ARIA15: When Alt Text Isn't Enough

No description exists for the painting. The alt attribute says "Lady MacBeth" — a name, not an explanation. For a screen reader user, that's where the image ends. The brushwork, the date, the artist, the context — gone. This is the gap WCAG Technique ARIA15 (opens in new window) is designed to close.

For the developer reading this: aria-describedby is not a replacement for alt. It's a second layer — a way to connect an image to descriptive text that already exists on the page. The distinction matters operationally, and getting it wrong produces either redundancy or silence.

The Technique

ARIA15 (opens in new window) describes a sufficient method for meeting WCAG 2.1 Success Criterion 1.1.1 Non-text Content (opens in new window). The technique is straightforward: when a short alt attribute cannot convey the full meaning of an image, use aria-describedby to point to descriptive text elsewhere on the page.

The W3C's own example is instructive:

<img src="ladymacbeth.jpg" alt="Lady MacBeth" aria-describedby="p1">
<p id="p1">This painting dates back to 1889 and is oil on canvas. It was created by John Singer Sargent, and represents ...</p>

The alt attribute provides the short name. The aria-describedby attribute points to a paragraph id that carries the longer description. The screen reader announces both. The sighted user reads the paragraph naturally — no hidden text, no duplicate content, no aria-hidden gymnastics.

One structural note worth emphasizing: aria-describedby accepts multiple space-separated IDs. A complex diagram could draw its description from two or three page elements simultaneously. This flexibility is part of what makes the technique powerful in practice, particularly for data visualizations, architectural drawings, or scientific figures where meaning is distributed across captions, legends, and body text.

Why This Matters

For a screen reader user encountering a complex image with only a short alt value, the experience is abrupt. The image exists. Its name is announced. Then the reader moves on — with no access to context that a sighted user absorbs visually.

This isn't a hypothetical friction point. It's a documented pattern. Our research on automated versus manual testing methodologies shows that automated tools catch missing alt attributes reliably, but they cannot evaluate whether the alt text is sufficient. A tool sees alt="Lady MacBeth" and marks the image as passing. A human reviewer — or a screen reader user — recognizes that the name alone conveys nothing about the work's content, date, medium, or meaning.

The gap between "technically present" and "actually useful" is where SC 1.1.1 lives. The criterion requires that non-text content have a text alternative that "serves the equivalent purpose." A name rarely serves the equivalent purpose of a painting.

Users with cognitive disabilities also benefit from explicit descriptions linked to images. When descriptive text appears visibly on the page — connected to the image through aria-describedby — it supports multiple comprehension strategies simultaneously. The visual and textual channels reinforce each other.

How It Works

The mechanism is a DOM relationship. aria-describedby takes the id of one or more page elements and exposes their text content as the accessible description of the target element. Screen readers typically announce this description after the element's accessible name and role.

The W3C's canonical pattern:

<!-- The image -->
<img src="ladymacbeth.jpg" alt="Lady MacBeth" aria-describedby="p1">
<!-- The description — visible to all users -->
<p id="p1">This painting dates back to 1889 and is oil on canvas. 
  It was created by John Singer Sargent, and represents ...</p>

Multiple IDs work the same way:

<img src="chart.png" alt="Q3 Revenue Chart" aria-describedby="chart-caption chart-methodology">
<p id="chart-caption">Revenue increased 14% in Q3, driven by enterprise subscriptions.</p>
<p id="chart-methodology">Data sourced from internal CRM. Figures exclude one-time licensing fees.</p>

The screen reader concatenates both paragraphs into a single description. The order follows the ID list, not DOM order.

One practical constraint: aria-describedby only accepts id values — not class names, not element types. Each referenced element must have a unique id on the page.

This table maps the three most common image-labeling mechanisms to their functional roles. ARIA15 sits in the middle column — it supplements the name, it doesn't replace it.

Applying This

For a development team, the practical question is: when does an image need this? The answer isn't always obvious, which is why this technique gets skipped.

In code review, flag any complex image — charts, diagrams, maps, paintings, photographs with narrative content — that carries only an alt attribute. Ask whether the alt text conveys equivalent information. If the honest answer is "no, but there's a caption below it," that caption should be wired up with aria-describedby.

In automated testing, tools will not catch insufficient alt text. This is a known ceiling. Our research on testing methodology gaps documents this clearly — automation handles detection, not adequacy. ARIA15 compliance requires human judgment about whether descriptions are sufficient.

In design, the technique works best when descriptive text is already part of the page layout — captions, figure notes, adjacent paragraphs. If the description exists only as a hidden element, it's inaccessible to sighted users and loses the technique's primary advantage. Design systems should treat image captions as first-class elements, not afterthoughts.

Quick wins: Audit image-heavy content pages — galleries, news articles, research publications — for complex images with thin alt text. Check whether adjacent captions have id attributes. If they do, adding aria-describedby is a one-attribute fix. If they don't, adding the id is the only additional step.

Longer-term: Content management systems should surface the aria-describedby field as an optional input when authors upload images, alongside the alt text field. This builds the technique into the authoring workflow rather than leaving it to developer intervention after the fact.

Higher education contexts warrant particular attention here. Museum collections, digital humanities projects, scientific repositories, and course materials all depend heavily on complex images. The Northeast ADA Center (opens in new window) has consistently documented that campus digital accessibility gaps concentrate in content-rich, image-heavy environments — precisely where ARIA15 applies most directly.

CORS Perspective

Viewed through the CORS framework, ARIA15 sits at an interesting intersection of all four pillars. From a community standpoint, the users most affected are screen reader users and people with cognitive disabilities encountering complex visual content — a population that includes students, researchers, and museum visitors who depend on institutional digital resources. Operationally, the technique has a low implementation cost when page structure already includes captions and descriptions, but requires content author training and CMS support to scale. The risk dimension is real: SC 1.1.1 is among the most frequently cited criteria in accessibility complaints and settlements, and insufficient alt text — even when technically present — satisfies neither the letter nor the spirit of the requirement. Strategically, organizations that frame this as a content quality issue rather than a compliance checkbox tend to implement it more durably; when descriptive text benefits all users visually and programmatically, the case for building it into workflows becomes straightforward.