344 JOURNAL OF LAW AND POLICY
Differences in academic training make the paradigm of
experimental validation testing for forensic authorship
identification more or less difficult to accept. Training in literary
criticism does not focus on empirical methods, while pure
computer science can bypass courses in experimental design.
However, in most branches of linguistics, empirical work is
mandatory. Psycholinguists design and run experiments testing
the theoretical constructs posited by linguistic theory (usually
from a generative theory), focusing on the cognition and
memory required to produce and process human language. The
validation testing described earlier is second nature to someone
trained in psycholinguistics (including child language acquisition,
psychology of literacy, and second language acquisition).
Even if the forensic linguist relates to the small community
of sociolinguists, the methods that the forensic linguist develops
should be recognizable as sociolinguistics. Historically,
sociolinguistics introduced a quantitative approach to midcentury
American linguistics and relied heavily on empirical data
collection, phonetic measurements, and experimental research
designs.^29 Therefore, when a forensic linguist asserts that his
academic training is in sociolinguistics, but his method is neither
quantitative, nor tested on ground-truth data, nor validated by
experiments, the disconnect between the forensic activity and the
academic world is startling to linguists, if invisible to attorneys
or judges.
III. THE FORENSIC COMPUTATIONAL LINGUISTICS APPROACH TO
AUTHOR IDENTIFICATION
Work in forensic computational linguistics began in the mid-
1990s, with funding from the National Institute of Justice.^30 By
INNOVATIVE APPLICATIONS ARTIFICIAL INTELLIGENCE (IAAI), 2009, at 60;
Chaski, Empirical Evaluations, supra note 16; Koppel & Schler, supra note
3.
(^29) Labov is considered the originator of sociolinguistics; his work is
characterized by quantitative, statistical analysis of naturally collected or
elicited linguistic behavior. See generally WILLIAM LABOV, THE SOCIAL
STRATIFICATION OF ENGLISH IN NEW YORK CITY (2d ed. 2006).
(^30) In 1995, I received a grant to validate linguistic methods for