You can have a flawless translation and still ship an embarrassing layout. Asian scripts don't follow the spacing, breaking and font assumptions that Latin typesetting is built on — and a designer who's only ever set English will hit every one of these traps the first time. Here are the ones that cost the most rework.

Thai, Khmer and Lao don't use spaces between words

These scripts run words together and break at phrase boundaries, not spaces. If your software breaks lines on spaces only, you'll get one giant unbroken line that overflows the frame — or breaks in the middle of a word, which reads as a spelling error to a native eye.

The fix is a layout engine that understands dictionary-based line-breaking. In Adobe InDesign that means the World-Ready Composer (or the Thai/SEA composer) plus a font that carries the right shaping data. Turn it on before you pour the text, not after.

Latin — breaks at spaces This is a phrase ↑ safe break points Thai — no spaces, break by dictionary นี่คือประโยคภาษาไทย breaks theengine finds
Latin breaks on spaces; Thai runs words together, so the layout engine must find break points itself.

Complex scripts need shaping, not just a font that "has the characters"

Khmer, Thai, Hindi (Devanagari), Tamil and Bengali stack and reorder marks: a vowel sign can sit above, below, before or after the consonant it modifies, and several glyphs combine into one cluster. A font can contain every codepoint and still render them in the wrong position if the application isn't doing OpenType shaping.

Quick test: paste a line of Khmer with subscript consonants into your tool. If the little stacked marks land beside the letters instead of underneath, your shaping is off — fix the engine or the font before you do anything else.

CJK line-breaking has its own rules (kinsoku)

Chinese, Japanese and Korean can break almost anywhere between characters — but not everywhere. Certain characters can't start a line (closing brackets, small kana, some punctuation) and others can't end one. Japanese typesetters call these kinsoku rules. Good software applies them automatically; bad output leaves a stray "。" stranded at the start of a line, which looks instantly wrong to a Japanese reader.

Don't fake bold and italic on CJK

Latin fonts have true bold and italic cuts. Many CJK workflows "fake" them by algorithmically smearing or slanting the glyphs — which muddies the strokes and, for italic, simply isn't a convention in Chinese or Japanese. Use a font family that ships a real heavier weight, and lean on size, color or weight instead of italic for emphasis.

Mind the mixed-script baseline and spacing

When Latin words, numbers or brand names appear inside CJK text (and they always do), the two scripts sit on slightly different visual baselines and want different spacing around them. Professional CJK fonts include the metrics to handle this; a Latin font with a bolted-on CJK fallback usually doesn't, and you get text that visually "bounces". Choosing one well-made pan-CJK family (and its proper fallbacks) solves most of it.

Embed fonts and check the licence

High-quality CJK and Indic fonts are large — tens of thousands of glyphs — and not every licence allows embedding in a PDF or app. Confirm embedding rights early, and always package or outline fonts before sending print-ready files, or the printer's machine will substitute something ugly.

A pre-flight list for multilingual DTP

  • Right composer on (World-Ready / SEA) before pouring text
  • A font that shapes the script, not just contains it
  • Kinsoku / line-break rules verified for CJK
  • Real weights — no faux bold or italic on CJK
  • Mixed Latin–CJK baseline and spacing checked
  • Font embedding rights confirmed; files packaged
  • A native speaker proofs the final laid-out file, not just the text

That last point matters most. Translation QA happens on the words; typesetting errors only appear once the words are in the layout. A ten-minute in-context review by someone who reads the language catches the stranded punctuation, the broken cluster and the clipped descender that no spell-checker ever will.