CHAPTER 6 HOW TO TRANSCRIBE RECORDINGS 125

What is transcription?
Data for qualitative researchers take many different forms but the spoken word is dom-
inates. Transcription is the process whereby a sound (or video) recording of the spoken
word is turned into written language for subsequent analysis. The spoken and written
word are different since the spoken has features lacking in the written word. Simple
examples are pitch, volume and pace characteristics. Although it is perfectly possible for
summaries of language data to be used for some research purposes, the use of transcripts is
extremely common. The written transcript is a verbatim (word-for-word) record of speech
or, sometimes, sections of spoken word particularly relevant to the research question. The
basic transcription choices in qualitative psychology are:

• The orthographic/secretarial/playscript transcription which reproduces solely the

words said, not how they are said. This is possibly the commonest form of transcrip-
tion.
• Jefferson transcription uses common keyboard symbols to provide additional infor-
mation over and above secretarial transcription. The additions include the way in
which the words are said, pauses in speaking, where speakers overlap, and so forth.
This system meets the needs of conversation analysis best (Chapter 10) and social
constructionist discourse analysis (Potter and Wetherell, 1987) (Chapter 9).
Research always imposes limitations on the data collected. Deciding to audio-record
a focus group discussion imposes constraints and deciding to video-record affects
the available data too. Taking notes during an interview influences things also. The
note-taking researcher might appear to be rather formal. Whatever choices are made
influence the data and, consequently, what can be done with the data. Much the same
is true of transcription: once a choice is made about who transcribes, what is tran-
scribed and how it is transcribed there are consequences. Potter and Hepburn (2009)
suggest:
Crucially, advocates of a straightforward orthographic or ‘play-script’ version
of transcript . . . often fail to appreciate that they are not a more neutral or sim-
ple record. Rather they are highly consequential transformations. For example,
orthographic transcript imposes the conventions of written language which are
designed to be broadly independent of specific readers. Such a transformation
systematically wipes out evidence of intricate coordination and recipient design.
It encourages the analyst to interpret talk by reference to an individual speaker or
focus on abstract relations between word and world. (p. 1)
By not indicating how words are being said, the analysis of secretarial transcrip-
tions is steered to interpreting the words as in formal written language. Consequently,
features of language used in everyday interaction to achieve certain ends are lost.
So if the researcher’s emphasis is on what language does then secretarial (orthographic
or playscript) transcription fails badly. Ultimately no transcription is the same as the
spoken words on which it is based.

Is a transcript necessary?
Not all forms of qualitative research necessitate the use of transcripts. The classic work
of Michel Foucault, which is the basis of critical discourse analysis, did not employ tran-
scripts of any description (Fairclough, 1993). For the critical discourse analyst, interviews

M06 Introduction to Qualitative Re 51202.indd 125 07/01/19 4:06 PM

 126 PART 3 QUALITATIVE DATA ANALYSIS

and transcripts are methodological tools enabling the identification of discourses – their
concern is not with the detail of how talk is constructed. Critical discourse analysts are
interested in the ideological dimensions of power embedded in text. In contrast, transcripts
are virtually universal in conversation analysis and social constructionist discourse analysts
which concern the mechanics of the construction of talk.
What do secretarial transcriptions omit that is present in interactions? Among the
additional non-verbal communication features found in interviews etc. are the following:

• Proxemic communication The use of physical space between conversing persons.

Think of the way in which women on television shows physically move their chair
away to indicate their loathing for an ex-partner refusing to accept paternity.
• Kinesic communication The body movements and postures which communicate
more than the spoken word alone. Folding one’s arms might be perceived as a lack
of openness to different opinions.
• Paralinguistic communication The changes in volume, pitch and other voice char-
acteristics. It is known from research that many people’s voices rise in pitch when
lying.
• Chronemic communication Variations in the pace of speech and silences. People
may speak faster when feeling emotional.
None of these appear in word-for-word orthographic/secretarial/playscript tran-
scriptions. The Jefferson transcription method partly includes these ‘missing’ elements.
No transcription method includes all of them. There are some aspects of human
interaction for which transcription procedures are being developed. Hepburn (2004)
sought methods by which crying may be described and transcribed by the researcher.
Discussions of the role of transcription in qualitative research suggest that there
is no transcription method that invariably is preferred. Although, transcription has
largely been standardised for conversation analysis and constructivist discourse analy-
sis where the Jefferson transcription system is required, this does not apply to all qual-
itative research. The transcription method used depends on the purpose and nature
of one’s research. Hammersley (2010) suggests numerous decisions to be made about
translation. These include the following:
• Do I need to transcribe any part of a particular recording and, if so, how much?
Not all qualitative research requires a transcript. Even if a transcript is needed, the
researcher may decide to transcribe only part of the recording. So selective transcrip-
tion is a possibility although clarity is needed about the reasons for selectivity.
• What elements of the recording be transcribed and how? One could represent the
sounds actually made rather than what was intended (e.g. transcribing regional
accents) or simply record the words heard correcting errors.
• Where there are many speakers, will the researcher try to indicate which individuals
a speaker is mainly addressing? In a focus group discussion two speakers may speak
to each other but say some things too quietly to be heard.
• Should non-word vocalisations be included (e.g. uh uh, coughs, intakes and
out-takes of breath, laughs, etc.)? They may or may not have meaning depending
on context. Sneezing has no alternative meaning but coughing can indicate all sorts
of things apart from being physiological.
• What should be done about pauses and silences? A decision is needed about whether
they are timed and, if so, just how? Is their precise length important down to the
millisecond or is it sufficient to note that they are perceived as long or short. Silence
may have various meanings. Who decides on the significance of pauses and silences?

M06 Introduction to Qualitative Re 51202.indd 126 07/01/19 4:06 PM

 CHAPTER 6 HOW TO TRANSCRIBE RECORDINGS 127

• Should physical gestures which accompany words be included?

• Just how should a page of transcript be laid out? Often it is done like a script for a
play but then how will overlapping utterances be identified?
• How should speakers be identified? Giving a name is indicative of gender, giving a
category such as parent and child suggests that these roles are important to under-
standing what is happening.
• What aspects should be selected for inclusion in a research report? Should the inter-
viewer’s question be given? Should the transcript in the report be exactly the same
as the researcher’s transcript or tailored especially for the report?
The general assumption across qualitative research is that ‘Transcripts are produced
for particular analytic purposes’ (Nikander, 2008, p. 225). The purpose of the anal-
ysis determines the features of the transcript. Transcripts do not have to be ‘perfect’
to be useful. Qualitative researchers should know the limitations of the transcription
methods that they use. They know that their recordings and transcripts are valuable,
in-depth representations of what is being studied – they are not that thing. Bucholtz
(2000) argues that transcription can be seen as a continuum. The extremes represent
what she calls naturalised versus denaturalised transcription (Figure 6.1). A transcript
may be anywhere on the dimension between these two extremes:

• Naturalism describes transcripts in which every nuance is captured if possible –

detail is important in these transcripts.
• Denaturalism describes transcripts where, for example, grammar is corrected and
‘noise’ such as stuttering, pausing and accents are eliminated. That is, idiosyncratic
elements are not included.
For some types of qualitative research, the choice of transcription method is estab-
lished by convention. But there are decisions to be made for other forms of qualitative
analysis. Oliver, Serovich and Mason (2005) suggest that the researcher should reflect
on the needs of their research and make decisions in keeping with these. Transcription
decisions have a significant impact on the outcomes of research. This is not only in
terms of what can be achieved. Conversation analysis without naturalistic transcripts
just would not work; on the other hand, thematic analysis, grounded theory and critical

Denaturalised transcription Naturalised transcription

(when the transcription is made to be (gives as much detail as possible such
like written language and might as pauses, stuttering, etc.).
include commas, paragraphing, etc.). Typical of conversation analysis
Typical of grounded theory

FIGURE 6.1 The naturalised-denaturalised dimension in transcription

M06 Introduction to Qualitative Re 51202.indd 127 07/01/19 4:06 PM

 CHAPTER 6 HOW TO TRANSCRIBE RECORDINGS 131

often mediators between the researcher and the participants in the research study. The
translator should be present throughout each step of the research, Berman argues.
Transcription methods vary substantially in qualitative research. The choice of
transcription method is not an arbitrary matter but the result of careful consideration
of the research task. This is not to suggest a new transcription method for every new
study. That would be pointless and futile. In psychology, the Jefferson transcription
method is almost invariably used in constructionist discourse analysis and conversation
analysis studies. It would be rarely, if ever, used by researchers using thematic analysis,
grounded theory, narrative analysis based on narrative theory, or interpretive phenom-
enological analysis. But it is the best documented available approach to transcription
and makes a good starting point to introduce the practicalities of transcription.

The Jefferson approach to transcription

Initially appearing somewhat daunting, the Jefferson transcription system uses many
symbols and conventions which make intuitive sense. For example, underlining denotes
emphasis and capitals indicate words said LOUDLY. Any difficulty is more to do with the
number of conventions in the Jefferson system since each individual convention is simple
in itself. The Jefferson transcription system uses characters available on standard computer
and typewriter keyboards. If they are prepared to spend time learning its basics, the
Jefferson transcription approach is easy for anybody to implement. The keystrokes used
in the Jefferson system do not necessarily have their normal meaning. Actually, it is more
correct to say that some symbols have both their normal and an additional meaning in the
Jefferson system. Thus capital letters have their conventional meaning but, as already men-
tioned, words in capitals are spoken louder than the other words around them. The use of
(brackets) is indicative of a word not identified with complete certainty from the recording.
Overall, these additional keystroke symbols tell the reader how the words in the transcript
are actually being said. The Jefferson system includes methods of indicating pitch, speech
volume, speech speed, emphasis within words, pauses or their lack of, overlapping speak-
ing, laughter and other non-words, and extra information from the transcriber. The main
Jefferson conventions are given in Table 6.1. Refer back to this table whenever neces-
sary to understand what is happening in a transcript. You may also spot that there are
slight differences between transcribers on certain matters of detail. The Jefferson system
evolved gradually. Not surprisingly then, new symbols were added at different stages in its
development. In addition, sometimes transcription symbols were replaced by others. This
occasionally adds somewhat to the problems involved in using Jefferson transcriptions.
In particular, it may be a problem when it comes to studying research which used slightly
earlier versions of the Jefferson transcription method. Hutchby and Wooffitt (1998) go
into more detail about Jefferson transcription.
Reading a Jefferson transcription is not too difficult, especially if you use Table 6.1
to check the meaning of the transcription symbols. For illustration, consider the
following study of neighbour dispute mediation. The transcript is of a section of a
recording of a meeting between two couples and a dispute mediator. The couples are
Graham and Louise (G and L) and Bob and Ellen (B and E). Bob does not appear in
this segment of transcript. In case you have any difficulties, you can find an annotated
version of the transcript later. This contains examples of each of the different features
in the transcript. The transcript is not as complex as some and it does not include every
transcription notation symbol but it makes a very clear starting point.

M06 Introduction to Qualitative Re 51202.indd 131 07/01/19 4:06 PM

 132 PART 3 QUALITATIVE DATA ANALYSIS

TABLE 6.1 The Jefferson transcription system (as described in Jefferson, 2004)

Jefferson symbol Example Usage

Pitch indicators
c absolcutely Pitch rise symbol. Bearing in mind that there are continuous variations in
the pitch of speech, c indicates that the following has an ‘unexpectedly’
markedly higher pitch than the previous speech. Multiple arrows may be
used to indicate greater extents of this.
T AbsolutelTy Again considering the normal variations in the pitch of speech, T indicates
that the following is spoken at a markedly lower pitch than the previous
speech.
* I’m sor*ry What follows the asterisk is spoken in a squeaky or creaky voice.
. sure. The word before is spoken with a falling intonation possibly indicative of
stopping though it does not have to be followed by a pause.
Speech volume indicators
CAPITALS for GOODNESS sake CAPITALS mark speech that is noticeably spoken at a louder volume than
the speech around it. This is assessed in terms of that speaker’s general
volume. It is incorrect to put a loud speaker’s talk in capitals throughout.
underlining For goodness sake This indicates speech spoken louder than the surrounding text though
underlining does not indicate things are quite so loud as the use of capi-
tals does.
°° And when I discovered The superscript ° or degree sign is used to mark the start and end of
that she had °died° noticeably quieter speech.
Speech speed indicators
>< Then I said>we’d better The speech between these signs is faster than surrounding speech.
hurry home<
<> On reflection, <I think> The speech between these signs is slower than surrounding words.
things are
Emphasis indicators within words
: de::licious The colons show the elongation of the previous sound. Multiple colons
can be used to indicate the extent of the elongation.
? right? The? indicates that there is a questioning (rising) tone whether or not the
speaker is grammatically asking a question.
underlining Kerumbs The use of underlining shows emphasis made in a word.
It indicates both where the emphasis is and the strength of the emphasis.
The emphasis can be in pitch or loudness.
Pauses or lack of
(0.1) Let me think (3.1) no I The numbers enclosed by round brackets (3.1) indicate the length of a
don’t remember pause in speech expressed in tenths of a second.
So (3.1) indicates a pause of three and one-tenths of a second. This is a
long pause and they are more typically 0.3, 0.5 and so forth. If they are
clearly a part of a particular speaker’s speaking then they are included in
their speech. If they do not involve a particular speaker then they are put on
another line of transcript. A new line is used where there is any uncertainty
about this.
(.) I was thinking (.) would The stop in brackets (.) can be described as a micropause which is appar-
you like a break ent but too brief to measure precisely.
- r-r-really The preceding sound is cut off by the hyphen.
[] I think the job= The equals signs shows that the next speaker begins without pause after
=no if you the first speaker.

M06 Introduction to Qualitative Re 51202.indd 132 07/01/19 4:06 PM

 CHAPTER 6 HOW TO TRANSCRIBE RECORDINGS 133

TABLE 6.1 (continued)

Jefferson symbol Example Usage

Overlapping speaking
[] I [think the] job Square brackets [ ] are used to indicate two speakers speaking at the same
[No if you] time.
Laughter and other non-words
Hhh hhh I’m sorry that it has The h indicates an audible breathing out – several can be used to indicate
come to this its length.
.hhh so to recapitulate.hhh The .h indicates audible breathing in. The h gives an indication of the
the committee has length of the breath.
voted against
heh heh heh heh HEH This is voiced laughter much as in ha ha ha. Since it is equivalent to a word
one can add other symbols.
h Do(h)n’t tick(h)le m(h)e The (h) indicates that laughter is occurring within speech.
Extra information from the researcher
((note)) ((Clare speaks in a mock Double round brackets enclose comments made by the transcriber which
Scottish accent)) indicate things such as the characteristics of the delivery or something to
do with the context of the speech, for example.
() the treasure is buried Round brackets with just space inside indicate that something has been
under the ( ) and it is said which the transcriber cannot recognise. The amount of space indi-
all mine cates the approximate length of the ‘missing’ word. Sometimes a query (?)
is used to the same effect.
(word) she had a (bunion) Round brackets with a word inside mean that the subscriber is not abso-
operation lutely certain what word has been said but believes that it is the word in
brackets.
(word)/(word) (nights)/(likes) Indicates two equally possible hearings of what was said
S John: S The use of the side arrow indicates that a particular line of transcript is
of special importance in terms of the analytic points being made by the
analyst.
Gaze (for video transcription)
______________ . . . . . . 62______________ . . . . . . Some transcribers use an unbroken line to indicate that a person is gazing
62 if you would just directly at another person while speaking and then a dotted line to indi-
cate the gaze has broken. Essentially it involves adding an extra line (with
the same number as the text) to indicate gaze.
*See the text of this chapter for details of how overlaps appear in transcriptions.

1G y’know it’s getting – it’s getting real serious this is (.) cbut the
2 lad keeps getting away with it (.) unfortunately (.) his mother hasn’t
3 got a bloke there (.) so she is talking in [front of the children
4L: [she’s not living there half
5 the time is she=
6G: =no she’s out at night and they are using it as a- a rendezvous for the
7 gang
[. . . ]

M06 Introduction to Qualitative Re 51202.indd 133 07/01/2019 20:59

 134 PART 3 QUALITATIVE DATA ANALYSIS

8G: that’s the whole top [and bottom of it

9L: [it’s like the dustbin left out for a week (?) on
10 [the pavement
11G: [IT’S ALL TO DO with this one lad (.) right (.) we’ve had report- we’ve
12 got connections at the school (.) they said ‘what’s the point of him
13 coming to school he knows nothing (.) he only causes trouble’ (0.5) so
14 [they never bothered about him
15L: [it’s like they’ve had words with this woman and can’t get through to
16 her from school you know [course (.) he’s left now so
17G: [(?)
[. . . ]
18G: (?) °no no° I mean the funny thing about this is that (.) in actual
19 fact (.) I mean the lady’s got to be responsible (.) she’s got to
20 be responsible [at the end of the day because [she’s never there [she
21E: [well she’s never there is she
22L: [() she’s effing and
23G: can’t control him
24E: yeah
(from Stokoe, 2003, p. 326)

The intellectual origins of Jefferson’s approach to transcription must be under-

stood. Her focus was on the fine detail of interaction between people. She was not
so much interested in the broad patterns of conversation but on how aspects of con-
versation are continuously shaped by what happened previously in the conversation.
In other words, speech is endlessly contingent on itself. On one famous occasion the
then British Prime Minister, Gordon Brown, performed a slip of the tongue in which
instead of saying that he ‘saved the banks’ he actually said ‘saved the world’ when
answering Parliamentary questions. Of course, this could be seen as a Freudian slip
which revealed what the Prime Minister truly was thinking. However, Atkinson
(2008) was reminded of Gail Jefferson’s (1996) paper ‘On the poetics of ordinary
talk’ in which she discusses how the choice of a wrong word was the consequence
of the words which immediately preceded it. So Atkinson checked what the Prime
Minister had actually said:

GORDON BROWN: ‘The first point of recapitalisation was to save banks that would otherwise
have collapsed and we’ve not only saved the world – erh – saved the banks.’

Notice that the extract contains a sequence of ‘w’ sounds – ‘was’, ‘would’, ‘otherwise’
and ‘we’ve’. Thus Gordon Brown’s error can be seen as an unfortunate recapitulation
of the ‘w’ sound rather than any manifestation of his personal psychology. Jefferson
was adamant that such speech-induced errors are not the consequence of hidden
psychological motivations as is the case with Freudian slips.

M06 Introduction to Qualitative Re 51202.indd 134 07/01/19 4:06 PM

 CHAPTER 6 HOW TO TRANSCRIBE RECORDINGS 135

Potter (2003) makes the point of Jefferson transcription clear using the fol-
lowing excerpt from a telephone call inviting someone over (Davidson, 1984,
p. 105):

A: C’mon down he:re, = it’s oka:y,

(0.2 sec)
A: got lotta stuff, = I got be:er en stuff

Potter writes:
Note the way the speaker upgrades the invitation. Why might this be? The likely
reason is that the pause of 0.2 of a second is a cue to an impending refusal.
Conversational actions such as invitation refusals are typically prefaced by some
delay, and research has shown that speakers modify their actions on the basis of
such predictions . . . (Potter, 2003, p. 82)
In other words, without the indication of a delay the interpretation of this short
snippet is less clear. The situation would not be rectified by later refusals or accept-
ances from the other person since the upgraded invitation may or may not be effective.
Remarkably, there appear to be no ‘rules’ governing how to do the orthographic/
secretarial/playscript approaches to transcription. Just what should go into the tran-
script and just what can be left out? For example, does the transcriber identify things
which cannot be heard clearly or do they insert their best attempt, when is it appro-
priate to transcribe the accent of the speaker, and what does one do about overlaps in
speaking? For this reason, it would be appropriate to use Jefferson transcription con-
ventions about these basic matters even when the additional linguistic codings which
Jefferson transcription involves are not used.

What the transcription symbols mean

Graham is the There is a brief, Graham says ‘but’

speaker for the first unmeasured, pause at a markedly higher
three lines in the speech pitch

1 G: y’know it’s getting – it’s getting real serious this is (.) cbut the Graham and Louise
are speaking
2 lad keeps getting away with it (.) unfortunately (.) his mother hasn’t
together
3 got a bloke there (.) so she is talking in [front of the children
4 L: [she’s not living there half
Graham takes over
5 the time is she= conversation from
6 G: =no she’s out at night and they are using it as a- a rendezvous for Louise without
the pause

M06 Introduction to Qualitative Re 51202.indd 135 07/01/19 4:06 PM

 136 PART 3 QUALITATIVE DATA ANALYSIS

7 gang
[...]
8 G: that’s the whole top [and bottom of it
9 L: [it’s like the dustbin left out for a week (?) on
Graham talks loudly
10 [the pavement
11: [IT’S ALL TO DO with this one lad (.) right (.) we’ve had report- we’ve
12 got connections at the school (.) they said ‘what’s the point of him
13 coming to school he knows nothing (.) he only causes trouble’ (0.5) so
A pause timed at
14 [they never bothered about him
half a second
15 L: [it’s like they’ve had words with this woman and can’t get through to
16 her from school you know [course (.) he’s left now so Transcriber cannot
hear
17 G: [(?)
[...] Some text omitted
18 G: (?) °no no° I mean the funny thing about this is that (.) in actual
no no is spoken
19 fact (.) I mean the lady’s got to be responsible (.) she’s got to
noticeably quietly
20 be responsible [at the end of the day because [she’s never there [she
21 E: [well she’s never there is she
Ellen begins talking
22 L: [( ) she’s effing and at the same time as
23G: can’t control him Graham then Louise
overlaps with Graham
24E: yeah

The development of transcription

Just who first instigated the orthographic/secretarial/playscript transcription of social
scientific data is unknown. Phonetic transcriptions of language began in the twenti-
eth century with systems such as the International Phonetic Alphabet. Gail Jefferson
(1938–2008) has been influential in disciplines like sociology and anthropology – and
more recently psychology. The questions which guided her research crossed discipli-
nary boundaries. Fortune led her to conversation analysis (Chapter 10) as it was being
developed by Harvey Sacks (1935–1975). She took one of his courses to complete
the requirement for her degree in dance. It so happened that Jefferson had previous
experience of doing orthographic/secretarial/playscript transcriptions from when she
had been a typist.
Jefferson began transcribing some of the lectures by Sacks from recordings. Sacks
was not the most prolific of writers and Jefferson’s transcriptions of his recorded
lectures are an important part of his legacy (e.g. Sacks, 1992). Eventually Jefferson
began her PhD under the supervision of Sacks. It was at this time that her methods of
tracing the really fine detail of interaction began to develop and evolve. Her method
was an attempt to combine both a precise record of what was said with the way in
which it was said. Consequently, rather than just note that the speaker laughs Jefferson
attempted to just indicate in detail of how that person’s laughter was combined into

M06 Introduction to Qualitative Re 51202.indd 136 07/01/19 4:06 PM

 138 PART 3 QUALITATIVE DATA ANALYSIS

elaborate systems if a simple transcript of words would suffice. One might initially
be guided by the approach taken by other researchers studying similar areas to yours.
The researcher who is interested in the processes which occur in conversation may well
need a more elaborate transcription procedure than a researcher who is interested in
the substance of what is said rather than how it is said. For the researcher interested
in conversation, the literal words used may be inadequate for understanding the pro-
cesses of conversation. A researcher who is interested in regional accents in conversa-
tion may find both the literal transcription and the Jefferson transcription inadequate
because neither adequately deals with how words sound. Phonetic transcription would
be more appropriate in this case though rare in psychological research. A transcription
method needs to be carefully evaluated to see whether it meets the purposes of your
research (Potter, 1997).
Irrespective of the transcription method chosen, there are some basic principles
which constitute valuable advice for those planning the transcription of spoken
language (O’Connell & Kowal, 1995):

• The principle of parsimony This suggests that a researcher should not transcribe
features of speech which are not intended to be part of the analysis. In other words,
only do in transcription that which will be helpful.
• Keep transcriptions in reports as simple as possible This means that transcriptions
provided in reports of your research should only include features which are impor-
tant in making your analysis intelligible to the reader. Thus, a Jefferson transcript
may not be appropriate even though it has been carried out by the researcher.
• Avoid creating a spurious impression of accuracy in your report O’Connell
and Kowal (1995) found that transcribers working on radio interviews actually
ignored the vast majority of short pauses when transcribing. About a fifth of short
pauses were included but the basis for their inclusion was unclear. Similarly, if the
researcher assesses the lengths of pauses subjectively then it is misleading to use the
convention (0.9) which implies a greater degree of accuracy in measurement than
was the case.
• Checking transcriptions One would not expect transcriptions to be error-free and
transcriber errors are common. Things such as verbal omissions and additions,
translocations and word substitutions may all occur in transcriptions. Consequently
there is a case for an independent checker to eliminate such errors.
Knowledge of transcription may come from basic training in its use but studying
how it is used in research reports enhances the learning process. You have already
had the opportunity of using the neighbour disputes transcript presented earlier. But
transcripts can be rather more detailed than that one, especially where the transcript
contains a lot of laughter or where the speaker breaks up a lot of words as they speak.
By reading relevant papers you will encounter the work of expert transcribers, which
initially looks extremely complicated. You will rapidly understand what is needed in
transcription better. When reading transcripts, pay attention to (a) understanding what
has been said and (b) understanding how it is said. Right from the start, you will find
some parts of most transcripts meaningful. In context, you might be able to have a
stab at some of the less obvious conventions such as how to indicate a quietly spoken
passage. Remember to check the main conventions by consulting Table 6.1. Of course,
to begin with you will consult this table regularly but gradually transcription becomes
second nature.
Initially choose a manageable section of your data to transcribe. This is good
training and will allow you to gain confidence before tackling the data in its entirety.

M06 Introduction to Qualitative Re 51202.indd 138 07/01/19 4:06 PM

 CHAPTER 6 HOW TO TRANSCRIBE RECORDINGS 139

Of course, you will also have queries to answer using Table 6.1 but, ideally, if you
have the chance to consult more experienced transcribers then ask them to review your
work. There are a number of websites which help with learning transcription, some of
which provide recordings to transcribe.
The main stages of a Jefferson transcription follow. They are based on the approaches
of Atkinson and Heritage (1984), Gumperz and Berenz (1993), Langford (1994) and
Roberts (2007). There is some flexibility as to the precise procedures. See also Table 6.2.

Step 1 Tuning into the recorded interaction Transcription is a focused activity rather than a com-
pletely routine process. You need to listen a number of times to the recorded interaction to
be transcribed. Repeated listenings should help you reach the following decisions:

• Am I going to transcribe all of the recording? While this seems to be an obvious

choice to make, it does have its consequences. In particular, transcription is resource
intensive. Complete transcriptions risks being a little rough or inaccurate. If parts of
the recording are unlikely to be analysed, there is little point in straining resources
by transcribing them.
• If not, what aspects of the recording shall I transcribe? In this case you need to
have a means of identifying the important parts for transcription. So, for example,
you may be studying how children of migrants construe their national identity. In
this case, you are looking for material on the recording dealing with this topic.
The boundaries may be determined, for example, by a relevant question from an
interviewer or a focus group leader. But content is not the only possible criterion.
For example, the researcher may be interested in how adult family members bring
children into family conversations, in which case, the boundaries will be around the
entry of a child into the family conversation. Boundaries are not necessarily clear.
The segment of recording should be listened to several times in order to achieve a general
understanding of what is happening. Effort should be made to identify each of the speak-
ers. This may not be easy, especially when the transcriber was not the data collector. Also,
the number of overlapping sequences has a considerable bearing on speaker identification.
This process can be difficult so is best carried out separately from capturing the words said
and the way they are said. This is also where a stereo-recording is most valuable.

TABLE 6.2 Steps in transcription

Step 3: adding Jefferson symbols

Step 1: tuning into the recorded Step 2: rough and transcribing sequencing
interaction transcription accurately
Listen to the recording several times Names of speakers for each segment of Add the fine detail of the transcription
conversation symbols taking special care with overlaps,
etc.
Decide whether all of the material The words said written down as pro- Make sure that the transcription is in its
needs transcribing nounced typically clearest form for readers
If not, decide what parts need to be Put any non-transcribable features in
transcribed brackets, e.g. ((gun fired))
Only use continuous capitals for loud
passages and nowhere else
Indicate pauses, etc. for later more
precise measurement

M06 Introduction to Qualitative Re 51202.indd 139 07/01/19 4:06 PM

 140 PART 3 QUALITATIVE DATA ANALYSIS

Step 2 Rough transcription Study Box 6.1 which provides basic advice on how a transcription
should be laid out, and also look at the transcription provided earlier. Remember that these
are style guidelines and that some things are probably better left until last. Inserting line
numbers is one such late task. The precise layout of a transcription involves judgement,
not simply the application of rules. Line length in transcription is constrained by the need
to insert overlapping speech clearly. This may involve trial-and-error until a satisfactory
solution emerges. Although in some transcription systems the line number refers to an
individual speaker’s turn, in Jefferson transcription the line lengths are arbitrary. Usually
they are kept to a moderate number of words. One way of doing this is to use the natural
groupings of words that occur in speech such as the number of words spoken before a
breath. But, clearly, there is room for variation in terms of how things appear on the tran-
script. This, normally, is of little or no consequence.
At the end of the rough transcription the transcriber should have recorded the following:

Aliases: false or • The names of the speakers for each turn in the conversation or interaction. It is best
assumed if aliases can be used to avoid data management issues as much as possible.
identity
• All of the words spoken, usually attempting to use the word sounds as spoken by
the speaker rather than how they would appear in standard English. For example,
‘summat’ for ‘something’ or ‘yer’ for ‘you’ or ‘your’. But this is not an area where
standardisation among transcribers is apparent. Many features of accents can be
represented reasonably accurately using conventional orthography (methods of

Box 6.1

PRACTICAL ADVICE
How to lay out a transcription

According to Potter and Hepburn (2009), the following way through the sequence. The line numbering is
is the best way of laying out a Jefferson transcription: fairly arbitrary and the same recording transcribed
by a different researcher may have lines of different
• Font It is important to use a proportional font oth-
length and perhaps more or fewer lines for the same
erwise the spacing of overlapping conversation, for
amount of the original recording. This arbitrariness is
example, is very difficult. Their recommended font is
important in that it gives the flexibility needed to be
Courier in 10pt size.
able to indicate overlapping speech.
• Line numbers Each line of a Jefferson transcription
• Layout (a) Use 25 mm (1 inch) margins at top, bottom,
includes a line number. Although these can be typed
left and right of the page and (b) use a code number
in manually, they can also be inserted automatically
for the extract and ideally include some indicator of
by Word. The important steps are (a) end each line
the source of the extract.
of transcript by pressing the Enter key to force a line
break; (b) select all of the lines which you want num- • Speaker’s/contributor’s name Have the speaker’s/
bered using your mouse; and (c) you will find num- contributor’s name in bold and try to clearly separate
bering in the options for paragraphing. Remember this from what they say with some space.
that line lengths are arbitrary so you can force line • Blank space Because you may wish to make notes on
breaks where you feel that it is convenient. The line the transcript, it is important to include a copious
numbering convention helps you identify an excerpt amount of blank space to the right of the text. Judi-
from a transcript since the line numbers will be part cious use of the Enter key will help you with that.

M06 Introduction to Qualitative Re 51202.indd 140 07/01/19 4:06 PM

 CHAPTER 6 HOW TO TRANSCRIBE RECORDINGS 141

writing down words). Phonetic spelling systems could also be used but these require
that the reader has some sophistication. Alternatively, use ‘pseudo-phonetic’ forms
as used in comic books (e.g. ‘b’cuz I luv ya’). All of these things make transcription
increasingly unreadable. There are circumstances where the precise pronunciation of
words may be crucial, for example, where one speaker speaks mocking the accent
of another.
• Any non-transcribable features e.g. when the speaker coughs or clears their throat.
These are put in double brackets ((clears throat)).
• Remember that capitals are not generally used in Jefferson transcription other than
for proper names which with a capital letter. However, you will find this ‘rule’
ignored in some transcriptions. Capitals are used form words spoken distinctly
loudly.
• Points where there are any pauses. You will probably find it easier to mark these
with brackets enclosing x’s (e.g. (xx)) as an indication that the fine timing should be
entered in the next phase.
The rough transcription may not meet all of these criteria at this stage. This is not
particularly important. The final transcription phase provides opportunity to correct
any remaining inadequacies. Two important things to consider throughout this stage
are issues concerning effective communications with the reader and how the transcript
contributes to this.
Transcription technology has improved rapidly since the typewriter’s era and
magnetic recording. Digital recording equipment (and video recorders) mean that
digital files are readily available for computer processing. These, generally, are of
a very high quality. Computer software is available which make for easy copying
and editing of these files together with easy search facilities. Faces and voices may
be disguised and names edited out of the recording for ethical reasons. According
to Potter:
The simplest way to transcribe is to work with two windows on a computer screen,
one running the audio file, the other running the word processor. Audio programs
are available that allow a stepwise movement through the file using a physical
representation of the wave form that is ideal for timing pauses and noting overlaps.
(Potter, 2003, p. 82)

Step 3 Adding Jefferson symbols and transcribing sequencing accurately This transforms the
intermediate transcription into its final form by inserting detailed Jefferson transcription
symbols as appropriate. Up to this point you have done the basic straightforward work,
including adding some of the Jefferson notation. However, there may be fine detail to add
and the initial layout may not be the ideal for the reader to follow what is happening. This
final stage concentrates on the detailed sequencing of the conversation and not merely the
words said. These symbols include the square brackets indicating overlapping or simulta-
neous utterances by two or more people. Examples are to be found in the Stokoe (2003)
excerpt (cited earlier in this chapter) but they are so important in Jefferson transcription
that they bear repeating. So look at the following, which uses square brackets []:

11 Gary: where do you think that we ought to [go out tonight?]

12 Sarah: [is there anything] on the television?

M06 Introduction to Qualitative Re 51202.indd 141 07/01/19 4:06 PM

 142 PART 3 QUALITATIVE DATA ANALYSIS

Square brackets are used to show when Sarah and Gary are speaking together at the same
time. Single brackets are used to indicate when two people start talking at the same time:

17 Sarah: I wouldn’t mind watching something

18 Gary: [well
19 Sarah: [like a documentary

Equals signs (=) are used to indicate latching, which is where another speaker takes
over the conversation from another speaker without a pause:

28 Sarah: you always want what you want on=

29 Gary: =what mce?

Things can be more complex in conversation and more than one speaker can latch at the
same time in which case square brackets ([) might be needed to indicate this. For example:

28 Sarah: you always want what you want on=

29 Gary: = [what mce?]
30 Shane: = [too right]

Of course, there are other speech characteristics which could be included – those which
indicate how individual words are said in the recording. These are common in Jefferson
transcriptions and explained in Table 6.1. Refer to this for clarification whenever necessary
but it can also serve as a memory aid to ensure that you have considered all of the different
transcription possibilities. As with most things, experience is essential in order to ensure
quality transcriptions which are useful both to the researcher but also to the reader.
Pauses in conversation are common. Largely in Jefferson transcription they are
signalled using (.) or (0.5) to indicate different lengths of pause (see Table 6.1).
However, look at the following:

38 Sarah: you choose

39 (.)
40 Gary: i’m not bothered
41 (.)
42 Sarah: are you sulking?

In this exchange the pauses are not attributed to either Sarah or Gary since they
are given a separate line. They are pauses in the conversation and not pauses in what
either Sarah or Gary are saying individually. If the pause was clearly attributable to,
say, Sarah, then it would appear in a line indicated as being said by Sarah.

M06 Introduction to Qualitative Re 51202.indd 142 07/01/19 4:06 PM

 CHAPTER 6 HOW TO TRANSCRIBE RECORDINGS 143

Remember that there are limits to any transcribing system and the features of speech
that you wish to include may not be part of the system. If you need to add additional tran-
scription features, which may not be part of the Jefferson system, then this is a choice open
to you. Of course, you need to carefully describe and explain any such additional coding.
Certain computer programs are often recommended to students to help them with
the process of analysis though you can if you prefer to leave this aside until you have
developed basic transcription skills. They do have advantages like enabling a degree of
‘noise reduction’ to improve sound quality. They also permit editing of sound excerpts.
Perhaps more importantly, they can display the waveforms of the recorded sound.
This allows very precise measurement of pauses in conversation and where sounds are
exceptionally loud or soft. The main programs to consider are Audacity and Adobe
Audition. See the Additional Resources section at the end of this chapter.

When to use Jefferson transcription

The decision to use Jefferson transcription rather than an orthographic, secretarial or
playscript word-for-word record should be a serious consideration (Table 6.3). If the
research takes a conversation analysis perspective then the Jefferson transcription system
is essential since there is an intimate association of Gail Jefferson’s work with the devel-
opment of conversation analysis. But not all researchers, by any means, are interested
in the approach to conversation taken by Jefferson and her colleagues. Some qualitative
researchers may be interested in what the participants in the conversations have to say
about topics relevant to their research question. For example, if the researcher is interested
in the life histories of sex offenders it is the substantive material about each offender’s life
history which can be found on the recording which is important to that researcher. Issues
such as how the offender ‘recovers’ from errors made in telling his life history are not likely
to be a particular interest of the researcher. So it is questionable whether using Jefferson
transcription would be beneficial in this instance. The economic cost of using Jefferson
transcription has to be evaluated against the likely research gains of using it. The resources
spent on the Jefferson transcriptions might be better spent on other things.
It is mainly where speech is being researched as social action that Jefferson
transcription comes into its own. This does not mean that all qualitative research-
ers always gain from using such a fine-grained transcription. There is, in research,
always a question of the level of analysis which needs to be employed. For example,
in general, Jefferson transcription is not too useful when the data analysis employed
is thematic analysis. Such an analysis is based on developing relatively broad cate-
gories which describe the contents of interviews, focus groups and so forth. A secre-
tarial or playscript transcription will be almost certainly all that is required in these

TABLE 6.3 When to use Jefferson transcription

Definitely use Jefferson Possibly use Jefferson No advantage in using Jefferson transcription so
transcription transcription use orthographic transcription
• Conversation analysis • Discourse analysis (especially • Narrative analysis
Potter and Wetherell version)
• Interpretative phenomenological analysis
• Thematic analysis
• Grounded theory phenomenological analysis

M06 Introduction to Qualitative Re 51202.indd 143 07/01/19 4:06 PM

 144 PART 3 QUALITATIVE DATA ANALYSIS

circumstances. Research in the interpretative phenomenological analysis tradition

(Chapter 13), narrative analysis (Chapter 14), grounded theory (Chapter 8) and the-
matic analysis (Chapter 7) does not usually benefit from Jefferson transcription.
Given that transcriptions are often available for further analysis by other qualitative
researchers, there is a case for fully transcribing the data using the Jefferson system.
This is simply because Jefferson transcription maximises the additional information
accessible to the secondary analyst. Without it, the value of the transcript is reduced.
It is also an argument against using ‘stripped down’ (less complex) versions of
Jefferson transcription such as Jefferson ‘Lite’ which is recommended by some
(e.g. Parker, 2005). But these are issues of some controversy in qualitative psychology
for which no definitive answer is available.

Evaluation of Jefferson transcription

It needs to be remembered what the Jefferson system does. The words said are recorded in
ways which suggest something of the way that they sound, though this does not amount
to a fully phonetic rendition. So you will see some words written down as they sound in
dialect, for example, adding to the difficulty of reading them. Table 6.4 provides an evalu-
ation of the advantages and disadvantages of the Jefferson transcription system.
O’Connell and Kowal (1999) are somewhat critical of some aspects of transcription
that appear in the psychological literature. They go so far as referring to some of the
‘standardisation practices’ in transcriptions as pseudo-scientific:

• They point to instances of elaborate Jefferson transcriptions which contribute noth-

ing to the author’s interpretation of their data and which, often, are not referred to
in the publication. Why transcribe things that seem to add no value?
• They question the breaking up of words when the transcript is intended to be read
by others. Thus indications of the prolongation (e.g. wa::s), pitch movement and

TABLE 6.4 Advantages and disadvantages of the Jefferson transcription system

Advantages of Jefferson system Problems with the Jefferson system

1. It records talk as experienced by participants in the 1. While the Jefferson symbols can sometimes be used
conversation and so keeps the analysis focused on this very precisely such as times in tenths of a second, other
rather than merely the words used. symbols such as : are less carefully defined.
2. Analysis of conversational interaction is facilitated by the 2. It is restricted in terms of what aspects of interaction it
system compared with a secretarial transcript. deals with. For example, it is not good for coding emo-
3. Even if the words are the focus of their analysis, it allows tion.
other researchers to more adequately check the original 3. Although the system may be modified, it tends to set the
analysis as the transcript is closer to what is on the format of and the parameters for what is transcribed.
recording. 4. Its origins in the days of typewriters mean that it does
4. It has gained dominance over other methods so can be not capitalise on the potential of computers to use col-
regarded as the standard system of notation. our and a range of characters, fonts and sizes.
5. By forcing the researcher to spend time in transcribing, it 5. It is very time consuming for the researcher to use.
encourages a more thorough approach to analysis. 6. There is disagreement about the value of Jefferson tran-
6. It requires skilled transcribers and cannot be carried out scription even among discourse analysts.
by, say, secretarial assistants.
7. One can use the line numbers to rapidly refer to a par-
ticular part of the transcript.

M06 Introduction to Qualitative Re 51202.indd 144 07/01/19 4:06 PM