H5P accessibility testing begins with the interactive content type a student actually uses: a quiz, interactive video, course presentation, branching scenario or matching activity. The central question for accessible multilingual H5P content is whether the learner can perceive the prompt, operate the activity, understand feedback and finish the same educational task without depending on one language or input device.
There is no responsible single blanket answer that every H5P activity is accessible. Content types differ, versions change and author choices can introduce barriers even when a type is considered generally suitable. H5P publishes current content-type accessibility recommendations, and effective testing also needs keyboard review, screen-reader checks where applicable, captions, focus behavior and correct-answer verification.
This longform guide explains how to choose an appropriate H5P content type, test keyboard interaction and focus, translate prompts and feedback without changing scoring, and make video or image alternatives work for real learners. The emphasis is on repeatable learning tasks across language editions, not merely on replacing English strings in an exported package.
Quick answer: what to do first
- Start with a real reader task in H5P and identify which part of interactive learning content types, feedback behavior, keyboard pathways, media and language editions the checker or review covers.
- Inspect meaningful headings, links, instructions, alternative text, data relationships and language accuracy before polishing the visual design.
- Use an automated checker to locate detectable issues, then review keyboard use, assistive technology and actual learner access where relevant.
- Keep assignment rules, scoring, dates, accepted answers, file versions and learner permissions aligned across language editions.
- Document who owns unresolved barriers and provide a usable equivalent path rather than describing a partial test as universal accessibility.
- Release only after the central task has been tested in its published context, with evidence of the language and device conditions reviewed.
What accessibility means in this authoring environment
Why content type selection is a design decision
Two H5P content types may teach a similar concept through very different mechanics. Selecting a type known to work well for a learner’s available input methods is often more effective than trying to patch an unsuitable interaction after publication. Check current official recommendations, not an old screenshot or generic claim.
What authors can and cannot control
Authors can usually control instructions, media descriptions, answer labels, feedback and some configuration. Navigation, widget markup and keyboard behavior also depend on the content type and runtime. An inaccessible mechanic may require an alternative content type or equivalent assessment.
Translation must preserve answer logic
Translate instructions and explanations, but verify the expected answer, accepted variants, punctuation handling and scoring rules. A word-order activity translated literally may become unsolvable; an answer key carried across languages can grade correct learners as wrong.
Testing happens in the embedded environment
H5P can appear in an LMS, an article, a presentation or another platform. The outer page’s keyboard order and language may change the experience. Run tests in the destination environment with the same devices, permissions and browser assumptions students actually have.
Scope matters: H5P accessibility depends on the specific content type, its version, how an author configures it and the learning environment. H5P’s content-type recommendations and current conformance notes should be reviewed before claiming accessibility. The central responsibility remains with the author and publisher: determine what the tool examined, what it missed and whether a reader can finish the task without privileged knowledge of the source-language edition.
A language edition is more than a replacement set of strings. Its readers need understandable requirements, consistent technical vocabulary, correct numbers and working routes to resources. For H5P, begin by selecting one authentic learner journey and making the destination and success criteria explicit before redesigning paragraphs. The comparison should be made with the actual activity state, not solely an editor screenshot.
Twelve-stage practical workflow
1. Choose an authentic reader task
Write the expected learner action and what would count as successful completion. Include the source and target language, assessment conditions and any access dependencies. A test without a defined outcome can identify attractive pages while missing a broken activity.
In H5P, For task selection, write down the precise activity and result that matter to the learner: where the journey starts, what must be found, what must be submitted or understood, and how completion is recognized. Avoid accepting an abstract phrase such as good usability as a test condition. Include a language edition and a meaningful learner role so the evaluation is reproducible. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
2. Locate the authoritative material
Locate the approved original and identify which course object, chapter, document or activity actually governs the instruction. Check whether an older export or translated copy is being mistaken for the current authority. Preserve stable identifiers.
In H5P, The source of truth may be an assignment setting, book metadata record, quiz configuration or an approved institutional policy rather than the surrounding explanatory paragraph. Capture those facts before editing text. If two records disagree, resolve the conflict with the accountable owner instead of choosing whichever expression sounds smoother in the target language. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
3. Determine tool coverage
Name the precise checker, edition and item reviewed. Do not present a result for one authoring surface as evidence about unrelated attachments or embedded services. Record external dependencies that need their own review.
In H5P, An editor-level accessibility check and a whole-resource audit are different evidence sources. Identify whether the tool reads HTML, a specific interactive component, a course file or a published export. Record what is excluded, because an unsupported inference about coverage can allow a serious barrier to survive even a successful report. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
4. Repair written instructions
Replace ambiguous directions with practical steps in a logical order. Make deadlines, exceptions and scoring conditions visible. Confirm that terminology is appropriate for the reader, then compare the words with actual configuration.
In H5P, Instructional clarity means spelling out the action, the object of the action and any conditions that change the result. Separate steps that must happen in order from optional suggestions. Test consequential negations, dates and scoring thresholds with a knowledgeable reviewer; those errors can invalidate a task even if the interface passes a structural check. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
5. Structure headings and lists
Use real headings, lists and other semantic relationships rather than creating structure through font sizes, spaces and colors. Review the resulting outline in every language edition, since text expansion can expose authoring shortcuts.
In H5P, Heading levels should reflect the ideas readers need to traverse. Use semantic lists for parallel items and ordered lists when a sequence is required. Then inspect the announced outline, not just the visual typography. A translated heading can expand across lines, but it should not silently become a different structural level. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
6. Describe meaningful visuals
Treat images, graphs and media as carriers of instructional meaning. Distinguish decoration from evidence, provide appropriately short labels and fuller nearby explanations, and preserve the concept that a nonvisual learner needs.
In H5P, For each visual, identify what a learner is expected to infer: an object, a trend, a sequence, a location or a comparison. A short alternative should communicate that purpose, while a complex graph or diagram may need adjacent detailed text or an accessible table. Do not replace essential meaning with a generic description of colors. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
7. Verify links and references
Make link names describe destinations and verify their actual targets and permissions. A translated label should not promise a translated resource unless that resource exists. Check downloads and any references to external services.
In H5P, A descriptive label is useful only if the destination delivers what it promises. Check whether references open in the correct language, whether file versions match the current edition and whether access is available without instructor privileges. Names such as Download guide should identify the actual format when that distinction changes the next action. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
8. Review tables and data
Check which cells, axes, measures and units belong together. Replace complex visual grouping with accessible headers and accompanying prose when necessary. Verify numbers independently from the linguistic revision.
In H5P, Check that a data structure announces the correct relationships, not merely a sequence of numbers. Every measurement needs an interpretable unit and category. Preserve decimal meanings, signs and boundaries during localization. When a table relies on nested merged headers, a simpler structure and a paragraph explanation may improve comprehension. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
9. Inspect media and interactions
Test pause, focus, captions, responses and equivalent routes in the actual environment. A check of static text cannot prove a dynamic activity works or that its feedback and grading are understandable.
In H5P, Interactive content needs review of states that a static page never displays: the initial prompt, focus movement, validation, error response, help, retry, success and final completion. For recorded media, examine captions and transcripts alongside their timing. When a particular mechanic excludes a learner, provide an equivalent way to demonstrate the intended skill. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
10. Test all language editions
Review full meaning in each selected language, including date formats, numbers, writing direction, terminology and names of actions. Ask a target-language reader to follow an unfamiliar task rather than approve sentences in isolation.
In H5P, Linguistic review should cover all visible and hidden states, not just the first screen. Check words of obligation, answer logic, dates, numerical formats, abbreviations and the text that announces a successful action. A target-language specialist should be able to reconstruct the same educational requirement without referring to the original page. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
11. Check learner access and devices
Use learner-level permissions and representative screen sizes. Inspect keyboard order, zoom, screen-reader behavior where appropriate, and the journey to and from the resource. Reopen the item after publication to catch context-dependent faults.
In H5P, Testing as a student can reveal inaccessible links, hidden settings, file permissions and return paths that an author preview misses. Use an appropriate test account and representative devices; inspect keyboard order, screen magnification and screen-reader output where relevant. A passing view at desktop width should not be assumed to work on mobile. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
12. Record evidence and release
Record the item, test date, locale, conditions, observation, fix owner and final decision. If a meaningful barrier remains, supply an accessible route and record the limitation. Re-test when course content or tool versions change.
In H5P, A release record should identify the specific version, locale, scope, expected outcome, observed result and person accountable for unresolved work. Treat a failure in the central task as a reason to repair or supply an equivalent route. Revisit the record after edits, exports, updates or newly copied learning resources; evidence has a version. Apply this reasoning to interactive learning content types, feedback behavior, keyboard pathways, media and language editions rather than treating the tool’s authoring interface as the only relevant evidence.
Applied accessibility and multilingual QA situations
The following 28 situations are practical editorial cases, not fictional claims about a particular institution. Each illustrates an observable failure, the consequence for a learner, a repair and a verification question. Where the named checker cannot inspect a feature, use an independent tool or manual task test. The value of a case is whether someone else can reproduce and resolve the problem.
1. Drag and drop lesson
The situation. A classification task requires students to drag animal names to habitats. Some learners cannot perform the operation with their input device. An editor may see a tidy page while a keyboard user encounters a different order of controls. Do not allow the editing preview to stand in for an actual learner journey.
The repair. Verify the content type’s current keyboard support and offer an equivalent operable classification task. Avoid inserting a translated string until the destination, associated feedback and required action have been checked. The interface alone cannot establish end-to-end correctness. For this drag and drop lesson situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Confirm that the linked destination, downloadable file and feedback language match the label shown to learners. A localized promise is not fulfilled by an unrelated English resource. In the specific case of drag and drop lesson, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
2. Interactive video pause
The situation. A question overlays a video just after an important verbal explanation. Students using captions may be reading when the prompt interrupts playback. Meaning can change when a label is shortened, a number is reformatted or an English-only example is left in a translated page. The review must test instructional consequences.
The repair. Check caption timing, pause behavior and return focus after answering. Choose the shortest wording that preserves the learning obligation without deleting a condition, example or exception. Concision should remove friction, not evidence. For this interactive video pause situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Ask a reviewer who did not write the page to follow its sequence and identify the next action. Their interpretation is stronger evidence than an author rereading familiar content. In the specific case of interactive video pause, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
3. True-false translation
The situation. An English true-or-false item relies on a double negative. The translated version changes the answer’s logical polarity. A technically valid heading or image description can still be unhelpful if it fails to explain the distinction the learner must make. Evaluate the educational task as well as the code.
The repair. Rewrite the prompt plainly and independently revalidate the keyed answer. For complex information, supply a layered explanation: a short accessible label, an adjacent description and a detailed alternative where needed. For this true-false translation situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Keep reproducible evidence: page or activity, language edition, test conditions, observation, owner and outcome. A screenshot alone rarely proves that the experience is operable. In the specific case of true-false translation, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
4. Missing image description
The situation. An image hotspot lesson asks learners to select the correct organ. The image has no meaningful description or alternate route for identifying the target. Longer words and scripts with different line-breaking behavior change layouts. Check the rendered state rather than treating the source-language screen as a reliable substitute.
The repair. Explain the organ features and provide an equivalent nonvisual question. Compare the repaired result with the original learning objective. The aim is equivalent understanding and action, not identical pixel dimensions or word counts. For this missing image description situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. A useful test asks an unfamiliar reader to complete the task without coaching and to explain what success looks like. Record any ambiguity rather than inferring success from silence. In the specific case of missing image description, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
5. Branching scenario labels
The situation. A workplace decision activity has choices called Continue and Continue. A screen reader cannot distinguish the consequences of the options. A clean automatic report is evidence about the checks performed, not a verdict on every barrier. Manual review is needed where judgment, interaction and teaching intent are involved.
The repair. Use decision-specific labels and verify the resulting branch in each locale. Keep a source-specific correction log so that the next revision does not silently recreate the same defect. Record decisions in terms that a new editor can apply. For this branching scenario labels situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Test the relevant state with only a keyboard and check a screen-reader route where appropriate. An action that depends on hovering or color remains vulnerable. In the specific case of branching scenario labels, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
6. Fill-in-the-blanks accents
The situation. A Spanish language gap exercise rejects an otherwise valid accented spelling variant. Correct learners lose points because accepted alternatives were not updated. Course owners should distinguish failures they can repair inside the authoring tool from those requiring an administrator, media specialist, platform vendor or alternative route.
The repair. Review exact-match rules and the pedagogical importance of diacritics. Check the item within the exact published context, including surrounding navigation and permissions. A standalone preview may behave differently from a real course. For this fill-in-the-blanks accents situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Read the target-language material aloud, including the crucial numbers and nouns. If the intended requirement cannot be restated confidently, revise before publishing. In the specific case of fill-in-the-blanks accents, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
7. Flashcards reversed
The situation. A bilingual vocabulary flashcard shows the answer before the question in one language edition. The retrieval task no longer measures the intended memory skill. If assessment conditions change between versions, learners can receive different grades for identical work. Consistency of rules matters more than literal sentence resemblance.
The repair. Reorder the face content and test first reveal, answer and reset behavior. Use the approved source of truth for the instruction before translating it. Keep identifiers, dates, scores and permissions under the same ownership as their corresponding learning objects. For this flashcards reversed situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Compare the visible instruction with the associated assignment or resource configuration. Differences in timing, access or scoring need correction at their authoritative source. In the specific case of flashcards reversed, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
8. Focus lost after grading
The situation. Submitting a multi-item quiz moves keyboard focus to an unrelated section. Learners do not know where the success or error feedback appears. The corrective action must be understandable to a colleague who did not create the original page. Keep the requirement, its context and its exact success condition together.
The repair. Ensure the current type and player return focus meaningfully or supply an alternative. Prefer semantic structure and plain instructions to cosmetic formatting. When a visual cue carries meaning, state that meaning in readable text as well. For this focus lost after grading situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Use a narrow viewport and enlarged text for the affected language. Verify that important labels do not disappear, overlap or become disconnected from their controls. In the specific case of focus lost after grading, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
9. Answer identified by color
The situation. A multiple-choice task marks the correct answer green without textual feedback. Students unable to distinguish green cannot understand what they learned. An editor may see a tidy page while a keyboard user encounters a different order of controls. Do not allow the editing preview to stand in for an actual learner journey.
The repair. Add explicit correct-answer feedback explaining the relevant concept. Avoid inserting a translated string until the destination, associated feedback and required action have been checked. The interface alone cannot establish end-to-end correctness. For this answer identified by color situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Confirm that the linked destination, downloadable file and feedback language match the label shown to learners. A localized promise is not fulfilled by an unrelated English resource. In the specific case of answer identified by color, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
10. Captions without speaker names
The situation. A two-person conversation video has translated captions with no speaker distinction. Learners misattribute a request and its response. Meaning can change when a label is shortened, a number is reformatted or an English-only example is left in a translated page. The review must test instructional consequences.
The repair. Add speaker identification where necessary and check synchronized caption timing. Choose the shortest wording that preserves the learning obligation without deleting a condition, example or exception. Concision should remove friction, not evidence. For this captions without speaker names situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Ask a reviewer who did not write the page to follow its sequence and identify the next action. Their interpretation is stronger evidence than an author rereading familiar content. In the specific case of captions without speaker names, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
11. Audio-only instruction
The situation. A dictation activity contains spoken directions absent from the visible text. Learners who cannot hear the instruction cannot begin the task. A technically valid heading or image description can still be unhelpful if it fails to explain the distinction the learner must make. Evaluate the educational task as well as the code.
The repair. Provide a transcript or an equivalently assessable alternative consistent with the objective. For complex information, supply a layered explanation: a short accessible label, an adjacent description and a detailed alternative where needed. For this audio-only instruction situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Keep reproducible evidence: page or activity, language edition, test conditions, observation, owner and outcome. A screenshot alone rarely proves that the experience is operable. In the specific case of audio-only instruction, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
12. Long German choice text
The situation. A translated choice label overflows the available quiz panel. The final qualifying phrase disappears on a small screen. Longer words and scripts with different line-breaking behavior change layouts. Check the rendered state rather than treating the source-language screen as a reliable substitute.
The repair. Test wrapping and redesign layout or wording without removing the condition. Compare the repaired result with the original learning objective. The aim is equivalent understanding and action, not identical pixel dimensions or word counts. For this long german choice text situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. A useful test asks an unfamiliar reader to complete the task without coaching and to explain what success looks like. Record any ambiguity rather than inferring success from silence. In the specific case of long german choice text, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
13. Arabic word order
The situation. A sentence ordering activity uses a left-to-right layout for Arabic tokens. The displayed order conflicts with the correct Arabic sequence. A clean automatic report is evidence about the checks performed, not a verdict on every barrier. Manual review is needed where judgment, interaction and teaching intent are involved.
The repair. Test direction and the answer key using an Arabic-speaking reviewer. Keep a source-specific correction log so that the next revision does not silently recreate the same defect. Record decisions in terms that a new editor can apply. For this arabic word order situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Test the relevant state with only a keyboard and check a screen-reader route where appropriate. An action that depends on hovering or color remains vulnerable. In the specific case of arabic word order, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
14. Numerical decimal separator
The situation. A calculation activity expects 3.5 while local instruction teaches 3,5. A valid numeric answer may be scored incorrect. Course owners should distinguish failures they can repair inside the authoring tool from those requiring an administrator, media specialist, platform vendor or alternative route.
The repair. State accepted notation and configure or explain numeric input behavior. Check the item within the exact published context, including surrounding navigation and permissions. A standalone preview may behave differently from a real course. For this numerical decimal separator situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Read the target-language material aloud, including the crucial numbers and nouns. If the intended requirement cannot be restated confidently, revise before publishing. In the specific case of numerical decimal separator, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
15. Keyboard trap
The situation. A learner tabs into an embedded interactive object but cannot reach the next page control. The entire surrounding course becomes unusable from the keyboard. If assessment conditions change between versions, learners can receive different grades for identical work. Consistency of rules matters more than literal sentence resemblance.
The repair. Test entry and exit paths and replace or repair the incompatible widget. Use the approved source of truth for the instruction before translating it. Keep identifiers, dates, scores and permissions under the same ownership as their corresponding learning objects. For this keyboard trap situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Compare the visible instruction with the associated assignment or resource configuration. Differences in timing, access or scoring need correction at their authoritative source. In the specific case of keyboard trap, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
16. Repeated question prompts
The situation. A quiz displays several items all headed Choose one with no contextual label. Screen-reader users lose track of which question they are answering. The corrective action must be understandable to a colleague who did not create the original page. Keep the requirement, its context and its exact success condition together.
The repair. Include the specific concept in question text and review announced grouping. Prefer semantic structure and plain instructions to cosmetic formatting. When a visual cue carries meaning, state that meaning in readable text as well. For this repeated question prompts situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Use a narrow viewport and enlarged text for the affected language. Verify that important labels do not disappear, overlap or become disconnected from their controls. In the specific case of repeated question prompts, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
17. Tiny hotspot targets
The situation. An image hotspot requires tapping a tiny region on a medical diagram. Touch and motor access become obstacles separate from medical knowledge. An editor may see a tidy page while a keyboard user encounters a different order of controls. Do not allow the editing preview to stand in for an actual learner journey.
The repair. Use larger targets or an equivalent structured-choice activity. Avoid inserting a translated string until the destination, associated feedback and required action have been checked. The interface alone cannot establish end-to-end correctness. For this tiny hotspot targets situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Confirm that the linked destination, downloadable file and feedback language match the label shown to learners. A localized promise is not fulfilled by an unrelated English resource. In the specific case of tiny hotspot targets, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
18. Narration volume mismatch
The situation. A course presentation overlays quiet speech with loud background music. Students may miss the central explanation even with correct translation. Meaning can change when a label is shortened, a number is reformatted or an English-only example is left in a translated page. The review must test instructional consequences.
The repair. Rebalance audio, provide reliable captions and separate meaningful content from ambience. Choose the shortest wording that preserves the learning obligation without deleting a condition, example or exception. Concision should remove friction, not evidence. For this narration volume mismatch situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Ask a reviewer who did not write the page to follow its sequence and identify the next action. Their interpretation is stronger evidence than an author rereading familiar content. In the specific case of narration volume mismatch, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
19. Incorrect feedback branching
The situation. A safety scenario translates No as Yes in one choice label. The visible choice no longer matches the destination branch. A technically valid heading or image description can still be unhelpful if it fails to explain the distinction the learner must make. Evaluate the educational task as well as the code.
The repair. Trace every decision path and compare its actual consequence against the label. For complex information, supply a layered explanation: a short accessible label, an adjacent description and a detailed alternative where needed. For this incorrect feedback branching situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Keep reproducible evidence: page or activity, language edition, test conditions, observation, owner and outcome. A screenshot alone rarely proves that the experience is operable. In the specific case of incorrect feedback branching, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
20. Quiz completion wording
The situation. The last button says Exit in translation but means Submit answers. Learners may leave thinking their results have not been saved. Longer words and scripts with different line-breaking behavior change layouts. Check the rendered state rather than treating the source-language screen as a reliable substitute.
The repair. Use explicit final action text and verify the score and completion state. Compare the repaired result with the original learning objective. The aim is equivalent understanding and action, not identical pixel dimensions or word counts. For this quiz completion wording situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. A useful test asks an unfamiliar reader to complete the task without coaching and to explain what success looks like. Record any ambiguity rather than inferring success from silence. In the specific case of quiz completion wording, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
21. Screen reader announcement order
The situation. A memory game announces the card’s decorative border before its content. Essential choices get buried in repeated noninstructional words. A clean automatic report is evidence about the checks performed, not a verdict on every barrier. Manual review is needed where judgment, interaction and teaching intent are involved.
The repair. Remove unnecessary spoken decoration and test choice identification. Keep a source-specific correction log so that the next revision does not silently recreate the same defect. Record decisions in terms that a new editor can apply. For this screen reader announcement order situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Test the relevant state with only a keyboard and check a screen-reader route where appropriate. An action that depends on hovering or color remains vulnerable. In the specific case of screen reader announcement order, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
22. Untranslated hint
The situation. A French geography game translates every prompt except the hint revealed after failure. The learner receives support only in the original language. Course owners should distinguish failures they can repair inside the authoring tool from those requiring an administrator, media specialist, platform vendor or alternative route.
The repair. Translate and review all states, including retry, hint and success messages. Check the item within the exact published context, including surrounding navigation and permissions. A standalone preview may behave differently from a real course. For this untranslated hint situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Read the target-language material aloud, including the crucial numbers and nouns. If the intended requirement cannot be restated confidently, revise before publishing. In the specific case of untranslated hint, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
23. Transcript detached
The situation. A branching video has a full transcript on a separate page without timestamps. Learners cannot connect decisions to the relevant segment. If assessment conditions change between versions, learners can receive different grades for identical work. Consistency of rules matters more than literal sentence resemblance.
The repair. Provide navigable transcript cues and an equivalent decision path. Use the approved source of truth for the instruction before translating it. Keep identifiers, dates, scores and permissions under the same ownership as their corresponding learning objects. For this transcript detached situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Compare the visible instruction with the associated assignment or resource configuration. Differences in timing, access or scoring need correction at their authoritative source. In the specific case of transcript detached, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
24. Mixed-language metadata
The situation. An H5P activity title and body use different languages without clear identification. Pronunciation and navigation can become confusing in some assistive setups. The corrective action must be understandable to a colleague who did not create the original page. Keep the requirement, its context and its exact success condition together.
The repair. Use supported language settings and consistent labels around embedded content. Prefer semantic structure and plain instructions to cosmetic formatting. When a visual cue carries meaning, state that meaning in readable text as well. For this mixed-language metadata situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Use a narrow viewport and enlarged text for the affected language. Verify that important labels do not disappear, overlap or become disconnected from their controls. In the specific case of mixed-language metadata, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
25. Timed question
The situation. A language-learning question advances before some learners can read the translated prompt. The time constraint tests reading speed rather than the intended concept. An editor may see a tidy page while a keyboard user encounters a different order of controls. Do not allow the editing preview to stand in for an actual learner journey.
The repair. Allow enough time or remove the timer unless timing is a valid assessment target. Avoid inserting a translated string until the destination, associated feedback and required action have been checked. The interface alone cannot establish end-to-end correctness. For this timed question situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Confirm that the linked destination, downloadable file and feedback language match the label shown to learners. A localized promise is not fulfilled by an unrelated English resource. In the specific case of timed question, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
26. Activity in an iframe
The situation. An H5P item works alone but loses keyboard navigation when embedded inside a course. The parent page and embedded player handle focus differently. Meaning can change when a label is shortened, a number is reformatted or an English-only example is left in a translated page. The review must test instructional consequences.
The repair. Retest in the real LMS context and check tab order across the boundary. Choose the shortest wording that preserves the learning obligation without deleting a condition, example or exception. Concision should remove friction, not evidence. For this activity in an iframe situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Ask a reviewer who did not write the page to follow its sequence and identify the next action. Their interpretation is stronger evidence than an author rereading familiar content. In the specific case of activity in an iframe, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
27. Correct-answer leakage
The situation. A translated memory card places the answer in the alternative text of its question image. Screen-reader users hear the response before attempting retrieval. A technically valid heading or image description can still be unhelpful if it fails to explain the distinction the learner must make. Evaluate the educational task as well as the code.
The repair. Describe the image without revealing the scored answer and offer equivalent clues. For complex information, supply a layered explanation: a short accessible label, an adjacent description and a detailed alternative where needed. For this correct-answer leakage situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. Keep reproducible evidence: page or activity, language edition, test conditions, observation, owner and outcome. A screenshot alone rarely proves that the experience is operable. In the specific case of correct-answer leakage, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
28. Version regression
The situation. A previously tested interaction behaves differently after an H5P library update. Stored QA results no longer represent the live runtime. Longer words and scripts with different line-breaking behavior change layouts. Check the rendered state rather than treating the source-language screen as a reliable substitute.
The repair. Retest a representative learner journey and record library and content-type versions. Compare the repaired result with the original learning objective. The aim is equivalent understanding and action, not identical pixel dimensions or word counts. For this version regression situation, decide whether the solution belongs in the authored content, the linked resource, the language review or the platform configuration. Do not make an unsupported promise about what H5P interactive content accessibility testing automatically checks.
How to confirm it works. A useful test asks an unfamiliar reader to complete the task without coaching and to explain what success looks like. Record any ambiguity rather than inferring success from silence. In the specific case of version regression, the review should establish that the learning requirement survives translation and that the student can perceive and complete the relevant action. If the fix is unavailable inside H5P, document a suitable equivalent route and who will implement it.
Questions educators and publishers regularly ask
Does a clean report guarantee accessibility?
No. It only reports the checks implemented for the inspected item. H5P accessibility depends on the specific content type, its version, how an author configures it and the learning environment. H5P’s content-type recommendations and current conformance notes should be reviewed before claiming accessibility. Human understanding, effective interaction and correct translation still need evidence.
Can one language edition stand in for the others?
No. Script direction, line breaks, technical vocabulary and local conventions can change the usability of a resource. Sample every published language, giving extra attention to complex layouts and consequential instructions.
What about an attached Word file or PDF?
Check the original file with an appropriate document-accessibility process. A result from H5P interactive content accessibility testing is not proof that arbitrary attachments or downloadable versions have correct tags, captions or reading order.
Who approves the learning meaning?
A subject specialist or accountable course owner should confirm the objective, answer logic, grading and conditions. A language reviewer checks clarity and terminology; an accessibility reviewer checks operation and alternative formats.
Is machine translation enough for a fast release?
Automatic drafting can help with noncritical text, but consequential instructions and feedback need informed review. Check numbers, negations, deadlines, role labels and the actions a learner is expected to take.
What should happen if the product lacks a needed feature?
Provide an equivalent accessible learning route, record the constraint and assign an owner for follow-up. Avoid silently deleting educational content solely to make an automated score improve.
How many pages should a team audit first?
Start with the highest-impact tasks: joining, reading essential instructions, completing an activity, submitting evidence and receiving feedback. Expand to a representative range of content types and languages.
What counts as good proof?
A dated record of the item, edition, user role, device or assistive method, expected outcome, observed result and correction. An unannotated screenshot or a generic statement that everything passed cannot reproduce the result.
How should technical terms be handled?
Preserve a controlled glossary, define unfamiliar abbreviations and check that the target-language term fits the scientific or professional concept. Consistency between text, feedback and assessments matters more than literal similarity.
When should review happen again?
Repeat the affected tests after content revisions, platform updates, course copies, altered permissions or export changes. A previously good result is not perpetual evidence about a different version of the learning artifact.
A practical release checklist
- The first three paragraphs of the reader-facing page state the main answer and the limits of the tool clearly.
- The document outline, link labels, alternatives, tables, captions and core content can be understood in the selected language.
- Deadlines, timezones, numbers, feedback, scoring rules and permissions match the authoritative course or publishing records.
- The critical task works with learner permissions, appropriate keyboard controls and the necessary media or document alternatives.
- Source and target editions were reviewed separately; language scripts, line length, mixed-direction text and localized examples were not assumed to be equivalent.
- Unresolved material barriers have named owners, transparent limitations and a workable equivalent route.
- The checked version, date, testing conditions and final publication decision are documented.
A responsible publication decision for H5P is not a certificate declaring all content accessible. It is a clear finding about this resource, this language edition and the tasks tested. Where a severe barrier prevents a student from completing an essential activity, preserve the educational objective and arrange an equivalent usable route before relying on that activity for assessment.
Authoritative product documentation and further reading
Product capabilities and settings can change. Review the current official materials, particularly before making institutional procurement, legal conformance or certification claims. Useful documentation: official H5P reference 1; official H5P reference 2; official H5P reference 3.
For the overarching principles of accurate communication across languages, read Master Art of Translation and the Translate Like a Pro series index. See also the related guide for an adjacent, different workflow. These links support orientation and comparison; no existing canonical article is replaced by this article.
A reader should leave knowing how to inspect and improve interactive learning content types, feedback behavior, keyboard pathways, media and language editions, what evidence a tool can actually establish, and how to make a defensible learner-centered release decision. The test is practical: can a new learner find the required material, understand the instructions, use the available controls and complete the same meaningful task in the intended language?