UX Glossary · S
Screen reader
Also: Assistive technology
Software that converts interface content to speech or braille, navigated largely by keyboard.
A screen reader announces interface content as synthesised speech or sends it to a braille display. The common ones are JAWS and NVDA on Windows, VoiceOver on macOS and iOS, and TalkBack on Android.
How people actually use them
Not by listening to a page top to bottom. Experienced users navigate structurally: jumping between headings, pulling up a list of links, moving through form fields, skipping to landmarks. Speech is often set far faster than a sighted person finds comprehensible.
This is why document structure matters more than any single accessibility fix. Headings in a logical order are the primary navigation mechanism, and a page where everything is styled text with no real headings is one long undifferentiated wall.
What breaks it
Non-semantic markup, where a div styled as a button is announced as nothing. Images with
missing or useless alt text. Form fields with no programmatic label, so the user hears
“edit, blank”. Content injected without notification, so a result appears silently. And
ARIA applied incorrectly, which is frequently worse than no ARIA at all.
The most useful thing a designer can do
Specify heading levels, label every field, and say what each image is for. Semantic structure is a design decision expressed in markup, not an engineering afterthought.
In practice
Turn one on for ten minutes. It is uncomfortable and it changes how you specify things permanently.