Poster Session

In order to encourage participation at the conference, especially for students at the 2025 LSA Summer Institute, we are continuing to accept poster submissions until JULY 10, 2025. Poster acceptances will be announced on a rolling basis until July 12, 2025. We invite posters on all kinds of projects at the intersection of criminal justice and linguistics, from completed research to works-in-progress to projects in the ideation phase. Please visit the registration and poster submission form.

Title: Language-based claims and the California Racial Justice Act

Authors: Tilden "Tilly" Brooks, Tanajia Moye-Green, Camryn Hayes and Kendal Fitzhugh

King and Jacobs (2024) discuss how racial bias manifests in the courtroom and name strategies to identify and attend to such biases. They highlight two pieces of legislation, the Racial Justice Act and the Racial Justice Act for All (collectively, the “RJA”), which bar the use of discriminatory language during criminal trials (Cal. Penal § 745(a)(2)). The passage of this law is, in principle, a step forward. However, research has revealed that successful RJA claims are few in number (MacLean, 2024) and that much confusion remains the implementation of its provisions (Chien et al., 2024). 

In this ongoing interdisciplinary work, we utilize methods from law, linguistics, and sociology to investigate cases involving the RJA’s prohibition on discriminatory language. Our main research questions are as follows:

 - What do RJA claims about language look like in practice?

 - How do judges react to language-centric RJA claims?

 - Which arguments are most successful in language-centric RJA cases? 

 - What barriers do parties alleging the use of discriminatory language in violation of the RJA face? 

We identify judicial opinions citing Cal. Penal Code § 745. We next collect the following information for each case: venue, court level, whether the case took place in federal or state court, context, relief sought, decision, and reasoning. To assess the linguistic aspect, we evaluate cases in which two kinds of claims arise: (1) claims in which discriminatory language is used to evoke racial bias, and (2) claims in which the alleged bias relates to the linguistic traits of a racialized minority.  In addition to qualitative coding, we plan to conduct interviews with legal practitioners.

A desired outcome of this work is a comprehensive illustration using descriptive statistics and case studies to demonstrate how the RJA can be and has been implemented as a means of advancing linguistic justice within the criminal legal system.

Title: The impact of social perception on message perception

Authors: Taela Dudley

It is well known that language ideologies, socially driven beliefs about how one should speak, can trigger assumptions about a person’s identity and character (Silverstein, 2003). For instance, Mainstream American English (MAE) and African American English (AAE) are dialects of English that drastically differ in how they are socially perceived–  the latter being extremely stigmatized due to racial and class bias. It has also been posited that people are afforded different levels of testimonial credibility purely as a factor of social status– a phenomenon referred to as ‘testimonial injustice’ (Fricker, 2008). The current study investigates how social bias and language ideologies influence a listener’s interpretation of different speakers, or how language ideologies contribute to testimonial injustice. This is explored through investigating interpretations of implausible sentences spoken in MAE and AAE.  

Gibson et al. (2013; 2017) found that implausible sentences (e.g., 'the mother gave the candle the daughter') could be interpreted in two ways depending on their surrounding context. When surrounded by other implausible sentences, they were interpreted literally, as implausible. When surrounded by plausible but ungrammatical sentences, they were reinterpreted plausibly under expectations of syntactic errors. The current study tests whether expectations about the speaker’s identity similarly produce different interpretations—specifically, whether listeners interpret messages differently based on speaker race.

This study compared Black (n = 77) and White (n = 74) listeners’ interpretations of implausible sentences spoken in MAE and AAE, and found that White listeners were more likely to give literal interpretations to sentences in MAE than AAE (SE= .289, p <.05). This effect was not observed amongst Black listeners, who interpreted both speakers the same. This suggests that the interpretation differences are socially motivated, given that prestige is the only perceptible difference between these two dialects.

These findings furthermore suggest that White listeners are more likely to take MAE speakers at face value, and more likely to assume AAE speakers to be error-prone, reflecting the deficit stigma historically linked to AAE (Rosa & Flores, 2017). This speaks to theories of testimonial injustice by demonstrating that a speaker’s believability is subject to the listener’s expectations for the speaker, which is influenced by social bias. 

Title: 💰❓✅👍Emoji Use in Legal Matters

Authors: Brianna O'Boyle

Be careful what you say or text, because it may end up costing you in court. A thumbs up emoji cost a farmer $80,000CAD when a thumbs up emoji was determined to be a contractual agreement (South West Terminal Ltd. v Achter Land, 2023 SKKB 116, 2023). While many people generally describe emoji as cute little images, they have began to permeate into online communication with over 90% of the online population using emoji (2015 Emoji Report in Goldman 2018). One issue that emoji have faced in the legal system is there have not been consistent manners in which judges or courts have treated them. On one hand some judges like Judge Forrest in the 2015 Silk Road trial allowed jury members to see posts that contained emoji, rather than only having them read aloud without emoji. Though on the other side sometimes emoji/emoticons have not been taken into consideration as evidence in people’s online communication, such as in the case of United States v. Elonis where the defense’s request to have texts read with :-P were denied. Relatedly, some have looked into how emoji can be used to lie or disguise criminal intent (Danesi 2012, Weissman 2024).

Another issue emoji have in the legal setting are their ability to be misinterpreted (Adam 2018,  Goldman 2018). There also are differences in which people from different cultures or contexts could interpret the same emoji (Miller et al. 2016, O’Boyle and Doyle 2023). This project proposal seeks to investigate comparable court cases as case studies, in which emoji have played an integral role as creating an argument or defense in understanding intent. Court cases would provide an ideal opportunity to analyze the linguistic meaning behind emoji use and how emoji in evidence contribute to case outcomes using discourse analysis and other methods. Additionally, there may be potential that critical discourse analyses may reveal deeper societal power structures if patterns emerge in how emoji are included or excluded as evidence in certain cases. Language is a complex, yet imperfect system—miscommunications occur daily, and affect many individuals, but language is a system that is resilient and ever changing. With the emergence of emoji and rise of CMC (computer mediated communication) it is even more important that we bring attention to their use and take steps to linguistically investigate how to interpret them; especially in instances that their use can have impactful, real world consequences such as in the eyes of the law.  

Title: Making sense of the police: Racial category work in talk about police–citizen encounters

Author: Hannah Fedder Williams, Georgetown University

Given the historical and current tensions between law enforcement and citizens in the U.S., it is unsurprising that many scholars have dedicated their work to analyzing police–citizen encounters and discourse about the police (e.g., Andrus 2019; Kidwell 2021; Kurtz & Upton 2017; McElhinny 2003). In Washington, D.C. in 2024, a controversial bill was passed, loosening restrictions on how officers can search and pursue citizens (Ward 2024). Despite frequent critiques of crime control in majority Black areas of the city (e.g., Schaffer 2024), to my knowledge there has not been recent work investigating this topic in D.C. Some microinteractional work in the U.S. has explored the relationship between race and policing, for example how race is covertly alluded to by officers making sense of their practices on the job (Jones et al. 2023) or overtly mentioned to make sense of police shootings in news media (Shrikant & Sambaraju 2021). As McElhinny (2003) argues, officers’ talk about policing not only perpetuates (potentially) harmful ideologies about policing, but also serves to reify the practices themselves. However, to my knowledge no work has explored specifically how citizens and police officers make sense of race(ism) in talking about policing in Washington, D.C. 

To address this gap, I use Membership Categorization Analysis to add to the body of discourse analytic work that examines how speakers bottom-up do racial categorization, following Shrikant & Sambaraju (2024). The data come from semi-structured interviews with 10 citizens and 11 police officers, the analysis of which is supplemented with field notes from 9 police ride-alongs in the city. The poster will present preliminary analytical findings showing how participants 1) foreground racial categories (e.g., ‘Black’, ‘white’) in self- and other categorizations to use them as an explanation or justifications of policing practices they have observed or conducted or 2) background racial categories to deny race or racism as meaningful for policing (foregrounding others such as gendered categories). As a law professor points out in describing her experience going through the police academy in D.C., her instructors’ only mentions of race were to “insist that it didn't matter” (Brooks 2021, p. 122). This study will help solidify how race indeed matters for people’s sensemaking of policing – whether they are officers or citizens – and hopefully inform empirically driven reform of law enforcement trainings. 

Title: Witness Recall for Conversation: An Empirical Check on the Hearsay Rule

Author: Susan Provenzano, Georgia State University

It is often said that a trial is a search for truth. To that end, the rules of evidence regulate the admission of facts; witnesses testify under oath; and cross-examination probes for lies and errors. A well-known body of empirical research shows that these measures fall short, however, in a critical respect: exposing inaccurate eyewitness recall and identifications.  This Article leverages empirical findings in the psychology of language to address an arguably more common but overlooked memory risk: that posed by the “hear-witness,” a person who honestly but falsely believes her memory for conversation is intact. 

Aided by a narrow hearsay rule and its twenty-odd exceptions, conversations come into evidence all the time. In criminal cases, co-conspirators, jailhouse snitches, and police can recount almost anything they heard a defendant say.  In civil cases, witnesses can repeat everything from overheard warnings to narrated events to excited accusations. This hands-off approach to hear-witnesses rests on unfounded assumptions about conversational recall, which empirical studies show degrades in just minutes to days, with only the conversational “gist” retained for the short term.  What is initially retained may later be lost or inaccurate, and when recounted at trial, may be tainted by the witness’s interpretations and corrupting influences hard to expose through cross-examination. 

In addition to this research, we present novel data on another conversational memory risk: even if a witness recalls what was said, she may mis-recall how it was said. In hearsay analysis, incriminating questions are often admitted at trial while incriminating declaratives are excluded – which makes sentence form recall a hearsay problem. The data also suggests that recall varies according to sentence form, a variability that matters for gathering evidence and preparing and examining witnesses. We conclude that the empirical research so far gives reason to believe that in the current unregulated state, hear-witness memory failings impede evidentiary accuracy and the search for truth in ways that may be invisible to the trial process. We make preliminary recommendations and lay out an agenda for further empirical study.

Title: Towards an Automatic Dialect Density Measure for African American English

Author: Marie Tano, Stanford University

This study, part of my dissertation, investigates how morphosyntactic features of African American English (AAE) correlate with officer behavior in high-stakes contexts like traffic stops. I develop a feature-based Dialect Density Measure (DDM) using GPT-4o to quantify AAE presence in transcribed speech across 15 morphosyntactic constructions (e.g., habitual be, multiple negation, zero copula). This fine-grained approach enables interpretable analysis of dialect salience and syntactic variation, with applications in sociolinguistics and responsible NLP. To assess model performance, I evaluate GPT-4o on 186 human-annotated sentences from the CORAAL corpus. While the model achieves near-perfect accuracy for high-surface-form features (e.g., ain’t, finna), it underperforms on syntactically complex or ambiguous cases (e.g., existential it), with results detailed through per-feature F1 scores and frequency correlations. These findings underscore the challenges of using large language models for racialized language analysis. The broader dissertation goal applies the DDM to police body-worn camera transcripts, examining potential links between AAE density and officer behavior (e.g., respect, escalation, credibility judgments). Ongoing work extends the DDM by incorporating fine-tuned LLaMA models, phonological features via self-supervised speech models, and human perceptual experiments.

Title: Hearing Blackness and Criminality: Examining Listeners Criminalization of Black American Speakers

Author: Dominique Branson

This study investigates whether U.S. listeners associate Black-sounding voices with criminality, and whether these associations vary by the speaker’s race and gender. One hundred U.S.-based speakers (Black and non-Black) provided brief speech samples, which were rated by 300 listeners for perceived racial identity (“voice racialization”). A separate group of 100 listeners participated in a psycholinguistic memorization task, recalling objects—both crime-related and non-crime-related—after hearing each voice. A “Black racialization score” (proportion of listeners perceiving a speaker as sounding Black) and a “criminalization score” (proportion of recalled objects that were crime-related) were calculated for each speaker. Regression models revealed that higher Black racialization scores predicted higher criminalization scores, with the effect significant only for Black men. These findings indicate that listeners more strongly associate voices perceived as Black—particularly Black men’s voices—with criminality. Results highlight the role of raciolinguistic bias in reinforcing gendered racial stereotypes, with implications for legal contexts and potential protections under Title VI of the Civil Rights Act. 

Title: Linguistic Dexterity amidst Linguistic Inadequacy: Narration, the legal tender

Author: Tasheney Francis

This paper is a part of a larger project examining televised interactions between traumatized Creole-speaking resident witnesses of an inner-city community and English-speaking jurists in a truth commission, held in an English-governed Jamaican courtroom. The residents, who experienced traumatic effects of a bloody military operation intending to arrest an alleged area-leader, were called to the stand as witnesses. However, the unexpected adversarial cross-examination style raises the question of how lay, Creole-speaking witnesses might linguistically construct a favourable identity to navigate the tenuous social and situational space, and meet their need for social justice? It is only an identity that is favourable to the communicative situation that permits communicative goals being met (Goffman 1981). Constructing such an identity seems onerous considering the already prevailing negative societal perception of the community with which the witnesses are associated, the divergent social and situational demands they must meet in this televised exchange, and the noticeable fact that they do not have the ‘appropriate’ linguistic tool, English. A detailed analysis of the interaction, within the framework of intersubjectivity, reveals the witnesses’ linguistic dexterity and creativity in facing and mounting these hurdles. The results show the linguistic resource of narrativization to be among the valuable tools with which they chisel their desired identity. Given the richness of the data, this paper narrowly focuses on how the witnesses create construction space within a discourse frame that renders them powerless, and unleash the power of narratives to get the audience to trade spaces and generate social action.  Through interactional alignment (Feyaerts, Bröne & Oben 2017) the witnesses took the discourse floor even during cross-examination to launch their narratives. With strategic framing (Hallahan 2008), emboldened by narrativization, witnesses created laminated frames (Wang 2018; Gordon 2009; Goffman 1981) in which they assigned and shifted blame to reconstruct a befitting identity, in this case, as the victims. The multimodal analysis further reveals how gestures plays a critical role in enabling enactment, which evokes moral reasoning and compels intersubjectivity. In so doing, their audiences can trade spaces with them, take on their perspectives, and this prompts their desired social action, and meets their communicative needs and social goals.