WRITING IMAGE DESCRIPTIONS
Alternative text, known as alt text, is a short, written description of an image that makes images accessible for people who are blind or have low vision and who use screen readers. Screen readers cannot read images, only text. The screen reader reads aloud the information on a computer screen and will tell the user there is an image present. If alt text is provided, the screen reader will read it aloud. If there is no alt text, you will know there is an image, but you won't be told anything about that image.
How Alt Text is Read
The Hidden Image Descriptions Making the Internet Accessible, a New York Times article from February 2022 opens with a series of empty boxes overlaid with short lines of text, noting that each line of text stands in place of an image. This is an example of alt text - a short, concise literal description of the image.
People who are blind or have low vision often rely on alt text when reading e-books with images. Alt text can also be detected and read aloud or translated into braille with screen readers, assistive technology software programs, apps, or browser extensions. Alt text is found in HTML code, and without a screen reader or assistive technology, it is not obvious. Some companies such as Meta, Microsoft, and Google have AI technology that automatically generates alt text. This can result in confusing, incorrect, and often misleading alt text for images.
Artist Alex Haagaard highlighted the pitfalls of AI-generated alt text with an exhibit that was part of #Crip Ritual, a virtual, multi-sited, participatory exhibition of artwork exploring themes of disability culture and ritual. Shitty Alt Text features printed digital images from Instagram with examples of actual auto-generated alt text inside a box with no image. The AI-generated alt text for the Mona Lisa, for example, is "Image may contain 1 or more persons," which obviously does not describe the actual painting in any way.
When alt text is not accurate, people who are blind or have low vision are misled about the visual content and while immediate and convenient, AI-generated alt text is not always accurate. Formed in 2020, Scribley, a certified Woman Owned Small Business, writes premium alt text and long descriptions for images and audio descriptions for video, written by actual people, with attention to detail, purpose, the meaning behind the image, context, and the reading level of the audience.
A long description provides more extensive and detailed visual information about an image. The length of the description, anywhere from a couple of sentences to multiple paragraphs, will depend on the complexity of the image. Long descriptions have a narrative structure, describing the most important content of the image first, followed by details that speak to its nuances. Unlike alt text, long description is visible to everyone, and usually appears beside or beneath the image. When writing long description, avoid repeating information that is already included in the accompanying text.
Developed by The DIAGRAM Center, a Benetech Global Literacy initiative, the Poet Training Tool is a web-based image-description resource with an interactive tool that guides users through the process of determining when a description is needed and how to write descriptions. It also has a section where you can upload your own images and practice writing descriptions.
Captions provide the reader with the basic information they need to understand an image and they should be clear, accurate, and complete. The appropriate format for captions can vary, but information should clearly identify people and locations and provide some context or background. Captions should be brief, one or two short complete sentences, written the present tense. When writing captions, don’t editorialize or make assumptions or characterize the content in descriptive terms, such as beautiful, dramatic, or horrifying.