THE INTEGRATION OF BANKING AND TELECOMMUNICATIONS: THE NEED FOR REGULATORY REFORM

(Jeff_L) #1
376 JOURNAL OF LAW AND POLICY

techniques but also for data collection and research
methodology. Neither forensic stylistics nor stylometric
computing is grounded in linguistic theory. Instead, both
forensic stylistics and stylometric computing are grounded in
conceptions of language that are common in prescriptive
grammar and literary criticism or focused on naïve conceptions
of language as a list of words or a list of function words. So
considering the “linguistics” in forensic linguistics, of which
author identification is a primary task, forensic computational
linguistics employs standard linguistics, while forensic stylistics
and computer science neither use linguistics in analytical
techniques nor theoretical underpinnings.
Second is the role of research in the approaches. In order for
the Daubert factors to be met, litigation-independent validation
testing on forensically feasible “ground-truth” data must be
conducted. Forensic computational linguistics has met this
challenge directly through the use of forensically feasible
“ground-truth” datasets such as the Chaski Writer Sample
Database. Independent of any litigation, validation tests have
been conducted, as reported earlier in this paper. These tests
have been run on forensically feasible data—that is, documents
which are short, in several types of genre and register, and
without any correction to grammar, spelling, or prescriptive
conventions about writing. Further, the data are ground-truth
data, where the authorship of each document is known; there is
no possibility that someone else was using a screenname or
posting blogs under a pseudonym. Finally, the validation test
research has resulted in a known protocol for what is needed to
apply the forensic computational linguistic methods; the test
results empirically limit the amount of data required. It is hoped
that both forensic stylistics and stylometric computing will
conduct the kind of research that forensic computational
linguistics performs, so that reliable methods of forensic
authorship identification can be offered to our courts.

Free download pdf