Artificial Intelligence and “untranslatable” words

Artificial Intelligence and “untranslatable” words

The translation industry has for decades relied on information technology to speed up its work. Now the impact of artificial intelligence is the biggest development. Automatic Translation solutions are increasingly equipped with “artificial neural networks” that simulate aspects of the human brain and provide computers with the capacity for “deep learning”. This means that the more translators use solutions like MemoQ, Memsource, Deepl or Google Translate, the more the artificial neural networks supporting them are extended. The software picks up new vocabulary, different writing styles and linguistic nuances. In other words, the machine learns from its mistakes and adapts to new situations. In effect, it “evolves”.

The results are increasingly revealing. A recent study by Memsource reveals artificial intelligence can reduce project costs by around 15%. This CAT (Computer Assisted Translation) tool’s deep learning function includes new AI technologies that can detect “untranslatable words” in any text. Names of people or companies, technical terms, acronyms, dates, repeated expressions and other elements are immediately translated… In the example cited in the report, a 150,000-word text in French, CAT tools allowed a reduction in the number of words requiring human intervention from 85,000 to 69,000.

These are not just private sector developments. In 2020, a group of researchers at Karlova University in the Czech Republic developed the “CUBBITT” automatic translation system that also applies deep learning principles. Their “Turing Test” using a number of press articles achieved impressive results, with more than half of the readers analysing the texts unable to distinguish between those translated by the machine and those translated by humans.


This has a number of implications that will soon change the landscape of the industry:

  • Project delivery times will get progressively shorter;
  • Translation costs will fall;
  • Professional translators will have to adapt to new technologies and invest in more training and software apps;
  • The increased speed of translation will mean translators will be able to do more work in less time.



{Suggested articles}

Leave a Reply

Your email address will not be published. Required fields are marked *