Did you know that a perfectly accurate subtitle can still leave a learner unable to answer a question about the lesson? Picture a science video: the teacher says, ‘Look at what changed,’ and the screen shows a line rising from 20 degrees to 35 degrees. Every spoken word appears in the captions. Yet a learner listening without the image—or reading a basic transcript—may not know what changed at all. The missing information was never spoken.
Captions are synchronised text for meaningful speech and audio in a video. Audio description communicates important visual information that the ordinary audio does not explain, while a descriptive transcript collects spoken, meaningful sound and relevant visual information in readable text. The immediate answer is that captions cannot automatically explain a silent diagram: the lesson needs its essential visual facts made available through spoken explanation, audio description and/or a suitable descriptive text alternative. The exact access requirements depend on the type of media and the relevant accessibility standard.
- Listen to the soundtrack without viewing the screen; record what a learner can genuinely learn from audio alone.
- Read the captions separately. Check meaningful speech and relevant sounds, not only the teacher’s main sentence.
- Identify silent diagrams, equations, labels, gestures or changes that carry essential knowledge.
- Supply those facts through naturally revised narration, separate audio description where appropriate, and a descriptive transcript.
- Keep invented example values clearly labelled and report actual data only when supported by the lesson.
- Ask a learner to answer an evidence-based question using each available format, then repair the missing information.
Three different jobs: speech, sound and sight
Captions primarily help a reader access speech and meaningful nonspeech audio while following the moving picture. Examples include ‘The solution is heated’ and ‘[glass clinks]’ when that sound matters to understanding. Translated subtitles may render dialogue in another language; terminology varies among platforms, so teachers should specify what the text contains rather than assume every ‘subtitle’ includes meaningful environmental sounds.
Audio description does a different job. It can state what is visible during pauses in speech: ‘The blue line rises from 20 °C at minute zero to 35 °C at minute three.’ That cannot be derived from an ordinary caption of ‘Look at the blue line’ if the graph was never explained aloud. The most efficient accessibility improvement is often to make the teacher’s original narration informative: ‘On the graph, temperature rises from 20 °C to 35 °C over three minutes.’ This helps many learners without needing a separate descriptive track for that moment.
A descriptive transcript offers another route. Instead of listing only the dialogue, it includes relevant visual events, speakers and sound cues in reading order. It can be searched, enlarged, translated carefully and used with assistive technology. A descriptive transcript is not simply a set of disconnected screenshots or a raw list of automatically recognised words.
Worked example: a 24-second invented science clip
Invented teaching storyboard, not a real experiment: a 24-second instructional animation displays a temperature–time graph. The vertical axis is temperature in degrees Celsius. The horizontal axis is elapsed time in minutes. In this made-up example, a blue line goes from 20 °C at 0 minutes to 35 °C at 3 minutes. No real thermometer, cohort outcome or school lab result is being claimed.
- At 00:00–00:06, the teacher says, ‘Here is the temperature graph.’ The caption can reproduce that speech.
- At 00:06–00:12, the teacher says, ‘Look how the reading changes.’ The caption can reproduce those words, but they do not give the plotted values.
- At 00:12–00:18, the screen silently displays the two endpoints while an arrow appears. These visual facts need explanation for a learner who cannot see the diagram.
- At 00:18–00:24, the teacher asks, ‘What was the rate of increase?’ The answer cannot be calculated from dialogue alone if no values were spoken.
For the invented graph, the temperature change is 35 − 20 = 15 °C. The elapsed time is 3 − 0 = 3 minutes. The average rate of increase over the displayed interval is therefore 15 °C ÷ 3 min = 5 °C per minute. This number describes the mean change per minute, not necessarily an identical change during every second. Without the endpoints and axes, a learner cannot justify the calculation.
A useful replacement narration is: ‘The blue temperature line begins at 20 degrees Celsius at zero minutes and reaches 35 degrees Celsius after three minutes. That is a rise of 15 degrees Celsius in three minutes.’ The captions can faithfully represent this revised spoken explanation. If the video also includes a silent gesture to an extra outlier or an essential colour-coded series, those features still need an equivalent explanation.
What a descriptive transcript can look like
A descriptive transcript for this clip could use short labelled passages such as: [Visual] A graph labelled ‘Temperature (°C)’ vertically and ‘Time (minutes)’ horizontally shows a blue line from (0, 20) to (3, 35). [Narration] ‘Look how the reading changes.’ [Visual] An arrow points from the starting dot to the ending dot. [Narration] ‘What was the rate of increase?’ This structure lets a reader locate the mathematical evidence even without the picture.
The point is not to write an essay about every visual pixel. Description should communicate information that matters to the learning goal. Decorative wallpaper, a teacher’s shirt colour or a routine camera zoom need not be described unless they change the meaning. In a geometry lesson, however, a dashed auxiliary line, a marked right angle or a changing position may be essential.
Diagnose the access failure instead of blaming attention
Before assuming a learner failed to pay attention, turn off the image and listen to the audio or ask a colleague to read only the captions. Can they answer the exercise? If not, the instructional evidence may exist only on screen. A learner may understand the mathematics perfectly but lack the referenced numbers.
- If the captions omit important spoken words, correct the caption transcription and timing.
- If the caption reads accurately but says only ‘this one’ or ‘over here’, revise spoken references and meaningful visual description.
- If an automatic caption confuses ‘fifteen’ and ‘fifty’, inspect the audio and mathematical plausibility before publishing.
- If a transcript lists dialogue but not the diagram, produce a descriptive transcript.
- If colour alone identifies which line is relevant, add labels, shape or explicit series names.
- If descriptions interrupt the main explanation, plan pauses, integrated narration or an appropriate alternative audio track.
Progressive practice for learners and teachers
Stage 1: hear, read and see three versions
Play a harmless fictional clip three ways: audio without picture; caption text without picture; and the complete visual video. Students identify facts that appear only in the visual version. Ask for a precise gap statement, such as ‘The start and end temperatures are absent from the speech.’ Avoid saying simply ‘There was no context.’
Stage 2: separate transcription from description
Provide five invented lines. For each, classify whether it records spoken dialogue, meaningful sound or visual information: ‘The water bubbles’; ‘[timer beeps]’; ‘The graph line reaches 35 °C’; ‘[music plays]’; ‘Notice the value.’ Discuss which of the statements are audible and which require image access. The same written sentence can serve different roles depending on the soundtrack, so students should classify based on evidence rather than wording alone.
Stage 3: rewrite a vague spoken reference
Replace ‘This gets hotter’ with ‘The thermometer reading rises from 20 °C to 35 °C across three minutes.’ Check that the rewritten line reflects the actual data on the diagram and does not introduce an invented trend. Then create captions from the corrected script, preserving sensible timing and speaker labels when necessary.
Stage 4: make a descriptive text alternative
Ask the learner to write two sentences describing the graph’s axis labels, endpoints and change. Have another learner, without the image, calculate the average rate from those sentences. If the second learner cannot work out 5 °C per minute, inspect whether the description omitted units, coordinates or elapsed time.
Stage 5: independently audit a new subject
Choose a short original clip with a map, historical timeline or geometry construction. The student must identify one silent visual fact, supply an honest explanation, check captions against speech and test whether another reader can complete a specific learning task. Transfer to a new subject is stronger evidence of understanding than memorising the science graph story.
Common mistakes when making a lesson accessible
- ‘Captions mean every visual element is explained.’ They do not necessarily convey information never spoken or sounded.
- ‘The automatic transcript is enough.’ Speech recognition can make errors and usually omits visual evidence.
- ‘Audio description must mention everything on screen.’ Focus on meaningful information relevant to the lesson.
- ‘A photo of the graph explains itself.’ Some learners cannot access the image, and many images are ambiguous without axes or units.
- ‘A label like 35 is self-explanatory.’ The unit, time and associated variable matter.
- ‘Adding more words always helps.’ An overloaded voice track may make comprehension worse; timing and clarity must be designed.
What parents, students and tutors can observe
A parent can ask whether the child can repeat what a graph shows from the narrated explanation alone. A teacher can assign one group to inspect captions and another to inspect visual meaning. Compare the missing facts they identify rather than comparing their reading speeds. For student-created videos, use a simple audit sheet: important audio present, visual facts explained, calculations supported and final transcript checked.
Observe whether a learner can correctly answer the same question using different accessible formats. Improvement may mean fewer missing axis labels, more accurate descriptions, clearer timing and a smaller gap between the conclusions reached from sighted and nonvisual access. It does not mean every child will achieve the same score; teaching materials, individual needs and test settings differ.
Limits, standards and primary sources
The W3C Web Accessibility Initiative explains that prerecorded video with meaningful audio generally needs captions for WCAG Level A, and that visual information can require additional access arrangements; the requirements vary across audio-only, video-only and synchronised media. Different WCAG levels distinguish certain descriptive-transcript and audio-description obligations. Use the W3C audio/video planning guide, captions guidance and transcripts guidance for exact distinctions.
The fictional graph is only a worked teaching example. Actual science recordings require verified axis values, units, times and context; inaccurate narration is worse than a clearly labelled limitation. A transcript may be helpful even where not specifically mandated under one conformance level. No single file format or automatic captioning service guarantees that a complex lesson is fully accessible.
Frequently asked questions
If the teacher reads every number aloud, do we still need description?
If all essential visual information is already meaningfully included in the ordinary soundtrack, the additional descriptive narration needed for that information may be minimal. Captions should still accurately capture meaningful audio when required.
Are subtitles and captions identical?
People and platforms use the terms differently. Captions are generally meant to represent meaningful speech and other audio information; subtitles often refer to dialogue, including translation. Check what the particular file actually includes.
What if the silent graph is an optional illustration?
If no essential instructional information depends on it, extensive description may not be needed. Test by asking whether the intended learning task can be completed using the other material.
Can one descriptive transcript help a classroom video?
Yes. It can make the meaning searchable and available in another modality. But meeting a particular accessibility requirement may also demand captions or audio description, depending on the media and standard.
Education is successful when the idea travels beyond one format
The central lesson is to distinguish what was said from what was shown and then make the essential meaning genuinely available. A pupil who learns to do this becomes a better scientist, writer, video editor and collaborator. The education objective is not merely to publish captions, but to check whether a person can use the evidence and reason from it independently.
Continue at the eduKate Education Hub, the Core Aim of Education hub and the related article Why Captions Cannot Explain a Silent Diagram. The verified Core Aim guide to graph alt text covers an adjacent still-image problem. Our Clementi Mathematics teaching reference illustrates the wider sequence of explanation, guided practice and independent checking.