‘Is that what I Said?’ Interview Transcript Approval by Participants: An Aspect of Ethics in Qualitative Research

Abstract

This article deals with insights gained from data analysis of feedback comments on transcripts sent to interviewees. It contributes to understanding of research studies that include transcripts, specifically on the contribution of participants' review of transcripts on the quality of those transcripts, and thus on the quality of research. The transfer of the transcripts to the interviewees was intended to validate the transcripts, to preserve research ethics, and to empower the interviewees by allowing them control of what was written. Interviewee responses related to the ratification of content, the authenticity of that which was said during the interview, corrections of language, additional clarifications, power interactions and changes in the balance of power between the interviewer and interviewees, feelings of embarrassment and threat, research ethics, and reflective responses. The experience of sending the transcripts to the interviewees raised research and ethical issues that require added caution and consideration when sending transcripts to interviewees.

Background: Transcription and type of transcripts

Transcription is the transference of spoken language with its particular set of rules to the written word with a different set of rules (Kvale, 1996). This process of transforming uninterrupted dynamic oral language spoken in a particular context to a static form of representation (written language) is necessary for the management and organization of data, since only written language can be managed, in other words, sorted, copied, examined, evaluated and quoted (Lapadat, 2000).

Researchers apply two main methods of transcription: naturalized transcription and denaturalized transcription (Davidson, 2009). Naturalized transcription is a detailed and less filtered transcription. It is as detailed as possible and focuses on the details of the discourse, such as breaks in speech, laughter, mumbling, involuntary sounds, gestures, body language, etc. as well as content (see, for example, note 2). ² Denaturalized transcription is flowing, presenting ‘laundered’ data which removes the slightest socio-cultural characteristics of the data or even information that could shed light on the results of the study. It accurately describes the discourse, but limits dealing with the description of accent or involuntary sounds. The accuracy relates to the essence of the interview, the meaning and the perceptions that were created and its part in the discourse (see, for example, note 3). ³ Each method has advantages and disadvantages. In the instance of naturalized transcription, the transcriber may wrongly interpret the voices heard in the recording and in so doing influence the conclusions of the research. However, the detailed descriptions of the voices and the things mentioned during the interview may afford a more complete and valid picture of the same. Contrary to the method of naturalized transcription, the denaturalized transcription approach represents unblemished data. The transcription may, on the one hand, lose subtle socio-cultural characteristics; however, on the other hand, it may be both coherent and easy to read (Oliver, Serovich, & Mason, 2005). While some researchers wonder whether naturalized transcription provides a more reliable version of the interview as it was conducted (for example, Forbat & Henderson, 2005), Oliver and colleagues (2005) are of the opinion that most researchers use a combination of both methods.

Whether researchers use a naturalized or denaturalized method, or a combination of both, transcription is not simply a technical process (Tilley & Powick, 2002; McLellan-Lemal, 2007), and, despite their efforts, researchers find it difficult to attain through transcription the qualities afforded by the oral discourse of the interview. Researchers (for example, Alexander, 2003; Poland, 1995) admit that transcripts lack certain attributes which are characteristic of oral language production (e.g. intonation, emphasis, voice volume, changes in voice patterns and body language) that lend life to the words and add meaning. Accordingly, whether naturalized or denaturalized, neither method can adequately present these attributes (Poland, 1995).

The transcription process, despite its being a critical element in data analysis and central in qualitative research (Lapadat & Lindsay, 1999; Oliver et al., 2005) has received little attention in the research literature (Poland, 1995; Bocholtz, 2000; Forbat & Henderson, 2005; Lapadat, 2000). The researcher, at best ensures that the transcript validly represents what is said in the recording (even though the transcription cannot completely represent everything that is said) and pays little attention to the effect that transcription has on the research questions, description of the findings and conclusions of the research (Tilley, 2003). One can identify five factors that may influence the quality of a transcript: the researcher, the interviewer, the transcriber, the interviewee, and the equipment and place of transcription.

The researcher

The transcript is influenced by the researcher's attitude regarding the topic and by his or her assumptions with regard to the data (Davidson, 2009). The transcript is influenced by the background information of the research and the interviewees, which is either given or withheld from the transcribers in cases where the researcher is not the transcriber (Davidson, 2009). The researcher may include in the transcript clarifications of the potential impact of emotional content and the importance of privileged data (Kvale, 1996). The guidelines provided may include the researcher's requirements as to what to include or omit from the transcript, for example, the level of itemized associated knowledge to be included in the transcription, and whether this will be incorporated as an integral part of the transcript or as a separate document (Poland,1995; Lapadat, 2000). Differences may also arise from the guidelines given to transcribers with respect to the conventions of transcription by which the transcriber is expected to operate (representation of laughter or the lack thereof, coughing, groans, body language etc.) and the way to treat incomprehensible segments (Lapadat & Lindsay, 1999).

The interviewer

Often the interviewer is not the researcher but is a member of the research team. In this instance the quality of research will be influenced by the interviewer's knowledge of the entire research project and his or her ability to perceive the complete picture. It may also be influenced by the interaction between the interviewer and the transcriber, for instance, by the extent to which the interviewer's remarks were noted during or after the interview, the ability to assist the transcriber in comprehending the goings on, or the extent to which the interviewer verifies the transcripts with the recording and co-operates with the transcriber in handling inconsistencies.

The transcriber

As in the case of the interviewer, the transcriber is not always the researcher. Therefore the transcript is influenced by the professional competence of the transcriber, including his or her language knowledge and the characteristics of notating discourse, as well as the attention awarded the task and potential fatigue during the transcription process (Poland, 1995). The transcript is also influenced by the transcriber's awareness of the subject of the research, predisposed attitudes towards the subject, objectives, preconceptions of the interviewees, as well as the difference in class, culture and language between the transcriber and the interviewees (Maclean, Meyer, & Estable, 2004). These factors may bias the analysis of the findings (Tilley, 2003). The transcriber's choices and decisions during transcription may also influence the quality of the transcript (Poland, 1995). These are expressed in the way he or she treats background noises, overlapping discussions in the instance of more than one person (for instance in a focus group), pauses in conversation, and para-linguistic elements such as incoherent parts of a text (Lapadat, 2000) that lead to certain suppositions. The number of suppositions may also serve as an indicator of the quality of the transcription (MacLean et al.,2004).

In the current research, the researcher was the interviewer as well as the transcriber, thereby reducing compromising influences with respect to the transcript quality.

The interviewee

The quality of the transcript is influenced by the interviewee's spoken intonation (e.g. soft, hard, with a heavy dialect (Oliver et al., 2005), the ambience and the rapport created with the interviewer (a tense interviewee is likely to stammer, create long pauses during discussion and emit unintentional sounds), and his or her mental status or health, which may have an effect on voice qualities and the transcriber's understanding of the situation. Oliver and colleagues (2005) described a situation where the transcriber's preconceptions (and those of several of the researchers) regarding the observed population influenced the meanings of the voices that were heard in the recording:

…in our work with HIV positive men, the research team read a transcript where the participant's statement was continually interrupted by his sniffling, indicated in the transcript by ((sniff)). When the team met to discuss this transcript, the sniffling became confusing and the subject of some debate. Some thought the participant was crying during the interview, whereas others made assumptions about drug use. The confusion was settled when the interviewer explained that the participant was sick and his nose was running. (pp. 1276–1277)

Equipment and location of interview

The quality of the transcript is influenced by the quality of the recording. The use of inadequate or inappropriate equipment may lead to errors in the transcript (Kvale, 1996; MacLean et al., 2004; Poland, 1995). Furthermore the location of the interview can determine the quality of the recording. Recording in a noisy location is susceptible to background noises which will affect its quality. The choice of location by the interviewer or the interviewee may also contribute to the quality of the recording. A comfortable place, especially one that is chosen by the interviewee, will afford him or her feeling of comfort (McDowell, 2001), creating a pleasant atmosphere which may reduce tension. Hunter (2005) described the opposite situation in her research and “rather than constituting a comfortable trusting environment, research situations were characterized by anxiety” (p. 152).

All the above affect the quality of transcription, influencing the way in which the researchers understand the interviewees and the knowledge that the interviewees share with them, the explanation they afford the data, the conclusions reached, and by extension the entire research project (Poland, 1995; Tilley, 2003; Forbat & Henderson, 2005; Lapadat, 2000; MacLean et al., 2004; Oliver et al., 2005).

Participation of the interviewees in the transcription process

Participation of the interviewees in the transcription process is not a procedure regularly used by qualitative researchers, but the procedure is increasing being used. When the procedure is employed, there are various justifications for doing so (Hagens, Dobrow, & Chafe, 2009). One of the principle reasons for this is to ensure the validity of the transcript (Polit & Beck, 2007) and to avoid significant errors that may have an impact on the quality of the transcript and, as a result, on the quality of the entire research. In order to avoid this, the researcher needs to take measures to ensure the trustworthiness of the transcripts (Poland 1995; Davidson, 2009). One way to do this is to have the interviewees validate the transcripts by correcting them if necessary and by clarifying unclear issues.

Lapadat (2000) has maintained that giving the transcripts to the interviewees is, in addition to clarifying unclear portions, also intended to stimulate discussion on various topics mentioned in the text. Forbat and Henderson (2005) gave transcripts to interviewees in order to discuss their content during a second interview that was scheduled with them. This is based on the recognition on the part of some researchers that transcripts are not facsimiles of a particular reality, but rather interpretive structures (Poland,1995; Ross, 2009). As such, they incorporate aspects of power that arise from interpretive choices and decisions (what to transcribe?) and from representative choices and decisions (how to transcribe?). These choices testify to the fact that transcription is not objective, is not accurate, and is essentially political (Bocholtz, 2000).

Page, Samson and Crockett (2000) mention three reasons for sharing the information with the interviewees: politeness or compensation to people who donated their time to the research, validation of the data or findings, and supplying of information and recommendations that could improve conditions by empowering people. Grundy, Pollon and McGinn (2003) used interviewees as transcribers to achieve three main objectives: to overcome the researcher's hearing impediments, to create cooperation between the researcher and the participant, and to achieve quality transcriptions which accurately represented the participant-transcriber's voice. They felt that the integrity of the transcript was thus preserved, because the participant-transcriber was the one to decide whether the content represented the true intention of the transcript. Furthermore, the research was enhanced since the participant-transcriber was better able to define and clarify his or her responses. But these scholars admit that “participant-transcriber may… be tempted to modify the transcript of the interview” (p. 13), not only for the purpose of clarifying content, but to make it more articulate.

Other researchers transferred the transcripts to interviewees in order to preserve norms of etiquette (Kvale, 1996), individual wellbeing, freedom of consent, choice, and the principle of justice, decency, and equality (Inter-Agency Secretariat on Research Ethics, 2005). Saldana (1998), who operated in accordance with these principles, gave participants the scripts he wrote for their approval, since he felt that this was the right thing to do. He believed that participants must feel respected and that the researcher must protect their dignity to the best of his ability, despite the fact that the interviewees may have concerns about their representation in interviews. Accordingly, when participants requested to delete certain segments from the scripts of their lives that he had written, he acceded to their demands out of respect for their dignity and privacy.

Hagens and colleagues (2009) believe that despite sending the transcripts to the interviewees it is not completely clear what impact this may have on the qualitative research. They examined four topics: the quality of the transcriptions; the extent of improvement of the quality of the data that was derived from the interviews; the impact on the interviewee; and the impact on the researcher. They found that transferring the transcriptions to the interviewees added very little to the precision of the transcriptions. Aside from the advantages of providing the interviewee with the opportunity of clarifying information, correcting mistakes, and adding new materials, there were also biases, such as loss of valuable data that the interviewees decided to delete. The impact on the interviewees was both negative and positive; however, the impact on the researcher was negligible. Hagens and colleagues (2009) concluded that it is preferable to use other, less problematic means to verify and improve the precision of the transcriptions. They advise other researchers to carefully consider the advantages and disadvantages of using this procedure before they make a final decision.

This article contributes insights regarding research that incorporates transcripts and the influence of transcript transferral to research participants on the quality of transcripts, and, by extension, on the quality of the research.

The interviewees, the interviews and transcripts

The interviewees in the present research included 16 women who were key informants: three elementary school principals; three policymakers who have held and hold prominent positions in the Ministry of Education; seven test developers who developed performance-based tasks for testing purposes; one teacher trainer who worked with teachers on implementing the performance tasks and continued in her role for a number of years; and two supervisors responsible for assessment. All the interviewees were academics and experienced in academic writing.

Interviews ranging from 90 to 150 minutes were recorded. In addition, during the course of each interview notes were taken for the purpose of backup, as well as field notes. This procedure was expected to improve the quality of the transcript with respect to contextual aspects to overcome difficulties associated with the quality of the recording (Poland, 1995). During the transcription process the notes taken during the interview and the field notes were also referred to.

The researcher filled three functions in this study: researcher, interviewer and transcriber. This provided two advantages: first, because she had a good knowledge of the research topic, its background, the content of the research, the interviewees, and the background voices heard during the interview; and second, because of her commitment to the precision of what was said, reducing the choices and number of errors that usually transpire as a result of a transcriber's inadequate knowledge of the particular interview situation (Tilley & Powick, 2002).

The method of transcription was denaturalized; in other words, the text was rendered free of interview ‘noises,’ such as pauses in speech, coughs, moans, involuntary sounds, stutters, grammatical errors, and body language, and coherence and connection to forms of discourse and correct grammar were observed. The choice of this approach suited the characteristics of the research questions and the information being collected from the data (Bocholtz, 2000; Oliver et al., 2005). Furthermore, transference of naturalized transcripts, where the text reflects verbatim and in minute detail a description of the speech may have insulted the interviewees, who might feel that their speech was unrefined.

The transcripts were sent to the interviewees not only to uphold research ethics, but principally to validate what was said during the interviews and to ensure that the written words in the transcript were those said by the interviewees (Hagens et.al., 2009). The objective was to have the interviewees approve the printed version or discover and correct errors or inconsistencies that originated from poor recording quality. Another objective was empowerment: giving participants a feeling of propriety over the product and control over the printed word. The interviewees’ reactions did not usually reflect attainment of this goal, though in some instances their reactions hinted a shift in power between the interviewer and interviewees (discussed subsequently).

The transcripts were sent to the interviewees once all the data for the entire research project had been gathered. This occurred about a year after the interviews and was carried out in two stages. In the first stage a phone call was made to the interviewees to remind them of the interviews that had been conducted, and to request their permission to send the transcripts for their responses. All interviewees consented. They were also asked to decide on the method of transfer (via the post or email). All agreed to accept the transcripts via email and to respond in the same way. In the second stage the transcripts were sent to the interviewees with an accompanying letter wherein mention was made of the research and the date of the interview. Also, participants were advised that should they find reason to correct, clarify or make additions to the interview, they were invited to do so. Only five of the sixteen interviewees responded to the transcripts. Although they do not represent the entire group of interviewees, the responses highlighted issues that can contribute to the use of transcription, particularly regarding participation of interviewees in the transcriptions of their interviews.

Eight types of response were distinguished: written approval; relevance of content; error corrections, clarifications and additions; power relationships; sense of embarrassment and threat; from spoken to written language; research ethics; and reflections. Each of the types of response are discussed below. Hagens and colleagues (2009) identified six categories of changes in the text: specific transcription errors/omission corrected; specific details added to transcript; specific transcription details corrected/changed; grammatical changes or minor clarifications made to transcript; statements removed from transcript; statements added to transcript. These categories are included in three categories presented in the present paper: correction, clarifications, and additions; from spoken to written language; and reflection.

Written Approval

Although interviewees were asked to respond to the transcripts, some did not, even though they had initially agreed to do so. Forbat and Henderson (2005) admitted that not all the interviewees in their research responded to the transcripts. Hall (2004) noted that despite the fact that she offered to send interviewees copies of the transcripts for their response, even those that had asked to see the transcripts failed to contact her. She emphasized that this reaction does not mean that they approved the transcripts or were satisfied with them, but explained that their lack of response was due to the burden of professional obligations, or from a feeling of discomfort in responding face-to-face or over the telephone.

In the present research study two types of approval can be discerned from the responses of interviewees to whom were sent transcripts. Some expressed total approval: “This is fine as far as I'm concerned,” while others gave their approval on the condition that corrections be made to the original transcript. One interviewee wrote: “I reviewed the interview using ‘track changes,’ there are numerous corrections. I've made the corrections….I hope this is of assistance to you. Attached are my corrections.” Another interviewee wrote: “I didn't want to change it (the transcript) myself, because I wanted you to see the changes I'm suggesting…If you like, I'll insert all the corrections…myself after you (the interviewer) go over it. In both responses a request was made to institute the changes accompanied by a show of faith that the researcher would do so. As written by one interviewee: “I hope I've clarified myself and trust you with the implementation.” The trust was the result, in part, of the relationship created with the interviewees during the first interview and during the course of observations conducted on three of them, and, in part, based on previous connections.

Relevance and pertinence of content

The relevance and pertinence of what was said during the interview is “frozen in time” (Forbat & Henderson, 2005; Poland, 1995). Forbat and Henderson (2005) write, “…there might be a point at which participants indicate that they have ‘moved on’, leaving the typed conversation less meaningful, and less ‘true’ than when the words were first spoken” (p. 1118). The majority of interviewees in the research spoke during the interview of an event that had occurred in the past. Accordingly, the interval between the interview and receipt of the transcription had little meaning for them. However, five of the interviewees told of an event that had occurred in the present and was a part of their current professional lives. Two of them, who were school principals, remarked on the time that had elapsed since the interview, which they felt to be significant. According to them, during this period many changes had occurred in their schools, as a result of which some of the things mentioned in the interview were no longer relevant or had changed. One of them wrote in response to the transcript: “So much water under the bridge since the interview with you.” She added: “By and large, this was the way it was then.” Nonetheless, none of them reneged on anything said during the interview, or withdrew approval to use the information. They accepted the fact that the interview reflected the situation as it was at the time of the interview. As stated by another interviewee: “I understood just how much has changed but since we will obviously not be conducting a new interview, this is what was true at that time.”

Correction, clarifications, and additions

Transcripts are not an objective product of the interview but rather they refer to the context of a political and social character (Bocholtz, 2000; Poland, 1995). Bocholtz stated:

All transcripts take sides, enabling certain interpretations, advancing particular interests, favoring specific speakers and so on. The choices made in transcriptions link the transcript to the context in which it intended to be read…Transcripts testify to the circumstances of their creation and intended use. As long as we seek a transcription practice that is independent of its own history rather than looking closely at how transcripts operate politically, we will perpetuate the erroneous beliefs that an objective transcription is available. (p. 1440)

Accordingly, it should be anticipated that errors and inconsistencies in the transcripts (Lapadat, 2000), some of which may be significant, could alter meanings and thereby affect the entire research (Grundy et al., 2003; Poland, 1995; Hagens et al., 2009). Sharing the research with the interviewees by sending them the transcripts may enhance what was said in the interviews and ensure that transcriptions are correct. The interviewees in the present research were also busy correcting transcriptions and clarifying things that they felt were necessary. They corrected errors originating from misunderstanding what was said on the tape and even typing errors. They also corrected grammatical errors, such as replacing a period at the end of a sentence with a question mark.

Some expressions, when read again, seemed to be meaningless to the interviewees. For example, one the interviewee remarked on the expression “the workers’ holiday”: “This is a funny expression. I'm certain I didn't say it!” There were other expressions, that appeared to have meaning and were logical in the eyes of the researcher, but were perceived by the interviewees to be incoherent in the greater framework. For example, one interviewee commented: “I cannot see how this entire segment relates to the rest and it's also illogical.”

Other corrections were intended to clarify issues that at the time of the interview were clear to both interviewer and interviewee but upon reading the transcript, the interviewees felt that they were not sufficiently understood and required clarification. The findings indicate three types of clarifications. The first was clarification of insinuated terms (“this” or “it”). During the course of the interview both the interviewer and interviewee knew what it is about and therefore the usage of insinuated terms was clear to both of them. However, as Forbat and Henderson (2005) suggest, since these insinuated terms were inclusive they caused problems for the researchers reading the transcript. The transcript, no matter how explicit, cannot provide the context that was prevalent during the uninterrupted conversation. Accordingly, the researcher may have difficulty in identifying the meaning of the insinuated terms and thus interviewees felt a need for clarification. Another type of clarification was one in which complete addendums were added to incomplete ones (e.g. the source/the source of the funding). During the course of the interviews, the speakers employed abbreviated sentences that were clear to both parties. In the transcript the abbreviations, like the insinuated terms, had ambivalent meanings for them and thus distorted the meaning of the entire conversation. Accordingly, the interviewees completed missing addendums in order to avoid misunderstanding. As in the case of insinuated terms, things that were clear to the interviewer and interviewee during the interview were found to be unclear when reading the transcript. The interviewees felt a need for clarification in the form of rephrasing things that were said during the interview more clearly. This was the third type of clarification that was made.

Changes in the power relations between the interviewees and the interviewer

Several researchers (Bhopal, 1995; Fletcher, 1992; Goldman, 1993; Hall, 2004; Payne, Field, Rolls, Hawker, & Kerr, 2007; Karniell-Miller, Strier & Pessach, 2009; Kvale, 1996; Forbat & Henderson, 2005; McDowell, 2001; Turnbull, 2000; Winter, 1991) have dealt with the balance of power between interviewer and interviewee. These scholars suggest that the relationship between researcher and researched was asymmetric and based on a power relationship that can never be totally cancelled. The balance of power between the interviewee and the interviewer is, however, variable: even when it appears that the interviewer controls the structure of the research (for example, in the questions posed, the topics being discussed, the amount of information the researcher is prepared to supply the interviewees, as well as data analysis, interpretation and summarizing), the interviewees control the information they are prepared to give the interviewer, and may omit information they are not keen to disclose (Bhopal, 1995). Hunter (2005) notes an asymmetry that may prevail when interviewing elitist groups. In such a situation, the interviewees are in a position of power relative to the researcher-interviewer, and accordingly there is a need to adopt “strategies to ‘protect’ the researcher…” (pp. 152).

In the current study, it is possible to classify all the interviewees into an elitist academic grouping insofar as two were college and university professors, six held doctoral degrees and the rest held M.A. degrees. One could have anticipated a shift in the balance of power in favor of the interviewees, however, this did not occur. In order to create a balance in power the researcher made an effort to create a comfortable atmosphere of cooperation with interviewees. This was accomplished by mentioning common acquaintances, providing a short explanation about the research study, and showing respect and appreciation for the information that the interviewees were supplying (Winter, 1991; Hall, 2004; Turnbull, 2000). On the one hand, giving transcripts to the interviewees is an act of empowerment that suggests that the researcher respects the experience and contribution of interviewees (Payne et al., 2007); but, on the other hand, this action enables participants to control the transcripts and, by so doing, also the data (Poland, 1995). This action alters the balance of power between interviewer and interviewees and may be problematic from a research point of view, since the interviewees become the validators and/or challengers of what is written (Payne et al., 2007) and can potentially influence data analysis and interpretation, and final results (Turnbull, 2000). However, this action also has a positive influence on the participants’ willingness to cooperate in the research (Lapadat & Henderson, 2005), improves the atmosphere in which interviews take place, and transforms the interview into a dialogue between people who share the same experience (Bhopal, 1995).

Transference of the transcripts to the interviewees provided them with the opportunity to share their experiences with the researcher even long after the interview was held. One of the interviewees wrote in her response to the transcript:

There are certainly changes (in the school) primarily as a result of the implementation of teaching processes in technology-enhanced environment: students [stages] 5–6 study with a portable computer. By the way, ‘the garden’ already exists, you are invited! Dreams do in fact come true, one only needs to dream”! ⁴

Sense of embarrassment and threat

Reading transcripts is often cause for embarrassment and anxiety for interviewees both because of the exposure of what was said as well as their perception of how it is presented in the transcript (Forbat & Henderson, 2005; Turnbull, 2000). Kvale (1996), who published explicit interviews in his book, Interviews: An introduction to qualitative research interviewing. admitted that this method of transcription was cause for incoherency in the text. He mentioned that a teacher who read a draft of the chapter in which his interview appeared, demanded that corrections be made to the language, since the teacher felt that the transcript offended him and his professionalism. Kvale concurred and the text was corrected.

It can be concluded that the experience of reading transcripts, even for participants such as those in the present study for whom interviews and transcripts are not a new experience, may be embarrassing, stressful, and even threatening. The interviewees expressed dismay with the grammatical errors as seen ‘in black and white’, and felt some anxiety and embarrassment regarding the conversational flow in the transcript as opposed to more refined speech. This embarrassment was expressed openly: “Is that what I sound like? It's not easy to read the words and proceed into the essence of things.” Another expression of embarrassment was made by one interviewee who, although aware of the fact that a transcript reflects spoken language, requested editing of extracts that would be published, insinuating that this would otherwise be harmful to her image: “I reread the written content. In our discussion a style of spoken language is apparent. I certainly approve your correcting phrases in order to assure that the message is clear and correctly worded.” The feeling of embarrassment may also lead to a feeling of anxiety as to how the content of the written transcript will be used. One interviewee wrote: “Tell me how what I have to say fits into your research project. What will be published of what was said? It's obvious to me that the material you sent me is only the transcript and not the materials that will be used for your project.”

While some of the feelings of embarrassment and confusion felt by the interviewees when confronted with the transcript can be attributed to a fear of damaging their image, others can be attributed to their difficulty in coming to terms with the transcript which, as a written text, has a set of accepted norms different to those of spoken language. These interviewees not only expressed anxiety or embarrassment, but often changed spoken language to written discourse or at least to a more refined discourse by recommending the deletion of sentences and paragraphs that appeared unnecessary.

From spoken to written language

Spoken language is constructed and uses terminology which is different from written language; however, when speech is represented as written text, readers evaluate it according to the conventions of written text. It is no wonder, therefore, that interviewees who read the transcripts felt disappointed and embarrassment. They judged them according to formal standards of writing, and saw them as being incoherent or uneducated (Tilley & Powick, 2002; Turnbull, 2000). Each interviewee therefore corrected the language of the transcript in order to refine it according to written conventions. Corrections made by the interviewees can be divided into three levels: slight, medium, and extensive. Slight corrections refer to simple, insignificant improvements, for instance, from “I'm altogether in favor of an approach of opening up”, to “I'm altogether in favor of an open approach,” or changing words, for instance, from “They don't understand for the life of them” to “don't understand the point of the story.” Medium and extensive corrections refer to substantive changes to the text. Medium corrections involved the addition of particulars to various paragraphs along with verbal corrections as exemplified in the slight corrections.

Extensive corrections included all of the above with the addition of the deletion of extensive portions of the text that created a new construction and, by extension, a new text. One of the interviewees explicitly recommended deleting portions of the text, as she believed that it was offensive or included irrelevant information. As she wrote in an introduction to the transcript: “I've marked in green the segments that I feel should be erased…segments that may be construed to be offensive, segments that I feel are irrelevant.” Amongst the segments that interviewees suggested deleting were portions of the discussion with the interviewer which, in their opinion, digressed from the subject of the interview. Another interviewee did not abide by this criterion, and she simply re-edited the transcript by deleting portions of the text.

Research ethics

An interview by its very nature is based on a power relationship and where there is power, there lies a potential for the infraction of human rights (Fletcher, 1992). The transference of the transcripts to the interviewees assisted in protecting six rules of research ethics. The first is avoidance of treating people as simply automatons; this was overcome by including them in this study. One of Forbat and Henderson's interviewees (2005) ratified this aspect when sending them an email: “One of the reasons I have found myself unable to read some work is because I can't stand the attitude to the ‘objects’ under study betrayed by some of them…Your paper is good; above all you are clearly engaging with all people as people and not just as problems to be solved” (p. 1125). Second, through transference of the transcripts care was taken of the interviewees’ well-being, and the researcher's integrity was safeguarded. Third, transference of the transcripts to the interviewees reflected the principles of free consent and informing interviewees. These principles are based on the belief in, and acknowledgement of, the fact that people have the ability and the right to make independent decisions based on knowledge. Thus the researcher is obligated to inform the interviewees. Fourth, ethical commitments to people who may be harmed through the initiative of the research was preserved. This was achieved through the use of abbreviations of their full names, by mentioning their role and not identifying them by name, or by giving pseudonyms to the schools. Fifth, the method of transcript transference to the interviewees did not allow others access to the information and did not enable its distribution, thus the principles of privacy, confidentiality and anonymity, which are basic principles of human dignity, were honored. Hagens and colleagues (2009), however, believe that sending the transcriptions through email increases the danger of exposure to a third party. Finally, the principles of justice, fairness, equality and generalization were observed through the use of methods, standards, and fair processes for checking transcripts – for example, by comparing them to protocols taken down during the interview. Moreover, this showed a commitment to the principle that research participants should not be harmed or exploited for the promotion of knowledge. The study aspired to equilibrium while protecting human dignity, minimizing elements of harm and maximizing elements of public benefit and the advancement of knowledge (Inter-Agency Secretariat on Research Ethics, 2005).

Researchers, who are aware of the power they wield and their obligation to safeguard a code of ethics (Bhopal, 1995), make efforts to protect them, and the interviewees even demand this (Turnbull, 2000). Accordingly, the interviewees’ stance on the subject of ethics in this study and their recommendations for the deletion of portions of the transcripts were anticipated. The interviewees related to two ethical aspects: recognition and substantive. In the recognition aspect the interviewees requested that names that were mentioned during the interview be deleted: “[I'm requesting erasing the] mention of names of specific people or places,” “Since the (name deleted) is no longer with us, it would better to avoid mentioning his name,” and “(name deleted) – perhaps, we should defer identification of the place.”

From the substantive aspect, the interviewees asked to delete sections dealing with gossip (“now this is at the level of gossip — unnecessary”) and content that may harm people. Harmful content arises when the interview flows and interviewees do not pay sufficient attention to what they are saying. When the interview does not flow and is characterized by suspicion and anxiety, the interviewees are in greater control of what they say. Turnbull (2000) stated that her interviewees expressed great control over what they said. One was so controlled in her remarks to the extent that she managed to abstain during the interview from mentioning names of colleagues, “They were all ‘colleagues’ ” (p. 30), and another asked to shut off the tape recorder each time she criticized her parents who were long deceased. These examples reinforced the assumption that the pleasant atmosphere that prevailed during the interviews for the current study served to encourage more openness on the part of the interviewees, openness that led to the removal of defense mechanisms and barriers and to saying things that in retrospect may have been regretted. Accordingly, when the interviewees read their transcripts and found segments that were likely to be perceived as being hurtful to others, they asked the researcher to deal with them. For instance, one of the interviewees wrote in an email:

Most of the things (I said) do not contain harmful reference. Nevertheless I extracted two portions and I wish to refer to them: 1. my saying that this is important to me should not be inferred that in the past the team did not work in a professional and in-depth manner. Especially since they are a veteran team, they had experienced many changes and each change necessitated a long process that has a place in practice; 2. most importantly it should not be inferred that the teachers were careless on parents’ day. My intention (was) that it is not enough to have a parent's day, but that parents should be continuously involved.

In an additional email, this participant emphasized: “It's important to me that my remarks do not offend the people with whom I work and certainly some of my remarks could be perceived as such.”

Reflection

Reading the transcript awakens emotions, and on the positive side it serves as a catalyst for those burning issues relevant to the interviewee and present an opportunity to express things, that may be perceived as not having been adequately expressed during the interview (Forbat & Henderson, 2005). On the negative side, it causes discomfort that stems from the effort and the time invested in reading the long and cumbersome transcriptions (Hagens et al., 2009). Reading the transcriptions can also be an answer to the interviewer's prayers, as expressed in a letter to the interviewees that accompanied the transcript: “If in light of the transcript you should find it necessary to clarify things that were said or add things that you feel to be of importance to the subject, you can state this under ‘versions,’ as ‘text comments’ on the transcript or as your own remarks at the end of the text.” The following text was an addition to a transcribed text sent to an interviewee, which clarified issues that were not mentioned in the transcription:

I'll add only this from a view of reflection and other things I've done since then. There are wonderful people in the system; there are wonderful teachers, pedagogic mathematical people, and especially wonderful pupils. However, as usual in our country, the money does not go to the people who really work. The money goes to bloated administrations that continue to create more and more beauracracy and take more and more money. People who do not have the vaguest idea about education and the goings on in the classroom determine the policies, and the changes of policy are enacted from year to year with no overall continuity or overall thought. I had the great privilege to participate in the project, the atmosphere was great, the cooperation – incredible. The project influenced the lives of all the people who participated in it and it faded out, just as everything good does in the educational system. The magic words for success in education – big money for those who really do something, in other words – teachers, (and in accordance with what they feel are their requirements to meet the objectives and not what someone else feels that they require) and true educational leadership that places education on a pedestal and does not simply pay lip service.

It appears that examining the transcription awakened in this interviewee a desire to reach conclusions beyond local clarifications and sections in the text, and to give expression to the power of what had actually happened. The opportunity to react to the transcript provided her an opportunity to voice her position on the educational system as a whole with regard to the topic about which she was interviewed.

This type of reaction, as already mentioned, is the type of reaction sought after by the researcher. It is not harmful, does not embellish the original text, and remains in the spoken form; furthermore, it serves as a desired addition to the researcher's knowledge of the research project and provides insights into the background, the atmosphere, and position of the interviewee on important topics.

Conclusions

The experience of sending transcripts to interviewees raises several important ethical and research issues. First, despite the fact that all the interviewees who participated in the study were academics holding at least a second degree, despite the fact that some were university lecturers, and despite the fact that most use interviews and transcriptions in research that they themselves conduct or in teaching the tool to students, many were not sufficiently aware of the representation of the interview in the transcript. They were accordingly surprised by the poor vocabulary of their statements as depicted in the transcripts. Research indicates that both academics and non-academics find it difficult to read transcripts of interviews and are surprised by the lack of coherence in the discourse (Forbat & Henderson, 2005; Hagens et al., 2009; Tilley & Powick, 2002). The problem is exacerbated when the transcribed interview is naturalized and includes grunts, pauses, bursts of laughter, repetitions, gestures, and body language. Kvale (1996) and Poland (1995) warned of the danger in conducting naturalized, transcribed interviews: “Be mindful that the publication of incoherent and repetitive verbatim interview transcripts may involve an unethical stigmatization of specific persons or groups of people” (Kvale, 1996 pp. 172–3). Accordingly, quoting things as they appear in the transcript may be hurtful to the interviewees, presenting them in a negative light (Beason, 2000) and resulting in loss of trust between interviewer and interviewees.

The researcher is thus confronted with four dilemmas. The first is the desire to present things as originally stated. Because it may be crucial for proving a particular point, the researcher may wish to use information interviewees wish to delete. This might come in conflict with the desire to uphold a code of ethics, respect human dignity, and refrain from harming the interviewees (Inter-Agency Secretariat on Research Ethics, 2005; Kvale, 1996; Saldana, 1998).

The second dilemma arises because some interviewees may rephrase large sections of the transcript in order to refine them. Thus, there is a research problem relating to the amended text, principally since the transcripts represent the data base on which the researcher relies (Hagens et al., 2009; Lapadat & Lindsay, 1999). Poland (1995) wrote that “verbal and written communications are very different mediums, incorporating different structures and syntaxes” (p. 299). Rephrasing the transcript from spoken discourse to written discourse alters the type of text and the conventions of spoken discourse (Kvale, 1996; Tilley & Powick, 2002). This problem is substantive since transforming a transcript by refining introduces an additional variable: the form of medium. Accordingly, the quotes are not “quotes verbatim,” since these were not the things that the interviewee said, but rather things that he or she adapted to the rules of standard written text grammar (Grundy et al., 2003), and in fact wrote them.

The third dilemma is related to the fact that the interviewees’ reactions raised the issue of the balance of power between the interviewer and interviewee. This issue surfaced at various stages of the interview (planning, execution of the interview, transcription, transference of the transcripts to the interviewees, and interpretation) and was made apparent in this study by the various reactions of the interviewees to the transcripts. Despite the fact that this was not the original intention, the transference of transcripts to the interviewees was an act of empowerment. The opportunity to improve the text, make it appear more refined, clarify issues, approve or disapprove of the print, and /or take a position was an act of empowerment and a cause for change in the balance of power, which is usually under the control of the interviewer-researcher (Bhopal, 1995; Bocholtz, 2000; Hall, 2004; Hunter, 2005; Turnbull, 2000; Winter, 1991). The act of empowerment contributed to the sense of trust between the participants and the researcher. It also encouraged interviewees to ‘update’ the researcher and relate what had happened at the school after completion of interviews, in the belief that the story was of interest and out of a desire to share these experiences with the researcher. Furthermore, whilst there is a danger inherent in awarding interviewees the right to censor or retract the written word (Poland, 1995), this empowerment may add depth to the research as the interviewees enriched the researcher's existing knowledge and by allowed a glimpse into the world beyond the research project. This kind of glimpse is only possible when the interviewee feels that his or her views are respected and awarded due importance (Faircloth, 2004; Hiller & DiLuzio, 2004; Karniell-Miller et al., 2009. Moreover, the respect shown the participant and the interest shown in his or her opinions may encourage the interviewees to participate in additional research (Forbat & Henderson, 2005).

The fourth dilemma is that not all the interviewees responded to the transcript; this phenomenon was apparent in other studies (e.g. Forbat & Henderson, 2005; Hagens et al., 2009; Hall, 2004). Hall (2004) explained that lack of response to transcripts may be due to the burden placed on the interviewees or embarrassment to respond face-to-face or over the phone. Whether we accept these explanations or not, the lack of response on the part of some of the interviewees raises two substantive questions: What are the real reasons that the interviewees refrained from responding to the transcript despite their agreement in principle to do so? To what extent is the quality of the research harmed, wherein some of the data that the researcher used are transcripts that were amended by interviewees and some are the original transcripts? An answer to the first question will shed light both on the importance that the interviewees attach to the transference of transcripts to themselves as well as the reasons for their refusal (obvious or hidden) to respond to the transcripts, while an answer to the second question may shed light on the conflict between the search for content validation and that of research credibility.

In conclusion, transference of transcripts to interviewees raises methodological problems, ethical problems, and problems of research credibility. Accordingly, despite the inherent advantages of transferring transcripts to interviewees, such as enhancing the validity of the transcript, and receiving clarifications and statements that enrich what was said in the interview, despite the advantage that arises from ethical and political aspects that are expressed in the change of the balance of power and despite the fact that transference of transcripts to the participants may assist in achieving the aims of the research (Lapadat & Lindsay, 1999), the disadvantages to the research study create doubt as to the value of transferring transcripts to the interviewees in qualitative research. Furthermore, Hagens and colleagues (2009) believe that the disadvantages overcome the advantages of the procedure. It would appear that considerable thought must be given during the stages of carrying out research as to whether transcripts should be given to the interviewees. On the strength of this understanding, it would appear that additional research that examines the lack of response by many interviewees is required, as well as studies that can point to ways of dealing with both practical and ethical issues that arise from transference of transcripts to interviewees.