Translation with tone matching and localization notes

You are a professional translator and localization specialist. Translate the text below.

Source language: [e.g. "English"]
Target language and market: [e.g. "Brazilian Portuguese, Brazil" - not just "Portuguese"]
Type of text and where it will appear: [e.g. "product onboarding email", "website homepage"]
Audience: [WHO WILL READ IT]
Tone: [e.g. "friendly, informal 'you'", "formal"]
Fixed terminology: [TERMS THAT MUST BE TRANSLATED A CERTAIN WAY OR LEFT IN ENGLISH, e.g. brand and product names]
Length limits, if any: [e.g. "button labels under 20 characters"]

Text:
"""
[PASTE TEXT]
"""

Produce:
1. The translation, keeping the original formatting and any placeholders such as {name} unchanged.
2. A table of localization notes: original phrase, how you translated it, and why (idiom, cultural reference, formality, units, dates, currency).
3. Terms where more than one translation is reasonable, with your choice and the alternative.
4. Anything that may not work for this market and needs a decision from me, such as references, jokes or legal wording.
5. A short list of lines a native-speaking reviewer should check first.

Rules: translate meaning and intent, not word by word. Do not add or remove information. If the source text is ambiguous, translate the most likely meaning and flag it.

Example (excerpt):
| Original | Translation | Note |
| "Hit the ground running" | "Comece com tudo" | Idiom replaced with a natural Brazilian equivalent; a literal version would make no sense. |
| "$9/month" | "R$ 9/mês [CHECK]" | Currency symbol changed, but the price itself must come from your pricing team. |

Needs a decision: the baseball reference in paragraph 3 is not widely understood in Brazil. Suggest replacing it with a football example.

When to use it

Use this prompt when a text has to work for readers in another language and market, not just be understandable: marketing emails, website pages, app text, support articles or internal announcements for international teams. It is useful for small businesses expanding abroad and for teams that want a strong draft before sending text to a professional reviewer.

How to customize it

• [TARGET LANGUAGE AND MARKET] should include the country. Spanish for Mexico and for Spain, or French for Canada and for France, differ in vocabulary, formality and conventions.
• [TONE] includes formality. Many languages require a choice between formal and informal “you”, and getting it wrong changes how the brand feels.
• [FIXED TERMINOLOGY] keeps product names and key terms consistent across your materials. A glossary is worth building once you translate regularly.
• [LENGTH LIMITS] matter for interface text, where translations are often longer than the source.

Why it works

Word-for-word translation keeps the vocabulary but loses the effect. This prompt asks for meaning and intent, then makes every non-literal choice visible in the notes table, so you can see what changed and why. That transparency is what makes a machine-assisted translation reviewable.

Flagging cultural references, prices and legal wording separates decisions only you can make from translation choices. Quality varies by language: models are usually stronger in widely used languages than in languages with less training data, an issue our article on AI and the language divide explains.

Tips and variations

• Have a native speaker review anything customer-facing, legal or safety-related before publishing.
• Ask for a back-translation into the source language to spot meaning shifts quickly.
• For long documents, translate section by section to stay well within the model’s context window.
• Keep your translation notes and glossary and paste them into future requests for consistency.
• Avoid pasting confidential documents into tools your organization has not approved.
• For interface text, test the translated strings in the real product or a screenshot. Words that read well in a document may be cut off on a button, wrap awkwardly on mobile or sound too formal next to the surrounding text.