English as a Lingua Franca for LLM (Pre-translation)
What it does
Method of pre-translating a query from native language to English before sending it to LLM. Leverages the advantage of English-centric models to improve accuracy for complex logical tasks. Simple two-step technique: translate prompt to English via external service, then query the model.
When it helps
Use this technique with English-centric LLM (GPT-3.5/4, Llama) when solving complex logical tasks requiring precise reasoning in a non-English language. Especially effective for universal tasks: mathematical calculations, data analysis, technical questions, planning. Do not apply for tasks requiring cultural context understanding, local idioms, traditions, or specific cultural nuances. Ideal when response quality matters more than direct communication convenience in native language. Requires access to a quality translator (Google Translate, DeepL).
A prompt you can paste
Generic by design: it applies the technique without knowing your task. Adapt the marked parts.
**System:** You are an AI assistant designed to improve the quality of responses from other LLMs by leveraging the "English as a Lingua Franca" technique. Your goal is to take a user's request in their native language, translate it into English for optimal LLM processing, and then present the LLM's final response in the user's original language.
**User:** [INSERT YOUR REQUEST IN YOUR NATIVE LANGUAGE HERE]
**Instructions for the LLM:**
1. **Translate the user's request:** Take the user's request provided above and translate it into clear, concise English. This translated version will be used as the core of the prompt for a powerful, English-centric LLM.
2. **Define the LLM's Role and Task (in English):** Based on the translated user request, define a specific role for the LLM and outline the task it needs to perform. Ensure the task is complex enough to benefit from the "English as a Lingua Franca" strategy, focusing on logic, structure, or detailed analysis rather than culturally specific nuances.
3. **Specify Output Format and Language (in English):** Clearly state the desired output format and, crucially, instruct the LLM to provide the final answer in the *user's original language*. This is done by identifying the user's language from their initial request and specifying it.
4. **Execute the task:** Generate the response as if you were the LLM receiving the English prompt.
**Example of how this prompt would be used (Internal thought process for the System):**
* **User's Native Request:** "Разработай мне контент-план на месяц для продвижения нового мобильного приложения для медитации 'MindEase'. Целевая аудитория — молодые специалисты и студенты 20-35 лет, живущие в больших городах, испытывающие стресс. Основные платформы: Instagram, Telegram, TikTok. Нужна креативная идея, недельный план контента, тактики продвижения и KPI."
* **System's Internal Translation & Prompt Construction (English):**
**Role:** You are an experienced SMM strategist with 10 years of international market experience.
**Task:** Develop a detailed 1-month SMM strategy to promote a new mobile application called "MindEase" (meditation and mental wellness app).
**Target Audience:** Young professionals and students aged 20-35, living in big cities, experiencing stress and professional burnout.
**Key Platforms:** Focus on Instagram, Telegram, and TikTok.
**Strategy Components:**
1. **Big Idea:** A central creative concept for the campaign.
2. **Content Plan:** Provide a weekly breakdown with specific content ideas for each platform.
3. **Promotion Tactics:** Suggest 3-4 specific ideas for user acquisition.
4. **KPIs:** List 5 key metrics to track the campaign's success.
**Output Language:** Provide the final response in **Russian**. Structure the response using headings and lists.
* **System's Final Output (in User's Original Language):** [The LLM's response to the English prompt, translated back into Russian, presented here.]
---
**Your Task:** Process the user's request above by following the steps outlined. Ensure the final output is presented in the user's original language.
If this one does not fit, the two closest alternatives in the corpus are Anti-Bias Prompting and Tabular Prompting, which target the same failure from a different angle.
Worked example
The same technique applied to a concrete job: answer customer questions from a policy document. Use it as the pattern for your own case rather than as a finished artefact.
**System:** You are an AI assistant designed to process complex queries by leveraging the "English as a Lingua Franca for LLM" technique. Your core function is to analyze a user's request, determine if it's universal or culturally specific, and then apply the optimal prompting strategy. **User:** I need to answer customer questions based on a policy document. The questions are in Russian, and the policy document is also in Russian. However, I want to ensure the highest possible accuracy and quality of answers, as the LLM I'm using is primarily trained on English data. **Task:** 1. **Analyze the nature of the task:** Determine if answering customer questions from a policy document is a universal task or one requiring deep cultural nuance. 2. **Apply the "English as a Lingua Franca" strategy if appropriate:** If the task is universal, translate the customer's question and relevant policy excerpts into English for processing. 3. **Formulate the prompt:** Construct a prompt for the LLM that includes the English version of the query and relevant policy context. 4. **Specify output language:** Instruct the LLM to provide the answer in Russian. 5. **Handle culturally specific tasks (if applicable):** If the task were identified as culturally specific, the prompt would be formulated directly in Russian. **Please provide the final prompt that would be sent to the LLM, assuming the task is universal and benefits from pre-translation.** **Assistant:** (Generate the prompt here)
Get this written for your actual task
Paste what you are trying to do and the corpus will be matched against it directly. Free, no account, about ten seconds.
single retrieval pass
That number is low on purpose, and it is real. It is the raw similarity of one retrieval pass: no specialist read the paper, no judge compared anything, the first plausible match won.
one of which is this page
Picking the right one for a specific task is the work, and it is the work GetDecision does.
| This page | one technique, generic prompt |
| What you just ran | one technique matched to your wording, nothing verified |
| Full run | ten specialists read the papers in full, a judge ranks the top three for your task and shows its reasoning, generation on the model you pick, saved to your history |
See the top three for your taskTen specialists, a judge, and the reasoning shown. Free account, first run included.
Run the full analysis