Yuka Kan'no's camera focuses on a deserted road, the landscape on either side filled with rocks and debris. Kan'no is filming a documentary in Rikuzen-takata City in Iwate Prefecture, Japan, which was destroyed by the Tohoku earthquake and tsunami on March 11, 2011. She interviews her mother, Yoshie, at the site of their former home, and though both speak Japanese, the film is subtitled for an English-speaking audience, thanks to efforts led by Purdue professor of Japanese Kazumi Hatasa.
"With everything gone, I have no idea what was where. I don't even know when I'm walking," Yoshie says, scanning the landscape, blinking.
"We only see roads," adds Yuka.
"But it definitely was here, I'm certain." Yoshie says, nodding slowly to back up her statement, looking anything but certain.
When Professor Kazumi Hatasa saw a news story on NHK World about a tsunami documentary by a Japanese university student, he wondered how it could reach more people. English subtitles would help, and the process of translating them quickly became an ideal teaching tool.
The language exercise was valuable, he said, because students translated a real-world experience rather than a simulated one. "At the same time, they thought about the disaster and thought about how the people thought, and to me that’s more significant."
Anna Pesut, a senior majoring in Japanese and Asian studies who worked on the translation, is also a musician. She translates Japanese song lyrics to English as one of her hobbies, and said the language of the film was more straightforward than lyrics, which often feature metaphors and symbolism. Despite this, she didn't find her role as translator easy. "It definitely made me realize that I need to sharpen up my skills in listening and understanding immediately. My strength is reading and writing, so that was kind of a little wake-up call," she says.
Hatasa has gained permission to use the segments again as educational material, and plans to continue sharing them with faculty at other universities, who now have one good English transcript as a means for comparison.
What people mean when they speak or write is far more complex than just the words they use. In a film, we can also interpret facial expressions and tone of voice, but those aren't available in written works, where the rhythm of the prose and the cultural background of the speaker might provide clues instead. Faculty at Purdue who work in translation face the difficult task of assimilating all of these elements, and doubly so, as they move from one language to another.
But each of us translates every day when we listen to others, even when we speak the same language, says Victor Raskin, distinguished professor of English and linguistics, because language underdetermines reality. When people speak to us, we recreate their meaning in our minds. If a friend tells us that a man walked into a room, we can picture that, and are generally satisfied with the information, without asking what he was wearing and where he was coming from—or for any number of details that would tell us even more about the moment being discussed. The same process happens in translating from one language to another. "Somebody says something, and what comes up in our mind is a piece of reality, a situation, and then we render that situation the best we can into a different language. So translation is always mediated by our knowledge of the world."
Raskin, who works in structural and computational linguistics, tries to address this shortfall in translation by working to improve machine translation of natural language. While current models use statistics to, in essence, replace one word for another, Raskin suggests that words don't correspond directly to other words, and focuses instead on semantics, or meaning. A table, for example, won't have a direct replacement in a culture where tables aren't used, but functionally, there is probably another item where people put food, for example. "There is this conceptual structure, which we call ontology, where all these meanings for words correspond to certain concepts and are interwoven by relations. It turns out that putting this in the computer helps."
It will take a long time, however, before a system of machine translation using ontological semantics is completed. And the 17 Purdue students helping to translate the Japanese documentary Kyo-o mamoru (Resilience – Protecting Today) in conjunction with students from 12 other universities, had just a few short weeks to create subtitles for the entire film in time for a public screening in April 2012. Professor Hatasa enlisted his colleague Atsushi Fukada, associate professor of Japanese and linguistics, to help with the technical aspects: finding subtitling software and creating several four- to five-minute segments of the film that could be assigned to a pair of students working together at Purdue or the other universities Hatasa recruited. Students first created a written Japanese transcript from the film's dialogue before translating that into English to create the subtitles.
The students' approach seemed to roughly follow one of two methods, said Hatasa, who has never undertaken translation work before. Some students suggested a word-for-word approach, in an attempt to avoid imposing too much of their own interpretation on its subjects. Others used a more psychological method of trying to understand what a subject was thinking or feeling, and then chose a particular phrase accordingly. "Both methods are honorable, intention-wise, and both are probably acceptable approaches. But they are definitely different approaches. I guess probably all translators face that kind of thing every day." The transcript for the film, he says, probably combined both styles in equal measures.
Translating ancient works, however, presents different challenges. Charles Ross, professor of English and director of the comparative literature program at Purdue, spent 12 years translating Matteo Maria Boiardo's fifteenth-century chivalric epic Orlando Innamorato (Orlando in Love) from the Italian. This key work, in the same romantic tradition as the Arthurian cycle, had never been translated into English. He retained its iambic pentameter and when possible the rhyming stanza ottava rima, but says that every word choice was a painstaking process informed by dictionaries and annotated editions of the original works.
Even within a line, the main unit of a verse translation, there is a beginning, middle, and end, and each section should be strong. Verbs lend strength in a way that prepositional phrases do not, for example, and inserting a rhyme at the end of a line also helps the next line begin robustly. The key to poetry, he explains, is its oral quality, so to focus on each syllable, Ross stepped away from his computer and slowed things down by working through each section by hand. "You say the line out loud, and that's why you write it out in fountain pen—you have to go slow, and just listen and hear that line."
Assistant professor of Spanish Yonsoo Kim has also worked on fifteenth-century translations, but her love of languages initially grew from her scientific background. Kim, who grew up in Argentina, studied to be an optician, and eventually moved to the U.S. to open an optical center. She returned to school to improve her English, and also decided to take a literature course in Spanish. "I was thinking, 'I'm a native speaker of Spanish, so I won't have any problem understanding the text.' And I realized that I couldn't understand it at all, because it was all about interpretation." Hooked, Kim combined her interests, and now studies medieval Spanish medical documents, translating them to modern Spanish. In her first book, she described how the nun Teresa de Cartegena, who was deaf, used medical discourse to work her way into the world of letters then dominated by men.
Kim soon realized that many similar texts were hidden in monasteries. Only one or two scholars might have access to each document, and after taking several years to translate it, would have no way to alert other scholars. So Kim enlisted Constantino Malagón Luque, a colleague in artificial intelligence at Nebrija University in Spain, to organize the Medieval Medicine Documents Identification System (MMEDIS). Their team is working on a computer program to perform automatic paleographic transcription, which would automatically read and translate the handwriting, abbreviations, contractions, and acronyms common to the period, so that the documents can be catalogued and accessed by all.
The program is still under development, but the results could be useful to many types of scholars. Colleagues in the Department of Dermatology at Columbia University are already scouring available documents for references to medieval treatments for cancer, which they hope may lead to new lines of research. In this spirit of interdisciplinary cooperation, Kim also helped found the Association of History, Literature, Science, and Technology (AHLiST), and is working with Luque on the development of a Humanities Integration Technology (HIT) system. This web portal, modeled after the music website "Songbird," will allow researchers to search for and hand-select specific content related to their work. If you wanted to compile resources on El Cid, the famous twelfth-century Spanish epic about the eponymous Castilian military leader, for example, a keyword search might allow you to see a cover of the manuscript and some pages of the text, along with a modern transcription or translations in several other languages. You could also choose to browse images of or journal articles on El Cid, and find out which archives hold these materials.
Purdue faculty embrace technology in their translation work, and the notion that computers will eventually replace humans in translation efforts is a false dichotomy, says Raskin. Computers are fast, never tire, and only make mistakes if we've programmed the errors in the first place. But they also handle simpler tasks, and aren't as creative or intelligent as humans. "At least at this point, the best results should be achieved in in hybrid teams; the computer does what it does best; people do what they do best," says Raskin.
And while Ross may sometimes use a fountain pen, he also created a video series called "First Lines" to introduce translation in his classes. Writers, faculty members, and graduate students who are native speakers in other languages read the first lines of famous texts in that language and loosely translate them in English, providing context from the text's culture of origin, as well.
"How should I put it?" multiple subjects in Kan'no's film ask, seeking help to accurately convey their thoughts on life and loss after the tsunami. Hatasa, Fukada, and their students asked themselves the same question, and in answering it, gave their subjects a new voice.