The short answer is “yes, definitely.”
Machine translation (MT), like Google Translate, has been such a convenient tool that many have used it at some point in life. It emerged around the 1950s and has gained strong momentum in recent years. Google, Microsoft, Amazon, and other tech giants have embraced MT, leading to the rapid development of this technology.
There are many different MT systems, with the four main types being Rule-based Machine Translation (RBMT), Statistical Machine Translation (SMT), Hybrid Machine Translation (HMT), and Neural Machine Translation (NMT). RBMT uses built-in linguistic rules, bilingual glossaries, and dictionaries to translate the source language. In contrast, SMT does not rely on the grammatical rules and knowledge of the language but utilizes computer algorithms to analyze large numbers of the content of each language pair to come up with statistical translation models.
HMT, as literal as its name, leverages the strengths of both RBMT and SMT. NMT, thought to be the most advanced system at this stage, uses neural network models (multiple processing devices modeled on the human brain) and deep learning to predict the probability of a set of words in sequence.
In fact, Google Translate, the most popular MT provider, is now powered by NMT. It has gradually replaced its predecessor, the much-less accurate “phrase-based” translation system, since 2016. So, how far has Google Translate improved? According to a study by Cornell University, it shows that Google’s NMT has managed to reduce translation errors by 60% compared to its “phrase-based” system.
Well, this is definitely a remarkable improvement. However, can we say that translators are no longer needed? The answer is clearly no for the time being as well as the foreseeable future. Put simply, MT has not been qualified yet to handle the high-priority content, aka the most challenging task a translator can get, let alone fully replace the translator. MT, like Google Translate, still fails, even when dealing with some seemingly effortless sentences.
A recent example of this is the U.S. Virginia state’s COVID site, which relied on Google Translate, mistranslated the English word “book” (to reserve in advance) into the Spanish word for “a written text that is published in printed form.” Not to mention also the translation errors on the welcoming banner during Chinese former premier Wen Jiabao’s state visit to Malaysia in 2011. Google Translate had mistranslated the title on the banner- “Official Welcoming Ceremony for HE Wen Jiabao’s State Visit to Malaysia” into “Official Welcoming Ceremony, along with him HE Wen Jiabao’s Officially Visit Malaysia.”
One could imagine how disastrous the outcome will be, particularly in a formal and professional setting, if we solely rely on MT? Again, let’s refer to some research to get a clearer picture. In 2017, Sejong Cyber University in Korea set up a competition between three MT and a group of human translators. Unsurprisingly, the machines failed to live up to expectations. While they were more rapid, they were not up to standard in terms of quality, with 90% of their translated texts being “grammatically awkward.” Another study by the Olive-View UCLA Medical Center and New York’s Memorial Sloan Kettering Cancer Center in 2021 concluded that Google Translate is not well-grounded to be used for medical information.
Clearly, the MT’s inability to consider each language’s context and cultural background has been its major shortcoming, which also justifies the necessity of human translators in translation services, especially in the professional setting. Some equate translating with writing (an art form), and this validly explains why the MT is incapable of this task.
Only humans have the knowledge of a language and cultural awareness and familiarity with contexts, idiomatic expressions, tone, and style. This enables them to discover the nuances between different languages and come up with an accurate and high-quality translation. Therefore, it comes as no surprise that human translators are still in high demand. The U.S. Bureau of Labor Statistics projects a 20% employment growth for interpreters and translators by 2029.
However, it is wrong to posit that MT does no good to the translation industry. On the contrary, the MT displays its strength when dealing with content that is high-volume, low-priority, needed fast, but where accuracy is unnecessary. Comments, social media posts, reviews, 24-hour chat facilities, and website help sections are all MT’s targets. MT is also advantageous in so-called “gist translation,” where users need to first understand a rough gist of material in a foreign language before deciding whether to consult a professional human translator.
Human translation can also combine with machine translation, leading to what is known as Post-Editing Machine Translation (PEMT). MT first translates the content to create the raw output, which is then proofread and edited by a human translator with the help of Computer-Assisted Translation (CAT) tools. The main benefits are said to be boosted production and quicker turnaround times. Yet, not every human translator welcomes this. Some criticize that PEMT actually wastes more time because the raw output by MT can sometimes be incomprehensible, and they have to start over from scratch.
Another combination between MT and human translation is “human in the loop” machine translation. It is a method where the MT can learn from and adapt to human feedback. Surprisingly, according to the CSA Research’s 2020 “The State of the Linguist Supply Chain” survey involving more than 7,000 linguists, 71% of the respondents favored “human in the loop.” This implies that human-machine deeper collaboration is becoming prevalent in the industry, which can significantly change how human translators work in the current setting.
Yet, one thing is for sure-human translators will still be needed, even always be needed in the future. Why? Because human translators will always be at the core of the translation industry, while MT is an essential complement (unless MT manages to think like a human translator). For now, if you want a high-quality and accurate translation, please look for a translation agency.
Fudea is committed to easing your burden in translation and localization efforts. We are a multilingual translation service provider focusing on professional translation and interpretation services for your business. If you want to learn more about our services, please contact us via our inquiry form.
