There are an ever increasing array of sophisticated tools that online translation companies have at their disposal these days. Technology is now paired with human translation to provide more efficient processes, greater accuracy and consistency across projects.There is some confusion among the public about some of these tools and people are unsure about what the impact is on the end result for their translations. Today we are going to clear up the difference between machine translation and translation memory in an effort to shed more light on the translation technology used in the market.
What is Machine Translation?
Machine translation requires no human involvement and systems can be programmed using rules based or statistical systems. Rules based is fairly straight forward; a set of language and grammar rules is established along with dictionaries for common words. The quality of the dictionary will greatly affect the end result using rules based systems. Statistical systems learn to translate through the analysis of data. When provided with a significant amount of data for a given language pair, a statistical system can deliver a relatively fluent sounding translation.
What is Translation Memory?
Translation memory is an intelligent database that stores useful segments of information for future use. Sentences, paragraphs or any section of a previous translation are stored to make life easier the next time around. This removes the need to translate the same phrase twice, improves consistency of translations and increases the efficiency of translations dramatically.
So What’s the Difference?
Statistical systems based machine translation and translation memory have some similarities, they both learn in a way, adding to the information to the database to build a more complete picture. From a professional translator's perspective both are very useful tools to help streamline their work. Translation memory is great piece of technology for speeding up translation work and although machine translation systems can be helpful to churn through large quantities of work, they still have a long way to go in the reliability and consistency stakes.