Applying Automatic Content Analysis Techniques to Legal Texts: How to Make Things With Constitutional Words
By enabling scholars to analyze large collections of legal texts quickly and without massive funding support, computer-based content analysis techniques promise to revolutionize comparative constitutional scholarship, at a time when an ever-greater number of legal documents are accessible on the internet at just a few mouseclicks. The CONREASON Project pioneers the application of supervised and unsupervised scaling methods, originally developed for the analysis of party manifestos, to the study of judicial opinions and other forms of legal writing.
Human language, not least in the legal field, is eminently complex.
But while we have not yet reached the point where computers can replace humans, automated content analysis methods have the potential to greatly amplify the scope, quality and depth of legal research. The CONREASON Project explores how automated content analysis techniques can be used to chart and reconstruct the evolution of entire lines of case law (e.g. the position of the German Federal Constitutional Court on European Integration over the past four decades) or to identify the dimension of disagreement in law review articles.
Human language, not least in the legal field, is eminently complex.
But while we have not yet reached the point where computers can replace humans, automated content analysis methods have the potential to greatly amplify the scope, quality and depth of legal research. The CONREASON Project explores how automated content analysis techniques can be used to chart and reconstruct the evolution of entire lines of case law (e.g. the position of the German Federal Constitutional Court on European Integration over the past four decades) or to identify the dimension of disagreement in law review articles.
What Is Automated Content Analysis?
Automated Content Analysis (ACA) lies at the intersection of several discplines: linguistics, computer science, statistics and the social sciences (including law). It works by making simplifying assumptions to exploit the rhetorical dimension of political and legal discourse. Judges, like politicians, choose their words so as to put the position they want to defend in the best possible light. As they change position, therefore, their choice of words tends to change as well. Based on this basic insight, ACA methods seek to make inferences about the actors' positions from the words they use.
ACA encompasses a whole family of computer-based techniques designed for the analysis of large text collections. In the CONREASON Project, we primarily focus on Latent Trait Scaling approaches, which come in the supervised (Wordscores) and unsupervised (Wordfish) variant.
ACA encompasses a whole family of computer-based techniques designed for the analysis of large text collections. In the CONREASON Project, we primarily focus on Latent Trait Scaling approaches, which come in the supervised (Wordscores) and unsupervised (Wordfish) variant.