When the first session of the Nuremberg trials opened on 20 November 1945, the International Military Tribunal faced a problem no court had ever solved: the proceedings had to run in four languages at once. The judges and prosecutors spoke English, French and Russian; the defendants and most of the witnesses, German. With traditional consecutive interpreting —the speaker talks, pauses, and the interpreter repeats— every hearing would have taken four times as long, and a trial that lasted ten months threatened to stretch into years.
The solution bore the name of a man almost unknown outside the profession: Léon Dostert, born in Longwy, France, in 1904, an emigrant to the United States, a Georgetown graduate and, during the war, General Eisenhower's personal interpreter. Appointed head of the tribunal's interpretation division, Dostert staked everything on something many considered impossible: that a human being could listen in one language and speak in another at the same time.
The hardware existed, though nobody had pushed it this far. IBM lent, free of charge, its Filene-Findlay system, an apparatus of microphones and headsets patented in the 1920s and tried out at the 1927 International Labour Conference in Geneva, where it had carried speeches translated in advance. At Nuremberg, for the first time, it would serve to interpret an unpredictable live debate. Five channels were wired into the courtroom: the first carried the original voice from the floor; the other four, simultaneous interpretation into English, Russian, French and German. Everyone present picked a language by turning a small dial on their headset.
The hardest part remained: finding people who could actually do it. The profession did not exist; the professional had to be invented too. Candidates first faced a screening in which Dostert's assistant, Peter Uiberall, asked them to reel off ten trees, ten car parts and ten farming tools in both their languages — anyone who hesitates over a plough in civilian life will hesitate over a legal term with the world listening. The best travelled to Nuremberg for a second test: a mock trial staged in the attic of the Palace of Justice with a provisional rig. According to the National WWII Museum, of some seven hundred candidates examined, only around five per cent proved suitable.
Those selected —thirty-six courtroom interpreters, organised into three teams of twelve— worked shifts of about eighty-five minutes: two teams alternated in the booths while the third rested or followed the proceedings as backup. The whole courtroom learned to live with them. A monitor controlled two lights aimed at the speaker: yellow asked for a slower pace; red meant stop and repeat. The very rhythm of the trial bent to the interpreters: no one was allowed to exceed sixty words per minute, dictation speed. Even Hermann Göring, the most notorious defendant, grasped where his defence would be won or lost: "Of course I want counsel," he said, "but it is even more important to have a good interpreter."
The experiment worked. By the time the sentences were read out on 1 October 1946, simultaneous interpretation had gone from reckless gamble to standard practice: the newly founded United Nations adopted the system, and Dostert kept breaking ground — he founded Georgetown's Institute of Languages and Linguistics and, in 1954, took part with IBM in the first public demonstration of machine translation.
From those booths to today
Eighty years on, every time a court interpreter takes the oath in a Spanish courtroom, or a sworn translator authorised by Spain's Ministry of Foreign Affairs certifies with their signature that a translation is faithful and complete, something of Nuremberg echoes: the idea that justice is only possible when every party understands —and is understood— with precision. Those thirty-six pioneers proved that translation is not an accessory to the law but one of its conditions.