Machine Translation Still Needs People

January 3, 2019 |

The worlds of international commerce, research and collaboration require translation services. It would be difficult for an international or multi-national company or research facility to hire expert linguists they need in order to conduct business and so the most common way of getting translation done is to go through a translation agency. However, for the past several years, there has been an ongoing trend towards reliance on machine translation. The evolution of machine translation has been quick and impressive. This summer, the CEO of a translation service told Forbes that within the next one to three years he expected machine translation to handle 50% of the current work in the translation industry.

Although machine translation is improving, and may be able to replace some of the current translation work, it will never replace the need for a human translator for a very simple reason, machines do not understand culture. Machines do word to word translation. If you’ve ever typed a word into Google Translate, or even used a dictionary, you know that simply being able to look up the meaning of a word is not the same thing as understanding a language or a sentence. Slang, idioms and even names vary not just from language to language but within a language or culture as well. Words and idioms also vary due to context, something a machine cannot understand. If your mother tells you to “shut the front door,” she probably means that you left the door open. However, if a teenager tells you to “shut the front door,” they mean that they’re surprised by what you just said. In some parts of the United States, if someone tells you that you “left the barn door open,” they mean you left a barn door open and a horse is getting out, in other places they mean that the zipper on your pants is undone. None of these differences would be understandable by a machine.

In addition to machines not understanding culture, they also fail to understand context. Even a relatively simple sentence such as “Before he left the house, the baseball pitcher left the pitcher of water on the left side of the room” could become nonsense when run through a machine translation service. Part of understanding the context of a word or a sentence is understanding the tone in which the sentence was written. Every piece of writing has a style and tone. Translating a document includes capturing that tone and style. Because a machine can’t tell the difference between academic writing, poetic writing, and humorous writing, a machine will never be able to fully translate a document. Due to regional accents, speech issues, and quixotic language usage, live translations and transcripts are even more difficult for machines.

It’s true that machine translations are becoming more efficient and the quality of the translations produced is becoming better. However, humans are still needed. After a machine creates a translation, there have to be copyeditors and proofreaders to ensure that the translation is grammatically correct and comprehensible and a human translator is still needed to localize the piece and enhance it for the target audience. As machine translation services improve, human translators will be able to use them more and more to supplement and improve their own work, but given that works are written for people by people, removing humans from the translation of the work would be foolish.

At Alpha Omega Translations we’ve been combining the best of technology and people for over 20 years. Let us help you with your machine and human translation needs.

Tags: , , , ,

Category: Business Translation

Skip to content