Forensic linguists try to #12;nd irregularities in language use, as irregularities provide
'signature' information that may withstand legal tests of certainty about,
for example, authorship identi#12;cation. Irregularities inhere in all levels of language
use, but here we focus on textual rather than spoken forms of language.
Yet, this research #12;eld is still nascent and a lot of work needs to be done though
existing methods give promising results. In this paper we will focus on an authorship
identi#12;cation method from the literary (Chaski, 2000) implemented in
a suite of computer programs developed by McCombe (2002) which was modi-
#12;ed by Medori (2004). Validation tests of the method are reported here, using
fabricated corpora as well as real texts. The results for the arti#12;cial and real
text data will stress that further research is necessary. Besides the bad results,
the program also gives very good results which, especially with regards to the
poor results, are very promising. The aim is to #12;nally reach an acceptance
of existing authorship identi#12;cation methods in courtroom, which would mean
that even written language can be named an evidence, a '#12;ngerprint' in court,
satisfying the Daubert Test1 for expert testimony.
|
