This post is adapted from an email thread in the DLF DAWG listserv.
Kristin Briney posted Writing Alt Text for a Scientific Figure on August 15th, 2024 where she provides a brief introduction and shares resources to guide the reader in describing scientific figures. In response to this post, members of DAWG shared other resources and experiences (Thanks to Wendy Robertson at the University of Iowa for sharing Briney’s post and Jasmine Clark for compiling and organizing the thread in blog form):
Direct Responses to Briney’s Post:
We’re working with our campus screenreader specialist on making our content accessible, and he’s pretty excited by the idea. Our grad student created the ALT text, and did so by dropping the images into UMGPT (our campus’ private secure ChatGPT clone), taking the generated captions, and cleaning them up. You can see examples of the resulting ALT text attached to the images in this draft internal presentation (which we ended up not using, but it’s useful backup for this conversation). The screenreader specialist liked these.
Here is our final version. Less ALT text, and more links to screenreader-adapted versions of the original spreadsheets from which we drew the data for the images.
– Courtesy of PF Anderson, University of Michigan
I love the formula for writing alt text for data visualizations, in Briney’s post. I ran WAVE on the link that Jen-chien sent.
alt text = *Chart type* of *type of data* where *reason for including chart*. *Link to source data.*
with the example
Column chart of research data availability where research data on the internet disappears at a rate of 2.6% per year. For underlying data, see “Figure2_UnavailableByYear.csv” file at https://doi.org/10.22002/h5e81-spf62
Embedded in this recommendation is the assumption that the researchers have published their data: in an open repository, in an accessible and usable format, with a persistent ID.
This recommendation seems as though it would also improve the article. If you can’t summarize your visualization in this format, then perhaps it doesn’t support your argument, or it isn’t needed, or it’s too complicated, or you don’t really understand your research!
I’m interested to hear the opinions from actual users of assistive technology. Is the advice in these posts helpful or unhelpful?
– Courtesy of Susan Hoover, University of Houston
Additional Related Resources:
Demo: AI to interpret charts for the visually impaired — arXiv Accessibility Forum 2024
<https://www.youtube.com/watch?v=kUsk7wVVQAo>
AI and Accessibility — arXiv Accessibility Forum Session 2024
<https://www.youtube.com/watch?v=mdzC9J78i6I>
Graphic Medicine in Disability Studies – Getting Started with Graphic Medicine in Disability Studies, by Ann Fox
– Courtesy of PF Anderson, University of Michigan
Dena Strong published a “data verbalization” post that shows “how to craft a simple and effective data verbalization…”
– Courtesy of Jen-chien Yu, Director Of Library Assessment, University of Illinois
Response to Strong’s post (shared by Yu above):
This is great! We’ve been having debates about the accessibility of comics when used in presentations. This:
“What type of visual is this?
What does it show?
Why does it matter?”
maps to what we’re calling functional ALT text, as contrasted with descriptive ALT text. For presentations, it’s about why is this on screen — is it decorative, a cognitive or symbolic placeholder for more complex content, is it communicating information, etc. For functional alt text, we keep the “what does it show” really simple and short.
What we are finding is that folk who were blind from birth have different desires from alt text than folk who lost vision later in life. Our screenreader specialist was blind from birth, and will tell us, “I don’t care if it’s pink. That doesn’t tell me anything. Just tell me why it’s there.” Folk who have memories of color and shape may actually want that information.
I also had tried putting the links to the screenreader-versions of data in the ALT text and have been instructed to NOT do this, because it’s annoying to have to listen to the computer read the URL aloud. I’ve also been told not to put the full links on the slide itself, but I balk at that, because for accessibility for me, I want to be able to see that before I click on it, and I don’t want to have to hover to see it. A suggestion provided by Ann Fox was to add the stripped down functional alt text in the presentation slides, and create a separate external Google Doc or downloadable file with enriched alt text. I’m working right now on my first effort to do this. It’s a lot of work. It looks something like this.
SLIDE 1
Image 1 (functional ALT text): …
Image 1 (descriptive ALT text): …
Image 1 (additional resources, links, data): …
Image 2 …
– Courtesy of PF Anderson, University of Michigan
A final response to the different resources shared:
I think those guides are indeed valuable resources to share with graduate students for their theses and dissertations—they offer helpful guidance across various areas.
Resources like the Do No Harm Guide, referenced from Writing Alt Text for a Scientific Figure by Kristin Briney, provide excellent frameworks for describing charts, such as the Four-Ingredients and Four-Levels Models, which can effectively address diverse user needs based on vision experiences, as PF mentioned. However, I’d like to offer a quick clarification on alt text for graphs that might help us apply these tags more effectively. Balancing alt text length is crucial; alt text within the <alt> tag is most effective when kept short and direct, ideally a single sentence conveying the key information.
For more detailed descriptions, best practice suggests using a <longdesc> tag or a linked long description with a <div> tag to add structure. This method enables screen reader users to navigate complex information more easily, reducing the risk of feeling “trapped” in lengthy, continuous alt text.
Another alternative is to link to a detailed description near the image, leading users to a separate page or section with a fuller explanation. This setup allows users to engage with image details at their own pace, focusing on specific information without becoming overwhelmed. For complex visuals, this flexible approach can enhance both accessibility and the user experience.
Just throwing this out there—I feel a bit panicked when I see people trying to put an entire block of text into alt text. It can easily end up overwhelming and confusing for screen reader users…
– Courtesy of Wen Nie Ng, MIS, UXD, CPWA, Digital Collections Librarian