4.4 Data analysis
4.4.3 Process of Analysis
The process of analysis consisted of several iterative ‘cycles of analysis’ (Jäger & Maier, 2009), with each cycle encompassing and connecting the various level and foci of
analysis described above. This section describes the two main cycles of analysis: within- text analysis and across-text analysis. Most authors, outlining analysis processes in various approaches to CDA stress the iterative, interwoven and non-linear nature of the analytic process, and avoid identifying specific steps (Jäger & Maier, 2009). Thus, while described separately, these cycles were repeated several times, at various points in time, sometimes performed parallel to another one, overlapped and fed into each other.
The first main cycles of analysis, within-text analysis, always began with an open reading of the text worked with. Reading an information brochure for the first time, I freely noted anything that came to mind on a paper copy of the text. To not constrain potential
interpretation early in the analytical process, these initial notes attended more to “the possibility that something interesting was going on, rather than [to] an indication of what it might be” (Wood & Kroger, 2000, p. 92, italics added). Further, as notes were done directly on the text, they also indicated where - with regard to the overall structure – ‘something interesting’ was going on; this was helpful later, when the organization of meaning across a text as a whole was analyzed. These first notes were very broad and contained various initial impressions, observations, and reactions, referring to various textual levels (i.e., form, function, content, and overall structure). For example, initially reading a text, I underlined and commented on words or sentences that drew my attention (e.g., “Keeping on the go… Who ‘keeps’ whom here? And from what?”), marked text sequences that struck me for various reasons (e.g., “This sequence sounds familiar, where have I read this before?”), noted questions that arose while reading (e.g., “Is there
mandatory reporting for physicians in Quebec?”), or recorded my reactions that specific text sequences evoked in me (e.g., “Reading this, I really don’t want to age”). Open reading also included a careful ‘reading’ of any other semiotic elements included in the text, such as photographs, symbols, check-lists, or self-tests. For instance, in ‘reading’ a photograph for the first time, I described any detail that I saw (such as “good teeth”, “white hair”, “sweater, leisurely wrapped around neck”, etc.). This strategy of detailed description facilitated me to step outside of taken for granted ‘ways of seeing’ (G. Rose, 2007) and to pay attention to elements of the photograph that would have gone unnoticed at first glance. Any other observations or reactions referring to other semiotic elements were noted (such as, “This cartoon on the title shows seven! alternative transportation options in the background: taking the bus, tram, subway, cab, biking, being driven by a younger person, and walking”).
Following the initial, open reading, an analysis sheet was set up for the information brochure I was working with. After compiling its bibliographical and contextual information, I conducted a reading of brochure guided by the analysis sheet. I now critically systematically explored the text using a theory-informed ‘lens’. For instance,
focused reading was guided by questions, such as “Who is defining the problem and who is addressed as having power to ‘fix’ it?”, “What technologies and practices of the self are the targeted audience called upon to participate in?”, and “How are aging
subjectivities constituted as ideal/non-ideal?”. Such theory-informed analysis enabled another level of critically reading a text. Especially when working with the very first texts, I was recurrently surprised by utterances, truth claims and social relations within a text that I had not noticed in my initial reading. The value of the analysis sheet to support systematic, theory-informed critical analysis is reflected in the following excerpt out of my reflexive notes (Notes, Oct. 26, 2010):
I am absolutely struck what one can discover in a text if one has a theoretical lens… I am just analyzing the first brochure, using the analysis sheet. While the sheet with all its many sections feels like a daunting task to fill out (...) I am struck (and excited!) by what it ‘makes’ (or better): helps me to see. There are so many things in the texts that I haven’t even been noticed before - even though, I felt that I already had done a pretty thorough first job when I did the initial reading (...) But now, when I go through the text again (...) having a particular question from the sheet in my mind, all these other things suddenly come to the forefront, to my attention, to the surface. It’s almost like a totally new text (...) Why didn’t I see these things before? Am I so used to read a text in a particular way? In which way? (...) I guess, today’s experience underlines that filling out the sheets is more than documentation of the analysis process; it seems to be another analytical strategy, drawing my attention to particular features and claims in the text, that I don’t ‘see’ and question anymore - as I am so embedded in these discourses too. The sheet helps reading beyond a text’s ‘surface’. It’s significant to have a theoretical framework in one’s work… Boy, today, I feel that I could analyze this one text only for at least a week… (Notes, Oct. 26, 2010)
Parallel to open readings and theory-informed readings, each text was also read using linguistic concepts and tools that have been suggested for textual analysis in CDA (J. E. Richardson, 2007). Reading a text ‘linguistically’ helped me to understand how meaning was created by form and function of particular word choices and sentence constructions, as well as how meaning was organized across a text as a whole. For instance, to draw out implicit meanings, I analyzed a text for particular word choices and sentence structures that have been described in linguistic analysis for particular functions. Richardson illustrates how certain word choices such “as the change of state verbs (stop, begin, continue) or implicative verbs (manage, forget) invoke presupposed meaning in their very use: ‘stop’ presupposes a movement or an action; ‘forget’ presupposes a great deal,
including an attempt to remember” (J. E. Richardson, 2007, p. 63). Textual analysis drew out a text’s underlying assumptions, implicit content, and absences, which were then linked to previous readings and al foci. For instance, in my initial readings of texts, I had marked a particular type of verb choices which were consistently repeated within and across texts (i.e., ‘remain’, ‘stay’, ‘continue’, ‘maintain’, ‘preserve’, and ‘keep’). Texts repeatedly called upon aging subjects to ‘remain’, ‘stay’, and ‘continue’ to be safe drivers, as well as to ‘maintain’, ‘preserve’, and ‘keep’ their body’s driving fitness. Applying linguistic tools, I investigated which presuppositions and meanings these particular verb choices shared (i.e., these verbs imply that something involuntarily changes to the worse, decreases or might get lost if nobody takes action to ‘keep’, ‘maintain’ or ‘preserve’ it), where in a text and in which contexts they were primarily used (i.e., “these verbs are dominantly used when safety or the aging body is brought up”, Notes, Nov. 3, 2010); and how actors and objects were linguistically related to each other in the process these verbs described. The findings of this analysis (e.g., “these verbs stress that one has to work actively
Besides critically reading singular texts in depth, I also repeatedly read texts ‘against’ each other. This type of reading formed the second key cycle of data analysis, that is, focused analysis across individual data pieces. Analyzing the texts as a whole, analysis in this cycle focused on similarities, variations, contrasts, repetitions, connections,
contradictions, and absences in content, form and function across texts. Often, an across- text analysis cycle began in following up, identifying and searching for commonalities and repetitions, and ended in specifically looking for differences, variations, and absences.
against an underlying process to ‘keep’ one’s status”, Notes, Dec.10) were then reviewed and related to findings of previous readings.
Across-text analysis happened at various times and levels within the overall process and combined emergent and planned ways and foci of analysis. That is, a particular focus of analysis was sometimes stimulated by the texts and sometimes systematically initiated. For instance, when I worked with a new information brochure, I began to notice similar word choices, parallel content, or a familiar line of argument as in already read
Once such formations emerged, they were systematically examined by reading texts parallel to each other and examining the whole body of texts for their existence, variations and absences.
Early on in the process I, for instance, noticed that many texts included numerical representations, such as measures of declining body function or accident statistics. In across-texts analysis, I investigated this observation further, by first identifying and marking all numerical representations in different texts and then analyzing what these shared, in which ways they differed, and what they were. Analysis then followed up on different observations and different paths. For instance, marking all numerical
representations across texts lead to the observation that numbers typically seemed to be ‘clustered’ in particular sections of the overall text. These sections were then investigated in-depth, but also with regard to the overall text. That is, it was investigated across texts, if other similarities with regard to other text sections could be identified.
As well, reading more and more texts, I began to notice that all texts seem to consistently provide ‘facts’ about aging and driving. Reading across texts, I marked all ‘facts’ about either aging or driving within each text, reviewed the filled out notes about aging and driving on all analysis sheets, and then investigated these texts sequences in-depth and parallel to each other. In a next step, categories were generated, that grouped similar content across texts in order to understand how aging and driving were constructed. As analysis proceeded, these categories were combined with other levels of analysis. For instance, while engaging in different cycles of analysis, it was noticed, that the
constructed ‘facts’ about aging and driving were shared similarities with regard to their content, but also with regard to the ways the content was constructed, such as by
frequently including numerical representations. As well, some of the generated categories seemed to follow a particular order to build the overall argument. At this point, various cycles of analysis were combined and analyzed together. For instance, I investigated if numerical representations played a role in constructing aging and driving and if so, the discursive functions they had in constructing a particular content.
As analysis proceeded, more and more al cycles and foci were combined. Once first understandings of the overall rhetorical structure and dominant themes across texts were taking shape, an initial al summary was written. This summary was reviewed and
reflected upon within an advisory committee meeting and it was decided that the overall al processes were effective and should proceed in the pursued ways.
Writing up the study represented the final phase of data analysis. As already described, the writing process in itself presents a valuable al strategy. For instance, selecting text sequences that could best illustrate the interpretation offered and finding words to frame this interpretation lead often to reviewing previous analytical notes, finding new aspects, or gaining a deeper understanding of a particular way of construction.
Ending analysis
Within critical discourse analysis, data analysis proceeds “until the analyst is satisfied that the research questions have been addressed and that a reasonable reading of the discourse (...) can be offered” (Wood & Kroger, 2000, p. 97). Analysis in CDA can be best described as “provisional until a fairly late stage (though it is ultimately always provisional)” (Wood & Kroger, 2000, p. 23)20
20
As its underlying epistemological perspective rejects the idea that only one analysis and interpretation is possible, analysis in CDA is ultimately always ‘provisional’.
. Thus, completeness of the overall analysis is achieved, when the analyst, based on her interaction with the data, ongoing reflexivity and adopted quality criteria, decides that data and analysis are theoretically and
methodologically sufficient to offer and justify an insightful interpretation and argument (Phillips & Hardy, 2002). In addition to ongoing reflexivity regarding this study’s analysis process and the decision to end analysis, the thesis advisory committee was also engaged in on-going discussions, as a group and as individuals. Part of the purpose of this more collective form of reflexivity was to receive input regarding the achievement of theoretical ‘sufficiency’ and the strength of the interpretation made.