Arbitration Document Translation and Language

Quick answer — In arbitration the language of the proceedings is usually agreed by the parties, so the practical problem is a document set spread across languages. Decide early what gets translated in full and what gets summarised.

Vitra.ai Universe keeps terminology fixed and every change traceable.

The language is chosen, the documents are not

Parties commonly agree the language of an arbitration, and the tribunal may give directions about translation. What nobody controls is the language the underlying documents were created in.

So a proceeding in one language routinely runs on evidence in three others, and the translation question becomes what to translate rather than whether to.

Full translation is not always the answer

Tribunals often accept relevance-based approaches, and translating a whole data room to submit a fraction of it wastes money that could go elsewhere.

The tribunal's directions govern this, and the decision is one for counsel. What the process should support is being able to change the answer cheaply, because what is relevant shifts as the case develops.

That argues for translating broadly and cheaply for review — the machine translation case — and then translating carefully what is actually going in.

Witness evidence needs particular care

A witness statement is a person's account in their own words, and translation can smooth away exactly the features that make it credible or that the other side intends to test.

Register, hesitancy, and the difference between stating a belief and stating knowledge all carry weight. A polished translation of an unpolished statement changes how the witness reads, which is why meaning-level review matters more here than style.

Terminology across the bundle

The same discipline as court filings applies, with an added complication: the parties may have used different terms for the same thing throughout their dealings. Recording those variants in a glossary rather than harmonising them is usually the right instinct — the difference in terminology may itself be part of the dispute. So the glossary here does a different job from the one it does on a transactional matter. It records what each party called things and when, rather than deciding what everyone should call them. Harmonising erases evidence, and it does so invisibly, because the harmonised bundle reads better than the real one. The decision about whether a variation matters belongs to counsel, and the pipeline's job is to surface it rather than resolve it.

Running it

Document translation preserves the structure and pagination that exhibits are cited by, translation memory keeps the bundle consistent, and quality control compares against the source rather than reading for fluency.

Confidentiality is usually an express obligation in arbitration, so processing location is a question to settle before anything is sent — on-premise where the undertakings require it.

FAQ

Who decides the language of an arbitration? Usually the parties by agreement, with the tribunal giving directions where needed. What nobody controls is the language of the underlying documents, so the practical question is what to translate.

Does every exhibit need full translation? Not necessarily. Tribunals often accept relevant extracts with the original attached, and translating an entire data room to submit a fraction of it is expensive. Counsel decides within the tribunal's directions.

Why do witness statements need special care? Because translation smooths the features that make a statement credible or that the other side intends to test. Register, hesitancy and the difference between belief and knowledge all carry weight.

Should inconsistent party terminology be harmonised? Usually not. If the parties used different terms for the same thing during their dealings, that difference may be part of the dispute, so it should be recorded rather than tidied away.