Eugenia Rho, an assistant professor in the Department of Computer Science at Virginia Tech, believes that the first moments of a vehicle stop are crucial, particularly when it comes to interactions with Black drivers. Rho and her research team conducted a study analyzing the language used by law enforcement officers during these stops and found that the officer’s initial 45 words can often indicate how the stop will unfold.
According to Rho, the study revealed a significant difference in how officers communicate with Black drivers during escalated stops that result in arrest, handcuffing, or searches compared to stops that don’t end in such outcomes. In escalated stops, officers tend to start with a command rather than providing a reason for the stop.
“We found that there’s a key difference in how officers talk to Black drivers during the first moments of stops that end in an arrest, handcuffing, or search versus those that don’t end in such outcomes,” said Rho, who leads the Society, AI, and Language (SAIL) research lab at Virginia Tech. “Simply put, the officer starts off with a command rather than a reason in escalated stops.”
The research, published in the Proceedings of the National Academy of Sciences, also discovered that Black men could often predict the outcome of a stop simply by listening to those initial 45 words, which typically lasted less than 30 seconds. Rho emphasized that there is a distinct linguistic pattern in escalated vehicle stops, as identified by trained coders, computational language models, and, most importantly, Black male citizens.
“There’s a clear linguistic signature to escalated vehicle stops. It was discerned by trained coders, computational language models, and perhaps most importantly, by Black male citizens,” Rho sai
Rho initiated this research while working as a postdoctoral researcher at Stanford University alongside Jennifer Eberdhardt, a professor of organizational behavior and psychology, and Dan Jurafsky, a professor of computer science and linguistics. The research team, which included researchers from the University of Michigan as well, analyzed audio recordings and transcripts from 577 vehicle stops that took place over a month in a racially diverse medium-sized city in the U.S. The data included stops resulting in arrest, handcuffing, or searches, but not those involving the use of force.
The team focused on Black drivers due to their disproportionate representation in the data. Rho explained that less than 1 percent of the escalated stops in their sample involved non-Black drivers. While both male and female drivers were included, the majority of escalated stops involved male drivers.
“We limited the study to Black drivers because less than 1 percent of the escalated stops included non-Black drivers in our sample,” Rho said. “We included both male and female drivers, but escalated stops were predominately male drivers.”
The data was used in two studies outlined in the research paper. The first study examined the language used by officers during the initial moments of a traffic stop, while the second study aimed to understand how Black men perceived those words. The paper also included a case study analyzing the first moments of the traffic stop involving George Floyd in May 2020.
In the first study, the researchers employed computational linguistics and hand annotation to analyze the transcripts, identifying various dialog acts such as greetings, commands, questions, reasons, and more. Dialog acts serve as roadmaps for conversations, indicating the speaker’s intention and guiding the flow of the discussion.
To account for factors that could affect the language used, such as the reason for the stop and the area’s crime rate, the researchers implemented controls during their analysis.
The study revealed that stops ending in escalation were almost three times more likely to begin with an officer issuing a command to the driver and 2.5 times less likely to provide a reason for the stop. Rho summarized the findings by stating that escalated stops often started in an escalated manner.
In the second study, the researchers played audio recordings of the traffic stops to a nationally representative sample of 188 Black male U.S. citizens, who varied in age, region, education, and political ideology. The participants listened to 10 stops at random, five of which resulted in escalation and five that did not. They were asked to assume the role of the driver and then surveyed about their feelings and predictions regarding the outcome of the stop.
The findings indicated that Black male participants used the language used by the officer as a guide to predict whether the stop would end with handcuffing, searching, or arrest. They predicted that 84 percent of stops involving orders without reasons would escalate. Moreover, they expressed concerns about the use of force in over 80 percent of stops involving orders without reasons, compared to only 47 percent of stops with reasons but no orders.
To investigate if the same linguistic pattern was present in stops involving the use of force, the researchers conducted a case study analyzing the initial moments of George Floyd’s encounter with the officer on May 25, 2020. Within less than 30 seconds, the officer issued 57 words through physical orders, while Floyd responded with 11 speech turns that included apologies, seeking reasons for the stop, asserting innocence, expressing fear, and pleading with the officer. However, the officer’s response to every dialog act from Floyd was simply another order.
Rho emphasized that, at a time when vehicle stops involving the use of force often attract national attention, it is important to understand interactions between police officers and citizens during more common vehicle stops. She highlighted that the average citizen is most likely to encounter law enforcement through vehicle stops, making it crucial to improve communication during these encounters.
While the studies provide valuable insights, Rho hopes that they will spark conversations about enhancing de-escalation training for law enforcement and fostering better relations between Black communities and the police.