Terjemahan dan lokalisasi

Not word-by-word translate, but foreign person feel this native-language write

Translation gig, much tougher than thought

Machine translation flavor too heavy, terminology inconsistent, cultural nuances lost

Google translate result, grammar correct but feels off—Japanese immediately see nggak native, Korean think honorific wrong. Machine translation biggest problem nggak mistranslate, sounds unnatural.

Terminology inconsistency worse. Same document, "user" sometime "用户" sometime "使用者", "deploy" sometime "部署" sometime "发布". Client see unprofessional.

Culture gaps also trap: Chinese "内卷" how translate English? "rat race" nggak fully match. Korean social politeness completely different from Chinese. This machine-solve impossible.

OpenClaw understand context, translation read like native speaker wrote

Tell OpenClaw source language and target, nggak word-by-word translate—understand full content meaning and vibe first, rewrite target language. Result read like local person write, nggak translation-accent.

Terminology consistency? build terminology table feed, whole document every terminology consistent. Multi-language batch translate? one-time output 5 languages, every version respect local language habits. This nggak translate, this localization.

Translation prompts, grab use direct

Batch translate until terminology management, cover translation full flow.

Multi-language translation one-go Golden prompt

Translate product doc to three languages: Japanese, Korean, Spanish.

Translation requirement:
1. Nggak word-by-word, rewrite natural target language style
2. Keep original tone and vibe (casual, conversational)
3. Proper nouns English keep (API, Token, OpenClaw)
4. Japanese version note honorific level, use "です/ます" style
5. Korean version note respect-language plus formal ending
6. Spanish version use Latin American Spanish (not Spain Spanish)

Each language separate output, label language code (ja / ko / es).

[paste original content]

Batch translate universal template. Remember specify target variant (Latin American Spanish vs Spain Spanish), else AI pick one default.

Establish terminology table keep translation consistency Advanced technique

Need build English-Chinese-Japanese-Korean terminology table.

Product core terminology, give unified translation:
- user → 用户 (not "使用者")
- deploy → 部署 (not "发布" or "上线")
- token → Token (keep English, no translate)
- prompt → 提示词 (not "指令" or "咒语")

Supply recommend translation other terminology:
[list your terminology need unify]

Output format: table (English / Chinese / Japanese / Korean / notes)

After establish table, translate this document, strict follow terminology table:
[paste doc to translate]

Build terminology table before translate: pro translator team standard practice. Prompt teach AI this discipline. Build once, after use always.

Check translation quality, correct unnatural expression Beginner-friendly

Check English translation, find unnatural or unidiomatic expressions:

Original (Chinese):
[paste Chinese original]

Translation (English):
[paste English translation]

Please:
1. Flag all "translation-accent" expressions (Chinese English, unnatural English habit)
2. Give more natural alternative
3. Grammar plus spelling check
4. Formality consistency (don't formal then casual)
5. Overall translation quality score (1-10) plus improvement suggestions

Finished translation self-check this. Especially useful check own English email and docs—massive Chinese English self-write totally unaware.

Batch translation setup

Bulk content translate, this setup help maintain consistency.

Batch translation prompt template

# Batch Translation Config

Source language: Chinese
Target languages: ja, ko, es, pt, fr, de, hi, bn, id

Translation rules:
  - Style: fit daily social context, nggak AI-flavor
  - Tone: casual, accessible, conversational
  - Proper nouns: keep English
  - Terminology table: reference glossary.json
  - Format: preserve HTML tags and Markdown structure

Output format:
  Per language separate file, filename messages_{lang}.py

We do exactly this

🌍 This website—openclaw.cocoloop.cn—support 11 languages, all translation this method. Go Product comparison and Model comparison check quality. Every language read like local person wrote, nggak machine translate.

💡 Translation nggak one-time deal. Content update, translation also update. Recommend lock translation flow Prompt template + terminology table combo, every update just run once.

Real scenario: go-global SaaS multi-language localization

SaaS product go-global, website need 8 languages

Project-management-tool startup, product push Japan, Korea, Southeast Asia, Latin America. Website, help docs, product copy 50k words, need translate 8 languages.

OpenClaw approach

Build terminology table first (200+ terms), batch translate per module. Each language have "review prompt" final check. 3 days full translation, cost less 1/10 pro translator price. Launch after user feedback: "your Japanese docs really well-written".

Professional translator company

Quote $15k, delivery 6-8 week. Get back need native speaker review, revise three rounds confirm. After content update must redo process again.

Translation pro tips

💡 Translation before clear target reader. Same Spanish, Mexico user versus Spain user style different.

⚠️ Legal docs, medical docs, contract terms—AI translate can only rough draft, final absolutely need pro translator or lawyer review. Risk too big if wrong.

Case ini membantu kamu?