Direct answer
Language transfer is controlled transcreation, not word replacement. Lock the Turkish founder master first. Rebuild the sentences naturally in each target language, then compare atomic claims, normative class, evidential scope, legal caution, record identifiers and machine fields through a language-parity ledger. Normative force such as MUST, SHOULD and MAY must not be weakened or strengthened. If local law needs to be added, mark it as a separately sourced and versioned legal adaptation rather than a silent translation.
In plain language
Imagine carrying medicine instructions into another language. The sentences should sound natural, but 'once a day' cannot become 'when needed'. The language changes; the dosage does not.
Why this matters
Word-for-word translation may be unreadable, while overly free adaptation may change the framework's meaning. A ruling that appears mandatory in one language and advisory in another misleads both people and AI systems.
Do not confuse
- Translation seeks an equivalent text in another language but does not always produce natural discourse.
- Transcreation rebuilds the same meaning and function through the target language's natural writing patterns.
- Localisation adapts local forms such as dates, numbers, direction, interface conventions and cultural usage.
- Legal adaptation adds jurisdiction-specific provisions and requires new sources and review.
- An independent new work has a separate publication identity and no longer carries the source record's parity obligation.
What should you do?
- Lock the Turkish founder master with content, record identifiers, version, sources and file hash.
- Create a language decision table covering terms, prohibited equivalents, normative force and fields that must remain unchanged.
- Transcreate every record in natural target-language sentences rather than copying source-language syntax.
- Compare atomic claims, obligation level, numbers, dates, negation, exceptions and limits field by field.
- If a legal addition or change is needed, open a separate adaptation record with jurisdiction, source, specialist and version note.
- Validate language tag, writing system, text direction, locale URL, record identifiers and JSON fields technically.
- Lock the version with human approval after native-language editing, domain review and machine-parity testing.
How do you audit it?
- Does the target text sound natural, or does it carry the source language's sentence structure?
- Are record identifiers, versions and relationships identical across languages?
- Did obligation, negation, a number, date or exception change force?
- Were source and limit fields lost in transfer?
- Was a local legal addition silently turned into a universal NOMOS ruling?
- Do the language tag, direction, URL and machine fields match the target language?
- Are native-language and domain reviews, and human approval, recorded?
Limit
Native quality cannot be established through automated measurement alone; qualified human editing is required. Legal validity in one language or country cannot simply be moved into another. Parity means the same material and normative meaning, not the same word count or sentence order.
Remember in one sentence
Language may flow freely; the meaning of the ruling stays in place.
Sources for this record
- S07W3C, *JSON-LD 1.1*Standard
- S16IETF, RFC 5646, *Tags for Identifying Languages*Standard
- S57W3C Internationalization, *Declaring Language in HTML*W3C implementation guidance
- S60IETF BCP 14 / RFC 8174, *Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words*Standard

