Will Machine Translation Replace Human Translators
Spoiler alert: no, not yet
Do the recent technological developments of free MT software like Google Translate have you worried about your future as as a translator? Are you currently hoarding food and weapons and reinforcing your house in anticipation of when the Machines reach human-levels of consciousness? Stop. Breathe. Have a cup of tea and read this article.
Humans v. machines
When I tell people that I'm a translator, a common response is “why?". Not “why” as in “that sounds like an interesting career path, why did you choose it?", but “that sounds completely redundant, why not just use Google Translate?". For the most part, I understand why people think this way, and am happy to give them a quick rundown on human translators vs. machine translators. However, recent developments in machine translation only seem to have increased this expectation that human translators will soon become obsolete. I do not believe this is the case at all, and personally think that people tend to base this idea on two main misconceptions related to machine translation and translation in general. But first! An (extremely brief) history of machine translation.
Machine translation, or “MT”, refers to any translation activity performed solely by software with no human intervention. While code-breaking machines had been widely used throughout the Second World War, MT as we know it today is based on an idea from American scientist and mathematician Warren Weaver, who wrote a letter in 1947 theorising the use of computers to translate human languages. MT began to be developed in earnest during the Cold War, and the biggest MT services today include Google Translate, Microsoft Translator, and Yandex.Translate.
Misconception 1: The rate of development means that MT will soon surpass human translators
MT has made significant advancements since the 1950s, and its continuing progress can be seen with the recent development of neural machine translation, which produces more accurate translations than ever, and has been adopted by most MT service providers including Google. It's easy to hear about these developments and believe that machine translations surpassing human ones is only a matter of time. However, this misconception has also been around since the Cold War, and it often leaves people disappointed. MT is far from consistently reliable, and factors such as the language combination and the translation direction can have a considerable impact on its accuracy. Even at its most accurate, MT often needs a human translator to perform post-editing work and check the text for inaccurate translations, strange phrasing, etc. As impressive as MT and its developments may be, it's nowhere near a substitute for a human translator.
Misconception 2: All languages are directly equivalent to each other
This could probably have been included as a weakness of MT, but the misconception that languages are equivalent to each other has often been voiced in discussions I've had about translation in general with non-translators or language specialists. If languages were any simpler than they are, MT probably would have replaced human translation long ago, as translation would just be a matter of matching a word in the source language with the equivalent word in the target language. Unfortunately, translation is rarely that simple, as linguistic features such as wordplay, idioms, slang, tone and style can rarely be translated with a “word-for-word” or literal strategy. Translation involves creativity — something that MTs cannot do.
Rage against the machine
I hope this post has made it clear that translators do not need to prepare for the impending doom of joblessness and an industry dominated by machines just yet. However, translators should not shy away from MT either. Not only can MT technology give us advantages in our work , such as speed and consistency, but the industry has come to expect these advantages as a standard, not an added extra. Far from raging against machine translation, we should embrace it.
Originally posted to LinkedIn on 18 July 2019.