A Computational Analysis of Constitutional Polarization

A new paper, “A Computational Analysis of Constitutional Polarization” by Columbia Law School’s David Pozen and Eric Talley found that the U.S. Constitution has come to enable, or even exacerbate, partisan strife.

The researchers recently examined the evolution of constitutional rhetoric on the floor of Congress from 1873 to 2016. First, they identified the hundreds of thousands of remarks that referred to the Constitution. Then they trained a machine-learning classifier to predict – based solely on the content of the remarks – whether Republicans or Democrats were speaking. If the algorithm finds this task hard to do, it implies that the parties are apt to talk in similar or overlapping ways. By contrast, if the algorithm performs this task with a high degree of accuracy, it implies that the parties are largely talking past each other. They found that, in the past four decades, it has become increasingly easy for an algorithm to predict whether any given constitutional remark was made by a Republican or a Democrat.

Scholars have imagined the Constitution as supplying a kind of cultural glue that disciplines disagreement and tempers partisan passions – but it is likely the opposite. Rather than transcending preexisting political divides, arguments framed in constitutional terms tend to mirror or magnify those divides.

PREDICTING PARTY AFFILIATION BY TEXTUAL CONTENT

Here is a piece published by the researchers in The Atlantic on June 2, 2019 while the newly released work was still a working paper: “Republicans and Democrats Are Describing Two Different Constitutions.” The article is the first to use computational methods to investigate the ideological and partisan structure of constitutional discourse outside the courts. The researchers applied a range of machine-learning and text-analysis techniques to a newly available data set comprising all remarks made on the U.S. House and Senate floors from 1873 to 2016, as well as a collection of more recent newspaper editorials. Among other findings include:

  • Constitutional discourse has grown increasingly polarized over the past four decades; 
  • Polarization has grown faster in constitutional discourse than in non-constitutional discourse; 
  • Conservative-leaning speakers have driven this trend; 
  • Members of Congress whose political party does not control the presidency or their own chamber are significantly more likely to invoke the Constitution in some, but not all, contexts; and 
  • Contemporary conservative legislators have developed an especially coherent constitutional vocabulary, with which they have come to “own” not only terms associated with the document’s original meaning but also terms associated with textual provisions such as the First Amendment.

Sign up for the free insideAI News newsletter.