Technology has tried to close the language gap with tools that facilitate real-time understanding of texts in English, trick, Chinese, Basque and any language. While some believe they are destined to leave translation professionals in the past, not a few wonder if technology will actually be able to overcome the language barrier one day.
You will be interested in:
In fact, the movie titles, the jokes in the streaming or the names of places in video games are just some of the everyday examples that reflect that the language is something much more complex than the simple literal translation of words or phrases.
Hispanic Heritage Month — which is celebrated from September 15 to October 15 and celebrates the contributions of this community in the United States — is a good time to ask not only if technology can overcome the language barrier, but What if we really want it to happen.
The idea of using computers to translate languages was proposed by the Englishman Andrew Booth and the American Warren Weaver in 1946, although the origins of machine translation even go back to the 9th century, with the work of the Arabic cryptographer Al-Kindi, who developed techniques for the systemic language translation. But what once seemed like science fiction is now reality.
We have all used the online translation tools of Google, Microsoft or DeepL on more than one occasion to understand a text written in another language, even those with which Spanish does not share romance roots, such as Russian or Japanese, for example. .
That is, if the internet provided the possibility of accessing information from anywhere in the world, advances in machine translation help us understand it and voice recognition tools even facilitate communication with people with whom we do not share a language. But not everything is perfect.
Google Translate, which recognizes more than 100 languages, uses a neural machine translation (NMT) algorithm, an approach based on an artificial neural network to predict the probability of a sequence of words.
In 2019, researchers at the University of San Francisco in California tested their ability to translate written medical orders in English into Spanish and Chinese. They found that it had an accuracy rate of 92 percent and 81 percent, respectively, but that 2 percent of the errors in Spanish and 8 percent in Chinese could cause “clinically significant harm.”
That machine learning applications are useful for solving everyday issues, but still insufficient for more sophisticated subjects reflects that they are not yet capable of overcoming the language barrier, but do we really want that to happen?
University of Queensland linguistics professor Michael Haugh explained in an article in The Conversation that in many languages the implicit language plays a key role, but sometimes what is not said is even more relevant. And these inferences, which differ between speakers and cultures, are not recognized by technology.
As an example, he explained that Chinese (Mandarin) speakers often decline a food offer when they are visiting, especially if they are not that close, although it is a way to test if it is a genuine offer. “Accepting an offer too quickly can also be considered rude,” he explained.
Technology will give us the ability to understand what is being said, but it will hardly allow us to recognize those inferences that are only possible thanks to an understanding between different cultures.