Between 2009 and 2011, a now convicted sexual offender in his early twenties was spending much of his time online persuading young girls and boys to produce and share with him sexual images and videos. To maximise his success, he would deceive and manipulate his victims by cycling through numerous different personas—a teenage boy, a young woman, a modelling agent, for example—seemingly trying to find the best fit for the person he was talking to.
Online anonymity is a significant hurdle in policing online sexual abuse, and cases like this are sadly common. As such, law enforcement agencies draw on expertise from a range of disciplines for support, including forensic linguistics. Given that this sort of online abuse occurs almost exclusively through language, linguists are in a unique position to assist police investigations by describing how language functions in various online criminal contexts as well as helping identify anonymous offenders through their language.
In the case mentioned above, the online abuse was enabled by multiple online identities. The man adopted 17 different personas. To understand how online abuse works, it’s important to consider two questions: First, what strategies did the offender use in the attempt to obtain images from victims? Second, did the 17 personas’ strategies vary, or were they linguistically consistent?
To explore these questions, it helps to understand how identity works. Rather than being some static, unchanging entity, identity is in some ways multiple, and at least in part, performed through language. Think, for example, about the various roles you might assume in a day- friend customer, boss, spouse, etc. And think about how your language choices might shift, subtly or dramatically, consciously or unconsciously, across each, enabling you to perform each of those different roles.
The first question is fairly straightforward. Across all 17 personas, the offender’s most common linguistic moves were associated with sexual and non-sexual rapport-building, maintaining/escalating sexual content, and assessing likelihood/extent of engagement. While these moves seemed to represent his overriding goals, he also made occasional use of more extreme and coercive moves which included overt persuasion and even extortion.
The second question got a bit more complicated. You might expect—as I did—a certain amount of variation between different types of persona. It seems reasonable to think, for example, that a mixed-race 15-year-old lesbian persona would differ somewhat from a white 19-year-old male persona. Not the case. Across all 17 personas, linguistically, the offender seemed to perform just two distinct roles. The first was sexual pursuer/aggressor, where he would determinedly pursue victims in an explicitly sexual manner. This typically involved lots of sexual content moves, occasional overt persuasion and extortion, and little rapport-building. His second role was something closer to friend/boyfriend, where he spent most time attempting to build sexual and non-sexual rapport and used almost no coercive strategies.
Most interesting about these opposing roles is that only one persona performed as the friend/boyfriend, where the other 16 personas were clearly sexual pursuer/aggressors, regardless of that identity’s age, gender, ethnicity, etc. So why did this one persona communicate so differently from the rest? News reports around the case, along with police records, verified much of the personal information he provided about himself when assuming this particular persona. In other words, this persona seemed less of a deceptive performance than many of the others, and more like the offender performing as himself, or at least the closest representation of what we might think of as the real person.
When analysing this offender’s linguistic moves, we had expected to find some small differences between personas of varying identity categories; males and females, or those of different sexual orientations, perhaps. What we did not expect was that we would be able to establish a single persona as being the most likely approximation of the real offender. These methods may prove useful in future investigations where offenders are known to be operating multiple online identities, at least in narrowing the pool of interesting online personas. Such analysis could also help where undercover officers are looking to emulate the linguistic strategies of people in this and other online criminal contexts.
It is imperative that we keep exploring new approaches to combating online crime, especially as offenders are taking ever-more sophisticated measures to mask their identities. Online anonymity remains a hurdle to policing online crime, but not a barrier. It’s clear that there’s a lot we can do to discover people’s real identities even when an offender’s language is the only available clue.
Featured image credit: “Black computer keyboard” by Florian Krumm. CC0 via Unsplash.