When translation falls short
I have the privilege of working across Asia Pacific, which I would argue is the most exciting, dynamic and diverse set of economies in the world. Our cultural and linguistic differences are challenging but are actually part of our strength as a region. But to fully leverage the dynamic differences, we need to bring the best of all of our economies together and that requires seamless translation. Technology is making this easier, but it’s still far from perfect.
I’ve been delighted to be able to include live, automated translation, to many of my virtual calls. This transcription, though, is far from perfect and there have been moments of confusion when like-for-like words are hard to match. Sentences describing innovation, competition and strategy can all end-up with disruptive or even military connotations when converted to other languages without appropriate context.
Even global brands have stumbled. Mercedes-Benz launched in China under a name that unfortunately translated as “rush to die.” KFC’s slogan “Finger-lickin’ good” came through as “Eat your fingers off.” And perhaps the most surreal: Pepsi’s “Come alive with the Pepsi Generation” became “Pepsi brings your ancestors back from the grave.” In a culture that reveres its elders, that’s worse than awkward.
When my book Information-Driven Business was translated for the Chinese market, there was a last-minute debate over something as fundamental as the word “business” itself. While English readers think of “business” as a broad, catch-all term, Chinese offers several nuanced options, each colouring the meaning in subtle ways. We debated whether the title should use shāngyè (商业), evoking the entire commercial landscape, or qǐyè (企业), emphasizing the enterprise as an organisation. In the end, the publisher settled on shāngwù (商务), a term that leans toward practical business dealings and professional affairs. It was a reminder that even a single word choice can shift how a book’s entire message is perceived, and that translation isn’t just about words, it’s about positioning, audience, and cultural resonance.
Machine translation has come a long way from the earliest, word-for-word, computer programs of the 1950s to fluid live translation embedded in virtual meetings. The dream of seamless cross-language communication feels tantalisingly close. In theory, translation seems simple: you take text in one language and produce its equivalent in another. But language isn’t a code you can crack with a dictionary. Context is everything and the last mile can completely undermine the conversation.
It’s challenging enough when translation involves choosing between a few word alternatives, but it becomes far more difficult, even for a human translator, when meaning depends on prior interactions, shared cultural cues, or interpersonal nuance. Anthropologist Edward T. Hall addressed this challenge through his framework of high-context and low-context cultures. In high-context languages, like Japanese, Arabic, and Korean, much of the meaning is derived from situation, tone, or social relationships, rather than words alone. In contrast, low-context languages such as English, German, and Dutch rely more on explicit verbal information.
This cultural difference presents a major hurdle for translation. Moving from a high-context language (like Japanese) to a low-context one (like English) often means adding information that was never stated outright. A Japanese message might politely sidestep a “no,” relying on shared understanding and social cues. But an English-speaking recipient expects a clear answer. The translator, human or machine, has to intuit what was meant, not just what was said.
Going the other way is just as fraught. The explicitness of English can come across as blunt or even rude when dropped into a high-context setting. A phrase that seems efficient to a Western audience might need softening, rewording, or selective omission to feel appropriate in Japanese or Korean. A literal translation can end up technically accurate and socially tone-deaf.
This is where machine translation, despite its progress, still struggles. Today’s AI models are remarkably fluent and fast, thanks to deep learning and billions of training examples. But most systems are pattern-matchers, not context-readers.
The best human translators ask lots of questions and prepare thoroughly before a live meeting. Some translation tools are beginning to take the same approach. Instead of treating each sentence in isolation, these tools use background knowledge to inform tone, terminology, and intent.
Startups like LILT and Unbabel are leading this hybrid approach, combining neural machine translation with human-in-the-loop feedback to continuously improve accuracy in enterprise settings. Meanwhile, platforms like Google Cloud Translation and Amazon Translate offer built-in features for uploading custom glossaries and style guides. The future of machine translation isn’t just faster or more fluent, it’s personalised, integrated, and capable of understanding the business context.
Why does this matter? Because most of us working across cultures don’t speak all the languages we engage with, and often, we don’t know what we don’t know. A document might look fine in translation but contain subtle distortions or social awkwardness only a native speaker would spot. And those small slips can erode trust.
In the Asia Pacific region, where linguistic and cultural diversity is our superpower, the promise of machine translation is immense. And we don’t have to wait for future breakthroughs to get started. We can already build better outcomes by giving our translation tools the context they need: glossaries, background documents and real examples of tone and usage. When we treat AI not as a black box, but as a partner that learns from our inputs, we move closer to our dream of seamless communication across languages.