How Do You Define "Troll"?
Claire Hardaker, a linguist at Lancaster University in the U.K., recently published an overview of "trolling," i.e., "behavior of being deliberately antagonistic or offensive via computer-mediated communication (CMC), typically for amusement's sake." In the wake of the media attention her work has received, Hardaker considers the varying definitions people have for the word "troll."
On the blog for Lancaster University's Centre for Corpus Approaches to Social Science (CASS), Hardaker explains:
Over the past fortnight, various broadsheets and media outlets picked up the story of my recent article, '"Uh…..not to be nitpicky,,,,,but…the past tense of drag is dragged, not drug.": An overview of trolling strategies' (2013), which came out in the Journal of Language Aggression and Conflict. Of the many thousands of comments collectively posted on each of those articles, one particularly interesting point that came through (out of many) was the general sense that there exists a single, fixed, canonical definition of the term troll which I ought to be using and had somehow missed.
So what is the definition of the term, troll? In my thesis, I spent a rather lengthy 18,127 words trying to answer precisely this question, and very early on I realised that trying to discover, or, if one didn't exist, create a clean, robust, working definition that everyone would agree with would be close to impossible. There are at least three major problems, which for simplicity's sake are best referred to as history, agreement, and change.
Hardaker finds that history isn't much of a guide, since there are many different historical meanings of the word, and it's unclear which of them gave rise to the online "troll." Agreement comes up wanting, too, since dictionary definitions (both professional and user-generated) differ widely. And the change of the online use of "troll" since its inception only adds to the term's semantic slipperiness. Undeterred, Hardaker writes:
How do we resolve these history, agreement, and language change problems? There is more than one solution to this problem, but mine was to use a lot of data — 80 million words of online data. From this, I extracted all the instances of trolling that I could find, and identified the consistent, major themes that users referred to as being a part of trolling. Once I had those themes, I amalgamated from them a generic, umbrella definition of troll. Crucially, this isn't typically based on how (alleged) trolls themselves use the term. In the data, loosely speaking, for every 1,000 examples of individuals talking about trolling (e.g. discussing whether they are being trolled, whether A is a troll, whether a certain behaviour counts as trolling) there was roughly only one example of a (supposed) troll — accused or actual — discussing their own intention to, or success at trolling. In other words, my definition of trolling is actually heavily — in my view too heavily — built on interpretation, rather than intention. However that may be, the final definition that I came to, as it currently stands is that trolling is...
the deliberate (perceived) use of impoliteness/aggression, deception and/or manipulation in CMC to create a context conducive to triggering or antagonising conflict, typically for amusement's sake.
Once I had established this deliberately broad definition, I set about identifying some of the consistent sub-themes, which I then published in my 2013 paper. As it turns out, this is a task that could take up many lifetimes. Given the size of the internet, even an 80 million word corpus is a drop in the ocean of all published online text, so the few trolling methods that I found are very unlikely to be all the types that exist. However, because of the proliferation of subtypes that began to emerge, I also quickly decided against giving each specific troll-like behaviour its own particular name (e.g. snert, griefer, etc.) since this seemed much more likely to increase confusion that reduce it.
In a nutshell, comprehensively defining even one meaning of one word is a never-ending task, and however well-researched a given definition at the conclusion, it is still highly unlikely that every user will perfectly agree with it. Added to that, the newer the word or meaning, the more abstract or "trendy" it is, and the more quixotic the field it belongs to, the wider the differences between individual definitions are ultimately likely to be.
Read Hardaker's complete blog post here, and read her commentary in The Guardian on the flavors of Internet trolls here. And check out our contributor Neal Whitman's thoughts about the word "troll" here.