The Dark Side of Over-Control in Translation ProjectsBy Volkan Güvenç, founder of Alafranga Language Solutions
Control is part of what a translation client pays for. A term list, a translation memory, a revision step and a QA report are how a buyer knows that a safety instruction says what the source says. We sell these things, and we would not run a technical project without them.
The problem starts when control is applied the same way to every text, regardless of what can go wrong in it. Then it costs time and money, and in some cases it makes the translation worse. The question is not how much control a project needs, but where.
|
▮Where strict control belongs
Some content leaves no room for interpretation. A wrong torque value, a missing warning or a shifted obligation in a contract clause has consequences outside the document. Here, close adherence to the source, enforced terminology and an independent second linguist are the right choice:
- Safety and compliance content: operating instructions, warnings, safety data sheets, declarations of conformity.
- Legal content: contracts, court documents and regulatory filings, where a change in wording can change the obligation.
- Medical content: instructions for use, labelling and clinical documents.
- Constrained text: UI strings, HMI and alarm texts with fixed character limits.
▮Where control turns into over-control
The same rules, copied to content with a different risk profile, start working against the result. These are the patterns we see most often:
- Word-for-word rules on marketing or web copy. A product page translated with the same source loyalty as a manual is accurate and reads like a manual.
- Review steps that check the same thing twice. Two reviewers with the same brief and the same checklist rarely find twice as many errors. They add a delivery day and a second invoice line.
- Preference edits treated as corrections. When a reviewer rewrites correct sentences to match personal style, the translator learns to stop making choices, and real errors become harder to see among the changes.
- Terminology rules that ignore the target language. Turkish adds suffixes to nouns. If a term check expects the exact form vana (valve), every correct vanayı, vananın or vanaya is flagged as an error. If the same rule is pushed into an AI or MT engine as a hard constraint, the engine may keep the base form and break the grammar of the sentence. The control is there, but it measures the wrong thing.
▮Why it happens
In most cases over-control is not a quality decision. It is a reaction to something missing earlier in the project.
- Unclear brief: when the purpose, audience and priorities of a text are not stated, supervision takes the place of instructions. Control becomes a substitute for clarity.
- Risk anxiety: one bad experience leads to rules that are applied to every project afterwards, whether they fit or not.
- Tools treated as judges: an automated QA check or an AI setting is assumed to be precise because it is automatic. It is only as precise as its configuration.