Moderating interviews in Arabic and English

Ask someone to describe a frustrating experience in their second language and you will get a shorter, flatter, more polite answer than the one they would have given in their first. The feeling is still there. The vocabulary for it is not.

For research across MENA, where a single study can span Gulf, Levantine and Egyptian Arabic plus English, language is not an operational detail. It shapes the data.

Let the participant choose, and record the choice

Ask for language preference at recruitment, not at the start of the session. It gives you time to assign the right moderator, and it prevents the awkward moment where someone politely accepts English because it feels easier for you.

Record the choice alongside the session. When you analyse, you want to know whether a pattern appeared across languages or only within one.

Dialect matters more than the label

"Arabic" is not a single setting. Modern Standard Arabic is the language of formal writing and news, and almost nobody conducts a casual conversation in it. Real conversation happens in dialect, and the distance between Gulf, Levantine, Egyptian and Maghrebi speech is substantial.

For moderation:

• Match the moderator's dialect to the participant where you can.
• Where you cannot, Egyptian and Levantine are widely understood across the region through media exposure.
• Do not moderate in Modern Standard Arabic. It makes the session feel like a formal interview, which is precisely the register you are trying to avoid.

Code switching is data, not noise

Bilingual speakers in the region switch languages constantly, and the switches are meaningful. People commonly move to English for technical terms, product names and workplace vocabulary, then back to Arabic for feeling, judgement and story.

When someone switches, notice it. A participant who describes the interface in English and their frustration in Arabic is telling you something about how they hold those two things.

Let it happen. Interrupting to enforce one language costs you the thing you came for.

Interface language versus thinking language

A very common situation: the product is in English, the user thinks in Arabic. Watch for the small pause where someone reads a label and translates it internally before acting.

That pause is a finding. It is the cost of the interface being in a second language, and it shows up as hesitation, misinterpretation of labels that seemed obvious to the team, and a preference for icons and layout cues over text.

If you support Arabic, test the Arabic. Right to left is not only a mirroring exercise. Watch where the eye lands first, which direction progress feels like it should move, and whether mixed content such as numbers, dates and English brand names reads cleanly.

Practical moderation notes

Open with the language question. "Would you rather we do this in Arabic or English? Either is completely fine for me." Say it in the language you would prefer them to feel comfortable declining.

Distance yourself from the design. In many settings, criticising work in front of its author is socially uncomfortable, and that instinct is stronger in some Gulf contexts. Say clearly that you did not build it.

Prefer tasks to opinions. "Show me how you would find the returns policy" produces behaviour. "Is the navigation clear?" produces politeness.

Allow longer silences. The pause before an honest answer is usually longer than feels comfortable, and longer still when the person is composing in a second language. Do not fill it.

Avoid idioms. They do not survive translation, and they make participants feel tested.

Note taking and transcripts

Take notes in the language of the session. Translating in the moment costs attention you need for the conversation, and early translation smooths away the specific wording that often matters most.

For analysis, keep the original phrasing alongside any translation. A translated quote is an interpretation, and the interpretation should be visible as one.

Automatic transcription handles English well and is improving for Arabic, but dialect accuracy remains variable. Treat an Arabic transcript as a working draft and check any quote you intend to rely on before it goes in a report.

Analysing across languages

When you pull themes across a mixed language study, watch for two traps.

The first is over-weighting the English sessions because they were easier to skim. The second is treating a difference between language groups as a cultural finding when it may simply be a fluency effect. People say less in a second language, and less is not the same as caring less.

Where a theme appears in only one language group, say so explicitly rather than presenting it as universal.

Building a bilingual study

If you are planning one:

1. Screen for language, including comfort level rather than a yes or no.
2. Assign moderators by language and dialect before scheduling.
3. Add time. Bilingual sessions run longer.
4. Record the language on every session.
5. Keep original wording in your notes.
6. Report by language group where the groups differ.

Running the sessions

UsersArabia recruits participants across MENA with language and proficiency captured in their profile, so you can screen on it and see it when reviewing applicants. Sessions can run with in-app video and automatic transcripts on Premium, or with your own call link on the free plan.

Set up an interview study or read about usability testing in the GCC.