Written by Stephanie Warner; copy edited by Meagan Kus
Recap of EAC-BC’s branch meeting on February 18, 2015.
I’m a fan of British murder mysteries and police procedurals, so it was a thrill for me to attend a talk by Dr. Lorna Fadden, a real-life language detective. Dr. Fadden is an assistant professor at Simon Fraser University, where she teaches introductory linguistics courses. She also researches in the following areas:
- Canadian varieties of English and how language manifests online
- discourse analysis (specifically, police interviews and internet luring)
Her website and SFU profile give detailed information on her research.
What exactly is discourse analysis?
According to Wikipedia, discourse analysis is a general term for a number of approaches used to analyze written, vocal, or sign language use of any signified semiotic event. More simply, the Linguistic Society of America describes discourse analysis as analysis of language “beyond the sentence,” or “what speakers do in conversation.” Dr. Fadden puts it this way: “What people say, how they say it, when they say it, and where they say it.”
Were they telling the truth?
How can we tell if someone is telling the truth? In ancient China, there was an early lie detector test called the rice test. Dry rice was put in the accused’s mouth. If someone is stressed, their mouth goes dry, so the dry rice would stick to their mouth, and they couldn’t spit it out!
Today, a forensic linguist can use audio software to break down components of speech. Forensic linguistics was used in the Exxon Valdez trial. The captain had been accused of drunk driving a giant oil tanker into the Alaskan coast. There is no breathalyzer test on the bow of a big tanker. Instead, investigators examined recordings of the captain’s speech one hour before, immediately after, and one hour after the accident. The speech displayed characteristics highly consistent with known effects of alcohol on speech, such as the ability to recall words and utter them in the right sequence. The captain was acquitted, but this speech analysis method has since begun to find acceptance in several courts.
Analyzing a police interview
Speech analysis can also be used on police interviews. For example, Dr. Fadden listened to a recording of a police interview to find out if the person was making false allegations. She wanted to find answers to these questions: Is the speaker’s linguistic performance consistent with what we expect from a competent native speaker in this context? Was the speaker linguistically adept enough to understand what was going on in the interview? In the end, Dr. Fadden concluded that the accused was not telling the truth. To determine this, Dr. Fadden looked at things like the following:
- speech rate in responses
- fatigue effect of a long interview
- turn latency (how long between the end of the investigator’s questions and the accused’s responses)
- pause ratio (how much pausing compared to how much speaking)
- specificity (generic to specific article use)
- cohesion (how each turn flows from the next)
Dr. Fadden noted that her work is “just one piece of the evidence used.” In addition to psychologist assessments, a forensic linguist provides “another layer of assessment” that can be used to demonstrate how fit a person is to perform in an interrogation setting such as a police interview or trial.
Who wrote the letter?
From the beginning of the written word, there have been debates about authorship. Biblical scholars debated who wrote parts of the Bible, and ancient Greek playwrights accused each other of plagiarism. An ongoing question has been whether Shakespeare authored his (or her, as it has been alleged) own work.
Authorship analysis is the comparison of documents to rule in or rule out possible authors. It’s a bit of a contentious issue because its effectiveness and accuracy are not thoroughly understood, and there is no set method for carrying it out. Dr. Fadden said, “It’s like a polygraph. It’s probably right, but sometimes it’s wrong.”
Dr. Fadden shared two examples of authorship analysis. In the first example, Dr. Fadden looked at menacing letters written to a company’s board of directors to determine where someone might be situated in the company. She found that pronoun use, among other things, indicated that the letters were not in fact written from the level of employees they claimed to be. In the second example, a custody dispute, Dr. Fadden examined whether letters allegedly written by young children were actually written by them. She looked at language that was atypical of what a child that age would write.
When doing an authorship analysis, there are certain things you can look for:
- quantitative features (words per sentence, letters per word, and number of unique words)
- qualitative features (favoured syntactic constructions; lexical choices; preference for literal and non-literal language; use of adjectives, adverbs, parentheticals, and non-restrictive relative clauses; and discourse markers, among other things)
Who wrote the blog?
At the end of the evening, editors became detectives. We were given two blog posts and had to analyze the language in them to decide who wrote a “mystery” blog post. It wasn’t an easy task! We were split 50/50. Despite my confidence, I was wrong.
I won’t be solving any mysteries soon, but at least the evening sated my curiosity about how a language detective works. Too bad forensic linguistics wasn’t around in Agatha Christie’s day—it would be fascinating to find out how Miss Marple would have used it to solve her crimes. By the way, I asked Dr. Fadden how her work had affected the way she watches crime dramas. She laughed and said she stays away from them and watches lots of “goofball” comedies!
Find out more
International Association of Forensic Linguists
Linguistic Society of America
Stephanie Warner is a fan of British murder mysteries and police procedurals. Her favourite police detectives are Inspector Jack Frost and Inspector George Gently.
Meagan Kus is a freelance copy editor and proofreader with an 18-year background in arts administration.
Image by Bigstock