Recently, a number of critical discourse analysts have been using the concept of 'manipulation' as an analytic tool to study political language. The attraction is clear:the concept is inherently critical , for by calling a particular use of language 'manipulation', the analyst is likely to be criticising that use. If the concept is to be analytically as well as critically, useful, then analysts must identify what is wrong with the language that they are identifying as 'manipulation'. This is where the difficulties lie. Different analysts are defining 'manipulation' in different ways, thereby highlighting different, even opposing, characteristics.In an important paper, Teun van Dijk (2006) has argued that manipulation occurs when the powerful abuse their power in order to influence those without power, in ways that are against the interests of the powerless but that are in the interests of the powerful (see also van Dijk, 2008). However, other analysts have denied that the formal possession of power need be involved in manipulation (Hasrati & Mohammadzadeh, 2012). Some claim that discursive 'manipulation' involves deceit because manipulators typically conceal what they are doing from their victims in order that the victims do not think ill of them (Blass, 2005;Harré, 1985;Pinto, 2004).Other analysts, however, suggest that manipulation involves using fallacious arguments, regardless of any intentional deceit (Danler, 2005;Kienpointner, 2005;Rigotti, 2005). David Buss, a cognitive social psychologist, explicitly defined 'manipulation' to exclude the possibility that malevolent or dishonest intention need be involved (Buss et al, 1987).On the other hand, some analysts claim that manipulation involves using specific fallacious arguments, regardless of any intentional deceit (Danler, 2005; 4 Kienpointner, 2005;Rigotti, 2005). Regarding political oratory, some critical analysts claim the key element of 'manipulation' is that speakers exert cognitive control over their audiences. According to Chilton, manipulation depends on the propagator controlling or dominating 'an intended receiver's mind' (Chilton, 2005, p. 17). In cognitive accounts, analysts suggest that receivers are encouraged, and even constrained, to process messages superficially so that they end up with misleading conceptions. Maillat and Oswald (2011) highlight the importance of concealment in this process because 'the success of manipulation lies in its covertness' (pp. 70-1; see also de Saussure, 2005).One might think that there is an impasse because any critical analyst, who wishes to use 'manipulation', is faced with different conceptions. Does manipulation involve conscious deceit, the abuse of power, erroneous arguments, covertness, superficial cognitive processing and so on? The present study aims to provide a way out of this impasse. Rather than adopting a top-down strategy of proposing a definition before proceeding to analysing examples, we offer a bottom-up strategy of advancing towards, rather than starting from, a definition. We will exa...