• No results found

Phenomenographic Bracketing and Reduction

A core approach in phenomenography is to revisit the data multiple times with multiple perspectives, choosing to either block out specific elements or drill down on others. This is called bracketing or reduction (Luft and Overgaard 2013). In total seven phases of bracketing and reduction were undertaken using the qualitative analysis tool NVivo (QSR International 2012). Each of the seven stages included up to five cycles of data analysis and are depicted below. The researcher coded in each phase, alternating between open, axial and selective coding, as data was identified, selected, categorised, deconstructed and then reconstructed. Open coding is a key grounded theory method that looks at developing concepts as codes and attaching this to the texts being analysed, in this case workshops, interviews, transcripts

and writing samples (Corbin and Strauss 2014). Axial coding was used to connect categories and concepts to each other within NVivo, another key method in grounded theory during qualitative data analysis. Selective coding was used to explore a core variable that related to all of the data.

Phase 1: Emergent, open coding of all workshops, interviews, video recall, transcripts, writing samples and researcher notes.

At this level, 480 references in thirteen data sources were explored across four themes that emerged from this phase: pedagogy, art form, dual state awareness and engagement. This involved three full cycles of analysis of the entire data set. The researcher experienced the workshops in person and took researcher notes. The researcher then watched each video in full and listened to each audio set in full as the composite video was created. Detailed notes were taken for each cycle in the researcher journal. The three workshop videos were then transcribed. The teacher interviews and video recall sessions were also fully transcribed. The final stage of this cycle involved open coding of all of the data sources in NVivo. This allowed confirmation of the key moments and frameworks for further reduction and bracketing in the later phases of data analysis. Free imaginative variation was predominantly used in this phase to determine a fixed identity for the phenomenon of dramatic engagement (Moustakas 1994). Phase 2: Emergent, open coding of language in transcripts from key moments.

Once the four themes of pedagogy, art form and dual state awareness were determined, the three key moments (one in each workshop) were analysed and openly coded. 177 references in three data sources were coded. This involved creation of 40 specific nodes and 75 memos in NVivo. This process involved three full cycles of analysis of each key moment. Initially each key moment was completely coded within the four themes. Each theme was then analysed, and secondary nodes created during another inspection. Secondary nodes that emerged at this point included items such as questioning, repetition, drama devices, active listening, role language, task language, self-correction and narrative tangents. The secondary nodes were then used to cluster memos. This began to add key insights into the phenomenon of learning language through dramatic engagement in process drama. Building these categories onto the framework allowed for further refinement of the thesis structure and specifically Chapters Four, Five and Six. Phase 3: Emergent, open coding of non-verbal communication in key moments.

During this reiteration of open coding, the three key moments were again analysed in detail in the same manner as the previous phase, through looking at the categories of characterisation, group dynamics, narrative, scaffolding and tension and how these were demonstrated through body language. This coding was conducted in three reiterations across three data sources, analysing 155 references. The first reiteration surfaced new categories of inspection, which included items such as eye contact, physical contact, gesture for meaning, emotive gesture, mime and laughter. These emergent themes were used in a secondary analysis to create new nodes and memos that were then clustered. This informed Chapter Four.

Phase 4: Axial coding of verbal and non-verbal communication strategies of four specific cases during each key moment.

The fourth phase looked at bracketing of four cases only. The key moments from each workshop were analysed but this time focus was on comparing and contrasting four variations – Hiro, Mateo, Mee and Kang. 124 references were analysed from the three data sources across four cases. Narrative, role and tension surfaced as new tertiary categories. Inside each of these categories, patterns emerged that are explored in Chapter Four. During this phase of axial coding, quantitative counts were made of turn- taking, role-switching and tension, further informing this chapter. The outcome space was charted, and the gaps identified and further explored. These gaps are where the phenomenographic outcome space was used to extrapolate new insights about the phenomena of dramatic engagement in language education and process drama. New types of role, narrative and tension were discovered and are explained in Chapter Four.

Phase 5: Axial coding of video recall transcripts.

This phase involved four cycles of analysis of the video recall interview transcripts. Applying reduction techniques, the primary category of dual state awareness was chosen to analyse the video recall transcripts. This revealed three emergent themes: metaxis, meta-emotion and metacognition, which are defined and explored in Chapter Six. A second inspection of the video recall transcripts within these themes revealed secondary nodes on an additional nine elements. 34 references were analysed across two data sources. These elements were agency; choice made in the moment; content and form; critical thinking and feeling; distance; motivation and curiosity; multiple states; regulation and moderation and self + group + ecosystem. This informed Chapter Six.

Phase 6: Selective coding of all participants across all sources.

As part of developing the theory and practice sessions of this thesis and looking at proof of concept to validate the emerging description of the phenomenon of dramatic engagement, a return was made to all data sources to cross-reference by the ten participants, the facilitator and researcher. 1,243 references were analysed over 13 data sources. This further informed Chapter Six.

Phase 7: Selective coding of major themes across all sources.

As an additional form of reduction and in order to determine four profiles of participants, a revisit was made of the thirteen data sources and analysed 1,349 references. This informed Chapter Seven. At each stage of the research, the literature and key texts were checked for new interpretation and understanding. The research was made in reiterative stages and choices made at each stage documented. This was validated with presentations and papers at conferences discussing phenomenography and the research process.