A corpus-based deconstructive strategy for critically engaging with arguments

My broad concern is with texts which aim to persuade an audience of a particular point of view on a particular topic – persuasion texts. Political speeches and newspaper editorials are examples of this text type. I put forward a strategy for critically engaging with such texts. Its focus is a persuasion text’s cohesion – how it is held together through its vocabulary and grammar. The strategy explores whether or not the cohesion of a persuasion text is unstable, if it deconstructs. Since a persuasion text’s credibility is dependent, amongst other things, on effective cohesion, showing where a persuasion text’s cohesion deconstructs diminishes its credibility. I call this critical reading strategy Electronic Deconstruction: ‘Electronic’ reflects the fact that the strategy draws on corpus linguistic method; ‘Deconstruction’ refers to the deconstructive approach of this strategy. An advantage of Electronic Deconstruction is that it can still facilitate critical engagement with an argument where it is difficult to reconstruct all its premises. This is because its evaluative focus is a persuasion text’s cohesive structure rather than its logical structure. To demonstrate Electronic Deconstruction, the article employs a case-study, a text written in 2008 by the late political journalist, Christopher Hitchens, which justified his continuing support for the 2003 Iraq intervention by US-led coalition forces. After highlighting a number of frustrations in identification of its arguments, and thus for critical assessment of its logical soundness, I show how Electronic Deconstruction as an alternative critical engagement circumvents incomplete reconstruction by doing the following: revealing that the text’s cohesive structure is unstable via an electronic deconstructive analysis which draws on a two-billion word corpus, the Oxford English Corpus. The article employs two corpus linguistic software tools: ‘Sketchengine’ (http://www.sketchengine.co.uk/) and ‘WMatrix’ (http://ucrel.lancs.ac.uk/wmatrix/).