What a dream: being able to immediately understand and speak the local language while abroad. Artificial intelligence brings us closer to this wish. What experts say about this future scenario.
In the cult novel “The Hitchhiker’s Guide to the Galaxy” no one needs to learn a language. Anyone who listens to the Babel fish, a fictional creature in the science fiction classic, suddenly understands all the languages of the universe. Is this a template for the here and now? Can we use artificial intelligence (AI) to make a similar leap across language barriers?
For the Berlin linguist Anatol Stefanowitsch, there are already simultaneous translators that could make learning foreign languages unnecessary. This is possible thanks to advances in machine learning. Computer-aided speech recognition and translation are now so good that they are sufficient for many everyday purposes. A corresponding app is often installed on modern smartphones.
The goal is: machine translation of language in real time. This means: The programs deliver the result without a break if possible. According to the Goethe Institute, the technical requirements for this already exist. Its experts differentiate between translating one language into another and learning and mastering a foreign language. The latter goes far beyond “simple simultaneous translation.”
Translation apps still fail when it comes to large tasks
A few years ago it seemed like magic when a menu was already translated on the display in front of the smartphone camera. Software on phones can now also help as an interpreter in a conversation in real time: sentences in a foreign language are converted into translated text on the display. This is how you can break the language barrier – albeit in a dialogue with delays.
Manufacturer Samsung goes even further with the translation of calls in its new Galaxy S24 smartphone, developed in close collaboration with Google. The idea is that you can reserve a table in a restaurant in another country without knowing the language. The software not only translates, but also pronounces the sentences with a computer-generated voice. The price, again, is breaks in the conversation.
The programs also sometimes have problems acoustically understanding what is being said. And although translation skills have improved over the years, an expression like “not the yellow of the egg” still becomes “not the yellow of the egg”.
The more the programs have to handle, “the more likely these apps are to fail,” explains Stefanowitsch from the Free University of Berlin. But he sees potential: “Speech recognition and translation are areas in which great progress can still be expected in machine learning.”
The foreign language sector is facing profound changes, according to the Goethe Institute. Because AI models like ChatGPT are developing rapidly. Experts believe they will “change both the learning process and the way we communicate.”
Why learning foreign languages still makes sense
And if that happens, does it still make sense to learn a foreign language? Stefanowitsch says yes. “I definitely think it’s valuable.” Because communication is not just about exchanging information, but also about dealing with each other on a human level. “In the future, we will not want to have a friendship or even a romantic relationship with a permanently intermediary app,” he explains.
In addition, people can only immerse themselves in other cultures to a limited extent, “as long as every utterance has to be translated by a computer,” says the linguist. Every language contains a different perspective on the world. This can only be experienced if the language is learned yourself.
This is also what the experts from the Goethe Institute are aiming for and cite as an example nursing staff in Germany, an increasing number of whom do not speak German as their native language. Here, natural language creates empathy. “Do we want to live in a world in which nurses communicate with their patients using simultaneous translation?” asks the Goethe-Institut.
New forms of foreign language teaching
Speaking English, lesson one: In a world with digital translation aids, teaching foreign languages at school seems old-fashioned. Nevertheless, Stefanowitsch believes that the children are “on the whole” learning in the right way. Language learning apps could complement school lessons, but not replace them, he explains. The Goethe-Institut assumes that the role of the teacher and teaching are changing – “away from the pure imparting of knowledge towards active support of the learners”.
When learning at home, students could then have the AI do their homework for them. This not only applies to foreign languages, but also other areas. “In the future, we will have to do without all types of homework that can be done by so-called AI applications,” predicts the linguist.
This is what simultaneous translations do with the native language
Machine-generated texts are very much based on a kind of average language. That’s why they “always sound very phrase-like and not very personal,” says Stefanowitsch. The Goethe-Institut sees an impact on dialects if machine simultaneous translators are used across the board. Reducing it to a standard language could put local and individual variations at risk. In addition: In order to ensure an error-free translation, there could also be a reduction in the amount of vocabulary. The linguist states: “It is better if translations are done by people rather than by machines.”