Deconstructing ‘responsibility’ in the food system
3.3 Document analysis
3.3.3 Applying a discourse analysis to the documents
The aim of analysing key documents pertaining to responsibility in aspects of the food system was to look beyond the apparent meaning, beyond the
‘dictionary definition’ of responsibility, and to explore how such texts construct and influence the reproduction of the concept of responsibility, and how they deploy it discursively. Early versions of the document analysis were, however, more descriptive of the documents and followed an unstructured, thematic
31 http://www.pepsico.co.uk/purpose/health/health-report-2010 accessed 09/05/11
32 http://uk.kantar.com/consumer/shoppers/2015/march-kantar-worldpanel-uk-grocery-share/
accessed 25/11/15
33 http://cr2010.tescoplc.com/~/media/Files/T/Tesco-Corporate-Responsibility-Report-2009/Tesco_CSR_2010.pdf accessed 09/05/11
course. It was only in later versions, that what was emerging was, after all, a Foucauldian discourse analysis and therefore further literature was consulted to hone this approach (e.g. Hall 2001a; Waitt 2010).
Within the Foucauldian discourse analysis ‘steps’ listed in Figure 3.2 (which were also used to examine the data generated in focus groups and interviews), I asked particular questions of the documents. These were based on guidelines specifically drawn up for analysing texts: who made, commissioned and owns the text, what are the relationships between the maker, owner and subject of the text and why was the text written? (Waitt 2010, p.227). Each document was interrogated using these questions, in combination with the steps in Figure 3.2, to consider “its inconsistencies, internal workings and small strategies of meaning” (Tonkiss 2004a). For example, the authorship, target audience and aims of a document affect the agenda within the text, whether that is explicit or concealed within more subtle discursive structures. Looking for such answers helped to discern “how language figures within social relations of power and domination; how language works ideologically” (Fairclough 2001, p.230).
This thesis touches on the role of ideology in government policy (see sections 4.4.4, 4.5.6, 7.4, 8.2); it should therefore be noted that although Foucauldian discourse analysis falls within the tradition of critical discourse analysis, it differs in at least one notable way from others that take a Marxist slant on ideology (Fairclough 2001). Foucault explicitly rejected the concept of ‘ideology’ in the Marxist sense of the dominance of the ideas of the ruling class, distorted to serve its power and interests; he opposed such reductionism and suggestion that the ruling classes created an ideology that overruled ‘truth’ (Hall 2001a).
Nevertheless, despite the focus on Foucault’s ideas, this thesis also borrows from the broader CDA tradition, using ‘ideology’ in the broader, more literal sense of a shared set of ideas, beliefs and values, as put forward by Bakhtin, who saw language use (or discourse) as ideological (Bakhtin 1981; Fairclough 2001).
The government and industry documents’ stated aims were deconstructed to seek any such underlying ideology and “domination” of the representation of responsibility. The texts were examined repeatedly, but in three key stages: first stage coding for description and organisation, second stage coding for analytic themes. The third phase involved further analysis such as described in steps 4-8 of Figure 3.2, namely, looking particularly for discursive strategies, efforts to create effects of ‘truth’, “rupture and resilience”, and silences in the discourse.
Additionally, the documents were scrutinised for the discursive construction of the different subject positions and of their supposed roles and responsibilities.
This was to account for the idea that “the human subject is decentred from being the agent of social development to being a product of social relations”
(Filmer et al. 2004, p.43). This itself gave insight into the powerful production of the subjectivities of the different groups of actors, for example, using the word
‘individuals’ sets up a contrasting position to using ‘the public’. The roles assigned to the different subject positions also provided opportunities for looking at the expectations of each of them, and how the architects of the food system positioned themselves at the diet-health intersection.
The analysis was an iterative process of identifying broad themes and then subthemes throughout the documents. It deliberately did not entirely “suspend pre-existing categories” (Waitt 2010, p.220); rather, in addition to seeking new themes, specific ones were explored based on the literature review and the goal of examining the discourse of responsibility. These included: responsibility itself, complexity, food choice, information provision, regulation and voluntary
agreements. The analysis explored the dynamic, relational nature of language, to further understand how responsibility is produced and communicated, and with what aim, in ‘official’ publications.
During repeated re-reading of the texts rhetorical devices were sought that advanced particular views, courses of action or ideologies. For example, the
‘complexity’ narrative around the diet-health problem was sometimes used
discursively rather than descriptively, to deflect responsibility away from the authors of the documents. It has been said that “the efficacy of discourse often resides in the assumptions it makes about what is true, real or natural” (G. Rose 2001, p.158) so analysis aimed to disrupt these assumptions by questioning the discursive structures that normalise certain uses or assumed meanings of words. Discursive formations were sought that helped to create effects of
“truth”; one such was the discourse of ‘choice’ which was recurrent throughout the documents (and the generated data) and consistently presented
responsibility for healthy eating as a matter of unmitigated choice for the general public. ‘Silences’ in the data that served as discursive strategies were also observed: for this it is necessary “to read against the grain of the text”
(Tonkiss 2004a) rather than reading “along” with it. One such silence was the influence that the food industry has over government policy through lobbying (Miller & Harkins 2010), which is a notoriously secretive business34.
3.3.4 Reflexivity
It is important to be aware of one’s own ‘agenda’, the discourses in mind, when reading the texts, which could – even inadvertently – colour what is and is not
‘seen’. This is particularly so, given the nature of the documents, presumably written with the aim of creating a particular impression: of an industry or government that is competent and committed to making a difference. It was therefore a fine balance between maintaining a critical, analytical eye, and a cynical one. If objectives of discourse analysis are “to disclose the created
‘naturalness’ of constructed categories, subjectivities, particularities,
accountability, and responsibility” (Waitt 2010, p.224) then it is also to do so without imposing one’s own preconceptions and constructions onto the data.
34 http://www.ft.com/cms/s/0/52c1bee0-9bfb-11e5-b45d-4812f209f861.html#axzz46fCVxMxu accessed 20/04/16
The analysis of the chosen documents was written up and results are presented in Chapter 4. It was done prior to data gathering and analysis from focus groups and interviews, which will be described in the following sections.