Expertise
I have recently begun work exploring whether developments in corpus-based computational linguistics are of any usefulness in legal study. Law may be characterized as a very large body of texts, too many texts (even when broken down into obvious subject matter domains) to be read directly. Is it possible to develop mechanisms for providing senses of these unread texts that do not rely too heavily (as present methods inevitably do) on jurisprudential filters? Statistics-based techniques for constructing models of grammar on the basis of regularities in large corpuses (Markov models and the like) may be useful in extracting other word patterns, perhaps of jurisprudential significance. The aim, in part, is something like that of on-going work in France (associated with the computer program Alceste) and work in computer-based literarty criticism (starting with the program TACT). My own work is at a very early stage, mostly in conjunction with a seminar -- Computer Analysis of Legal Texts -- about to be offered for the second time. I am working with opinions of the United States Supreme Court dating from 1937, and generating word frequency lists via TACT, in the hope of testing whether these lists suggest anything about the Court's work in this period (a well-known transition time in constitutional law, for example) which is in some sense new.