How do you translate a ThingLink interactive image, 360-degree tour or branching scenario into another language without leaving essential instructions in the source language? ThingLink translation is unusually dependent on where a learner clicks. A scene may show a picture with a dozen hotspots, while a scenario adds decisions that link multiple scenes. Professional ThingLink localization must preserve not just readable text but the route by which a learner discovers evidence, makes a choice and receives useful feedback.
This ThingLink multilingual translation guide covers automatic translation of interactive scenes and scenarios, manual corrections, language selectors, language-specific sharing and embeds, updated text, accessible viewing, captions, image labels and learning-path verification. It is written for educators, museum curators, trainers, translators and institutions distributing interactive learning experiences across multiple language communities. It explains how to manage the translated content that belongs to the platform separately from media or linked resources that may need their own treatment.
ThingLink’s official help guides, updated July 21 and September 29, 2026, describe automatic translation into more than 40 languages on eligible plans. Crucially, the vendor states that translating a branching scenario does not automatically translate the ThingLink scenes inside it. It also says source edits do not automatically update translations and that translated rich text can lose formatting. Those limits are not minor details: they determine whether a published multilingual tour remains accurate after the source changes.
The answer in one minute
Start by listing every scene, hotspot, text tag, decision block and linked resource in the source tour. Confirm the eligible account plan and select source and target languages through the supported translation control. Translate scenes and scenarios as separate content layers where needed, correct important terminology manually, inspect formatting and update translations after source edits. Share either a learner-selectable multilingual version or an intentionally language-specific link or embed. Test complete paths in the actual viewing context, including screen-reader and keyboard access where applicable.
Learn the actual ThingLink translation boundaries
Understand scene versus scenario
A scene is an interactive medium such as an image, video or 360-degree environment with tags. A scenario is an experience that connects blocks and can include scenes. ThingLink documentation explicitly distinguishes their translation scope. If you translate the scenario’s blocks but never translate the scenes it references, the learner may see localized branching instructions followed by original-language hotspots. Treat the two levels as related but separately verified assets. A project register should record which languages are complete at each level.
Automatic translation and manual corrections
ThingLink describes selecting an original language and target languages, then saving and translating. It provides an editing route for corrections. This is useful for many short labels, but automatic drafts can mishandle proper nouns, museum object names, terms of safety and classroom vocabulary. A reviewer should inspect strings where they appear, not as disembodied sentences. In a scenario, the difference between ‘enter this room’ and ‘do not enter this room’ is a learner-path decision, not a stylistic preference.
Changes demand a second translation event
The official guide warns that updated original text is not automatically propagated. A changed tag may appear in the source language even while the older version of the scene was translated. Establish a source-change record and rerun or revise the affected translations after every substantive change. Then inspect the result rather than assuming an automatic refresh always preserves manual correction or formatting. A multilingual tour is a maintained publication, not a one-time export.
Language-specific links can reduce friction
ThingLink documents both a viewer language selector and links or embeds that open in a chosen language. Each can be useful. A museum may prefer a selector so visitors choose freely; a school may distribute direct Spanish and Arabic links in separate learning portals. Neither option by itself proves every internal scene is localized. Test the exact shared link on a device without editor permissions and make sure the learner sees the intended text, navigation and accessible descriptions.
Formatting and nontext media
ThingLink warns that translated text may lose formatting, including emphasis. Visual hierarchy matters when warnings and optional explanations share a tag. Reapply emphasis to important conditions and ensure an accessible equivalent exists. Treat image-embedded labels, speech in videos, subtitles, documents and 3D content separately. The official guide says 3D model scenes cannot yet be translated through the described scene feature. Do not promise a one-button transformation of every media type.
Accessible viewing and learner dignity
ThingLink presents an Accessibility Player, caption options, alt text and Immersive Reader support. Those can reduce barriers but must be validated with the actual project and language. Alternative text should convey the relevant idea, not merely name a photo. Captions should tell the same story as the spoken narration. The learner must be able to activate necessary tags, identify where they are and return to the decision without losing the thread of the tour.
Step-by-step multilingual production workflow
1. Choose the unit of translation
Inventory each standard scene, scenario and embedded item. Record which languages are intended for the whole project and which scenes are external or unsupported. Save evidence that can be checked later: the exact content layer, target language, input version and learner-facing result. Never conclude that the whole project is translated because one tag looks good in a preview. The important standard is what another person can read and do in the target version, with the same underlying facts and decisions.
2. Approve the master text and facts
Read all original hotspots, decision conditions, safety statements, measurements and wrong-answer explanations. Fix ambiguities before multiplying them across languages. Save evidence that can be checked later: the exact content layer, target language, input version and learner-facing result. Never conclude that the whole project is translated because one tag looks good in a preview. The important standard is what another person can read and do in the target version, with the same underlying facts and decisions.
3. Confirm availability and plan permissions
Check the current ThingLink subscription, translation language support and controls rather than relying on a generic advertisement. Do not send sensitive content into an unapproved service. Save evidence that can be checked later: the exact content layer, target language, input version and learner-facing result. Never conclude that the whole project is translated because one tag looks good in a preview. The important standard is what another person can read and do in the target version, with the same underlying facts and decisions.
4. Translate standard scenes
For each supported image, video or 360 scene, use the available Translate menu to set source and target languages. Check the viewer selector and sample tags after processing. Save evidence that can be checked later: the exact content layer, target language, input version and learner-facing result. Never conclude that the whole project is translated because one tag looks good in a preview. The important standard is what another person can read and do in the target version, with the same underlying facts and decisions.
5. Translate scenario blocks separately
Open each branching scenario and translate its text. Then confirm that its embedded ThingLink scenes have been localized independently. Review the route from scenario entry to final outcome. Save evidence that can be checked later: the exact content layer, target language, input version and learner-facing result. Never conclude that the whole project is translated because one tag looks good in a preview. The important standard is what another person can read and do in the target version, with the same underlying facts and decisions.
6. Correct terminology and formatting
Use the supported manual editing path where available. Check scientific terms, place names, warnings, formatted emphasis and strings that wrap on smaller devices. Save evidence that can be checked later: the exact content layer, target language, input version and learner-facing result. Never conclude that the whole project is translated because one tag looks good in a preview. The important standard is what another person can read and do in the target version, with the same underlying facts and decisions.
7. Release with the intended language route
Choose either the visible language selector or a target-language link/embed as appropriate. Confirm permissions and accessibility mode, and check what happens when a learner opens a direct link. Save evidence that can be checked later: the exact content layer, target language, input version and learner-facing result. Never conclude that the whole project is translated because one tag looks good in a preview. The important standard is what another person can read and do in the target version, with the same underlying facts and decisions.
8. Maintain language parity after source revisions
When a source tag or branch changes, refresh or edit the affected language editions and re-test them. Keep a map of revisions, reviewers, release links and remaining untranslated media. Save evidence that can be checked later: the exact content layer, target language, input version and learner-facing result. Never conclude that the whole project is translated because one tag looks good in a preview. The important standard is what another person can read and do in the target version, with the same underlying facts and decisions.
Worked example: a multilingual coastal wetlands tour
Stage 1
A science museum builds a ThingLink 360-degree coastal wetlands tour. The learner visits three scenes: a tidal channel, a protected nesting site and a community monitoring station. A scenario asks which intervention should happen first when a water-quality alert arrives. The intended audience speaks English, French and Malay. A bilingual subject reviewer should be able to demonstrate what a learner must understand at this point and how the translated scene or scenario supports that conclusion. Where a media element cannot be translated by the built-in route, record the independent replacement or accessible alternative before approving the experience.
Stage 2
The team freezes the source facts and maps hotspot labels, explanatory cards, measurement units, video narration and branch conditions. The word ‘salinity’ must not be confused with ‘pollution’, and the tide chart must keep its original dates. A curator signs off the source decision rule and distinguishes temporary observation from immediate public warning. A bilingual subject reviewer should be able to demonstrate what a learner must understand at this point and how the translated scene or scenario supports that conclusion. Where a media element cannot be translated by the built-in route, record the independent replacement or accessible alternative before approving the experience.
Stage 3
Using its enabled account features, the team translates each supported scene into the two target languages. It then translates the scenario separately because the official documentation warns that scenario translation does not automatically cover the scenes. A reviewer opens every linked scene from each branch to check for remaining English. A bilingual subject reviewer should be able to demonstrate what a learner must understand at this point and how the translated scene or scenario supports that conclusion. Where a media element cannot be translated by the built-in route, record the independent replacement or accessible alternative before approving the experience.
Stage 4
The French reviewer changes an automatically generated place name back to its approved official form and restores emphasis to a warning sentence that lost formatting. The Malay reviewer tests whether the hotspot label fits the mobile view and whether the word chosen for ‘monitoring station’ matches local educational usage. A bilingual subject reviewer should be able to demonstrate what a learner must understand at this point and how the translated scene or scenario supports that conclusion. Where a media element cannot be translated by the built-in route, record the independent replacement or accessible alternative before approving the experience.
Stage 5
One video contains an English recorded interview and on-screen labels. These elements are tracked as separate media work rather than assumed to be automatically translated by the scene’s text function. The project owner either supplies appropriately captioned media or a clear alternative that gives learners the evidence needed for the task. A bilingual subject reviewer should be able to demonstrate what a learner must understand at this point and how the translated scene or scenario supports that conclusion. Where a media element cannot be translated by the built-in route, record the independent replacement or accessible alternative before approving the experience.
Stage 6
Finally, the institution tests a French direct link, a Malay embed and the multilingual selector in a non-editor viewing session. A new learner must interpret the same tide evidence and select the same safe response. After a source update, the team marks the affected hotspots and scenario blocks for retranslation and review. A bilingual subject reviewer should be able to demonstrate what a learner must understand at this point and how the translated scene or scenario supports that conclusion. Where a media element cannot be translated by the built-in route, record the independent replacement or accessible alternative before approving the experience.
Thirty practical localization cases
Use the following cases as separate quality tests. A ThingLink project may not contain every feature, but every feature it does contain should pass an appropriate language, interaction and access check.
1. Untranslated hotspot tag
A museum visitor opens a translated 360 image and sees an English annotation about a historic object. Because this information is reached through an interaction, a translator who reads only the opening page will not see the problem. Trace the exact click sequence and evaluate the words in context. The risk is specific: The surrounding page may appear localized while the essential evidence remains in the original language. The practical intervention is: Open every hotspot in the target-language view and track fields that require separate corrections. Record the corrected wording, scene identifier and tested language so the same issue does not return during the next update. The evidence of completion is: A visitor without English can state what the object is and why it matters. The general lesson is: Translation coverage is measured at each interaction point. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
2. Scenario block versus scene
A learner chooses a French branch that opens an English-language interactive photograph. Separate a language defect from a delivery defect. The text may be accurate but attached to the wrong branch, hidden by formatting or opened through a misleading link. The risk is specific: Scenario translation does not automatically translate its embedded scenes, according to official help. The practical intervention is: Translate and review the scene as a separate asset before confirming the branch’s readiness. Keep a short source-to-target decision note; this is especially valuable when terminology or a branch condition is contested. The evidence of completion is: The French branch and all linked hotspots remain understandable without language switching. The general lesson is: Connected assets need independent sign-off. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
3. Original text changed later
A safety training owner revises an English instruction after Spanish tags were approved. Test from the learner’s point of view rather than the editor’s. The correct result is the ability to interpret evidence and make the intended decision using the available controls. The risk is specific: The old translation may not refresh and may display a mixture of source and target languages. The practical intervention is: Mark changed segments and explicitly refresh or edit the target versions, then retest conditions. Repeat the learner path after correcting the issue, because a valid text change may reveal a separate visual or navigation problem. The evidence of completion is: Every released language shows the new instruction and the same required behavior. The general lesson is: Maintenance is part of localization. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
4. Wrong proper noun
An art-gallery tag identifies a named indigenous artwork and its creator. Preserve the invariant: a factual measurement, place identity, direction, permission, chronology or learning objective. Natural target wording may differ substantially while that invariant remains constant. The risk is specific: Automatic translation may substitute a generic noun or alter an official cultural name. The practical intervention is: Consult the curator’s terminology record and manually correct protected names in each language. Check this result in the actual release link rather than assuming the authoring preview is identical to the visitor experience. The evidence of completion is: The object’s identity and attribution match the authoritative catalogue. The general lesson is: Fluency cannot override provenance. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
5. Formatted warning loses emphasis
A translated first-aid tag no longer bolds the words describing an emergency exception. A translation may look fluent while undermining the task. Review the intended action, then check how the words interact with the scene, video, hotspot or decision. The risk is specific: ThingLink warns translated rich text can lose formatting that originally signalled priority. The practical intervention is: Inspect emphasized text and restore or replace clear emphasis without relying on colour alone. Invite a reader who has not memorized the source tour; familiarity can conceal target-language confusion. The evidence of completion is: A learner can distinguish warning, instruction and optional note on the target screen. The general lesson is: Formatting can carry safety meaning. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
6. 360 compass direction
A virtual field trip directs students to the northern ridge of a landscape. Look for the failure after the first click, not only before it. In multimedia learning, the hardest-to-find untranslated instruction often appears one step beyond the screen that receives the most review. The risk is specific: A right-to-left interface choice could be mistaken for a reason to reverse geographic direction. The practical intervention is: Keep actual compass relationships and validate target-language labels in their scene positions. Document any remaining supported-feature limitation honestly and supply an alternate learning route when required. The evidence of completion is: All learners identify the same physical ridge regardless of reading direction. The general lesson is: Geography is invariant under localization. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
7. Language-specific embed
A bilingual training portal embeds one language edition of an interactive image. Because this information is reached through an interaction, a translator who reads only the opening page will not see the problem. Trace the exact click sequence and evaluate the words in context. The risk is specific: A copied embed may accidentally open the default source version instead. The practical intervention is: Select the intended language in the share workflow and test the copied code on the destination page. Record the corrected wording, scene identifier and tested language so the same issue does not return during the next update. The evidence of completion is: The embed starts in the advertised language with the expected tags available. The general lesson is: Distribution settings form part of the translation. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
8. Visible language selector
A public-health outreach tour lets visitors switch languages themselves. Separate a language defect from a delivery defect. The text may be accurate but attached to the wrong branch, hidden by formatting or opened through a misleading link. The risk is specific: Visitors may find a selector but still encounter untranslated content behind some tags. The practical intervention is: Confirm all essential scenes and scenario blocks before advertising a language as complete. Keep a short source-to-target decision note; this is especially valuable when terminology or a branch condition is contested. The evidence of completion is: A first-time visitor switches languages and completes the same route successfully. The general lesson is: A selector is a promise of usable content. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
9. Hidden language selector
A school prefers separate direct links rather than a public language menu. Test from the learner’s point of view rather than the editor’s. The correct result is the ability to interpret evidence and make the intended decision using the available controls. The risk is specific: Hiding a selector without correct direct links can trap learners in the wrong language. The practical intervention is: Provide clearly labelled language-specific links and verify viewer access permissions. Repeat the learner path after correcting the issue, because a valid text change may reveal a separate visual or navigation problem. The evidence of completion is: Learners reach their assigned edition without an editor account or guesswork. The general lesson is: Frictionless access reduces avoidable exclusion. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
10. Clone of translated scene
A trainer clones a previously approved bilingual safety scene for a new cohort. Preserve the invariant: a factual measurement, place identity, direction, permission, chronology or learning objective. Natural target wording may differ substantially while that invariant remains constant. The risk is specific: Cloned translations may include outdated policies or examples even if copies retain language data. The practical intervention is: Treat the clone as a new revision and compare its text, hazards and language states with current policy. Check this result in the actual release link rather than assuming the authoring preview is identical to the visitor experience. The evidence of completion is: The clone reflects the newly approved instructions in every published language. The general lesson is: Reuse does not remove review responsibility. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
11. Multiscene tour gap
A geography tour includes six locations but only the opening scene is translated. A translation may look fluent while undermining the task. Review the intended action, then check how the words interact with the scene, video, hotspot or decision. The risk is specific: A user may assume the language setting applies universally to all scenes. The practical intervention is: Make a per-scene completeness matrix and test the full tour from start to finish. Invite a reader who has not memorized the source tour; familiarity can conceal target-language confusion. The evidence of completion is: A learner can travel through all required scenes in the selected language. The general lesson is: Coverage is a route-level property. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
12. 3D model scene limitation
A technical museum experience includes an interactive 3D machinery model. Look for the failure after the first click, not only before it. In multimedia learning, the hardest-to-find untranslated instruction often appears one step beyond the screen that receives the most review. The risk is specific: The official standard-scene feature does not promise translation of 3D model scenes. The practical intervention is: Provide an approved alternate explanation or supported language route and state any remaining limitation. Document any remaining supported-feature limitation honestly and supply an alternate learning route when required. The evidence of completion is: The learner receives enough accessible target-language information to meet the objective. The general lesson is: Unsupported media must be handled explicitly. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
13. Captioned interview
A community history tour embeds a video interview in the original language. Because this information is reached through an interaction, a translator who reads only the opening page will not see the problem. Trace the exact click sequence and evaluate the words in context. The risk is specific: Translated hotspots cannot be assumed to translate recorded speech or subtitles. The practical intervention is: Provide reviewed captions, an alternate transcript or a suitable localized interview asset. Record the corrected wording, scene identifier and tested language so the same issue does not return during the next update. The evidence of completion is: A muted or non-source-language viewer can understand the evidence cited by the tour. The general lesson is: Media language is distinct from tag language. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
14. Image text embedded in photo
A science exhibit photograph contains printed labels on the apparatus. Separate a language defect from a delivery defect. The text may be accurate but attached to the wrong branch, hidden by formatting or opened through a misleading link. The risk is specific: A text translation function cannot be presumed to rewrite pixels in the image. The practical intervention is: Prepare a localized graphic or a clear tag and accessible description explaining the labels. Keep a short source-to-target decision note; this is especially valuable when terminology or a branch condition is contested. The evidence of completion is: The learner identifies the correct instrument and component without source-language literacy. The general lesson is: Pixels may contain curriculum. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
15. Accurate measurement unit
A weather tag states rainfall was measured in millimetres during a particular period. Test from the learner’s point of view rather than the editor’s. The correct result is the ability to interpret evidence and make the intended decision using the available controls. The risk is specific: An automatic draft may confuse measurement unit, scale or interval. The practical intervention is: Preserve numeric values and unit meaning in a locked reference, then review the target explanation. Repeat the learner path after correcting the issue, because a valid text change may reveal a separate visual or navigation problem. The evidence of completion is: The learner reports the same physical observation in each language. The general lesson is: Quantities must survive translation. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
16. Branching safety condition
An industrial tour asks whether an emergency door may be opened before authorization. Preserve the invariant: a factual measurement, place identity, direction, permission, chronology or learning objective. Natural target wording may differ substantially while that invariant remains constant. The risk is specific: A missed negative or conditional phrase can reverse the safe course of action. The practical intervention is: Use bilingual safety review and test both the permitted and prohibited choices. Check this result in the actual release link rather than assuming the authoring preview is identical to the visitor experience. The evidence of completion is: The chosen branch reflects the same safety rule across languages. The general lesson is: Decision rules require subject review. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
17. Scenario wrong-choice feedback
A learner takes an incorrect route through a biodiversity decision exercise. A translation may look fluent while undermining the task. Review the intended action, then check how the words interact with the scene, video, hotspot or decision. The risk is specific: The target feedback may shame the learner or explain the wrong biological mechanism. The practical intervention is: Review the pedagogical reasoning and tone of every unsuccessful outcome. Invite a reader who has not memorized the source tour; familiarity can conceal target-language confusion. The evidence of completion is: The learner understands the error and can try again with corrected evidence. The general lesson is: Feedback must teach, not merely label. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
18. Immersive Reader limitations
A multilingual learner uses reading assistance while exploring an image-based narrative. Look for the failure after the first click, not only before it. In multimedia learning, the hardest-to-find untranslated instruction often appears one step beyond the screen that receives the most review. The risk is specific: The presence of reading support does not mean every embedded text and media element is accessible. The practical intervention is: Test the supported reading mode and provide alternative descriptions where needed. Document any remaining supported-feature limitation honestly and supply an alternate learning route when required. The evidence of completion is: A learner can access the central evidence through available assistive pathways. The general lesson is: A feature badge is not a complete accessibility audit. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
19. Accessibility Player navigation
A keyboard user needs to find tags in a complex 360-degree environment. Because this information is reached through an interaction, a translator who reads only the opening page will not see the problem. Trace the exact click sequence and evaluate the words in context. The risk is specific: A visually obvious hotspot may be hard to reach or identify without a pointer. The practical intervention is: Test the Accessibility Player and keyboard workflow in the target publication. Record the corrected wording, scene identifier and tested language so the same issue does not return during the next update. The evidence of completion is: The user can locate required content and return to the tour without confusion. The general lesson is: Navigation equality matters alongside translation. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
20. Alternative description
A graph image illustrates a rapid increase in particulate concentration. Separate a language defect from a delivery defect. The text may be accurate but attached to the wrong branch, hidden by formatting or opened through a misleading link. The risk is specific: An alt label saying ‘chart’ does not communicate the trend or measurement. The practical intervention is: Write an accurate target-language description of the relationship and important values. Keep a short source-to-target decision note; this is especially valuable when terminology or a branch condition is contested. The evidence of completion is: A screen-reader user can answer the same question about the graph. The general lesson is: Descriptions should convey evidence. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
21. Long text in narrow tag
An expanded target-language paragraph appears in a small hotspot popup on a phone. Test from the learner’s point of view rather than the editor’s. The correct result is the ability to interpret evidence and make the intended decision using the available controls. The risk is specific: The popup may require excessive scrolling or obscure the visual reference it explains. The practical intervention is: Simplify structure, preserve the factual concept and check legibility at mobile size. Repeat the learner path after correcting the issue, because a valid text change may reveal a separate visual or navigation problem. The evidence of completion is: A learner reads the whole instruction and still sees the referenced scene. The general lesson is: Layout affects meaning. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
22. Unapproved machine text
A company training scenario includes a live customer case and identifiable employee details. Preserve the invariant: a factual measurement, place identity, direction, permission, chronology or learning objective. Natural target wording may differ substantially while that invariant remains constant. The risk is specific: Sending protected examples to a translation feature may breach organizational policy. The practical intervention is: Use fictionalized cases or obtain approval for the actual processing route before translation. Check this result in the actual release link rather than assuming the authoring preview is identical to the visitor experience. The evidence of completion is: The lesson tests the same decision without exposing unnecessary personal information. The general lesson is: Data minimization accompanies responsible localization. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
23. Regional language variety
A public exhibit targets speakers of different regional varieties of Portuguese. A translation may look fluent while undermining the task. Review the intended action, then check how the words interact with the scene, video, hotspot or decision. The risk is specific: One version may sound unfamiliar or imply a different public agency or service. The practical intervention is: Choose the required locale and validate terms with a reviewer familiar with the intended audience. Invite a reader who has not memorized the source tour; familiarity can conceal target-language confusion. The evidence of completion is: Visitors recognize the same institution and procedure without ambiguous terminology. The general lesson is: Language labels are more than national flags. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
24. Mixed-script label
An Arabic industrial tour uses Latin machine identifiers beside right-to-left instructions. Look for the failure after the first click, not only before it. In multimedia learning, the hardest-to-find untranslated instruction often appears one step beyond the screen that receives the most review. The risk is specific: Identifiers can appear rearranged or be mistaken for words to translate. The practical intervention is: Keep machine codes invariant and test rendering direction around labels and numbers. Document any remaining supported-feature limitation honestly and supply an alternate learning route when required. The evidence of completion is: A learner reads the correct identifier and follows the intended sequence. The general lesson is: Codes should remain stable. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
25. Tour jump link
A hotspot opens another scene through a link whose label was localized. Because this information is reached through an interaction, a translator who reads only the opening page will not see the problem. Trace the exact click sequence and evaluate the words in context. The risk is specific: The label may promise an external source while actually opening a second internal scene. The practical intervention is: Test the destination and use a verb that reflects what clicking does. Record the corrected wording, scene identifier and tested language so the same issue does not return during the next update. The evidence of completion is: Visitors correctly predict whether they will enter another scene or leave the tour. The general lesson is: Navigation text must be truthful. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
26. Scene title in catalogue
A school distributes multilingual tour links through a shared learning portal. Separate a language defect from a delivery defect. The text may be accurate but attached to the wrong branch, hidden by formatting or opened through a misleading link. The risk is specific: Multiple language variants have the same English title and confuse learners. The practical intervention is: Label each released route with a clear language and revision in the portal. Keep a short source-to-target decision note; this is especially valuable when terminology or a branch condition is contested. The evidence of completion is: A student opens the intended version on the first attempt. The general lesson is: Findability is part of publication. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
27. Source scene and scenario drift
An English scene is revised but its existing scenario contains older branching text. Test from the learner’s point of view rather than the editor’s. The correct result is the ability to interpret evidence and make the intended decision using the available controls. The risk is specific: Even within one language, route and scene claims can contradict each other. The practical intervention is: Compare scenario logic and associated scene content before refreshing each translation layer. Repeat the learner path after correcting the issue, because a valid text change may reveal a separate visual or navigation problem. The evidence of completion is: Every branch uses consistent current facts across scenes and languages. The general lesson is: A translation register should map dependencies. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
28. Automated translation overtrust
A beautiful German draft uses the wrong term for a safety inspection. Preserve the invariant: a factual measurement, place identity, direction, permission, chronology or learning objective. Natural target wording may differ substantially while that invariant remains constant. The risk is specific: Natural phrasing hides a material operational difference from the source. The practical intervention is: Have a specialist reviewer simulate the actual work decision and correct the term. Check this result in the actual release link rather than assuming the authoring preview is identical to the visitor experience. The evidence of completion is: The worker chooses the intended inspection rather than a different process. The general lesson is: Expert verification outranks fluency. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
29. Review ticket without location
A tester reports that the Malay version is confusing somewhere in the virtual tour. A translation may look fluent while undermining the task. Review the intended action, then check how the words interact with the scene, video, hotspot or decision. The risk is specific: The editor cannot reproduce a vague complaint across dozens of tags and scenes. The practical intervention is: Capture project, scene, tag, language, source revision and screenshot with the defect. Invite a reader who has not memorized the source tour; familiarity can conceal target-language confusion. The evidence of completion is: A second reviewer reproduces and confirms the correction quickly. The general lesson is: Precise evidence improves quality control. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
30. New owner inherits the tour
A museum curator leaves after releasing five language editions and new staff need to update them. Look for the failure after the first click, not only before it. In multimedia learning, the hardest-to-find untranslated instruction often appears one step beyond the screen that receives the most review. The risk is specific: No one knows which original scene or translated tag matches the changed exhibit. The practical intervention is: Archive the scene and scenario map, language settings, manual corrections and link register. Document any remaining supported-feature limitation honestly and supply an alternate learning route when required. The evidence of completion is: The next curator can update the facts and release all target routes safely. The general lesson is: Sustainable translation survives staffing changes. This is not merely a wording preference; it determines whether the intended audience can accomplish the learning task from start to finish.
Four-week implementation plan
Week 1 — Structure and source audit
Map each image, 360 media scene, video, scenario block and linked resource. Choose the target languages and verify the account’s translation eligibility. Play every branch in the original language, correct contradictory facts and define the decisions learners should make. Prepare a glossary for technical terms, official names, safety conditions and measurements. Assign a language reviewer and a subject reviewer, and record which assets will not be covered by automatic translation. Produce a short baseline test with screenshots and the source revision.
Week 2 — Translation at both content levels
Translate supported scenes using the platform’s source and target language selection and separately translate each scenario. Open hotspots, inspect text that appears only after a particular decision, and correct awkward or inaccurate translations through the documented manual route. Compare scene titles, tag labels, numbers, place names, links and formatted warnings. Prepare separate target captions, graphic labels or alternatives for embedded media and unsupported 3D scenes. Confirm that each required scene and each scenario branch is represented in the completeness matrix.
Week 3 — Accessible full-route testing
Run the project as a visitor in each target language, starting with a fresh browser or learner account. Follow a correct and an incorrect branch. Test the language selector or direct-language link as the project requires. Examine mobile popups, keyboard navigation, Accessibility Player behavior, descriptions, captions, mixed-direction scripts and long strings. Ask a reader unfamiliar with the source tour to explain a key piece of evidence. Fix the most consequential errors before spending time on minor visual polish. Record exact reproduction steps and retest.
Week 4 — Publish, share and maintain
Release approved languages with the exact viewer links or embed codes that were tested. Check access permissions and verify the intended initial language. Give teachers or curators a link register and a clearly labeled list of known limitations. Establish a change rule: any substantive source edit requires a check of the corresponding translated scene or scenario, because the system does not automatically update every translated string. Archive language-specific corrections, reviewer approvals and the source-to-target dependency map so future staff can maintain the experience reliably.
Frequently asked questions
Does ThingLink translate interactive scenes?
Its official help describes automated translation for eligible supported image, video and 360 scene types on specified plans. The operational answer should always be confirmed in the account and project being released. Test one representative path before extending the process to a complete multilingual tour; a completed translation command is not the same as an approved educational experience.
Does translating a scenario also translate its scenes?
No. The official scenario guide explicitly says its constituent ThingLink scenes need separate translation. The operational answer should always be confirmed in the account and project being released. Test one representative path before extending the process to a complete multilingual tour; a completed translation command is not the same as an approved educational experience.
How many languages are supported?
Vendor documentation describes more than 40 languages, with an enumerated list that may change. Confirm your exact locale and plan. The operational answer should always be confirmed in the account and project being released. Test one representative path before extending the process to a complete multilingual tour; a completed translation command is not the same as an approved educational experience.
Will revised source text automatically retranslate?
No. The official guides say to refresh translations after editing source text. The operational answer should always be confirmed in the account and project being released. Test one representative path before extending the process to a complete multilingual tour; a completed translation command is not the same as an approved educational experience.
Can translators correct machine drafts?
Yes. ThingLink documents manual correction workflows for translated text. The operational answer should always be confirmed in the account and project being released. Test one representative path before extending the process to a complete multilingual tour; a completed translation command is not the same as an approved educational experience.
Does formatting always survive translation?
Not necessarily. The help articles warn that rich-text formatting may be lost. The operational answer should always be confirmed in the account and project being released. Test one representative path before extending the process to a complete multilingual tour; a completed translation command is not the same as an approved educational experience.
Can I link directly to a selected language?
ThingLink documents language-specific links and embed options in addition to the viewer language selector. The operational answer should always be confirmed in the account and project being released. Test one representative path before extending the process to a complete multilingual tour; a completed translation command is not the same as an approved educational experience.
Are 3D model scenes covered?
The standard scene translation guide says 3D model scenes cannot yet be translated through that feature. The operational answer should always be confirmed in the account and project being released. Test one representative path before extending the process to a complete multilingual tour; a completed translation command is not the same as an approved educational experience.
How do I handle accessibility?
Test the actual Accessibility Player, reader support, alt text, captions and keyboard behavior for each target edition. The operational answer should always be confirmed in the account and project being released. Test one representative path before extending the process to a complete multilingual tour; a completed translation command is not the same as an approved educational experience.
What belongs in the maintenance register?
Keep source and target revision, scene/scenario relationships, corrected terms, approved languages, viewer links and retest outcomes. The operational answer should always be confirmed in the account and project being released. Test one representative path before extending the process to a complete multilingual tour; a completed translation command is not the same as an approved educational experience.
Official references and architecture links
These vendor sources describe the features and current boundaries used in this guide. Plans, controls and supported languages may change; verify the present account configuration.
- ThingLink: translating media projects (updated July 21, 2026)
- ThingLink: translating scenarios (updated September 29, 2026)
- ThingLink: accessibility features and Immersive Reader
Related eduKateSG material: Master Art of Translation — Education and Learning Systems; Translate Like a Pro 4350 — Genially; Translate Like a Pro 4300 — Adapt Learning. This article has a distinct scope: ThingLink’s scene-and-scenario structure, multilingual viewers and the requirement to synchronize separately translated project components.
The professional objective is a visitor who can enter a scene, find evidence, choose an appropriate route and explain the meaning in their own language. A multilingual interface is valuable because it makes that sequence possible; the ultimate quality measure is the learner’s independent understanding.