Listen to each channel deliberately
A two-channel asset can contain a spatial stereo mix, two microphone feeds or different languages. Do not decide its meaning from channel count alone. For media you own, listen to a known passage from each channel separately and preserve the source's production notes.
XMLTV's audio metadata describes bilingual material as separate languages in the two channels of one stream. That is a useful metadata concept, not proof that an arbitrary two-channel file has that structure. Verify the actual content before selecting a processing path.
If the channels carry different languages, playing them together can make both harder to follow. A standard spatial downmix task is therefore the wrong starting assumption.
Choose the required language assets
Decide whether the destination needs separate language tracks, one selected language or a documented bilingual presentation. Keep labels and source-channel associations explicit. Copying one channel for a chosen mono language may be intentional, but it must be based on verified content.
Imagine an owned archive with one narration language on the left and another on the right. Combining both at equal level would not create a useful translation. A production handoff should instead identify each language and produce the agreed delivery assets.
Review the resulting language versions for completeness, level and timing. Preserve the original two-channel source and record any mapping. A successful export with two sounding channels does not establish that the audience received the intended language choice.
Questions answered
Does every two-channel recording contain stereo space?
No. Identify its actual channel roles.
Should two different narration languages be summed into mono?
Usually that obscures both. Choose an explicit language-delivery plan.
Sources & further reading
Official references for the subjects discussed. Service terms and device software can change.