---
name: Accessibility grouping
slug: accessibility-grouping
category: accessibility
status: published
created: 2026-08-21T00:00:00.000Z
modified: 2026-08-21T00:00:00.000Z
definition: Collapsing a cluster of elements, an icon plus a label plus a value,
  into one thing assistive technology stops at, instead of three stops that mean
  nothing apart.
aliases:
  - name: shouldGroupAccessibilityChildren
    source: uikit
  - name: merge semantics
    source: flutter
  - name: importantForAccessibility
    source: android
  - name: accessibility element
    source: uikit
  - name: screen reader focusable
    source: android
tags:
  - assistive-tech
relations:
  contrastWith:
    - presentational-children
    - group-label
    - owned-element
  variantOf: []
  partOf: []
  seeAlso: []
implementations: []
sources:
  - title: Semantics in Compose, Android Developers
    url: https://developer.android.com/develop/ui/compose/accessibility/semantics
  - title: View.ImportantForAccessibility, Microsoft Learn
    url: https://learn.microsoft.com/dotnet/api/android.views.view.importantforaccessibility
  - title: Accessibility, React Native
    url: https://reactnative.dev/docs/accessibility
demo: inline
exhibit: false
useWhen: a card reads as five fragments instead of one item
---

A [screen reader](/screen-reader) user moves through a screen one element at a time, and
the unit they move by is the accessibility element rather than the view. A comment row
drawn as an avatar, a name, a role and a timestamp is four views, and left alone it is
four stops: a pair of initials, a name, a job word, and a time with nothing attached to
it. Grouped, it is one stop whose announcement carries all four pieces in the order they
are read, which is the same row costing a quarter of the gestures and meaning four times
as much.

Every platform gives you the switch, under a different name. Compose merges a subtree with
`Modifier.semantics(mergeDescendants = true) {}`, and its `clickable` and `toggleable`
modifiers already do it, which is why a tappable card usually behaves correctly before
anyone thinks about it. The Android view system sets `importantForAccessibility` on the
node, whose values are `IMPORTANT_FOR_ACCESSIBILITY_AUTO`, `YES`, `NO` and
`NO_HIDE_DESCENDANTS`, where important means, in the platform's own words, that the view
fires accessibility events and is reported to accessibility services that query the
screen. React Native builds the label for you when you do not supply one: on a touchable,
it is constructed by concatenating all Text node children separated by spaces. The
composed string is still an [accessible name](/accessible-name), so the same rules apply
to it, including the order it comes out in and whether that order matches
[reading order](/reading-order).

This is the screen-reader analogue of a [tab stop](/tab-stop). A tab stop is one position
the Tab key reaches even when the widget standing there holds many focusable parts; a
group is that same compression applied to the screen reader's own cursor, which visits far
more than Tab ever does, since it stops on text and images too. The two are set
independently and can disagree: a card can be a single tab stop and still be four swipes.

The cost is that grouping throws the internal structure away. A summary row is exactly
right for it, because there is nothing inside worth reaching separately. A row with an
avatar that opens a profile, a title that opens the item, and a menu button that opens
five commands is exactly wrong for it, because collapsing it hides two of the three
actions from anyone who cannot see them to tap them. Grouping answers how many elements
there are; its round mate, [accessibility trait](/accessibility-trait), answers what each
one is announced as, and a card with the right traits on every child can still be four
stops that mean nothing apart.