Intonation Units in Spoken Interaction (2007)

8/10/2019 Intonation Units in Spoken Interaction (2007)

1/34

Intonation units in spoken interaction:Developing transcription skills

JUURD H. STELMA and LYNNE J. CAMERON

Abstract

This paper describes the transcription process and the development of tran-

scription skills in a research project using recorded spoken interaction as its

main data. The spoken data was transcribed using intonation units, and the

paper traces the development of the first authors skills in identifying such

intonation units. Intertranscriber checks of transcription, involving three re-

searchers, were used to highlight ways in which the identification of intona-

tion units could be improved. Subsequent re-transcription of the data high-

lighted stretches of talk that included many hesitations, false starts, and

speech used to regulate ongoing spoken interaction. These features were

linked to low levels of intertranscriber agreement. It is argued that the ex-

isting literature on intonation units does not address how to best deal with

this quality of spontaneous spoken interaction. The paper concludes with an

agenda that may be used to improve the quality of transcription in similar

research projects, and to develop the transcription skills of the researchers

that are responsible for transcription.

Keywords: intonation units; transcription; transcription skills; intertran-scriber checks; spoken interaction; spontaneous talk.

1. Transcribing spoken interaction

In a 1979 landmark paper on transcription, Elinor Ochs (1979: 44) set out

to consider with some care the transcription process (emphasis added).

She went on to say,

We consider this process (a) because for nearly all studies based on [verbal] per-

formance, the transcriptions are the researchers data; (b) because transcription

is a selective process reflecting theoretical goals and definitions; and (c) because,

18607330/07/00270361 Text & Talk273 (2007), pp. 361393

Online 18607349 DOI 10.1515/TEXT.2007.015

6 Walter de Gruyter


2/34

with the exception of conversational analysis . . . , the process of transcription

has not been foregrounded in empirical studies of verbal behavior. (emphases

added)

A main focus of Ochss argument was to raise awareness about the eectsof various features of re-presenting spoken interaction in the form of tran-

scripts. She discussed at length the significance of page layouts, top-to-

bottom and left-to-right biases when reading transcripts, the represen-

tation of verbal versus nonverbal features of interaction, and the use of

transcription symbols. She not only made what was previously not-so-

obvious obvious, she also built a careful case for treating transcription as

something theoretical in nature.

According to Lapadat and Lindsay (1998: 5), the period following

Ochss contribution has been characterized by the following progressionof perspectives: the search for [transcription] conventions, acceptance of

a multiplicity of conventions, and [then] abandonment of standardization

in favour of contextualized negotiation of method (cf. also Lapadat and

Lindsay 1999). Standardized notation systems suggested in the literature

are usually designed to facilitate transcription according to a particular

perspective on spoken interaction. For example, conversation analysis

employs the Jeersonian notation system (cf. Atkinson and Heritage

1984: ixxvi), which is designed to explore the moment-to-moment

unfolding of turn-taking in interaction. Other systems are more eclectic,

but even so make their orientations explicit. For example, Gumperz and

Berenz (1993: 119) suggest a comprehensive set of conventions designed

to reveal the functioning of communicative signs in the turn-by-turn in-

terpretation of talk. Finally, some systems are designed to account for

more specialized features of spoken interaction, such as Du Boiss (1991)

suggested standard notations for transcribing talk into intonation units,

which have as their theoretical basis the connecting of mental processing

and speech production (cf. also Du Bois et al. 1993). This latter notationsystem will be discussed in detail later in this paper, as the transcription in

the current research in part draws on Du Boiss notation system.

The argument for standardized notation systems was followed by de-

bate about the usefulness of such standardization (Lapadat and Lindsay

1998). This debate has included attempts at empirically deriving agreed

upon notation systems (e.g., Dressler and Kreuz 2000), suggestions for

general design principles to base transcription conventions on (Du Bois

1991; Edwards 1993), and arguments against standardization; for exam-

ple, Muller and Damico (2002: 303) suggest that because transcriptioninvolves recurrent interpretive cycles that filter, shape, and even recreate

data, there is no such thing as the complete transcript, and Cook

(1990) argues that one cannot claim objectivity when trying to re-present

362 Juurd H. Stelma and Lynne J. Cameron


3/34

contextual features in transcription. What has emerged from this debate

is what may be described as aconsideredapproach, where standardization

is valued but not mandatory (cf. OConnell and Kowal 1999). Gumperz

and Berenz (1993: 119) exemplify this line of thinking when they pointout that their notation system is not designed to record everything

that can be heard, and adding: Yet, at the same time, we seek to

remain as comprehensive and attentive to detail as possible in showing

what the phenomenological or perceptual bases of our interpretations

are.

Just as in Ochss (1979) original contribution, the later debates have fo-

cused on transcription as re-presentation or interpretation, as something

which is more or less consistent, and as being informed by theoretical po-

sitions and analytic concerns. Where the term processis used, it is used in

Ochss sense that the act, or activity, of transcription should be guided by

an awareness of the interpretive and theoretical nature of transcribing

talk (e.g., Green et al. 1997; Muller and Damico 2002). This focus neatly

avoids the very messy reality of actually doingtranscription. Lapadat and

Lindsay (1998: 21) argue that empirical examination of transcription

processes, products, and their implications is singularly lacking in the re-

search literature. With some exceptions, the empirical studies that do

exist generally focus on transcripts as products. For example, Lapadat

and Lindsay (1998) compared the dierent transcriptions produced by

students enrolled in a graduate course on language development. Romero

et al. (2002) explored how subjects reading of the same segment of talk,

transcribed using dierent notation systems, compared to the original

audio recording of the talk. Roberts and Robinson (2004) studied the in-

tertranscriber (they used the term inter-observer) agreement of four re-

searchers, all transcribing the same segment of talk using the Jeersonian

notation system. Finally, OConnell and Kowal (1994: 140) explored how

transcripts inevitably include errors because like all language users, tran-scribers are in search of meaning. While valuable contributions, and a

step in the right direction as they oer the field of transcription a base of

empirical evidence for discussion, these contributions nevertheless lack

any focus on the actual processof transcribing talk.

A first exception to this picture is a number of comments in passing

made about the process of transcribing talk. For example, Lapadat

(2000: 204) suggests that the process of doing transcription . . . promotes

intense familiarity with the data, which leads to the methodological and

theoretical thinking essential to interpretation, and more specificallythat transcribing talk in interaction slows it down and focuses the

researchers interpretive eye, allowing him or her to become intensely

familiar with the data, and to draw meanings out of them (2000: 215).

Intonation units in spoken interaction 363


4/34

Lapadat (2000: 216) also suggests that a researcher should keep an audit

trail of decision points while transcribing, in the end using this record to

produce a code book of what was transcribed, how, and why. Again,

these are useful comments indicating the importance of approaching tran-scription in a systematic manner, but they do not address in any empirical

manner the process of actually doing transcription.

A more notable exception to the general picture is a small set of recent

studies that have explored the experiences of researchers doing transcrip-

tion work. Gregory et al. (1997: 295) comment on the progressively more

common practice of hiring transcribers, made possible by the increasing

funding available to qualitative research, and how, in their field of health

research, this creates a need for examining the emotional laboring and

work worlds of transcribers. In a series of articles, Tilley (cf. Tilley

2003a, 2003b; Tilley and Powick 2002) has explored, through interviews

and narrative accounts, the particular experiences of transcribers who

themselves are not researchers on a particular project, and the challenges

this creates within the research projects. Finally, Bird (2005: 246) pro-

vides a personal narrative of her own initial and growing relationship

with the process of transcription across a series of transcription tasks

she undertook. These studies provide holistic accounts of transcribers ex-

periences in qualitative research projects. They do not, however, address

the particular challenges involved in doing the narrow types of tran-

scription often required in discourse analysis or applied linguistics re-

search projects. The studies, nevertheless, suggest that doing transcription

is a complex process, and that to uncover the real complexity of this pro-

cess may well require further direct exploration.

Associated with the scarcity of studies on the process of doing tran-

scription is the observation, made by several authors, that the researchers

who actually do transcription are often postgraduate students or research

assistants (e.g., Lapadat 2000; Tilley 2003a, 2003b). With this realizationmust come the recognition that these research assistants are developing

their transcription skills concurrently with the activity of preparing tran-

scripts for a project (Bird 2005; Tilley 2003a, 2003b; Tilley and Powick

2002). Conspicuous in this regard is Lapadat and Lindsays (1998: 21) ob-

servation that transcription seldom appears as a topic of consideration in

the education of future researchers and practitioners who will be employ-

ing transcription in their work. McLellan et al. (2003: 73) come fairly

close to the mark when they point out that training data managers, tran-

scribers and proofreaders [tasks often done by research assistants] ishighly variable given the research structure, the setting, the type and

volume of data collected, the data produced, and the analytic approach

taken. This highlights a number of implications that need careful consid-



5/34

eration in research projects employing research assistants to do transcrip-

tion. For example, how much time it takes to prepare quality trans-

criptions may need to be balanced against the time needed to develop

transcription skills, and the involvement of more senior researchers inthe development of the transcription skills of the research assistants needs

to be made explicit.

The present paper addresses these gaps in the literature, on the one

hand documenting the process involved in the transcription of spoken in-

teraction using intonation units in an applied linguistics research project,

and on the other hand describing the gradual development of the tran-

scription skills of the researcher with primary responsibility for preparing

the transcripts for the project. To achieve this, we next provide an intro-

duction to the project from which this paper emerged, take a closer look

at the literature on intonation units and the transcription of these units,

and then outline our own view of what constitutes quality transcription.

This is followed by a detailed description of the transcription process we

engaged in and the intertranscriber checks we conducted. The final part

of the paper is an in-depth exploration of one researchers developing

skills in transcribing spoken interaction using intonation units, employing

as evidence both the products and the processes of transcription in the

research project. We conclude the paper by presenting an agenda for the

development of transcription skills and quality of transcription that

emerged from our close engagement with the transcription process in our

project work.

2. The research project

The transcription process and the development of transcription skills de-

scribed in this paper took place in the context of a research project whosemain aim was to explore the dynamics of metaphor use in conciliation

talk.1 The spoken data in this project consisted of about 3.5 hours of

video- and tape-recorded spoken interaction between two participants; a

perpetrator of a bombing and the daughter of a victim of this bombing.

For the most part, the interaction consisted of extended speaker turns,

with the listener producing occasional back channel responses. Only

sometimes did the interaction consist of shorter, more interactive turns.

The spoken interaction was nevertheless unscripted, and thereby sponta-

neous. The purpose of the interaction between the two participants was tolisten to each other, or to understand the story or journey of the Other,

as experienced before, during and after the bombing (Cameron and

Stelma 2004; Cameron forthcoming).



6/34

3. Intonation units in spoken interaction

The project team, i.e., the first and second authors of this paper, agreed at

an early stage to transcribe the conciliation talk using intonation units.We were attracted to intonation units because there is evidence that these

units can account for the inherently dynamic interplay between speaking

and thinking (Chafe 1994, 1996), thereby responding to the research aim,

which was to explore the dynamics of metaphor use in talking and think-

ing in a context of post-conflict reconciliation.

Our use of the intonation unit (henceforth IU or IUs) is based primar-

ily on Wallace Chafes (1994, 1996) extensive work on naturally occurring

language data from a range of discourse contexts and languages, building

a case for a relationship between consciousness and language. In particu-

lar, Chafe (1996: 39) suggests that consciousness is a process in which

remembering, imagining, evaluating, and speaking come together to pro-

duce what we know as thought and language.2 One of the constant prop-

erties of consciousness, according to Chafe, is that it has a focus and a

periphery, much like human focal and peripheral vision. Moreover, focus

is a dynamic property, as there is a restless movement from one focus to

the next; consciousness does not stand still (1996: 38). In terms of speak-

ing, the restless movement, or dynamic nature, of consciousness manifests

itself in speaking in the form of IUs (Chafe 1994, 1996). Chafe (1996: 40)

underlines the dynamic nature of consciousness, and the dynamic rela-

tionship between consciousness and speaking, by pointing out that IUs

(and consciousness) are produced in a series of brief spurts, typically be-

tween one and two seconds long.

This coupled relationship between consciousness and IUs, produced

in spurts, is supported by a number of arguments for the IU as a

cognitive unit (cf. Park 2002), as well as arguments for how the IU,

and intonation more generally, does work in spoken interaction (cf.Couper-Kuhlen 2001; Wennerstrom 2001). One argument is based on

the observation that IUs, on average between 4 and 5 words long (Chafe

1994: 65; Crystal 1969: 256), are smaller than clauses. For this reason,

IUs may be constrained by something other than grammatical structure.

Cognitive constraints, such as how much information can be active in

consciousness at one time, might be such an alternative explanation

(Chafe 1994). On the other hand, some IUs are clause length. Since

clauses often encode propositions, it may be argued that IUs themselves

are vehicles for basic cognitive processes of information storage anddiscourse processing (Park 2002: 639). However, this observation may

equally well be explained as a confluence between grammatical and into-

nation structure in discourse processing (cf. Ford and Thompson 1996).



7/34

A possibly more intriguing observation is how IUs seem to be encoding a

single message (Kreckel 1981), a single new idea (Chafe 1994), or a single

unit of information (Halliday 1967). In Chafes (1993: 37) framework,

this includes substantive IUs, which are the contentful stretches of speechthat include ideas of people, objects, events, and states, and regulatory

IUs, which function, in one way or another, to regulate the flow of infor-

mation. There is also evidence that dierent tunes marking boundaries

between so-called intonational phrases,3 or sets of intonation units, can

be used to interpret discourse-relevant relationships. That is, boundary

tunes between intonational phrases communicate meaning to interlocu-

tors (Pierrehumbert and Hirschberg 1990). IUs are thus seen as playing

both cognitive and interactive roles in spoken discourse.

This perspective on consciousness and language is consistent with the

second authors conceptualization of the dynamic role of metaphor use

in thinking and speaking (Cameron 2003). Hence, we felt that transcrip-

tion using IUs was appropriate for representing the dynamic unfolding of

mental and interactive processes in spoken data, and for facilitating the

investigation of metaphor use in the conciliation talk.

4. Identifying and transcribing intonation units

A central challenge, and a nontrivial one we shall claim, is the task of

identifying IUs in spoken interaction. Cruttenden (1986: 36) observes

that many linguists assume that the phonetic correlates of boundaries be-

tween intonation-groups are far more straightforward than they actually

are. Cruttendens definition of intonation groups is similar to Chafes

IUs, as well as our own understanding of IUs. Cruttenden further pro-

poses that the diculty of identifying boundaries between intonation

groups depends on whether the verbal data is speaking prepared textsor more spontaneous speech. Cruttenden also suggests that adults into-

national competence can be very variable. Background knowledge in the

topic of a conversational event, whether a spoken event was rehearsed,

and probably also the extent to which new ideas are being generated by

the speech, may all aect the identifiability of IUs (Wichmann 2000).

For example, the IUs of an experienced news anchor, reading from a tele-

prompter, would presumably be more distinct, and thereby more easily

identifiable, than the IUs of the spoken interaction of, say, two people

who meet by chance in the street.The literature is quite clear on how complete IUs may be recognized.

Chafe (1994: 58) suggests that the following six features be used in identi-

fying boundaries between IUs: (i) changes in fundamental frequency, or



8/34

pitch (in musical terms, each key on a piano represents a dierent pitch),

(ii) changes in duration or tempo (manifesting itself as shortening and

lengthening of syllables and words), (iii) changes in intensity or loudness

(including stress and/or accents), (iv) alterations between vocalization andsilence (i.e., pausing), (v) changes in voice quality (e.g., creaky voice),

and (vi) changes in speaker turn. Chafes account thus produces six char-

acteristics of prototypical IUs:

1. Pitch usually includes a resetting of the pitch baseline (as in a step-

up or step-down in the pitch level) and a recognizable final pitch

contour (e.g., falling or rising);

2. Duration usually includes increased tempo at the beginning (as in

a shortening of syllables and/or words), and then a gradual slowingdown toward the end (as in a lengthening of syllables and/or words);

3. Intensity usually includes one or more syllables and/or words spoken

more loudly;

4. Pausingis often preceded or followed by pausing (but may also con-

tain pauses within its boundaries);

5. Voice quality sometimes begins or ends with a creaky voice or

whispering;

6. Speaker turnmay sometimes be associated with a change of speaker.

Cruttendens (cf. 1986, 1997) description of criteria for identifying what

he calls intonation groups adds useful detail to Chafes characteristics.

In our own use of IUs as a tool for making sense of the dynamics of

metaphor use, we have come to view Cruttendens intonation groups

and Chafes intonation units as similar units of speech. There are some

dierences, however. Cruttenden makes a distinction between external

and internal criteria for intonation groups. Cruttendens (1997: 2934)

external criteria for identifying intonation groups include: (i) pausing

(unfilled and filled), (ii) anacrusis (increase in speech rate at the start ofan intonation group), (iii) lengthening of syllables (at the end of intona-

tion groups), and (iv) changes in pitch level and/or pitch direction on un-

accented syllables.

Cruttenden distinguishes these external criteria from prosodic features

that are internal to intonation groups. For example, he argues that a

step-up and step-down in pitch level is sometimes associated with an ac-

cented syllable, in which case it should not be interpreted as an intonation

group boundary. To help resolve potential ambiguity, Cruttenden (1986:

34) points out that accents in connected speech normally fall only on syl-lables that are lexically stressed. Hence, a step-up or step-down in pitch

level followed by a lexically stressed syllable or word is not an intonation

group boundary. Cruttendens internal criteria for intonation groups,



9/34

then, are (i) an intonation group must contain at least one stressed sylla-

ble, and (ii) there must be a pitch movement to or from at least one ac-

cented syllable.

Another potential ambiguity is that pausing, anacrusis, and syllablelengthening may all, according to Cruttenden, happen in the middle of

intonation groups, especially in the case of hesitation. To distinguish

between intonation group boundaries and hesitation, Cruttenden (1997:

35) suggests the following heuristic: if the features of pause, and/or ana-

crusis, and/or syllable lengthening divide an utterance into two part-

utterances either one of which does not have the minimum internal struc-

ture of an intonation group, then any combination of these features is

taken as a hesitation rather than a boundary between intonation groups.

The later parts of this paper will show how this added detail, provided by

Cruttendens external and internal criteria, helped us make sense of IUs

in our own data.

A final point we would like to make about the identification of IUs is

related to the distinction between prepared and spontaneous talk, intro-

duced earlier in this paper. There is a question about the extent to which

IUs, with complete intonation contours, as described by, e.g., Chafes

(1994) characteristics listed above, reflect spoken interaction adequately.

Cruttenden (1986: 36) points out that in the case of spontaneous speaking

any clear and obvious division into intonation-groups is not so apparent

because of the broken nature of much spontaneous speech, including as

it does hesitation, repetitions, false starts, incomplete sentences and

sentences involving a grammatical caesura in their middle. Wichmann

(2000: 21) also observes that spontaneous speech will to a greater or

lesser extent display syntactic and prosodic disfluencieshesitations,

repetitions, incomplete utterances. In sum, the extent to which IUs are

complete will vary a great deal when speakers have to deal with the

real-time dynamics of spontaneous speech.The research by Lindsay and OConnell (1995), reviewed above, sug-

gests that hesitation phenomena and sentence fragments can cause partic-

ular diculties in transcription of spoken discourse. Even so, there is little

information in the transcription literature on how to deal with this often

broken, or fragmentary, nature of spontaneous speech. The literature

does mention the possibility of incomplete intonation contours, called

fragmented (Chafe 1994) or truncated (Du Bois et al. 1993) IUs. How-

ever, there is little detail on how such fragmented or truncated IUs are

recognized, or how the presence of these aects the identification of morecomplete intonation contours. As we will see, the presence of such trun-

cated IUs was one of the features of our data that caused most problems

in the project transcription process.



10/34

5. Quality and the transcription process

Given the challenges we faced in the transcription of the project data, a

statement about our position on the transcription process is necessary atthis juncture. Our position may be summarized as post-positivist and re-

alist. We reject the simplistic notion of real, or what Mishler (1991, cited

in Lapadat and Lindsay 1999: 7374) calls nave realism, in which a

transcription is held to be a true re-presentation of a discourse event. In-

stead, we align ourselves with the social realist approach developed by

Sealey and Carter (2004: 126), which requires us to take account of the

irreducible subjective realities of human consciousness and being while

also allowing that degrees of objectivity are possible in applied linguistic

research. We can thus avoid the inaction dictated by an extreme interpre-

tivist or constructionist position. To some extent, every researcher who

listens to an audio-recorded conversation will hear something dierent;

people vary in the acuity of their hearing, in their awareness of melody

and intonation, and in other neurological or physical factors. However,

this variation does not rule out reaching useful levels of agreement be-

tween researchers on what was said in audio-recorded talk, or useful dia-

logues between researchers to improve the quality of a research process

such as transcription. People share sucient neurological and physical

features, and, in the research context, usually sucient common sociocul-

tural and language understandings, for a level of agreement to be reached

that will produce a transcription that can be good enough for the partic-

ular research purposes. This, then, is similar to what we earlier referred to

as the considered approach that seems to have emerged, again echoing

Gumperz and Berenzs (1993: 119) advice to remain as comprehensive

and attentive to detail as possible in showing what the phenomenological

or perceptual bases of our interpretations are.

Hence, researchers working together on a project will need to setacceptable levels of agreement, or alternatively detail how disagreement

is dealt with to increase the consistency of later transcription processes.

Meeting acceptable levels of agreement, and/or dealing with disagree-

ment in ways that improve the consistency of transcription processes, will

be a measure of the quality of a transcription. The quality of a research

report will be enhanced by explicit statements about what is selected from

all that is possible to transcribe, and why, and about the levels of agree-

ment reached in transcription, and how. From our perspective in this pa-

per on the development of transcription skills, the process of learningto transcribe will include: learning about how irreducible subjective real-

ities aect what is heard and transcribed; learning about inter-researcher

variation; developing skills and strategies to work with these in the pro-



11/34

duction of transcriptions to agreed levels of quality; acquiring knowledge

and skills to write about these decisions and processes in written reports

of the research.

6. Stages of the transcription process

The project transcription process included four stages. The first tran-

scriber (the 1st author of this paper) prepared an initial transcript of a

five-minute sample segment of the spoken data. This was followed by

two additional researchers, one being the second author of the paper, pre-

paring separate transcripts of the same five-minute sample segment of the

data. The first transcriber then used the results of these intertranscriber

comparisons to reflect on his transcription skills, and subsequently to de-

velop his skills in identifying IUs further. This included the production of

what we have called an enhancedtranscription of the five-minute sample

segment of data. In the last stage of the transcription process, the full spo-

ken data set was transcribed by the first researcher.

Table 1 provides an overview of the dierent transcript versions of

the five-minute sample segment, as well as who prepared them. These

are referred to in the following discussion. See Appendix 1 for the tran-

scription conventions and Appendix 2 for short illustrative segments of

the transcripts.

The first transcriber was employed as a research fellow, and had overall

responsibility for preparing the project transcripts. Prior to the start of the

project, this researcher had experience of transcribing bilingual spoken

classroom interaction, as part of his doctoral study, using IUs. He had

also done some transcription using IUs for an earlier pilot stage of the

project reported upon here. However, the use of IUs was not central to

the aims of these previous research experiences, and he did not have anyformal training in the transcription of IUs or any other from of transcrip-

tion. The second transcriber (the second author of this paper) was the

principal investigator in the research project, with overall responsibility

for processes and outcomes of the research (of which transcription was

Table 1. Transcriptions of the five-minute sample segment

Transcriber Version

First transcriber Transcript 1Second transcriber (experienced, but not with IUs) Transcript 2

Third transcriber (experienced with IUs) Transcript 3

First transcriber Enhanced transcript



12/34

only one component). She had no first-hand experience of transcription

using IUs, but had extensive experience of transcribing discourse using

other discourse units and notational systems. The third transcriber acted

in an advisory capacity, oering feedback on the aims, processes, andoutcomes of the project as a whole. He had extensive experience of tran-

scribing spoken interaction using IUs, both as a doctoral researcher and

later on a number of research projects using spoken interaction as pri-

mary data.

The initial transcript (Transcript 1) of the five-minute sample segment

of spoken interaction was prepared from a digitized audio recording,

played back using VoiceWalker software (Du Bois 2000). This software

is designed to automatically step through an audio file; playing a few

seconds of the audio file, then taking a small step back before again

playing a few seconds. This stepping through the audio file, in a two

steps forward and one step back manner, facilitates the transcription of

spoken interaction without the use of the more traditional tape recorder

and pedals system. The software allows both the forwards and backwards

steps to be adjusted, and also allows the researcher to repeat steps when

needed, thereby allowing for dierent types of spoken discourse, varying

quality of recordings, and researchers dierent styles of transcription.

Extract (1) shows a segment from the beginning of Transcript 1, illustrat-

ing the product of the first step of the transcription process.

(1) (Sample extract from Transcript 1 [by the first transcriber])

there is no right,

. . for me,

. . to sit here and be forgiven.

. . . if you understand me.

I mean in a sense there is the political thing,

I knew what I was doing,

. . and I would even defend actions I have taken,

The second and third transcribers prepared their transcripts under some-

what dierent conditions. They were given a verbatim transcription of the

five-minute segment of spoken interaction, presented as a continuous

block of text as per Extract (2).

(2) (Sample verbatim transcript segment for intertranscriber checks)

there is no right for me to sit here and be forgiven if you understand

me I mean in a sense there is the political thing I knew what I was

doing and I would even defend actions I have taken

Consistent with the aims of the intertranscriber checks, and working with

a digitized audio file, the task of the second and third transcribers was to



13/34

segment the continuous block of text into IUs. Although a somewhat dif-

ferent task, we felt that this approach was justified, especially considering

previous research that has established high agreement between experi-

enced transcribers identification of words and sounds only (Roberts andRobinson 2004). The second and third transcribers were free to change

the words in the text they received according to their own hearing.

It was understood, by all the researchers involved, that the project was

working with Chafes (1994) notion of IUs. For this reason, no formal

definition of IUs was provided, and the intertranscriber checks were

based on rules as published rather than rules as agreed through discus-

sion (cf. Oelschlaeger and Thorne 1999). Just as the first transcriber did,

the second and third transcribers also indicated the transitional continuity

of each units intonation contour. Transitional continuity is defined as

the degree of continuity that occurs at the transition point between one

intonation unit and the next (Du Bois et al. 1993: 53). We will discuss

the marking of transitional continuity in greater detail later in the paper.

At this point, it is sucient to note that there are three types of transi-

tional continuity: final, continuing, and appeal. If an IU did not have an

identifiable transitional continuity, it was marked as truncated (see dis-

cussion of truncated IUs below). The specific instructions for the inter-

transcriber checks were:

Insert line breaks corresponding to intonation unit boundaries (press the RE-

TURN key at the end of each intonation unit you identify);

Mark the end of each intonation unit for transitional continuity. Please use:

a periodfor final

a comma for continuing

a question markfor appeals

Alternatively, mark intonation units as truncated. Please use --

The second transcriber additionally recorded pauses and marked promi-nent words. The second transcriber listened to the audio file using the

same VoiceWalker software as the first transcriber. The third transcriber,

being a Mac user, commented that there is no Mac version of Voice-

Walker, so Im just rewinding on Quick Time.

7. Intertranscriber checks

We struggled for some time to find a meaningful way to compare the dif-ferent transcript versions. On the face of it, what we were doing seemed

like a straightforward inter-rater check that could be reported in terms

of levels of agreement and would tell us something about the reliability



14/34

of Transcript 1 prepared by the first transcriber. This would be commen-

surate with a positivist notion of transcription, where talk is an observ-

able behavior that can be transcribed completely and accurately (Lapadat

2000). There are, in the literature, some precedents for such inter-rater

checks, some of which have developed quite sophisticated statistical mea-sures (e.g., Roberts and Robinson 2004).

Table 2 shows a simple count of IUs identified by each of the tran-

scribers, as well as how many IUs were transcribed as identical. The re-

sults show that the two intertranscribers identified more IUs overall, 186

and 218 IUs respectively, as compared to the first transcribers 159 IUs.

This result was not very encouraging. Moreover, of the 186 IUs that the

second transcriber identified, only 93 IUs were identical in the first re-

searchers transcript. Table 2 also shows that the numbers are similar

when the first and third transcribers performances are compared.Not only were these results discouraging, we also felt that this positivist

view did not fit the purpose of our intertranscriber checks, which was to

improve the quality of transcription in the conciliation talk project. That

is, we felt that any comparisons between the transcripts should facilitate

the development of the first researchers transcription skills.

Extracts (3) and (4) contain parallel representations of a segment of

talk from Transcripts 1 and 2 and Transcripts 1 and 3, respectively. The

annotation in these parallel transcripts illustrates the approach we finally

used in order to produce a more meaningful comparison of the three tran-

scripts. The lines in these parallel transcriptions are numbered to facilitate

line by line comparisons of IUs identified by the three transcribers. The

additional annotation, in the right-hand column in Extracts (3) and (4),

records dierent types of agreement, using Transcript 1 as the point of

reference. We recorded the following types of agreement:

identical IUs;

IUs with a common initial boundary;

IUs with a common final boundary; talk identified as two or more IUs by the intertranscribers, but that

has the same overall initial and final boundaries as a single IU identi-

fied by the first transcriber.

Table 2. Total number of IUs identified and IUs agreed as identical

Total IUs identified IUs identical as

in Transcript 1

Transcript 1 159

Transcript 2 186 93

Transcript 3 218 98



15/34

(3) (Comparison of Transcript 1 and Transcript 2)

Transcript 1 Transcript 2

1 and you know Im er -- and you know Im -- 2 er er3 er Im aware thats like er -- Im aware thats like er -- 4 its a part of your sort of spiritual make up to, its a part of your sort of spiritual make

5 . . to confront . . the situation. . . to confront . . . the situation, 67 . . er and er . . move on from it. . . .(1.0) er and er -- 8 . . .(1.0) move on from it.9 but again er -- but again er -- 10 . . .(1.0) dealing you know, . . .(2.0) dealing -- 11 you know,12 like having to handle that, like having to handle that, 13 you know and -- . . .(1.0) you know and -- 14 . . or the enormity of it, or the enormity of it,

(4) (Comparison of Transcript 1 and Transcript 3)

Transcript 1 Transcript 3

1 and you know Im er -- and you know Im er, 23 er Im aware thats like er -- er Im aware thats like er, 4 its a part of your sort of spiritual make up to, its a part of your sort of spiritual make5 . . to confront . . the situation. to confront-- 6 the situation,7 . . er and er . . move on from it. er and er, 8 move on from it.

9 but again er -- but again er, 10 . . .(1.0) dealing you know, dealing you know like, 1112 like having to handle that, having to handle that. 13 you know and -- you know and [uh], 14 . . or the enormity of it, or the-- 15 enormity of it.


16/34


17/34

heard only one, and one instance where the second transcriber identified

as many as four IUs corresponding to the one that the first transcriberidentified. Finally, there were 17 instances where the first and second

transcribers identified the same initial boundary and 11 instances where

they identified the same final boundary (see Extracts [3] and [4] for illus-

tration). The final row in Table 3 records the number of IUs uniquely

identified by the second transcriber. Table 4 shows the same comparison,

but now between the first and the third transcribers.

We found this more subtle analysis more productive for our purposes

than a straightforward inter-rater check. For one, it begins to explain

why the first transcriber identified only 159 IUs, as compared to the sec-ond transcribers 186 IUs and the third transcribers 218 IUs. Although

the absolute level of agreement reported in Table 2, counting only IUs

fully agreed upon, appears low, the first transcribers identification of

Table 3. Identification of IUs in Transcript 2 as compared to Transcript 1

Instances when the first and second transcribers agreed on the identification of an IU 93

Instances when the same talk was identified as one IU by the first transcriber, but as

two IUs by the second transcriber

21


three IUs by the second transcriber

4


four IUs by the second transcriber

1

Instances where both the first and the second transcriber agreed on the initial

boundary of an IU

17

Instances where both the first and the second transcriber agreed on the final

boundary of an IU

11

IUs uniquely identified by the second transcriber 7

Table 4. Identification of IUs in Transcript 3 as compared to Transcript 1

Instances when the first and second transcribers agreed on the identification of an IU 98


two IUs by the second transcriber

33


three IUs by the second transcriber

6


four IUs by the second transcriber

2

Instances when the same talk was identified as one IU by the first transcriber, but asfive IUs by the second transcriber

1

Instances where both the first and the second transcriber agreed on the initial

boundary of an IU

9

Instances where both the first and the second transcriber agreed on the final

boundary of an IU

9

IUs uniquely identified by the second transcriber 5



18/34

boundaries between IUs appeared promising. That is, the more subtle

analysis shows that the real problem, as it were, was that the first tran-

scriber did not identify enough, or all, boundaries between IUs, thereby

identifying fewer IUs overall. This was productive for the purpose of de-veloping transcription skills, as it indicated a specific area that the first

transcriber could pay attention to, namely improving his identification of

boundaries between IUs.

8. Developing transcription skills

In order to improve his skills in recognizing IU boundaries, the first re-

searcher again consulted Chafes six characteristics of IUs, reviewed in an

earlier part of this paper. Chafe (1994: 59) uses the following typical IU

to illustrate four out of his six characteristics:

. . and so the hall isreal long%.

. . .(.36) [the next intonation unit]

Pausing: This IU is preceded by a very brief pause (indicated by two dots)

and followed by a slightly longer measured pause of 0.36 seconds.

Duration: The IU starts with three words spoken with an accelerated pace(transcribed in small print: and so the). Such accelerated speech at the begin-

ning of an IU is also called anacrusis (Cruttenden 1986: 39). Note also

that the final word in the IU is transcribed with an equal sign following

the vowel (long), indicating that this is spoken more slowly. There is

therefore a general pattern of slowing down from the beginning toward

the end of the IU. The significance of this for identifying IU boundaries

is that when the speech speeds up again this may indicate the start of a

next IU.

Pitch: Of the three prominent words in the IU (hall, real, and long),

the first word (hall) has the highest pitch (as measured in hertz), the sec-

ond word (real) has a slightly lower pitch, and the final prominent word

(long) has the lowest pitch. This, then, is a declining pitch pattern within

the IU. IUs may often be part of yet larger units, spanning several IUs,

and with a marked pattern of declining pitch from IU to IU. These larger

units are variously called declination units (Schuetze-Coburn et al. 1991)

and paratones (Yule 1980). Declination units may help to identify indi-

vidual IU boundaries, with each successive IU marked by a slight reset-ting of the pitch baseline within an overall pattern of declining pitch

across units (cf. Schuetze-Coburn et al. 1991). The IU is also character-

ized by a falling final pitch contour (marked by a period).



19/34

Voice quality: The end of the last word of the IU (long) is spoken with a

creaky voice (marked by the percent sign).

A similarly rigorous identification of IUs in the project data encounteredsome diculties. The first transcriber initially found Chafes use of a pe-

riod to mark sentence-final falling pitch, a comma to mark contours that

are not sentence-final, and a question mark to represent an appeal, po-

tentially confusing. Du Bois et al. (1993) was helpful here. They also use

period, comma, and question mark to distinguish intonation contours as

being either final, continuing, or an appeal. According to Du Bois et

al. (1993: 53), thistransitionalcontinuity of intonation contours will have

various realizations, one of which is the final pitch movement of IUs. In

the above example, from Chafe, the falling pitch contour at the end of theIU is associated with a final intonation contour. In sum, the first re-

searcher resolved to re-learn the use of these familiar symbols, the punc-

tuation marks, to describe the intonation contours of IUs, and he used fi-

nal pitch movement of IUs as a heuristic to aid the identification of these

intonation contours. At the same time, the identification of final pitch

movement turned out to be helpful in the identification of IU boundaries.

The first transcriber encountered two additional diculties. Firstly, so-

called changes in voice quality at the beginning and end of IUs (e.g., Cha-

fes creaky voice) were dicult to recognize. Secondly, distinctly discern-ible declining patterns of pitch level were not found within or across the

identified IUs. We suspect that these may be rather subtle characteristics

of IUs, and that extensive experience may be needed before one may be

able to use these productively for identifying IU boundaries. We did,

however, notice that IU boundaries were sometimes associated with a

slight re-setting of the pitch baseline; i.e., IUs sometimes started with a

slightly lower or higher pitch than the preceding talk. In addition, we also

added prominence as a characteristic of IUs in our data. That is, we often

saw that IUs in our project data contained at least one prominent syllableor word.

The careful re-transcription of the five-minute sample segment, allow-

ing for the above subtleties, resulted in the following characteristics of

IUs identified in the project data:

1. Pitch: Beginning with a re-setting of the pitch baseline and having a

recognizable final pitch movement

2. Duration: Beginning with a shortening of syllables/words (i.e., ana-

crusis) and ending with lengthened syllables/words3. Prominence: Including at least one clearly prominent word, achieved

with altered intensity, pitch, or a combination of both

4. Pausing: Being preceded and followed by a pause



20/34

Note that these characteristics include elements not only from Chafe

(1994). We also include one of Cruttendens (1986, 1997) internal criteria;

i.e., that there should be at least one prominent word or syllable in an IU.

We accept, however, that the way we recognize prominence is somewhatdierent than the pitch movement to or from an accented syllable that

Cruttenden uses as an internal criterion for identifying intonation groups.

An important tool helping the first transcriber to arrive at, and practice

the use of, the above criteria in his identification of IUs, was what we call

an enhanced transcription. The conventions used in this enhanced tran-

scription build on Chafes (1994) narrow transcription of IUs, as illus-

trated above; include elements from Du Bois et al. (1993); and add our

own notation where necessary. Example (a) below illustrates the en-

hanced transcription conventions, and thereby also the four criteria we

used to identify IUs in the data. The enhanced conventions are, then, a

way to make transparent the decision-making involved in the identifica-

tion of IUs in the five-minute sample segment of the conciliation talk. By

contrast, the conventions used in the final transcription of the full data set

were guided by what we needed to re-present for our exploration of met-

aphor dynamics. Hence, in the later full transcription we used the more

economical conventions illustrated by Example (b).

a. . . .(2.0) _back in the ^moment\b. . . .(2.0) back in the moment.

In Example (a), there is a two second pause at the beginning of the IU

(marked by the three dots and parentheses indicating the length of the

pause). At the beginning of the IU there is a resetting of the pitch baseline

(marked by the underscore). The first three words of the IU are acceler-

ated (anacrusis; marked by the smaller font size). The first syllable of the

word moment is prominent (marked by the caret). The first and second

vowels of the word moment are both lengthened (marked by the equalsigns). Finally, the IU ends with falling final pitch (marked by the back-

ward slash). In Extract (6) the conventions are applied to a longer stretch

of talk from the first researchers enhanced transcription of the five-

minute sample segment of conciliation talk. The line numbering in this

extract, as well as in subsequent extracts, corresponds to the line number-

ing in the enhanced transcript (see Appendix 2 for an illustrative

segment).

(6) (Illustration of conventions used in the enhanced transcription)39 Pat: and you ^know Im er --

40 . . .(1.0) er Im ^aware thats like er --

41 . . its a part of your sort ofspiritual ^make up to\



21/34

42 . . .(2.0) to con^front/

43 Jo: [hmh]

44 Pat: . . .(1.0) _[the] situ^ation/

45 . . . _and er --46 . . . move on from it\

47 . . .(1.0) buta^gain/ --

48 _er

49 . . .(3.0) _^dealing /--

50 _you know\

51 like. . ^having to handle that\

52 . . . _you know\

53 and er

54 . . .(1.0) _or the. . e^normity of it\

A number of things are evident from an examination of the enhanced

transcription in Extract (6). First of all, not many IUs encompass all the

characteristics of our typical IU. That is, any single IU is identified using

only a subset of the characteristics. This observation was uniform across

all parts of the five-minute sample segment. It is also evident that the talk

contains many incomplete utterances (often marked as truncated IUs)

and several so-called regulatory IUs (e.g., you know and but again; cf.

Chafe 1993). This less fluent nature of the talk was not uniform acrossthe five-minute sample segment of conciliation talk. However, we noticed

that less fluent stretches of talk were associated with lower levels of

agreement in the earlier intertranscriber checks, and also that the first re-

searchers enhanced transcription of these stretches was very dierent as

compared to his initial transcription.

Extract (7) exemplifies how the transcription of another stretch of less

fluent talk changed a great deal, from the first researchers initial tran-

scription and the enhanced transcription. Just as in Extract (6), the en-

hanced transcript in Extract (7) includes a number of truncated IUs (lines146, 152, 154, and 155) and some regulatory IUs (you know in lines 152

and 157).

(7) (Transcript 1 and the enhanced transcript for a less fluent stretch of

talk)

Transcript 1 Enhanced transcript

146 and that . . that er -- . . and that --

147 . . .(1.0) _that um\148 that pain that loss . . was shared

by . . by everyone.

. . _that ^pain

149 _that ^loss



22/34

150 . . . _was ^shared by\

151 . . by everyone\

152 you know and after that, . . .you know an- --

153 andafter that154 . . er the pain on on every side, . . .(1.0) _um --

155 _ the^pain on --

156 on^every side\

157 you know, . . _you know\

158 I felt it. . . _I felt it\

Extract (8) shows a stretch of talk that was more fluent, in the sense that

it includes fewer truncated IUs (only one in Extract [8]: when you come-- in line 12) and fewer regulatory IUs (only one clear example in Extract

[8]: I mean in line 13). Such more fluent stretches of talk were also

common in the five-minute sample segment.

(8) (Enhanced transcription of a more fluent stretch of talk)

8 Pat: theres ^no right

9 . . .(1.0) for ^me

10 . . .(2.0) _to sit here and be for^given\

11 . . .(2.0) if you understand me

12 Jo: hmh

13 Pat: _ I mean

14 in a ^sense theres the political thing

15 Iknew what I was ^doing\

16 . . and I would even de^fend actions Ive taken

17 _etcetera

18 _^but

19 _when you come --

20 when it comes ^down to it\ 21 . . I am ^sitting with somebody whos aected by\

22 . . . _^my actionns

The more fluent stretches of talk, as in Extract (8), were also dierent in

that they showed high levels of agreement in the earlier intertranscriber

checks, and the transcription of these stretches of talk did not change

much between the first researchers initial transcription and the enhanced

transcription. The parallel representation in Extract (9) illustrates how the

transcription of the more fluent stretch of talk in Extract (8) did notchange much between the initial transcription (Transcript 1) and the en-

hanced transcription; only two IUs, in lines 14 and 22, are unique to the

enhanced transcription.



23/34

(9) (A more fluent stretch of talk from Transcript 1 and the enhanced

transcript)

Transcript 1 Enhanced transcript8 there is no right, theres^no right

9 . . for me, . . .(1.0) for ^me

10 . . to sit here and be forgiven. . . .(2.0) _to sit here and be

for^given\

11 . . . if you understand me. . . .(2.0) if you understand me

12 hmh

13 I mean in a sense there is the

political thing,

_I mean

14 in a^sense theres the political thing

15 I knew what I was doing, I knew what I was ^do:ing\

16 . . and I would even defend

actions I have taken,

. . and I would even de^fend

actions Ive taken

17 etcetera. _etcetera

18 but, _^but

19 when it come -- _when you come --

20 when it comes down to it, when it comes ^down to it\

21 I am sitting with somebody

who is aected by . . my

actions.

. . I am ^sitting with somebody

whos aected by\

22 . . . _^my actionns

It seems, then, that stretches of talk that included a lot of truncated and

regulatory IUs were particularly challenging for the first transcriber. Suc-

cessfully dealing with these stretches of talk was therefore instrumental

for the development of his transcription skills. By contrast, the transcrip-tion of stretches of talk with fewer truncated and regulative IUs were as-

sociated with higher levels of intertranscriber agreement, and changed

much less across the dierent stages of transcription.

The final stage, then, was the transcription of the full data set, putting

the skills that the first transcriber had developed to work. This transcrip-

tion did not use the enhanced transcription conventions. Rather, the proj-

ect now reverted to a more simple set of conventions agreed as sucient

for the purposes of the research project. Extract (10) illustrates the con-

ventions used in this final transcription.

(10) (Illustration of conventions in the final transcription of the concili-

ation talk)



24/34

39 Pat: and you know Im er --

40 . . .(1.0) er Im aware thats like er --

41 . . its a part of your sort of spiritual make up to,

42 . . .(2.0) to confront.43 Jo: [hmh]

44 Pat: . . .(1.0) [the] situation.

45 . . . and er --

46 . . . move on from it.

47 . . .(1.0) but again --

48 er,

49 . . .(3.0) dealing --

50 you know,

51 like . . having to handle that.

52 . . . you know,

53 and er,

54 . . .(1.0) or the . . enormity of it.

9. Conclusion

In this paper, we have highlighted issues around the development of tran-

scription skills, in particular the problems of working with spontaneous

talk, such as that found in spoken interaction on emotionally charged

topics. Our position is that the process of transcription is key to rigorous

research involving spoken interaction data, and the present paper oers,

we believe, a unique description of the challenges of the transcription pro-

cess in one particular project, as well as the ways that these challenges

were overcome. In the following, we summarize this process of develop-

ing transcription skills in our research project. This summary may, at thesame time, be seen as a set of recommendations, or an agenda, that can

inform the transcription process of researchers engaged in other similar

projects working with spoken discourse data.

Our starting point was the decision to transcribe the conciliation talk

using IUs. Working toward this aim, we trialed our transcription using

a five-minute sample segment of talk. The first transcriber began by tran-

scribing the segment based on definitions of IUs available in the litera-

ture. Next, we involved two more experienced intertranscribers in the first

transcribers learning process. Following this, the first transcriber madehis decisions transparent using our enhanced transcription conventions.

Finally, we developed project-specific conventions, criteria, and proce-

dures for the transcription of the full data set.



25/34

The first step, then, was to prepare an initial transcript of the five-

minute sample segment of conciliation talk based primarily on the char-

acteristics of IUs outlined by Chafe (1994), but also consulting Crutten-

den (1986, 1997). However, as Chafe (1994: 62) points out, and we our-selves discovered at the end of this first step, skills in the transcription of

IUs requires both instruction and practice, and takes time and eort to

develop. In particular, the later intertranscriber checks, as well as the sub-

sequent externalizing of the transcribers decisions, showed that the crite-

ria for identifying IUs were only partially internalized by the first tran-

scriber in this first step of the transcription process.

We, next, had two additional researchers transcribing our five-minute

sample segment of conciliation talk. Both of these transcribers brought

unique and valuable experience to the task. Although we originally set

out to compare the three resulting transcriptions, we found this to be a

less fruitful exercise for the development of the first transcribers skills.

Rather, anticipating the necessarily more autonomous task of transcrib-

ing the full data set, the first transcriber worked through the two addi-

tional transcriptions now available to him, noted both dierences and

similarities, what the nature of these dierences and similarities were, and

then reflected on whether these indicated areas where his own transcrip-

tion might change and/or improve. For example, the detailed intertran-

scriber analysis showed that IU boundaries had been correctly identified,

but that a number of potential IU boundaries had been missed. With this

insight, and re-engaging with the literature on IUs, the five-minute sample

segment was re-transcribed. In sum, this exercise may be described as lis-

tening, noticing, and reflecting using already prepared transcripts, in this

case the transcripts of two experienced researchers. The exercise, then,

involved learning with rather than from the transcripts of the more

experienced researchers. More generally, we believe that intertranscriber

checks should be geared toward improving the quality of transcription,rather than acting as a simple objective measure validating the reliability

of transcription. This does not mean that we dismiss the need for reliabil-

ity across transcribers. Rather, our experience tells us that such reliability

can only follow from a more general concern with the quality of tran-

scription within a project.

In the following step of the transcription process, the first transcriber

used the enhanced transcription conventions (see Extract [6]). This exer-

cise was instrumental in making transparent the decision-making process

involved in the identification of IUs. It also acted to highlight particularchallenges of transcribing IUs with consistency. For example, the first

transcriber found that paying equal attention to the beginnings and ends

of IUs increased the consistent identification of IUs. This also highlighted



26/34

characteristics of IUs, such as anacrusis and re-setting of the pitch base-

line at the beginning of IUs, as well as syllable lengthening and terminal

pitch contours at the end of IUs. Importantly, this equal focus on both

the beginnings and ends of IUs was subtly dierent than the concernwith boundaries between intonation units, which had been the focus

of the first transcriber at the earlier intertranscriber stage. Making the

decision-making processes more transparent also highlighted that identi-

fying IUs was less problematic in the case of more fluent stretches of

talk, and more problematic in the case of less fluent stretches of talk.

From this, we became aware that the existing literature provides insu-

cient detail on how to deal with hesitations, false starts, and talk used to

regulate ongoing spoken interaction, as well as the eect of these trun-

cated intonation contours on the transcription of co-present more com-

plete intonation contours. Finally, making the transcription decisions

transparent using our enhanced conventions helped ensure the develop-

ment of transcription skills to a level where, when the first transcriber

went on to transcribe the full data set, he was working not at the limits

of his skill level, but within his skill level.

Finally, the development of project-specific conventions, criteria, and

procedures was an ongoing aim, aecting all stages of the transcription

process. At the beginning of the project, we decided to transcribe the con-

ciliation talk using IUs. This was based on our intuitions that such a tran-

scription would facilitate the investigation of the dynamics of metaphor

use in the conciliation talk. In later stages, more fine-grained decisions

were made about what features of the spoken discourse to represent in

the transcripts. These decisions were not taken in isolation from the first

transcribers gradually developing skills. Rather, his developing skills in

the transcription of IUs acted at times to create aordances, i.e., opening

up new possibilities for what could be included in the final project tran-

scription, and at times as a constraint, i.e., what the first transcriber coulddo within his skill level limited what could be included in the final project

transcription.

One final activity that may usefully be added to this agenda for devel-

oping transcription skills is the possibility of learning from other tran-

scribers experiences. This may involve talking to other researchers with

similar responsibilities within a research project, or with more experi-

enced researchers who in the past have had such transcription responsibil-

ities. It may also include interacting with the emerging research methods

literature that describes the experiences of transcribers (e.g., Gregory et al.1997; Bird 2005; Tilley 2003a, 2003b; Tilley and Powick 2002). The pre-

sent paper is a further contribution to this emerging research methods

literature.



27/34

Appendix 1: Transcription conventions

Discourse feature Convention Description

Transitionalcontinuity:

,.

?

ContinuingFinal

Appeal

Final pitch

contour:

/

\

_ (at the end of IU)

Rising

Falling

Level

Pitch baseline: _ (at the start of IU) Re-setting of the pitch

baseline

Truncation: -

--

Truncated word

Truncated intonation unit

Prominence: ^ Prominent syllable or word

Accelerated pace: and so the Small print

Slower pace: Equal sign

Pauses: . .

. . .

. . .(2.0)

Short pause (


28/34

etcetera.

but,

when it come --

when it comes down to it,I am sitting with somebody who is aected by . . my actions.

and er --

. . thats er --

. . . theres no preparation in the world for that.

I dont think.

. . you know its er --

3X certain X4--

. . I think it is unique.

and er --

. . I couldnt possibly have anticipated,

er --

. . your response and what have.

I was aware from speaking to certain people,

how . . you saw this as a journey etcetera.

Transcript 2: Second transcriber (experienced, but not with IUs)

Pat: you know,

its . . broken some sort of taboo here.

and er --

it goes into that territory,

and er --

theres no ^right --

. . .(1.0) ^for ^me,

. . .(1.0) to sit here and be forgiven.

. . .(3.0) if you understand me?

I mean in a sense theres the political thing,I knew what I was doing,

. . .(1.0) and I would even ^defend actions Ive taken,

etcetera.

but --

when you come --

when it comes down to it,

I am ^sitting with somebody whos ^aected by

. . .(1.0) ^my actions,

. . .(1.0) and er --

. . .(1.0) thats er --

. . theres no ^preparation in the world for that.

I dont think.



29/34

you know its er something --

. . .(2.0) I think its ^unique,

and er --

. . .(2.0) I couldnt ^possibly have anticipated--

. . .(2.0) um

. . .(2.0) ^your response . . and what have you,

I was aware from speaking to certain ^people,-

how--,

. . .(1.0) you . . saw this as a journey etcetera.

Transcript 3: Third transcriber (experienced with IUs)

Pat: you know its--

broken some sort of taboo here,and er it,

goes into that territory,

you know,

of er,

theres no right.

for me.

to sit here and be forgiven,

if you understand me.

I mean in a sense theres the political thing,

I knew what I was doing.

and I would even defend actions I have taken,

etcetera.

but--

when you come--

when it comes down to it,

I am sitting with somebody whos aected by--

my actions.and er,

thats er,

theres no preparation in the world for that.

I dont think.

you know its er,

something--

I think its unique.

and er,

I couldnt possibly have anticipated,um,

your response,

and what have I was aware from speaking to certain people,



30/34

how--

you--

saw this as a journey etcetera.

Enhanced transcript: First transcriber

1 Pat: you knowits

2 . . broken some sort ofta^boo here/

3 Jo: . . . hmh

4 Pat: and er --

5 it goes intothat ^territory/

6 _you know\

7 . . . of er --

8 Pat: theres ^no right9 . . .(1.0) for ^me

10 . . .(2.0) _to sit here and be for^given\

11 . . .(2.0) if you understand me

12 Jo: hmh

13 Pat: _I mean

14 in a ^sense theres the political thing

15 Iknew what I was ^doing\

16 . . and I would even de^fend actions Ive taken

17 _etcetera

18 _^but

19 _when you come --

20 when it comes ^down to it\

21 . . I am ^sitting with somebody whos aected by\

22 . . . _^my actions

23 . . .(1.0) and er --

24 . . .(2.0) thats er --

25 . . .(2.0) theres no preparation in the ^world for that\26 Jo: [hmh]

27 Pat: . . . _[I] dont think\

28 . . .(1.0) _you know its er --

29 . . something --

30 . . .(1.0) I think it is unique\

31 . . and er\

32 . . .(2.0) I couldnt possibly have an^ticipated\

33 . . .(1.0) _er

34 . . _^your response and what have\35 I was a^warefrom speaking to certain people/

36 . . .(1.0) how . . . y- you --

37 . . saw this as a ^journey etcetera\



31/34

Notes

1. The project, Using visual display to investigate the dynamics of metaphor in concilia-

tion talk, was supported by the UKs Arts and Humanities Research Board under itsInnovation Award scheme. We acknowledge that support, and also thank the partici-

pants in the talk for giving permission to use the data.

2. This view of consciousness does not necessarily dismiss a role for unconscious mental

processes. In fact, Chafe (1996: 39) concedes that both the content . . . flow and manage-

ment of consciousness is probably in large part unconsciously determined. Our assess-

ment is that the exact division of labor, between conscious and unconscious processes,

does not impact upon the process of transcribing intonation units.

3. Pierrehumbert and Hirschberg (1990) suggest that intonational phrases are made up of

one or more intermediate units. These intermediate units are roughly similarly to what

we in this paper call intonation units.

References

Atkinson, J. M. and Heritage, J. (1984). Transcript notation. In Structures of Social Action:

Studies in Conversational Analysis, J. M. Atkinson and J. Heritage (eds.), ixxvi. Cam-

bridge: Cambridge University Press.

Bird, C. M. (2005). How I stopped dreading and learned to love transcription. Qualitative

Inquiry11 (2): 226248.

Cameron, L. (2003).Metaphor in Educational Discourse. London: Continuum.Cameron, L. J. (forthcoming). Patterns of metaphor use in reconciliation talk. Discourse &

Society.

Cameron, L. J. and Stelma, J. H. (2004). Metaphor clusters in discourse. Journal of Applied

Linguistics1 (2): 107136.

Chafe, W. (1993). Prosodic and functional units of language. InTalking Data: Transcription

and Coding in Discourse Research, J. A. Edwards and M. D. Lampert (eds.), 3343. Hill-

sdale, NJ: Lawrence Erlbaum.

(1994).Discourse, Consciousness, and Time: The Flow and Displacement of Conscious Ex-

perience in Speaking and Writing. Chicago: Chicago University Press.

(1996). How consciousness shapes language.Pragmatics and Cognition 4 (1): 3554.

Cook, G. (1990). Transcribing infinity: Problems of context presentation. Journal of Prag-matics14: 124.

Couper-Kuhlen, E. (2001). Intonation and discourse: Current views from within. In The

Handbook of Discourse Analysis, D. Schirin, D. Tannen, and H. E. Hamilton (eds.),

1334. Malden, MA: Blackwell.

Cruttenden, A. (1986).Intonation. Cambridge: Cambridge University Press.

(1997).Intonation, 2nd ed. Cambridge: Cambridge University Press.

Crystal, D. (1969).Prosodic Systems and Intonation in English. London: Cambridge Univer-

sity Press.

Dressler, R. A. and Kreuz, R. J. (2000). Transcribing oral discourse: A survey and a model

system.Discourse Processes 29 (1): 2536.Du Bois, J. W. (1991). Transcription design principles for spoken discourse research. Prag-

matics1 (1): 71106.

(2000). VoiceWalker: A discourse transcription utility. URL: 3http://www.linguistics

.ucsb.edu/resources/computing/download/documentation.htm4[accessed 27 May 2005].



32/34

Du Bois, J. W., Schuetze-Coburn, S., Cumming, S., and Paolino, D. (1993). Outline of dis-

course transcription. In Talking Data: Transcription and Coding in Discourse Research,

J. A. Edwards and M. D. Lampert (eds.), 4589. Hillsdale, NJ: Lawrence Erlbaum.

Edwards, J. A. (1993). Principles and contrasting systems of discourse transcription. In Talk-

ing Data: Transcription and Coding in Discourse Research, J. A. Edwards and M. D.

Lampert (eds.), 331. Hillsdale, NJ: Lawrence Erlbaum.

Ford, C. E. and Thompson, S. A. (1996). Interactional units in conversation: Syntactic, in-

tonational, and pragmatic resources for the management of turns. In Interaction and

Grammar, E. Ochs, E. A. Scheglo, and S. A. Thompson (eds.), 134184. Cambridge:

Cambridge University Press.

Green, J., Franquiz, M., and Dixon, C. (1997). The myth of the objective transcript: Tran-

scribing as a situated act. TESOL Quarterly31 (1): 172176.

Gregory, D., Russell, C. K., and Phillips, L. R. (1997). Beyond textual perfection: Tran-

scribers as vulnerable persons.Qualitative Health Research 7 (2): 294300.

Gumperz, J. J. and Berenz, N. (1993). Transcribing conversational exchanges. In TalkingData: Transcription and Coding in Discourse Research, J. A. Edwards and M. D. Lampert

(eds.), 91121. Hillsdale, NJ: Lawrence Erlbaum.

Halliday, M. A. K. (1967).Intonation and Grammar in British English. The Hague: Mouton.

Kreckel, M. (1981). Communicative Acts and Shared Knowledge in Natural Discourse. Lon-

don: Academic Press.

Lapadat, J. C. (2000). Problematizing transcription: Purpose, paradigm and quality. Interna-

tional Journal of Social Research Methodology 3 (3): 203219.

Lapadat, J. C. and Lindsay, A. C. (1998). Examining transcription: A theory-laden method-

ology. Paper presented at the Annual Meeting of the American Educational Research As-

sociation, San Diego, CA, 1317 April (Eric Document Number ED 419 821).

(1999). Transcription in research and practice: From standardization of technique to in-

terpretive positionings. Qualitative Inquiry 5 (1): 6486.

Lindsay, J. and OConnell, D. C. (1995). How do transcribers deal with audio recordings of

spoken discourse.Journal of Psycholinguistic Research 24 (2): 101115.

McLellan, E., MacQueen, K. M., and Neidig, J. L. (2003). Beyond the qualitative interview:

Data preparation and transcription. Field Methods15 (1): 6384.

Muller, N. and Damico, J. S. (2002). A transcription toolkit: Theoretical and clinical consid-

erations.Clinical Linguistics & Phonetics 16 (5): 299316.

Ochs, E. (1979). Transcription as theory. In Developmental Pragmatics, E. Ochs and B. B.

Schieelin (eds.), 4372. New York: Academic Press.

OConnell, D. C. and Kowal, S. (1994). The transcriber as language user. In The Dynamicsof Language Processes: Essays in Honor of Hans W. Dechert, B. Guillermo (ed.), 119142.

Tubingen: Gunter Narr.

(1999). Transcription and the issue of standardization. Journal of Psycholinguistic Re-

search28 (2): 103120.

Oelschlaeger, M. L. and Thorne, J. C. (1999). Application of the correct information unit

analysis to the naturally occurring conversation of a person with aphasia. Journal of

Speech, Language, and Hearing Research 42: 636648.

Park, J. S.-Y. (2002). Cognitive and interactional motivations for the intonation unit. Stud-

ies in Language26 (3): 637680.

Pierrehumbert, J. and Hirschberg, J. (1990). The meaning of intonational contours in the in-

terpretation of discourse. In Intentions in Communication, P. R. Cohen, J. Morgan, and

M. E. Pollack (eds.), 271311. Cambridge, MA: MIT Press.

Roberts, F. and Robinson, J. D. (2004). Interobserver agreement on first-stage conversation

analytic transcription.Human Communication Research 30 (3): 376410.



33/34

Romero, C., OConnell, D. C., and Kowal, S. (2002). Notation systems for transcription:

An empirical investigation. Journal of Psycholinguistic Research 31 (6): 619631.

Schuetze-Coburn, S., Shapley, M., and Weber, E. G. (1991). Units of intonation in dis-

course: A comparison of acoustic and auditory analyses. Language and Speech 34: 207

234.

Sealey, A. and Carter, B. (2004). Applied Linguistics as a Social Science. London:

Continuum.

Tilley, S. A. (2003a). Challenging research practices: Turning a critical lens on the work of

transcription.Qualitative Inquiry 9 (5): 750773.

(2003b). Transcription work: Learning through coparticipation in research practices.

Qualitative Studies in Education 16 (6): 835851.

Tilley, S. A. and Powick, K. D. (2002). Distanced data: Transcribing other peoples research

tapes.Canadian Journal of Education27 (2/3): 291310.

Wennerstrom, A. (2001). The Music of Everyday Speech: Prosody and Discourse Analysis.

Oxford: Oxford University Press.Wichmann, A. (2000). Intonation in Text and Discourse: Beginnings, Middles and Ends. Har-

low, Essex: Pearson Education/Longman.

Yule, G. (1980). Speakers topics and major paratones. Lingua 52: 3347.

Juurd H. Stelma received a Ph.D. in Education from the University of Leeds in 2003 and is

currently a Lecturer in TESOL in the School of Education at the University of Manchester.

His main research interest is the development of methodology for exploring the dynamical

nature of language in use. Address for correspondence: School of Education, Humanities

Devas Street Building, University of Manchester, Oxford Road, Manchester, M13 9PL,

[email protected].

Lynne J. Cameron is Professor of Applied Linguistics in the Centre for Language and Com-

munication at the Open University. Her research seeks to understand how language is used

in building understanding between people, particularly through metaphor. She has published

widely on the use of metaphor in dierent settings, including the co-edited Researching and

Applying Metaphor (1999, with Graham Low) and her book Metaphor in Educational Dis-

course(2003). Address for correspondence: Centre for Language and Communication, Fac-

ulty of Education and Language Studies, The Open University, Milton Keynes, MK7 6AA,

[email protected].



34/34

Intonation Units in Spoken Interaction (2007)

Documents