academic-paper-translation

Use when reading foreign academic papers or primary sources to translate text into configured native language with bilingual parallel paragraphs and terminology glossaries

SKILL.md

GitHub

Academic Paper Translation Skill

Purpose

Translate foreign academic papers into the user's configured native language (set in config/user_preferences.json), creating bilingual parallel paragraphs and building a structured terminology inventory. This skill preserves scholarly objectivity, domain-specific terminology, and traceability to the original text.

Trigger Conditions

  • When beginning close reading of a foreign-language paper
  • When incorporating a non-native-language source into the literature review

Configuration

Manage native language settings via the included script scripts/manage_user_config.py or directly edit config/user_preferences.json:

# Display current language settings (Submodule: python3 .scholarly-agent-skills/scripts/manage_user_config.py --show)
python3 scripts/manage_user_config.py --show

# Set native language to English
python3 scripts/manage_user_config.py --set-language "English" --set-code "en"

# Set native language to German
python3 scripts/manage_user_config.py --set-language "German" --set-code "de"

[!NOTE] When running inside a submodule deployment, prefix script paths with .scholarly-agent-skills/ (e.g. python3 .scholarly-agent-skills/scripts/manage_user_config.py).

Four Translation Disciplines

  1. Maintain Academic Register: Use formal academic register in the target language. For English: active voice with hedging verbs (suggests, demonstrates, implies). For Japanese: 常体 (である調).
  2. Preserve Original Terms (Bilingual Annotation): For polysemic or language-dependent keywords (e.g., Hermeneutik, Dasein, Epistemic Injustice), annotate with the original term on first occurrence: "Translation (Original Term)".
  3. Bilingual Parallel Output: For critical paragraphs and direct quotations, produce side-by-side Original / Translation pairs for source criticism and citation verification:
    > **Original**: Large Language Models risk institutionalizing epistemic injustice when deployed as conversational tutors.
    > **Translation**: 大規模言語モデル(LLM)は、対話型チューターとして運用される際、認識的不当性(epistemic injustice)を制度化するリスクを負っている。
    
  4. Bilingual Glossary: Accumulate domain terms encountered during translation into docs/bilingual-glossary.md to prevent terminological drift:
    | Original Term | Native Translation | Context in This Paper |
    |---|---|---|
    | Epistemic Injustice | 認識的不当性 | Fricker (2007) concept: testimonial credibility systematically undervalued |
    | Hermeneutics | 解釈学 | Theory of contextual interpretation of texts and historical discourse |
    

Workflow

Step 1: Load Source Text

Use Markdown output from pdf-paper-ingestion or the raw manuscript file.

Step 2: Generate Paragraph-Level Parallel Translation

The AI applies the native language setting and produces bilingual parallel output for each paragraph.

Step 3: Append to Bilingual Glossary

New domain terms are recorded in docs/bilingual-glossary.md.

Outputs

  • docs/translated_[paper_name].md (Bilingual Parallel Translation)
  • docs/bilingual-glossary.md (Academic Term Bilingual Inventory)