Five small rules for Persian on the web
The details that separate a translated site from one that was designed in Persian.
Draft for review. This piece has not been published yet.
Most bilingual sites are English sites with Persian poured in. You can tell within a second, usually from something small. These are the five small things I now check on every Persian page I build.
1. Never the middle dot
The · separator is everywhere in English metadata. In many Persian typefaces it is almost indistinguishable from the Persian zero ۰. Use the Persian comma ، instead.
2. Persian digits in running text
۱۴۰۵ in a Persian sentence, 1405 in an English one. Identifiers such as order numbers or versions stay Latin, but they get isolated.
3. Isolate every Latin token
A product name, an email or a version number inside a Persian sentence can reorder the words around it. Wrapping it in <bdi> (or using Unicode isolates) keeps the sentence in the order it was written.
4. No letter-spacing, more line-height
Tracking breaks Persian joins. Persian also needs taller lines than English, roughly 1.9–2.0 for body text, because of its vertical rhythm of dots and descenders.
5. Choose the fallback font on purpose
If the web font fails, what renders? On many systems the default fallback for a Latin font stack contains Arabic glyphs that look nothing like your design. Put a Persian-capable fallback in the stack yourself.
None of these are hard. They are just invisible until someone who reads Persian opens the page.