I've been following this with some interest and I think I'm beginning to see how this is like nesting dolls. In the kernel is the actual interview in the native language - the primary source(s) which is actually a video or audio recording. The next layer is the interviewers transcription layer - a secondary source that may or may not compare & contrast various primary sources to create a collective composite which others who can read the native language are free to comment, modify, dispute, etc.

That's great - don't monkey with success. So far all that has been done is to convert one or more oral sources into a written condensation that is now open to all native readers (or listeners if you use text to speech technology) and that information is coherent within the context of that culture.

Where all this seems to fall down is that some insist on forcing? this material into English so a researcher will not be inconvenienced with the task of learning the native language. That doesn't make much sense, as the ethnographer or other researcher needs to understand that culture enough to access the primary source and understand the cultural context - which means they must know the native language anyway.

Work based on this cultural collection of material that will ultimately be published in English should be written by this researcher or a translator who can make it make sense in English. Running the original material through a mechanical translator is fraught with errors and misunderstandings.