The author Seanan McGuire relayed a story during a convention panel on accessibility some years ago that stuck with me. During an industry panel, audience members could preview a comic issue at the front of the room while the panelists were taking moderator and audience questions. Only one person could examine the preview at a time, and the people involved were selected at random from the audience by raising their hands at intervals throughout the panel.
Also in attendance was a hard of hearing (HoH)/Deaf person. As such, an American Sign Language (ASL) interpreter relayed the spoken information from the panel to the person as an effective accommodation. However, the delay between when the speakers had finished and the interpreter had relayed the speech through sign meant that every time a new volunteer was selected to review the comic, a hearing person had already been chosen by the time the interpreter had finished relaying that the comic was available again. Seanan, as someone conversant in ASL, noticed the situation and, in slight breach of protocol regarding interpreters, directly signed to the person and asked if they wanted to see the comic. At the person’s enthusiastic yes, Seanan was able to direct the attention of the person in charge of the comic preview to make sure the HoH/Deaf person saw the comic at the next opportunity.
This story is a good example of the difference between accommodation and accessibility. Reasonable accommodations for disabilities are required by the Americans with Disabilities Act (ADA). Title II talks about State and Local governments, which usually includes libraries, and Title III covers businesses that are open to the public, which should cover most libraries not covered under Title II.1 The biggest difficulty with accommodations as the prevailing standard is that it places the burden upon the person with the disability to disclose that disability and then make a request to be included. Once a request for accommodations is made, the disabled person gets to wait and hope that their request isn’t rejected under one of the many permissible ways to deny a request, even with the ADA’s requirements. Accessibility doesn’t wait for a person to make a request or file a suit, it provides ways of inclusion and participation up front for a person without requiring disclosure or identification of a disability. If inclusion is one of the goals of a library or library system, then accessibility is one of the necessary means to that goal.
In addition to inclusion goals, many people without disabilities use accessibility features to increase their enjoyment or engagement with digital materials. Successful implementations of accessibility features benefit people outside the target group who might not consider themselves disabled at all, but come to appreciate and use accessibility features and improvements to their physical and online spaces.2 Much like how curb cuts made navigating sidewalks and physical spaces easier for more than wheelchair users, accessibility tools make navigating and engaging with digital works easier. Subtitles and captions are excellent for catching dialogue that has been mixed too quietly or in situations where turning up the volume would disturb someone sleeping, working, or attending a virtual meeting. Descriptive audio tracks make it possible to follow along with a work if you have to keep one or both eyes on something else in the room. “Alt-text” is named for the “alt” attribute in the “img” tag of Hypertext Markup Language, or HTML, that provides a textual description of an image. It benefits visually impaired people who want a description of an image online, people who turn off images to save bandwidth (or browse with text-only browsers), and people who look at an image and don’t get why it is included. Additionally, alt-text is the perfect place to explain a meme or a joke.
The COVID-19 pandemic forced a pivot from in-person gatherings to virtual offerings in most aspects of life, whether socializing with friends, attending school, work, or events, or maintaining body, mind, and health. The already staggering pace of online content creation increased even more as people started podcasts, vlogs, picture diaries, and uploaded all kinds of media to their favorite social sites. Libraries that engaged in virtual programming met, sang, danced, told stories, did science and art, and got to know parts of their communities that had not set foot inside a library building.
Accompanying the shift to the Internet and virtual venues was an increased interest in accessibility features and settings for participants and creators alike. While many library employees became minor and major stars of their community on camera, the necessary elements to make those performances accessible to broad audiences didn’t always follow. Why didn’t accessibility get baked into all of this new content? Because accessibility takes either time or money to include. Libraries and other content creators realized they could not provide completely accurate transcriptions without a pool of captioners and transcribers that cost more money than they had to spend. In response, technology companies scrambled to find a software solution to accessibility. For the most part, whether in presentation software, meeting software, or the uploading of video to popular video-sharing sites, the solution provided by technology companies was speech-to-text engines that provide automatic captions for presentations, meetings, and programming. It became immediately apparent that these engines were not going to provide perfect fidelity and a 100% accurate transcript. Depending on the context of the speech or the accent of the speaker, they might not be accurate at all. In my experience, speech-to-text consistently mangles non-white names, technical terms, or acronyms, many of which are crucial to understanding the audio content. Despite the flaws of these automatic options, they do provide more accessibility than completely inaccessible videos. Similarly, automatic creation of descriptive text for images by processing the images through a trained algorithm would be better than undescribed images, but is not the same as humans providing descriptive text and context for images.
Many of these speech-to-text engines, computer captioning, and transcription programs are included in the standard cost of a program or service. Their somewhat more accurate cousins can be purchased at what appear to be reasonable rates and then deployed across the organization for much less cost than hiring and/or training humans to do the work and giving them the time to do the work. Because some is better than none, and with the volume of material that needs captioning and describing, “good enough” seems to be the standard selected for accessibility, rather than “accessible.” “Good enough” programs look increasingly attractive as a solution once you look at the amount of time needed to provide accessibility for images, audio, and video.
For example, I recorded a ten-minute lightning talk video for the WLA conference this year. The recording, editing, and synchronization of the audio and video tracks for that ten-minute video took two or three hours just to get the core product in place and then two more hours of waiting for the video and audio of the final product to render. For additional accessibility, however, I also decided to include a subtitle file along with the video, so that someone with a hearing impairment would not only get to see the information on the slide deck, but also see what I was saying. The creation and inclusion of the subtitle file added four more hours’ work before the video was fully ready for upload. Seven hours of work on a ten-minute lightning talk seems like a lot of time to invest. As it turned out, the platform that the conference selected for displaying the video didn’t recognize the embedded subtitle track, and didn’t offer anywhere to upload the subtitle track separately. Despite the presence of accessibility features baked into the original recording, anyone hoping to watch the presentation with subtitles-for accessibility or ease reasons-was completely out of luck.
To mitigate the issue, and so that there would be a record of the content of the presentation after the conference, I took to heart the request of Sumana Harihareswara, notable technologist and stand-up comedian, to provide a text version of the talk as well.3 While it was less of a burden to upload, since I had the script and slides already created to serve as a base for the final textual version, there was still work involved in formatting, embedding links, and otherwise transforming the talk into a hypertext state that would get the information across to someone reading as well as someone watching. That amounted to more of my time spent on something that I wanted to persist and be accessible to others.
All told, I probably spent five hours in the production process for the video and five to seven more hours on building and embedding the subtitle file and building and checking the “with slides” and “without slides” hypertext versions of a ten-minute talk. I put in this extensive effort without any guarantee any of the conference attendees watching the video or reading the text version of the talk would require or use the accessibility features that I had built in. If the talk had been forty minutes or an hour, many more hours of my time would have been spent on providing accessibility features. At my current wage, all of that time spent would have cost more than $500 to produce the ten-minute talk. If I wanted to save the time, I would have had to find someone willing to caption the talk and/or produce a transcript of the talk and pay them a well-deserved amount of money to produce those artifacts. A cursory internet search for human transcription suggested that for the ten-minute talk, the process of producing a transcript within a few days (or weeks) ranged from a low potential cost of $20 (assuming no difficulties) through $32.50 (also assuming no difficulties). If the talk had been live, the captioning could have been $11.10 (likely a little more) for that same block of time, assuming it was possible for the captioner to tie in to the presentation system. The result of hiring a professional would have been spending less than a tenth of the cost of having an accessibility-interested amateur do transcription and subtitle timing or attempt to have someone do real-time captioning for a live event.4
For one individual, doing a single talk or event, the cost of captioning and/or transcription seems particularly manageable. For an institution to provide those services for all of their programs at all of their programming times, that cost is multiplied by the number of virtual programs that need work. Quickly, the reasonable cost for one program becomes potentially budget-breaking to provide for all programs. The potential expense in the abstract moves accessibility back toward the realm of accommodation, back toward using “good enough” tools because they are what the budget appears to afford, and back toward holding programs, meetings, and events in person instead of online to avoid the additional cost of online accessibility.
I don’t yet see enough active demand for accessibility at live and virtual events from attendees for a captioner/transcriber to pitch hiring them for a full-time library job in most places. However, my library system produces enough material in programming and marketing that one or two employees who provide captioning, transcription, and subtitling for all of those materials would have enough work for full-time employment. Not knowing if or how often accessibility features are used tempts the budget-conscious organization to leave accessibility up to the individual staff member creating content instead of employing or contracting a professional to provide this service as part of standard practice, and as a result, accessibility suffers.
It would be a shame if, having demonstrated that we can reach people who haven't been part of library services before, we went back to old ways of doing things out of concern for time, cost, or a belief that going beyond the minimum requirements of ADA is unnecessary. Or that we settled for “good enough” accessibility when it’s possible to do it correctly and consistently with the application of time and money. Accessibility is not a perk or something that can be settled with “good enough” tools that consistently miss key things. Keeping the connections made online and providing good digital library services requires having a plan for consistent, high quality, accessible materials and programming. Perhaps most selfishly, getting used to the idea of including accessibility as a core component now will make it easier to keep using library services when accessibility stops being optional and instead becomes necessary, for age and disability comes for us all in the end.