Man and machine have made significant progress in the field of translation. From retooled apps to deep learning algorithms, translation is faster and more accurate than ever before. More languages are supported across more platforms and more inputs–text, voice and camera. At the same time, authentic language, accurate translation, and contextualized understanding continue to be challenge.
How much of a challenge? Microsoft had an epic fail with Tay and Samsung’s virtual assistant Bixby was DOA. Recent studies and unser testimony also point to biases baked into the code, indicating that we are not quite ready to ditch the dictionary yet.
There is certainly a great deal at stake in the race to crack–or really create–the universal language code. The universal translator is not near, but that doesn’t mean the end is nigh. There is a lighter side to language folly. I happen to have seen two in the past few days. These prove that no matter how dependent we are on algorithms, we are still in command because we can express ourselves thoughtfully and laugh when the results aren’t quite ready for prime time.
Case #1: One, Two, Three Garcías
On April 15th, three outfielders with the same surname started for the Chicago White Sox. This marked the first time on the history of Major League Baseball that this occurred in any starting outfield–or any outfield combo for that matter. This was significant news for the Chicago White Sox organization, MLB and and the #PonleAcento pushers–García, not Garcia– like me.
This news came my way via @loswhitesox on Twitter, one of the many Spanish language feeds I follow. I’m in the habit of checking the translation via Bing–partly for curiosity and partly to contribute better translations where I can. On this occasion, the original tweet and translation looked like this:
Where to begin with this one? There are really two fundamental problems in the translation: one concerning the word “jardineros” and one concerning the word “partido.” To Bing’s credit, the primary dictionary definition of the word “jardineros” is in fact “gardeners.” Ditto for “partido” and “political party.” Unfortunately, what the dictionary definition and the advanced algorithms don’t take into account is any contextual understanding. In the world of baseball, “jardineros” means “outfielders” and “partido” means “game.” This is unequivocal and to miss this given the context clues is unacceptable.
This is a pretty clear case of not seeing the forest for the trees; in an attempt to get the discreet words right, the global, contextualized meaning is lost. Now, I’m not a coder and I’m not contracted by Bing, Google or any other outfit to advise their operations. (Perhaps the wild success of this post can change that?!) Despite my limitations, I have to believe that there is a way to tag words with contextual cookies to say that if you are talking baseball, “partido = game,” but if you are talking politics, “partido = party.” There are thousands of examples: colloquial uses, culinary uses, high culture uses, popular culture uses.
Now, there are regional differences and idiomatic expressions that will complicate this endeavor exponentially. What to do? Either set to work making the tags more nuanced or say “no más” beyond what can be considered shared/standard language. I’d recommend we really forge ahead with the former.
This case brings me back to my first year teaching when a student turned in an assignment with the enigmatic but eternally memorable line “Yo jugo jarra.” Think on that.
This case brings me back to my first year teaching when a student turned in an assignment with the enigmatic but eternally memorable line “Yo jugo jarra.” Think on that. Put yourself in the shoes of a freshman studying Spanish for the first time. Think about all the contextual clues missed and misuses of parts of speech. Stumped? “Yo jugo jarra” is mixed-up misuse of “I play pitcher.” As above, the issue here is the parsing of water pitcher “jarra” and baseball pitcher “lanzador.” I hope that sharing this example here can serve as a reminder to check the context clues and go with the appropriate selection.
Case #2: “In your presence my front is not felt”
Mixing the game of telephone with translation can cause either high tensions–in the world of “partidos políticos”–or hijinx. Jimmy Fallon chooses hijinx.
What does this bit show us? First, English cum Hungarian cum English pop interpreted by Anne Hathaway isn’t half bad. “Talent” is “talent” in English, Catalán, Czech, Dutch, and Estonian…probably more too. (Bing nails it at the word level.) Second, lyrics of pop songs, while comical in translation, stand up better to twists and turns than epic poetry might. Third, and most significantly, this bit shows us that even though Google Translate is leading the way for man-machine synergy in translation, we need to work on our own uniquely human skills to translate and evaluate based on not only syntax but common sense. We need to understand the current technology’s limitations and appreciate our own abilities as language learners. As Gloria Gaynor sort of said “I am aware that I’m aware/ I live all my live was yes.” This truth is universal.