Did you know that a school graph can have an excellent title and still leave a reader unable to tell what it says? Imagine a student using a screen reader to examine a line graph. The image announces only ‘chart’. The reader cannot hear the values, the units or the turning point. The scientific reasoning might be perfectly good, but the evidence has been locked inside a picture.
Alternative text (alt text) is a text alternative for an image. For a complex graph, a short alt description should identify the graphic and its key message, while a nearby full text description or accessible data table preserves the detailed values and relationships. This two-part approach follows the World Wide Web Consortium’s Web Accessibility Initiative guidance for complex images. The immediate answer is not to squeeze every data point into an enormous alt-text sentence: give the graphic a concise identity, and make the underlying information available in readable text.
Why a caption alone does not solve the problem
A caption tells a sighted reader what a figure represents; it may appear below a picture and might describe the main conclusion. An alt text conveys the image’s information or purpose to readers who cannot use the image visually. A data table gives access to exact values. They can support one another, but they are not identical jobs.
Suppose the caption says ‘Reading progress across three months’. That is too vague for a reader who needs to know which groups are compared, the time units and whether one group overtook the other. If the image is the only place where those values appear, the report is harder to check, quote or reuse even for people who can see it.
Worked example: describe a fictional reading-time chart without losing the values
The following observations are invented solely for teaching. They are not measured pupil performance or a claim about any real school. Imagine a line graph whose horizontal axis lists May, June and July, and whose vertical axis lists minutes of reading per week. Two sample groups, A and B, have these values. Each plotted point represents a weekly total recorded for the named example month; the figure does not claim to represent a monthly average.
| Month | Group A (minutes) | Group B (minutes) |
|---|---|---|
| May | 120 | 100 |
| June | 140 | 125 |
| July | 130 | 160 |
A poor alt text is: ‘Colourful line graph with two lines’. Colour and the presence of lines do not tell the reader what is being compared. A better short alternative is: ‘Line graph of weekly reading minutes for two study groups, May to July; Group A peaks in June, while Group B rises throughout and leads in July. Full values appear in the adjacent table.’ This names the chart, its units and timescale, the main pattern and where to find exact numbers.
The full text description can add: ‘In May, Group A records 120 minutes and Group B 100. In June they record 140 and 125. In July Group A records 130 and Group B 160.’ The table then allows a reader to inspect any individual value or compare values independently. Do not put a conclusion such as ‘Group B studied harder’ into the alt text; these are reading-minute totals, not a measure of effort or comprehension.
Hand-check the calculations before describing the trend
- Group A from May to June: 140 − 120 = +20 minutes per week.
- Group A from June to July: 130 − 140 = −10 minutes per week.
- Group B from May to June: 125 − 100 = +25 minutes per week.
- Group B from June to July: 160 − 125 = +35 minutes per week.
- July group difference: 160 − 130 = 30 minutes per week in Group B’s favour.
Check the direction by comparing each pair of neighbouring table cells. The proposed alt text says Group A peaks in June, which is true because 140 is greater than 120 and 130. It says Group B rises throughout, which is true because 100 < 125 < 160. It says Group B leads in July, which is true because 160 > 130. The conclusion is now supported by an independently readable data source.
A graph’s job and its accessibility description should match
For a simple icon or purely decorative shape, the appropriate alternative may be an empty alt attribute, so assistive technology can skip decoration. But a graph used to support an argument is informative. A student should not label a meaningful chart as decorative merely to avoid writing a description.
For a simple visual with just one takeaway, concise alt text may be enough. For a detailed graph, map, flowchart or scientific diagram, provide more than one brief sentence if the reader needs exact data, relationships, direction arrows or scale markings. The W3C explains that complex images require a short identifier and a more complete text equivalent, which can be a structured description or table. Its broader images tutorial also distinguishes informative, functional and decorative imagery.
Diagnostic: what is missing from the student’s description?
Ask the student to compare two descriptions of our example: (1) ‘Students read more’. (2) ‘Group B’s weekly recorded reading minutes rose from 100 in May to 160 in July; Group A rose from 120 to 140 in June, then declined to 130 in July.’ The first has no stated group, baseline, unit or dates. The second gives enough information to reconstruct the trend, but would still benefit from an accessible table for all individual points.
A second diagnostic uses the fictional claim ‘Both groups improved every month.’ The table disproves it: Group A’s total fell from June to July. If a learner repeats the erroneous claim in an image description, accessibility text does not repair the reasoning. The same evidence-checking standards apply to both image and words.
An accessible routine for students submitting a project
- Name the purpose: what question is this chart supposed to answer? Decide whether exact values matter.
- Check the plotted evidence: confirm each value, axis label, unit, timeframe and legend against the original data.
- Write one concise alt description: say what kind of chart it is, what is compared, and the essential finding. Point to the fuller description where appropriate.
- Add a nearby text alternative: a clear paragraph or properly structured table with headers and the values necessary to understand the graph.
- Do not encode meaning by colour alone: name series clearly in the description, and use labels, markers or other distinctions where practical.
- Try an independent read: hide the image and ask another learner whether they can answer the original question using only the text.
Progressive guided-to-independent practice
Guided: improve an unusable sentence
Start with ‘A chart showing results’. Prompt the learner to add the topic, unit, groups and time. A useful first revision is ‘Line graph comparing weekly reading minutes for Groups A and B from May to July’. Then ask for the key change, and finally for the location of the full values.
Supported: check a one-line description against data
Give the student three fictional percentages: 10%, 12%, 11% of participants in successive weeks. The description ‘a steadily rising percentage’ is false: the third figure falls by one percentage point. A defensible description says the percentage rose from 10% to 12%, then slipped to 11%. The accessible text must not smooth away a reversal just to make the visual sound encouraging.
Independent: write the alternative without seeing the figure
Use a new fictional bar chart: Library visits by Year Group X are 18 on Monday, 22 on Tuesday and 14 on Wednesday. Ask the student to create a short alt description and a three-row text table from the figures alone. Self-check: Tuesday is the highest at 22, Wednesday is lowest at 14, and the units are visits, not visitors unless the data actually records unique people. The daily range is 22 − 14 = 8 visits.
Parents, teachers and observable progress
For a parent reviewing a project, a useful question is: ‘If I could not see this image, what information would I lose?’ A teacher can make the same test by handing the table and description to another pupil who has not viewed the graph. Can that pupil identify the highest and lowest values, the units and a reversal in direction? A learner who can do this consistently has moved beyond merely naming chart types.
Track improvement over two attempts: record whether the learner names the subject, units, series, timescale and trend accurately, and whether a complete data alternative exists. Do not make a long alt text a target in itself. The goal is to give a reader the necessary meaning with as little avoidable friction as possible.
Limitations: accessible text is not a licence to overstate
The example data do not establish why reading time changed, whether reading skill improved, or which activity caused an increase. If the chart shows estimates, uncertainty or missing observations, those qualifications also matter in the text equivalent. Different publishing software handles tables, image descriptions and screen readers differently, so the author should check the actual output rather than relying only on an editor’s preview.
Frequently asked questions
Should alt text repeat every cell in the table?
Not usually when the full table is clearly available nearby. The short alternative identifies the graphic and its main meaning; the table supplies detailed values. Make sure the table is accessible and its relation to the chart is clear.
Does a caption automatically become alt text?
No. Some tools display captions but still require a separate image alternative. Inspect what the publishing tool actually stores and what assistive technology receives.
Can I describe a rise as ‘successful’?
Only if you have an appropriate, explained success criterion and supporting evidence. Rising values alone are not necessarily good, and an image description should avoid invented causes or evaluations.
What if the school requires a screenshot of a graph?
Keep the screenshot if requested, and supply the accessible description or table as a permitted companion. Follow the submission format while ensuring the evidence remains understandable to readers who cannot rely on the image.
Further reading and learning transfer
The companion eduKateSG guide How to Compare Two Charts When Their Axes Use Different Units helps students check the units before writing descriptions. How to Read a Chart Rebased to 100 Without Confusing Growth with Size adds another important check for visual claims. For general image distinctions, see Making School Images Accessible with Alt Text.
Explore subject routes at the eduKateSG Learning Hubs and see the diagnosis-to-independent-check approach in the Clementi Secondary 1 Mathematics reference. The educational aim is not simply compliance: it is to make the learner’s evidence understandable, testable and available to everyone.