>>> answer it will have probably different consequences for our  methodology.
>>>> Also different use cases like internet and intranet (especially  when it
>>>> comes to scripting for JAWS or other screen readers in closed
>>>> environments)
>>>> might have an impact. I'm thinking about if we could find for the  tests
>>>> of
>>>> intranets something better than just "optional" without reducing the
>>>> audience of our methodology in whole.
>>>> Best
>>>> Kerstin
>>>> Von: Denis Boudreau [mailto:dboudreau@**accessibiliteweb.com<dboudreau@accessibiliteweb.com>
>>>> ]
>>>> Gesendet: Montag, 2. April 2012 15:40
>>>> An: RichardWarren
>>>> Cc: Eval TF
>>>> Betreff: Re: Using AT for evaluation
>>>> Hi Richard,
>>>> I would also like to weigh in with Richard here. All too often,  screen
>>>> reader testing is considered a luxury that can be done without. I  am
>>>> one of
>>>> those who think that an evaluation cannot be considered complete  unless
>>>> some
>>>> screen reader testing has been conducted - and ideally, not only by a
>>>> developer, but really by a "real" end user with a visual  impairment,
>>>> using
>>>> the assistive technology regularly. The same could be said of other  end
>>>> user
>>>> using other tools for other disabilities or limitations, but at the  very
>>>> least, screen readers.
>>>> There are always things that are brought up with AT testing that  cannot
>>>> be
>>>> flagged using only a checklist. Some of the things that come to  mind are
>>>> links used for buttons that really should be coded as<button>, an
>>>> overwhelming number of heading elements in a page (big menus and  fat
>>>> footers
>>>> anyone?) or quite obviously, any script that opens up or reveals
>>>>   content in
>>>> a page. I recently had big surprises simply with phone number  formats
>>>> and
>>>> how screen readers read them. That was another real eye opener (no  pun
>>>> intended).
>>>> This is why I tend to follow this pattern personally:
>>>> * testing the web page with a screen reader
>>>> * using an automatic checker for basic problems
>>>> * running manual testing to complete the audit
>>>> And whenever I am being offered the budget to do so, calling in a
>>>> visually
>>>> impaired friend or two who can push those tests much further that  my
>>>> sighted
>>>> self can push them.
>>>> /Denis
>>>> On 2012-03-29, at 6:48 PM, RichardWarren wrote:
>>>> First – sorry I missed the last half of the teleconference – system
>>>> crash.
>>>> I wish to add to the discussion on using AT in evaluation. I  believe it
>>>> is
>>>> important to use a screen reader at the very least before  completing an
>>>> evaluation. We do the normal stuff first (it is not fair to ask a  blind
>>>> user
>>>> to struggle if we already know that the site is impossible for  them).
>>>> But as
>>>> soon as we are happy that a site is reasonably good we always ask
>>>> someone to
>>>> check with their screen reader. Most times their comments re- inforce
>>>> what we
>>>> have found (often with better phrasing<G>). But just occasionally  they
>>>> find
>>>> something that our other systems do not pick up. For example the word
>>>> “accesskeys” sound completely Russian unless it is written “access
>>>>   keys”, or
>>>> “access-keys”.
>>>> I strongly believe that the audio output needs to be checked  properly.
>>>> If
>>>> you look at our outline procedure sent to this list on 26 Feb you  will
>>>> see
>>>> that we find someone who is new to the site to use a screen reader.  This
>>>> approach gives us a high level of confidence in our final evaluation.
>>>> Richard
>>>> Technical Manager
>>>> Website Auditing Limited
>>>> http://www.userite.com
>>>> This e-mail is confidential. If you are not the intended recipient  you
>>>> must not disclose or use the information contained within. If  you have
>>>> received it in error please return it to the sender via  reply e-mail and
>>>> delete any record of it from your system. The  information contained within
>>>> is not the opinion of Edith Cowan  University in general and the University
>>>> accepts no liability for  the accuracy of the information provided.
>>>> CRICOS IPC 00279B
>>> --
>>> Michael S. Elledge
>>> Associate Director
>>> Usability/Accessibility Research and Consulting
>>> Michigan State University
>>> Kellogg Center
>>> 219 S. Harrison Rd Room 93
>>> East Lansing, MI  48824
>>> 517-353-8977
>> --
>> Detlev Fischer
>> testkreis - das Accessibility-Team von feld.wald.wiese
>> c/o feld.wald.wiese
>> Borselstraße 3-7 (im Hof)
>> 22765 Hamburg
>> Tel   +49 (0)40 439 10 68-3
>> Mobil +49 (0)1577 170 73 84
>> Fax   +49 (0)40 439 10 68-5
>> http://www.testkreis.de
>> Beratung, Tests und Schulungen für barrierefreie Websites

Shadi Abou-Zahra - http://www.w3.org/People/shadi/
Activity Lead, W3C/WAI International Program Office
Evaluation and Repair Tools Working Group (ERT WG)
Research and Development Working Group (RDWG)
Received on Wednesday, 4 April 2012 18:05:27 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:40:20 UTC