You are here

Human Vs Machine Translation: Who Wins?

Human Vs Machine Translation: Who Wins? main image

Guest post: Ashley Harris

The goal of finding a machine that can think on the level of a human continues. In a recent test, researchers pitted the Google Translate program against a living, professional translator. Unfortunately for those with visions of artificial intelligence (AI) suddenly taking over most human tasks, the machine translation failed to provide any real competition for the human transcriber. Nevertheless, all is not lost for high-tech fans; the research evaluation found clear benefits to using the Google application in certain specific cases.

To arrive at the above conclusions, the research team put the two parties through a series of tests. Round one was based on a Spanish text to be translated into standard English. This request presented no problems whatsoever for the human. She was able to translate the sentences in a way that provided the reader with a clear understanding of the true meaning of the text.

This didn’t go so well for the machine translation program. Unlike people, computers cannot yet comprehend implied meanings. Thus, the Google-based translation was of the word-by-word variety. This style lost the overall meaning of the entire text. The results were clear. Humans are much better at analyzing written texts than machine translation applications.

So, should people forget about using Google Translate? Well, not so fast. In reality, the service can prove valuable to those who just need to get the gist of a text. Second language learners, for example, who just need to know if their translation is within the general ballpark, are likely to find the application useful. Likewise travelers who want a rough idea of something they see written in a foreign language can benefit from quick access to a machine translation.

The second part of the experiment involved an audio transcription. Though the results were closer than in the first round, in the end, humans once again proved the better translators. Google Translate took the Spanish it heard and then copied it quite flawlessly. However, the translation into English produced some errors. Most significant of the problems was an inability to locate English words for some Spanish words. The machine just left these words in Spanish. In contrast, the human translator again gave a nearly perfect performance.

For some time to come, it seems we’ll still need to rely on other humans, whether it be teachers, transcriptionists or tour guides, to accurately interpret languages. Yet machine translation is certainly not useless, and often a good quick check mechanism. However, research results demonstrate that computers cannot compete with fluent humans when it comes to analyzing the full meaning of a text.

Human Vs Machine Translation

Ashley Harris is outreach coordinator for Connect Theory and is working closely with Verbal Ink on content development and refinement.

Related categories:

Guest Writer's profile image
Written by Guest Writer
 

Want to leave a comment?

Please login or register to post
comment above our articles

0 Comment