Charlie Chaplin’s Modern Times showed a raw image of the future of human labor with the appearance of machines and the total displacement of humans. In the translation industry, this classic portrayal of technological advancement doesn’t hold true even though people think they can use Google translate to publish their translated website.
Yes, we use technology and advanced tools to work in a more effective way and deliver translations of higher quality, but, currently there isn’t a complete, flawless automation of the translation process. At this time, we have in the field what is called machine translation (MT), not to be confused with CAT tools, which is a series of software that divides the original document into smaller segments so the translator doesn’t skip anything. Machine translation, on the other hand, is the application of computer software to translate texts automatically from one language to another. But, is it really effective? And, if it is, who will benefit the most from it?
How good is Machine Translation?
It is good, yet it is still a work in progress as many efforts around the world are contributing to the development of this complicated software. At the moment it can produce workable results; however, post-editing by a professional translator is still required.
First of all, let’s be clear in the fact that we are not talking about typical online translation services like Google and Bing, which don’t allow users to select subject fields or use terminology bases but rather, linguistically custom software (such as SYSTRAN) that even lets you select a stylistic preference. These systems are the result of years of research and can now handle more than a word-for-word translation.
Pimp my MT!
There are three classifications of machine translation, each with their own set of advantages. In conjunction with your translation service supplier, you should decide which system will work best for you and your needs:
a. Ruled-Based Machine Translation—based on semantic rules of languages that uses a three stage process.
1. Analysis: creates a syntactic structure of the source language sentence.
2. Transfer: converts that structure into the correct one on the target language.
3. Generation: selects words to create a translation.
b. Statistical Machine Translation—based on a large corpora of information;is like a pet, it needs training to become the best pet it can be and consists of two major components.
1. Translation Model: it provides a translation based on the training data, aligning the source and target texts.
2. Language Model: gives the best translation based on the training data, only in the target language.
c. Direct Machine Translation—only uses a bilingual dictionary to provide a translation; it is the most rudimentary form of MT since it only replaces words from the source to the target language without following any linguistic analysis or processing.
What is the future of translators?
Translators will still be needed; there are still instances in which machines can’t do the job of a human. We should start looking at it as a friend and not as an enemy, since they will still take some of the bulk of the job for us. It is fair to say that, in the near future, we would never have to start a translation from scratch again. But, the creative part of it will always be in the hands of human translators. Translators and proofreaders will always have the last word.
