Machine translation (MT) is a hot topic these days. Computers and algorithms can produce better translation results than ever before. But what does this mean for the client who requires high-quality translations in a timely manner?
Advancing from Statistical Method Translation (SMT) to Neural Machine Translation (NMT)
A significant advance in automated translation dawned in 2016 when Google Translate adopted neural machine translation (NMT), an approach that is generally superior to the statistical method translation (SMT) that had powered Google Translate since its 2006 launch. As a result of NMT and other technology advances, Google Translate and a handful of other translation platforms have become useful in producing quality translations for publication, business, technical and legal purposes.
Machine Translation Post-Editing (MTPE) and the Changing Role of Human Translators
Human translators remain key players, however, providing the human touch to create excellent translations in a process known as machine translation post-editing (MTPE). The professional translator breathes life into the bare-bones MT, through skillful adjustment of word placement, vocabulary, writing style, and other elements of the translation that machine translation engines often get wrong. A relatively new discipline in the translation industry, MTPE is more than just proofreading or copy-editing by another name, although MTPE incorporates both of those processes.
Generally, NMT has not come much closer to achieving the quality of human translation than when it began on Google Translate three years ago. In a recent test by researchers, of 63 sentence output by NMT, 60 had errors. Most were errors of acceptability – the quality of the translation – or adequacy, the correctness of the translation.
Machine translation is an effective tool in the hands of a professional translator. It can save translators valuable time, allowing for greater quality control. MT engines analyze the source document and human translators make the final decisions on any translated text and are free to retain, edit, or totally re-write the automatic translation.
What About “Other” Languages and Industry-Specific Terminology?
Today’s MT platforms tend to do better with widely spoken languages, such as French or English, although even in such languages MT errors are common. Still, more errors occur in languages with complex grammar and syntax; think Japanese or Russian. If a word has more than one meaning or changes its meaning depending on the syntax, machine translation cannot match a trained human translator.
For the average user, Google Translate can be a great aid to understanding, since it’s free, fast, and convenient. However, the platform makes many mistakes and the quality varies widely, depending on the language pair and the nature of the source material. Legal translations, for example, are often too much for machine translation, especially if the source document is from a country with characteristically archaic legal terminology, such as Uruguay or Mexico.
… and Cultural Specifics of the Target Audience?
Since there is no quality control in machine translation platforms, the average user may be unaware of serious errors in the output. Another drawback of machine translation versus a professional translator is the lack of what is called “transcreation” – the practice of adjusting a translation to ensure it matches up with the culture and communication standards of the target audience. A human translator takes a vast array of such factors into consideration when crafting a translation, while a machine translation is incapable of doing so.
Machine translation is a valuable tool that is here to stay and will likely continue to improve. It cannot replace the human translator, however, when it comes to understanding and incorporating the context, metaphors, slogans, idioms, and language nuances that come into play in creating a “perfect” translation.
Contact us and we will ensure that your translation incorporates the latest technology and the guiding hand of a professional, experienced translator, in order to produce the end result you need and deserve.
Comments are closed.
