A phone placed on the table today can do something that until a few years ago seemed almost like airport science fiction: it listens to a voice, breaks it into data, recomposes it in another language and puts it back into circulation as the conversation progresses. In video calls, in automatic subtitles, in social content dubbed on the fly, the distance between two languages has become much shorter. OpenAI has presented voice models also designed for real-time translation, Google Translate has introduced functions for live conversations in over 70 languages, Meta has been working for years on systems capable of moving from speech to speech and from text to text in dozens of languages. The AI translator now fits in your pocket, in your browser, in your earphones, in meetings. He does his job with a speed that a human being can only look at with a hint of admiration and a hint of annoyance.
At that point the temptation arrives precisely: investing months, years, effort, wrong pronunciations, irregular verbs, embarrassments at the restaurant and notebooks full of notes is still needed. The most comfortable answer would be to leave everything to the car. On the other hand, human beings have always done it: they offload part of their mental work onto the tools. Writing has made memory easier, the calculator has taken a lot of sweat out of arithmetic, artificial intelligence enters the same family of cognitive prostheses. Used well, it can open up enormous access, help those who study, make it easier to travel, work, ask for information, follow a lesson, understand a document.
The point to keep in mind, however, concerns the difference between a tool that broadens capabilities and a tool used to completely avoid effort. When it comes to language, something greater than technical competence would disappear. A form of mental, cultural and emotional friction would disappear. That kind of slow gymnastics in which you stop on a sentence, look for a word, get the order wrong, feel your brain protesting and then, little by little, it begins to find a way.
The fatigue that remains in the head
Anyone who has tried to learn a language knows that little domestic humiliation: you want to say a simple thing, like “yesterday I went,” and you find yourself staring into space as if at a forgotten ATM PIN. The grammar gets stuck, the right word arrives two seconds later, the sentence seems to have come from a robot with a fever. Yet part of the value lies right there.
In learning psychology we talk about “desirable difficulties”: conditions that make studying less fluid immediately, but help retention, understanding and transfer in the long term. In the case of vocabulary and languages, some difficulties force us to retrieve information, compare contexts, manage interference between different systems. The work appears slower, less brilliant, even frustrating, and for this reason it can have a better impact on the memory.
Studying a foreign language means keeping together sounds, rules, gestures, registers, exceptions and small betrayals. A word is truly learned when it stops being an isolated piece and begins to have a smell, situation, tone. The right verb in a formal email, the expression that can be used with a friend, the one that sounds cold even if the dictionary translates it well. The AI translator can resolve the meaning in a second. The brain, however, when it works within a language, does something else: it trains itself to choose.
This effort sets in motion memory, attention, cognitive control, flexibility. Managing multiple languages also means keeping competition between similar words at bay, understanding which code to use at that moment, adapting to who you have in front of you. They are small, continuous, unspectacular requests. Added up over years they become a form of mental exercise. The so-called cognitive resilience, i.e. the brain’s ability to maintain efficient functions even with age, is also built through demanding and repeated activities. Speaking, listening, reading and thinking in multiple languages is part of this type of activity.
Clicking “translate” avoids the obstacle. Sometimes it’s very useful. If you need to understand a sign, respond to an urgent message, orient yourself in a foreign city, bless the button. If it becomes the only way to stay in front of a language, however, the obstacle disappears along with the training.
Less miracles, more nuances
A very seductive formula has been circulating about multilingualism for years: the “bilingual advantage”. Put like this, it seems like a medal hanging around the neck of those who speak multiple languages, a kind of cognitive superpower good for any occasion. The research, as often happens, ruins the party a little and makes it more interesting. Some studies have found benefits in attention, working memory, or inhibitory control; others observed much weaker or no differences. The best photography is more selective.
A recent study published in Scientific Reports analyzed 94 adults between 18 and 83 years old, using visuospatial and auditory tests related to working memory, attention and inhibition. In practice, researchers have observed how people process information seen in space, patterns, shapes, positions, and heard information, such as sounds or verbal stimuli, while having to remember, concentrate, or ignore distractions. Multilingualism was measured as a spectrum, without roughly dividing participants into “monolinguals” and “multilinguals.” This allowed us to include more realistic experiences: languages spoken at different levels, more or less intense daily use, different family and cultural histories.
The result avoids both easy enthusiasm and hasty rejection. On most tasks, monolinguals and multilinguals performed similarly. The most interesting difference emerged in visuospatial working memory: in older adults, a greater diversity of linguistic experience was associated with better results in that specific area. In fact, the authors speak of selective effects, linked to age, the type of task and the stimulus modality, rather than a general advantage valid for everything.
This is an important caution. Learning a foreign language doesn’t automatically turn anyone into an attention genius. However, it can contribute, together with other mentally rich activities, to keeping some functions alive. The same epidemiological literature has in various works linked bilingualism or multilingualism to a later onset of neurodegenerative conditions such as Alzheimer’s, even if the mechanisms remain debated and must be treated with caution.
The most useful thing, perhaps, is precisely in removing the veneer of the miracle from the language. Speaking multiple languages means training complex cognitive habits over time. A bit like walking a lot, reading, playing, studying, continuing to learn. No magic wand. Rather a practice that leaves traces.
Translating words, inhabiting a language
Machine translation runs strong on speed and accessibility. In many situations it works well, sometimes it surprises. However, it remains a technology based on pattern recognition, even as those patterns become increasingly refined. Difficulties emerge where language stops being just information and becomes social position, humour, modesty, implication, register, collective memory. Recent studies on machine translation continue to point to problems with cultural context, idioms, ambiguity, wordplay, and languages with less data available.
A well-translated sentence can fall flat if it loses the way it is said. This can be understood in a very popular scene of Love Actuallywhen Jamie, played by Colin Firth, tries to ask Aurelia to marry him in broken Portuguese. The scene holds up precisely for that: for the clumsiness, for the exposition, for the effort put into public view. A perfect version read by software would have delivered the content. The body of the sentence would have been missing.
The difference is all here: translating allows you to understand, learning allows you to participate. A language carries within it habits, hierarchies, ways of joking, phrases that sound affectionate in one culture and invasive in another, silences that weigh differently. This is why learning a language is also about the possibility of entering, with respect and slowly, into the way in which other people organize the world.
Some testimonies collected in the study on multilingualism show this very well. One person says he thinks in Telugu, a language spoken mostly in southern India, and remembers numbers and counts in English. Another describes Afrikaans as the language of the heart, the one for expressing intense emotions, while English remains the language of work and daily life. Here we are far from the simple replacement of one word with another. We are inside identities that change rooms depending on the language used.
Artificial intelligence will continue to change the study of languages. It can correct pronunciations, build personalized exercises, simulate conversations, reduce the shame of the first attempt, give immediate feedback. It would be absurd to treat it as a single, compact threat. The AI translator is already a powerful ally, especially for those who have to move between languages for work, study, migration, treatment or emergencies.
The part that remains human is the part that asks for time. The language learned poorly, then better, then almost well. The ear that begins to recognize a joke. The brain that stops translating every word and starts thinking directly in there. The small pride of making yourself understood without crutches. A car can get you to the other side of the sentence. To truly understand a language, however, you still have to walk.
You might also be interested in:
