Google Translate works similar to the human brain: it learns and abstracts from a huge amount of real life data. This means, it learns and improves its performance slowly over time. As laughable Google Translate results might have been a few years ago so impressive they are often by now. Of course, it will not be hard to find some ridiculous translations even now in particular when it comes to correct grammar. However, if you are aware of the service’s strength and weaknesses you have got an awesome tool at your fingertips.
The more common an expression occurs on the web and in real life the more accurate the service tends to be.
You can judge the reliability of a translation by making a ranking of linguistic elements:
- Common phrases (e.g. How are you?)
- Fixed expressions, proverbs, etc. (e.g. easy as pie)
- Collocations (e.g. stale water)
- Words in context (e.g. admitted to hospital)
- Single words in isolation (e.g. admit)
- Complex sentences (e.g. After having been admitted to hospital he quickly recovered).
Improved single word translations
In the case of complex sentences, nobody should expect by now to get completely perfect results, even though the free Google Translate has long overtaken all commercial products in this area. In the case of single words Google has recently added two enormously helpful features:
- Ranking by frequency of use
- Synonyms to double check on the correct meaning
Google has not only improved its translations but also the user experience by adding voice recognition and speech output. In addition the mobile app offers handwriting recognition as well as OCR (optical character recognition), which allows you to take a photo of, say, a sign in a foreign language and you immediately get the translation.
Of course it is often a good idea to double check with one of the many great online dictionaries. However, I tend to use these less and less these days as Google Translate has become my primary dictionary. I think we can expect further improvements of Google Translate, both as far as the quality of translation is concerned as well as regarding user experience.