Designing for Cyrillic and Latin: Multilingual Typography

The short answer

An interface in Russian, Uzbek and English runs into three problems:

1. The font must render Cyrillic and Latin with equal quality, including Uzbek-specific characters.
2. Text length varies by language, and the same button may not fit the translation.
3. Language rules — plurals, number and date formats — affect how strings are built.

All three are solved at the design stage, not after translation.

How to choose a font

Check a typeface against a concrete list:

• Cyrillic is properly designed, not bolted on. Compare "ж", "ы", "д", "л" with Latin letters of the same weight — they should feel like one family in stroke, width and character.
• There is a Cyrillic italic. True Cyrillic italics change letterforms ("т", "д", "и"). Without a real italic, the browser just slants the upright, and it shows.
• Extended Cyrillic. Uzbek Cyrillic needs "ў", "қ", "ғ", "ҳ". In Google Fonts some of these live in the cyrillic-ext subset rather than the basic cyrillic one, so it has to be included explicitly.
• Uzbek Latin marks. The letters o‘ and g‘ use a mark that looks like a turned comma, while the tutuq belgisi (ma’lumot) uses an apostrophe. Make sure the characters you choose exist in the font and do not fall back to a mismatched glyph from a system font.
• Enough weights. At minimum regular, semibold and bold. Browser-synthesized bold looks crude.

Safe starting points are typefaces with broad language support: Inter, Manrope, Roboto, Noto Sans, IBM Plex Sans, PT Sans. You still need to test them on your own text.

A handy test string:

O‘zbekiston, g‘oya, ma’lumot, sun’iy intellekt
Ўзбекистон, қоғоз, ғоя, ҳаёт
Съешь же ещё этих мягких французских булок
The quick brown fox jumps over the lazy dog

The Uzbek apostrophe: what to know

Users type o‘ in different ways: with a plain apostrophe, a backtick or the correct typographic mark. That leads to practical rules:

• In UI and content, agree on one character and use it everywhere.
• In search and filters, normalize all variants to one, or "g‘oya" typed two ways becomes two different words.
• In forms, do not treat a plain apostrophe as an error.
• In capitals, check how O‘ and G‘ look in headings — in some fonts the mark sits too low or too far away.

Text length across languages

Russian and Uzbek phrasing is often longer than English. Uzbek is agglutinative: suffixes attach to a word, and a single word can get very long ("foydalanuvchilarimizga"). What follows from that:

A time-saving habit: lay out the design in the longest language first, or use pseudo-localization — artificially lengthened strings that reveal weak spots before real translations arrive.

Language rules that affect layout

• Plurals. Russian has three forms ("1 файл", "2 файла", "5 файлов"), Uzbek nouns do not change after a number ("5 ta fayl"), English has two forms. Do not glue strings from fragments — use plural-aware templates.
• Word order. "Delete file" and "Faylni o‘chirish" are built differently, so a variable may land in a different place in the string.
• Numbers and dates. Thousand and decimal separators and date formats differ: render them through localization tools, not hardcoded strings.
• Text in images. Any text inside an image needs a separate version per language — move text into the layout whenever possible.

Common mistakes

• Choosing a font from its Latin specimen without checking Cyrillic.
• Loading only the basic Cyrillic subset, so "қ" and "ғ" render in a different font.
• Fixed button widths tuned for English.
• Truncating with an ellipsis where the full name matters.
• Testing layouts only with placeholder text.