There's two levels here: 1. The screenreader transforms the visual structure into something understandable for the human behind it 2. The human tries to make

There's two levels here: 1. The screenreader transforms the visual structure into something understandable for the human behind it 2. The human tries to make sense of it and maintain a mental model of what the screenreader is presenting to them Semantics are key for #1, but #2 can still lose track of what's happening on a complex form. For that, it helps if the human can keep trying, and fix whatever they missed on the next pass.