Translation technology is an important part of the industry. Many translation companies see the use of technology as a selling point, but how does it really affect their services? Today we’re looking at what translation technologies mean for the customer and translator.
Why use translation technologies?
Before we see why linguists use translation technologies, we need to look at what they are (and aren’t). Translation technologies isn’t a direct synonym for automatic translation. The term refers to the tools and programs language professionals can use to help with translations.
In short, translation technologies support professional linguists to achieve their best work, helping with quality control, speed, and accuracy.
Examples of translation technologies
A translation memory is an important tool that can help with both speed and accuracy. It works like a database and stores sections of text that have already been translated. When a translator comes across the same words again, the translation memory suggests the previous translation.
Translation memories really came into use during the 1990s and have become indispensable since.
When we translate, it’s important to use the right term every time for clarity and consistency. Imagine you’re using a manual to set up a new internet router and need to know which cables fit where. If the instructions alternate between cable, wire, and lead indiscriminately, it will quickly become confusing. Likewise, brands often prefer to use the same terms across their communications for consistency, and mistranslations of technical texts can have serious consequences.
A terminology management system helps with this. It might extract, implement, and maintain the right terminology, for example through a translation glossary that acts like a dictionary of pre-approved terms with information about how they should be used.
Although translation technologies isn’t a synonym for automatic translation, machine translation tools fall under the category. This might be done in tandem with a human translator through machine translation post-editing, where a trained linguist edits the computer translated text. This can improve speed and works well for bulk translations where creativity isn’t important.
A quick history of machine translation
Machine translation has developed significantly over the years. While it doesn’t give accurate enough results to replace the work of human translators, it can assist them in their work. Here are a few examples of types of machine translations.
Rule based translation – demonstrated in 1954
IBM demonstrated their revolutionary translation technology in public for the first time in the 1950s. By attaching tags (grammatical rules) to words, the team were able to turn Russian sentences into comprehensible English in a matter of seconds.
This type of translation is usually thought of as ‘rule based’, and is no longer one of the most popular types of machine translation.
Statistical translation – widespread in the 1990s
Although it had been around for a while, statistical translation came into its own from the 1990s onwards. Statistical machine translation examines previously translated texts to determine what the most likely translations are. It works with statistical models rather than grammatical rules and often with phrases rather than individual words.
Neural machine translation – breakthroughs in 2016
Neural machine translation is modelled on neural pathways in the brain. It uses a complex system to map language sequences, and considers words and phrases in context, rather than independently.
In 2016, Google announced the launch of neural machine translation for its products, which is still in use in Google Translate today. Today, many major automatic translation platforms use this approach along with machine learning to create increasingly accurate results.
Translation technologies at Future Group
At Future Group, we believe in using the right technologies to create the highest-quality translations. This includes MTPE and translation management systems.