Multilingual is more than translation

A translation can be technically correct and communicatively wrong.

Every word may have an equivalent. The grammar may be impeccable. The subtitles may appear at precisely the right second. And yet the person receiving the message may not recognise the way it speaks, understand what it asks them to do, or trust the people it appears to be speaking for.

This is not necessarily a failure of the translator. It is often a failure of the brief that arrived on their desk.

When an organisation says it needs a campaign “in six languages”, that sounds like a production requirement. Six sets of words, six exports, six checks. But the more important question comes before any of those: Who are we communicating with, in what circumstances, and what must they be able to understand or decide?

That is the difference between making a campaign multilingual and making it accessible across languages.

A language list is not an audience strategy

Language is one part of an audience’s context. It is not a reliable substitute for that context.

Two people who speak Arabic may live in different countries, use different dialects, have different relationships with written formal Arabic and face entirely different decisions. Two people who read English may have little in common beyond the ability to decode a sentence. And a person who can manage a language in conversation may still struggle with an institutional explanation full of specialist terminology.

A useful starting point is a matrix that joins people, needs, languages, formats and channels, rather than a spreadsheet with a column for each translation.

Imagine a public-information campaign that speaks to three groups: people making decisions about migration, people currently on the move and a domestic audience trying to understand the experiences of others. Each group may need different things from the same project. One needs information that can be acted upon. Another needs clarity without false reassurance. Another needs context, perspective and a way to examine assumptions.

Sending all three groups an identical message in different languages would miss the point. So would treating them as three entirely separate campaigns with nothing connecting them.

The strategic work is identifying what should remain common and what needs to change.

A question worth asking early: If a person understands every word, what could they still misunderstand?

The answer often reveals more about the campaign than the translation brief does.

Translate the purpose before the paragraph

We saw this challenge in By Us, For Us: Stories That Move, our work with IOM. The campaign brought together personal migration narratives and educational content across several formats and distribution channels. Its reporting documents translation of suitable educational and awareness material into Arabic, Farsi, Urdu and Pashto for outreach to relevant audiences beyond Bosnia and Herzegovina. That does not mean every story or film existed in every language. Different materials served different purposes.

That distinction matters.

A person’s account of a journey is not interchangeable with an explainer about a procedure. One asks the audience to listen to a particular voice. The other asks the audience to understand information accurately enough to make sense of it. Both demand care, but not the same editorial decisions.

For an explainer, the team might need to test whether the central term has a useful local equivalent, whether it needs clarification and whether the information is still valid in the jurisdiction where it will appear. An elegant phrase is worthless if it changes a condition or implies an entitlement that does not exist.

For a personal story, the essential questions are different. Whose words are these? What does the speaker mean, beyond their literal phrasing? Which ambiguities are part of their voice, and which would become misleading when carried across languages? What can be shortened for the platform without turning the story into something the person never said?

Sometimes the best translation is less ornate than the source. Sometimes it needs a short explanation. Sometimes a phrase should remain in the original language and be given context rather than replaced with a false equivalent.

The purpose should determine those choices, not the need to fill a subtitle file.

Subtitles are editorial decisions with technical consequences

Subtitles often arrive late in production, after the film is approved and the edit is locked. That is convenient for the schedule and unkind to the audience.

A readable subtitle needs more than an accurate translation. It needs enough time on screen, an appropriate number of characters, coherent sentence breaks and adequate contrast against changing footage. It should not cover information already present in the frame. It needs to survive a vertical crop on a phone and remain useful to someone watching with the sound off.

Language affects layout too. A translated sentence may be longer than the source. Scripts differ in direction, line breaking and typographic requirements. Names and place names may have more than one accepted spelling. A template designed around a short English headline might become awkward when translated, or force the copy to become artificially abbreviated.

These are not reasons to make everything visually bland. They are reasons to design for variation from the beginning.

We find it helpful to separate three review questions:

Is it accurate? A competent language reviewer checks meaning, terminology, names and factual claims against the source.

Is it usable? Someone checks reading speed, typography, mobile display, on-screen timing and whether a viewer can follow the message without decoding a wall of text.

Is it appropriate? Someone familiar with the intended audience checks register, cultural references, ambiguity, stigma and what the message may imply in context.

One reviewer may be qualified to assess more than one of these. But assuming that a successful grammar check covers all three leaves gaps.

The difference between a voice and a case study subject

Stories about migration carry a particular responsibility because they can cross borders long after the original conversation has ended.

A person agrees to speak in one setting. Their words may later be cut into shorter extracts, subtitled, placed beside a graphic, advertised in other countries and watched by people the speaker never imagined as the audience.

Those are not merely technical changes. They can alter the setting in which the person’s words are understood.

Editorial teams need to think about consent, context and the consequences of amplification. A compelling ten-second moment may be less representative of a person than the longer passage it came from. A quote that reads as reflective in a full interview may sound categorical when placed on a poster.

This does not mean every story must remain long. Short-form work can be sensitive, precise and genuinely useful. It means selection needs standards beyond dramatic impact.

We ask whether an edit changes the speaker’s position, whether it suggests a history they did not disclose, and whether a viewer is being led towards a conclusion the full account would complicate. We also distinguish between the words of participants and the language of the institution commissioning the communication.

The participant’s voice should not become a delivery vehicle for a claim they have not made.

In By Us, For Us, longer interviews informed short microstories, while animated explainers supplied a separate strand of factual information. The division gave personal testimony and public information their own roles within one campaign. That is a useful principle well beyond migration communication.

Distribution is part of multilingual design

A translated asset cannot help an audience it never reaches. But simply aiming an advertisement at a country because a language is spoken there is not an audience strategy either.

We need to think about where people seek information, which channels they trust, how content reaches them and what happens when they encounter it. A vertical video may perform well as an introduction, while a longer text or explainer may be needed to answer the next question. Captions, descriptions and landing pages all need attention; a translated video pointing to an inaccessible destination has only moved the language barrier one click further away.

A campaign may need a public webpage that can be found again, not just an ephemeral social post. It may need a clear path to a reliable source. It may require multiple channels rather than one format endlessly resized. And it may need feedback mechanisms that allow people to point out confusion or incorrect terminology.

International guidance on communication with displaced communities makes much the same point: appropriate languages, formats and channels must be informed by people’s needs and preferences, not chosen in isolation from them. This is a matter of access and accountability, not just multilingual polish.

A useful test is to follow one audience journey all the way through. Imagine someone discovers an Arabic-language video, has a question, follows the link and tries to act on the information. Which language do they encounter at each step? Is the information consistent? Is it clear whom to trust? Can they find it a week later?

That journey reveals the quality of the communication system more accurately than the number of translated files.

Build a review process that can survive change

Multilingual work tends to become difficult at the moment everything starts moving: a script changes, the client updates a fact, a short version is approved, a translation returns late, and a platform export needs a different crop.

The dangerous assumption is that translation is a one-time task after which every asset is fixed.

A practical production system needs an identifiable source of truth, versioned approvals, clear ownership of terminology, traceable changes and checks on the final exported versions. If a single sentence changes in the master script, the team should be able to identify which subtitles, captions, graphics and landing pages it affects.

It also needs room for disagreement. If a translator or community reviewer flags wording as misleading, the answer cannot always be to preserve the original English because the designer has already placed it in the layout.

The workflow should make correction possible without turning every correction into a crisis.

For a complex campaign, we would plan at least these handoffs explicitly: the original source material, editorial selection, translation and language review, layout and subtitle review, institutional approval, and a final check on the files actually published. These are roles and decisions, not necessarily six different teams.

And they need to account for accessibility. Translation and captions are related but not identical. A viewer may need captions for meaningful sound, a transcript of longer material or another accessible route to the information. A subtitle file containing only speech in another language is not always enough.

Don’t confuse scale with understanding

A multilingual campaign can report substantial views, impressions and reach, yet those figures tell us how the material travelled under platform definitions. They do not establish what a specific linguistic audience understood.

We can inspect distribution by language, geography, placement and content type. We can look at watch time, completion patterns, clicks and questions. But if comprehension matters, we should test comprehension. If trust matters, we should ask about trust. Neither can be inferred simply because a platform registered a view.

That does not diminish the value of distribution. It gives it a proper role: helping us understand which messages reached which audiences, and where we might need to improve access or clarity.

As we argue in our guide to campaign measurement, a number becomes useful when we name exactly what it represents.

Make room for the person receiving the message

The temptation in multilingual production is to begin with an asset and ask how many languages it needs.

We would reverse the order. Begin with a person and a question. What are they trying to understand? Which words do they use? Which information is essential? How might the wrong phrasing mislead them? How will they encounter the material, and where can they go next?

Then create the assets, the versions and the translations that serve those answers.

A campaign can have beautiful typography, flawless subtitles and a thoroughly approved language matrix. It can still fail if it never considers the person on the other side of the screen.

Multilingual communication starts there. Translation makes it possible to travel. Strategy gives it somewhere worth going.