The C compiler I wrote in the 80's could produce error messages in several languages. I was pretty proud of it. But nobody used that capability, they preferred the English messages. A huge problem was finding someone bilingual in English software jargon and X software jargon.
So yeah, one spoken language for programming is more efficient.
Also, the D programming language accepts foreign language identifiers. Does anyone use that feature? I'm sure someone will say they do, but they are a very tiny, tiny minority.
https://nationalpost.com/news/canada/official-bilingualism-c...
There are more important things like producing documents only once, and presenting ideas only once. I went to a fully bilingual university, and every brochure and every sign and every everything was produced and printed twice.
Then there are the huge things: right now there are 7000 languages, so any given person only gets to enjoy a tiny fraction of the cultural product produced by those 7000 speakers.
There'd be a massive gain in the efficiency of consumption if some random shepherd in Nepal had a blog that some random doctor in Mexico could read natively and be inspired.
Or if some kid rapping in Bucharest about his problems were natively understandable to some kid listening in Brazil.