Most "bilingual" identity work is one language designed and one language translated. The Latin wordmark gets the typography study, the grid system, the kerning pass. The Arabic gets Google Translate and whatever font ships with the OS. That's not bilingual — that's monolingual with a courtesy copy.
Why this keeps happening
Arabic and Latin don't share a baseline logic. Arabic is connected, contextual — a letter's shape changes depending on its neighbors. It reads right-to-left but numbers still read left-to-right. A Latin grid system dropped onto Arabic text doesn't just look wrong; it fights the script's actual structure. Treating it as a translation step instead of a design problem is why so many regional brands have a confident English mark and an apologetic Arabic one.
The result is a brand that performs like two separate companies depending on which language you meet it in first.
Building one system, not two compromises
The right approach starts the type study in both scripts at once — picking an Arabic display face and a Latin display face that share the same weight logic, the same optical size behavior, before either wordmark is finalized. Spacing tokens get defined per script, not shared, because Arabic and Latin have genuinely different rhythm. The lockup rules (icon position, clear space, minimum size) get tested in both directions from day one, not retrofitted after the English version ships.
A bilingual identity isn't two logos that happen to share a color palette. It's one system that knows how to speak twice.
Nile Pharma Food's identity is a direct example: the Arabic wordmark carries equal typographic weight to the English one — same confidence, same craft — because the brief never treated Arabic as the fallback. That's the actual bar for "bilingual," and it's reachable on any project, not just the ones with budget for two design teams.