Like most of the industry, my position on using Google Translate has always been that Google Translate is way too flawed to rely on for anything more important than reading the latest letter from Oncle Françoise or writing to Zia Maria. Overall, I still feel this is true. You wouldn’t want to use Google Translate for mission-critical information such as legal content or intricate instructions on how to operate a nuclear reactor. But, it is getting better.
There are two reasons that Google Translate is improving:
- It uses a Statistical Machine Translation engine
- The community is helping to add and correct translations
Types of Machine Translation
There are three types of MT engines:
- Rules-based
- Statistical
- Hybrid
A Rules-based MT (RBMT) engine is programmed with linguistic rules. Content is translated according to the rules. While it can be a very accurate way to translate via machine, the process of programming the rules is extensive and time-consuming. And the engine has a very difficult time processing content that doesn’t follow the rules.
A Statistical MT (SMT) engine constantly learns translations by pattern matching. As more content is analyzed, the engine learns more patterns. It then selects the most statistically likely version of a requested translation. In order for an SMT engine to be useful, it has to be trained by processing an enormous amount of content. This is also why SMT engines improve over time.
A Hybrid engine combines rules and statistics. Theoretically, it has the best chance of producing an accurate translation.
Google Translate is an SMT engine. Over time, it has processed a huge amount of content. The more content it processes, the more accurate it becomes. This is particularly true when a user is presented with multiple translations and picks a preferred version. The engine learns from the selection.
Help From the Community
Crowdsourcing is the second key to the improvements in Google Translate. Google recently announced the Google Translate Community. According to Google,
In the new community, you’ll find options to help with a variety of things, including generating new translations and rating existing ones. Over time, you’ll find more ways to contribute, as well as get more visibility into the impact of your contributions and the activity across the community.
If you don’t want to join the community, you can still contribute to translation accuracy. If you find a translation that you’d like to improve, you can click “Improve this translation” and provide a translation of your own.
I don’t think this means that you will want to use Google Translate for your medical records any time soon. However, it will go a long way towards increasing the accuracy of Google Translate over time.
One of the best ways to get better quality translation from any type of MT engine is to optimize your source content first. At Content Rules, we use sophisticated linguistic software and experienced editors to help ensure that your source content is squeaky clean. For information on our content optimization services, click here.
Val is on the Advisory Board for the Technical Communications Program at the University of North Texas. When not working with customers or students, Val can be found sitting behind her sewing machine working on her latest quilt. She also makes a mean hummus.
- Preparing Content for AI: 6 Reasons Why You’re Not Ready - August 29, 2024
- How to Be Inclusive in the Workplace: My Experience as a Hard of Hearing Person - August 12, 2024
- How to Improve AI Performance? Do This One Thing to Your Content - May 20, 2024
