Imagine for a moment.
You show a snowy landscape to an Eskimo and to an Aztec.
They observe exactly the same reality.
The same snow.
The same landscape.
The same physical phenomenon.
Yet, in their minds, they do not see quite the same thing.
This idea may seem counter-intuitive.
However, it lies at the heart of a work that has become a reference in the world of translation: Les problèmes théoriques de la traduction (The Theoretical Problems of Translation), published in 1963 by the French linguist Georges Mounin.
And more than sixty years after its publication, one of the diagrams presented in this book still strikingly illustrates the current limits of artificial intelligence.
When languages divide up the world differently
In his book, Georges Mounin draws on an example that has since become famous.
He compares the way Eskimos and Aztecs lexically organise the universe of snow.
For the Eskimos, several distinct realities exist where other languages see only one.
- Falling snow.
- Snow already on the ground.
- Powdery snow.
- Hardened snow.
- Snow blown by the wind.
All these realities deserve different words because they play a concrete role in everyday life.
For the Aztecs, however, these distinctions do not hold the same importance. Several notions separated by the Eskimos are grouped into broader categories.
The result is fascinating.
Both peoples observe the same physical phenomenon.
But they do not mentally divide it up in the same way.
The real problem of translation
It is often imagined that translation is simply a matter of replacing one word with another.
As if every term naturally had its exact equivalent in every language in the world.
In reality, this is rarely the case.
Each language is a particular way of organising reality.
Each culture chooses what deserves to be distinguished, specified, or on the contrary, grouped together.
This is precisely why some words seem untranslatable.
And it is also why word-for-word translation often produces mediocre results.
The translator’s job is therefore to reconstruct meaning.
Not simply to replace words.
What this teaches us about artificial intelligence
AI-based translation tools are impressive today.
They translate quickly. They often produce grammatically correct text. They allow you to understand a foreign document in just a few seconds.
But they still face the same fundamental problem described by Georges Mounin in 1963.
AI excels when there is a clear statistical correspondence between two languages.
It becomes more fragile when conceptual boundaries shift.
In other words, when one culture distinguishes five realities where another sees only one.
Or vice versa.
In these situations, the issue is no longer linguistic.
It becomes cultural.
And it is precisely here that human expertise retains its full value.
To translate is to translate a worldview
The best translators do not simply translate sentences.
They translate contexts, usages, cultural references, intentions.
- An international contract.
- A marketing campaign.
- Technical documentation.
- A legal judgement.
- Or even an entire website.
All these types of content require more than a simple linguistic conversion.
They demand a subtle understanding of what the text truly means.
A 60-year-old lesson… more relevant than ever
The example of the Eskimos and the Aztecs reminds us of an essential truth.
Languages are not just dictionaries.
They are different ways of seeing the world.
That is why quality translation never relies solely on technology.
Even in the age of artificial intelligence.
And it is probably the reason why the profession of translator still has a bright future ahead.
Need a professional translation?
At Alpis Translation and Interpretation, we combine linguistic expertise, cultural understanding and human quality control to ensure translations that are faithful to the true meaning of your documents.
Request your quote and let’s discuss your project.