Equivalence is not full replacement; some sound information is lost in transcription

What it is

The goal is usable parity, not item-by-item replacement. Sound expresses language, tone, spatial position, simultaneous events, and emotion at once; transcription covers only part of that. Acknowledging this lets effort go where comprehension is actually at stake instead of claiming a complete transcription.

Why it happens

The limitation is inherent to the encoding: text is discrete and sequential, sound is continuous and concurrent. Concurrency cannot stay parallel in text and must be serialized, imposing an order; spatial information—what came from which direction—has no coordinate in plain text; emotion and urgency can only be approximated at the level of wording. The loss is structural rather than a matter of poor execution. Equivalence design can compress the loss below the point where key judgments change, and can tell readers which dimensions are not represented.

Studying it

Gap analysis measures where equivalence actually ends: enumerate the information dimensions audio carries, mark which the transcription covers, and have text-only readers perform tasks that depend on those dimensions, logging failures. Measures include task success and which missing dimension each failure maps to. Comparing gains across increasing transcription depth shows whether more description is still in a useful range.

Where it stops holding

When a task depends only on linguistic content, equivalence is nearly complete—a plain dialogue transcript, for instance. When it depends on spatial audio, concurrent events, or emotional judgment, extra means (spatial cues, event description, prosody annotation) are needed just to approach parity. If those means cost more than they return, the right move is to define the scope honestly rather than claim equivalence.

Applying it

• State explicitly which sound dimensions are covered and which are not, rather than promising full equivalence.
• Cover first the dimensions that affect judgment and action, expressing space and concurrency in text or interface structure.
• Offer a route for uncovered dimensions, such as replaying the original audio or consulting an event log.
• Verification: have text-only readers complete tasks depending on space, concurrency, and emotion, log the failures against the dimension list, and confirm effort lands where impact is highest.

Related

• Within the group: D2.12.1 Captions carry the words but rarely restore tone or urgency · D2.12.2 Non-speech sound needs description, not only dialogue transcription
• Adjacent: D2.12.3 Caption delay destroys the timing cues that sound carried · J2.03 Coverage of captions and text alternatives
• Search terms: accessibility parity · information loss · media equivalence