Captions Without Content: The F8 Failure That Erases Deaf Users

A caption track exists. The video plays. And a deaf user still misses the punchline, the warning, the plot twist — because the captions left it out. This is the failure pattern documented in WCAG 2.2 Failure F8 (opens in new window): captions that exist but omit dialogue or important sound effects. The presence of a caption file is not the same as accessible captions. That distinction carries real legal weight.

The Failure

Failure F8 addresses a specific violation of Success Criterion 1.2.2 Captions (Prerecorded) (opens in new window), which requires synchronized captions for all prerecorded audio content in synchronized media. The failure occurs when caption text omits material content — either dialogue or non-speech audio that carries meaning.

The W3C's own description is precise: captions that lack "all of the dialogue (either verbatim or in essence) as well as all important sounds" are not real captions. The document gives two concrete failure patterns:

// Failure Pattern 1:
Caption text contains dialogue but omits
important sound effects entirely.
// Failure Pattern 2:
Caption text omits dialogue during
portions of the material — gaps in coverage.

One important clarification from the source: simplifying spoken text for readability is acceptable and standard practice. Captions do not need to be verbatim transcripts. What they cannot do is disappear — drop a warning alarm, skip a character's line, or go silent during a scene where audio carries critical information.

This failure applies to all technologies. There is no platform exemption.

Why This Matters

For a compliance officer reviewing a video library, incomplete captions are easy to miss in a checklist review. The file is there. The track is enabled. The automated scan returns no error. But the content is wrong — and no automated tool catches that.

This is precisely the gap documented in our research on automated versus manual testing: automated tools detect the presence of a caption track, not its accuracy or completeness. The compliance officer who relies on automated scans alone will certify as accessible a video that fails a deaf user at every critical moment.

Consider what incomplete captions mean in practice. A training video for a government agency where the safety instructions appear only in audio — no caption. A university lecture where the professor's aside about the exam is not captioned. A public health video where the urgent alert tone is never identified. In each case, the deaf or hard-of-hearing user receives a degraded version of the content that sighted, hearing users access in full. That is not equal access. It is the structural exclusion that Title II of the ADA (opens in new window) and Section 504 of the Rehabilitation Act are designed to prevent.

The legal exposure here is not hypothetical. DOJ enforcement and private litigation have repeatedly addressed caption quality — not just caption presence — as a compliance requirement. Settlement agreements in this space often mandate quality review processes, not just caption file submission. As our analysis of post-settlement compliance patterns shows, organizations that treat captioning as a checkbox frequently return to litigation because they never built the internal processes to ensure caption quality.

The Fix

The corrected pattern requires that captions account for all meaningful audio content — speech and non-speech alike.

// Incomplete caption (Failure F8):
[00:00:14] "We need to evacuate now."
[00:00:22] "The door is on the left."
// Missing: alarm sound at 00:00:10 that
// triggers the dialogue — context is lost.
// Compliant caption:
[00:00:10] [Alarm blaring]
[00:00:14] "We need to evacuate now."
[00:00:22] "The door is on the left."

Sound effects that carry narrative or instructional meaning — alarms, music that signals tone, crowd noise that establishes context, sound cues that explain a character's reaction — must be identified in brackets. Dialogue must be present for all speakers throughout the full duration of the content.

SC 1.2.2 does not require perfection in every word. It requires completeness in meaning. The standard is whether a deaf user receives equivalent information to a hearing user. That is the test.

Applying This

For development and content teams, F8 catches are almost entirely a human review problem. Here is what a practical quality process looks like:

In code review: Caption files (.vtt, .srt) can be reviewed for obvious gaps — long silent periods in the timestamp sequence that do not correspond to intentional pauses in the audio. If a 90-second video has a 40-second gap in caption timestamps, that is a flag worth investigating.

In QA testing: Watch the video with audio muted and captions enabled. Ask: does the captioned experience convey the same information as the full audio experience? If a sound effect matters to understanding the content, it must appear in the caption track.

In procurement: Caption quality standards should appear in vendor contracts. "Captions provided" is not sufficient contract language. Specify that captions must include all dialogue and meaningful non-speech audio per SC 1.2.2.

In content workflows: Auto-generated captions from platforms like YouTube or Teams are a starting point, not a finished product. They routinely miss speaker identification, sound effects, and accuracy in specialized vocabulary. A review step before publication is not optional for compliant content.

The compliance framework research we've published documents how organizations often mistake process artifacts — caption files, accessibility statements, audit reports — for actual compliance. F8 is a clean example of that gap: the artifact exists, the compliance does not.

Reference Table: SC 1.2.2 Compliance Requirements

CORS Perspective

From a risk and legal priority lens, F8 sits in a deceptive compliance zone: it is the kind of failure that passes automated scans, survives surface-level audits, and still constitutes a clear legal violation under SC 1.2.2. For government entities subject to Title II and the DOJ's 2024 web accessibility rule, caption quality is not a soft expectation — it is an enforceable standard. The strategic risk is compounded by scale: organizations with large video libraries may have hundreds of non-compliant caption tracks, each one a discrete violation. The community impact falls most directly on deaf and hard-of-hearing users who have no alternative access path when captions fail. Building caption quality review into content workflows — not just caption file generation — is the operational fix that prevents this failure from accumulating at scale.