Archive for the ‘experiments’ Category
This last year, I’ve been happy that engineering teams are asking me to talk about accessibility earlier and earlier in their development process. A few months ago, one team asked me to look at a prototype that was a bunch of images with a few hot spots. I’m used to getting code and running my typical test pass. But, in this case, I couldn’t.
At first, I thought about what it would take to make the prototype accessible, about the HTML that they would need to write. But, this was an MVP (minimum viable product) to get quick feedback on their design. It didn’t seem worth the investment and it didn’t feel agile enough. So, how was I going to get feedback about the usability of the non-visual user experience at the same time they were getting feedback about the visual user experience?
I took the prototype to a friend who is blind and I pretended to be a screen reader. Instead of him interacting with the computer, he told me what keys he would press and I responded as I understood a screen reader would respond.
“To start,” he said, “I would press the down arrow 4 times.”
I responded by reading the first 4 lines of text on the page.
He said, “Hmm. This page doesn’t have any headings or landmarks?”
I broke character for a second, “Yes, it does. It has 10 headings and 4 landmarks.”
“Oh, well the screen reader would have told me that when the page loads and then I would have used that information to get a sense of the layout of this page.” He said.
“Ahhh-ha! Right. Yes, I forgot that part. Let’s start again.” And I returned to my role as screen reader.
For the next 30 minutes we “played” with the site. I learned where I had holes in my understanding of some of the strategies that people use when navigating a new site with a screen reader. I evolved my knowledge about how a screen reader announces certain aspects of a web page.
I went into the session with a list of recommendations I was planning to make to the engineering team. I left with a new list of recommendations and set of strategies to use the next time I play the role of a Screen Reader.
Many thanks to Kelly Ford for collaborating with me on this experiment.
[Note: This is based on the idea of a Wizard of Oz experiment, common UX research mechanism.]
Yesterday, I presented as part of a social media event and had an awesome time. The other speakers were stimulating and fun.
Today, I wanted to publish my slides on my website and on slideshare. So, I opened Acrobat to take a look at the accessibility of the PDF. I knew that the images would not have text equivalents so I was prepared to add those. I was not prepared for the following 4 hours of frustration…which is resulting not in an accessible PDF but in this blog post.
First off, Open Office Impress did not generate a tagged PDF despite me checking the checkbox. boo!
Secondly, when I generated XHTML instead of PDF, I lost all of the formatting and images. boo!
Thirdly, Acrobat only saves about 45 characters worth of each of the descriptions of the images despite giving me a text box that will allow me to enter at least 256 characters (I’m guessing because that’s the limit in the HTML 4.01 spec). boo!
I learned a valuable lesson today: all future decks will start and end in XHTML.
The book…in process. This is a screen shot taken with my phone then emailed to flickr. I wanted to see what they would do with the text.
First, I edited the title of the photo on my phone from random text (imgnnn) to “Testing” then I created an email with “Testing for text” as the subject and “Will this text show as description?” in the body. Flickr used the email subject as the title of the photo and the body as description. Cool!
On the flickr page for this photo, the actual alt is null. How do you folks feel about that? Since the title is a good equivalent, does the image itself need an equivalent in the alt attribute? I certainly want an alt present on the image element – even if it is null – to indicate that it has been considered. I could repeat the title as the alt value, but that would be redundant. Too bad there isn’t a way to associate the heading with the image to indicate they create a single semantic unit, but then again, this whole page is metadata for this image…in a sense the whole page is a single semantic unit.
If you all had this debate while I’ve been “away,” please point me to the archives. Thanks!