Multilingual training
A training can exist in several languages at once. Each language is its own version of the same training, linked back to the language it was translated from, so you keep one course with many language variants, not many separate courses.
This page explains how to create translations, how to control which languages a course allows, and how to see at a glance which translations are complete, which are missing, and which have fallen out of date.
How learners get their language
Learners read a training in the language of the app they are using. When a learner opens a training, the system looks for an approved version in their language:
- If one exists, the learner reads that version.
- If none exists, the learner reads the assigned version and sees a banner explaining that no version is available in their language yet.
Either way, the completion record (evidence) always stores the exact version the learner actually read. The app's language switcher shows a small check next to each language that has content, so a learner can see what is available before choosing.
The training detail page
Opening a training from the Training materials list takes you to its detail page, the home for everything about that course. It is one scrolling surface, built the same way as a controlled document's detail page, so both read alike:
- A properties summary at the top shows the reviewer's at-a-glance facts (source language, recertification, category, last updated). The labelled Show all properties button uses the same panel icon and tooltip as the controlled-document detail page, and opens the full properties drawer (allowed languages, AI translation, scheduling, and identifiers). If any version is awaiting approval or a translation has gone out of date, a small indicator links straight down to the version list.
- The Versions section is the heart of the page. It lists one card per revision, newest first, each collapsed by default. A card shows the version number, its review state, and, when your organization uses more than one content language, a small coverage chip telling you how many languages of that revision are ready, so you can tell at a glance whether a revision is fully translated without opening it. Create new version sits next to the Versions heading, because creating a version is something you do to the version history rather than to the training as a whole. It opens the authoring flow, and what it produces is a new version awaiting review, never a change to the version learners are taking.
- The Used in section lists the campaigns that assign this training.
Language is a property of each revision, not of the training, so there is no page-level Languages control. You work with languages inside the card of the revision you are translating, as described next.
Languages live inside each version card
There is no separate Languages tab. Expand a revision card and its languages appear inside it, one chip per content language your organization has enabled, with the source language marked. Pick a chip to work with that language:
- Preview opens a read-only view of that language's content, exactly as a learner sees it.
- A language you have not created yet shows as missing with a Translate to {language} button. A translation that has fallen out of date is flagged so it can be reviewed and re-translated.
- A translation still awaiting approval shows an Approve action for a reviewer; an approved-but-inactive version shows Activate.
Picking Translate to {language} launches the translation workspace described next.
Creating a translation
There are two ways in:
- On the training's detail page, expand a revision card and click Translate to {language} on a missing language (or the re-translate action on an out-of-date one).
- From the Training materials list, open the menu on a training and choose Translate with AI.
Either way a focused translation workspace opens. It is a single-purpose screen built for translating, not the full authoring flow:
- The source content sits on the left, read-only, so you translate against it directly.
- The target language sits on the right, with the AI's draft already in place. You edit each region (Introduction, Closing summary, and the Assessment) and can re-run Translate with AI on any region.
- Write a change reason (or let the AI draft one) and send for review.
The AI draft is a starting point, not the final word. A translation is not live until a different person approves it in the Quality Review Queue (segregation of duties), exactly like an original-language version.
Document-anchored trainings follow the controlled document
If the training's content is a controlled document (it shows the document verbatim), its languages are tied to that document. Translating into a language the document already has an official, effective variant in pins that variant, so the learner reads the effective translated SOP, not a machine re-translation. The body of the document is never machine-translated; only the Introduction, Closing summary, and Assessment are. If the document has no effective variant in that language yet, the workspace blocks and tells you to translate the source document first, with a link straight to it. When the source document already has that language approved but not yet released, the message instead asks you to make the document variant effective first, so you know the difference between "the translation does not exist" and "it exists but has not been published". This keeps the training and the document in lock-step.
When the source document gains a language the anchored training does not have yet, the language coverage matrix highlights that gap so you know the training can be translated right away (the official variant already exists to pin). See below.
You do not need the matrix to notice this. On the training's detail page, expand a revision card and pick a language the training does not have yet. For a document-anchored training the card reads the source document's languages and tells you exactly where you stand:
- The source document already has that language effective, so the card offers Translate to {language} and notes that translating pins the official variant.
- The source document does not have that language, so the card replaces the translate button with a link to Open source document, and tells you to translate the source document first.
- The source document has that language approved but not yet effective, so the card tells you to make the document variant effective first, again with a link to the document.
This is why a training can still show a language as missing even though the controlled document it is built from already has that language: the two are separate records, and the training needs its own version. The card now points you straight to the next step.
Restricting languages per course (governance)
Some courses are security-relevant and must not be machine-translated, or should only ever be published in a fixed set of languages. Open Manage course languages from the header language bar's overflow menu (⋮):
- Allowed languages - tick the languages this course may be published in. Leave it empty to allow all enabled languages. When set, the language bar only offers those languages.
- Disable AI translation - turn this on for courses where machine translation is not trusted. The "Translate with AI" action disappears for that course, and the system refuses any AI translation request for it.
These rules are enforced by the system, not just hidden in the interface. An AI translation into a language the course does not allow is rejected.
Seeing coverage and out-of-date translations
The Training materials list shows a row of language chips on each training so you can see at a glance which languages it is published in. A chip turns into a warning when that translation has gone out of date. This happens automatically when the master version's sources change, so the translation needs to be reviewed and re-translated.
Use the Needs attention filter at the top of the list to show only trainings with an out-of-date translation. A training that needs attention carries a Needs attention chip in both the list and its detail page. In the list, hover or focus the chip to see exactly why, such as which translations are out of date and which source-document languages have not been translated yet. On the detail page the chip opens a short panel with the same reasons plus a shortcut to the version list, where you translate or review each language, and a link into the Help Center.
For the full picture, open Language coverage from the toolbar (or visit Training materials → Language coverage). This is a matrix of every training against every supported language, showing whether each language is effective, approved, pending review, or missing, and which ones are out of date. The cells are clickable: open a published language in its detail page, or start a translation for a missing or out-of-date one straight from the matrix. You can also export the matrix to CSV for an auditor or a translation plan.
For a document-anchored training, the matrix also folds in the languages of its source document. When the document already has an effective variant in a language the training has not been translated into, that cell is highlighted as ready to translate, and the training is flagged under Needs attention. The CSV export lists these source languages so you can plan the outstanding training translations.
Tracking completions in a fallback language
When a learner completes a training in a language other than their own (because no approved version existed in their language), the system records it. You can review these on the Reports page under the By language tab, and export them to CSV. This helps you prioritise which translations to create next.
Related pages
- Settings and scheduling - Signature, recertification, and scheduling controls
- Versions and recertification - How versions accumulate
- Review and submit - Submitting for approval
- Evaluation dashboards - The By language report