WCAG 1.2.8: The Full-Text Alternative That Reaches Everyone

You've added captions. You've added audio description. And you've still left out a significant group of disabled people. WCAG 1.2.8 exists precisely for them — and understanding what it actually requires reveals something important about how we think about multimedia accessibility.

THE REQUIREMENT

Success Criterion 1.2.8: Media Alternative (Prerecorded) (opens in new window) sits at Level AAA under WCAG's Perceivable principle. It requires that a text-based alternative be provided for all prerecorded synchronized media — meaning video content that combines both audio and visual tracks.

But "text alternative" here doesn't mean a transcript of the dialogue. It means something more complete: a document that reads like a book, capturing everything happening in the video. Visual context. Actor expressions. Physical actions. Non-speech sounds — laughter, ambient noise, off-screen voices. The sequence of descriptions mirrors the sequence of the media itself. If the video includes interactive prompts ("press now to answer"), the text alternative must provide parallel functionality, typically through hyperlinks.

This is sometimes called a media alternative or descriptive transcript, and it's distinct from audio description. Audio description is constrained by the pauses in existing dialogue — it can only fit so much. A media alternative has no such constraint. It can describe everything, fully, in the order it occurs.

Technique	What It Provides	Satisfies 1.2.8?
G69 (opens in new window)	Full text alternative for time-based media	Yes
G58 (opens in new window)	Link to alternative adjacent to media	Yes (placement)
H53 (opens in new window)	Using the <object> element body as alternative	Yes (HTML method)
G159 (opens in new window)	Full text alternative on separate page	Yes
Audio description alone	Narrates visual gaps in dialogue	No — insufficient
Dialogue-only transcript	Misses visual information entirely	No — documented failure

The documented failure mode, F74 (opens in new window), is providing a text alternative that fails to include all audio and visual information. That's the most common way teams think they've met this criterion and haven't.

WHY THIS MATTERS

SC 1.2.8 centers DeafBlind users — people whose vision is too poor to reliably read captions and whose hearing is too poor to reliably follow audio. This is a population that existing multimedia accessibility techniques consistently fail to fully serve.

Captions serve Deaf users who can see the screen. Audio description serves blind users who can hear. A media alternative serves people who can do neither reliably. They access it through refreshable braille displays, reading the text alternative at their own pace, in their own time. That's not an edge case to optimize away — it's a fundamental access need that no other technique addresses.

But the impact extends further than DeafBlind users specifically. People with cognitive disabilities often benefit from a full text alternative because they can move through content non-linearly, re-read sections, or process information without the time pressure of video playback. People with low bandwidth or situational limitations — reading in a noisy environment, on a slow connection, in a context where they can't play video — get complete access to the content through text. A well-written media alternative is genuinely useful to a wide range of people, not just those it was designed for.

When this requirement goes unmet, the exclusion is total. There's no workaround. A DeafBlind user can't approximate access to a video through captions or audio description — those channels simply don't reach them. The content is inaccessible in a way that's categorical, not partial.

HOW TO MEET IT

A sufficient media alternative isn't a dialogue transcript with some description bolted on. It requires deliberate construction.

What to include:

• All spoken dialogue, in sequence
• Descriptions of significant visual information (setting, who is on screen, what they're doing)
• Non-speech audio (laughter, alarms, music that carries meaning, off-screen voices)
• Text of any on-screen text that isn't spoken aloud
• Actions and expressions that carry communicative weight
• Interactive elements translated into functional text equivalents (links, buttons)

What the document should feel like: The W3C's own guidance describes it as reading "like a book." That's a useful benchmark. If someone who has never seen the video reads your media alternative and comes away with the same understanding as someone who watched it, you're in the right territory.

Placement: G58 (opens in new window) specifies that the link to the alternative should be immediately adjacent to the media element. G159 (opens in new window) allows it on a separate page — but the link to that page must be findable. Burying it in a footnote or a help section doesn't satisfy the intent.

The overlap with 1.2.3: SC 1.2.3 (Level A) allows organizations to satisfy that lower-level requirement either through audio description or through a media alternative. SC 1.2.8 at AAA requires the media alternative specifically. Teams that used a media alternative to satisfy 1.2.3 may already have the content — they just need to verify it's complete and properly linked.

APPLYING THIS

SC 1.2.8 is Level AAA, which means it's not required for baseline WCAG 2.1 or 2.2 AA conformance. But that framing can mislead teams about when it matters. Government contracts, procurement requirements, and organizational accessibility policies increasingly reference AAA criteria selectively. If your organization has committed to AAA conformance on any content, or if you're producing content for DeafBlind audiences specifically, this criterion is active for you.

Automated testing will not catch failures here. A tool can detect whether a <video> element has an associated text alternative — it cannot evaluate whether that alternative is complete. Our research on automated vs. manual testing found that automated tools catch at most 37% of accessibility barriers in comprehensive audits. Media alternatives are squarely in the 63% that requires human judgment.

In design: Build the media alternative into the content production workflow, not as a post-production task. When scripts are written, the foundation for a media alternative exists. Augmenting a script with visual descriptions during production is significantly less expensive than reconstructing visual context after the fact.

In development: Use H53 (opens in new window) (the <object> element body technique) where applicable, or implement a clearly labeled, adjacent link using G58 (opens in new window). The link text should be descriptive — "Read the full text alternative for [video title]" — not "transcript" (which implies dialogue only) and not "accessibility version" (which is vague).

In QA: Manual review against a checklist that maps each segment of the video to the corresponding text alternative section. Spot-check non-dialogue moments specifically — those are where alternatives most commonly fail.

For organizations navigating multiple overlapping standards frameworks, The Compliance Framework Paradox is worth reading alongside this criterion — the overlap between 1.2.3, 1.2.5, and 1.2.8 is exactly the kind of multi-standard complexity that creates organizational confusion about what's actually required.

COMMUNITY PERSPECTIVE

From a community lens, SC 1.2.8 is a clarity test for organizational commitment to DeafBlind inclusion. The criterion sits at AAA precisely because it requires more effort — but that effort reflects the real cost of reaching a population that every other multimedia technique leaves behind. Organizations that treat AAA as optional-by-default are making an implicit choice about which disabled users matter. That choice has consequences: for DeafBlind users who encounter inaccessible video content, exclusion isn't inconvenient — it's complete. The Southeast ADA Center (opens in new window) consistently emphasizes that meaningful accessibility means reaching the full range of disability experience, not just the majority. SC 1.2.8 is where that principle becomes concrete.