Are human translators becoming obsolete?

Machine translation services have made high-quality translations widely available, but they have changed the role of translators

• By Wesley Lewis / Contributing Journalist

Chris Findler says the introduction of neural machine translation software has reduced the demand for human translators.

“I am pessimistic about the future of traditional translation work,” says Findler, senior lecturer in translation and interpreting at National Taiwan Normal University (NTNU).

Online translators such as DeepL Translator, Yandex, and Babylon offer accurate translations in dozens of languages, which means a human translator may no longer be needed for some jobs.

Photo: Wesley Lewis

The growing influence of machine translation software is irreversible. Translation software can use artificial neural networks and large databases to accurately predict word sequences and provide nuanced expressions in their results. The presence of AI in the field cannot be ignored and translators of various specialties will have to adapt to new job descriptions.

“With artificial intelligence taking over, some translation jobs will look more like editor jobs,” Findler says.

Translation AI could be seen as a threat to jobs, but it has also created a new post-machine editing role. Findler says there is a strong demand for translators to edit and add a “human touch” to texts that have already been machine translated.

Photo: Wesley Lewis

Technical or media interpreters and translators are more likely to be threatened or impacted by AI. Findler says interpreters could eventually go extinct due to AI’s ability to recognize speech and instantly turn it into text. He adds that museum and literary translation is less threatened.

In Findler’s experience, Taiwanese museums always want to hire translators, but companies prefer to opt for the more cost-effective option of using an online translation service. In addition, technical translators tend to deal with texts with repetitive language, which can more easily be translated automatically.

Perry Svensson, the former head of translation at the Taipei Times, says about 40-50% of his freelance Chinese-to-English and English-to-Swedish work is post-machine editing. It corrects errors that the machine translator might have made.

Photo: Wesley Lewis

He adds that most of his translation work five years ago was not already translated automatically and required him to translate from scratch.

Svensson often uses DeepL Translator for articles and then makes minor edits to the translated text. Machine translators come in handy when faced with multiple projects and deadlines.

Although useful, free machine translation services have reduced the number of freelance translation jobs. Svensson translates for newspapers, logistics companies, as well as codes of business ethics and product manuals. His advice to translators is to have multiple specialties in case AI makes human translators obsolete in a specific area.

Photo: Wesley Lewis

“If you like photography, your knowledge of this field will make for better translations,” says Svensson.

Ines Tsai (蔡淑瑛) works on television and translates for tourist programs, among other things. She says auto-generated captioning technology exists, but it has yet to replace human translators.

In her work, she also comes across jobs where she is in charge of editing machine-translated subtitles. She says much of her work involves capturing colloquial speech, jokes, idioms and other aspects of language that are still difficult for a machine to register and translate correctly.

For the moment, literary translators remain the least concerned by machine translation software.

“Literary translation is a creative task that requires in-depth knowledge of multiple languages, cultures, and literary traditions,” says Ji Lianbi (計連碧), a graduate student in literary translation at Boston University.

Ji says software can understand the basic meaning of a sentence, but a word-for-word translation is often inauthentic and less expressive in the target language.

Anna Zelinska-Elliot, director of literary translation at Boston University, says universities should be required to prepare translation students for a more technology-driven profession. She adds that non-literary translation courses should address the impacts of machine translation in their field.

While literary translation isn’t as impacted by AI, Zelinska-Elliot says universities should familiarize students with various online dictionaries, but not necessarily online translators.

As a college professor, Findler fears that students will be discouraged from studying translation due to improved AI. He thinks universities should merge interpreting and translation programs to produce well-rounded students and inform students about the changing job market.

Although AI has yet to completely replace human translators, there is no way to avoid its encroachment on their work.

For Findler, the future of the translation field is set in stone, adding that many traditional translation jobs “are going to go down the horse and buggy route.”