Facebook removed Burmese translation feature: Why machine translation can never replace human translators
Few days ago, Facebook’s spokeswoman indicated that the American social-media company had “switched off” the Burmese translation feature for posts and comments on Facebook after a Reuters article showed unacceptable flaws of this “instrument”.
In this article, we don’t dig deep into the story. Instead, it is about why machine or automatic translation can never replace human translators.
- Machines cannot understand culture
Facebook was accused of failing in its efforts to combat vitriolic Burmese language posts about Myanmar’s Rohinya Muslims. Even in last August (2018), United Nation’s investigators claimed that Facebook had been a “useful instrument for those seeking to spread hate” against the Muslin minority group. This is, however, understandable as the automatic translation feature obviously do not have the complexity to comprehend slang, idioms, names,… and thereby fail to understand the consequences of any statements, not just the ones relating Rohinya Muslims.
Meanwhile, native linguists who are well trained and versed in the languages and are able to recognize and understand cultural terms are skilled enough to find suitable equivalents in the target languages.
- Machines cannot be as fast as human translators in catching up with languages’ evolutions
It can’t be denied that new words and phrases are consistently created and developed in any language. For this aspect, machine translators are definitely not be able to pick on these evolutions as fast as human translators who are experts on localization and also different dialects in the language they specialize in. For machines, to keep up with the evolutions, they need to be regularly programmed and updated to learn new words, phrases and the context to provide appropriate translation, which requires great efforts also from humans.
- Relate words to context – a real challenge for machine translators
For a words that have different meaning in different context, it is a real challenge for machine to provide precise translation. Words with more than one meaning have to be related to the context to help determine the appropriate meanings. This no doubt can only be handled by human translators. That’s also the reason why machine translators often give out target text with illogical meaning.
- Artificial intelligence is astonishing but cannot match human intelligence, yet
IT engineers are working to improve the efficiency of artificial intelligence in general and automatic translation to be specific. Google translate tool can be named as a typical example as the quality of the translations it produces are getting more understandable. However, it is still the human translators who put in the finish touches to ensure the precision of the machine output, let alone local references. The complexity of languages is something that only human intelligence can figure out. In fact, these devices are developed to support human translators but not to force them out of the field.