Images of text: accessible or not?

Images of text are intended to be seen, but what if we design them so they are intended to be perceived?

Images of text are broadly in use throughout the Web. They are used to market products, convey complex information, and to present text in a stylised way that is efficient to design and produce. Images of text are an unsolvable problem, but only if we design them to be so.

When thinking about how and why we need to make these design choices accessible, we can always seek guidance from the W3C’s alt Decision Tree, and from the rationale behind the Web Content Accessibility Guideline’s Success Criterion 1.4.5: Images of Text. Let’s look a little deeper into the complexity of images of text.

Images of text can be:

• Logos (not subject to the same stringent WCAG requirements as other types of images of text)
• Complex infographics and diagrams
• Graphs and maps
• Stylised presentations of text

Sometimes images are the accessible option

Images of text can be the accessible option for some people:

• People who are semi-literate or who have a language-processing disability, and can understand diagrams more readily than text
• Complex information can be more cognitively digestible in a graph or infographic
• As they are made of maths and not of pixels, scalable vector graphics (SVGs) can be used and later be visually customised to a user’s preferences.

For many, images remain inaccessible

Not all people can perceive images of text, and they present some significant accessibility barriers:

• Pictures are inherently unperceivable by people who are blind.
• Pictures of words may be visible by people who are dyslexic and people who are colour-blind, but visibility does not equate to cognitive or linguistic understanding. 
• Colour contrast issues are not an inherent flaw of images of text, but they are often  defected in accessibility audits and are commonly seen throughout the Web.

How to solve?

So, if images of text aren’t going anywhere, how do we make them accessible?

If the image is relatively simple, we can replicate the text in the <alt> attribute or aria-label for screen reader users. While we have resolved the problem for one group of users with this approach, it does not solve for people with print disabilities who require the ability to customise text to their own requirements. 

In most cases, providing a programmatic duplication of content is the solution for inaccessible images of text.

In the case of a graph, this might mean also presenting the data in a semantically-defined tabular format. Placing these two components adjacent to one another may receive design push-back, but the presence of duplicated content is the solution to providing inherently inaccessible content to an audience.

Conclusion

The question is not whether images of text are inherently or always accessible or inaccessible. The real question is whether the information they convey can be perceived and understood by everyone who needs it.

When images of text are used, equivalent alternatives should be provided so that users who cannot perceive or customise the image are not excluded. Accessibility is not about avoiding images of text altogether, it is about ensuring that information remains available regardless of how it is presented.