Resumen de (Dis)trusting the text: detecting identity deception in used car classified ads
This paper reflects on a piece of social/expLanatory research on textuaL deception conducted in the late 1990s before formal/utiLitarian approaches to textuaL deception invoLving large-scale statistical analysis and a-theoreticaL discovery procedures became the norm. Following the Birmingham tradition of data-driven discourse analysis and corpus linguistics, the study began with a real-world problem: how to unmask small dealers or rogue traders masquerading as private sellers in the leading UK used-car magazine, Auto Trader. In the original study, 100 Auto Trader classified advertisements were analysed and a typicaL order of structural elements was established. Prototypical linguistic realisation patterns were then identified for both tagged 'trader' adverts and untagged 'private' ones. The weaker claim of the study was that traders' linguistic realisations of the genre were typically different from those of the private seller. This led to the stronger claim that it might be possible to identify hidden traders through their linguistic choices. The current paper sets the original study in historical context, troubles some of its assumptions, and notes the paradox of trusting a text to reveal deception when the very nature of deception is that the text cannot be trusted.
