Applies to: A track or text record is labelled und or otherwise described as unknown.

The useful distinctions

  • The BCP 47 subtag und denotes undetermined language.
  • Unknown language is different from known absence of linguistic content.
  • A title’s language or a country association does not prove the language of a track.

Keep uncertainty in the record

An unknown label can be inconvenient, but assigning a likely language hides the information gap instead of resolving it. The viewer may recognise the programme’s production origin or understand its translated title while still having no basis for naming the track’s language. Preserve the supplied label and identify the component it describes. A source can correct incomplete metadata more effectively when the original uncertainty has not been replaced by an unsupported guess.

Use observations with clear limits

If the viewer confidently understands a short passage, that observation can help, but it should be attached to the exact edition and segment. If they do not recognise the speech, say so. A silent scene does not prove that the whole track has no linguistic content. Likewise, readable menu text does not identify the programme’s spoken language. Keep those limits visible while asking for the intended language information.

Hypothetical worked example

A fictional audio option on edition UL-5 is labelled und. The viewer records that label and notes that they cannot confidently identify the speech near 06:20. They ask the edition owner for the intended audio-language description. They do not relabel the option from the listing’s flag, call it silent because the opening has no dialogue or assume that und names a specific spoken language.

A focused way to proceed

  1. Keep the unknown label exactly as supplied.
  2. Identify whether it concerns spoken audio, text or another component.
  3. Record only observations the viewer can confidently make and ask the source to clarify the label.

When to stop or ask for help

Request authoritative component-language metadata while retaining the original unknown label and bounded observations.