Balancing linguistics and ethnography in a study of business news
In the introduction to this edited volume, Rampton et al. (2015) write that ethnography may serve as 'a way of enriching a fundamentally linguistic project' (ethnographic linguistics) or 'alternatively linguistics can be a way of helping researchers with a range of different backgrounds to reach deeper into the ethnographic description of social or institutional processes' (linguistic ethnography). Work that combines linguistics and ethnography in journalism studies tends to blur the distinction between ethnographic linguistics and linguistic ethnography. My own research on business news (Van Hout and Jacobs, 2008; Van Hout and Macgilchrist, 2010; Van Hout, 2010; Van Hout et al., 2011) serves as a case in point.
Between October 2006 and March 2007, I made 47 research visits to the business newsdesk of De Standaard, the largest Dutch-language quality newspaper in Belgium. Overall, conditions of access were excellent.
I was granted a workspace next to one of two journalists whom I had met during preliminary research interviews. Upon simple request, I was given a guest account which gave me access to the editorial platform, the newsroom diary and mailbox. Moreover, I had all access to the various editorial meetings and was able to make extensive fieldnotes, collect newsroom documents such as press releases, internal memos and working documents, conduct formal and informal interviews and log reporters' writing processes in real time.
During my fieldwork, the business newsdesk employed thirteen professional journalists (one female, twelve male): a desk chief, a section editor and eleven 'editors' (Dutch: redacteurs), i.e. senior reporters assigned to fixed domains ('news beats') for the print edition of the newspaper: banking, finance, energy, transport, technology, labour, media, marketing and human resources. In addition, the economics newsdesk hired the services of a self-employed financial markets journalist and occasionally relied on freelancers and interns. Two editors from the online newsdesk were permanently assigned to the economics newsdesk to content manage the .Biz website, the online version of the print edition. Apart from the section editor and the two online editors, all business reporters were in their mid-thirties to early fifties and had considerable professional experience, some at other newsdesks, others at rival newspapers or broadcast media.
Business news comprises both 'hard' news about financial markets, corporate finance and government expenditure and 'soft' news about personal finance, consumer products, marketing and technology. Despite its prestige (cf. the Financial Times or the Wall Street Journal) and omnipresence in quality news media, business news remains an under examined practice of professional journalism. However, we do know that business newsrooms are flooded with company PR, government announcements, industry surveys, newswire messages and news alerts (Doyle, 2006) and that business journalists and business professionals form an increasingly interdependent relationship (Grunberg and Pallas, 2013). When I began the research for this project in 2006, I was primarily interested in how preformulation, i.e. the newspaper-like style of press releases (described at length in Jacobs, 1999), impacted business journalists' work and enlisted ethnography as 'method' (Lillis, 2008): a procedure for collecting talk around texts.
However, what captured my attention once I began fieldwork were not isolated discourse events - the act of reproducing preformulated texts - but rather the densely intertextual social practices 'of research, information selection, collegial coordination, and editorial conversation' (Boyer and Hannerz, 2006) that characterise newsmaking. Each time a journalist lifts a quote from a telephone interview and inserts it into a news story, a truth claim is reported and a journalistic stance is encoded. It is through this intertextual lens that I gradually learned to see journalism practice as a recursive and relatively stable process of textual organisation and coordination. This insight turned my gaze away from applied linguistic notions of text towards ethnographic approaches to context (Lillis, 2008), genre (Briggs and Bauman, 1992) and writing (Barton and Papen, 2010). In other words, by trying to make sense of what I was observing, my epistemological focus shifted from using ethnography as a methodological add-on (ethnographic linguistics) to embedding discourse analysis in an ethnographic framework (linguistic ethnography).
Ethnography privileges participant perspectives, speaks to theoretical claims made elsewhere and favours thick descriptions of worlds inhabited by informants - and their researchers. These tenets cast a wide net which I tried to haul in empirically by tracking newsroom text trajectories (Blommaert, 2005) across time and space. I started with the identification of newsworthy events via their selection and negotiation during editorial conferences and followed their transformation into newspaper articles. Embracing linguistic ethnography's openness towards data and methods, its commitment to participation, and its aesthetic of smallness and slowness, I combined discourse analysis and computer-assisted writing process analysis (Latif, 2008) to 'deepen' my ethnographic descriptions of business journalists at work.
More specifically, I used two software applications that recorded in real time what journalists did on their computer screens. Inputlog (Leijten and Van Waes, 2006) is a Windows based keystroke logging tool that records keyboard strokes and mouse movements. Camtasia Studio® is an online screen registration tool which records computer screen action (Degenhardt, 2006). Both applications ran in the background and did not interfere with normal computer operations. Collecting the keystroke logging data was preceded by a period of 'informant recruiting' in which I asked for reporters' informed consent. I offered them a written agreement detailing the objectives of my study, permission to use the data for scientific purposes and data veto right. Out of the twelve reporters (the desk chief did not write stories on a regular basis), I actively recruited nine. Four accepted my invitation to participate and two declined. Two computers were declared unresearchable and another reporter lost interest because the logging software slowed his computer down.
The software added contextual depth to my observations by contrasting retrospection and social action. I asked journalists to comment on excerpts from the screenvideos during so-called 'cue-based retrospective interviews' (Perrin, 2013). These authorial metacomments provided insights about journalists' understandings of news writing, journalism and power. The software made it possible to compare what journalists said about their writing practices and what they actually did on screen. Crucially, these contrastive observations enabled me 'to explore what's significant and at stake for writers at specific sociohistorical moments and, importantly, thus to engage with what is significant contextually' (Lillis, 2008).
Along the way, I developed a four step research protocol that involved
- (i) identification of stories within the news beats of my key informants;
- (ii) asking for reporter permission to record their writing process; (iii) data recording and storage; and (iv) conducting a retrospective interview as soon as the reporter had filed the story for copy-editing. From these data, I extracted a core set of 18 cases that document empirically how particular stories entered the newsroom, were discussed during staff meetings, and produced and filed for copy-editing. The cases were selected for their specificity and not as representative samples (Small, 2009). Every logged story comprised at least four data files: the Inputlog data file, the Camtasia screenvideo, an audiofile of the retrospective interview and copies of the source texts. In addition, relevant working documents, email messages, interview notes made by the reporters and fieldnotes were included. I made rough transcripts of the screenvideos and retrospective interviews and had Inputlog generate analytical files from the raw logging data. These analytical files range from detailed records of the recorded writing session, including input actions (keyboard, mouse click or movement, window nagivation) to linear (i.e. line by line) representations and descriptive statistics about the writing session (total number of characters and words produced, and average pause times).
Analytically, the challenge was 'to keep the research centred on the tension between text and social practice and not to drift into either a text-free analysis of social situations or a decontextualized analysis of text' (Scollon, 1998). In what follows, I illustrate this balancing act between text and professional practice by analysing how an unplanned news story made its way through the newsroom and into the newspaper.