The legal system is concerned with interpreting and applying existing laws to real world circumstances. In the U.S., the final arbiters of these laws are the nine justices of the U.S. Supreme Court. The cases presented before the court generally involve disputes between two parties, the petitioner and the respondent. However, in some cases third parties provide additional information to the court as amici curiae (‘friends of the court’). The present study uses corpus statistics and moral foundation theory to explore and compare the types of arguments brought up by the different parties during oral arguments over the past 6 decades, with a focus on amicus curiae arguments and how they affect the results of the case. Overall, the results show that both amici and respondents rely on moral arguments more frequently than petitioners do.

The legal system is concerned with interpreting and applying existing laws to real world circumstances. In the U.S., the final arbiters of these laws are the nine justices of the U.S. Supreme Court. Over the past few decades, political leanings have played an increasing role in how these justices are appointed. As such, it seems apt to investigate the role that moral reasoning plays in their world views and their decision making. The present study uses corpus statistics and moral foundation theory to explore the types of arguments brought up by different justices during the 2004-06 terms. More importantly, the study investigates how the use of moral language by advocates arguing in front of the court influences the vote of the justices. The results show that increased use of moral language and rhetoric, especially loyalty, can affect the outcome of the case.

During conversations, it is not uncommon to notice that interlocutors start using similar words and grammatical structures.This alignment of language use is thought to help comprehension, as well as lead to an alignment in underlying represen-tations. In the context of negotiations, the degree to which parties exhibit such an alignment can indicate the likelihoodof reaching an agreement. The present study expands this notion to the courts and uses corpus statistics to examine therelationship between the alignment of semantic content during oral arguments and the decision reached by the justices.The analysis demonstrates that lawyers that align their language with that of the justices are more likely to have a decisionin their favor. Additionally, as befits the power dynamic between justices and lawyers, lawyers are more likely to aligntheir language with the justices than the justices are to align their language to that of the lawyers.

Many theories of discourse structure rely on the idea that the segments comprising the discourse are linked through inferredrelations such as causality and temporal contiguity. These theories often suggest that the information needed to determinethe relation can be found when the discourse is interpreted through the application of world knowledge. However, Sanders(1997) found that the interpretation of ambiguous relations can be affected by the discourses genre. Similarly, Sagi (2006)reported that participants were faster to interpret discourse relations when they were preceded by the same discourserelation. The present study demonstrates that exposure to discourse relations such as result (e.g., John passed Mark ina marathon. He won.) or explanation (e.g., John ... He was in great shape.) can affect the interpretation of subsequentambiguous relations encountered in an unrelated context. This result suggests that discourse relations are representedindependently of the context in which they appear.

The disambiguation of pronouns is a complicated process that has been shown to be influenced by both linguistic andcognitive factors. In particular, readers prefer an interpretation that is causally likely. For example, in the sentence pairJohn accused Mark of stealing a car. He called the police, readers judge that the antecedent of he is more likely to beJohn than Mark because of the perceived causal link between the accusation and calling the police. I will describe newresults that explore how the presence of negation affects such interpretations (e.g., He did not call the police). While,as expected, negation disrupts the perceived causal link, this disruption does not affect the choice of antecedent (John isstill the preferred antecedent). This suggests that readers identify the unnegated causal relationship when interpreting thenegated sentence. The implication of this result to models of pronoun disambiguation will be discussed.

Many of the important decisions we make have moral implications. Moral Foundations Theory (Haidt & Joseph, 2004) identifies 5 distinct styles of moral reasoning that may be applied to such decisions. This paper explores how reading text that emphasizes one of these styles might affect our reasoning. After participants read a series of tweets that emphasized the Fairness/Cheating foundation they exhibited an increased reliance on this style compared to when they read tweets emphasizing the Care/Harm foundation. This affected participants’ answers to a questionnaire designed to measure the perceived importance of the different foundations, as well as in their rating of the foundations evident in other tweets. Interestingly, this effect was short lived and was not observed for the Care/Harm foundation. These results suggest that exposure to the moral reasoning of others might temporarily influence what moral arguments we are likely to accept and employ.

Recent advances in generative language models, such as ChatGPT, have demonstrated an uncanny ability to produce texts that appear to be comparable to those produced by humans. Several key empirical results related to human processing of language, such as analogical reasoning, have been replicated using these models. Nevertheless, there are some important differences between the language generated by these models and language produced by humans. In this paper, I examine how LLMs performs on two pronoun disambiguation tasks reported on by Rhode, Levy, and Kehler (2011) and Sagi and Rips (2014). While LLMs performed reasonably in these tasks, their responses demonstrate stronger language-based biases while the influence of world knowledge, such as causal relationships, was lessened. Because LLMs replicate language produced by humans, these results can help shed light on which aspects of language use are directly encoded in language and which require additional reasoning faculties beyond language processing.