At its core, the study of the law -- broadly conceived -- is concerned with texts: statutes, constitutions, judicial opinions, and briefs. These texts contain the rules, judgments, and arguments that make up and shape the law. Ironically, quantitative empirical scholarship on the law has largely ignored these documents, and with good reason. To date, few tools have been available that would allow systematic analysis of a large number of legal texts in a cost-effective manner. In the fields of computational linguistics and computer science, however, there have been important advances in automated content analysis. In these disciplines, tools have been developed that permit the analysis of patterns and interrelationships in a text. Unfortunately, political scientists -- and scholars of judicial politics in particular -- are largely unaware of these methodological tools. The central purpose of the proposed conference is to overcome this gap between the disciplines. The conference will bring together political scientists, legal scholars, computational linguists, and computer scientists to explore how modern automated content analysis tools might be used to analyze complex legal documents.
The purpose of this grant was to organize a conference to bring together political scientists, legal scholars, and computational linguistics to explore the potential for computerized content analysis of legal texts, specifically judicial opinions. At its core, the study of the law – broadly conceived – is concerned with texts: Statutes, constitutions, judicial opinions, and briefs (and even transcripts of oral arguments). These texts contain the rules, judgments, and arguments that make up and shape the law. Ironically, scholars in the social sciences who use quantitative approaches to study how law is created, and how it shapes social and political outcomes, have largely ignored these documents, and with good reason. To date, few tools have been available that would allow systematic analysis of a large number of legal texts in a cost-effective manner. One critical consequence of this is that scholars have been constrained in their ability to develop more nuanced understandings of the interactions between institutions that create law (e.g., the courts, Congress, state governments, federal agencies) because doing so requires more fine-tuned measures of the content of legal texts (e.g., the legal rules developed in judicial decisions.) Suppose, for example, that a scholar would like to investigate the extent to which lower courts "follow" the Supreme Court in their opinions -- an important question if we want to understand the extent to which law is "equal" across the United States. To do so, it would be useful to have a measure that estimates how closely lower courts adopt the reasoning and legal standards set forth by the Supreme Court. Similarly, if a scholar is interested in the development of case law in a particular area over time, it would be valuable to be able to place opinions relative to one another to trace the evolution of the law (for example, how has the treatment of "privacy" changed over the last thirty years, and in response to what forces?). Over the last two decades, developments in computing technology have led to the emergence of techniques in computer science and computational linguistics that are devoted to the automated analysis of large collections of texts. These develpoments are extremely promising for addressing the kinds of questions in legal scholarship posed above, but the two academic communities do not usually interact. The conference was designed to overcome these disciplinary divides and to foster cooperation across scholarly communities in order to make progress in addressing the questions posed above. As a result of the exchanges at the conference, a number of cross-disciplinary research efforts are now under way to make concrete progress on automated content analysis of legal texts.
