If your center teaches English, French, and Arabic, you live a reality most education software ignores: your daily work crosses two writing directions, three keyboard layouts, and student populations who navigate interfaces in different languages.
Most assessment tools were designed in English, for English, with other languages added as translation files. The difference shows up exactly where it hurts: in the assessments themselves. Here is what multilingual assessment actually requires.
RTL is a layout, not a translation
Arabic support is often advertised as "Arabic interface available." That covers the menus. The real test is the assessment content:
- Does a fill-in-the-blank question flow right-to-left, with blanks in the right positions?
- Do matching exercises mirror correctly, so the logical "first column" is on the right?
- When a question mixes Arabic text with a French word or a number, does the bidirectional text render in the right order?
- Is the timer, the progress bar, the navigation all mirrored?
Open a demo account and build one real Arabic exercise before believing any feature list. Five minutes of building reveals what no sales page will: misplaced punctuation, blanks that jump to the wrong side, and mixed text that scrambles.
Mixed-language questions are the norm, not the edge case
Language teaching constantly crosses languages inside a single question. Instructions in French for an English exercise; an Arabic question quoting an English sentence to translate; vocabulary items pairing scripts.
This means the unit of language is the text run, not the document. A tool that forces "this assessment is in Arabic" as a single global setting fights how language teachers actually write. What you want is the ability to mix scripts freely inside one question, with each part rendering correctly, the same way it does in a well-formatted Word document.
The interface language is the student's choice, not the content's
A French course taken by an Arabic-speaking student should show French content inside an Arabic interface, if that is what the student understands best. Buttons like "Next question" and "Submit" are not part of the pedagogy; they should be in whatever language reduces friction.
Decoupling content language from interface language sounds obvious, yet plenty of platforms bind them together. For centers in the Maghreb and the Gulf, where a single classroom can include francophone, anglophone, and arabophone digital habits, the decoupling matters every single session.
Grading across languages
Two practical points that surface late and painfully:
- Accents and hamzas. Auto-grading must let the teacher decide whether "eleve" matches "élève" and how strictly to treat hamza variants. There is no universally right answer: a beginner course may tolerate what an advanced course penalizes. The tool must make it a setting, not a hard-coded opinion.
- AI grading needs to work per language. If you use AI evaluation for open answers, verify it grades French and Arabic responses with the same care as English, and that your rubrics can be written in the language of the course. A model grading Arabic answers through the lens of English-language training is a quiet failure mode worth testing explicitly.
The shortlist test
When evaluating any platform, run this 20-minute test: build one exercise per language, including one mixed-language question; take each as a student with the interface in a different language; and grade one open answer in each language. Most of the market fails before the third step.
Multilingual is not a feature checkbox. For centers like yours, it is the product. Choose accordingly.