the-BA : Grant

By Wendy Barnaby

David Hodgson was convicted of murdering Jenny Nicholl in February, 2008. Her body has never been found. Hodgson was convicted partly because, in text messages he sent on her phone after she disappeared, he spelled "myself" as "meself". In her own text messages, Nicholl had spelled the word "myself".

Delivering his Joseph Lister Award Lecture later today, Dr Tim Grant, Deputy Director of the Centre for forensic Linguistics at Aston University, will describe how advances in forensic linguistic research make it possible to identify who has sent certain text messages.

Dr Grant and his colleagues have developed a method, based on one used in marine ecology, to determine how closely the text of messages sent by one person resemble those sent by another.

In the Nicholl/Hodgson case, for example, Grant would analyse the messages known to have been sent by Nicholl before her disappearance, for about a dozen characteristics of their linguistic style. This would include whether "I am" was written "im" or "I am", and whether there was a space between the terms in phrases like "go2shop".

Messages known to have been sent by Hodgson would be similarly analysed. A third analysis would be done on the messages believed to have been sent after some harm had occurred to Nicholls. This third group would then be compared with the other two, to see whether their style was closer to those authored by Nicholls or by Hodgson.

Grant's method refines one used in marine ecology to compare similarities in ecosystems in different places. It quantifies people's style of text writing, and assigns a numeric measure of stylistic difference between any two texts.

"This encourages a move from expert opinion-based evidence to more methodologically rigorous and empirically tested techniques," he said.

Later this month, the Council for the Registration of Forensic Practitioners is opening a linguistic subregister in its science and engineering panel, which indicates confidence in the method.

So far, court evidence has relied on older methods of assessing the authors of texts. The new method has not yet been used in court, but Dr Grant expects it will be in three to four years.

The new method can help discriminate between texts sent by men and by women. Women's texts tend to be interpersonal, whereas men's are generally sent to make arrangements.

"This evidence would not be admissible in court," said Dr Grant, "but we could help the police by saying that we think that a certain text is more likely to have been sent by a man than a woman."

The method also shows how people who text each other frequently grow more similar in their texting style.

Dr Grant is sometimes called upon by companies who want to identify the author of malicious emails sent from anonymous accounts.

"In these kinds of communication it is relatively easy to be, or at least feel, anonymous," he said. "New technologies have created an anti-social phenomenon of mass anonymity, and the ability to identify the writer can only be beneficial for society."

Grant has put together a database of about 8000 texts from 900 authors, to establish base rate information for certain features in texting language. The public can contribute to his research by submitting their own text samples to http://www.forensiclinguistics.net/texting.
Read more in the Financial Times, the Press Association and online at ScienceDaily.com.