Thanks, I read the page but am not sure to understand how it relates to the original question. Can you decode ? Does it explain why creating labels/aliases with code 'mul' or 'mis' fails using the API ?
Sorry, I wrote "when the language value is unknown", but this was not precise : the original data we want to synch already contains data using code "mis" (as a valid ISO639-2 iso code, see
https://www.loc.gov/standards/iso639-2/php/code_list.php), and as I read it was supported by Wikidata, I attempted to use that code directly through the API, with no luck. Similarly, original data uses code 'mul' (also a valid ISO639-2 code).