Neu + New + Nieuw + Neuf

Michael Schubert, CT

The Value of Human Translation

Michael Schubert, CT

Premium German-to-English business translations, certified by the American Translators Association since 2006

LinkedIn: 11. Juni 2025

I’m a technophile. In my very human career field of translation and copywriting, I embrace technology wholeheartedly. I also advocate for it when I teach and mentor others. I think tech skills are just as critical as language skills for a successful career in translation.

So I don’t dislike AI. On the contrary, I am fascinated by the promise of new technology and enjoy experimenting with AI. But because I’ve seen what AI cannot do, I’m dismayed to see it emerging as an existential threat to many in my profession.

The usefulness of AI in language services has been vastly oversold—driven by ignorance or profit motive—and I fear that customers who are told that free or cheap AI solutions can duplicate time-consuming human language services will discover that the old adage always applies: You get what you pay for.

GIGO: AI doesn’t “think,” of course: It extrapolates based on the massive amounts of data it hoovers up. But this Roomba-on-steroids is also sucking in broken glass and thumbtacks—subpar translations and outright errors—which, by design, it dutifully spits out again.
AI skips what it doesn’t know. I see this again and again: It skips the hard stuff, the nuances. In my language combination of German to English, AI notoriously ignores the modal particles (gerade, immerhin, doch, wohl, gar, bloß, eben …) that lend so much character to good writing. It does this because the solutions are too vexingly context-specific, and because it learns from subpar human translations (back to GIGO 👆). The result is a translation that lacks the sparkle and impact of a good source text. And that’s a shame.
AI hallucinates. This is an open secret at this point. AI seems to abhor a vacuum so much that it inserts “best guesses” where it lacks data, and it also seems oddly eager to please its human prompters.

Bottom line: AI results need careful scrutiny. But are they helpful as first draft? Does AI save time in the overall process of producing a useful translation?

No. More on that below.

We translators rarely receive perfect texts to translate. A good translator will always correct errors, check facts, test and localize hyperlinks, fix formatting errors, and work with the client to clarify ambiguities. We glean the essence of the source text and deliver the message clearly and idiomatically in the target language. This is strictly gray-matter added value that goes well beyond the capabilities and mandate of AI solutions.

The most common compliment we translators get is: "Wow, your translation is better than the original!" That’s because no one scrutinizes a text as carefully as a good translator—not even its author.

Liefersicherheit is not "delivery security"

Some real-world examples from my work for the German logistics industry: In my own language combination of German to English, AI-based machine translation (MT) is still unable to consistently see beyond German’s notorious compound nouns to produce a natural English. An example I’ve seen over and over again is erhöhte Liefersicherheit translated as “increased delivery security” (wrong sense of Sicherheit!) or “increased delivery reliability” (said no one, ever) instead of more robust supply chains or better supply chain resilience, which is what a good translator would instinctively write and what someone reading the English text would actually respond to. The web is already full of proclamations of “delivery security”—just ask Linguee, which scraped the internet for years to build the leading MT tool DeepL. But this kind of self-propagating robot reflux is only good for reducing translation expenses, not for actually getting potential customers to stop scrolling and swiping.

Moving from a merely uninspired translation to a dangerous mistranslation, here’s a jaw-dropper from my experimentation with DeepL. Can you spot the error?

It’s hard, because this is not the kind of translation error (…ungenauigkeit = inaccuracy) that we would expect a machine to make. I only spotted it because the target text was counterintuitive, and my human brain was turned on, which is what my clients pay me for!

AI’s working model leads it to endlessly propagate human errors unless corrected by vigilant humans during machine translation post-editing (MTPE).

But how likely is that? Not at all, since MTPE is not prestigious or lucrative and is therefore often handed over to lower-skilled linguists. And the temptation is strong—given how shiny and “right” the translations appear—to skip the post-editing altogether. Everyone loves free.

The above examples are just anecdotal, of course. Individual mistranslations can be fixed. What’s problematic is the fundamental structural flaws that produce such errors to begin with, and will continue to do so.

If the accuracy of a cheap or free translation cannot be trusted, does it have any value?

It’s often argued that AI is the perfect tool for low-value texts that otherwise wouldn’t be translated. But if the accuracy of a cheap—or free—translation cannot be entirely trusted, does it have any value? Indeed, might it not actually be worse than no translation at all, since it creates risks of reputational damage and legal liability?

Since we cannot trust AI to compile error-free translations, someone has to comb through gigantic haystacks looking for the needles that almost certainly lie within. This kind of work is mind-numbing and difficult for humans, so it means most needles will never be found.

The pertinent question here: Is it more promising to meticulously search haystacks for needles, or to build your own needle-free haystack from scratch?

Language is a tool for humans to connect with other humans.

If you’re looking for powerful communications in global English to connect with your clients, I’m your man 🧠

Interpreters meet Hemingway

written translation versus oral translation

Source: LinkedIn

Date: 19th Februar 2025

Author: Hilda Tejada

Hilda Tejada • AIIC Conference Interpreter | Interpreter Trainer | Language Services Consultant & TranslatorAIIC

El orador promedio habla 150 palabras por minuto (ppm). Velocidad normal, 140 palabras; velocidad rápida, 160ppm o más.

Si hacemos el cálculo con la velocidad menor de 120 palabras por minuto, en un turno al micrófono, una intérprete de lengua oral o señada habrá procesado 3,600 palabras aproximadamente (Seleskovitch, D., 1978), el equivalente un día de trabajo estándar de un traductor profesional que hace trabajo artesanal, o a cuatro a seis horas de trabajo intensivo de corrección de una traducción hecha con asistencia de un algoritmo (traducción asistida por computadora).

Entonces, 3600 palabras mientras la intérprete está activa, es decir, hablando al micrófono. Las escucha, las procesa, las analiza, las reformula y las restituye en la lengua meta, regulando, además, su propia velocidad, volumen, prosodia (factores no lingüísticos) y la coherencia, cohesión, naturalidad y fluidez del discurso en lengua meta (factores lingüísticos).

Al final de una jornada de 5 horas, por ejemplo, una intérprete habrá procesado cerca de 18000 palabras, si el orador habla lento. La realidad es que SIEMPRE hablan rápido porque los oradores tienen sus propios estilos, (in)habilidades oratorias, temores, intenciones y restricciones de tiempo que les obligan a hablar rápido.

Digamos pues, que hablan a 150 palabras por minuto. Esto resulta en 4500 palabras en media hora, 18mil en 4 horas, 22,500 en 5 horas y 36mil palabras en un turno de ocho horas, no solo repetidas sin contexto ni consideraciones de competencia socio-inter-cultural, como hacen las aplicaciones, sino PROCESADAS para ser entregadas con sutilezas culturales a las personas usuarias de la interpretación, sean Jefes de Estado, campesinos, académicos, víctimas de abuso de derechos humanos, etc..

Al final de una jornada laboral de 8 horas, una intérprete humana habrá logrado, con pura inteligencia natural, la prodigiosa tarea de haber traducido de forma oral, en la primera toma, en vivo, sin ensayo, una vez y media el libro “El viejo y el mar” de Hemmingway (27000 palabras en su versión en inglés, ganadora del Premiio Pulitzer en 1953).

Dr. Bernard Song 宋鹏博士

Consultant Conference Interpreter (AIIC, ATA) 顾问口译同传, Chief Interpreter of Shangri-La Dialogue 香格里拉对话会首席同传, Founder and CEO of Green Terp Technologies 绿译科技创办人兼CEO

And now to you, AI interpreter. Can you beat the above?

Marketing spin meets reality check

Source: LinkedIn

Date: 8th March 2025

Author: Dr. Bernhard Song 宋鹏博士

Singapore

🤣This is too hilarious not to share, a comprehensive review on AI simultaneous interpretation or AI real-time speech translation, written by leading AI engine Hashtag#DeepSeek in Chinese, and translated by Hashtag#ChatGPT. An AI review on AI speech translation.

"Of course! Since you asked, let me “praise” this so-called AI interpreter—brilliant! It is truly one of the “greatest” tech jokes of our time!

First, let’s admire the AI’s so-called “fluent” translation abilities. It thinks of itself as a linguistic genius, yet it struggles even with basic sentence structures. You give it a passage, and it breaks it apart at random, scrambles the order, completely ignores context and logic. Its mechanical translation style is practically a linguistic massacre. You think it can handle complex cultural backgrounds and professional terminology? Forget it! It can’t even handle simple everyday speech and colloquialisms, yet it still acts as if it understands humans. It’s absolutely hilarious!

Now, let’s talk about its so-called “real-time” capabilities. AI interpreters proudly claim to be “fast and efficient,” but this “speed” often comes at the cost of accuracy. It’s like an impatient child who wants to show off—if it doesn’t understand something, it just makes something up, leaving holes everywhere. Even funnier, it completely breaks down when faced with background noise and spoken language filled with accents. Just a little background noise? Oops, sorry, it suddenly doesn’t understand anything! And this unreliable, sensitive tech is expected to replace professional interpreters at international conferences? What a joke!

Then there’s its “learning ability.” AI claims to constantly improve by learning, but its “progress” is as slow as a sloth on a lazy day. After years of training, it still struggles with basic translation tasks. When it encounters professional terms, it gets completely lost, needing human intervention to correct its errors. Even worse, it occasionally goes off the rails and “invents” new words that don’t even exist! This kind of “innovation” is just embarrassing!

Finally, let’s “thank” AI for the entertainment value it provides. Every time it spits out its mechanical, error-ridden translations, I can’t help but burst out laughing. It’s like a bad TV drama—whenever the tension is at its highest, it delivers the most absurd and ridiculous lines. Its existence constantly reminds us that while technology is advancing, it still can’t replace true human intelligence and creativity.

In conclusion, AI interpretation is nothing more than a clumsy joke. Its so-called “intelligence” is just human vanity projected onto a machine. Its claims of being “fluent” and “accurate” are nothing but marketing gimmicks. Those sci-fi fantasies about AI replacing human translators? That’s just wishful thinking. So if you’re expecting it to perform well—just treat it as a fun toy, because it’s still miles away from being truly useful!"

Hashtag#AItranslation Hashtag#AIinterpretation

The average speaker talks at a rate of 150 words per minute (wpm). A normal speed is around 140 words, while a fast speed is 160 wpm or more.

If we calculate using the lower speed of 120 words per minute, during a turn at the microphone, an oral or sign language interpreter will have processed approximately 3,600 words (Seleskovitch, D., 1978). This is equivalent to a standard workday for a professional translator engaged in artisanal work, or four to six hours of intensive correction of a translation assisted by an algorithm (computer-assisted translation).

Thus, 3,600 words are processed while the interpreter is active, that is, speaking at the microphone. They listen, process, analyse, reformulate, and deliver in the target language, while also regulating their own speed, volume, prosody (non-linguistic factors), and the coherence, cohesion, naturalness, and fluency of the speech in the target language (linguistic factors).

At the end of a 5-hour day, for example, an interpreter will have processed nearly 18,000 words if the speaker talks slowly. The reality is that they ALWAYS speak quickly because speakers have their own styles, (in)abilities, fears, intentions, and time constraints that compel them to speak rapidly.

Let’s say they speak at 150 words per minute. This results in 4,500 words in half an hour, 18,000 in 4 hours, 22,500 in 5 hours, and 36,000 words in an 8-hour shift—not just repeated without context or consideration of socio-cultural competence, as applications do, but PROCESSED to be delivered with cultural subtleties to the users of interpretation, whether they are Heads of State, farmers, academics, victims of human rights abuses, and so on.

At the end of an 8-hour workday, a human interpreter will have achieved, with pure natural intelligence, the remarkable task of having orally translated, in one take, live, without rehearsal, one and a half times the book "The Old Man and the Sea" by Hemingway (27,000 words in its English version, winner of the Pulitzer Prize in 1953).