Machine translation (MT) is becoming more common in our industry, with many enterprises and language service providers using it to meet the increasing demand for localization. While there are still serious shortcomings in terms of quality, machine translation can be a boon for very big projects.
Ahead of the game
memoQ recognized this trend early on, and invested in MT support to empower our users, both corporate and individual. We want to make sure all of our users have the tools they need, regardless of the size or scope of their projects.
As part of memoQ version updates, we’ve introduced numerous and robust integrations with MT engines. Now in memoQ 9.3, we’ve added support for Amazon MT.
Amazon MT
One great feature of Amazon MT is custom terminology, dubbed Named Entity Translation Customization. A fundamental issue with machine translation in general is the inability for it take context into account, particularly when it comes to names which are unique to certain brands, organizations or industries.
When a MT engine stumbles upon such a name, usually it handles it in one of these ways:
- Translates the term, if it’s not recognized as a proper noun.
- Keeps it in as per the source text, if recognized as a proper noun.
But in the world of marketing, not everything is black or white. Sometimes brand names in one country manifest as something entirely unrelated in another. To illustrate, let’s look at the example of Japanese pen brand Sailor. In China, it’s known as “写乐”, which also happens to be the name of the iconic woodblock print designer Toshusai Sharaku.
The above approaches would fail in this instance. Amazon is trying to resolve this issue by allowing translators to create custom terminologies, while memoQ provides an interface for you to upload or select terminologies.
Please note that you will need an AWS account to make use of Amazon’s pay-per-use service.
16 machine translation service vendors through plugins
In addition, we’ve also made improvements to our existing MT plugins. We know that Google MT and Microsoft Translator, two of the most frequently used MT engines, sometimes have trouble with certain types of documents, especially those with lots of tags. Manual correction of embedded tags is not fun, and is prone to human error. In version 9.3, we’ve improved tag handling with these two MT plugins.
We also give you a choice: you can send your request to Google MT and Microsoft MT as either XML or HTML.
