You are here

The Science of Understanding Yoda

Seed funding supports research using technology to learn foreign language
Posted September 15, 2015
CREDIT: Concept by agencybcg.com CREDIT: Concept by agencybcg.com

What if you could learn a language just by reading a Buch you enjoy?

If your brain automatically translated "Buch" to book you just got a taste of a new method of foreign language instruction currently under investigation thanks to seed-grant funding from the Science of Learning Institute, a priority of the Rising to the Challenge campaign.

Serious obstacles to learning a new language — such as German — are mastering the sheer size of the vocabulary and the rules that govern the language. But imagine that you began to learn German by reading an "English" novel that included German syntax. The experience might be something like trying to understand the Star Wars character, Yoda. For example, you might encounter sentences such as "I would gladly a bread loaf eat." Without much effort, you would start to become accustomed to the feel of German word order and usage. 

Thanks to a seed-grant program made possible by support from an anonymous donor, researchers are exploring whether technology can help students learn a language by absorbing its grammar and vocabulary from digital texts that mix a smattering of new language into old, progressively adding more and more of the new language as students’ comprehension increases.

"The idea is to learn a foreign language without really actually sitting down and learning it; just kind of reading it," says Philipp Koehn, the study’s principal investigator and a professor in the Whiting School of Engineering's Department of Computer Science.

The seed grants are intended to jumpstart research that crosses traditional academic silos, allowing investigators to enter uncharted territory by partnering with colleagues from other divisions to create a synergy of knowledge and discovery.

With the grant, Koehn and colleagues from his own department and the School of Education hired a graduate student to work on the computer algorithms they hope will create a fluid hybrid of two languages, gradually adjusting elements like word order and vocabulary complexity to match students' increasing readiness. The next step will involve testing the program with students studying foreign languages at the Krieger School of Arts and Sciences. Eventually, the researchers hope to have enough data to take the investigation to the next level.

The idea first came to Jason Eisner, Koehn's co-investigator and an associate professor of computer science, when he was studying French in high school. The language seemed easy, he thought, except for all that vocabulary. He imagined a novel that started in English, added French word order, and slowly replaced English words with French ones.

"I never wrote that book," Eisner says, "but it seemed to me more recently that maybe we could generate this more easily by using machine translation" — for example, the Google "Translate" function. If it works, he says, learning a new language will be more like a child learning a first language, taking in linguistic patterns through exposure to topics of interest.

Chadia Abras, associate professor and associate dean for online teaching and learning in the School of Education, contributes expertise that grounds the technology in concepts of education theory, aligning the technology with what is known about language acquisition. Once the program is in the testing phase, Abras will lead the evaluation process.

The colleagues are a great match, Abras says, with Koehn and Eisner's background in system building and machine learning, and hers in theories of teaching, learning, and language acquisition.

"We needed each other, because each of us on our own would not have been able to obtain a grant or complete this study," she says.

"We’re grateful to the donor who made it possible to do this exploratory research," Eisner agrees. "We never would have done this without the Science of Learning grant because I have no track record with educational technology, so I wouldn’t have applied for — or gotten — a grant in this new area. It's a way to try something that’s speculative, new, and exciting."