Translating Names and Acronyms

When I first wrote about this problem, the typical culprit was a translator reaching for a dictionary. Today the culprit is just as likely to be a machine translation engine or a large language model that renders a company name into Arabic in a fraction of a second — confidently, fluently, and wrong. The tools have changed. The error has not.

Translating names and acronyms is less troublesome in Latin and Indo-European languages than in Arabic, for several reasons:

First, most of those languages share an alphabet, so names and acronyms carry across with little adaptation. Arabic does not share this advantage. Second, Arabic historically does not use acronyms; it is a descriptive language that prefers full phrases to abbreviations. Third, many sounds in Latin and Indo-European languages have no exact equivalent in Arabic, and vice versa, which opens up multiple valid transliterations of the same name. Fourth, Arabic has no upper- and lower-case distinction, which removes one of the visual cues that signals “this is an abbreviation, leave it alone.”

These structural gaps are exactly where automated tools stumble. A neural engine sees “Dubai Electricity and Water Authority” and, lacking any instruction to the contrary, translates it — because translating is what it was built to do. It has no concept of a registered brand name unless that name already appears, consistently, in its training data.

Why this matters more now, not less

The Arabic localization industry remains young in important ways. Many translators still receive little formal training in modern localization, and there is no regulatory body or professional standard governing terminology and the adaptation of new names. AI has poured fuel on this gap. Volume has exploded, turnaround times have collapsed, and machine output is increasingly published with minimal review. The result is that an incorrect name can now propagate across hundreds of pages — and back into the next model’s training data — before anyone notices.

The challenge

Translating the names of companies, projects and entities remains one of the most common errors in Arabic texts, whether produced by a person or a machine. In our training workshops we put it plainly: it is not the translator’s job to name the client’s company. That is the client’s decision. As a simple test, we ask attendees to render DEWA (Dubai Electricity and Water Authority) into Arabic. We routinely get five or more different versions in the room. Run the same prompt through several AI tools and you will get a similar spread — which is precisely the point. Fluency is not authority.

How to adapt a name into Arabic

Work through these steps in order. If one does not yield a confident answer, move to the next.

1. Check the client’s own assets first. The official Arabic website, logo, slogan or brand graphics are the highest authority. If the client has already committed to a name, your job is to match it exactly — not improve on it.
2. Research trusted sources. If the client’s materials are silent, look to outlets read by professionals in the field: established regional business press, official government portals and registries, and the entity’s own verified social channels. Treat anonymous web results and AI summaries as leads to verify, never as final answers.
3. Ask the client. If research fails, ask whether an approved Arabic name exists. If it doesn’t, that is your opening to propose one and have it formally approved — so it becomes the reference for everyone who comes after you.
4. Ask the community. Failing all of the above, post the question to a reputable translators’ community such as ProZ or Translators Café. Colleagues are helpful but not infallible; weigh several answers with a healthy dose of skepticism before deciding.

What about acronyms?

The same steps apply, with one change. In place of step three, consult a specialist — someone in the field fluent in both the source and target languages — because the client often has not considered how the acronym should behave in Arabic. This is also where AI is most likely to fail silently: it will happily expand, translate or invent an Arabic acronym that no one has ever used.

Lock it down: build a termbase

In an AI-assisted workflow, researching a name once is not enough — you have to make the machine respect it every time. Once a name or acronym is verified, record it in a termbase or glossary and feed it to your translation tools so the approved form is enforced automatically. This single discipline prevents the most damaging modern failure: the same entity appearing under three different names across one project because three different tools, or three different sessions, each made their own guess.

The rule has not changed in a decade, and AI has made it more important, not less: never translate proper names or acronyms. They are matters for research, not translation — and the faster and more fluent our tools become, the more deliberately we have to apply that rule.

Dr. Ali Mohamad is CEO and Senior Researcher at HOC.