At the University of North Carolina (UNC), Chapel Hill, the archives and special collections unit recently announced a Conscious Editing Initiative that merits consideration for anybody writing alternative text or accessibility descriptions. The Conscious Editing Initiative seeks to redress inequities in the descriptive language and framing used by the archival community. Specifically, the initiative aims to redress descriptions that presume a white experience and perspective.
This project at UNC Chapel Hill is important to note because of the attention it draws to positionality. What biases and perspectives do you bring to the work of writing alt text and accessibility descriptions? Not only that but what language is appropriate when describing the perceived age, gender, race, or economic status of the people in images? Should some information be withheld when we consider that alt text is publicly accessible metadata? How do our language choices potentially impact the subjects of photos?
As an example of these issues in practice, consider the undated Edward S. Curtis photograph labeled only “Man in Native Dress on Rocks in Water.” Among other things, this description conflates the identities of indigenous communities while appealing to white stereotypes and assumptions about a non-specific “native” identity.
Another example comes from a 1971 photograph featuring mostly white men and one Black woman at the International Conference on the Biology of Whales. The woman, Sheila Minor Huff, was the only person left unidentified in the photo’s caption and she was only found after a period of intense online sleuthing. Gaps in image metadata are important, as are the privacy preferences of the individuals involved, and all of these things must be weighed when creating alt text.
In “It’s Complicated: Negotiating Accessibility and (Mis)Representation in Image Descriptions of Race, Gender, and Disability,” Bennett, et al., discuss some of the difficulties outlined above and recommend practices for image describers based on interviews carried out with screen reader users with intersecting BIPOC identities. For instance, interviewees pointed out the difference between alt-text language that describes appearances versus language that presumes identity. Appearance-based language may point out skin tone, hairstyles, makeup, and accessories while identity-based language posits race, gender, and disability status. Interviewees were opposed to descriptions that assume an identity but they also encouraged alt-text creators to follow the preferences of the individuals in a given photo whenever these are known—including preferences about identity-first versus person-first language when describing disability.
In addition to the above, Bennett, et al., advise considering the needs and preferences of screen reader users when deciding how to write alt text descriptions. Appearance-based descriptions often prove useful to screen reader users in particular scenarios—for instance, during a conversation about representation in media.
As you consider this topic further, you may want to take a look at the Cooper Hewitt Guidelines for Image Description. These guidelines provide multiple examples of long image description and recommend best practices for describing skin color, gender performance, age, and physical appearance.
To learn more about how the disability community is discussing tensions between access and ethics in alt text description, see the following:
- Kudlick, C., & Schweik, S. (2014). Collision and collusion: Artists, academics, and activists in dialogue with the University of California and Critical Disability Studies. Disability Studies Quarterly, 34(2).