Making interview transcripts real: the reader’s response

Realizing meaning

The function of data is to facilitate the communication of a message from one entity to another. Messages and how they are given meaning is at the heart of sociological endeavour, and fundamental to our understanding of work and employment. Consequently, this note is important for scholars who seek to explore what might be going on as they ‘realize’ the meanings of their participants. During verbal communication a multitude of messages are continuously sent and received by both speaker and hearer. It is said that much of what is communicated during an interpersonal exchange comes under the umbrella of nonverbal communication (Mehrabian, 1972) and is often below the level of conscious awareness (Berrey, 1988). This non-conscious judgement is thought to account for a significant proportion of the conversation, yet has little or nothing to do with what is being discussed (Burnett and Motowidlo, 1998; Hollman and Kleiner, 1997). Nonverbal cues include body movements; paralinguistic voice behaviours such as pitch, intonation, pace and elongations; and facial expression.

Verbal language emerged in face-to-face communication where vocalizations are typically accompanied by visual cues as part of speech events (Beattie and Shovelton, 2000). When visual cues are absent, such as during telephone conversations, people hear more diverse audible cues: with the absence of one mode of communication affecting the functioning and impact of another (Fichten et al., 1992). Scholars have also noted the interdependence of communication modes where combining visual with verbal messages changes attitudes (Rossiter and Percy, 1980) and the juxtaposition of images with text has an impact on what you read (Barrett and Barrington, 2005).

Readers comprehend text on three levels: by considering the specific words on the surface level; encoding the ideas conveyed within the text at the textual level; and via the information described but not directly mentioned at the situational level (Van Dijk and Kintsch, 1983). The situational level has been described as the ‘big picture’ that readers draw when engaging with text (Lehman and Schraw, 2002: 740). Studies of communication often examine its apparent inbuilt richness. Face-to-face is often considered the richest mode of communication and text, in the form of prose or numbers, perceived as the leanest (Daft and Lengel, 1986). However, subsequent research has found that neither the encoding of the message nor the medium through which it travels regulates its richness or its impact (Boland, 1991 cited in Lee, 1994). Studies conclude that the richness, or otherwise, of communication is a complex and ever-changing product of the interaction between the mode; content and context; sender and audience; form and function; medium; and fluency of the message.

The form of natural talk

Scholars have identified that dysfluency in communication impacts the way in which the audience evaluates the content of the message for both verbal and textual modes. An examination of the extant literature, however, has highlighted that the manner in which an audience appraises the dysfluencies of each mode, and the impact of this appraisal, differs.

Verbal communication is notoriously disrupted: speech contains a range of interruptions including pauses, interjections and revisions. This represents what Goffman (1981) calls natural talk. The level at which natural talk shifts into the category of dysfluent speech was assessed by research in the 1970s at three dysfluencies in 100 intended syllables. However, studies from the 1990s onwards saw a shift in this level of assessment (Roberts et al., 2009). This shift is reflected in articles published in the Journal of Sociolinguistics, among others, where our evolving speech patterns (reflecting multi-culturalism and other societal changes) have resulted in increased linguistic variations (Sharma, 2011): extenders such as ‘and stuff’ (Cheshire, 2007), fillers ‘um’ and ‘er’, ‘like’ and ‘be like’ are linguistic dysfluencies on the increase in certain cultures (Cheshire et al., 2011). As a result, recent research has noted that interruptions can be up to 14 variations per 100 intended syllables and the speech is still considered natural talk.

These interruptions do not require the conscious intent of the speaker (Provine et al., 2007). Furthermore, and significantly, they are noticed and yet disregarded by the hearer (Arnold et al., 2004). Thus, the customary interruptions of natural talk means that the hearer hears – and as a result, is cognizant of – the verbal disruptions and yet skips the blanks and retrieves just the information delivered by the speaker in order to engage in the conversation. Yet, the increasingly interrupted nature of our natural talk and its potential impact when reading verbatim transcripts in qualitative studies has not been considered in the literature (also noted by Bezemer and Mavers, 2011). This research note addresses this gap, and examines how scholars suggest we interact with textual dysfluencies.

The form and function of text

Despite the contention about what makes good textual communication, most agree that text needs to be written with consideration of its intended reader and needs to be coherent. Reader consideration is having regard for pragmatics, which includes the commonality of language and contextual experience, along with an assessment of the reader’s knowledge of the specific words used (Iser, 2000). Coherence is ‘the extent to which text segments are linked structurally to themselves and information in memory’ (Lehman and Schraw, 2002: 738). Local coherence is achieved via the unambiguous reference to adjacent text segments; whereas, global coherence relates to the reader’s ability to construct and integrate inferences into a situational model (Lehman and Schraw, 2002). Coherence is required for readers to make sense of the text and leads to comprehension.

Comprehension is used here as a purposefully contentious term: orchestrating the debate of whether or not there is a meaning to be comprehended. The notion that a reader can comprehend the meaning of a text presupposes that the text holds a meaning, and that this meaning exists independently of the reader’s experience. It is the challenge of the view that there exists a single true meaning of a text that underpins literary criticism and the reader-response theories of Fish (1980) and Iser (1978). The exposition of the processes that surround meaning making, often core to qualitative research studies, make an examination of the lessons of literary criticism and reader-response theory of great interest.

Literary criticism and reader-response theory emerged as a reaction to those who commented on literary prose, who held the text at the centre of the analysis, and were given licence, or somehow acquired authority, to dissect the text to expose the hidden meaning. Reader-response theorists, as the name suggests, moved the emphasis towards the examination of the reader and their experiences in the act of reading, and consider the text as the stimulus to this action. Literary denotes, among other things, the telling of stories. Drawing on oral histories and personal accounts to air the experiences of employees is the backbone of much of our understanding of the sociology of work. The rich narratives of first-hand accounts continue to be popular in employment studies (Taylor et al., 2009) and are being sought in quarters in which they have previously been overlooked (Pratt, 2009). Consequently, reader-response theory is helpful in furthering our understanding of the possible implications of our engagement with the product of much narrative-based research: verbatim transcripts.

Reader-response scholars typically focus on literary fiction and poetry, the authors of which have strategies of leaving pauses and gaps (Fish, 1980). Plot twists, changes of pace, juxtaposed settings, incomplete accounts and partial sentences represent intentional dysfluencies in the narrative (Iser, 2000). These gaps allow room for the reader to make inferences: to bring into play their ‘faculty for establishing connections − for filling in the gaps left by the text itself’ (Iser, 1972: 284–5); and to gain stability in an act of making sense (Rosenblatt, 1978). The development of reader-response theory was largely phenomenological: scholars’ belief that readers immerse self into the text in this manner had no evidential basis bar the apparent ability of one text to speak to different readers in a variety of ways (Iser, 1978).

However, subsequent research examining how readers assimilate textual communication echoes this early work. Studies found that readers ‘ruminate’, ‘generate additional inferences’ and then produce a workable integrated model which incorporates the currently active entities and relevant background knowledge (Zwaan and Radvansky, 1998: 166). This background knowledge is held in the reader’s memory (Ericsson and Kintsch, 1995). Meaningful sentences are stored in long-term working memory, but sentences that are interspersed with interruptions or incoherencies are stored in short-term working memory with long-term working memory being scanned for connections to enable the reader to make sense of the text (Ericsson and Kintsch, 1995). Put differently, in making sense of dysfluent text the reader integrates the text with their own words − based on their own experiences (Cornelius and Boos, 2003) and background knowledge (Lehman and Schraw, 2002) − which then produces a revised narrative (Zwaan and Radvansky, 1998).

This integrative process is implicated in Golden-Biddle and Locke’s (1993) study of how ethnographic texts convince readers. Golden-Biddle and Locke identified three aspects: authenticity, plausibility and criticality. The aspects of criticality that they argue lead to readers being convinced is when a text ‘probe[s] readers to re-examine the taken-for-granted assumptions’, achieved by ‘carving out room to reflect, provoking the recognition and examination of differences and enabling the reader to imagine new possibilities’ (1993: 595). These communicative practices create a ‘place for readers to stop’ and to ‘take time out’ (1993: 610); that is, they purposefully interrupt the text, resulting in textual dysfluencies. As a consequence, readers ‘fill in’ the missing information’ (Rapp and Taylor, 2004: 999) which then becomes a ‘rich structure that connects text and knowledge as well as personal experiences’ (Ericsson and Kintsch, 1995: 230). The effect of this integration is not only that the reader makes sense of the text, but that they do so by ‘filling in’ the gaps in the narrative with themselves, their experiences, with the result that the narrative becomes a co-creation of the researcher and the researched which resonates and rings ‘true’. This process of ‘filling in’ offers one possible response to my question − ‘Why do I seem to agree with them more now I’ve read what they said?’ − it suggests that ‘them’ has become ‘us’.

Reflecting on research

Questioning the research context and the nature and impact of the interaction between the researcher and the researched − be it animal, plant, participant or text − is to be reflective of our research practice. Research is not undertaken in a vacuum. It is central to knowledge creation and dissemination, whether to students or practice, and is set against the backdrop of increased pressure in higher education. The pressures affect the three pillars of academic work − research, teaching and engagement − with scholars required to demonstrate impact in each. The pressures are therefore underpinned by a multi-directional pull on limited resources and this means that the processes of research are under increased scrutiny. Allied to this is the growing requirement for qualitative research to demonstrate ‘quality’ and robustness in the evidence-based climate (Duncan and Harrop, 2006).

In this climate, the question of whether qualitative approaches can demonstrate the necessary rigour frequently raises its head (Spencer et al., 2003). The question may be old, yet the response continues to suffer from being held-up against the apparent precision and objectivity of quantitative approaches. As a result, the demonstration of robustness in qualitative research remains an important issue. Qualitative research − certainly the interview, focus group or ethnographic approaches − often requires more time than other traditions and, as a result, can absorb greater resources. Consequently, the resource pressure may influence the practice of qualitative scholars, resulting in a reduction in these approaches to research, and/or academics may feel under pressure to spend less time engaging with the data.

Another prospect is that the research process may be increasingly segregated, with aspects being undertaken by different (and cheaper?) personnel. Research assistants may interview, transcriptions might be produced externally or via voice recognition software (Park and Zeanah, 2005), with another academic reviewing the transcripts. Additionally, scholars have greater mobility with multi-cultural research on the rise (Kim, 2012), and with the prospect of expanding the richness of our knowledge there comes the increased possibility of the use of during-interview interpreters (Williamson et al., 2011) or post-interview translators (Temple, 2008). Segregation in the research process may of course be the choice of the academic and for some may be a long held custom-and-practice. Whether by virtue of cost−benefit, tradition, geography, language, or other then arguably there are a wider range of instances which may result in either less engagement with, or a growing detachment from, data as they move from source to transcript. Thus, the current research context introduces the possibility of another gap (to be filled?) in the conversation.

Considering the influence of the research context is to be reflexive. Reflexivity can be understood as examining one’s own personal, possibly non-conscious, reactions when undertaking research; exploring the dynamics of the researcher–researched relationship; and focusing on how the resultant research is co-constituted, polyvocal or socially situated (Denzin and Lincoln, 2005; Finlay, 2002). Within an interview context, reflexivity is to be aware that the researcher and the researched affect each other mutually and constantly (Alvesson and Skoldburg, 2000). The lessons from these studies apply here (indeed, are demonstrated here) and yet it is the mutual and continuous nature which differentiates the interview from the reader-text setting.

During conversation we share a social space and this gives rise to multi-modal and multi-directional communication: a chorus of verbal, visual, bodily and/or paralinguistic language. In this setting, whether we are speaker or hearer, we are interactants: accounts are co-constituted since any gaps in speech are ‘filled’ with one or other of these modes, emanating from all parties by virtue of their attending to the same communication event. In contrast, in the realm of reader-text the context shifts from communal to individual, and any gaps in the talk cannot be filled by the speaker (writer): only by the reader.

I now provide two examples which consider the transition from communal to individual. It is acknowledged that they may not travel, and that this is best ‘tried out’ with your own data. The first extract is from an interview with a senior manager in a large public service organization. This portion of the interview centred on his approach to performance management, he said:

I mean I’m well, we’re not here for anything other than er…you know … getting things done like. I know they … er … they might not like it … or um … me … but what’s … what’s that to me really. They have to just er to just get on and do what … you know… what’s to be done. There’s targets aren’t there and er well there you are.

During the interview, what I heard him say − verbally and visually − was: ‘I’m here to get things done. They might not like it but they just have to get on and do it.’ His tone of voice and body language made clear that he had little regard for the detrimental impact of target setting on members of his team. His facial expression said ‘tough’: employee engagement was low on his to-do list. However, on reading the interview transcript some time later, I found myself ‘filling in’ the dysfluencies with a different voice: specifically, a voice which spoke of a dislike for the pursuance of targets and a concern, albeit a resigned concern, for the effect on his team. I found myself agreeing with this ‘new’ empathetic participant, that indeed both he and his team do just have to ‘get on and do’.

The second extract is taken from an interview with a nurse in an NHS hospital. It presents another example of a differing ‘reading’ of an interview when in situ compared to the verbatim transcript. This portion of the interview centred on her colleague’s attitude and behaviour. She said:

Thing is when she comes here … well, she has a mortgage to pay and that. She cares about the patients, but … the thing is she doesn’t matter in it all … what matters is the patients … yeah like I say she’s got a lot on.

During the interview, this remark was delivered in a harsh manner and was wholly dismissive of the work−life balance issues that her colleague was experiencing. Yet, when I read the text sometime later I ‘heard’ a very different comment, namely: recognition of the good nursing care that her colleague was delivering, the repeating of her colleague’s deferential attitude, along with concern for the pressure that her colleague was under − concern that was not evident during the interview. These two examples highlight to me, on reflection, that had I engaged with the interview transcripts only, the messages I would have received, and subsequently delivered to my audience, would have differed from those I took from the interviews.

Conclusion: responding to the inevitable, ‘so what?’

This research note has explored the impact of the changing form of our natural talk on the textual representations contained within verbatim transcripts and, drawing largely on reader-response theory, examined the possible effect of readers’ engagement with this talk-as-text. The response to the inevitable ‘so what?’ question draws together the argument that runs throughout this note and has its foundations in the reader-response literature. If the lessons from this literature are accepted, then it is acknowledged that self-in-text integration is nothing new. However, there are a number of societal or contextual changes which might have an impact on scholars working in the field; they are: the increasingly interrupted nature of our natural talk; a growing interest in narrative-based research; the increasing resource pressure and evidence-based climate in higher education; and a range of other factors, including internationalization or virtual teams, which may result in a greater segregation in the research process. I argue that one or more of these factors may affect the representation of research transcripts; influence the way in which we engage with data and conduct our analysis; and, maybe most importantly, impact on the lessons we pass onto our audience (as highlighted by the examples from my own research).

This statement is firmly set in the ontological world where ‘on one hand, how researchers see data and the meaning attributed to it is what makes data useful, interesting, and a contribution to knowledge. On the other hand, our biases and perspectives influence interpretation throughout analysis’ (Weston et al., 2001: 384). However, this world accepted and embraced, what are the suggestions that result from this note for qualitative scholars in their pursuance of rigour and quality in their research? Firstly, academics are called to be mindful of how they engage with research transcripts. They are asked to be cognizant of the possible impact of their reading of talk-as-text. Researchers should take time to consider how they act upon the text while it acts upon them. Secondly, transcripts should not be used as the sole source of data wherever possible. More than one interviewer, if resources allow, would offer an opportunity for another viewpoint on the data. Further, researchers should engage with audio, video, or field-notes alongside the text. This could lead to the addition of in-text comments such as ‘said ironically/sarcastically/enthusiastically’ which offers a sense of the tone of the remark. This is not to suggest the ‘truth’ will be discovered or findings more ‘correct’, but it offers greater exposure to the different communicative modes and thus to a wider experience on which to report.

Thirdly, if the research process is segregated between personnel, for whatever reason, then those involved in the research should ensure that they engage in a discussion of their individual sense-making and any attributions made to the transcripts (also recommended by Weston et al., 2001). This approach is more likely to lead to the emergence of the core constructs and demonstrate rigour to the researcher, research team and any third-party audiences. Fourthly, and finally, the assumptions and inferences made when reading the transcript should be made transparent. This may mean the transcripts being annotated by the academic − maybe in the form of Finlay’s (2002) meta-reflective voice − making explicit their reflections such as noting what they thought was meant by what was said as opposed to what was said (Bezemer and Mavers, 2011). The audience may agree or disagree, and one would hope they would do one or other, but are at least provided with the information. The audience referred to here may be other members of the academic team, research group members, peers, reviewers or funders. This audience offers the prospect of engaging in a (albeit sometimes one-way) form of conversation, with the aim of providing ‘evidence’ of the robust processes of qualitative research.

In closing, I would like to highlight that communicative ‘misunderstandings’ may also have implications for work and employment more widely. Considering the changing nature of the workplace, the spread of virtual teams, home-working and working on-the-move (Felstead et al., 2005) all give rise to an increased distancing between colleagues. The by-product of this is an escalation in textual communication, whether through email or virtual workspaces, with varying degrees of ‘fluency’. Reflecting on the lessons of this note, the increased use of textual communication may result in teammates considering themselves more alike, which could give rise to a more harmonious team or, alternatively, greater isomorphism. As a result, future research might wish to consider the form of communication between colleagues; this might add another dimension to our understanding of how workplaces ‘work’. Furthermore, scholars might consider the ‘usefulness’ of ambiguous textual communication: those with an interest in conflict management or arbitration, for example. In addition, researchers and practitioners who wish to explore, or further the development of, learning organizations, tacit knowledge or policy-to-practice, might also examine how understanding may be disseminated and assimilated when transmitted in dysfluent text.