The five junior faculty members in UChicago’s Department of Linguistics are investigating a wide range of phenomena using a variety of methods. From theory to experiment to engineering, these assistant professors are “bringing fresh perspectives to traditional lines of inquiry in the department, and opening up creative and exciting new ones,” says department chair Chris Kennedy, William H. Colvin Professor.
BUILDING BETTER AI
Allyson Ettinger’s research on humans and artificial intelligence systems is “motivated by a combination of scientific and engineering goals.” On the human side, she uses computational methods to model and test hypotheses about the brain’s processing of language. “As a sentence unfolds word by word, how is the brain accessing the meanings of those words and combining them into meanings of phrases? What is the role of probabilistic understanding?” On the AI side, she works on establishing reliable assessments of the linguistic capabilities of natural language processing (NLP) systems.
“Recent years have seen impressive and dramatic success in the performance of NLP systems in AI, leading many to conclude that humanlike language understanding may have been achieved in these systems,” Ettinger says. Her research group uses insights from linguistics and cognitive science to clarify how we understand human-based standards, adapting controlled methodologies from the study of humans to develop more precise assessments for language capabilities in AI systems. In particular, she focuses on testing their ability “to perform robust and systematic compositional meaning understanding”—that is, to combine the meanings of smaller parts of language to produce more complex meanings.
CONSTRUCTING A SENTENCE
Erik Zyman also investigates how complex linguistic expressions are built out of simple building blocks. As a theoretical syntactician, he is interested in “identifying the fundamental operations that build syntactic structure in human language.” One basic operation, which linguists call Merge, combines two words or phrases into a larger unit. Although Merge is fundamental to many versions of generative syntax—an approach according to which grammar is governed by deep laws—Zyman says the operation is difficult to define with precision while both satisfying the relevant conceptual requirements (elegance, simplicity, and others) and accounting for the properties that the relevant syntactic structures have in human language. In a forthcoming article in the journal Syntax, Zyman says he “develops a novel formal definition of Merge that overcomes some drawbacks of previous ones while building on their strengths.”
Zyman aligns himself with a tradition of seeking order beneath complexity: “The second-century Alexandrian syntactician Apollonius Dyscolus was convinced that syntax is fundamentally orderly and rationally comprehensible. Nearly two millennia later, Noam Chomsky, building on a view of Galileo’s, stated that ‘Nature is in fact simple and it’s the task of the scientists to show how that’s the case.’ I agree with both of them.”
CONSTRUCTING AN IDENTITY
Two scholars are examining the social construction of identity through language.
Sharese King, a Neubauer Family Assistant Professor, investigates the relationship between race, place, and language, examining cross-regional variation in African Americans’ identity and speech. Inspired by her family’s linguistic diversity, King asks “how African Americans perform Blackness across contexts and the role language has in such performances.” She has conducted research with speakers in Bakersfield, California; Rochester, New York; and, most recently, in Chicago through the Chicagoland Language Project, a collaboration with Northwestern University that seeks to “document the diversity of language and life” in different South Side neighborhoods.
King’s research aims to address the social and political consequences of racialized language. In addition to publishing in journals such as Language, the Annual Review of Linguistics, and the Journal of Sociolinguistics, she has been featured on podcasts, trained court reporters on transcribing African American English, and coauthored an op-ed in the Los Angeles Times with UChicago psychologist Katherine D. Kinzler arguing that “anti-racist reforms by institutions and individuals will require thinking about race and speech—and opening our minds to the equal value of speech in all its forms.”
Tinks (Royâle) Bermúdez approaches the construction of race—and gender—by looking at humor, investigating how “counterhegemonic humor and counternarratives” can change people’s self-understanding and mediate their relationship to oppressive forces. Bermúdez has examined how the Naso (Teribe) people of Panama use humor to engage in social commentary and mock Indigenous stereotypes, and they argue that a similar process, which they call euphoric transmutation, underlies practices of resistance by transgender people of color. The dominant language representing race and gender, they say, “doesn’t allow us to connect with our bodies or ourselves, and it makes us exist in a state of freeze or dissociation.” Countering dissociation, euphoric transmutation—which can involve teasing or other playful inversions of power relations—is meant to be “an active process in everyday life that changes things from the bottom up and gives safety and joy to oppressed people.”
Bermúdez also investigates how interspecies relationships—between humans and both animals and plants—give us a deeper connection to our own bodies and to our selves; in particular, how “practicing interrelation” with other living beings may encourage a better form of relation to transness and Blackness among humans.
CHOOSING YOUR WORDS
“We sometimes lose sight of how impressive language production really is because we do it all the time,” says Monica Do, who uses experiments to better understand the relationship between thought and language. “Language involves making choices,” she says. These choices include not only deciding what to talk about but also what form an utterance will take. “We can say ‘A cat attacked your houseplant.’ Or ‘Your houseplant was attacked by a cat.’ Or simply ‘Your houseplant was attacked,’” Do says. “What makes us choose one way of saying things over others?” (She says we might choose the final expression if our cat did the attacking.)
Decisions like these—which can be explicit or subconscious—are the subject of Do’s experiments, which use techniques such as eye-tracking to investigate the many cognitive and social factors affecting speakers’ linguistic choices in English and other languages. “It’s not easy, a little like trying to read people’s minds,” she says. “But the insights we get from our experiments can help to shed light on the commonly utilized but poorly understood task of spontaneous language production.”