Did you know? A beautiful school graph can hide its actual lesson from a reader who cannot see the image clearly. A pupil may spend an hour choosing colours for a science fair poster and only a minute writing ‘graph’ in the image description. That one word does not tell the reader what was measured, what changed or why the result matters.
Accessible graph text gives a reader the graph’s essential information without relying on sight. For a simple chart, name the kind of chart, the measured quantities and units, and the main comparison. For a complex or data-rich chart, add a separate readable description or data table. This is a practical writing and reasoning skill: pupils must understand the graph before they can explain it to someone else.
The World Wide Web Consortium Web Accessibility Initiative guide to complex images distinguishes a short text alternative from a longer equivalent description for charts, diagrams and other substantial visual information. Its images tutorial also explains that the useful alternative depends on the image’s purpose. The made-up school examples below are teaching illustrations, not observed pupil results.
A chart is a message made from data, not merely a decoration
Imagine a science report about how a toy car travels on three different ramp surfaces. The graph has an x-axis showing surface type and a y-axis showing distance travelled in centimetres. A visually bold chart might use bright coloured bars. Yet ‘red bar, blue bar, green bar’ tells a reader almost nothing unless the colours are explicitly linked to surfaces and numerical measurements.
The writer’s first job is to identify the question the graph answers. It might be: ‘On which surface did the model car travel the farthest under the conditions of this class test?’ If the chart does not answer a question the writer can state plainly, the issue may be the chart design itself, not merely a missing description.
Three levels of explanation: image name, takeaway and full values
- Image identification: ‘Bar chart of a toy car’s distance on three ramp surfaces, measured in centimetres.’ This says what the picture contains but does not yet convey the results.
- Short takeaway: ‘The car travelled farthest on smooth card and least far on felt.’ This gives the headline comparison when it is supported by the figures.
- Full equivalent: write the surface names and measured distances in accessible text, including the measurement units and enough context to interpret them.
Do not force a paragraph of every possible detail into one tiny alt attribute. In a school website or presentation, a short alternative can point to a nearby section headed ‘Graph data and description’. That section can provide readable values and an interpretation. For a printed poster, an accompanying explanatory paragraph or accessible digital version can serve the same goal.
A complete, hand-checked example: a class borrowing chart
Consider this fictional bar chart, used to make the arithmetic transparent. It shows how many books were borrowed from a classroom library on five consecutive school days: Monday 12, Tuesday 16, Wednesday 9, Thursday 18, Friday 15. Each number counts borrowing transactions in the imaginary example; it does not claim that a pupil finished reading a book, that all borrowers are unique or that these are actual school statistics.
Check the total before describing it: 12 + 16 = 28; 28 + 9 = 37; 37 + 18 = 55; 55 + 15 = 70 borrowing transactions. Thursday has the largest daily count, 18. Wednesday has the smallest, 9. The difference between those days is 18 − 9 = 9 transactions. The mean per school day is 70 ÷ 5 = 14 transactions. None of these sums needs a visual bar colour to be understood.
A weak alt text would say, ‘Graph of books’. A slightly better one would say, ‘Five bars, with the tallest on Thursday.’ A useful short version could say: ‘Bar chart of classroom library borrowing, Monday to Friday: Thursday highest at 18 transactions and Wednesday lowest at 9. Full daily values are provided below.’
The corresponding visible description could say: ‘The chart plots books borrowed per day, not books read or unique readers. The daily counts are Monday 12, Tuesday 16, Wednesday 9, Thursday 18 and Friday 15 transactions, for a five-day total of 70. Thursday exceeds Wednesday by 9.’ A screen-reader user, a pupil viewing a small phone and a parent skimming printed notes can now retrieve the same core message.
Why the distinction between borrowing and reading matters
Suppose a pupil writes ‘Thursday was the best reading day.’ The chart shows borrowing counts, not the number of pages read, reading time, comprehension or enjoyment. Calling the highest bar the ‘best reading day’ turns a measured quantity into an unmeasured judgment. Accessible description should be just as faithful to the evidence as a well-written science conclusion.
Other tempting substitutions are equally unsafe. The chart does not show that 70 different pupils borrowed books; the same person may borrow more than once. It does not prove that a library event caused the Thursday rise. It does not show whether Friday’s lower number reflects a shorter opening time. Those would require additional records or a different study design.
What the axes and units must tell the reader
Before writing the description, inspect the title, x-axis, y-axis, units, series labels, scale and any missing values. A bar chart with ‘Number of books borrowed’ on the y-axis communicates a count; one labelled ‘Proportion of pupils borrowing’ would require a denominator. If the vertical scale is truncated, do not imply that bar-height differences start from zero when they do not.
If a chart uses two series—for example, fiction and nonfiction books—name them and explain which comparisons matter. If the y-axis is in kilograms, write kilograms rather than silently converting to grams. If a time series includes a missing Wednesday value, say ‘Wednesday not recorded’, not ‘Wednesday zero’. The text description must be as careful as the data collection.
Turn the skill into a learner diagnostic
Hand students a simple graph without showing its author’s paragraph. Ask them to answer four things: What was measured? In what unit? Which result was highest or lowest? What conclusion would go beyond the evidence? Their answers reveal whether they are merely reading visual shapes or actually interpreting data.
Consider a student who correctly identifies the tallest bar but calls the unit ‘pupils’ when the axis says ‘transactions’. That is a unit-reading error. Another might copy all five numbers accurately but state that more borrowing means better comprehension. That is an inference error. A third writes ‘chart’ despite knowing all the values; that is a communication error. Each diagnosis suggests a different next lesson.
Guided practice: describe the results, not the appearance
Step 1: recover labels and numbers
Give the learner a fictional science chart showing plant heights at the end of a week: Pot A 11 cm, B 14 cm, C 13 cm. Ask for a plain sentence: ‘A bar chart shows end-of-week plant height in centimetres: A 11 cm, B 14 cm, C 13 cm.’ No growth-rate claim is possible without starting heights.
Step 2: write the essential comparison
Now ask for one evidence-based addition: ‘B was tallest at 14 cm; A was shortest at 11 cm; the difference is 3 cm.’ The calculation is 14 − 11 = 3. If a pupil writes ‘B grew fastest’, stop and identify the missing starting measurements. This is the moment that turns accessible writing into scientific reasoning.
Step 3: make a short alternative and a long description
For the full chart, the learner drafts one concise image alternative that names the visual and its central finding. Next, they write a nearby paragraph with all three values and units. They check the two pieces do not contradict each other and that a reader can locate the full description.
Step 4: transfer without the model answer
Give the learner a different chart: for example, a line graph of a beaker’s temperature at several timed observations. Ask them to note the axis units, initial and final values, any missing interval and whether the line proves a mechanism. The assignment is complete when another pupil can answer the graph question from the description alone.
A realistic classroom workflow for accessible graphs
- Before graphing: keep the data table, original labels and units. Do not discard records after drawing the chart.
- When designing: use legible labels and patterns or direct labels where colour differences alone would be insufficient.
- When describing: name the measured quantities, state the central result and add numerical detail in a long description or table when needed.
- When checking: ask a classmate to interpret the written description without viewing the chart, then compare their understanding with the actual graph.
- When revising: correct unsupported verbs such as ‘proved’, ’caused’ and ‘improved’ unless the study actually supports them.
The W3C guidance is about making meaningful content accessible, not merely ticking an alt-text checkbox. A very small decorative design may appropriately have no informational alternative. A chart conveying evidence has a different purpose: withholding its data in a visual-only form leaves an important part of the lesson unavailable.
Parents and teachers: measure what improved
For parents: ask the child to describe a chart while the image is hidden. If you can reconstruct the key result and units, the explanation is doing its work. Ask, ‘What could we not know from this graph?’ to build restraint, not anxiety.
For teachers: assess a first and second attempt using observable criteria: accurate labels, accurate units, preserved key figures, clear takeaway, location of complete values, and absence of unsupported inference. A learner who fixes an incorrect unit has made specific progress even if the resulting sentence is not elegant.
For learners: read a short alt description aloud. Does it make sense without saying ‘this’ or ‘that’? Does it distinguish ‘borrowed’ from ‘read’? Does it reveal any assumptions? When the text faithfully communicates the evidence, it is functioning as an independent explanation, not an afterthought.
Limitations: an alternative cannot repair a misleading chart by itself
A readable paragraph cannot rescue a chart built on wrong numbers. Double-check the data and arithmetic first. When values are approximate, say so. When graphs are complex or interactive, a short alt line may be insufficient; provide detailed explanations, accessible tables or another usable view suited to the audience and technology.
One more boundary matters: this guide does not claim that every school website will become fully accessible by adding a sentence. Accessibility includes navigation, keyboard operation, structure, contrast and more. We are focusing on a precise, teachable contribution: conveying graphical information without requiring the visual image.
Frequently asked questions
Is ‘graph showing results’ good enough?
Usually not for a graph that carries important information. It does not identify the values or the conclusion, and a reader cannot use it to answer the graph’s question.
Should I put all thirty data points in the alt text?
For a complex graph, keep a concise identification and direct readers to a properly structured, accessible data table or longer description. The full text equivalent should convey the essential information rather than forcing an unwieldy caption.
Do I need to describe every colour and grid line?
Describe visual features when they carry meaning, such as a colour coding that distinguishes series. Decorative grid lines are normally less important than labels, values, data relationships and the point of the chart.
Can I say the chart ‘proves’ the cause of a change?
Only when the evidence and study design warrant that strength of claim. A time trend alone often cannot separate alternative causes. In most classroom charts, ‘shows’, ‘records’ or ‘is consistent with’ is safer.
What if the graph includes a missing value?
Identify it as missing, not zero, and do not interpolate a replacement without clearly saying that it is an estimate and explaining the method.
Keep the learning connected
An accessible description is a small act of intellectual honesty. The child learns to identify what is measured, preserve the meaning, show the numbers and explain uncertainty. Related practice includes checking charts with hidden spreadsheet rows and distinguishing summary from evaluation.
Find more practical examples in the Core Aim of Education learning hub and eduKateSG Education Hub. The immutable Clementi learning reference shows the broader diagnosis-first habit: explain, practise, independently check and transfer. The goal is not more words about a picture. The goal is a reader who can understand its evidence.