Carole E. Chaski – Institute for Linguistic Evidence
Course time: Tuesday/Thursday 3:30-5:20 pm
2336 Mason Hall
Linguistics as a Forensic Science introduces students to the current state of the art in forensic linguistics. Students learn the legal standards that linguistic evidence must meet, how linguistic research has produced methods that meet these standards, as well as examples of methodological failure. Cases and rulings are discussed in the context of methodological issues for linguistics, and to demonstrate the seriousness of legal standards. Examined in detail are linguistic methods for author identification, text classification, intertextuality and linguistic profiling. Most forensic linguistic methods attempt to identify, individuate or classify texts, so automatically texts are seen as instances of either individual or group variation (i.e. the method must be able to categorize texts as belonging to different individuals, the method must be able to classify texts as belonging to a particular type of text, the method must be able to identify texts as coming from a person with a certain level of education or dialect, and so forth).
The paradigm which students learn in this course is one in which (1) universal principles provide methodological grounding for the analysis of variation, (2) texts are analyzed for the instantiation of syntactic and semantic properties, (3) the instantiations are quantified, (4) the quantifications are subjected to statistical analysis, (5) the statistical analysis is subjected to validation testing for error rates. This paradigm –known as computational forensic linguistics– poses several challenges to linguistics as a science, such as, the choice of levels and units for linguistic analysis of forensic texts for specific tasks, the predictability of linguistic behavior, tools for analysis of variable linguistic behavior, and the model of language which is both circumscribed or determined by universal principles but at the same time instantiated in group and individual behaviors. Thus, computational forensic linguistics provides a proving ground for how universal principles ground analysis and method so that individual and group variability can be accurately captured and then used for prediction –the core of scientific endeavors.
Current forensic linguistics methods exemplify the tension between universality and variability. The ways in which different methods embrace universality or variability have either enabled or prevented linguistic methods from reaching error rates low enough for legal use. Admissible methods that have successfully met the scientific rigor required for legal evidence combine analysis based on universal principles of linguistic structure with statistical analysis of linguistic variability. On the other hand, methods which have focused on variability to the exclusion of universal principles have failed methodologically to produce repeatable results or low error rates, and have thus not met legal standards and are generally ruled as inadmissible. The computational forensic linguistic paradigm embraces variability as the core of most forensic linguistic problems, with universal structural principles as the primary analytical approach for solving these problems. Only this synergistic approach — a structural-behaviorist approach— actually works to produce feasible forensic linguistic methods that are theoretically grounded, replicable and reliable.
Students in this course should have already taken an introductory linguistics course. Students in may also find the Institute courses on R and Python to be good courses to take at the same time but they are not required.