Forensic linguists identify criminals by their writing style
Image: First fingerprints taken 1859/60 by William James Herschel in the public domain
Graphology, or handwriting analysis, has long been used to match suspects to the ransom notes that they definitely didn’t write, I swear Officer, it was my identical twin brother. But with the advent of email, SMS and cheap home printing, very little textual evidence is handwritten anymore. Enter forensic linguistics.
The way you write – the length of your sentences, your use of punctuation, or your intractable belief that ‘professional’ should have two Fs in it – creates a linguistic ‘fingerprint’ that can be used to identify you. Forensic linguists have been tasked with examining blackmail letters, death threats, potentially faked suicide notes and even historical items, such as the famous ‘Bixby letter’, supposedly penned by Abraham Lincoln, but a matter of fierce debate.
In 2009, 27-year-old Amanda Birks died in a house fire, from which her children and husband just barely escaped. She had sent multiple texts to friends and family on the night of the fire, including one telling her mother she was going to spend the evening ‘relaxing with candles’. Investigation revealed marital troubles and her husband Christopher Birks became a suspect. Forensic linguists compared Amanda’s final texts with ones sent over the previous weeks, as well as to Christopher’s own texts. They determined that he had used her phone to help make the crime look like an accident. Faced with this evidence, he plead guilty and was sentenced to life in prison.
In another case, it’s the police who were under the magnifying glass. Paul Malone, a suspect in a series of armed robberies, was interviewed about his movements on the day of the fourth robbery while a police officer took notes, which Malone then signed his name to. He claimed that, after he signed the document, police filled in blank lines with incriminating items.
The Centre for Forensic Linguistics was asked to compare the disputed sections of the statement to determine if they were written at a later date. CFL noted striking differences – in the undisputed sections, the grammar was fairly loose, often omitting subjects and definite articles, whereas the disputed sections had more formal, structured sentences. They reported that this was consistent with the undisputed sections being quick notes taken during an interview, and the alleged additions being written less hurriedly and more thoughtfully. They also mentioned changed line indentation and handwriting size. (In this case, to CFL’s disappointment, the presiding judge dismissed their analysis and Malone’s appeal was overturned.)
CFL’s work has also allowed them to identify a person’s native language through their writings in English, and to successfully impersonate specific individuals, including children subject to online grooming. Police assume the identity of the child, using analysis from forensic linguists (Do they say ‘haha’ or ‘lol’? What emoticons do they use, and how often?) to make it convincing.
Linguistic analysis of Donald Trump's twitter account suggests that it is shared by two writers: Donald himself, who writes inflammatory stuff from an Android device, and a second, more calm, moderate writer who uses an iPhone. The Unabomber, famously, was convicted partially on the basis of distinctive syntactical tics in his manifesto.
Some of the most exciting breakthroughs have come through combining forensic linguistics with the unprecedented power of modern data-mining and machine learning. For her PhD thesis, Carnegie Mellon University student Emily Kennedy created an app to combat sex trafficking. Masquerading as legal sex workers, traffickers post thousands of ads on publicly accessible websites every day. They regularly switch up their names, phone numbers and locations, making them difficult to track.
But their spelling, grammar and punctuation habits remain the same. Kennedy’s app, Traffic Jam, sifts through the astronomical amount of data, looking for ads with same linguistic ‘fingerprint’ and grouping them together. This allows detectives to build up a picture of the trafficker’s operations, track their movements and eventually gather enough evidence to make an arrest. Traffic Jam is used by around 50 organisations throughout the US and has rescued over 120 victims of human trafficking.
