How do I translate a SCORM course into another language?

You can either export the text for a translator and reassemble the package afterwards, or use an authoring tool that translates the course directly. The second route avoids the two most common failures: a package corrupted during reassembly, and an interface left in the source language. Whichever you choose, the buttons, menus, module names and quiz feedback all have to change as well as the body text.

Why does my translated course look broken?

Almost always layout rather than language. Translated text is a different length to the source, so headings wrap and overflow their containers. Right-to-left languages such as Arabic and Hebrew invert the whole layout, not just the words, so a course that only mirrors the text ends up with navigation on the wrong side. Check a translated course at phone width, where there is least slack.

Should all languages be in one SCORM package or separate ones?

Separate packages, in almost every case. A single package with a language picker means the LMS sees one course with one completion status and one score, so you cannot report on how learners performed per language, and every learner downloads every language. One package per language, assigned to each audience, keeps your reporting meaningful.

Why is text inside my images still in the original language?

Translation tools extract text from HTML, XML and JSON, not from pixels. Any wording baked into a diagram, chart or screenshot has to be recreated per language by hand. The practical fix is preventative: keep text out of images and put it in a caption or a labelled list beside the image, so it travels with the translation.

Does translating a course break SCORM tracking?

It can. A SCORM package holds HTML, JavaScript, a manifest and media in a specific structure, and tools that unpack and reassemble it may corrupt character encoding, the manifest or the tracking calls. The symptoms are subtle: quizzes display but never report, or completion never registers. Always upload a translated course to your LMS, complete it, and check the report before signing it off.

How to Translate a SCORM Course Into Another Language (Without Breaking It)

Translating a course sounds like a text problem. Export the words, send them to a translator, put them back.

Then the Arabic version arrives and the navigation is on the wrong side. The German headings overflow their buttons. The quiz still reports, but every question shows as “Response 1” in your LMS. And three months later someone updates one slide in the English master, and you rebuild all nine languages by hand.

eLearning translation is not a text problem. It is a layout, packaging and maintenance problem that happens to involve text. Here is what breaks, in the order you will hit it.

1. The layout breaks before the language is even wrong

This is the first thing to go, and the most expensive to fix late.

Translated text is rarely the same length as the source. German and Finnish typically run longer than English; Chinese and Japanese run shorter. A heading that fits neatly on one line in English wraps to three in German, pushing everything below it out of place.

Then there is direction. Arabic, Hebrew, Persian and Urdu read right to left — and that is not a text setting, it is a layout inversion. Navigation moves to the other side. Progress bars fill the other way. Icons that imply forward motion now point backwards.

iSpring’s own localization guide is blunt about the consequence: translated content often does not fit the original layout because of differing text volume and orientation, and an otherwise correctly translated text can ruin the appearance of your entire course. Their recommended fix is extra work from your design team, with the delays and cost that implies.

What to check: open the translated course at a phone width, not just on your monitor. Most layout failures show up first at narrow widths, where there is no slack for longer strings.

2. Text baked into images does not get translated

Standard translation tools extract text from HTML, XML and JSON. They cannot extract text that is part of a picture — a labelled diagram, a screenshot of software, a process chart exported from PowerPoint.

That text stays in the source language while everything around it changes. Learners notice immediately, and it reads as carelessness rather than a technical limitation.

There is no clever fix. Someone has to recreate those images per language, and a flat per-word quote will not cover it. The practical answer is preventative: keep text out of images wherever you can. Use a caption below the image, or a labelled list beside it, so the words live in the page and travel with the translation.

3. The package breaks quietly

This is the failure that survives review and reaches learners.

A SCORM package is a bundle of HTML, JavaScript, a manifest and media, held together by a specific structure. Generic translation tools that unpack and reassemble it can corrupt exactly the parts your LMS relies on: character encoding, the manifest, tracking calls.

The symptoms are specific and easy to miss in a quick check — navigation that no longer advances, quizzes that display but never report, completion that never registers, or interaction data that arrives as unlabelled “Response 1, Response 2” columns.

What to check: do not sign off a translated course by clicking through it. Upload it to your LMS, complete it as a learner, and look at the report. Confirm the score arrives, completion registers, and question-level data is readable.

4. Then the source changes, and you do it all again

The first translation is the cheap part. The maintenance is where budgets disappear.

Every language you ship is another artefact to update. Change one policy in the English master and someone must re-translate that section, repackage, re-test and re-upload nine files. Miss one, and a market is training on outdated compliance content — which is worse than not translating it at all.

This is why localization costs are usually underestimated. A quote covers the first pass. The second year is the real number.

Should all languages live in one package?

You will see tools that bundle every language into a single SCORM file with a language picker, and it sounds tidier: one upload, one course record.

In practice it creates a reporting problem. The LMS treats that package as one course with one completion status and one score, so you cannot tell how learners performed in Arabic versus Spanish, or which languages people actually chose. File size multiplies too, since every learner downloads every language.

Colossyan — who build a SCORM translation tool — put it plainly: technically it works, but almost nobody does it, and the standard approach is one package per language assigned to each audience. That is worth knowing before a single-file workflow locks you into reporting you cannot use.

One course per language is not a limitation. It is what keeps your LMS reporting meaningful.

What a good translation workflow actually looks like

Whatever tool you use, these are the properties worth insisting on:

The interface travels with the content. Buttons, menus, module names and quiz feedback all have to change, or you ship translated content wrapped in English chrome. This is the single most common half-finished translation.
Direction is handled as layout, not text. Arabic should invert the course, not mirror the words inside a left-to-right shell.
The export is a real SCORM package, with language and direction written into it, so the LMS and screen readers know what they are handling.
The result is editable. Machine translation is a strong first draft, not a final one. You need to be able to fix a term without going back to the vendor.
Re-translating is cheap. If updating the source means starting the whole cycle again, you will stop updating it.

Doing it in SCORMBuilder

In SCORMBuilder.ai, translation is a button in the editor. Pick from 35+ languages and it builds a complete, editable copy of the course in that language — body text, headings, lists, quiz questions and options, module and lesson names, and every button in the player.

Right-to-left languages invert the whole course, not just the words, and the language and direction are written into the exported package so your LMS and assistive technology read it correctly. The copy is a normal course: open it, change a phrase, adjust a quiz, export to SCORM 1.2, 2004, xAPI or cmi5. Your original is untouched.

Two things it deliberately does not do. Narration is not carried over — an Arabic course playing English audio is worse than one with no audio, so you regenerate it in-language. And puzzle answers stay in the source language, because a translated crossword answer cannot fill the same grid; you are told which blocks need rebuilding rather than discovering it later.

It is included in the $29/month plan, so translating into a second language costs nothing extra — which matters, because the point is not translating once. It is being able to keep nine languages current without dreading the next update.