Features Chosen for Analysis
The three sections above represent this preliminary engagement with the data as a set of questions suggested by the language of each of the texts. The process modelled here was repeated in the case of all 20 texts in each data set. The process now became interactive and recursive. The long- list of questions generated was followed by preliminary analysis, some of which proved fruitful, and some less so. Some of the features noted turned out to be relatively rare. Others did not appear to generate a productive analysis. For example, a review of sentences that pivoted on the conjunction “as” (for example, in the headline “BP profits soar as oil giant accelerates spill clean-up” (The Evening Standard [London], 27.4.2010)) showed that the construction was quite common, but otherwise uninteresting. Some of the features which were identified as significant were well-identified and researched, for example, the grammatical analysis of modality. Some appeared to be less standard, for example, my particular understanding of listing and categorisation in this data, which required additional identification and description work.
Through this process of identification and preliminary analysis, I drew up a list of nine features for particular investigation for this set of texts, and these were:
- 1. Naming of events.
- 2. Naming of people.
- 3. Categorisation.
- 4. Genre.
- 5. Intertextuality.
- 6. Modality.
- 7. Metonym.
- 8. Metaphor.
- 9. Discourses.
The number nine is not significant and the list is not exhaustive. Three data sets offering a total of 60 texts provide a rich source of linguistic data with potential for analysis against a wide range of dimensions. Five, ten or twenty features might have been equally appropriate for investigation, depending on the nature of the data set and the purpose of the research
Fig. 7.1 N ine linguistic features for analysis
work. Nevertheless, the final nine features were the ones that not only seemed to be pertinent to an understanding of how the BP events were represented linguistically in the media but also showed movement and development across the data.
The term “features” is a convenient shorthand; in fact, the linguistic areas suggested as important are disparate including grammatical systems such as modality; rhetorical usages such as metaphor; pragmatic and semantic processes such as naming and categorisation; and supra-textual concerns such as genre, discourses and intertextuality. It is at this point in the analysis process that the chosen features could be reconnected with Barthes’ framework of semiotic levels. Those features that operated at the level of a word or word group (such as the naming of events) were considered to be a single “sign”, in the sense of a building block for meaning. Those features that represented systems or codes were considered to be at the level of “code”; for example, the modality system, genre and inter- textuality. At the level of “myth” were two rhetorical tropes—metonymy and metaphor—that serve to add additional connotative meaning to denotative representation. Finally, an analysis of discourses sought to uncover the ideological motivation of some of the language choices at the other semiotic levels. Figure 7.1 shows how the selected features for the BP data relate to the framework of Barthes’ semiotic levels first introduced in Chap. 3-
Of these nine feature types, some would probably be common to most linguistic studies of this kind—for example, an analysis of social actors, genre, intertextuality and discourses is likely to be relevant to most data. Other features such as categorisation or modality might be more relevant to some data than to others.
The Immersion Stage in Summary
This second immersion stage served two main purposes:
- 1. Gaining familiarity with the data. Reading and rereading the data gave an invaluable sense of the texts which, while not pre-empting the findings from the depth analysis, suggested connections and hypotheses relating across texts and data sets as well as within individual text items.
- 2. The output of Stage 2 was a defined set of discursive features for study, which were suggested by the data themselves, rather than predetermined.