More than a decade ago, neuroscientist Ev Fedorenko asked 48 English-speaking people to perform tasks such as reading sentences, memorizing information, solving math problems, and listening to music. While they did this, she scanned their brains using functional magnetic resonance imaging to see which circuits were activated. If, as linguists have proposed for decades, language becomes linked to thought in the human brain, then language processing areas will be activated even during nonlinguistic tasks.
Fedorenko’s experiment, published in 2011 in the journal, showed that when it came to arithmetic, musical processing, general working memory and other non-linguistic tasks, the language areas of the human brain showed no response. Contrary to what many linguists claim, sophisticated thought and language are separate things. One does not require the other. “We have this highly specialized place in the brain that doesn’t respond to other activities,” says Fedorenko, an associate professor in the Department of Brain and Cognitive Sciences (BCS) and the McGovern Institute for Brain Research. “It is not true that critical thought needs language.”
The experimental design, using neuroscience to understand how language works, how it evolved and its relationship to other cognitive functions, is at the heart of Fedorenko’s research. Along with his colleagues Roger Levy and Ted Gibson, he is part of a unique intellectual triad in MIT’s BCS Department. (Gibson and Fedorenko have been married since 2007). Together, they have established a long-standing collaboration and have built a significant body of research focusing on some of the most essential topics in linguistics and human cognition. Working in three independent laboratories – EvLab, TedLab and Computational Psycholinguistics Lab – researchers are motivated by a common fascination with the human mind and how language works in the brain. “We have a lot of interaction and collaboration,” Levy says. “It is a landscape of broad collaboration, intellectually rich and diverse.”
Using a combination of computational modeling, psycholinguistic experiments, behavioral data, brain imaging, and naturalistic language substantial data, researchers also answer a fundamental question: What is the purpose of language? Of all the possible answers to the question of why we have language, perhaps the simplest and most obvious is communication. “Believe it or not,” says Ted Gibson, “that is not the standard answer.”
Gibson came to MIT in 1993 and joined the Linguistics department in 1997. Over time, he realized that he was developing an intellectual orientation that distinguished him from his peers. For example, the field of linguistics has been significantly influenced by the ideas of Noam Chomsky, now Institute Professor Emeritus and Professor Emeritus of Linguistics at MIT, who holds that language and thinking are inextricably linked and that the capacity for language exists before learning. Gibson wanted to take an approach that could generate recent tests of some basic ideas. He began to emphasize creating his own forms of quantitative research to explore some of the outstanding issues. “The data you can get can be much broader if you get a lot of people through experimental methods,” Gibson says, adding, “I felt like I had to break it down in detail and see if there was truth to these claims.”
Three decades after starting at MIT, Gibson finds the collaborative research at BCS compelling and provocative, pointing to recent ways of thinking about human culture and cognition. “Now we are at a stage where it is not just about the arguments against. We have a lot of positive things about what language is,” he explains. Levy adds: “I would say that all three of us believe that communication plays a very important role in language learning and processing, but also in the structure of the language itself.”
Levy points out that the three researchers completed PhDs in different subjects: Fedorenko in neuroscience, Gibson in computer science, Levy in linguistics. However, for many years before their paths finally converged at MIT, their shared interest in quantitative linguistic research led them to closely follow and be influenced by each other’s work. The first collaboration between the three took place in 2005 and focused on language processing in Russian relative clauses. Gibson recalls that around this time Levy presented what he described as “a wonderful piece of work” that helped him understand the connections between language structure and communication. “Communication pressures drive structures,” Gibson says. “Roger played a key role in this. He was the one who helped me think about these things a long time ago.”
Levy’s lab focuses on the intersection of artificial intelligence, linguistics, and psychology, using natural language processing tools. “I try to use the tools provided by mathematical and computer science approaches to language to formalize scientific hypotheses about language and the human mind and to test these hypotheses,” he says.
Levy points to ongoing research between himself and Gibson on language comprehension as an example of the benefits of collaboration. “One of the most important questions is: why does language understanding fail, why does this happen?” Together, the researchers applied the concept of the “distorted channel,” first developed by information theorist Claude Shannon in the 1950s, which states that information or messages become distorted during transmission. “Language understanding develops over time and involves continuous integration of the past and the present,” says Levy. “Memory itself is an imperfect conduit that transmits the past from our brain just now to our brain now to support effective language understanding.” Indeed, the richness of our linguistic environment, the experience of hundreds of millions of words in adulthood, can create a kind of statistical knowledge that guides our expectations, beliefs, predictions and interpretations of linguistic meaning. “Statistical knowledge of language actually affects the limitations of our memory,” says Levy. “Our experience shapes our memory of the language itself.”
All three researchers say they share the belief that by following the evidence, they will eventually uncover an even bigger and more complete story about language. “This is how science works,” says Fedorenko. “Ted trained me along with Nancy Kanwisher, and both Ted and Roger are very data-driven. If the data doesn’t give you the answer you had in mind, you don’t just have to keep pushing your story. You come up with new hypotheses. Almost everything I did was like that.” At times, Fedorenko’s research on parts of the brain’s language system surprised her and forced her to abandon her hypotheses. “For one project, I came with an earlier idea that there would be some separation between the parts dealing with combinatorics and word meanings,” she says, “but every element of the language system is sensitive to both. At one point I thought to myself, this is what the data is telling us and we have to accept it.”
The works of researchers pointing to communication as the constitutive goal of language open up recent possibilities for probing and studying a foreign language. The standard claim is that human language has a dramatically richer lexicon than the language of animals, which have no grammar. “But often we don’t even know what other species communicate with,” Gibson says. “We say they can’t communicate, but we don’t know. We don’t speak their language.” Fedorenko hopes this will open up more opportunities for cross-species linguistic comparisons. “Understanding where things are similar and where they are different would be extremely useful,” he says.
Meanwhile, the potential applications of language research are far-reaching. One of Levy’s current research projects focuses on how people read and operate machine learning algorithms based on the psychology of eye movements to develop proficiency tests. By tracking the eye movements of people who speak English as a second language as they read texts in English, Levy can predict how well they perform in English. This approach may one day replace the English as a Foreign Language test. “It’s more of a hidden measure of language, rather than a test that allows you to play to a much greater degree,” he says.
Scientists agree that some of the most stimulating opportunities in the neuroscience of language lie in enormous language models that provide recent opportunities to ask recent questions and make recent discoveries. “In the neuroscience of language, stories about how the brain processes language have been limited to verbal descriptive hypotheses,” Fedorenko says. Computationally implemented models are now surprisingly good at using language and show some degree of adaptation to the brain, he adds. Now researchers can ask questions such as: What actual computations do cells perform to derive meaning from strings of words? “You can now use these models as tools to gain insight into how people might process language,” he says. “And you can take models apart in a way you can’t take apart a brain.”