JSTOR Digital Stewardship Services: Guidelines for writing alt text

Consider the image’s location, audience, and use:

• What collection(s) is the image part of?
• Is it publicly available or limited to institutional users or specific researchers?
• Is it being used in a specific class or exhibition?

Alt text is context-dependent, not image-dependent.

The same image may need different alt text depending on where it appears, how it’s used, and who the intended users are.

• Describe the image’s content, function, and purpose.
• Include only the information needed to understand the image and its purpose.
• Don’t include unnecessary visual details that don’t contribute to meaning.

• If the surrounding text fully describes the image, use concise alt text such as “Chart illustrating data provided in Description field.”
• Don’t verbatim duplicate text from the page within the alt text.

• Place the most important information first.
• Meaningful details that enhance understanding should come second.

It’s most important to convey the information someone actually needs to understand the image's content and function.

This also lets people choose how much detail they want to engage with when they encounter the image.

• Never begin with “image of…”
• Include the image’s medium or materiality when that information is important to its context and meaning.
  • Example: Painting, black and white photograph, map, faded newspaper

In many cases, the type of image is important for understanding it, especially when the image is complex.

Screen readers announce that something is an image, so “image of” creates unnecessary redundancy.

• Provide all meaningful text from the image in text form, either as part of the alt text or adjacent to the image.
• If the text is long, it should be provided separately.

Meaningful text is any text that is part of the image's purpose and meaning. Access to meaningful text in an image is critical for understanding the image.

Note: In some cases, this may include potentially harmful or biased language. This language should not be omitted if it’s meaningful to the image.

If you’re publishing to JSTOR, review our Community Guidelines for more information on handling hateful or discriminatory language.

• Describe only what is in the image.
• Use neutral, objective language. 
• Don’t interpret or speculate about the image’s content.

• Keep alt text around 1–2 sentences maximum.
• Don’t introduce jargon or new terminology within alt text.
• Include color only when meaningful to the image meaning/context.

Alt text that is too long may cause mental fatigue.

If the meaning of the image can’t be conveyed concisely, use a long description.

• Alt text should be consistent with the writing on the page where the image is located.
• Include any information conveyed in the image that’s alluded to in the surrounding text of the page in the alt text.

• Write alt text in proper sentence case.
• End alt text with a period and a space.
• Ensure all text is spelled correctly.

Provide all text equivalents in the original language of the non-text content.

Example: A print with a French caption should include the caption text as written in French. A translation of the caption may be provided elsewhere in the item’s metadata.

• Avoid unnecessary special characters that may be read ambiguously by assistive technologies.
• Ensure punctuation is intentional and screen-reader friendly.
• Diacritics should be included as necessary.

Some screen readers can’t interpret special characters, and may announce them inaccurately.

Screen readers can interpret diacritics correctly, so those should be included when necessary.

Don’t rely solely on sensory characteristics to describe image content or purpose.

Sensory characteristics refers to how something looks, sounds, feels, etc.

For many people, sensory characteristics alone are limited or unhelpful for understanding an image.

It’s appropriate to include them when meaningful to the content, but it shouldn’t be the only way you describe something. See also WCAG SC 1.3.3 Sensory Characteristics.

• Only describe visible physical attributes when relevant to meaning.
• Include meaningful objective description of what is visually present about a person.
  • Examples: Skin tone, hair color, visible disability aids (wheelchair, cane, etc), clothing style, etc.
• Don’t assign identity categories (e.g. race, gender) unless information is confirmed and meaningful to the image/collection.

A person’s visual appearance is often meaningful to the image. When meaningful, it should not be omitted.

Using appearance-based language (such as “light-skinned” or “long-haired”) instead of identity-based language (such as “Latino” or “female”) helps prevent misidentification and the reinforcement of bias and stereotypes.

• Describe gender only when the person's gender is verified from contextual details and important to the context of the image. 
  • If gender can’t be verified, use neutral terms (person, they/them), especially for photographs of people.
• Describe skin tone using non-ethnic terms when skin tone is clearly visible.
  • Examples: light-skinned, dark-skinned, etc.
• Don’t make assumptions about a person’s race or ethnicity.
  • Only identify if ethnicity/race is confirmed via contextual details and relevant to the image’s meaning.

Descriptive information about people is important for blind and low-vision people and is often meaningful to an image’s content and purpose.

Omitting this information can unintentionally reinforce unconscious bias, as the audience is left to fill in the gaps. However, it’s equally important not to reinforce bias by making assumptions about people, particularly when describing photographs of people.

Making assumptions about a depicted person’s identity can reinforce harmful stereotypes. Only objective information should be provided unless those details are confirmed via provided context and important to the image's meaning.

• Whether you choose to describe or not describe people, be consistent regardless of appearance or assumed identity categories.
• You may have different strategies for different collections. In that case, ensure you are consistent within each collection.

Whether you choose to describe people’s appearances or not, all alt text within a given collection should follow the same approach. 

Consistency is important for accessibility, as it helps someone build mental models of your content.

Consistency also helps prevent unintentional bias. For example, noting skin tone exclusively for images of people of color, but not for images of white people, can reinforce the assumption that whiteness is the “default” state of being.

• Provide all meaningful details, including if an image contains harmful or stereotypical imagery.
• Avoid omitting or censoring meaningful information, even when uncomfortable or potentially offensive. (Diagram Center Image Description Guidelines.)

The goal of alt text is to communicate the full meaning of the image.

In some cases, this may mean describing the biased or harmful nature of imagery or using potentially offensive words featured in the image.

For example, a propaganda poster that features a stereotypical caricature of a particular group should be described that way, as that is meaningful to the image. If someone couldn’t see the poster, and that information wasn’t described, they wouldn’t be able to fully understand the image’s content and purpose.

Note: If publishing to JSTOR, be sure to review our Community Guidelines and ensure any potentially harmful or offensive images fall within the use cases described.