What is the difference between translation and interpretation?
Translation works with written text. Interpretation works with spoken or signed language, in real time. They are separate professions requiring different skills, and practitioners of one are not automatically competent at the other.
Translation allows time. A translator can research terminology, consult references, consider several renderings, revise, and have work reviewed. The output is expected to be polished and accurate, and it is permanent — errors are visible and citable. Translators typically specialise by field, because legal, medical, technical and literary translation each require domain knowledge as much as language skill.
Interpretation happens in the moment. There is no time to look anything up, no revision, and no second attempt. The interpreter must comprehend, convert and deliver continuously, which is cognitively extraordinary.
The main modes:
Simultaneous — interpreting while the speaker continues, with a lag of a few seconds. Used at conferences and the UN, usually from a booth. The cognitive load is so high that interpreters typically work in pairs and swap every 20–30 minutes, which is a professional standard rather than a preference.
Consecutive — the speaker pauses and the interpreter delivers a segment, often working from specialised notes. Used in medical appointments, legal settings and business meetings.
Whispered (chuchotage) — simultaneous interpretation delivered quietly to one or two people.
Sight translation — reading a written document aloud in another language, which sits between the two disciplines.
What both require beyond language: cultural knowledge, since much meaning is not in the words; subject expertise; and judgement about what to preserve when an exact equivalent does not exist — idiom, humour, register and legal terms of art frequently have none.
Machine translation has changed the work substantially, with much professional translation now post-editing machine output. Interpretation has been far less affected, because real-time speech, accents, overlapping talk and context remain hard.