Assistive technology

Definition

Assistive technology is the collective name for hardware and software people use to perceive and operate interfaces: screen readers that speak or braille the page, magnifiers that enlarge a region of it, voice control that clicks by name, switch devices that scan through controls, and eye trackers that point without hands.

Wider than screen readers

Web writing tends to use "screen reader" as shorthand for all of it, and the shorthand distorts decisions. A voice control user needs visible labels that match accessible names, or the command "click Send" finds nothing. A switch user needs every control reachable in a sensible sequence, because scanning through them one by one makes wasted stops expensive. A magnifier user reads the page through a small moving window, which is why content that appears far from where the action happened goes unnoticed. Each tool stresses a different property of the page, and "works in a screen reader" covers only one of them.

The shared contract

What all of these tools have in common is what they read: the accessibility tree the browser computes from your markup. Native elements with real names, honest roles, and maintained states serve every tool at once, which is why the cheapest accessibility work is choosing the right element in the first place. No tool can repair a control the markup never described.

What this means for testing

You cannot test every tool, and you do not need to. Keyboard-only operation is the base contract most devices build on: if every task completes with Tab, Enter, Space, and the arrow keys, switch access and much of voice control inherit the result. One screen reader pass catches the naming and announcement gaps. The accessibility tree panel in browser developer tools verifies the contract all the tools share, in seconds, with nothing installed.

Common failures

• Visible button text that differs from the accessible name, breaking voice commands.
• Feedback rendered far from the control that triggered it, invisible under magnification.
• Custom controls that work by mouse and screen reader but never by switch, because focus cannot reach them.
• Testing reduced to one screen reader on one browser and generalized to everything.

Review question

Pick the page's most important task and ask which of these completes it: the keyboard alone, speech output alone, and voice commands matching the visible labels. Each answer exercises a different tool family, and the fixes are almost always in the markup, not in the tools.