The rhetorical language of numbers: the politics of criminal statistics
In the past, different theorists have expunged the 'scientificity' of official (criminal) statistics, and have questioned their accuracy and completeness, the processes through which they are produced, and the structural relations of their production. Despite these criticisms, there has been no moratorium on their use. In fact, so long as the 'obvious' caveats and qualifications are ventured, it is pretty much business as usual, and statistical and numerical information remains the primary resource for the manufacture of plausible, convincing and authoritative statements about the 'crime problem'. Indeed, as Maguire (1994) has argued, a salient feature of almost all modern discourse about 'crime' is the emphasis placed upon terms associated with quantification and measurement: 'volume', 'extent', 'growth', prevalence', 'incidence', 'trends', and so on. Moreover, in political and media debates, trends in aggregate crime figures are often put forward as evidence of failures or successes in criminal justice policy, or are treated as a barometer of declining standards in parenting, education and morality, and/or the persistence of sexist, racist and homophobic attitudes.
Quantitative accounts, whether produced from annual criminal statistics, the British Crime Survey, local surveys and/or academic research, are explicitly, and invariably implicitly, given more credence than 'subjective', 'anecdotal', 'speculative', qualitative versions of events. Quantification, it seems, conveys a sense of 'transparency' and objectivity', and calculation tends to be regarded as an impersonal, mechanical routine devoid of human emotion, desire and bias. Yet, given the centrality of the language of quantification to political claims-making about 'crime and punishment' and 'law and order', its status as a non-rhetorical medium must be seriously called into question. This paper, then, is concerned not only with the discursive ways in which statistical and quantitative accounts are used and mobilised in political debate on the 'crime problem', but also how numerical formulations are rhetorically organised to produce specific argumentative effects. The analysis centres on the recent publication of annual criminal statistics (January, 2000) which, on the face of it, evidence a rise in crime and the short-term failure of New Labour's crime reductive strategy. Informed by Foucault's archaeological and geneaological 'theories of discourse', and taking advantage of recent developments in rhetorical analysis, the paper examines broadsheet news coverage of this 'discursive event' noting how different politico-ideological points of view are marshalled, conveyed and legitimated through a 'rhetoric of quantification'.
Archaeology, genealogy and rhetoric
The concept of 'discourse', and methods of 'discourse analysis' owe a considerable debt to Michel Foucault whose insights revolutionized theoretical ideas on the relationship between language, knowledge and power, and socio-political change (Foucault, 1970; 1972; 1973; 1977; 1978). In this article, I make use of Foucault's 'archaeological' and 'genealogical' work in which he puts forward, on the one hand, a theoretical framework for understanding 'discursive formations' as 'rules' for constituting areas of knowledge (Foucault, 1972); and on the other, a series of theoretical claims which emphasise the mutually constitutive relationship between discourse and power, and the discursive nature of social transformation. In Truth and Power (1980: 131), Foucault neatly summarises the theoretical shift in emphasis, and the relationship between archaeology and genealogy; he states:
"Truth" is to be understood as a system of ordered procedures for the production, regulation, distribution, circulation, and operation of statements. "Truth" is linked in a circular relation with systems of power which produce and sustain it, and to effects of power which it induces and which extends it. A "regime" of truth.
There are three major strands to his archaeological thesis. Firstly, Foucault argues for a constitutive view of 'discourse'; that is, he suggests that 'discourse' (speech and writing), constitutes the social, including 'objects' (such as the 'crime problem', 'fear of crime', 'community safety'), and 'subjects' (such as 'offenders', 'victims', 'police officers', 'judges', 'lawyers') and marks out the field of their possible transformation - such that 'offenders', for instance, may be re-described as 'persistent', 'sex' or 'petty'. Secondly, he asserts the primacy of 'intra-discursivity' and 'inter-discursivity'. In the former, he suggests that discourse is organised and structured in such a way that its various modes of description, deduction, definition, assertion and application characterize its 'architecture'. 'Inter-discursivity', on the other hand, recognises that a discourse co-exists with others, and draws upon them in complex ways. So, for example, criminological discourse may claim territorial rights to a particular field of knowledge and style of inquiry, but it borrows from and deploys the insights of, for example, medical, psychiatric, economic, sociological, biological and legal discourses1. Thirdly, Foucault argues that 'discourses' open up a field of possibilities for the creation and development of theories and propositions, not all of which may be realised.
Foucault's genealogy, however, is not concerned with the 'architecture' of discourse so much as its function in the social processes and political practices of modern society. Power, he argues, is implicit to our everyday activities; it is exercised in the ways in which 'we' gather knowledge of others. Thus, through police patrols and interviews, identity parades, court trials, Neighbourhood Watch Schemes, the production of official statistics and press reports, knowledges (discourses/'truths') about a 'criminal population' are generated. At the same time, 'new' techniques of power are developed on the basis of these knowledge(s) - such as 'CCTV surveillance', 'target-hardening', 'zero-tolerance policing', 'fast-response', 'community policing' and 'vigilantism' - suggesting the interrelationship and reciprocal nature of power/knowledge. Discursive practice is, therefore, profoundly political; power struggle occurs both in and over discourse and is the spring of social change and transformation.
Of course, Foucault has his critics, and comprehensive critiques of his work are available elsewhere - see, for example, Dreyfus and Rabinow, 1982; Hoy, 1986; Dews, 1987; Fraser, 1989; McNay, 1994. In this context, Fairclough's (1994) contention that Foucault is 'too abstract' and 'confusing' in the way he describes 'discursive practice' would seem to be the most pertinent; that is, Fairclough complains that whatever the merits of Foucault's insights, they are not (apparently) empiricised in 'real' texts or 'actual' events and people. Fairclough may have a point, but neither Foucault's archaeology nor genealogy was intended as a form of empirically-oriented textual (or linguistic) analysis (McNay, 1994: 69). There is, of course, a long and rich tradition of textually-oriented discourse analysis, both structuralist and post-structuralist, which does concern itself with the empirical nature of discourse, and the linguistic resources and processes which discursive practice involves2. The branch of this scholarship which is of most relevance here is that devoted to rhetorical analysis; that is, a concern to examine discourse both for the way it is organized to make argumentative cases, and for the way it is designed to undermine alternative versions of events - see, for example, Billig, 1985, 1987, 1989, 1991. If, as Potter and Wetherell (1994: 49) argue, the distinction between Foucault's abstract theory of discourse and empirically-driven textual theories of discourse is an artificial one which is no more than a convenient heuristic, then there is no good reason to make either/or theoretical choices. It makes good sense, then, in this context, to locate a textually-specific rhetorical analysis within Foucault's archaeological and genealogical framework, thus maximising the analytical yield. As Foucault has stressed:
... there is no reason for describing (an) autonomous layer of discourse except to the extent that one can relate it to other layers, practices, institutions, social and political relations, (1989: 18).
In the remainder of this paper, I put this analytical schema to work by examining the broadsheet coverage of the recent publication of annual criminal statistics. I outline, briefly, the main features of the news story and the political context in which it emerged, and go on to firstly, identify the variety of numerical formulations about the 'crime problem' which were used to illustrate and emphasise different elements of the 'story'. Secondly, I consider how these formulations were drawn upon to make argumentative cases about, for example, localised police effectiveness, the growth of internet crime and geographical variation in police performance. Finally, I consider how these rhetorical positions are mobilised and deployed in a wider discourse which continuously transforms and legitimates varied conceptualisations of the 'crime problem'.
'Rise in offences sets alarm bells ringing'
Under the banner 'tough on crime, tough on the causes of crime' the New Labour government has carved out its place in the political landscape of 'law and order'. Addressing a conference of academics, policy-makers and criminal justice professionals at the University of Hull in July, 1998, Joyce Quin (Minister of State at the Home Office) proclaimed that New Labour had a 'new approach, with new ideas, new themes, new emphases and new actions' which constituted a 'proactive and positive approach to removing the threat of crime and disorder' (Quin, 1998: 185). On the back of a wide-reaching consultation process, research evidence, a review of best practice, and the investment of £400 million of government monies, a series of inter-related measures have been introduced which establish a comprehensive legislative and policy framework within which an effective crime reduction strategy can be put into practice, nurtured and realised (Home Office, 1999). The flagship legislation, Crime and Disorder Act, 1998 (CDA), is a key element in this process. The publication of official statistics on recorded crime in January, 20003 is significant for a number of reasons. First, the figures cover the twelve month period since the implementation of the CDA (October 1998 - September 1999), and thus provide an early indicator of its impact on crime reduction. Secondly, for the first time, criminal statistics include a breakdown of offences based on individual 'basic command units' (BCUs, i.e. police divisions) within force areas, allowing focused geographical comparisons to be made. Consequently, the government announcement (on 18th January) that recorded offences marked the first rise in crime in six years, was the catalyst for an intense broadsheet debate within which the efficacy of New Labour's approach to criminal justice matters was hotly contested. As The Times (19/01/00) suggested, and as the sub-title (above) confirms:
The Government's credibility on law and order was seriously undermined yesterday with publication of figures showing the first rise in recorded crime for six years.
This opening statement incites a politics of the 'rise in crime'. It signals a 'discursive event' in which different points of view are exchanged, and (re)conceptualisations of the 'crime problem' are made possible. What is significant, in this context, is how this politics involves a struggle over 'numbers' as well as 'words'. In the next section, I identify the range of numerical expressions and different modes of quantification which characterise the discourse and lay the groundwork for the development of specific argumentative points. The newspapers examined were The Times, The Guardian and The Daily Telegraph, specifically the editions published on 19 January.
A lexicon of numerical expressions
While by no means exhaustive of the range of possible formulations, seven different kinds of numerical statements appeared in these newspapers'' coverage of the 'rise in crime'. By far the most common mode of expression was that of proportion - '2 per cent increase in sexual offences'; 'two-thirds of all robberies'; '24 of the 43 police force areas'. Proportional statements were buttressed by expressions of frequency such as '350 attacks per 1,000 population'; 'six times more likely'. A third, and commonly used numerical form was that of precision - for example, '2.2 per cent'; '5.2 million'; '663,800 offences'; 'population of 4211'; '36.9 offences per 1,000 homes'. Fourthly, some newspaper texts employed a quantification mode which expressed magnitude - for instance, 'the rise of 115,000 recorded crimes brings the total to 5.2 million' - although this form was much less marked than proportion, frequency and precision. Fifthly, a few statements took the form of equations - for example, 'offences have risen by 29% or 70,000'; 'there were 406 offences of violence, or a rate of 2.9 per 1,000'. Sixthly, there were many instances of the use of approximation which included statements such as 'nearly 6%'; 'more than half of the total'; 'fewer than 10%'; 'roughly 16 per cent'. In addition to these discrete modes of quantification, there were combinational possibilities which facilitate comparison and contrast across different forms of numerical expression. So, for example, some statements combined precision with equation, magnitude and approximation - 'crime grew by 8.7 per cent or 80,285 offences to a total of over 1 million'; others combined approximation with magnitude, equation and proportion - 'offences top the million mark with an 80,000, or 9% rise in offences'. Moreover, The Times (19/01/00) displayed a series of league-tables and maps adding an elaborate visual and spatial dimension to the numerical expressions which appeared in the main body of the text. Indeed, as Potter et al (1991: 343) suggest, visual displays function as 'parallel commentaries' which reinforce textual numerical expressions.
These different quantifiers enable readers to draw up similarities and contrasts across the data, as well as seek out patterns and trends in 'rising crime'. However, what was clear from my reading of the press coverage was the additional inclusion of non-numerical expressions which gave qualitative meaning to numerical statements. Thus, the different manifestations of 'rising crime' across offence types and police force areas/BCUs were variously evaluated as 'slight', 'sharp', 'marked', 'significant', 'relatively small', 'very low', 'huge', 'only', 'typically', 'predominantly', 'rarely', 'usually', 'few', 'mainly' and 'just'. Similarly, press reports utilised non-numerical quantifiers to position numerical statements on a spectrum of extreme cases to show, for example, the 'biggest', 'largest', 'worst', 'smallest' and 'least' instances of a 'rise in crime'. The rich lexicon of non-numerical quantifiers give sense to the numerical forms, and increases the range of possibilities for inserting 'number' into argument. The task now is to explore how this kind of discursive 'architecture' is put to work to produce specific argumentative (rhetorical) effects.
The rhetoric of quantification
One of the most important features of any newspaper 'story' is its 'angle' or 'argument'. In the following three extracts, the statistical evidence of 'rising crime' is exploited to put forward different arguments about (and different constructions of) the crime problem'.
Offences increased by 2.2 per cent in the year from October 1998 to September 1999. There were 114,000 more offences taking the total to just over 5.2 million. Fraud and forgery, sexual offences, thefts from the person and violence against the person all rose.... The figures make it clear that increases in the Metropolitan and West Midlands force areas were largely responsible for pushing up the national total. In the Met, crime grew by 8.7 per cent or 80,285 offences to a total of over 1 million. In the West Midlands there was an increase of 16 per cent or 48,755 offences (The Daily Telegraph, 19/01/00).
The 2% overall increase in the crime figures recorded by the police, for 12 months to September 1999, masks a 29% rise in fraud and forgery, which includes credit card crime on the internet, and a 19% rise in street robberies in Birmingham, leeds, London and Manchester. While violent crime rose by 6.3%, crimes against property, including burglary, rose by just 2%. Drug offences, mainly possession, fell by 9% (The Guardian, 19/01/00).
Yesterday's figures showed wide regional variations, with recorded crime rising by 16 per cent in the West Midlands, 12 percent in Bedfordshire and 8 per cent in the Metropolitan Police area. However, it fell in 24 of the 43 police areas of England and Wales, including by 10 per cent in Lancashire, by 6 per cent in Durham and 7 per cent in Cheshire (The Times, 19/01/00).
For The Daily Telegraph the 'crime problem' centres on the effectiveness of two specific police force areas - the Metropolitan and the West Midlands; The Guardian, on the other hand, emphasises the growth of 'net-crime' and 'street robberies' in four metropolitan force areas; while The Times draws attention to the geographical variability of police performance. These arguments involve different sorts of quantification and utilise a range of numerical (and non-numerical) expressions to accomplish a variety of rhetorical goals.
Consider, the first of these extracts: 'offences increased by 2.2 per cent' is a precise proportion which conveys a sense of 'smallness' about the rise in recorded offences. Juxtaposed with a total of 'just over 5.2 million', has the effect of shifting the focus onto an altogether different order of 'number'. The contrast between the 'smallness' of '2.2' and the 'millions talk' of the totality constructs a minimal picture of 'rising crime'. Moreover, the approximation combined with the precision of 'just over 5.2 million' implies a level of magnitude beyond which counting is rendered pointless. The insertion of the equation '114,000 offences' is, then, crucial to bridging these extremes of number; expressed as 'hundred(s) of thousands', readers are actively encouraged to see the rise in crime as 'substantial' rather than 'small'. The text goes on to list those offences which have risen over the previous year. The absence of any reference to property crime (such as burglary, thefts of and from vehicles, vehicle interference, criminal damage, handling and theft from shops), suggests that the list is designed to be read as emblematic rather than complete. At the same time, the list makes no reference to the relative rate at which these offences have risen. The absence of quantifiers here suggests that the overall rise in crime, rather than the detail of it, is of primary analytical concern. Indeed, since the article claims that 'the Metropolitan and West Midlands police force areas were largely responsible for pushing up the national total', this is the only 'fact' which needs to be demonstrated quantitatively. Thus, a combinational numerical expression is introduced which establishes that 'Met crime' has increased by 8.7 per cent (ie. four times the overall increase of 2.2 per cent); that this equates with 80,285 offences (ie. over two-thirds of the 'absolute' rise in recorded offences); and that the 'Met' total is so high that it can only be approximated as 'over 1 million'.
Similarly, the total number of offences recorded for the West Midlands force area is omitted (i.e. 353,035) since it cannot be implicated in the 'millions talk' of the 'Met', or seen as significant within the total number of recorded offences ('just over 5.2 million'). Rather, the emphasis is restricted to the West Midlands' proportionate rise in recorded offences, (16 per cent), and the absolute number of offences that this rise involves (48,755). Through this rhetorical route, the 'fact' that the West Midlands has experienced almost eight times the overall percentage rise in crime, and over two-fifths of the 'absolute' rise in recorded offences, may be sufficient to convince readers that the force, with the 'Met', is 'largely responsible for pushing up the national total'. It is immaterial that these two force areas appear to have experienced an absolute rise in crime (i.e. 129,040 recorded offences) which actually outstrips the total rise in crime (114,000 recorded offences). By suppressing all quantifiers of decreasing crime levels across different force areas (as well as offence categories), the rhetorical effect is to exaggerate the rise in crime and lay the primary responsibility for it onto two specific force areas.
By comparison, the range of numerical expressions deployed in the second extract is far less elaborate. The commentary uses proportion to identify the overall rise in crime (i.e. 2%), and as with the first extract, this implies the 'smallness' of it. But whereas the Daily Telegraph uses a combinational quantifier to render this 'smallness' as 'substantial', this article does so through a different numerical route. The preferred tactic here is to deconstruct the '2%' to show a more complex picture of events. Moreover, since the figure 'masks' more than it reveals, readers may be encouraged to adopt a more critical, 'realist' stance towards statistical evidence. Thus, readers are alerted to the proportionate rise (and fall) in crime across a range of offence categories which are listed in a descending order of magnitude. The rhetorical effect of this technique is to invent for the reader a 'hierarchy of concern' about particular offence types, with 'net crime' positioned at its apex. At the same time, and in contrast to the first extract, this listing purports to be 'complete' rather than emblematic. For example, care is taken to expand offence categorisations so that readers are clear that 'fraud and forgery' includes credit card crime; that 'crimes against property' includes burglary. Similarly, the listing clarifies and specifies observed changes in offence levels; thus, readers are in no doubt that 'street robberies' have (only) risen in Birmingham, Leeds, London and Manchester, and that the fall in the recorded number of 'drug offences' is 'mainly' due to the decrease in possession offences. The 'completeness' of the coverage is thus accomplished through both a vertical movement of inclusion (i.e. the creation of a 'hierarchy of concern') and a horizontal one (i.e. clarification and specification).
The third extract appeared some way into the article from which it was taken. However, apart from a statement that 'the 2.2 per cent increase in offences caused alarm in Downing Street', the extract is the first occasion where numerical expression is used to convey a specific argument - that the rise in recorded offences is regionally variable. In the same way that the second extract constructs a 'hierarchy of concern' about different offence categorisations, the article uses proportion to construct an informal 'league table' of police performance. Moreover, the additional numerical statement that recorded offences 'fell in 24 of the 43 police force areas of England and Wales' subtly classifies police force areas into two distinct groupings - those that are successful and those which are failing. This quantifying manoeuvre enhances the rhetorical effect of the commentary in as far as success/failure provides a further dimension of interest not present in the contrastive quantification of the rising/falling levels of recorded offences.
Rhetoric and the discursive construction of the 'crime problem'
These commentaries and the numerical statements which they deploy set up what Potter et al (1991: 339) refer to as a 'preformulation' - that is, an argument is put forward which, at the same time, anticipates (and encourages) the formulation of specific counter-argument(s). 'Preformulation' has a specific rhetorical effect; that is, it can make subsequent responses seem predictable, hackneyed or even jargonistic. So, for example, in response to the Daily Telegraph's general commentary on the 'rise in crime', the Home Secretary was reported to have said that 'the rise was 'slight' and that offences had dropped overall since the Government came to power in 1997'; while The Guardian's concern with the growth in 'net crime' elicited an 'expert' response that although some of the 29% increase was due to increased reporting and changes in the counting rules, cyber-crime was increasing'. At the same time, The Times' focus on regional variation prompted the Chair of the Police Federation to comment that 'the solution is real investment to expand policing and strong leadership to reassure and drive the service with determination during a period of low morale and confusion'. There is, of course, an important process of selection involved here - who responds, where the response appears in the text, how much of the response is reported, is a crucial aspect of journalistic and editorial work. More importantly, the arguments presented in these three broadsheets offer different conceptualisations of the 'crime problem' which themselves are selectively drawn from a wider field of discursive possibilities.
Consider, for example, how in the previous three months, the 'crime problem' has been variously centred on 'the shortage of police recruits' (The Guardian, 01/10/99); 'the rising incidence of murder and rape' (Daily Telegraph, 11/10/99); 'the leniency of the courts' (Daily Telegraph, 12/10/99); 'the fear of victimisation' (The Guardian, 13/10/99); 'the demographic glut of young men of offending age' (The Times, 13/10/99); 'the under-recording of violent crime' (Daily Telegraph, 13/10/99); 'an increase in racist incidents' (The Guardian, 28/10/99); 'an increase in armed robbery and aggravated burglary' (The Guardian, 28/11/99); 'a property crime wave' (The Times, 30/11/99); 'a samurai sword attack' (The Guardian, 30/11/99); and 'a misuse of stop and search' (The Guardian, 18/12/99). In other words, the 'crime problem' and all those 'events' and 'people' which constitute it, is a shifting, fluid and contingent phenomenon. It is continuously transformed in and through discourse, and through that transformation a 'new' politics of crime control is promoted and desired. In this instance, the 'event' of 'rising crime' has induced a politics of intervention premised on improving the effectiveness of two specific police force areas, stemming the tide of cyber-crime, and reducing geographical variation in police performance. Indeed, subsequent articles developed these themes, transforming (innocuous) numerical formulations into political proposals for change. Of course, these are not new insights - such accounts lie at the heart of constructionist approaches to 'social problems'. What I hope to have added to these perspectives is an appreciation that the language of quantification, not solely the processes and practices of measurement and enumeration, provides an important resource for constructing specific points of view. In other words, the language of quantification - conventionally viewed as 'neutral', objective' and 'hygienic' - is a rhetorical medium through which criminal (and other social) statistics are made meaningful, and different political standpoints are expressed.
In the course of this article I have attempted to delineate some of the practices through which numerical and non-numerical forms of quantification are marshalled and organised to form specific arguments. The material I have used is neither representative nor comprehensive, but I would argue that it is indicative of press reportage of statistical trends. The analysis would have benefited from an in-depth examination of all the textual data (not solely the three extracts), the use of a broader discourse analytic technique (such as semiotics) and, indeed, an exploration of editorial and journalistic processes. For all that, I have demonstrated the point that quantification is as rhetorical a medium as more qualitative forms of expression, and as such, is crucial to critical argument and political involvement. At the same time, 'the point' is also a rhetorical construction designed to survive potential criticism and alternative views of the language of quantification.
I am indebted to Professor Diane Richardson, Colin Clark and John Kennedy for their valuable comments on an earlier draft of this article.
Bakhtin, M. (1981), The Dialogic Imagination, Austin: University of Texas Press.
Bakhtin, M. (1986), Speech, Genres and Other Late Essays, Edited by C. Emerson and M. Holquist, translated by V. W. McGee, Austin: University of Texas Press.
Barthes, R. (1973), Mythologies London: Paladin.
Billig, M. (1985), 'Prejudice, categorization and particularization: from a perceptual to a rhetorical approach', European Journal of Social Psychology, 15: 79-103.
Billig, M. (1987), Arguing and Thinking: A Rhetorical Approach to Social Psychology, Cambridge: Cambridge University Press.
Billig, M. (1989), 'The argumentative nature of holding strong views: a case study', European Journal of Social Psychology, 19: 203-223.
Billig, M. (1991), Ideologies and Beliefs, London: Sage.
Clark, C. and Pinch, T. (1988), 'Micro-sociology and micro-economics: selling by social control' in N. Fielding (ed) Actions and Structure, London: Sage.
Derrida, J. (1978), Writing and Difference, London: Routledge and Kegan Paul.
Dews, P. (1987), Logics of Disintegration: Post-structuralist Thought and the Claims of Critical Theory, London: Verso.
Dreyfus, H. L. and Rabinow, P. (1982), Michel Foucault: Beyond Structuralism and Hermeneutics, London: Harvester Wheatsheaf.
Eco, U. (1990), The Limits of Interpretation, Bloomington, Indiana: University of Indiana Press.
Fairclough, N. (1994), Discourse and Social Change, Cambridge: Polity Press.
Foucault, M. (1970), The Order of Things: An Archaeology of the Human Sciences, London: Tavistock.
Foucault, M. (1972), The Archaeology of Knowledge, Translated by A. M. Sheridan Smith, London: Routledge.
Foucault, M. (1973), The Birth of the Clinic: An Archaeology of Medical Perception, Translated by A. M. Sheridan Smith, London: Tavistock.
Foucault, M. (1977), Discipline and Punish: The Birth of the Prison, Harmondsworth: Penguin.
Foucault, M. (1978), The History of Sexuality: Volume I, Harmondsworth: Penguin.
Foucault, M. (1980), 'Truth and Power' in C. Gordon (ed) Power/Knowledge: Selected Interviews and Other Writings 1972-1977, Brighton: Harvester.
Foucault, M. (1989), Foucault Live: Interviews 1966-84, New York: Semiotext(e).
Fraser, N. (1989), Unruly Practices: Power, Discourse and Gender in Contemporary Social Theory, Cambridge: Polity Press.
Home Office (1999), The Government's Crime Reduction Strategy, London: HMSO.
Hoy, D. C. (ed) (1986), Foucault: A Critical Reader, New York: Basil Blackwell.
Kristeva, J. (1981), Desire in Language, Oxford: Basil Blackwell.
Kristeva, J. (1986), 'Word, dialogue and novel' in T. Moi (ed) The Kristeva Reader, Oxford: Basil Blackwell.
Maguire, M. (1994), 'Crime statistics, patterns and trends: changing perceptions and their implications' in M. Maguire, R. Morgan and R. Reiner (eds) The Oxford Handbook of Criminology, Oxford: Clarendon Press.
McCloskey, D. (1985), The Rhetoric of Economics, Brighton: Wheatsheaf.
McNay, L. (1994), Foucault: A Critical Introduction, Cambridge: Polity Press.
Potter, J. and Wetherell, M. (1994), 'Analyzing discourse' in A. Bryman and R. G. Burgess (eds) Analyzing Qualitative Data, London: Routledge.
Potter, J., Wetherell, M. and Chitty, A. (1991), 'Quantification rhetoric - cancer on television', Discourse and Society, 2, 3: 333-365.
Povey, D. and Cotton, J. (2000), 'Recorded Crime Statistics: England and Wales, October 1998 to September 1999', Home Office Statistical Bulletin, Issue 1/00, London: Home Office.
Propp, V. (1968), The Morphology of the Folk Tale, Austin: University of Texas Press.
Quin, J. (1998), 'The Labour Government's new approach to criminal justice', Policy Studies, 19, 3/4: 185-189.
Said, E. (1978), Orientalism: Western Conceptions of the Orient, London: Penguin.
Saussure, F. de (1974), Course in General Linguistics, London: Fontana.
Simons, H. (1990), 'A perspective on the rhetoric of the scientific research report and an illustration: "Drug Pushing" in a Medical Journal Article', mimeo, Temple University.