Questioning Transcription: The Case for the Systematic and Reflexive Interviewing and Reporting (SRIR) Method

Abstract: The recording and verbatim transcription of interviews is often considered to be one of the more tedious but necessary aspects of the in-depth qualitative research process. While transcription is undoubtedly a necessary methodological tool for researchers focusing specifically on discourse and language, it has also been widely adopted by researchers across the social sciences, and is sometimes advocated as a means of inherently improving the rigour of qualitative research. Based on recent experience from fieldwork in rural China, where I had initially expected to utilise the verbatim transcription method, in this article I critically assess the role of transcription in the design, implementation, and outcome of cross-cultural multilingual qualitative research. I argue that, in certain cases, verbatim transcription can limit the kind of information that may be considered valuable as data, and delay the processes of data reduction and analysis, thus separating the researcher from the fieldwork event. In response to these critiques, I propose an alternative approach to collecting, categorising, coding, and analysing qualitative data: the systematic and reflexive interviewing and reporting (SRIR) method. The SRIR method utilises semi-structured and unstructured interviews conducted by two or more researchers. After completing an interview, researchers engage in reflexive dialogue, and jointly write interview and analysis reports. In this way, the SRIR method begins the process of coding and analysis in situ, thus facilitating critical engagement with emergent themes during fieldwork rather than afterwards. The method is, therefore, ideally suited to research projects that are designed to be open ended and flexible, in order to follow up on new information and potentially even change focus.

Key words: fieldwork; transcription; interviewing; survey methods; epistemology; grounded theory methodology; rapid rural appraisal

4. Grounded Data Reduction, Reflecting Local Realities, and Future Methodological Development

In this article, the systematic and reflexive interviewing and reporting (SRIR) method for collecting, organising, coding, and analysing rich qualitative data during and after in-depth fieldwork is outlined and elaborated. The SRIR method was developed as an alternative to the method of recording interviews and transforming them into text through verbatim (i.e., word-for-word) transcription prior to coding and analysis. While verbatim transcription is undoubtedly necessary for research focusing specifically on discourse and language, it is also widely utilised by researchers across the social sciences, and is advocated in many prominent textbooks and guides used to train qualitative researchers (ATKINSON & HAMMERSLEY, 2007; BRYMAN, 2008; CRESWELL, 2007). For instance, in the popular book "Constructing Grounded Theory: A Practical Guide Through Qualitative Analysis", Kathy CHARMAZ states:

"Coding full interview transcriptions gives you ideas and understandings that you otherwise miss. In contrast, coding from and across notes might give you a wider view. It can, however, contribute to grounded theorists going around the studied phenomenon ... [and] risks constructing superficial analyses" (2006, p.70). [1]

This has resulted in a widespread expectation that qualitative researchers—particularly graduate students and early career researchers—use the verbatim transcription method, both to ensure that they are not missing anything, and to demonstrate rigour.¹⁾ [2]

The SRIR method was initially designed in response to this expectation (both from the University of Leeds where I was studying for my Ph.D. and from myself) that I use the verbatim transcription method for a research project that entailed several phases of in-depth and open-ended qualitative fieldwork. After the first round of fieldwork, I critically reflected on the epistemological and practical implications of utilising the verbatim transcription method for research relying primarily on semi-structured and unstructured interviews. In the case of my research project, I identified a number of problematic issues that would have arisen from the act of transcription—particularly with regard to the ways in which the method reduces data and defines what kind of information is valuable, thus fundamentally shaping the knowledge that is produced. Ultimately, I decided that the verbatim transcription method would have been unsuitable for the type of research I was undertaking. [3]

As an alternative to verbatim transcription, the SRIR method gathers data primarily through semi-structured and unstructured interviews conducted jointly by two or more researchers. During the interview process, researchers note responses, observations, feelings, hunches and preliminary analyses—only recording interviews if possible and convenient. After the interview is complete, researchers engage in reflexive dialogue, and write up systematic interview reports (SIRs) that include responses and some direct verbatim quotations, as well as non-verbal data. Every two weeks researchers jointly analyse the SIRs alongside other observational, documentary, and visual data, and write up preliminary analysis reports (PARs). [4]

The SRIR method begins the process of coding and analysis in situ, and facilitates critical engagement with emergent themes during fieldwork. It also allows researchers to reflexively reduce data into manageable chunks through reflexive dialogue and the creation of SIRs. This contrasts with transcription, which reduces data by stripping out non-verbal information through the textualisation of recordings after fieldwork has been completed. The SIRs and PARs represent rigorously and reflexively co-produced, co-reviewed, and pre-analysed data outputs that can be imported into qualitative data analysis (QDA) software packages for deeper analysis alongside other types of data collected during fieldwork. In this way, the SRIR Method provides the basis for pluralistic data collection and analysis. It allows researchers to broaden their analytical lenses, and encourages openness in identifying areas of primary importance for inquiry during fieldwork, as well as the types of information that need to be collected in order to most effectively analyse emerging themes. [5]

The key features of the SRIR method are flexibility and openness. This means that, while it is presented in this article as an alternative to a reliance on verbatim transcription to generate data, the two methods are not actually mutually exclusive. Indeed, the SRIR method is most effective when utilised in coordination with some selective verbatim transcription of recorded interviews to triangulate meaning and identify key quotations for use in research outputs, thus allowing participants to speak with their own voices. It should also be acknowledged that many researchers already use a combination of the techniques outlined in the sections below. In this article, I seek to systematise (and validate) these widely used techniques, and integrate them into a flexible and pluralistic methodological approach that provides a systematic framework for the collection, organisation, and analysis of data collected during qualitative fieldwork. This has the potential to contribute to the methodological literature on working in pairs and small teams—particularly groups bringing together emic and etic perspectives. The SRIR method also has great potential to more effectively produce anonymous data outputs, thus making it possible to open up data sets, making qualitative research more transparent. [6]

The procedure for using the SRIR method presented in this article developed out of my recent experience (2012 to 2013) of conducting in-depth qualitative fieldwork exploring the linkages between financial services, livelihoods, and local socioeconomic development in rural Jiangxi Province, China. Over the course of fieldwork I conducted 78 semi-structured interviews lasting approximately one hour each (48 of which I was able to record), and I engaged in 38 unstructured interviews—that were more conversational in style—of high relevance to my research that I was not able to record.²⁾ My initial research design included plans to complete verbatim transcriptions of all my interviews in an attempt to establish validity and rigour. This was based on expectations from my institution, and also my own interdisciplinary social science research training, which had instilled in me the belief that, whenever possible, verbatim transcription should be completed. However, I immediately encountered problems with this approach, and realised that recording and transcribing every interview and conversation would be practically infeasible for the type of open-ended and flexible research project I was engaging in. I also came to the conclusion that it would be methodologically unsuitable, as transcription would focus my data collection efforts on types of information that I regarded to be of less importance—i.e., the literal responses of interviewees rather than more holistic observation and interpretation of the fieldwork experience as a whole. This caused me to reflect on how the verbatim transcription method actually shapes the generation and interpretation of data—framing and defining what is considered valuable, meaningful and true. [7]

The rest of the article is organised as follows. In Section 2 I examine some of the main problems that can arise from recording and transcribing qualitative data based on my fieldwork experience. In Section 3 the SRIR method is outlined—particularly focusing on collaborative interviewing, reflexive dialoguing, and the production and analysis of SIRs and PARs. In Section 4 I begin by illustrating some of the fundamental differences between the SRIR Method and other approaches, especially with regard to how and when data reduction occurs and which kind of information is considered valuable. I then go on to explore the potential for future methodological development, improvement of transparency, and integration with other qualitative and quantitative approaches. In Section 5 I conclude by summarising the key strengths and limitations of the SRIR method, and highlighting the importance of continuous critical methodological reflection. [8]

One of the main criticisms levelled at social science research that relies primarily on qualitative methods (e.g. unstructured/semi-structured interviews, informal conversations, focus groups, and observation) is that the data collected are subjective, unscientific, and lacking in transparency—particularly in comparison with positivist approaches, which seek to emulate the natural sciences by producing research that is generalisable, verifiable, and reproducible (BRYMAN, 2008). In response to these criticisms, some researchers have attempted to clean up and quantify the messy elements of qualitative data collection and analysis through the creation of standards that qualitative research must meet in order to be considered rigorous (CHI, 1997; WERNER, 1998). One such standard that has emerged over the past few decades is the recording and verbatim transcription of interviews, conversations, and focus groups in an attempt to ensure rigour by producing a data output that can—hypothetically—be subjected to the same scrutiny as the types of data used in positivist research (HALCOMB & DAVIDSON, 2006; LAPADAT, 2000; McLELLAN, MacQUEEN, & NEIDIG, 2003; POLAND, 1995; WERNER, 1996). Many of the most widely-used research methods textbooks and guides depict transcription as a desirable step in the qualitative research process (CHARMAZ, 2006; CRESWELL, 2007), while acknowledging that it is not always possible due to time and resource constraints (ATKINSON & HAMMERSLEY, 2007; BRYMAN, 2008). [9]

The widespread acceptance of verbatim transcription as a desirable element in research methodologies across the social sciences is somewhat surprising given the established body of literature arguing that the act of transcription is, in fact, a situated practice and a theoretically-laden methodological choice that has significant epistemological implications for research outcomes (GREEN et al., 1997; OCHS, 1979; SKUKAUSKAITE, 2012). I have found these critiques to be particularly salient in my own experience of working in qualitative area studies, where much research is grounded in in-depth interviews about sensitive subjects and conducted across cultures in a variety of languages and dialects, rendering recording and transcription especially challenging for a variety of reasons. The prospect of recording and transcribing interviews proved to be particularly problematic for the open-ended and flexible research design of the project discussed in this article, as it limited my ability to quickly follow up on leads and change analytical focus when new information was presented. In the next section I outline the most troubling aspects of the verbatim transcription process for my research, and in Section 3 I illustrate how the SRIR method was developed to address these issues. [10]

Transcription necessitates the audio recording of interactions with research participants; however, this is not always possible or desirable (WERNER, 1999). Often, people do not want to be recorded, or the environment is not conducive to producing a recording of sufficient quality. For instance, some of my interviews were conducted with street vendors at a busy township market: there was nowhere to sit or to place a recording device and the background noise made any recording virtually unusable. Other times, participants were happy to have an informal conversation with me, but became withdrawn and non-communicative after I requested their consent to record. Additionally, in my experience, an interview with a single participant can spontaneously transform into a group interview if friends, family, or neighbours become curious. Some of these impromptu participants come and go throughout the conversation making it difficult to request their consent to be recorded for ethical purposes, or identify them during the transcription phase in order to put their responses into context. [11]

These examples illustrate the fact that important information is often presented when the recorder is off. Therefore, relying on verbatim transcription to transform interviews into data implicitly limits the type of information that is collected. While this narrowing of the analytical lens is necessary for research that focuses on the precise use of language and discourse, it can close off important areas of inquiry for more exploratory and open-ended research projects. In the case of my research on microcredit, some of my most important discoveries occurred when I was not able to record. This situation prompted me to consider how this crucial, but unrecorded, information could be collected, valued, and adequately analysed alongside other types of data. [12]

It is widely recognised that transcription is technically very difficult (DiCICCO-BLOOM, & CRABTREE, 2006; HALCOMB & DAVIDSON, 2006; MARKLE, WEST, & RICH, 2011; McLELLAN et al., 2003; OLIVER, SEROVICH, & MASON, 2005). Spoken language is structured differently from written language. In particular, incomplete sentences, pauses, self-corrections, partial utterances, and background noise all make producing a truly complete verbatim transcription basically impossible, and research has shown that different transcription formats significantly influence interpretation of the text (MISHLER, 2003). Admittedly, there has been rapid technological improvement in machine transcription, and recordings can now be linked with text files in many QDA software packages (EVERS, 2011). However, speech recognition software still proves to be essentially as time consuming as manual transcription (DRESING, PEHL & LOMBARDO, 2008). Even assuming technology allows for a perfect transcription seamlessly linked with the original audio in the near future, the audio-text combination would still leave out vital non-verbal modes of communication, such as posture, facial expressions, and gestures that may change the meaning of what is being said. Researcher observations, feelings, hunches, and positionality are also left not reflected upon in a verbatim transcription. [13]

These issues of technical accuracy, and accuracy of meaning, are exacerbated when working across cultures and languages. Because I work in a second language (Chinese), creating highly accurate verbatim transcripts proves extremely difficult, not least because some of my interviewees have strong accents, and interviews are sometimes conducted in local dialects (with the help of local research assistants) that do not completely exist in written form. Even if semi-accurate verbatim transcription could be accomplished, it would still leave out the ways in which things were said, which can quite often be more important to the meaning than the actual words themselves. Ultimately, this points to the importance of both linguistic and cultural fluency (VEECK, 2001), and the need to maintain flexibility and reflexivity during all stages of the research in order to achieve a useful and trustworthy understanding of what people actually mean. Reflecting on these issues during fieldwork prompted me to consider alternative ways to more effectively and holistically capture the interview experience in order to better reflect the information and different types of meaning that were being presented through verbal and non-verbal communication, as well as the wider context surrounding the interviews themselves. [14]

In my own experience of attempting to work with verbatim transcription, I found that in certain cases the method pushes critical engagement with the field to a later time when memories and feelings associated with the fieldwork event have faded. While a certain amount of distance when analysing data collected during fieldwork can provide new insights—entirely postponing analysis until a date long after interviews have been completed serves to flatten and de-contextualise the data. This issue is compounded by the fact that verbatim transcription is an extremely time-consuming process (MARKLE et al., 2011), which further separates the researcher from the actual empirical act of conducting fieldwork. During the analysis phase after my first round of fieldwork, I discovered that it takes me at least 10 hours to complete the verbatim transcription of a one-hour interview in standard Chinese. Therefore, the 48 interviews I was able to record during the fieldwork discussed in this article would have required 480 hours of transcription, or 60 full eight-hour workdays. Admittedly, I could have hired research assistants to transcribe some of the interviews, but the cost would have been high and I would have still needed to monitor the work in order to ensure a standard level of quality. In the end, even if transcription could have been completed accurately and in a timely fashion, I would have still been required to read and code hundreds of thousands of words. [15]

Because the process is so time consuming, I became concerned that the transcription phase of research would be hurried due to the pressure from time limitations to move on to the analysis and writing-up stages of my project. I felt that this would have probably produced transcription data that lacked contextual clarification or was simply inaccurate. This would have resulted in research that glossed over the complex emergent realities that I was interested in discovering during intensive engagement with the field, in favour of simplistic analysis confirming or disproving my pre-existing assumptions and theories. In this way, transcription would have reinforced the dichotomy that the field is for collecting data and the office is for analysis, when my project actually required continuous and iterative critical analysis of emerging themes during fieldwork in order to build theory from below. For this reason, I decided that it was important to develop methodological tools that would permit me to actively engage with the interview data while in the field, and allow the data to drive my research agenda going forward. [16]

Reflecting on the issues highlighted above led me the realisation that the verbatim transcription method would also shape the construction of truth, meaning, and knowledge in ways that would have been unsuitable for the research I was undertaking. This is because transcription relies primarily on the interviews that can be recorded in good quality and transformed into usable text documents. In practice, this often means that interviews which put forward a clear and coherent narrative—often with willing participants speaking in a standard accent/dialect and presenting themselves in a way that fits into the researcher’s conceptualisation of the project—are considered valuable and form the basis for the research findings. Meanwhile, other types of information that I considered to be critical to my own research, such as observations and non-verbal forms of communication, are afforded less importance or even ignored all together. [17]

This is hugely significant for the evidence base of research that relies on verbatim transcription to produce data. By transforming speech into text and largely eliminating observation and non-verbal communication, verbatim transcription necessitates the valorisation of the written word over other types of data. Meaning is reduced to the literal words that are spoken and transcribed. Knowledge is derived from the textualisation of the fieldwork experience. While the narrow focus on precise wording is necessary for research examining discourse and use of language, it is not necessarily appropriate for all research in the social sciences. This urge to capture experiences and interactions, and reproduce them exactly in written form is, perhaps, rooted in our tendency to equate the written word with validity and permanence (an affliction that is particularly acute among members of the academy). As Caius Titus is renowned to have said in the Roman Senate: "Spoken words fly away, written words remain" (Verba volant, scripta manent). [18]

Ultimately, the attempt to pin down the interview through its transformation into text leads to an epistemology that sees truth and the resultant knowledge as emanating from words that have been textualised—a process that necessarily strips away some of the context. This proves problematic for open-ended and flexible research that seeks to interrogate the continuously unfolding dynamics between researcher and research participant, and sees the research project as a co-production, which is replete with different types of meanings and understandings that need to be interpreted critically, reflexively, and iteratively. In a sense, the act of transcription is an inherently positivistic endeavour, and thus unsuitable for much research adopting an interpretivist approach. It suggests that meaning is the result of combinations of words in a certain order and is, therefore, implicitly objective in an almost mathematical way—i.e., word + word = meaning; meaning + analysis = knowledge. Interpretivist research approaches, however, see meaning and knowledge as being emergent from dynamic and socially constructed discursive processes that must be understood within their own contexts. [19]

Reflecting on these issues made me realise that transcription is, in fact, a situated and theoretically-laden practice that has epistemological implications for the reduction, coding, and analysis of data (GREEN et al., 1997; McLELLAN et al., 2003; OCHS, 1979; SKUKAUSKAITE, 2012). From this perspective, it became immediately obvious that verbatim transcription would not be the most suitable method to transform the contextually rich information I had been collecting into useful data that would allow me to analyse the ways in which rural financial services are perceived and understood in the context of Chinese rural development. This realisation prompted me to look into alternative methods, such as reflexive journalling (FASICK, 1977; HALCOMB & DAVIDSON, 2006), enhancing recall through stimulus transcription (WERNER, 1999), and partial transcription (McLELLAN et al., 2003). However, while each of these approaches certainly have their own strengths, I needed to develop a systematic set of procedures that would not be seen as overly subjective and insufficiently rigorous by my institution, which was expecting the use of verbatim transcription. Additionally, I found these approaches to be more conducive to research conducted by a single researcher working alone. I was working with research assistants and it was, therefore, important for the research to produce data outputs that could be interpreted and analysed by multiple people at different points in the project.³⁾ [20]

For these reasons, I sought a means of jointly producing data outputs with my research assistants that could be comparatively analysed using a QDA software package, and which would serve as a clear audit trail for others wishing to assess the rigour of my analysis and validity of my findings. Therefore, I decided to develop a set of methods that would allow me to plan, conduct, and analyse the fieldwork experience holistically in a systematic, reflexive, and dialogical way. After a substantial amount of experimentation, this resulted in the creation of the SRIR method, consisting of four overlapping and iterative stages: 1. collaborative outlining and interviewing; 2. reflexive dialoguing, and the production and initial analysis of SIRs; 3. collection of observational, documentary, and visual data for comparison and triangulation, and the creation of PARs; and 4. importation of all data into a QDA software package for integrated analysis. Appendix A provides an overview of the SRIR method. [21]

Prior to conducting formal semi-structured interviews, my research assistant and I created flexible interview outlines (FIOs) tailored to different respondents.⁴⁾ These FIOs were used to guide interviews and structure note taking (see Appendix B for an example of the first page of an FIO used during the third phase of fieldwork). After each interview we discussed the suitability of the questions and identified important information that we may have missed. We then reflected on ways in which the FIOs could be improved and adapted them accordingly. These FIOs were, therefore, dynamic and open to allow the research to change focus and direction, and to follow up on the numerous "unknown unknowns" that presented themselves during fieldwork (CHAMBERS & LOUBERE, 2017, p.38). [22]

During interviews, my research assistant and I took turns asking questions. This made it easier to take notes without lengthy pauses in the conversation. In addition to literal responses to our questions, we collected demographic information about the respondent(s) and categorised their relative socioeconomic status based on their housing conditions and material possessions. We also took note of their non-verbal signals, attitudes, and demeanours, the interview environment, our own intuitions and feelings, and other general observations. When interviewees said something of particular significance that could be quoted, a note was made next to the question it was in response to so that it would be possible to return to the recording at a later time to get the exact wording. If the interview was not being recorded, we would pause to quickly write down the exact words that were said in our notes. [23]

Usually, two interviews were conducted each day—one in the morning and one in the afternoon. The evenings were spent writing the SIRs (one written by myself and the other written by the research assistant) and engaging in reflexive dialogue about what had been discovered during the day. SIRs were written in Microsoft Word using the FIOs as the templates. Each person used a unique text colour so as to easily distinguish between writers. When interviewees brought up a new topic or presented unexpected information, a note was made in the report and, where appropriate, the FIO was adapted for future interviews. Important observations and preliminary analyses were written at the end of the reports. [24]

After completion, the SIRs were exchanged and the second writer added missing information, observations, analyses, and identified areas where they disagreed with the initial report. We then engaged in reflexive dialogue about the interviews, as well as our personal experiences of the day’s work and the wider research project more generally. During this time we discussed points of interest and disagreement in the reports (referring back to recordings if possible). We also tried to make sense of what we had learned and put it into a wider context. In total, this process took between two to three hours for each set of two interviews (less than 1/5th the time for a full verbatim transcription). It (co)produced concise and rigorously co-reviewed SIRs of between 3,000 to 5,000 words in length. These SIRs contained participant responses, rich contextual information, and preliminary analyses. Appendix C shows the first page of an SIR (the entire SIR is six pages long). The pink text in Chinese and English was written by the research assistant, and the black text in English was written by myself. As can be seen in Appendix C, the SIRs contain important demographic and background information of the interviewees, but this is easily made anonymous. This means that the SIRs can be made available as part of an open and transparent data set. [25]

In addition to formal semi-structured interviews, my research assistants and I also engaged in unstructured and unrecorded interviews with people on a daily basis, which often provided important insights into issues related to the research. Therefore, over the course of fieldwork 38 SIRs were written for unstructured interviews using essentially the same process outlined above, but without an FIO to guide the report writing. The data encapsulated in these SIRs proved to be invaluable, and would have been largely lost had I relied on verbatim transcription, due to the fact that I was unable to record these encounters. [26]

Throughout fieldwork I also systematically collected observational, documentary and visual data (including photos and videos), and kept a personal fieldwork journal. Approximately every two weeks my research assistant and I compiled all of the different types of data that had been collected and engaged in reflexive dialogue. This allowed us to triangulate findings, and make comparisons within and across the fieldwork sites. Key themes were explored and theorised, and areas of primary importance for further examination were identified. [27]

We then jointly wrote up PARs outlining how we understood the data and the fieldwork experience at that time. PARs were written using a similar technique to the SIRs, with my assistants and I using different coloured text to differentiate our contributions. In the PARs we highlighted differences of opinion and potential areas for theoretical development. The FIOs were then amended to refocus on areas that we determined to be of most importance for the research.

After returning from the field I digitally scanned any remaining hard data, and imported all the SIRs, PARs, the fieldwork journal, observational data, documentary data, and visual data into the widely-used QDA software package NVivo. This left me with a large well-organised digital data set that had already undergone preliminary in situ analysis. Once in NVivo, this data set was subjected to further integrated and comprehensive coding, categorisation, and analysis to theorise directly from the data in a grounded theory mode.⁵⁾ Figure 2 is an example of an SIR that has been imported into NVivo for analysis—codes can be seen on the right side of the figure as coding stripes. [29]

SIRs formed the main body of data, and since they had already undergone preliminary analysis in the field, it was relatively straightforward to identify codes and undertake deeper analysis. Since the SIR text was colour coded by author, it was possible to see where ideas had originated, and even to trace the development of analysis over the course of the reflexive dialogue process. Because I also conducted my literature review using NVivo (BEEKHUYZEN, 2008), I was able to identify how themes that emerged during my fieldwork related to the theories in the literature. Additionally, because I recorded many of the interviews and had already made note of the location of important quotations in the SIRs, it was possible to return to the recordings and do selective transcription to obtain verbatim passages for use as evidence during the writing-up stage.

Figure 2: Systematic interview report undergoing coding and analysis in Nvivo. Please click here for an enlarged version of Figure 2. [30]

The process of conducting research using the SRIR method outlined above directly addresses the problematic issues that I personally experienced (highlighted in Section 2) when considering the possibility of recording and transcribing interviews for my research project. In particular, the SRIR method does not require interviews to be recorded, so data that is collected when recording is impossible or undesirable are valued equally. It also does not rely on the accurate textual reproduction of the spoken word—a time-consuming endeavour, which often neglects important non-verbal forms of communication in favour of the precise words that are spoken. More importantly, the SRIR method prompts researchers to engage with their fields reflexively, dialogically, and analytically throughout the research process. This locates interpretation and analysis during the fieldwork event, rather than after fieldwork has been completed. [31]

In this way, the SRIR method represents a shift away from an epistemology that understands knowledge as reified, static, and concrete—i.e., interviews "written in stone"—to one that instead sees meaning as being co-produced through discursive action and embedded within specific contexts. This epistemological approach relies on reflexive dialogue between different types of people in order to critically challenge engrained assumptions, as well as "ethnocentric views and partial interpretations of empirical data" (THUNØ, 2006, p.246). Open and reflexive dialogue of this nature allows a plurality of types of knowledge to come to the fore. My use of the SRIR method outlined above particularly benefited from the inclusion of both the emic (insider) knowledge from my local research assistants and the etic (outsider) knowledge that I brought to the field. Insider-outsider research of this kind has the ability to broaden and nuance findings by looking at the field through multiple sets of eyes (BRÖCKERHOFF & KIPNIS, 2014). This has the potential to produce new theoretical perspectives and to challenge entrenched epistemological assumptions about what type of information is important and where analytical lenses should be focussed. [32]

Ultimately, the SRIR method facilitates critical engagement with emergent themes during fieldwork, rather than afterwards. The SIRs represent detailed, concise, and rigorously co-reviewed data blocks that contain participant responses, rich contextual information, and preliminary analyses and coding. Therefore, the use of SIRs—in conjunction with the PARs and other types of data collected—allows for time and energy to be focussed on deepening analysis after fieldwork is complete. Moreover, by identifying a number of significant quotations during the interview for selective verbatim transcription and inclusion in final research outputs, the SRIR method can also directly give voice to the people whose opinions, understandings, and aspirations are at the heart of the research being conducted. [33]

4. Grounded Data Reduction, Reflecting Local Realities, and Future Methodological Development

As a flexible and pluralistic mode of conducting research, the SRIR method can be considered a grounded theory approach to data collection and analysis. Indeed, many of the critiques of transcription found in this article have also been identified in the grounded theory literature. As GLASER points out: "Tape recording, I have warned over and over in my books, is too much coverage and too slow to get to analysis because of waiting for type written form" (2013, n.p.). As a grounded response to the limits of transcription for exploratory and open-ended research, the SRIR method prompts researchers to engage with the field in an open but systematic way that allows concepts and theories to inductively emerge from the empirical data (CHARMAZ, 2006; GLASER & STRAUSS, 1967). It is also inherently constructivist, as the co-production of SIRs and PARs, and their analysis both during and after data collection, allow researchers to construct "levels of abstraction directly from the data and, subsequently, gather additional data to check and refine ... emerging analytic categories [culminating in] an abstract theoretical understanding of the studied experience" (CHARMAZ, 2006, p.6). In this way, the SRIR method contrasts not only with verbatim transcription, but also with top-down methodological designs—such as quantitative survey methods. These approaches often impose pre-set theoretical constructs on the research, defining what is and is not important before fieldwork even begins, thus prematurely closing avenues of inquiry (CHAMBERS & LOUBERE, 2017). [34]

The SRIR method is, therefore, epistemologically distinct from verbatim transcription or quantitative survey approaches. However, this is not only due to a different understanding of where knowledge derives from and how knowledge is structured. It is also the result of different data reduction points—i.e., the moments when data is reduced to make categorisation and analysis more manageable. The reduction of data during research is something that is not often explicitly acknowledged. However, it is a necessary step in any methodology, and by examining where and how different methodological traditions engage in data reduction it becomes possible to understand their underlying epistemologies, and get a better sense of the types of knowledge they will produce. [35]

In survey methods, for instance, data reduction takes place prior to fieldwork through the creation of questionnaires that predetermine what questions will be asked (and what answers will be received), thus clearly framing the types of information that will and will not be collected. Verbatim transcription reduces data after fieldwork through the transformation of the interview experience into text, thus stripping out non-verbal information. The SRIR method reduces data during fieldwork through the process of reflexive dialogue and report writing. Grounded openness is maintained throughout, and questions are adapted and refined as the research unfolds and the primary areas where analysis should be focussed become clearer. It provides a means of flexibly and reflexively determining which kinds of data are more important in different contexts—sometimes opting to focus on direct quotations, in other cases giving more weight to the way in which something was said, a facial expression, the physical environment, or perhaps the group dynamics that were occurring before, during, or after the interview. In this way, the SRIR method also engages in the necessary task of reducing data into manageable chunks for academic analysis. However, these chunks are flexibly produced and pluralistic in the types of data they contain. Indeed, the SRIR method adheres to the grounded theory dictum espoused by GLASER that "all is data" (2001, p.145).

For this reason, the SRIR method is particularly suitable for observing, describing, and analysing the emergent, dynamic, and complex realities that are presented during in-depth fieldwork. In a sense, the SRIR method could be considered a revolution back to the methodological transformation instigated by CHAMBERS and others in the 1980s and 1990s through the development of rapid rural appraisal (RRA) (CHAMBERS & LOUBERE, 2017).⁶⁾ RRA was originally a reaction to what CHAMBERS (2008, p.72) refers to as the "long-and-dirty" methods of traditional ethnography and large-scale surveys, which he describes as being hugely time consuming, inefficient, and often producing inaccurate or irrelevant data. The RRA toolbox includes methods such as sketch mapping, diagramming, transect walks, and flexible interviewing (to name a few), which seek a "rigour of cost-effectiveness through exploratory iteration and trade-offs between relevance, accuracy, and timeliness, ignoring inappropriate professional standards. They apply the principles of optimal ignorance and proportionate accuracy (or appropriate imprecision)" (p.67). [37]

Similarly, the SRIR method seeks to be both more efficient and effective through systematic and reflexive data collection and analysis—aiming to achieve a "rigour of trustworthiness" rather than replicable and generalisable analysis (CHAMBERS & LOUBERE, 2017, p.39). Throughout SRIR fieldwork, unnecessary data is reduced—but only after critical and joint reflection—allowing researchers to continually refocus their analytical lenses. The SRIR method is also fundamentally participatory in nature, as it relies on continuous dialogue, "critical scepticism, triangulation and crosschecking with peers" (ibid.). It seeks to give research participants a greater role in shaping the research trajectory through their involvement in the iterative and reflexive creation of categories—i.e., simply listening to the people who are the subject of the study to discover what areas and topics are most important for examination. This is in contrast to surveys or transcription, where conceptual schemes tend to be imposed onto the field either before or after fieldwork has been completed. [38]

As a flexible set of methods, SRIR has the potential to be developed and utilised in more collaborative and participatory ways in the future. The SRIR method could be adapted for use with larger research teams. For instance, a team of 10 researchers could form five working pairs, each separately conducting interviews and writing up SIRs. The whole team could then have daily group dialogue sessions where the key themes, commonalities and differences between and across the interviews would be teased out and written up in co-produced PARs that would serve as data for analysis, while simultaneously informing future research directions and analytical approaches. More ambitiously, the SRIR method could be extended to entire communities in a participatory mode, allowing research participants themselves to be involved in the interviewing, reflexive dialoguing, writing up of SIRs, and preliminary analyses. [39]

As mentioned above, the SRIR method also has great potential to make qualitative research more transparent, providing an open-access audit trail for the collection, organisation, and interpretation of data, thereby enhancing the trustworthiness and believability of the analytical process, findings and conclusions. While some researchers have advocated increasing transparency by embedding video or audio into research outputs (MARKLE et al., 2011), original recordings such as these, as well as verbatim transcriptions themselves, are difficult to anonymise, as they are often interspersed with personal information making it easy for listeners or readers to identify individuals or places. Therefore, in order to protect the identity of research participants (and adhere to ethical standards), researchers often cannot make their primary data public. [40]

SIRs, alternatively, are easily anonymisable (see Appendix C and Figure 2), and could be uploaded to an open and publically viewable online database or wiki (BRÖER et al., 2016). Researchers involved in the project would be able to edit the reports directly on the database using a tracked changes function so that the time and author of each alteration would be recorded. Research outputs could then directly reference the online database through hyperlinks or footnotes, giving readers access to the primary data used as evidence for findings. PARs and explanations of how codes and categories were identified could also be posted online, allowing readers to make their own judgements about the analysis underpinning the research. This would dramatically increase transparency and go a long way to dispelling many of the criticisms of qualitative research related to opacity and subjectivity. [41]

In this way, the SRIR method has the potential to open up and illuminate qualitative data sets, allowing multiple researchers to access the information and work on different aspects of the data in the same way as quantitative research that utilises data from large databases. The SRIR method is also uniquely suited for use in mixed-methods approaches combining both qualitative and quantitative tools to gather data—also referred to as q-squared (SHAFFER, 2013). For instance, it could be used to follow up on large quantitative surveys in order to crosscheck and triangulate findings, or to provide deeper long-run analysis of patterns of correlation and causality observed in previous studies. During this process, the SRIR method could be used to identify other important areas or gaps for examination by future large-scale surveys or statistical analysis with existing data sets. Thus, the SRIR method and quantitative approaches could continually and iteratively feed into each other. [42]

There is also scope for the SRIR method to be integrated into studies using participatory statistical approaches aimed at giving research participants a larger role in the identification of areas where measurement is needed, and then undertaking the generation and assessment of the numbers themselves (HOLLAND, 2013). Reflexive dialogue, SIRs and PARs could be used to document how and why local people determine certain areas or topics are more important than others for measurement, and to illuminate the ways in which research participants generate numbers and define categories. In this way, the SRIR method could provide the deep data foundations underpinning quantitative analysis by providing much needed transparency and reliability with regard to what numbers are deemed important, where numbers come from, and how these numbers are arrived at. This would provide the means for researchers, research participants, and those utilising research to dig into findings, examine claims and even feed back into the research itself. [43]

In this article, the SRIR method has been outlined as an alternative to the method of recording and transcribing interviews for analysis. Based on critical reflection during a project where I was expected to utilise the verbatim transcription method, I highlighted a number of problematic issues with transcription for cross-cultural multilingual qualitative research—particularly in relation to the act of recording, accuracy, and language in the transcriptions, and the ways in which the transcription process results in physical and temporal separation from the fieldwork event. These issues illustrate that transcription is not simply a desirable step in the qualitative research process, but rather a situated and theoretically-laden act (GREEN et al., 1997; McLELLAN et al., 2003; OCHS, 1979; SKUKAUSKAITE, 2012). The transformation of the interview experience into a transcript does not capture a more authentic version of the truth. Rather, it is a methodological choice that defines which kinds of information are considered important, thus framing the types of knowledge that are produced. The SRIR method was developed as a direct response to the limitations of verbatim transcription for my own research. Unlike transcription, which implicitly valorises the literal words that are spoken and reduces data by stripping out non-verbal information after fieldwork is complete, the SRIR method values a plurality of data, and undertakes data reduction during fieldwork through reflexive and collaborative dialogue. [44]

The SRIR method is a fundamentally inductive and grounded approach to collecting, coding, and analysing fieldwork data. Researchers discuss which areas and themes are important, and create flexible schedules of questions that can be adapted to include or exclude information and follow up on leads. Reflexive dialoguing and the co-production of SIRs and PARs allows for the inclusion of not only the exact words that are spoken, but also other types of information. This includes hunches, feelings, non-verbal communication, visual data, in situ analyses, etc., which would often be left out of a transcript or never become visible in top-down approaches that limit the type of data collected prior to fieldwork. The SIRs and PARs, therefore, represent a data format that compiles various types of information which are determined to be important by researchers and research participants through systematic reflection and dialogue. This allows for plurality and flexibility, in some cases collecting more information from non-verbal communication, visual data, or observation, while at other times drawing on direct quotations and statements. At the same time, the SIRs allow for these pluralistic data to be contained in an output that can be analysed using a QDA software package alongside other data collected during fieldwork. In this way, the SRIR method seeks to broaden what is considered valid data, and produce more holistic understandings of dynamic and complex social processes. [45]

This does not mean that verbatim transcription cannot be used effectively (or may even be necessary) in certain types of research, or that the SRIR method is necessarily always a better way of transforming contextually rich information gathered in the field into useful data. Indeed, the SRIR method is most effective when utilised in coordination with selective verbatim transcription that can be used for analysis and to give voice to participants. In other words, in this article I do not aim to disparage verbatim transcription or imply that the SRIR method is superior to other methods. Rather, I seek to highlight the importance of continual reflection and openness on the part of researchers when designing and implementing research projects. The key contribution of this article is the elaboration of a systematic, transparent, and flexible set of methods to collect, organise, and analyse a plurality of data in pairs or small groups. The version of the SRIR method outlined in this article is not without its own limitations. For instance, it would need to be adapted substantially in order to work for larger sample qualitative studies conducted by big research teams. That being said, the method is flexible, compatible with other approaches, and conducive to adaptation in different contexts. For this reason, the SRIR method has the potential to be integrated into participatory research projects and mixed methods approaches in order to deepen analysis and provide the means to open up new areas of inquiry through dialogical reflection undertaken in a systematic, collaborative, and inclusive way. [46]

First and foremost, I would like to thank all the research participants who took time out of their busy lives to speak with me during fieldwork. This research would not have been possible without the involvement of SHEN Qiu and PAN Lijuan, who were both wonderful companions in the field. I owe a debt of gratitude to Ivan FRANCESCHINI, Christian SORACE, Jörg WIEGRATZ and two anonymous reviewers for their helpful comments on earlier drafts. Finally, I would like to thank Katja MRUCK for patiently guiding this article through the review and revision process. The usual disclaimers apply.

1) This conclusion is not only based on the many examples of transcription being advocated in research methodology textbooks, but also from my own experience of discussing this situation with other researchers during a number of cross-disciplinary conferences and seminars on methodological challenges in qualitative fieldwork. The expectation persists despite the existence of a large body of literature critically assessing the situated, political and theoretical nature of transcription (GREEN, FRANQUIZ & DIXON, 1997; OCHS, 1979; SKUKAUSKAITE, 2012). <back>

2) For more on semi-structured and unstructured interviewing see BRYMAN (2008). <back>

3) During this research project I worked with three different research assistants, but only with one at any given time. In the first two phases of fieldwork I was with my first assistant, in the third fieldwork trip I worked with both the second and third research assistants, and in the fourth and final phase of fieldwork I worked only with the second research assistant. Therefore, the SRIR method was initially designed for pairs of researchers. However, the method is sufficiently flexible to allow for the incorporation of larger research teams and even to include research subjects in the collection and analysis of data in a participatory mode. For more on expanding and developing the SRIR method see Section 4. <back>

4) In the case of this research project, different FIOs were created for borrowing and non-borrowing farming households, microenterprises, local officials and financial institutions. <back>

5) For more on grounded theory coding, see CHARMAZ (2006), GLASER and STRAUSS (1967). For an overview of how to use NVivo to code and analyse data, see BAZELEY and JACKSON (2013). <back>

6) RRA further developed into participatory rural appraisal (PRA) and other participatory approaches (CHAMBERS, 2008). <back>

Atkinson, Paul & Hammersley, Martyn (2007). Ethnography: Principles in practice. London: Routledge.

Bazeley, Patricia & Jackson, Kristi (2013). Qualitative data analysis with NVivo (2nd ed.). London: Sage.

Beekhuyzen, Jenine (2008). Conducting a literature review: A puzzling task. Australian Association for Research in Education Annual Conference, Brisbane, Australia, 2008, http://www.aare.edu.au/data/publications/2008/bee08127.pdf [Accessed: May 11, 2017].

Bröer, Christian; Moerman, Gerben; Wester, Johan Casper; Malamud, Liza Rubinstein; Schmidt, Lianne; Stoopendaal, Annemiek, Kruiderink, Nynke; Hansen, Christina & Sjølie, Hege (2016). Open online research: Developing software and method for collaborative interpretation. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 17(3), Art. 2, http://nbn-resolving.de/urn:nbn:de:0114-fqs160327 [Accessed: May 11, 2017].

Bryman, Alan (2008). Social research methods (3rd ed.). Oxford: Oxford University Press.

Chambers, Robert & Loubere, Nicholas (2017). Liberating development inquiry: Freedom, openness and participation in fieldwork. In Gordon Crawford, Lena J. Kruckenberg, Nicholas Loubere & Rosemary Morgan (Eds.), Understanding global development research: Fieldwork issues, experiences and reflections (pp.27-46). London: Sage.

Charmaz, Kathy (2006). Constructing grounded theory: A practical guide through qualitative analysis. London: Sage.

Chi, Michelene T.H. (1997). Quantifying qualitative analyses of verbal data: A practical guide. Journal of the Learning Sciences, 6(3), 271-315.

Creswell, John W. (2007). Qualitative inquiry & research design: Choosing among five approaches. London: Sage.

DiCicco-Bloom, Barbara & Crabtree, Benjamin F. (2006). The qualitative research interview. Medical Education, 40(4), 314-321.

Dresing, Thorsten; Pehl, Thorsten & Lombardo, Claudia (2008). Speech recognition software—An improvement to the transcription process? Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 9(2), Art. 17, http://nbn-resolving.de/urn:nbn:de:0114-fqs0802174 [Accessed: May 11, 2017].

Evers, Jeanine C. (2011). From the past into the future. How technological developments change our ways of data collection, transcription and analysis. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 12(1), Art. 38, http://nbn-resolving.de/urn:nbn:de:0114-fqs1101381 [Accessed: May 11, 2017].

Fasick, Frank A. (1977). Some uses of untranscribed tape recordings in survey research. The Public Opinion Quarterly, 41(4), 549-552.

Glaser, Barney G. (2001). The grounded theory perspective: Conceptualization contrasted with description. Mill Valley, CA: Sociology Press.

Glaser, Barney G. & Strauss, Anselm L. (1967). The discovery of grounded theory: Strategies for qualitative research. Chicago, IL: Aldine.

Green, Judith; Franquiz, Maria & Dixon, Carol (1997). The myth of the objective transcript: Transcribing as a situated act. TESOL Quarterly, 31(1), 172-176.

Halcomb, Elizabeth J. & Davidson, Patricia M. (2006). Is verbatim transcription of interview data always necessary? Applied Nursing Research, 19(1), 38-42.

Holland, Jeremy (Ed.) (2013). Who counts?: The power of participatory statistics. Rugby: Practical Action Publishing.

Lapadat, Judith C. (2000). Problematizing transcription: Purpose, paradigm and quality. International Journal of Social Research Methodology, 3(3), 203-219.

Markle, D. Thomas; West, Richard Edward & Rich, Peter J. (2011). Beyond transcription: Technology, change, and refinement of method. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 12(3), Art. 21, http://nbn-resolving.de/urn:nbn:de:0114-fqs1103216 [Accessed: May 11, 2017].

McLellan, Eleanor; MacQueen, Kathleen M. & Neidig, Judith L. (2003). Beyond the qualitative interview: Data preparation and transcription. Field Methods, 15(1), 63-84.

Mishler, Elliot G. (2003). Representing discourse: The rhetoric of transcription. In Nigel Fielding (Ed.), Interviewing (pp.12–37). London: Sage.

Ochs, Elinor (1979). Transcription as theory. In Elinor Ochs & Bambi B. Schieffelin (Eds.), Developmental pragmatics (pp.43-72). New York: Academic Press.

Oliver, Daniel G.; Serovich, Julianne M. & Mason, Tina L. (2005). Constraints and opportunities with interview transcription: Towards reflection in qualitative research. Social Forces, 84(2), 1273-1289.

Poland, Blake D. (1995). Transcription quality as an aspect of rigor in qualitative research. Qualitative Inquiry, 1(3), 290-310.

Shaffer, Paul (2013). Q-squared: Combining qualitative and quantitative approaches in poverty analysis. Oxford: Oxford University Press.

Skukauskaite, Audra (2012). Transparency in transcribing: Making visible theoretical bases impacting knowledge construction from open-ended interview records. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 13(1), Art. 14, http://nbn-resolving.de/urn:nbn:de:0114-fqs1201146 [Accessed: May 11, 2017].

Thunø, Mette (2006). In the "field" together: Potentials and pitfalls in collaborative research. In Maria Heimer & Stig Thøgersen (Eds.), Doing fieldwork in China (pp.245-261). Honolulu: University of Hawaii Press.

Veeck, Gregory (2001). Talk is cheap: Cultural and linguistic fluency during field research. Geographical Review, 91(1-2), 34-40.

Werner, Oswald (1996). Notation for transcribing conversations and interviews. Field Methods, 8(1), 11-12.

Werner, Oswald (1998). Short take 24: Do we need standards for ethnography? Field Methods, 10(1), 1-3.

Werner, Oswald (1999). When recording is impossible. Field Methods, 11(1), 71-76.

Nicholas LOUBERE is an associate senior lecturer in the study of Modern China at the Centre for East and South-East Asian Studies, Lund University. His research examines socioeconomic transformation in contemporary China, with a particular focus on microcredit and migration.

Centre for East and South-East Asian Studies
Lund University
Sölvegatan 18, 223 70, Lund, Sweden

Loubere, Nicholas (2017). Questioning Transcription: The Case for the Systematic and Reflexive Interviewing and Reporting (SRIR) Method [46 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 18(2), Art. 15,
http://nbn-resolving.de/urn:nbn:de:0114-fqs1702152.

Forum Qualitative Sozialforschung / Forum: Qualitative Social Research (FQS)

ISSN 1438-5627

Creative Commons Attribution 4.0 International License

	Quantitative Surveys	Verbatim Transcription	SRIR Method
Moment of Reduction	Before fieldwork	After fieldwork	During fieldwork
Method of Reduction	The creation of inflexible questionnaires that limit the type and quantity of information that will be collected during fieldwork	The transformation of the recorded interview into text, thus stripping out non-verbal modes of communication, as well as empirical observation of the interview experience	Systematic and reflexive dialogue between those involved in the research resulting in a constant and iterative process of determining where research should focus and what kinds of data should be collected
Type of Knowledge Produced	Targeted answers to specific questions	Interpretation of the exact and literal words that are spoken during formal recorded interviews	Holistic interpretation of the fieldwork experience as a whole underpinned by a comprehensive audit trail of all stages of the research process