# Sign language (/docs/reader-support/sign-language)



`videos.json` holds sign-language clips. Unlike the other sidecars, its keys are page indexes; use the section id embedded in the filename to match it to the current document’s `data-section-id`.

<PackageSnippet title="Sign-language map reference" path="content/i18n/en-US/videos.json" />

```ts
const sectionId = fileName
  .replace(/^sl_/, "")
  .replace(/\.(mp4|webm)$/, "");
```

Compare that id with the page you are displaying. There is no need to reproduce the exporter’s page indexing.

In V1, the keys are `video-<pageIndex>` and filenames use `sl_<sectionId>.mp4` or `sl_<sectionId>.webm`. Treat a missing key, an unexpected filename, or an unavailable asset as no clip for that page.

## Choose a reader-owned surface [#choose-a-reader-owned-surface]

Sign-language clips describe the current page context, so a picture-in-picture player is a good default. Close it when the reader navigates to a page with no clip. Glossary-term videos are separate: they arrive in the individual glossary entry rather than `videos.json`.

<Callout type="warn" title="Do not show a dead control">
  Gate the player on both `features.signLanguage` and a clip existing for the current section.
</Callout>