Abstract
Transcription is a practice central to qualitative research, yet the literature that addresses transcription presents it as taken for granted in qualitative studies. In this article the author provides a review of three decades of literature on transcription between 1979 and 2009. The review establishes core understandings and issues that have informed the transcription literature, including the ways it is said that transcription is overlooked in qualitative research. Discussion of the literature raises the need for more empirical studies that examine transcription in qualitative research, and suggests specific questions that qualitative researchers might address in relation to transcription and its reporting.
Introduction
Much of the literature that has examined transcription has drawn on a seminal work by Ochs (1979) in which she proposed, and demonstrated, that transcription was theoretical in nature. Ochs wrote from the perspective of child language studies, but her consideration of transcription addressed issues that had broader implications for researchers employing transcription. She illustrated, for example, how notation of talk and interaction needed to vary to meet specific goals of individual studies. Her central claim that “transcription is a selective process reflecting theoretical goals and definitions” (p. 44) still stands as unrefuted (Duranti, 2007).
Although Ochs (1979) put transcription matters “on the table” three decades ago, researchers who have addressed transcription in the literature have continued to assert that inadequate attention has been given to transcription in qualitative research (Oliver, Serovich, & Mason, 2005; Tilley, 2003b). This neglect is evident in the underdiscussion of aspects of transcription by researchers, the lack of empirical accounts of transcription (Lapadat & Lindsay, 1999), and inattention given to the problematic nature of transcription in research reports (Tilley, 2003a) and in the training of researchers (Bird, 2005). An important consequence is that many qualitative researchers naturalize what is an interpretive process (Duranti, 2007) and present transcripts as transparent rather than the result of a series of choices in need of explication (Ochs, 1979).
The purpose of this article is to consider transcription in relation to qualitative research through a review of the transcription literature produced since Ochs's (1979) groundbreaking consideration of transcription. The review is organized into three broad areas: how transcription is defined and understood, how transcription is conducted, and how transcription is reported in research studies. Each section is used to outline issues raised in the literature and to examine how issues have been addressed or might be addressed in the future. I conclude that the literature provides a plethora of claims about the inattention given to transcription and suggests wide scope for empirical studies of the transcription process, including how qualitative researchers understand and practice transcription and how transcription has been reported in qualitative inquiry.
Method
The review addressed two research questions: What are key understandings and issues evident in the transcription literature between 1979 and 2009? and What are the implications of these for qualitative research? Data collection was bounded chronologically and consisted of searches of the literature that encompassed journals in the field of linguistics and its subdisciplines (e.g., psycholinguistics, sociolinguistics), journals of applied studies (such as Language and Education, Qualitative Health Research), journals for the social sciences where language is a focus (e.g., Language and Social Interaction), and key qualitative research journals (e.g., International Journal of Qualitative Methods, Qualitative Inquiry). Two major published reviews of the literature (Baker, 1997; Lapadat & Lindsay, 1999) and reference lists from relevant journal articles were used to locate further materials. Books specifically addressing transcription were encompassed in the review, and some handbooks about qualitative research were also examined when the literature suggested that transcription was underaddressed in these. The published literature encompasses more than 46 journal articles and 30 book chapters or sections from books.
The analysis of the literature began with a first reading of the literature to determine the problems identified as a focus for research, the approach to addressing the problem (conceptual, empirical, etc.), the findings of studies, and conclusions reached. This information was recorded in a table in which the literature was organized chronologically and updated as new materials were identified.
Materials were then read numerous times and gradually coded. Coding was usually of sections of articles. Some initial categories were employed to do this, but others emerged during the process. As analysis proceeded, superordinate categories were developed. For example, sections of some articles about transcription and computers were coded Computers. Later these were allocated to Transcription and Technology. The analytic process was an iterative one, so the table was updated when new materials were identified, some articles were coded after the superordinate categories were developed, and some categories were shifted when they appeared to have a better fit in another superordinate category. Finally, superordinate categories were organized into the three broad areas that shape the review.
How transcription is defined and understood
The transcription literature shows increasing points of agreement about how transcription is understood. In common are views of transcription as a process that is theoretical, selective, interpretive, and representational. Important differences in the literature relate to distinctive theoretical and methodological positions about how transcription should represent language and how researchers approach transcribing language to understand the world. Although claims that transcription is taken for granted are threaded throughout the literature, recent literature increasingly raises the issue in relation to qualitative research.
Shared understandings about transcription
The literature on transcription provides many definitions of transcription. All illustrate a central claim in the literature: that transcription is theoretical in nature (Ochs, 1979, 1999). For example, conversation analysts have defined transcription as situated practice (Mondada, 2007) that provides accounts of a social and moral order (Baker, 1997). From the perspective of linguistic anthropology, transcription can be viewed as cultural practice or cultural activity and transcripts as artifacts that possess “temporal-historical dimensions” (Duranti, 2007, p. 302). From within sociolinguistics, transcription has been defined as a political act (Green, Franquiz, & Dixon, 1997) whereby the recording of speech “reflects transcribers' analytic or political bias and shapes the interpretation and evaluation of speakers, relationships and contexts depicted in the transcript” (Jaffe, 2000, p. 500).
Further, transcription is understood to reflect theory and to shape it (Du Bois, 1991) as researchers “reflexively document and affirm theoretical positions” (Mischler, 1991, p. 271) during the process of transcription and analysis. Ochs (1979) illustrated this reflexive relationship through an examination of how transcription of timing phenomena in children's talk provides detailed information about their linguistic competence and leads to new understandings of what that competence entails. Hepburn (2004) set out to deliberately illustrate “the value of producing a detailed enough transcript to allow exploration of crying as a topic of analysis” (p. 252) from the perspective of conversation analysis (CA). Hepburn's study established how only through the production of a detailed transcript of crying can its interactional features be described and explicated.
Transcription is also considered to be a representational process (Bucholtz, 2000; Green et al., 1997) that encompasses
what is represented in the transcript (e.g., talk, time, nonverbal actions, speaker/hearer relationships, physical orientation, multiple languages, translations); who is representing whom, in what ways, for what purpose, and with what outcome; and how analysts position themselves and their participants in their representations of form, content, and action. (Green et al., 1997, p. 173)
This elaboration reminds that transcription is not merely the mechanical selection and application of notation symbols. Instead, researchers make choices (Kvale, 1996), and these represent some actions, in certain ways. Choices are integrally related to theoretical positions and how researchers locate themselves and others in the research process (Jaffe, 2007). Bucholtz (2000) established how choices about representation of talk in transcripts encompass power relations, a phenomenon that has been examined in great detail in the sociolinguistic literature related to transcription (see, for example, Jaffe, 2000). Questions of identity and representation in relation to transcription are also evident in some earlier and influential discussions of transcription (see Jefferson, 1996; Ochs, 1979; Preston, 1982). Baker's (1997) review of the transcription literature in literacy research articulates specific issues in representing teachers and students in transcripts of classroom interaction.
Transcription entails a translation (Slembrouck, 2007; ten Have, 2007) or transformation of sound/image from recordings to text (Duranti, 2007). The process is a selective one whereby certain phenomena or features of talk and interaction are transcribed. Rather than being a problem to overcome, selectivity needs also to be understood as a practical and theoretical necessity (Cook, 1990; Duranti, 1997). Because it is impossible to record all features of talk and interaction from recordings, all transcripts are selective in one way or another. Selectivity needs to be acknowledged and explained in relation to the goals of a study rather than taken to be unremarkable. As Ochs (1979) put it, “A more useful transcript is a more selective one” (p. 44) as extraneous information makes a transcript difficult to read and might obscure the research purpose.
Transcription that encompasses translation from one language to another presents an especially complex and challenging situation. It might require the use of interpreters, for example, and transcribers other than the researcher if the researcher is not a native speaker of the language used by research participants (Moerman, 1996). The use of more than one language in a transcript has implications for its layout (see Duranti, 1997, for detailed discussion) as well as raises issues of representation and power (Bucholtz, 2007b). Some aspects of the latter are explored in detail in Vigouroux's (2007) examination of an ethnographer's approach to transcribing with research participants when the language of the participants is not the first language of the researcher. The study provides a rich account that is presented using dual transcripts in French and in English. Riessman's (2006) account of interviews and transcriptions illustrates the cultural considerations that come into play when interviewers are translators for the researcher and transcribers. Slembrouck (2007) has asserted that “the question of translation-of/in-transcription can be expected to become even more central to discourse and social science research” (p. 825). She noted multilingual complexities that provide particular challenges for representation in transcripts including overlap and aspects of prosody.
Conceptualizing points of difference
A number of researchers have sought to develop conceptual frameworks to address the range of approaches to transcription. Bucholtz (2000) considered transcription using a continuum that presents two extremes in the range of transcription practices. She termed these naturalized and denaturalized transcription. Naturalized transcription occurs when written features of discourse have primacy over the oral, so written down talk exhibits many features of written language that do not actually occur in spoken talk. For example, commas, full stops (periods), and paragraphing are incorporated. Bucholtz referred to naturalized transcription practices as literacized. Denaturalized transcription preserves the features of oral language such as “ums” and “ers.” Bucholtz highlighted that the more a transcript retains the features of spoken language, “the less transparent it becomes for readers unaccustomed to encountering oral features in written texts” (p. 1461). Readers unused to denaturalized transcription practices might find the transcripts odd looking and difficult to read.
Oliver et al. (2005) posited a continuum to encompass the range of transcription practices. The continuum has naturalism at one end and denaturalism at the other. For these researchers, naturalism describes transcription practices that seek to provide “as much detail as possible” and denaturalism as practices where “idiosyncratic elements of speech (e.g., stutters, pauses, nonverbal, involuntary vocalizations) are removed” (pp. 1273–1274). Naturalism and denaturalism are said to “correspond to certain views about the representation of language” (1274). In a naturalized approach to transcription “language represents the real world” (p. 1274). In a denaturalized approach, the view is that “within speech are meanings and perceptions that construct our reality” (p. 1274). A number of analytic methods are discussed in relation to these two kinds of transcription practices. For example, it is suggested that denaturalized transcripts are suited to methodologies such as grounded theory and critical discourse analysis (Oliver et al., 2005). CA is viewed as a naturalized approach. Oliver et al. argued that researchers might end up with transcripts that do not match their research objectives if they fail to think about transcription style before beginning transcription of recorded data (p. 1274).
Lapadat (2000) related differing approaches to transcription to epistemological assumptions and related research paradigms. According to her, positivist views are reflected in approaches to transcription which take transcripts to be transparent, and transcription to be a manual task that produces an accurate rendering of recordings. The assumption underlying positivist approaches is that talk can be objectively presented (Green et al., 1997). Interpretivist perspectives take transcripts to be “theoretical constructions” (Lapadat, 2000, p. 208) and transcription to be a representational and interpretive process (Green et al., 1997; Mischler, 1991) whereby researchers make choices about what to record, and how, in transcripts. Approaches in between are the “muddle in the middle” (Lapadat, 2000, p. 207), which encompasses approaches where transcription is ignored entirely to the practices of CA where a standardized approach is employed rather rigorously.
The conceptions outlined here are useful for understanding distinctiveness in relation to transcription, particularly within qualitative research where numerous methodologies are employed, so there is a vast difference between CA and critical discourse analysis, for example, although researchers from both perspectives might develop and analyze transcripts. In the literature, differences about transcription play out further in relation to what is considered to be data and what is analysis. For some researchers transcription is a fundamental analytic tool (Bucholtz & Du Bois, 2006; ten Have, 2007), analytic procedure (Hutchby & Wooffitt, 1998), or “form of analysis” (Duranti, 1997, p. 137). For others, analysis begins after the development of a transcript (see Ochs, 1979, for example). Some accounts of transcription take research recordings to be data (Coates & Thornborrow, 1999; Mondada, 2007), whereas others view transcripts as data (see Johnson, 2000; Ochs, 1979).
From standardization to reflectivity
In a review of the transcription in qualitative inquiry, Lapadat and Lindsay traced the emergence of a “methodological literature” (1999, p. 65), which was limited largely to discourse analysis, conversation analysis, and speech pathology practice. They characterized the literature as evincing a shift from concerns for standardization to interpretive positionings (p. 64); however, their comprehensive review informed a critique that called for close examination of transcription in the research process as the literature was still dominated by narrowly focused accounts of transcription.
Much of the early literature about transcription focused on the need for standardized approaches to transcription within disciplines such as psycholinguistics (in the study of child language development) and sociolinguistics (use of phonetics to represent dialects of English, for example). In the case of child language development, researchers established how transcription influenced findings about age of acquisition of specific features of language (Bloom, 1993) and the analysis of language data in computer databases (Edwards, 1993b). Ochs presented a basic notation for child language studies, and those who followed her argued the importance of a standard approach to establish age of acquisition of features of talk used by very young children.
The literature provides detailed information about some central notation systems such as Jefferson notation (Atkinson & Heritage, 1999), the discourse transcription of Du Bois (1991), HIAT (Erlich, 1993), Gumperz's transcription system (Gumperz & Berenz, 1993). O'Connell and Kowal (1994, 1999) provided examinations and comparisons of a number of these systems in relation to their “usefulness and adequacy” (p. 82). An edited book by Edwards and Lampert (1993) presented a number of notation systems within the broad context of research that seeks to transcribe and then code data. A number of these authors have written from the perspective of child language, where the challenges of recording speech produced by very young children raise particular issues in relation to recording utterances.
Although the argument has been made for one standardized approach to transcription (Du Bois, Schuetze-Coburn, Cumming, & Paolino, 1993), the varying theoretical and methodological perspectives of researchers mean that they develop transcripts that necessarily differ from those of others (Lapadat, 2000). Early calls for standardization in areas such as child language research were generally replaced by an emphasis on the necessity to employ principles for notation systems (Bloom, 1993; du Bois, 1991; Edwards, 1993a, 1993b). For example, Edwards set out a comprehensive set of principles that illustrate important considerations for employing specific symbols in notation. Discussions of principles or guidelines for notation systems suggest that researchers, especially neophytes, should think very carefully before embarking on developing a transcript using an ad hoc transcription system (see, for example, Du Bois, 1991; Műller & Damico, 2002).
According to Duranti (2007) most researchers do develop a hybrid system of transcription. Apart from acknowledgement of some well-known researchers who use their own systems (Cook, 1990), there appears to be no research cited in the literature that examines how widespread this practice is among researchers in general. Although developing a hybrid system might be a theoretical and methodological necessity (Blommaert, 2007), it is not without its challenges (Bucholtz, 2007a), in which case, guidelines (O'Connell & Kowal, 1999) and procedures might be useful for researchers because at least they illustrate the complexity required to develop notations systems that serve the purpose of specific studies.
The explicit consideration of specific systems and principles for developing these sits in contrast with more recent reflective accounts of transcription. These remind that “rather than seeking standards and conventions, interpretive researchers rely on critical reflection and contextualized negotiation of method” (Lapadat, 2000, p. 210). Bird (2005) documented changes in her assumptions about transcription during postgraduate research. Tilley (2003a, 2003b) provided understandings of her take on transcription from the perspective of critical feminist methodology. A number of researchers have concluded that there are benefits for researchers in engaging “afresh” with transcription through reflection (Slembrouck, 2007), even when drawing on a specific notation system or being guided by established transcription practices. Reflections should focus on various aspects of the transcription process as it proceeds rather than being a reinventing of the wheel (Bucholtz, 2007b). Accounts of this kind provide useful insights into the experience of transcribing or learning to transcribe; they do not seek to provide methodical guides.
Overlooking the relationship between theory and transcription
Although the literature on transcription provides broad agreement (Baker, 1997) on the interpretive and constructed nature of transcripts, transcription continues to be overlooked by those analyzing language data in the social sciences (Bird, 2005; Kvale, 1996). Specifically, it is claimed in the transcription literature that the relationship between theoretical assumptions of a study and transcript development are frequently left implicit or ignored. Lapadat and Lindsay (1998) notably asserted that when researchers do not explain the theoretical perspectives that inform transcript development or address choices made in transcription, they assume a “default position” whereby they appear to take transcripts as transparent, thus inadvertently present a positivist position on transcription by omission. Lapadat and Lindsay (1998) illustrated this claim in a small study of how postgraduate students approached transcript development in a coursework assignment that required transcription of recorded data. The researchers found that the students took the transcription process for granted, with the result that transcripts were understood not as constructions but, rather, as providing “objective” accounts of recorded data.
The literature shows an increase in articles written from the perspective of qualitative research since Lapadat and Lindsay (1999) produced their conceptual review of the transcription literature. Largely, though, there is an absence of empirical studies that address how transcription is understood by researchers from within qualitative research, how qualitative researchers relate transcription to their theoretical approaches in specific research projects, or how failure to relate transcription to theoretical assumptions impacts on the achievement of research goals in qualitative research.
How transcription is conducted
In this section I address some of the “nuts and bolts” practices that are integral to transcription. These include how to develop a transcript, working with transcribers, and using technology in the transcription process.
Descriptions of how to approach transcription
The literature provides some accounts of how to approach doing transcription. ten Have (2007) has described how to approach the development of a transcript in conversation analysis (CA). He suggested that it is done best in rounds, with a different feature of talk focused on each time until all specific features of interest in talk are systematically recorded. One should begin with what has been said and then address how words were said (record intonation, gaps in talk, and so on). ten Have's account provides an approach to systematically managing the encoding of numerous features of talk that are of interest.
From the perspective of CA, analysts should develop transcripts by moving back and forth between recordings and transcripts (Ashmore & Reed, 2000; Pomerantz & Fehr, 1997).
Researchers prefer to develop their own transcripts as transcription is regarded as analysis within the perspective (Hutchby & Wooffitt, 1998). They prefer also to record information made salient within the recordings, although arguments have been made in the literature for the use of information from field notes (Hamo, Blunt-Kulka, & Hacohen, 2004). On the whole, the process of transcription proceeds in tandem with repeated examination of recorded data (Silverman, 1998). Transcripts might alter as the analytic process progresses (Mondada, 2007) and different interactional phenomena become of interest. Ashmore and Reed (2000) provided a valuable critique of the CA perspective on the relationship between recordings and transcripts, and Lapadat and Lindsay (1999) provided an outline of some of the criticisms of what is taken to be context, and thus represented, in transcripts developed by conversation analysts.
The transcription system used in CA was developed by Jefferson (1985, 1996). It is employed universally by those working from the CA perspective, has been very influential on approaches to transcription more broadly (Gumperz & Berenz, 1993; Ochs, 1979), and is regarded as having become “a near-globalized set of instructions for transcription” (Slembrouck, 2007, p. 823). According to Hutchby and Wooffitt (1998), a CA transcript “embodies in its format and in the phenomena it marks out the analytic concerns” that drive the work of conversation analysts (p. 76). Transcript notation encompasses two types of concerns; the dynamics of turn-taking and the characteristics of speech delivery (p. 76). Jefferson notation encompasses symbols to represent aspects of each. As well, researchers might develop additional symbols where examination of a certain phenomenon requires it (2001; ten Have, 2007) but employ similar ways of presenting transcripts in the reporting of studies (Psathas, 1995). Psathas and Anderson (1990) and Atkinson and Heritage (1999) have provided detailed outlines of the notation system.
Gee, Michaels, and O'Connor (1992) have also provided a discussion and example of how transcription might be approached. They examined two transcripts of the same data and outlined how each was constructed. The approach to transcription is informed by ethnopoetics and is also outlined in further detail by Gee (1999). The researchers' suggested one way to go about transcription from the perspective of discourse analysis. They also provided commentary in relation to how different transcripts of the same text result in different analyses to suit the research purpose of each analysis. Rapley (2007) used a number of transcriptions of the same recording to show how transcription can be approached using levels of detail that fit with research purposes.
Stelma and Cameron (2007) offered a very detailed, reflective account of the transcription process and of the experience of developing transcription skills. In particular, the article focuses on the recognition and transcription of intonation units in spoken data, and on achieving agreement between transcribers (the authors in this case). How disagreement is addressed is the focus for substantial discussion. Comparisons between transcripts of the same recorded data illustrate how transcribers hear “differently.”
Because data can be transcribed in many ways (Coates & Thornborrow, 1999), researchers need to think about transcription carefully before beginning the development of a transcript (Lapadat, 2000). Lapadat described her approach to the development of three transcripts, showing how differing research purposes result in a different approach to transcription in each case. Rather than providing a single approach to doing transcription, her point is to illustrate that decisions need to be made about transcription before it begins. In this way, researchers are able to produce a transcript that enables research questions to be addressed through the application of an approach to transcription that is suited to the needs of the specific study.
For some researchers, transcription will be a continual process of revision (Coates & Thornborrow, 1999); that is, transcripts can be reworked for particular research purposes, resulting in transcripts of the same recording that might differ markedly when produced for different purposes (ten Have, 2007). Riessman's (2006) explication of two different versions of a research interview illustrates how researchers might deliberately seek different representations of the same data to construct research participants in particular ways. Mondada described the changes that occur to transcripts over time and pointed out,
These changes are not simply cumulative steps towards an increasingly better transcript: they can involve adding but also subtracting details for the purposes of a specific analysis, of a particular recipient-oriented presentation, or of compliance with editorial constraints. (p. 810)
Mondada (2007) proposed a distinction between working transcripts and edited transcripts, whereby the end purpose for transcription (such as publication) might result in transcripts that differ from those used for the purpose of analysis during the research process. Luebs (1996) examined use of transcription in publications as an aspect of her doctoral research; in her dissertation she presented and analyzed the views of researchers and publishers.
What is especially notable about many of the above accounts of doing transcription is that they have received very little acknowledgement in the numerous handbooks that address research methods and various methodologies in qualitative research. Qualitative research handbooks in general provide little acknowledgement of the complexity of transcription (Halcomb & Davidson, 2006) and rarely cite the literature or address the theoretical, methodological, or practical issues pertaining to transcription.
Working with transcribers
Researchers' use of hired transcribers is integral to the production of transcripts in many qualitative research projects. Tilley and Powick (2002) have examined the use of hired transcribers, particularly in relation to the trustworthiness of transcripts and their analysis. This qualitative study is one of only a few that provides empirical accounts of transcription (see also Lapadat & Lindsay, 1998). Tilley and Powick examined transcripts of interviews with hired transcribers and drew out a number of issues. These include absence of direction given by researchers to hired transcribers, researchers' use of transcripts rather than returning to data, transcribers' omission or alteration of words when transcribing, and ethical matters related to confidentiality agreements for hired transcribers.
Suggestions for avoiding transcriptionist errors or corrections have been provided by MacLean, Myer, and Estable (2004). These include spot-checking of transcripts as they develop and the provision of a specific notation system and examples for transcribers to use as a guide. Employing professional transcribers is one way to address trustworthiness (Dressler & Kreuz, 2000), although transcribers need instruction in relation to the specific purposes of individual studies and the transcription requirements of these. McLellan, MacQueens, and Neidig (2003) outlined a specific transcription protocol used to ensure consistency in transcripts developed for qualitative analysis employing computers. They suggested also that a proofreader is useful but needs to be familiar with the transcription protocol, the research topic and terminology, and “the vernacular used by interviewees” (p. 72).
Transcribers might also have notions of what a “good” transcript looks like, and these might reside in notions of (written) correctness (Tilley & Powick, 2002). Grundy, Pollon, and McGinn (2003) have considered the contribution that the use of research participants can make to the trustworthiness of studies and the effectiveness of the transcription process.
Transcription and technology
How transcription is done has always had an integral relationship with the technology available to record data and to transcribe it. Early in the past century, anthropologists recorded talk of informants directly onto paper. This had a number of associated difficulties, not the least of which was the need for informants to speak slowly for their talk to be recorded (Duranti, 1997). The advent of audio recording devices made it possible to review language data many times over to produce transcripts of naturally produced language (Sacks, 1995). The shift from audio recordings to video meant an increased availability of information to record and more choices that needed to be made about what to record (Lapadat, 2000). Computer technology has made available the possibility of large databases at the same time as it has made important the standardization of notation across these (Bloom, 1993; Edwards, 2001; MacWhinney & Snow, 1992). Use of computer software enables effective management of qualitative data but requires protocols that guide the systematic development of transcripts (McLellan et al., 2003). Most recently, a small number of researchers have written about transcription in relation to multimodality due to the advent of digitalized technology that makes it possible to view recordings and develop transcripts on the computer screen using software programs developed specifically for transcription. Examination of programs such as Transana and Clan (Mondada, 2007), for example, have been used to illustrate how technology alters the transcription process and challenges the notion of “transcript.”
Some researchers have shown how image and written text can be combined to produce new forms for transcripts. Norris (2002) provided an account of transcription in her study of children's use of television and computers during play dates in homes to argue and illustrate that the shift to transcription of action other than language has important implications for transcription. Specifically, it makes relevant, and possible, multimodal transcripts that encompass images recorded on screens (such as computers and video monitors) and foreground these over language. Norris argued, from the perspective of multimodal discourse analysis, that transcription conventions developed for the tape recorder are not best suited for video camera data. Instead, Norris (2004) has developed an alternative approach to transcription, which she describes and illustrates step-by-step using images from her own research. Plowman and Stephen (2008) have used Comic Life software to produce comic strips that encompass video stills and speech bubbles to represent selected recordings of interaction
Numerous possibilities and constraints on transcription are integrally related to computer technology and its development (Du Bois, 1991; Zukow, 1982). Recently, the literature shows reference to the affordances that technological developments present in relation to addressing transcription problems through the development of alternative methods for presenting data. Whereas some researchers have looked to digital technologies to replace transcripts (e.g., Kvale, 1996; Slembrouk, 2007), others suggest that the affordances of new technologies come hand in glove with new complexities for transcription (Mondada, 2007). A challenging issue is whether transcription systems developed for “old technologies” can effectively represent data recorded with new technologies (Norris, 2002).
How transcription is reported in research studies
This aspect of the literature targets the perceived neglect of researchers to address transcription in written reports of studies. This absence is made salient in relation to the reporting of qualitative studies in final reports and in journal articles that arise out of studies. Trustworthiness of qualitative studies is raised as a question when transcription is overlooked. Reflection on the research process is emphasized and commentary about transcription viewed as essential in the reporting of research.
Reporting on issues of transcription quality and trustworthiness
Transcription and transcript development are neglected in written reports of research (Duranti, 2007; Lapadat & Lindsay, 1999). As mentioned previously, Lapadat and Lindsay (1998) asserted that transcripts are frequently left to stand alone; that is, they are presented in qualitative research reports will little or no explication about the theoretical perspectives and methods used to develop them through the process of transcription. Instead, Lapadat (2000) proposed that
transcription decisions and processes employed during data collection and analysis need to be explained clearly and thoroughly in the write-up. When standardized procedures are used, a few words will suffice, but when researchers contextualize and negotiate method as a means of interpretive seeing, there is no shortcut to explicit description. (p. 217)
Although other researchers have concluded similarly that transcription is not addressed sufficiently in reports of research (Tilley & Powick, 2002; Tilley, 2003a), few studies appear to address this aspect of reporting qualitative research. Wellard and McKenna (2001) examined 42 articles reporting qualitative research published in one year in journals in the field of nursing. They analyzed methods of transcription, details of transcription process, and use of citations to support the approach taken and found that 66% of the articles provided no detail other than to report that interview data were transcribed. In the remaining articles, either verbatim or full was used to label the transcription process, but these terms were not defined or explained.
Issues of transcription quality and trustworthiness are central to transcription within qualitative approaches to research as transcripts are used not only for analysis but as evidence of that analysis (Duranti, 2007) and the researcher's analytic claims (Ashmore & Reed, 2000). It would appear that it is particularly in relation to addressing trustworthiness that qualitative researchers must be held accountable for their approaches to transcription and the transcript that results. Specifically, researchers should note and query the trustworthiness of implicit accounts of transcript development (Easton, McComish, & Greenburg, 2000; Lapadat, 2000).
Poland (1995) has provided a typology of errors that can occur in transcription and suggested the need for methods sections of research reports to provide
steps taken to ensure audiotape quality, the directions provided to transcribers, and an assessment of the trustworthiness of transcription, based on a review of selected transcripts in the context of an explicit acknowledgement of the interpretive nature of the transcription process. (p. 306)
Lapadat (2000) addressed transcription in relation to issues of quality and provided some general considerations to address it. These include thinking before beginning to transcribe, including in relation to recording; having clarity of purpose; “doing quality transcription” (p. 215), including through the accounts provided of these in write-ups.
Addressing transcription unevenly
It has been noted that researchers within certain disciplines have been far more explicit than others (O'Connell & Kowal, 1999). For example, discussion of transcription is more evident in the field of linguistics and its disciplines, such as sociolinguistics and some areas of psycholinguistics. The social sciences in general are considered to overlook transcription in research, although CA has been acknowledged throughout the literature for its influence on understandings of transcription and transcription notation (Duranti, 2007; Ochs, 1979). The dissertation work of Luebs (1996) stands alone in presenting a substantial study of transcription, including transcribers' choices and decisions about transcription and the use of transcription in journal articles. Transcribers in Luebs's study included sociolinguists, discourse analysts, and conversation analysts. She found a strong association between transcription practices in disciplines and the practices of researchers within these.
Further, it appears that transcription is discussed more in relation to some methods than others. For example, there are small number of publications about variants of ethnography and transcription (Hamo et al., 2004; Poland, 1995; Tilley, 2003b; Vigouroux, 2007), and CA researchers continue to address transcription, particularly in handbooks or chapters about the methodology (Silverman, 2000). Other researchers have written from the perspective of discourse analysis (Bucholtz, 2007b; Gee, 1999). It appears from this review that it is more difficult to find discussion of transcription in relation to other qualitative research methodologies or analytic methods. For example, Oliver et al. (2005) noted the absence of commentary about transcription in the literature related to critical discourse analysis and that “effort must be expended to find useful guidance about transcription in grounded theory” (p. 1278).
For Lapadat (2000), the problem in relation to qualitative research is this: “If we do not accept the notion of one true reality that can be uniquely recorded and fully represented in written text, how do we do and evaluate transcription?” (pp. 209–210). The literature suggests that researchers need to clarify aspects of the transcription process for themselves through reflection (Bucholtz, 2007a); make aspects of the transcription process explicit to others involved in research, such as paid transcribers (MacLean et al., 2004) and research participants (Grundy et al., 2003); and articulate the transcription process in written reports of research (Bucholtz, 2007; Lapadat, 2000).
Discussion
The transcription literature across three decades has emphasized the need for researchers to be explicit about transcription. Calls for explicitness in the accumulated body of literature now encompass numerous aspects of transcription, including its theoretical assumptions (Ochs, 1979), the relationship between theoretical perspective and selection of transcription method (Coates & Thornborrow, 1999), selectivity (Stelma & Cameron, 2007) and the “partial nature” of transcripts (Coates & Thornborrow, 1999, p. 596), choice of data conventions (Edwards, 1993b), what analysis the transcript is “good for” and how its development relates to the original recording (Duranti, 2007), and trustworthiness of transcripts and the transcription process, including steps taken to ensure quality (Poland, 1995).
The matter of why transcription remains so underaddressed in qualitative research is curious. In the literature we find various explanations: Little attention is given to it in the training of researcher students (Bird, 2005), methods texts continue to foreground data collections methods such as interview conduct (Halcomb & Davidson, 2006), the literature is spread widely across the journals of a range of disciplines, and issues of transcription in some disciplines do not interest researchers in others (Luebs, 1996). In addition, this review suggests that the literature is constituted by numerous conceptual considerations and illustrations of how and why transcription matters, repeated claims about the inattention given to transcription in qualitative inquiry, but very few empirical studies that respond to these claims. It is important to note here that the commentary regarding the taken-for-granted approach to transcription is driven by qualitative researchers and has slowly escalated in the past decade.
Although the literature asserts that transcription is taken for granted within qualitative inquiry, it might be useful if finer distinctions are made in relation to this claim. For example, it appears that that there is a paucity of studies that address, and illustrate, how transcription is approached within certain research methods (Oliver et al., 2005). Given that it can be expected that researchers will show differences in approaches to transcription, the identification of this gap suggests opportunities for researchers within a wider range of methodologies to publish detailed descriptions and explications of how transcription is approached within the methods they employ. This would contribute to the growing body of literature that shows how approaches to transcription differ and why they need to do so to fit with the diverse theoretical and methodological perspectives of researchers and the studies they design.
Further, a salient feature of the recent transcription literature is that it frequently draws on and examines transcription of interview data. Examinations of perspectives, practices, and issues related to other kinds of recorded data are needed. For example, recordings of classroom lessons present particular challenges for transcribers due to large numbers of coparticipants in the setting (Davidson, 2004). Discussion of the complexities of transcribing other kinds of multiparty talk would also make a valuable contribution to the developing body of literature, as would publications about transcription that encompasses translation, a situation particularly pertinent to publication of research articles in leading journals that publish only in English.
Finally, this review of literature suggests a plethora of questions that might inform empirical studies of transcription so as to address claims in the literature and inform our practices as qualitative researchers. How is transcription reported in journal articles, reports or theses? What accounts are provided of transcription choices in relation to theoretical perspectives on research problems, and how do choices made relate to methodological approaches to research problems addressed in studies? What transcription systems do qualitative researchers employ or do they develop their own hybrid systems? How are hybrid systems constituted and explained? What literature do qualitative researchers draw on when reporting approaches to transcription? How do qualitative researchers address the trustworthiness of their transcripts and approaches to transcription? Do researchers make choices about transcription but fail to make those choices apparent in the reporting of research? More important, does it matter? The body of literature that understands transcription as interpretive, selective and representational, overwhelmingly says yes.
Conclusions
Three decades since the groundbreaking publication by Ochs (1979), the literature about transcription now presents a substantial body of work, albeit spread across numerous fields and disciplines, and dominated largely by issues relevant to linguistic research. The literature provides a growing consolidation of illustrations of transcription as theory, as selective and partial, as representative and as interpretive. However, the impact of the literature on researchers' approaches to transcription appears to be found in the continuing claims about the absence of explicit attention to transcription in research. While this absence provides rich ground for those who write about transcription matters, it still needs to be asked why transcription continues to be overlooked to such a large degree and to determine specifically how transcription is addressed across and within studies in qualitative inquiry.