Those crazy Googlers

One of my favorite institutions in the world is Google.  Sure, thier terrifyingly complete data collection does leave doubts in my mind, and I do believe that they’re working with both the NSA and the CIA, but I couldn’t live without the services they offer me.  They’re the institution that’s most dramatically hurling us towards “Everything-That-Ever-Was-Available-Forever,” a phrase coined by Patton Oswald that describes the day you can fire up your e-reader/laptop/tablet and, for a nominal fee, download any combination of words, images and sounds that ever was created.  A large part of that is making texts from other languages available, because only a fraction of everything worth reading is written in English, and a slightly larger fraction is ever translated into English.  For the most part, most texts aren’t going to be worth the time, effort and cost to translate into other languages.  But if it was possible to provide a translation that’s free and significantly better than current translating services, then it would be possible to provide this service automatically in the ETEWAF world.

Enter Google.  Using the power of large data sets and the internet, their Google translate is already the industry leader and eons beyond what I used when writing 250 words in Spanish II.  Thier academic team is approaching the most difficult aspect of translation: poetry.  Because every word and line break in poetry is carefully considered, there arise significant issues even in a direct, literal translation.  Do you preserve the form or meaning?   Is it even possible to translate the form? This specific word has a connotation which is significantly different from its denotation.  Which do you use?

Google Machine Translation

What Google seems to have done here is (and take with salt; IAmNotAComputerScientist) is use a system which kills branches on the tree before they can take down the house.  A Google translation starts with a large computer (fittingly) which, like Deep Blue, searches out all the possible translations for any one given word.  In their technical paper, they describe the way that they take a given translated word and cut it out because it wouldn’t work with the meter, saving CPU cycles that would’ve been wasted if that one word had continued to be treated as a possibility.  It seems like it still needs work, but it’s certainly fascinating.  Ultimately, I don’t feel as if it’s a threat to translators at all; it’s just bringing us closer to ETEWAF.

Google Academic Paper on Machine Translation of Poetry

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: