Some of these possible translations are used more frequently than others. So researchers often use that data to help build translation systems.Įven the most common phrases have lots of possible translations. For example, the European Parliament translates their proceedings into 21 languages. Luckily, there’s lots of double-translated text already sitting around in strange places.

In the same way that the Rosetta Stone was used by scientists in the 1800s to figure out Egyptian hieroglyphs from Greek, computers can use parallel corpora to guess how to convert text from one language to another. This double-translated text is called parallel corpora. Making Computers Translate Better Using StatisticsĪfter the failure of rule-based systems, new translation approaches were developed using models based on probability and statistics instead of grammar rules.īuilding a statistics-based translation system requires lots of training data where the exact same text is translated into at least two languages. The way we speak English more influenced by who invaded who hundreds of years ago than it is by someone sitting down and defining grammar rules. Human languages are full of special cases, regional variations, and just flat out rule-breaking.

The problem is that human language doesn’t follow a fixed set of rules. It didn’t work reliably for real-world documents. Unfortunately this only worked for simple, plainly-structured documents like weather reports. Some of the smartest linguists in the world labored for years during the Cold War to create translation systems as a way to interpret Russian communications more easily. Linguists came up with complicated rules and programmed them in one-by-one. This is how the earliest machine translation systems worked. That worked! If we just keep adding more rules until we can handle every part of grammar, our program should be able to translate any sentence, right?

Translate picture language