Conducting Multilingual Surveys

When your participants speak different languages, the language of your survey isn't a detail — it's a participation barrier. A francophone community member who receives an English-only invitation to an English-only survey is less likely to respond, and if they do, the quality of their responses may suffer. A Punjabi-speaking parent asked to evaluate a school program in English may not participate at all.

This guide covers how to build surveys that reach people in their preferred language, from the survey itself through the email invitation that delivers it.

What the platform supports

The platform supports 10 languages for survey content: English, French, Spanish, Punjabi, Mandarin Chinese, Cantonese, Tagalog, Arabic, German, and Italian. Your platform interface runs in English or French, but the surveys and polls your participants see can be in any of these languages.

The architecture is language-agnostic by design. Participants answer questions by selecting options that have predetermined identifiers. A French participant choosing "Bleu" and an English participant choosing "Blue" are selecting the same option — the data is identical. Your analytics work the same regardless of which language the responses came from.

Adding languages to a survey

After creating your survey in the builder, open the Languages tab in the settings panel. You'll see your survey's default language (the one you wrote it in) and the ability to add additional languages.

When you add a language:

Machine translation runs immediately. Every translatable field — the survey name, description, question text, option text, block headings, consent text — is translated automatically. This gives you a working draft in seconds.
You can optionally assign a translator. If you have team members with the TRANSLATOR role, you can assign one to review and refine the machine translation. The platform filters the picker to show only translators who have declared that language as one they work in.
You can also assign a reviewer. For surveys where translation accuracy matters (research instruments, consent text, anything with legal implications), you can add a second person who reviews the translator's work before the translation is considered complete.

Machine translation is good enough for many contexts — internal surveys, quick polls, community feedback. But for research instruments where precise wording affects validity, and especially for consent text where accuracy has legal implications, human review matters.

The translation workflow

When you assign a translator, a structured workflow begins:

The translator receives a notification — an in-app notification and an email with a direct link to the translation workspace.

The workspace is a dedicated full-page editor. It shows each translatable segment with the source text on top and the target text below, grouped by survey page. The translator can edit the machine translation, approve it as-is, or rewrite it entirely. Keyboard shortcuts (Tab between segments, Ctrl+Enter to save) make the work efficient.

Consent text gets special treatment. The consent page is translated whole: the heading, explanation and purpose you wrote, the platform's own sentences on that page, and each optional consent's label and description all appear as segments, and each carries a warning: "This text has legal implications. Professional translation is recommended. It is also what the participant's consent record will certify." The warning does not stop a machine draft from being published; publishing does stop when a language you offer has no consent text at all, because consent never falls back to another language. What is published in each language is what the consent record stores, word for word — and what the participant reads: the page is drawn from the recorded text, not from a separate catalogue. The platform's own sentences on that page — the labels, the retention sentence, the two buttons — arrive in English and French already in the platform's reviewed wording, marked as platform text in the workspace; a machine draft never replaces them there unless you edit them yourself.

Participants read that page in the language that is theirs. A survey opens in the language a participant already chose (a link that names one, an earlier visit), else the language the platform knows for them — the language their invitation was sent in, the language on their group membership, the language they last answered in — else the language their browser reports. Where Quebec participants are in scope the survey opens in French unless the participant has chosen a language, with their own language one tap away; the consent record keeps both the language the page opened in and the language they consented in, so the switch is theirs and it is on record. The same text, with its hash, appears in the participant's portal, in their data export, in a right-of-access bundle, and in the ethics-board package, which carries the consent page in every offered language.

When the translator is done, they click "Mark Complete." If you assigned a reviewer, the translation moves to "In Review" and the reviewer is notified. The reviewer can approve or send it back for revision. If there's no reviewer, the translation goes straight to "Complete" and you're notified.

If you change the source text after translation, the system detects it. Changed segments are marked as "stale" — they show an amber badge in the workspace, and the translator receives a notification telling them how many segments need re-review. This prevents stale translations from silently shipping.

Working with translators

Setting up translator access

Translators are team members with the TRANSLATOR role. To add one:

Go to Account → Team and invite a new member
Select the Translator role
Choose which applications they can access
Select which languages they translate — this is how the platform knows who to suggest when you're assigning translation work

Translators can view form structure and edit translations, but they cannot create forms, see individual results, or change survey settings. They see exactly what they need and nothing more.

Translator language capabilities

Each translator declares which languages they work in. This is set in their User Settings under "Translation Languages" and can also be configured during the invitation process. When you assign a translator to a French translation, only team members who have French in their language list appear in the picker.

The translator's home base

Translators have a "My Translations" page (/user/translations) that lists all their assigned work, grouped by priority — tasks needing action first, then in-progress work, then completed tasks. Each item links directly to the workspace. This is where translators go when they log in.

Multilingual email campaigns

Translating your survey is half the story. The email invitation that delivers it should also speak the participant's language.

Participant language preferences

When you add members to a participant pool, you can specify their preferred language. This is an optional field — if not set, it defaults to the application's default language. There are three ways to set it:

Manual add: When adding members one at a time, select their preferred language
CSV import: Include a locale column (or language, preferred_locale, langue) in your CSV file
Self-registration: When participants join via an invite link, the platform detects their browser language and uses it as the default

Campaign language setup

Email campaigns are authored in English and French directly (side by side in the composer). For additional languages:

Save your campaign draft
In the Additional Languages section, add the languages you need
Machine translation generates subject, preheader, and body for each language
Optionally, open the translation workspace to review and refine

How sending works

When a campaign is sent, the platform checks each recipient's preferred language:

If the language matches an authored version (English or French): that version is sent
If the language matches a translated version (e.g., Spanish from the translations overlay): the human-reviewed version is used if available, otherwise the machine translation
If the language doesn't match any available version: the campaign's default language is used

A participant who prefers Spanish receives the Spanish version of the email. The email links to the survey, which also has a Spanish version. The entire experience — inbox to submission — is in their language.

Practical considerations

How many languages do you need?

Start with the languages your participants actually speak. Check your community demographics. A survey for a Toronto neighbourhood might need English, Mandarin, Cantonese, and Punjabi. A Quebec community needs English and French. A national survey might need both plus Spanish and Arabic.

Adding a language with machine translation only takes seconds. The question isn't "can we add it?" but "is the quality good enough for this context?" For a quick community poll, machine translation is often fine. For a research instrument that will inform policy decisions, invest in human review.

Machine translation quality

The platform uses professional-grade translation services. Quality is generally good for straightforward text — survey questions, option labels, basic descriptions. It's less reliable for:

Culturally specific concepts that don't translate directly
Technical or domain-specific terminology (medical, legal, social work)
Tone and register — formal vs. informal may not carry over correctly
Consent and legal text — where precision matters for compliance

For these cases, assign a human translator. The machine translation gives them a solid starting draft to work from, which is faster than translating from scratch.

Right-to-left languages

Arabic is fully supported with RTL text direction. The translation workspace automatically switches text direction for RTL languages, and the survey renderer does the same for participants.

Translation and publishing

You can publish a survey before all translations are complete. If a participant accesses the survey in a language that's only partially translated, the platform falls back to the default language for any untranslated segments. The publish dialog warns you if translations are incomplete, so you're making an informed choice.

Setting it up: step by step

Build your survey in your primary language as usual
Open Settings → Languages and add the languages your participants need
Machine translation generates drafts automatically
Assign translators if you want human review (optional but recommended for research)
Translators work in the full-page workspace and mark complete when done
Add members to your participant pool with their preferred language (or include a locale column in your CSV import)
Create your email campaign, author it in English and French, then add additional languages in the Languages section
Send — each participant receives the email in their preferred language, linking to a survey that's also in their language

Summary

Multilingual surveys aren't a feature you turn on — they're a commitment to reaching people where they are. The platform makes the mechanics easy: machine translation for speed, human review for quality, structured workflows so nothing falls through the cracks, and locale-aware email delivery so the experience is seamless from inbox to submission.

The hard part isn't the technology. It's deciding which languages matter for your community, investing in quality translation where it counts, and making sure your participant data includes language preferences so the system can do its job.