Arabic that feels native, not mirrored

Mirroring the layout is the right starting point. Reading order, navigation, progress indicators, sliders, back and forward arrows, and the placement of primary actions all reverse. But mirroring is a rule with exceptions. Media controls such as play and rewind usually keep their direction. Charts with a time axis often stay left to right. Phone dials, clocks and many logos do not mirror at all.

We keep a written exceptions list for every product so designers and developers apply the same judgment. Icons need a case-by-case check: a mirrored arrow is correct, a mirrored check mark or a mirrored letter is wrong. The audit stage finds these one by one in your existing screens and gives each fix a priority, so a team with a live product can repair the worst problems before a full redesign.

Real Arabic interfaces mix scripts constantly: a customer name in Arabic next to an invoice number, a product code, a URL, or an English brand. Without correct markup these strings reorder themselves and punctuation lands on the wrong side. The fix lives in the code as much as the design: setting the dir attribute, isolating inline Latin runs and testing with realistic data. The W3C Internationalization pages on bidirectional text are our working reference.

Typography needs the same care. Arabic is a joined script with context-dependent letter shapes, so text must never be split by character or truncated in the middle of a word. It needs more vertical space than Latin, and weights that survive small sizes on mid-range Android phones. Numerals are a decision, not a default: Western digits are common in commerce and banking apps, Arabic-Indic digits appear in some official and healthcare contexts. We expose this as a setting, and let the choice follow your audience.

Interface wording is design work. We write in clear Modern Standard Arabic at a Saudi business register, short and direct, avoiding calques from English such as literal translations of button labels. A good test is whether the text could have been written by an Arabic speaker with no English source. We also write error and empty-state messages so they tell the user what to do next, not just what went wrong.

Then we test. A native-speaker usability round of five to eight participants, run in Arabic, finds problems an audit misses: a label that is correct but unnatural, an icon that reads as the opposite of its meaning, a form that asks for information in the wrong order. A typical engagement runs two to three weeks for an audit and fix plan, and four to eight weeks if it includes redesign of the main flows.