That Standards Guy



That Standards Guy is the online persona of Karl Dawson, a web developer living and working in Ipswich, England.

I'm a member of the Guild of Accessible Web Designers and the Web Standards Group and team member at Accessites—an awards site to recognise accessible and usable websites.

I specialise as a front-end developer and worry about the minutae of semantic (X)HTML and CSS, accessibility, microformats, typographic rhythm and grid design. I also care about the user experience and remind myself constantly of visitor site goals when working with clients and their aims.

That Standards Guy is proudly powered by WordPress using my own “StrictlyTSG v3.0” theme. Site Policies.

Stay up to date via the RSS feed. What’s RSS?

How useful are accessibility evaluation tools?

Please update your bookmarks, on 1 September 2006 this article was updated and re-released via Accessites. This revised article is replicated on my blog for comments.

Best regards, Karl

6 Responses to “How useful are accessibility evaluation tools?”

  1. Are you SURE this one can be fully automated?
    9.1 Provide client-side image maps instead of server-side image maps except where the regions cannot be defined with an available geometric shape.

    As I read it, what you can automate is “have you provided a client-side image map” or even at best (which I think is unlikely) “do you have a server side image map that is not currently defined with an available geometric shape”. As it is unlikely that the automated tool will know the purpose of the server-side image map, I imagine it would be impossible for an automated tool to determine whether or not a server-side image map could be adjusted to still suit the purpose and also fit an available geometric shape.

    Okay, I’m going to the far end of a fart on this one, but I think this is another one to move into the second category…

  2. Yeah, I’ve read that checkpoint a few times now and I think a tool would check for a client-side image map and award a tick in the box if found. If a server-side image map was detected, then you would need to perform a manual check for equivalent text links. I don’t have much, if any experience with image maps but if you have a legitimate reason for it to be server-side then you will have to manually check it.

    I don’t know, I’m inclined to leave it in the first group because I think the intention was to check for a client-side image map only. But then again, if it were server-side you’d need to manually check for redundant text links to pass it. Even as the article stands now, 5 wholly automatic checks out of 65 is pretty poor going. Removing the last Priority 1 checkpoint from the list is academic if you were thinking of buying a tool after reading this…

  3. Karl, good article, I would agree with Jack though and add that if you are saying that you would check to detect an image map, and then manually check for equivalent text links, then that is very similar to checking an image for an appropriate alt text. There’s no tool that can check the links to see if they are equivalent.

  4. Cheers Grant, 2 to 1 in favour of kicking 9.1 out (but I don’t really disagree).

    When I get a few mins, I’ll amend - so kick me in a week if not done. I just wanted to quickly acknowledge your comment :)

  5. Okay then, while I’m being awkward:
    12.4 - an automated checker can check if a label is associated with a control but not that it’s associated with the right control. Eg Name and Address labels and fields. How does it know which is which unless it can understand the purpose to which that form field will be put. Some manual checking still required?
    4.3 - it can check if a primary language is identified. But unless it’s really clever, will it actually determine if it’s the right one? If you put “en” and then spell “colour” will it tell you that you’re wrong ?(should be “en-gb” for us UKers) If not, then again it’s a manual one.

    I’ll let you keep the validate and deprecate ones though :-)

  6. haha Jack, cheers! ;)

    I assume 12.4 checks for matching for and id attribute values - as to whether you code them correctly in the first place is a matter for manual QA checking, yeah.

    4.3 is naughty because it uses the word “primary” with the word “natural”. W3C have already used the phrase primary language to mean metadata about the document as a whole with “natural language” equating to the text-processing language that is more specific and defined via the lang or xml:lang attribute. Thankfully, the accompanying guidelines for 4.1 and 4.3 sort out the ambiguity.