Cognitive psychology will suggest that both reading and listening ultimately sets off the same understanding in your brain anyways. The question is then why listening to programming code is so different to listening to natural languages. If you look at the nature of listening and seeing, the former is immediate (you have to hear the sound exactly when it is generated, with no memory short of echoes) while the latter is not equally time-constrained and allows revisiting – it can be more of an external memory.

That suggests that we actually DON’T build internal models of the code that are quite incomplete, and only work on parts while the rest is stored and accessed externally.

And yes, quite different from natural language comprehension! It actually implies easy, repeated access to the code is essential, and that readability is even more important for code than natural language. Do you agree with my reasoning?

But you raise a good research question! One can build a good argument for either case – and even if it to me seems obvious that we read code in a similar way as natural language text, I can’t know that for sure. Well-designed experiments and research could help sort that out, until then it’s only going to be speculation. My discussion above is more a minor exploration or a starting point to think more about it, reading more into it is going further than the data can support.

Computer languages express ideas differently and so comprehension would require different efforts. Using readibility tools for natural languages on computer languages seems dubious, at best.