How Can You Test Screen-Reader Labels on a Chart?
From the earlier reference collection. Read the current software blog.
Verify that a chart communicates its measures, values, and trend without requiring a reader to interpret its visual marks.
Start with the chart’s purpose
Write down what a reader should learn from the chart: the measure, its unit, the time or category range, and the main pattern. Then check that those facts are available to someone using a screen reader. A label such as “Line chart” identifies a form, but says little about the data or why it matters.
Use a small fixture with known values. For example, a daily series of 12, 15, and 11 readings can support a summary that names the period, unit, and rise-then-fall pattern. Avoid relying on a visual screenshot as the only evidence of accessibility.
Inspect the accessible name and description
Check the accessibility tree or screen reader output to confirm that the chart has a meaningful name and a concise description. The name can identify the subject, while the description can state the range and important trend. Verify that the units are spoken or otherwise clearly associated with the values.
Be precise about what the chart summary claims. If the data show a local peak but no overall increase, the description should not call the series steadily rising. Test an all-equal or incomplete series too, so the description does not invent a trend when the values do not support one.
Make the underlying values reachable
A trend summary is useful, but it does not replace access to the plotted values. If the chart provides a data table, inspect it with keyboard navigation and a screen reader. Confirm that row and column headers identify each date or category and that values include units where necessary.
If users can focus individual marks, test that each point’s accessible text identifies its label and value. Check focus order and ensure that keyboard users can reach the data without moving through a long sequence of decorative controls. The exact interaction depends on the chart design, so test the implemented interface rather than assuming a particular chart library behaves correctly.
Test changes and nonvisual use
Repeat the checks after changing a filter or selecting a new period. The accessible name, summary, and table should update with the visible chart; stale text can mislead even when the drawing is correct. Also test empty and error states so the user receives a clear explanation when no points are available.
A practical acceptance check is to navigate to the chart without looking at the screen and answer three questions: what is measured, what values are available, and what broad pattern appears? If the interface supports those answers without guesswork, its labels and data access are doing useful work.