Marcos Mesa Sam Wordley/Shutterstock
Michael Hunter, a student at the University of North Carolina, was found dead in his bed. His roommate, Joseph Mannino, called 911. An autopsy indicated that Hunter had died from a mixture of drugs, including an overdose of Lidocaine, an anesthetic. But when the pathologist found an injection mark on his arm, with no sign of a needle near him, the police opened an investigation.
Hunter had roomed with Mannino, a med student, and Garry Walston, a landscape architect. Mannino had access to Lidocaine, and admitted he’d given it to Hunter for migraine headaches. He added that Hunter had recently discovered he was HIV-positive, so perhaps he’d used the Lidocaine to end his life. Mannino even gave police a printout from a computer disc that contained apparent suicide notes to Hunter’s friends and relatives.
However, Walston told police that Mannino and Hunter had been angry at each other. In fact, Mannino had been in the process of moving out when Hunter was found dead.
The suicide notes were key: Had Hunter really written these documents? Without actual handwriting to work with, who could make a determination from just content analysis?
As it happens, a growing group of professionals can offer just this service.
I recently attended a workshop on forensic linguistics, sponsored by the Association for Linguistic Evidence (TALE). The organizer was Dr. Carole Chaski, president of ALIAS Technology and founder of the Institute of Linguistic Evidence, Inc. (ILE), a nonprofit agency that supports research on the validity and reliability of language-based author identification.
Linguistic analysis involves making a detailed analysis of the content of a questioned document to compare it with what its potential author writes and/or reads. The basic premise is that no two people use language in exactly the same way. The pattern of unique differences in each person's use of language and the repetition of those traits throughout his or her writing provide the internal evidence that links (or fails to link) a person to the questioned writing.
When analyzing a sample, such as a suicide note, forensic linguists examine the subject’s other writings or—with unknown subjects—search text databases that could contain similar language habits. The sample's language can help establish the writer's age, gender, ethnicity, level of education, professional training, and ideology.
Key items are vocabulary, spelling, grammar, syntax, and punctuation habits. Other kinds of textual evidence might include borrowed or influential source material, document formatting, and the physical document itself.
Chaski’s method is based on generative grammar, a touchstone in linguistics for the past five decades which originated in the work of Noam Chomsky. This approach analyzes the rules of grammar in a particular language to form an algorithm with which to predict which combinations of words will form grammatical sentences.
Unlike some “stylistics” analysts who use subjective judgment, Chaski relies on standard statistical tests. She is known for her approach to what's called “the keyboard dilemma,” i.e., the difficulty with identifying authorship of documents written with a keyboard to which multiple users have access. To address this issue, she applies cross-validated syntactic analysis.
That is, a questioned document will show distinct syntactic patterns—the unconscious way in which a person automatically combines nouns, verbs, adverbs and prepositions to create phrases—and each is counted statistically.
With a program of her design, Chaski has tested hundreds of linguistic variables with several statistical procedures, achieving up to 95 percent accuracy for author attribution from blind samples.
For the Hunter case, she agreed to analyze the supposed suicide notes.
She was provided numerous samples of both Hunter’s and Mannino’s writing. Then she applied her program.
None of Hunter’s known exemplars (written documents) exhibited certain key syntactical items that were evident in the suicide notes, so Chaski concluded with a high degree of probability that he had not written them. However, there was reason to believe that Mannino might be the author. (An HIV test also showed that Hunter was not actually HIV positive, increasing suspicion against Mannino.)
Mannino was arrested, and admitted at his trial that he’d written the notes. He was found guilty of involuntary manslaughter.
As a forensic scientist, Chaski is working hard to reduce the errors of subjective analysis (junk science) and improve the reliability and validity of ILE’s approach. With the increased use of computers for creating documents of all types, this is a service that law enforcement cannot do without. The more it's grounded in scientific standards, the better.