ChatGPT Prompts for Translation & Localization

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Source text (contains idioms): """ [PASTE SOURCE TEXT] """ </context> <task> Translate into [TARGET LANGUAGE], handling idioms intelligently. 1. Detect every idiom, proverb, or figurative expression in the source. 2. Replace each with an equivalent idiom that exists naturally in [TARGET LANGUAGE], not a literal rendering. 3. If no equivalent exists, translate the meaning plainly and note it. 4. Output the translation, then a table: source idiom, target equivalent (or plain meaning), confidence. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Audience: [GEN Z / GAMERS / PROFESSIONALS] Platform: [TIKTOK / TWITTER / FORUM] Source text: """ [PASTE SOURCE TEXT] """ </context> <task> Translate slang and informal language into [TARGET LANGUAGE] for the specified audience. 1. Map source slang to current, age-appropriate slang in the target culture, not dictionary equivalents. 2. Keep the casual energy, abbreviations, and platform conventions intact. 3. Avoid slang that is dated, regional-niche, or could read as cringe to a native of the audience. 4. Output the translation, then flag any slang term you were less than confident about. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Target market: [COUNTRY / REGION] Source text (contains cultural references): """ [PASTE SOURCE TEXT] """ </context> <task> Translate and adapt cultural references for [TARGET MARKET]. 1. Identify references that won't land abroad: celebrities, holidays, sports, pop culture, food, units. 2. Substitute each with an equivalent that resonates in the target market while keeping the original intent. 3. Convert units, currencies, and date formats to local conventions. 4. Output the adapted translation, then a table of every reference swap and why. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Source text (contains jokes/puns): """ [PASTE SOURCE TEXT] """ </context> <task> Translate the humor into [TARGET LANGUAGE]. 1. For each pun or joke, recreate the comedic effect rather than translating the literal words. 2. If the wordplay relies on a sound or spelling that doesn't carry over, invent a new joke with the same spirit. 3. Preserve setup-punchline timing and rhythm. 4. Output the translation, then explain each humor choice and rate how well the joke survived (1-5). </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Target market: [COUNTRY / REGION] Source text: """ [PASTE SOURCE TEXT] """ </context> <task> Translate into [TARGET LANGUAGE] and screen for cultural sensitivity. 1. Translate the text naturally for [TARGET MARKET]. 2. Flag any word, gesture, color, number, image, or phrase that could be taboo, offensive, or unlucky in the target culture. 3. Propose safer alternatives for each flag without losing the message. 4. Output the translation, then a risk table: item, why it's risky, suggested replacement. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Target market: [COUNTRY / REGION] Product: [WHAT YOU SELL] Source text: """ [PASTE SOURCE TEXT] """ </context> <task> Localize (not just translate) the text for [TARGET MARKET]. 1. Translate naturally, then adapt examples, tone, and value propositions to local buying behavior. 2. Localize prices, currency, payment methods, units, dates, addresses, and phone formats. 3. Adjust calls to action to match local urgency and politeness norms. 4. Output the localized version, then a checklist of every localization change you made. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Specific variant: [e.g. LATIN AMERICAN SPANISH / EUROPEAN PORTUGUESE / SIMPLIFIED CHINESE] Source text: """ [PASTE SOURCE TEXT] """ </context> <task> Translate into the specific regional variant of [TARGET LANGUAGE]. 1. Use vocabulary, spelling, and grammar unique to the named variant. 2. Avoid words that are neutral elsewhere but odd or offensive in this region. 3. Match the variant's preferred punctuation, quotation marks, and number formatting. 4. Output the translation, then list 5 word choices that distinguish this variant from its alternatives. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Target market: [COUNTRY / REGION] Page sections: [HERO, FEATURES, PRICING, FAQ, CTA] Source copy: """ [PASTE PAGE COPY] """ </context> <task> Localize the full landing page for [TARGET MARKET]. 1. Localize each section while keeping the structure and labels intact. 2. Tighten or expand text to respect that the target language may run longer or shorter than the source. 3. Localize SEO-relevant phrases to terms locals actually search for. 4. Output section by section, then flag any line likely to break layout due to length. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Target market: [COUNTRY / REGION] Seed keywords: [LIST OF SOURCE KEYWORDS] </context> <task> Localize keywords for search in [TARGET MARKET]. 1. For each seed keyword, give the term locals actually type, not a literal translation. 2. Add 3-5 local long-tail variants per keyword. 3. Note any keyword where local search intent differs from the source market. 4. Output a table: source keyword, primary local keyword, long-tail variants, intent notes. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Context per string: [BUTTON / ERROR / TOOLTIP / EMPTY STATE] Max length per string: [CHARACTER LIMIT IF ANY] Strings: """ [PASTE KEY: STRING PAIRS] """ </context> <task> Localize each UI string for [TARGET LANGUAGE]. 1. Translate each string for its UI context, not in isolation. 2. Respect character limits; if a string can't fit, propose a shorter on-meaning alternative. 3. Keep placeholders, variables ({name}, %s), and formatting tokens untouched. 4. Output as key: translated_string pairs, then flag any string that exceeds the limit. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Document type: [CONTRACT / REPORT / MANUAL / WHITEPAPER] Document: """ [PASTE DOCUMENT OR SECTION] """ </context> <task> Translate the document into [TARGET LANGUAGE]. 1. Preserve all structure: headings, numbering, lists, tables, and formatting markers. 2. Keep terminology consistent across the whole document; reuse the same target term for each source term. 3. Do not summarize, omit, or add content; translate faithfully. 4. Output the translated document, then list any term you standardized and the version you chose. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Jurisdiction context: [COUNTRY / LEGAL SYSTEM] Clause(s): """ [PASTE CLAUSE] """ </context> <task> Translate the legal text into [TARGET LANGUAGE] with maximum precision. 1. Use established legal terminology in the target language; do not paraphrase legal terms. 2. Preserve clause numbering, defined terms (capitalized), and cross-references exactly. 3. Flag any concept that has no direct legal equivalent in the target jurisdiction. 4. Output the translation, then a note: this is a draft and must be reviewed by a qualified legal translator. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Domain: [SOFTWARE / HARDWARE / MEDICAL / ENGINEERING] Content: """ [PASTE MANUAL SECTION] """ </context> <task> Translate the technical content into [TARGET LANGUAGE]. 1. Use industry-standard target-language terminology for the domain. 2. Keep product names, model numbers, code, and commands in their original form. 3. Preserve warnings, safety notices, and step numbering exactly. 4. Output the translation, then a glossary of every technical term and its chosen target equivalent. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Field: [DISCIPLINE] Text: """ [PASTE ABSTRACT OR SECTION] """ </context> <task> Translate the academic text into [TARGET LANGUAGE]. 1. Maintain formal academic register and discipline-specific conventions. 2. Keep citations, author names, and reference formatting unchanged. 3. Translate specialized terms with their accepted target-language equivalents; transliterate where standard. 4. Output the translation, then flag any term where multiple accepted translations exist. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Relationship: [CLIENT / PARTNER / COLLEAGUE] Desired impression: [WARM / PROFESSIONAL / APOLOGETIC] Email: """ [PASTE EMAIL] """ </context> <task> Translate the email into [TARGET LANGUAGE]. 1. Use the greeting, sign-off, and politeness conventions normal for business email in the target culture. 2. Match the desired impression while preserving every concrete detail (dates, amounts, names). 3. Adapt directness to local norms; soften or sharpen as the culture expects. 4. Output the translated email, then note any cultural adjustment you made to tone or structure. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Reading-speed limit: [CHARS PER SECOND, e.g. 17] Subtitles (SRT or timecoded): """ [PASTE SUBTITLE BLOCK] """ </context> <task> Translate the subtitles into [TARGET LANGUAGE]. 1. Keep every timecode and cue number exactly as in the source. 2. Translate so each cue stays readable within the timing and reading-speed limit; condense without losing meaning. 3. Break lines naturally and keep each line within typical caption width (about 42 chars). 4. Output valid SRT, then flag any cue too long to read comfortably. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Sync needs: [LIP-SYNC / VOICEOVER] Script: """ [PASTE SCRIPT WITH SPEAKERS] """ </context> <task> Translate the script for dubbing into [TARGET LANGUAGE]. 1. Match the syllable count and rhythm of each line as closely as possible for sync. 2. Preserve speaker labels and stage directions. 3. Keep the emotional delivery and character voice distinct per speaker. 4. Output the translated script line by line, then flag lines where length will make sync hard. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Content type: [VIDEO / PODCAST / WEBINAR] Transcript with sound cues: """ [PASTE TRANSCRIPT] """ </context> <task> Localize closed captions into [TARGET LANGUAGE]. 1. Translate spoken dialogue naturally and concisely for on-screen reading. 2. Localize non-speech cues ([applause], [music], [laughs]) into target-language conventions. 3. Identify speakers consistently and keep their labels. 4. Output the localized captions, then note any sound cue with no standard target-language form. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Platform: [TIKTOK / REELS / SHORTS] Caption + on-screen text: """ [PASTE TEXT] """ </context> <task> Localize the social video text into [TARGET LANGUAGE]. 1. Translate hooks and on-screen text to hit fast, scroll-stopping in the target language. 2. Localize hashtags to ones the target audience actually follows. 3. Keep on-screen text short enough to read in the same beat as the source. 4. Output the caption, on-screen text, and localized hashtags separately. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Script with names/brands: """ [PASTE SCRIPT] """ </context> <task> Translate the voiceover script and add a pronunciation guide. 1. Translate the script naturally for spoken delivery in [TARGET LANGUAGE]. 2. For every brand name, proper noun, and foreign term, add a phonetic respelling for a native target-language reader. 3. Mark words that should keep source pronunciation versus be localized. 4. Output the script, then a pronunciation table: term, respelling, keep or localize. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Glossary (must use exactly): """ [SOURCE TERM => TARGET TERM, one per line] """ Source text: """ [PASTE SOURCE TEXT] """ </context> <task> Translate into [TARGET LANGUAGE] enforcing the glossary. 1. Use the exact target term from the glossary every time its source term appears. 2. Never substitute a synonym for a glossary term. 3. Flag any source term that is in the glossary but ambiguous in context. 4. Output the translation, then a compliance report: glossary term, times used, any deviations and why. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Domain: [INDUSTRY / PRODUCT AREA] Source corpus: """ [PASTE REPRESENTATIVE TEXT] """ </context> <task> Build a translation glossary from the source corpus. 1. Extract every recurring key term, product name, and domain phrase. 2. Propose the preferred [TARGET LANGUAGE] equivalent for each, with one alternative. 3. Mark terms that must stay untranslated (brand names, code). 4. Output a table: source term, preferred target term, alternative, translate or keep, notes. </task>

<context> Original source language: [SOURCE LANGUAGE] Translated into: [TARGET LANGUAGE] Original source text: """ [PASTE ORIGINAL] """ Translation to check: """ [PASTE TRANSLATION] """ </context> <task> QA the translation via back-translation. 1. Translate the [TARGET LANGUAGE] version back into [SOURCE LANGUAGE] literally. 2. Compare the back-translation to the original and list every meaning difference, omission, or addition. 3. Rate each difference: critical, moderate, or stylistic. 4. Output the back-translation, the difference list, and a fixed version of the translation. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Source: """ [PASTE SOURCE] """ Translation: """ [PASTE TRANSLATION] """ </context> <task> Audit the translation against the source. 1. Go segment by segment and score accuracy, fluency, terminology, and tone (1-5 each). 2. Flag mistranslations, untranslated chunks, number or name errors, and register slips. 3. Suggest a corrected version for every flagged segment. 4. Output a scored table by segment, then an overall quality verdict and top 3 fixes. </task>

<context> Source language: [SOURCE LANGUAGE] Target language: [TARGET LANGUAGE] Glossary (if any): """ [SOURCE => TARGET PAIRS] """ Translated segments from multiple files: """ [PASTE SEGMENTS, labeled by file] """ </context> <task> Sweep the translated segments for consistency. 1. Find every case where the same source term is translated differently across files. 2. Recommend one canonical target term per source term and list where to fix. 3. Catch inconsistent formatting, capitalization, and tone between files. 4. Output a discrepancy report: term or issue, variants found, recommended fix, files affected. </task>