[show-current-language-name]
Before AI Interpretation, There Was a Room Full of Women in Headsets
In November 1945, inside the Palace of Justice in Nuremberg, twenty-four defendants sat behind glass, four languages competed for the same courtroom, and a technology that had never been tested at this scale was switched on for the first time. No one in that room — judges, prosecutors, journalists — had ever heard interpretation happen simultaneously, in real time, live. It had never been done before. And in the booths making it happen, working alongside their male colleagues, were women.
A trial that couldn’t happen without a new invention
Nuremberg needed to run in English, French, Russian, and German at once — for four Allied judicial teams and for the German defense. Consecutive interpretation, the standard of the day, would have stretched the trial for years. Instead, the tribunal adopted the IBM Filene-Finlay system: a network of headphones and a switchboard connecting the courtroom to soundproofed booths, where interpreters spoke the proceedings into microphones in real time. IBM installed it at no charge; the US government covered only shipping.
Leon Dostert, the linguist who headed the tribunal’s Translation Division, spent months recruiting the people who would actually make the system work — combing Switzerland, Belgium, and the Netherlands for multilingual graduates, many from the Geneva School of Interpreters, who could hold four languages in their heads and speak one while listening to another. Teams of twelve interpreters rotated through each day of testimony. There was no training precedent for this. No one had done it before, so no one could be trained for it — they simply had to be capable enough to invent the discipline as they went.
Among them was Elisabeth Heyward, recruited into the English-to-French booth. By her own account, she stepped into a live simultaneous interpretation booth for the first time the day after she arrived in Nuremberg — and discovered, mid-session, that she could actually do it. That was the entire onboarding process for the birth of a profession. Decades later, historical accounts of the trial would sometimes note what the women interpreters looked like before they noted what they’d accomplished. What they’d accomplished was inventing simultaneous interpretation as we now know it — the same conceptual technology behind every real-time interpretation platform in use today.
The hidden half of Bletchley Park
While interpreters were being recruited for Nuremberg, a different kind of linguistic war work had already been running in Britain for years, and it stayed classified so long that most of the people who did it never got credit in their lifetimes. At Bletchley Park, the codebreaking operation that shortened the war in Europe, women made up roughly 75% of the nearly 10,000-person workforce by 1945.
Not all of that work was cryptography in the strict sense — a significant share of it was linguistic. Codebreaking Italian, German, and Japanese traffic required people who could read the language underneath the cipher, not just the mathematics of the cipher itself. Mavis Lever was nineteen years old, with only rudimentary Italian, when she was recruited into Dilly Knox’s research unit. She identified the word fragments that cracked open Italian naval Enigma traffic — work that directly enabled British intelligence ahead of the Battle of Cape Matapan. Knox’s verdict on the victory: it was won “by the grace of God and Mavis.” She was twenty when she went on to help break the German Abwehr cipher. Rena Stewart, another Bletchley codebreaker, later put her language skills to a very different use: translating Adolf Hitler’s will.
Because the work was classified under the Official Secrets Act, most of these women couldn’t tell anyone what they’d done — not their families, not each other — for decades. It took researchers a five-year project after declassification just to reconstruct how central their contributions actually were.
A pattern, not a coincidence
There’s a third thread worth naming alongside these two: in the Pacific theater, Japanese-American women — Nisei — served as military translators for the US, doing essential intelligence work while their own families, in many cases, were held in internment camps back home. It’s a more complicated story, and a heavier one, but it belongs in the same frame: at every major turning point where language stood between information and action, women were doing the linguistic work that made the action possible, and getting comparatively little of the recognition.
From the booth to the platform
What’s changed since 1945 is the technology, not the mission. The IBM Filene-Finlay system and the women who staffed its booths solved a problem that had never been solved before: how to make understanding happen between people who don’t share a language, in real time, at the scale a moment demands. AI-powered interpretation solves the same problem today, at a scale and speed those Nuremberg interpreters — inventing their discipline live, one unprecedented session at a time — could never have imagined, but couldn’t exist without the discipline they built from nothing.
On Women’s Equality Day, it’s worth remembering that the booth came before the platform, and that women were in it from the start.