the first certificate in english examination speaking test ... · 92 galaczi bigger range of speech...

34
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/233065072 Peer–Peer Interaction in a Speaking Test: The Case of the First Certificate in English Examination Article in Language Assessment Quarterly An International Journal · April 2008 DOI: 10.1080/15434300801934702 CITATIONS 73 READS 557 1 author: Some of the authors of this publication are also working on these related projects: Research methodology View project Evelina D. Galaczi University of Cambridge 20 PUBLICATIONS 173 CITATIONS SEE PROFILE All content following this page was uploaded by Evelina D. Galaczi on 19 August 2015. The user has requested enhancement of the downloaded file.

Upload: doankhanh

Post on 16-Aug-2018

216 views

Category:

Documents


0 download

TRANSCRIPT

See discussions, stats, and author profiles for this publication at:

https://www.researchgate.net/publication/233065072

Peer–Peer Interaction in a Speaking Test: The Case of

the First Certificate in English Examination

Article  in  Language Assessment Quarterly An International Journal · April 2008

DOI: 10.1080/15434300801934702

CITATIONS

73

READS

557

1 author:

Some of the authors of this publication are also working on these related projects:

Research methodology View project

Evelina D. Galaczi

University of Cambridge

20 PUBLICATIONS   173 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Evelina D. Galaczi on 19 August 2015.

The user has requested enhancement of the downloaded file.

This article was downloaded by: [University of Cambridge]On: 19 August 2015, At: 02:32Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: 5 Howick Place, London, SW1P 1WG

Language Assessment QuarterlyPublication details, including instructions forauthors and subscription information:http://www.tandfonline.com/loi/hlaq20

Peer–Peer Interaction in aSpeaking Test: The Case ofthe First Certificate in EnglishExaminationEvelina D. Galaczi aa University of Cambridge ESOL Examinations ,Published online: 06 May 2008.

To cite this article: Evelina D. Galaczi (2008) Peer–Peer Interaction in a SpeakingTest: The Case of the First Certificate in English Examination, Language AssessmentQuarterly, 5:2, 89-119, DOI: 10.1080/15434300801934702

To link to this article: http://dx.doi.org/10.1080/15434300801934702

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all theinformation (the “Content”) contained in the publications on our platform.However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness,or suitability for any purpose of the Content. Any opinions and viewsexpressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of theContent should not be relied upon and should be independently verified withprimary sources of information. Taylor and Francis shall not be liable for anylosses, actions, claims, proceedings, demands, costs, expenses, damages,and other liabilities whatsoever or howsoever caused arising directly orindirectly in connection with, in relation to or arising out of the use of theContent.

This article may be used for research, teaching, and private study purposes.Any substantial or systematic reproduction, redistribution, reselling, loan,

sub-licensing, systematic supply, or distribution in any form to anyone isexpressly forbidden. Terms & Conditions of access and use can be found athttp://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

LANGUAGE ASSESSMENT QUARTERLY, 5(2), 89–119, 2008Copyright © Taylor & Francis Group, LLCISSN: 1543-4303 print / 1543-4311 onlineDOI: 10.1080/15434300801934702

HLAQ1543-43031543-4311Language Assessment Quarterly, Vol. 5, No. 2, February 2008: pp. 1–49Language Assessment Quarterly

ARTICLES

Peer–Peer Interaction in a Speaking Test: The Case of the First Certificate in English

Examination

Peer–Peer InteractionGALACZI Evelina D. GalacziUniversity of Cambridge ESOL Examinations

This discourse-based study turns its attention to paired test-taker discourse in theFirst Certificate in English speaking test. Its primary aim is to focus on fundamentalconversation management concepts, such as overall structural organisation, turn-taking, sequencing, and topic organisation found in the dyadic test-taker interactionin 30 pairs of test takers. The analysis highlights global patterns of interaction,termed “Collaborative,” “Parallel” and “Asymmetric,” and salient features of inter-action characteristic to each pattern. A second goal of the study is to investigate apossible relationship between the patterns of peer-peer interaction and test scoresfor “Interactive Communication.” The results show a close relationship betweentype of talk and test scores, thus providing validity evidence for the test scores. Thestudy discusses implications for a more accurate understanding of the construct ofconversation management underlying the FCE examination, and for the empiri-cally-based construction of assessment criteria for speaking tests.

One of the most promising methodological developments in the last 15 years in theinvestigation of speaking tests has been the introduction of qualitative methodolo-gies, which have offered the testing community new ways of investigating perfor-mance testing issues and in the process have proven to be “effective andilluminating tool[s]” (McNamara, Hill, & May, 2002, p. 222) in the analysis of

Correspondence should be addressed to Evelina D. Galaczi, University of Cambridge ESOL Exam-inations, 1 Hills Road, Cambridge CB1 2EU, Great Britain. E-mail: [email protected]

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

90 GALACZI

aspects of the validity of oral tests. Many excellent research endeavors have con-tributed by now to the development of the field of discourse-based studies in orallanguage assessment (e.g., Brown, 2003, 2005, 2006; Egbert, 1998; He, 1998;Ikeda, 1998; Iwashita, 1998; Johnson, 2001; Kormos, 1999; Lazaraton, 1991,1992, 1996, 1997, 2002; Moder & Halleck, 1998; O’Loughlin, 2002; O’Sullivan,2002; Ross, 1992; Swain, 2001; Yoshida-Morise, 1998; Young, 1995; Young &Milanovic, 1992, to name a few). These research studies have established aresearch agenda within language assessment, which has gone beyond examiningthe ratings of speaking tests to an analysis of the turn-by-turn sequential interactionin oral tests. This microanalytic focus has complemented quantitative methodolo-gies used in assessment research and has extended the focus of analysis to the oraltest discourse itself. As such, this research trend has responded to van Lier’s (1989)now-classic appeal to “obtain an ‘insider’s’ view” (p. 489) of the language profi-ciency interview to make it a more useful means of assessing oral proficiency.

This study turns its attention to the paired speaking test format, with a specificfocus on the First Certificate in English speaking test (henceforth FCE). In 1996,Cambridge ESOL (part of Cambridge Assessment) introduced the paired speakingtest format in most of its Main Suite examinations in response to current thinkingon the assessment of oral language skills and the need for a broader range of tasksthat would elicit different aspects of oral proficiency. The test typically involvestwo candidates and two examiners (a trio format, which is quite rare, is used incases where there is an uneven number of candidates at an examining session, andthe last candidates joins the final pair of candidates to form a group of three). Thepairing of candidates allows for a more varied sample of interaction, that is, candi-date–candidate as well as candidate–examiner. The pairing of examiners allowsfor marks to be given from two differing but complementary perspectives. Theexaminer who participates in the test (the “interlocutor,” in Cambridge ESOLterms) provides a global mark based on a holistic scale; the nonparticipatingexaminer (called the “assessor”) awards four analytical marks on “Grammar andVocabulary,” “Discourse Management,” “Pronunciation,” and “InteractiveCommunication.” There is no explicit task achievement rating scheme, and candi-date output is assessed based on performance on all four tasks. Both assessmentsare based on a nine-band marking scheme, where 1.0 is the lowest score and 5.0the highest; half scores can be awarded as well. The test comprises four tasks: aninterview between the examiner and candidate, an individual long turn from eachcandidate, a two-way interaction between the candidates, and a three-way interac-tion between the candidates and examiner. The use of a multiple-task formatallows for different types of talk and candidate output to be generated (e.g., ques-tion and answer, uninterrupted long turn, discussion1).

1For a more detailed overview of the FCE, visit http://www.cambridgeESOL.org

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 91

Although the paired task has been used for almost a decade now and variousresearch studies have investigated relevant issues (ffrench, 1999; Ikeda, 1998;Iwashita, 1998; Kormos, 1999; Lazaraton, 1996; Lazaraton & Frantz, 1996;Nakatsuhara, 2006; Norton, 2005; O’Sullivan, 2002), there is still a shortage of pub-lished studies that focus on the nature of talk elicited by the paired format. The studypresented here aims to fill that niche and investigate the talk which is co-constructedwithin dyads of test takers involved in a paired FCE test task from a ConversationAnalysis (CA) perspective. There were two basic issues guiding the study:

1. Patterns of interaction: What are the patterns of interaction in Part 3: Two-way Collaborative Taskof the FCE? What are the salient discourse features of these interactionalpatterns?

2. Relationship between patterns of interaction and Interactive Communica-tion (IC) score2: Is there a relationship between the patterns of interaction and the IC scorereceived? How do the interactional patterns impact the IC score?

THE PAIRED SPEAKING TEST FORMAT

The construct of speaking ability underlying the Cambridge ESOL speaking tests isinformed both by theoretical advances such as theories of communicative languageability (Bachman, 1990; Bachman & Palmer, 1996) and by pedagogical develop-ments. Such theories of language ability include a conversation management compo-nent and presuppose the need for oral tests to provide opportunities for test takers todisplay a range of their communicative competence, which taps into conversationmanagement ability. One of the advantages of the paired format is that it can providesuch opportunities for peer–peer discourse because in some tasks the interaction isnot, as Skehan (2001) put it, “orchestrated by the assessor” (p. 169) but is instead con-trolled to a larger extent by the candidate. The test as a whole is, naturally, still con-trolled by the assessor, but the assessor does not take part in some of the paired tasks.

The empirical validation work on this test format has shown that, unlike the con-ventional interview test format, which results in tasks with asymmetrical participa-tion, paired oral test tasks are more symmetrical in the interaction possibilities theycreate (Egyud & Glover, 2001; Iwashita, 1998; Kormos, 1999; Lazaraton, 1997,2002; Taylor, 2001), they elicit a wider sample of learner performance with a

2Interactive communication was the assessment criterion chosen because the current focus ofinvestigation—conversation management—is most closely related to the interactive communicationcriterion in the FCE speaking marking scheme.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

92 GALACZI

bigger range of speech functions (ffrench, 1999), they provide increased opportuni-ties for test takers to display their conversational management skills (Kormos,1999), and as a test format they are received positively by the test takers (Egyud &Glover, 2001). In the words of Skehan (2001), paired test tasks “enable a widerrange of language functions and roles to be engineered to provide a better basis fororal language sampling with less asymmetry between participants” (p. 169).

Although empirical investigations of paired test interaction are testimony totheir usefulness as assessment instruments, discussions in the literature havealerted us to a number of caveats associated with paired tests. A fundamental con-cern inherent in paired test tasks has been discussed by McNamara (1997) andSwain (2001) in their excellent and thought-provoking papers on the nature ofinteraction in speaking tests. Both authors cautioned testers about the problem ofinterpreting individual scores based on a jointly constructed interaction amongparticipants and highlighted the importance of investigating the co-constructedpeer–peer discourse in a paired speaking test, as it has serious implications for thescoring of individual performance. As Swain (2001) noted, “in a group, the perfor-mance is jointly constructed and distributed across the participants” (p. 296),which implies a need for language testers “to be able to measure accurately the[solo] performance of test-takers interacting in a small group setting” (p. 296). Thispoint echoes Lazaraton’s (1997) contention about the fundamental codependenceof the two test takers in terms of the interaction they produce: “A sense of ‘taskcompletion’ was hard to evaluate for one test taker without taking into account thecontributions of the other” (p. 157). In light of the difficulties associated with therating of jointly constructed performance, McNamara’s (1997) simple question,“Whose performance is it, anyway?” addresses the heart of the matter and under-scores the need for a better understanding of the co-constructed peer–peer interac-tion in paired oral tests.3 A thorough investigation of the joint construction inpaired tests, which is the aim of the study presented here, is likely to lead to fairerand more substantiated inferences made on the basis of test scores because it willprovide more accurate insights into the relationship between the language outputof the candidate and the score received. Such a deeper understanding holds impli-cations for test development because it contributes to a more accurate understand-ing of the paired test format and the underlying construct of conversationmanagement. It could also inform rating scale construction and rater training as itprovides insights into the relationship between task, language output, and scoresand allows for terms in assessment scales to be empirically derived and operation-ally defined. Finally, it holds implications for discourse analysis, as it providesinsights into the structure of the paired test as a speech event in its own right.

3It is important to note that the issue of co-construction is not exclusive to paired tests but alsoapplicable to the singleton format, as Brown (2003) convincingly showed.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 93

METHOD

Participants

The sample in this study comprises 30 paired dyads (60 test takers) who took thespeaking part of the FCE in either June or December 1999. The study participantscame from 11 different first language backgrounds (plus one unknown), with75% speaking either Italian (n = 24), Chinese (n = 12), or French (n = 9) as theirfirst language. In terms of gender, 75% were women (n = 45) and the remainderwere men. Most of the dyads consisted of either two women (53%) or a man anda woman (43%), and 80% comprised speakers with the same first language.

Data Collection

The data were collected by FCE test administrators during six routine administra-tions of the test (June 1999 in Italy, Taiwan, Switzerland, and the UK; December1999 in China and the UK). Videotaping was not used as part of the data collec-tion procedure. Given the quality of information that could have been obtainedfrom a video recording of the participants in interaction, this presents a limitationto the investigation. However, the setting up of video cameras would not havebeen feasible during data collection, as it would have been intrusive to the actualspeaking test. The lack of videotaped data is considered to be one of the inevita-ble trade-offs between breadth and depth of the database. It made the investiga-tion of nonverbal issues such as gestures impossible, but because that is not anaim of the study, this limitation was considered marginal.

Materials

The FCE task chosen for analysis was “Part 3, Two-way Collaborative Task,” whichinvolves the two test takers in a two-way discussion and is the only FCE speakingtask that consists entirely of peer–peer talk. The task is set up by the examineraccording to the instructions specified in the “interlocutor frame.” The interlocutorframe is a script for the examiner that specifies the language to be used when intro-ducing and managing a task and the prompts to be given to the test takers. The inter-locutor frame is used for the purposes of standardisation, because by adhering to thescript, examiners ensure that all candidates are treated equally. An example of aninterlocutor frame and the accompanying visual can be seen in Figures 1 and 2.

Data Analysis

Data Transcription Procedures. The interactions were transcribed usingCA conventions following Atkinson and Heritage (1984). The choice of

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

94 GALACZI

transcription conventions (given in the appendix) was dictated mainly by theneed for an accurate transcript that would include microlevel features such aslength of pauses, hesitations, overlaps, and interruptions. In an effort to ensurehigher accuracy and thus reliability of the transcription, all the transcripts werechecked a second time approximately 1 month after the initial transcription.Despite the care taken to produce accurate transcripts, I acknowledge that theprocess of transcription is by definition a selective undertaking (Ochs, 1979). Toaddress such theoretical problems associated with CA transcription, I used thetranscripts in conjunction with the recordings during the analysis, as recom-mended in the latest guidelines on CA research (“Some Guidelines,” 2003).

Procedures for Analysis. In line with CA methodology, the “data explora-tion strategy” was employed (Lazaraton, 2002; ten Have, 1999) in this study. It isimportant to point out that even though CA principles and techniques were used,a purist CA methodology was not adopted, and terms that come from differentanalytic perspectives, such as conversational floor or backchannel (Yngve,1970), have also been employed. In addition, quantification, which goes againstpurist CA methodology, was used in the analysis.

The issue of quantification in CA is a controversial one, with lack of consen-sus in the literature about its status within this analytical tradition (Heritage,1995; Heritage & Roth, 1995; Lazaraton, 2002; Linell, 1998; Psathas, 1995;

FIGURE 1 Examiner frame for “Part 3, Two-way Collaborative Task,” FCE speaking test.Reprinted from FCE: First Certificate in English Handbook (2002).

Now, I’d like you to talk about something together for about three minutes. I’m

just going to listen.

I’d like you to imagine that a busy international hotel is looking for staff for the

holiday season. Here are some of the jobs available.

[‘Interlocutor’ places in front of the test takers a sheet with pictures of various hotel

jobs, such as porter, barman, chef, pianist, cleaner, waiter, and receptionist.]

First, talk to each other about how difficult it would be to do these jobs without

training. Then say which job you would each find the most difficult.

You have only about three minutes for this. So, once again, don’t worry if I

stop you, and please speak so that we can hear you. All right?

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 95

Schegloff, 1993; ten Have, 1999). Schegloff and Psathas argued against codingand quantification in CA analyses largely because of the “reductionist” nature ofcoding schemes (Psathas, 1995, p. 8), which limit the phenomena being analyzedto a finite set of features. Similarly, Linell noted that “the operationalization ofconcepts, which is part of the development of coding categories, usually impliesdisregarding some of the actual complexities of the data” (p. 178). The argumentis that a coding system, no matter how detailed and reliable, cannot capture allthe complexity and depth of interaction. Although I acknowledge the view that“quantification is not a substitute for analysis” and that “we need to know whatthe phenomena are, how they are organized, and how they are related to eachother as a precondition for cogently bringing methods of quantitative analysis tobear on them” (Schegloff, 1993, p. 114), I also believe that quantification has apotentially useful role as an auxiliary method to CA research. More specifically,it has the potential for providing insights complementary to the interpretation ofthe data. Quantification that follows careful analysis of individual cases and usesmeaningful categories emerging out of the CA analysis is, therefore, warranted andhighly valuable, as has also been advocated by Heritage and Roth (1995). The pos-sibility of a fusion of these two methodologies is a belief that underlies this study.

FIGURE 2 An example of a visual prompt for “Part 3, Two-way Collaborative Task,” FCEspeaking test. Reprinted from FCE: First Certificate in English Handbook (2002).

waiter

cleaner

pianist chef

barman

porter

receptionist

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

96 GALACZI

Following the analytical steps outlined by Pomerantz and Fehr (1997), “topi-cal sequences” emerged as the most appropriate unit of analysis. The issue oftopic and topic management is notoriously difficult and complex. Brown andYule (1983) noted that “formal attempts to identify topics are doomed to failure”(p. 68). Their pessimistic prognosis lies largely in the difficulty of determiningone abstract, overarching topic in the discourse. This relates to the notion of “rel-evance” (Grice, 1975) and the fact that what conversations are “about” is a veryrelative concept. At a higher level of abstraction, for example, a group of topicsmay merge into one broader topic. As Ainsworth-Vaughn (1992) succinctly putit, “almost all gaps in the referential path can be crossed by doing inferentialwork” (p. 412). A further complexity is the difficulty of drawing boundariesbetween the new pieces of referential information presented sentence by sen-tence, as noted by Brown and Yule (1983). Often, talk may drift imperceptiblyfrom one topic to another in stepwise progression (Button, 1991; Button & Casey,1984; Jefferson, 1984), thus posing difficulties in identifying what the topics are,or a topic could be segmented from other topics (Button & Casey, 1984).

Acknowledging the complexities of dealing with topic in a systematic man-ner, rigorous criteria for identifying topics were established in this study. A “top-ical sequence” was operationalised as the spate of talk that referred to a specificvisual discussed by the test takers. In other words, a sequence began with the firstturn that initiated discussion related to a certain visual prompt, and everythingrelated to that prompt was part of one topical sequence. As such, in this stage ofthe analysis, each transcript was divided into discrete topical sequences at pointsin the transcript that indicated a topic shift based on the visual prompts availableto the dyad. This allowed for a systematic and consistent analysis of topic andtopical moves for the purposes of this study.

In addition to characterizing the topical sequences and moves within the topi-cal sequence, the analysis further focused on their form, because as Pomerantzand Fehr (1997) noted, there are different ways in which something can be done(e.g., statement vs. question) and the choice of one alternative over another canbe meaningful. The timing (i.e., pauses and gaps) and means of obtaining turns(by self-selection or other-selection), starting turns (after a gap, overlap, or latch),or terminating turns (forcefully or voluntarily yielded) was also the focus of anal-ysis. Finally, the dyad members’ roles were classified as collaborative, noncol-laborative, dominant, and passive based on the occurrence of the aforementionedfeatures in the whole transcript. The roles the different speakers oriented to wereconstructed on a turn-by-turn basis, and the analysis showed that in the majorityof cases those roles remained stable throughout the development of the task. In afew cases the speakers oriented to two different roles as the interaction unfolded.Based on the roles oriented to, four interactional patterns emerged, termed “col-laborative,” “parallel,” “asymmetric,” and a “blend,” which are discussed in theFindings section.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 97

Once the patterns of dyadic interaction were established, inter- and intracoderreliability were checked to accomplish external reliability for the findings. Anindependent researcher (a discourse analyst who was unfamiliar with the data)was given a brief training session which involved explaining the categories thatdistinguished between the various patterns of interaction and illustrating the gen-eral aim of the analysis with two norming tests. The independent coder then read/listened to 11 randomly selected transcripts (approximately one third of the dataset) and assigned each one an interactional pattern. In addition, the independentcoder was told that she could assign two patterns of interaction to a specific dyadif the interaction was a “blend” of two interactional styles. The intercoder agree-ment was calculated to be 91% (using Hatch & Lazaraton, 1991), which was con-sidered an acceptable level of intercoder reliability.

In addition to the intercoder agreement estimate, intracoder reliability wasestimated by recoding all the transcripts for distinct patterns of interaction2 months after the initial analysis. The intracoder reliability was calculated to be96%, which was considered to be an acceptable level of intracoder reliability.

FINDINGS

The remainder of the article provides an overview of the patterns of interactionco-constructed by the test-taker dyads and the underlying dimensions distin-guishing them. This is followed by a discussion of the relationship between thepatterns of interaction and IC scores.

The Dimensions of Mutuality, Equality, and Conversational Dominance

The analysis revealed that the test-taker dyads were “producing topicality”(Maynard, 1980) in different ways: Some were working in a collaborative man-ner and sustaining topics over longer stretches of discourse; others were workingin a “solo versus solo” fashion; still others emerged as asymmetrical in theamount of effort each speaker put toward completion of the task. As noted ear-lier, the three distinct patterns of interaction that emerged were termed collabora-tive, parallel, and asymmetric. In addition, a “blend” was found, which containeddiscourse features of two interactional patterns. The patterns of interaction weredistinguishable based on the dimensions of “mutuality” and “equality”(Damon & Phelps, 1989; Storch, 2002). These two concepts refer to the creationof shared meaning from one turn to the next (mutuality) and to the work distribu-tion among the participants (equality; see also Jones & Gerard, 1967, for the sim-ilar notions of “interactional contingency” and “goal orientation”). FollowingStorch (2002), the two dimensions were conceptualized as a continuum, eachranging from low to high.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

98 GALACZI

The analysis further indicated that the dimension of conversational dominanceintersected with the dimensions of mutuality and equality. Based on Itakura(2001), conversational dominance was operationalised as the quantity of talk(“quantitative dominance”), interruptions (“participatory dominance”), and ques-tions (“sequential dominance”). As such, each of the interactional patterns wasfurther subdivided into two subpatterns based on the overall high or low level ofconversational dominance. An important finding that emerged from the analysisof conversational dominance was the differential distribution of features of con-versational dominance. In the case of the collaborative dyads, conversationaldominance was mostly “sequential” (Itakura, 2001), that is, it controlled the direc-tion of the interaction through questions and next-speaker selection. In the paralleldyads, the conversational dominance was mostly “participatory” (Itakura, 2001),that is, it involved competition for the floor, and in asymmetric dyads it was quan-titative, with one dyad member participating noticeably more that the other. In thecase of the collaborative dyads, the conversational dominance features were facil-itative and other-centered, as they signaled engagement between the interlocutorsand moved the task forward in a mutual fashion. In the parallel dyads, the conver-sational dominance was self-centered, as it often violated speaker rights.

Overall, the study found that most FCE test-taker dyads co-constructed a dis-course that was collaborative (30%, n = 9 dyads), parallel (30%, n = 9 dyads), ora blend, consisting of either collaborative/parallel features (23%, n = 7 dyads) orcollaborative/ asymmetric features (7%, n = 2 dyads). A small subset (10%, n = 3dyads) oriented to asymmetric interaction. Furthermore, the use of conversationaldominance strategies was generally low in the majority of the dyads and occurredin only 23% of the dyads (n = 7 dyads).

Collaborative Interaction

Interactions that exhibited high mutuality and high equality were termed collabo-rative. In collaborative dyads the two participants shifted in their interactional rolesbetween listener and speaker; they developed both their own topics and supportedthe development of the other person’s topic. They were, as such, both “self-focused” in their interactional behavior and “other-focused.” The most salient dis-tinguishing feature of the collaborative dyads was the topic extension move, whichsignaled engagement with the other speaker’s ideas and displayed an ability to bothsay something that relates to what has been said before and introduce somethingnew, which is a basic duty of conversationalists (Tracy & Moran, 1983).

The analysis also found that some collaborative dyads oriented to a discourserich in high conversational dominance features characterised by short turns, rapidspeaker change, questions, avoidance of gaps between turns, supportive overlaps,and frequent acknowledgement tokens, resembling a “high-involvement” style oftalk (Tannen, 1981).

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 99

Collaborative talk was structurally manifested in exchanges of the following type:

Speaker A: Topic initiation Speaker A: Topic initiation + Topic extensionSpeaker B: Topic extension OR Speaker B: Topic extension + Topic initiationSpeaker A: Topic extension Speaker A: Topic extension

Interactionally, it exhibited the characteristics given in Table 1.The interaction between Danielle4, a Swiss woman, and Peter, a Swiss man, is

used to briefly illustrate various features of collaborative talk. Peter and Danielleare asked by the interviewer to discuss services that should be included in anewly opened student café. Several visual prompts were given for this task to aidthem with the discussion. What follows in Excerpt 1 is an exchange betweenthese two speakers. In this excerpt, as in all the excerpts to follow, only a shortpart of the whole exchange is given for illustrative purposes.5 The bold phrasesmark the new topics that are introduced.

4All names have been changed.5A more detailed illustration of the different patterns of interaction can be found in Galaczi (2004).

TABLE 1Interactional Characteristics of Collaborative Talk

Dimension Interactional Characteristics

High mutuality • Topic expansion of self- and other-initiated topic• Uptake and expansion of the prior speaker’s interactional contributions,

and not just topic repetition or recycling• Development of topics over several turns (multi-turn topics)• Less reliance on the visual prompts and more “context independence”

(Young, 1995) in the development of topics• Cohesion and coherence between turns in the form of (a) minimal

responses as acknowledgement tokens (e.g., “yes”), (b) agreement (e.g., “I agree with you”), (c) syntactic cohesion (e.g., “also”), and (d) lexical cohesion (e.g., repetition and paraphrase of an idea)

High equality • Balanced quantity of talk between the two participants• Balance between the two participants in the use of topic initiation

moves and topic expansion moves

Conversational dominance

Low conversational dominance dyads:• Listener support in the form of acknowledgment tokens• Listener involvement in the form of overlaps and latchesHigh conversational dominance dyads:• Frequent use of follow-up questions• Frequent instances of overlaps and latches• Collaborative floors (Coates, 1996) where the participants completed

each other’s utterances

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

100 GALACZI

Excerpt [1] P = Peter; D = DanielleTask: New student café(3 turns have been omitted)

1 P (.3) [(i do]n’t know) [((laughter))]2 D [yep] [((Laughter.))] yeah, and I think it’s also imp-

ortant to have something to have something like a snack bar, =P = Hum hmmD yeah, [because it’s]

3 P [yeah (me too)].4 D yeah. (.3) and coffee, and, (.3) BUT, (.3) you shouldn’t forget

this um, this uh, gambling (.3) things. (.4) because ah thereare some people, especially you, ((Laughter))

P %yes% =D = who like to uh, (.3) to calm down with these things, and i

think (.3) u:m, (.3) if you’re in school and study the whole dayyou- you need something to- to relax and I think that’s not bad.

5 P yeah, but it’s, it’s uh, very important that uh, this (.3) automatsah they have to, to be in a- in a different ro[om].

6 D [ah], in a separateroom, ye [ah. I agree with you] yeah.

7 P [yes. In a separate] room, because, (.) because all the-all the students ah want to do ah-

D = yeah.=P = their <homew[ork>]

8 D [may]be they’ll feel disturbed orso[mething like that]

9 P [yeah, aha hmm], aha hmm. AND (.) w- with- with the snackbar, I agree with you, (.3) but uh, (.5) I think uh, the pricesshould be, should be very (.) less, as less as possible.

This exchange is especially rich in terms of collaborative features: Two topicsequences are developed (initiated by Danielle in Turns 2 and 4) with severaltopic extension moves that extend both self- and other-initiated topics. The mostimportant topic development move from the point of view of task fulfillment isthe topic extension move of both self- and other-initiated topics as it fulfills thetask requirements and provides evidence of language proficiency. Danielle’stopic initiation move in Turn 4 (“gambling things,” i.e., electronic games) isdeveloped with topic extensions from both speakers, lasting six turns. Excerpt 1also illustrates the smooth shift in responsibilities of both speaker and hearer.This dual listener/speaker role is basic to conversationalists who, as Maynard(1980) noted, engage in a division of labor in which one party assumes theresponsibilities of topical speaker and the other offers ongoing displays of hearing

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 101

and understanding. They are, in topic management terms, doing topic shift worktogether. Danielle and Peter’s effortless switching between these roles is a criti-cal feature of collaborative talk in this testing event.

This exchange is also illustrative of many links between turns, such as agree-ment (e.g., Turns 3, 6, and 9), syntactic cohesion (e.g., “also” in Turn 2, “thisautomats” in Turn 5, “they” in Turn 8), and lexical cohesion (e.g., “separateroom” in Turn 6 and repetition in Turn 7). There are, furthermore, numerousinstances of listener support in the form of acknowledgment tokens (e.g., Peter’s“hum hmm” in Turn 2 and “yes” in Turn 4; Danielle’s “yeah” in Turn 7) andlistener involvement in the form of latches and overlaps (e.g., Turns 5 to 9),showing high engagement between Peter and Danielle and their sensitivity to theno-gap-no-overlap rule of conversation (Sacks, Schlegloff, & Jefferson, 1974).

The discussion now turns to the talk of a dyad that works collaborativelyexhibiting high equality and mutuality and yet constructing discourse rich in con-versational dominance moves. This is the case with Eric, a Taiwanese man, andMyra, a Taiwanese woman. They are discussing summer jobs. Their openingexchange is given in Excerpt [2].

Excerpt [2] I = Interviewer; E = Eric; M = MyraTask: Summer jobs

I and please speak so that we can hear you1 E ok which one you (want) prefer?2 M A:::H in my opinion we- I wanna choose a hotel3 E I will choose a hotel too ((laughter))4 M YEA:H?5 E yeah6 M why?

(.7)7 E well (.3) basically: (.8) I like to (inform) I like to negotiate with people

M A h[a]E [%you] know% DEAling with people

8 M yeah [me to-]9 E [talk] to people (you [know)]

10 M [yeah] because you know I like to go traveling everywhere so::(.8) if I can working in hotel I can (.) see a lot of people’s different-ah [their from different] fro- (.6) different (from) countries

11 E [Yeah different people] yeah especially when (inside) =12 M = yes I:[we can ]

E [((laughter)) ya:h]13 M we can practice all English and ah maybe French [or Spa]nish with

them =14 E [yeah] = french [Spa ]nish

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

102 GALACZI

15 M [yeah] [( ) benefits]16 E [yeah you can get] ( ) yeah (.6) SHOP?

(.8)17 M N[O: I] don’t think so18 E [yeah] it’s too boring =19 M = yes too boring I don’t- (.3) I don’t like to be suffer from (.3) bore-

dom all the summer20 E yeah i do:n’t. (.6) know (.5) back and forth pick all those milk shelves

[yeah]21 M [how] about a (.3) farm?

Similar to the interaction between Danielle and Peter, these two interlocutors aredoing “substantive topical talk” (Jefferson, 1993, p. 24), that is, they are highlyinteractionally engaged and orient to mutual topic development. Even just a cursoryglance at this transcript, however, reveals the high-involvement nature of this inter-action (Tannen, 1981), characterized by supportive overlaps and interruptions,questions, short turns, rapid speaker change, avoidance of gaps between turns, andstrong listener support in the form of echoes and frequent acknowledgement tokens.

What distinguishes this extract from the previous one are the questions thatthe two interlocutors ask of one another (Turns 1, 4, 6, 16, 21). This is one ofthe main discourse characteristics that distinguishes between the low and highconversational dominance subgroups within the collaborative pattern of inter-action. Danielle and Peter used just one question in their entire interaction. Incontrast, Eric and Myra got through five questions in the space of 21 turns. Aquestion has potential power in the interaction as it imposes conversationallimitations on the next speaker’s contribution. The selected speaker has nochoice but to take the floor and must furthermore be relevant as dictated byGrice’s (1975) maxim of relevance. In the case of Myra and Eric, it is impor-tant to note that both participants used questions as topic extension moves andneither one played a dominating role in the interaction. Furthermore, the ques-tions were woven into the interaction as follow-up questions that function astopic extension moves and thus develop the task further in a very conversa-tional manner. In contrast, as the discussion of parallel high conversationaldominance dyads indicates, the latter dyads used a different type of conversa-tional dominance, namely, participatory dominance (e.g., interruptions), whichwas not facilitative and led to a great deal of competition for the floor.

Parallel Interaction

The second pattern of peer–peer interaction found in the FCE oral tests resem-bled “solo” versus “solo” interaction. In such cases both speakers initiated anddeveloped topics (high equality) but engaged little with each other’s ideas (low

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 103

mutuality). Topic extensions of other-initiated topics were rare in parallel talk,and instead the speakers focused on developing their own contributions. Toparaphrase Teasley and Roschelle’s (1993) term, the participants were creating“parallel problem spaces” (as opposed to a “joint problem space”), which existedalongside one another, but did not come into any substantial contact. Unlike thebalance in the roles of speaker and listener observed in the collaborative dyads,the speaker role here was much stronger than the listener role.

Similar to the collaborative dyads, the dimension of conversational domi-nance was at play in this pattern of interaction, and two parallel talk sub-groups were distinguished. In the low dominance dyads, the interaction wascharacterized by lengthy gaps between turns. The interaction in the highdominance dyads abounded in competition for the floor in the form ofinterruptions.

Parallel talk was structurally manifested in exchanges of the type:

Speaker A: Topic initiation + Topic extensionSpeaker B: Minimal acknowledgement + Topic initiation + Topic extensionSpeaker A: Minimal acknowledgement + Topic initiation + Topic extension

Interactionally, it exhibited the characteristics given in Table 2.The interaction between Ella, a Swiss woman, and William, a Swiss man, is

used to illustrate the interactional features of parallel talk. Excerpt 3 gives thebeginning of their discussion on the task of “hotel jobs”:

TABLE 2Interactional Characteristics of Parallel Interaction

Dimension Interactional Characteristics

Low mutuality • Frequent initiation moves and expansion of self-initiated topics• Limited expansion of other-initiated topics• Pro-forma topic ratification (e.g., just “yeah”)• Fast topic decay• Lack of listener support (e.g., rare acknowledgment tokens)• Few cohesive links between turns (e.g., rare syntactic and lexical

cohesion)

High equality • Quantity of talk balanced between the two participants• Balance between the two participants in the use of topic initiation

moves and topic expansion moves

Conversational dominance

Low conversational dominance dyads:• Extensive use of hesitations and fillers (e.g., ah hum)• Frequent gaps between turns

High conversational dominance dyads:• Frequent instances of competitive overlap and interruptions

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

104 GALACZI

Excerpt [3] I = Interviewer; E = Ella; W = WilliamTask: Hotel jobs

I and please speak so that we can both hear you.1 E → all right. (2) um2 W well, I think the most difficult (.5) for me: would be: (.4) play

piano because (.4) every[body]E [uhm]W know, I can’t play ((laughter))

→ (1)3 E ((laughter)) u:m, I agree absolutely with you. (.6) u:m, ((laughter))

(.6) maybe: the most EAsiest job would be cleaner, but it’s not(.6) really satisfying.

→ (1)4 W → what I could do is- (.5) maybe the job as a waiter. (1) I think it

could be: (.5) done.5 E hmm. hm (.5) well, I would be (your order) too nervous, I sup-

pose ((Laughter)) (.5) Um, (.5) maybe: the most suitable job forme would be: (.) <receptionist>, (.6) because I like (.6) um, Ilike foreigners. I like to: (.8) to talk with them, and (.6) I likealso different language. (.8) so, I think it would me the most(.6) [suitable for me].

6 W [(yeah )] good (for you) definitely7 E ((laughter))

(.6)8 W well the chie:f, (.6) this is not that bad. (.5) also =

E = ah hmm(.5)

W I could (.) learn by doing. (.8) a bit more about (.) cooking.9 E ((laughter))

10 W learn by doing.11 E → Ah (2) uhm

The spate of talk given in Excerpt 3 consists mostly of initiation moves withno substantive topic work by the second speaker: In the space of 11 turns, thereare as many as five topic initiation moves (Turns 2, 3, 4, 5, 8). Out of the five top-ics developed by Ella and William, only three get extended over 2 turns (piano,waiter, and receptionist), and each time the topic extension is very basic. Oppor-tunities for topic extension are not taken up, and instead topics are prematurelyclosed with minimal topic ratification (Turns 3, 5, and 6), laughter (Turns 3, 7,and 9), and gaps between turns (at the arrows). Topics decay rather than gothrough a collaborative development and conclusion. In the 11 turns given in the

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 105

excerpt, there are five gaps longer than 1 second and (Turns 1, 2, 3, 4, 11). Suchlong gaps, strongly indicative of trouble in the conversation (Jefferson, 1993)clearly signal the lack of alignment between the speakers.

The discussion now moves to parallel talk which involves a high level of con-versational dominance. The excerpt is taken from the talk between Petra andElena, two Italian women. The task they are discussing involves selectingunusual sports to be included on a new sports channel. Excerpt 4 illustrates theirtopic development strategies:

[Excerpt 4] P = Petra; E = ElenaTask: New sports channel(33 turns have been omitted)

1 E [this ahm] this is the: (.3) one (.3) I like best. (.4) be[cause]2 P → [yes] maybe because ah [I don’t like ]3 E → [( it’s the one )] the Olympic games I’m always [watching this]4 P → [Ah yes yes] in fact also because these ones are ah- those ones

are not so:5 E violent =6 P = yes this one I think it’s (.3) [BORING] ((laughter))7 E [BORING] ((laughter))8 P I don’t [know]9 E → [becau]se (.) i: dislike (.4) [( )]

10 P → [( )] ah yes in fact I think ah this one also is very difficultbut is ah=

E = ah yesP but is ah ( ) ((laughter)) you know (.6) [and this]

11 E [and this] is the nicest (.) to see I think (.3) becau[se]12 P → [yes] this one and [this one]13 E → [wind] wind surfing [i::s ] the best (.) th[ing]14 P [ah yes] with (.) with the sea also because I: ah tried no I: tried

only once this one but a:h i’m not able ah=

Just a cursory glance at the short exchange in Excerpt 4 reveals fast topic decay andhigh competition for the floor in the form of interruptions and competitive overlap (atthe arrows). Interruptions are clear instances of competition for the floor and viola-tions of an interlocutor’s right to an unimpeded turn. Overlap, on the other hand, canbe both cooperative and competitive. In the case of collaborative high conversationaldominance dyads (Eric/Myra), the frequent overlaps did not result in a topic shift butextended the prior topic or provided support for the speaker. Competitive overlap, onthe other hand, results in both a change of speaker and a change of topic, as seen here.

Another salient feature of the interaction between Elena and Petra is the inci-dence of initiate → minimal response + initiate exchanges (Turns 2, 4, 6, 10, 12)

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

106 GALACZI

where Elena initiates a topic, Petra responds with a minimal response (“yes,” “ahyes”) and then initiates a new topic herself. The response move, which generallyindicates high mutuality in an interaction, is here often just a minimal acknowl-edgment. Petra, who is the main user of this interactional strategy in this dyad,uses the response slot with a response marker of cohesion (“yes”), but thatmarker is in fact just a turn-taking device, which is then often followed by atopic shift. Turns that contain a minimal acknowledgment followed by a topicshift are not uncommon in conversation, as noted by Jefferson (1993). Suchminimal acknowledgment moves establish cohesion with the previous turn butdo so in a pro forma manner and as such become instances of “topical disen-gagement” (Jefferson, 1993, p. 11) where “the display of recipientship isfleeting—a merest nod to the other’s materials before/while launching one’sown” (p. 9). In the case of a collaborative test task, such interactional behav-iour shows lack of mutuality and works against the successful completion ofthe task.

Asymmetric Interaction

The third pattern of interaction found in the data set involved dyads that orientedto different discourse roles, one dominant and one passive, with moderate mutu-ality in topic development. Its main characteristic is the unbalanced quantity oftalk and topic development contributions in the dyad, with one interlocutor lead-ing the interaction and the other taking a secondary role.

From the perspective of conversational dominance, this pattern included twodifferent subpatterns of interaction: one where a dyad member truly appropri-ated the task and took the floor with conversational dominance moves such ascompetitive overlaps and interruptions. In the other subpattern, the dominantparticipant oriented to that role as a result of the passive role and lack of initia-tive of the other dyad member. Often he or she had to prompt the other dyadmember with questions. These different roles have also been recognized inclassroom work as “expert/novice” and “dominant/passive” configurations(Storch, 2002).

Asymmetric talk was structurally manifested in exchanges of the followingtype:

Speaker A: Topic initiation and topic extensionSpeaker B: Minimal acknowledgment (“yes” or “I agree”)Speaker A: Topic extensionSpeaker B: Minimal acknowledgment (“yes” or “I agree”)Speaker A: Topic initiation

Interactionally, it exhibited the characteristics given in Table 3.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 107

To illustrate the salient points of this pattern of interaction, I will discuss theinteraction between Dana, a Taiwanese woman, and John, a Taiwanese man.Consider Excerpt 5, in which Dana and John discuss summer jobs.

Excerpt [5] I = Interviewer; D = Dana; J = JohnTask: Summer jobs

I and please speak so that we can hear you ok?(3)

1 D (so) you wanna start?2 J ok sure (.5) a:hm (.5) ok we have FARM (.) shop and hotel (.5)

a:hm it looks like ((cough)) this column here with the (farm)(.5) ahm (.3) requires more labor

→ (.6)3 D [yes]4 J [so] it’s more physical (1) I think yi-

→ (3)5 D They may keep (.4) ((laughter)) the body shape (good) ((laughter))6 J Right ((laughter)) right a:h7 D ah:m

→ (1)8 J → this one shop it looks more (2) ahm (5) looks more pleasant

because it’s (.3) you just stand there (.) right?9 D yeah [you can] (take)

10 J [usually]11 D can contact with people: =12 J = right you get to meet new customers and all that

TABLE 3Interactional Characteristics of Asymmetric Interaction

Dimension Interactional Characteristics

Moderate mutuality • “Solo” development of topic mainly by one person• Rare expansion of other-initiated topics

Low equality • Unbalanced quantity of talk • lack of balance in terms of topic initiation and topic extension moves

Conversational dominance

Low conversational dominance dyads:• Gaps between turns• Questions by one dyad member inviting participation

High conversational dominance dyads:• Competitive interruptions and overlaps

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

108 GALACZI

→ (1)13 D [hmm]14 J [hote:l]

→ (2) for what I know from ah about hotel is that ah (.4) sometimesthe customer (.) complains a lot (.7) and you have to (face) (.)ah (.) the customer and a:h (.8) to be proactive (.3) before theycomplain too much ((laughter)) SO (.3) this one I think the- a:hyou have to be able to (.3) make sure that (.3) ah you satisfyevery (.) customer’s needs.(2)

15 D (but) so for a summer jobJ HmD think a:hm

(.5)16 J and don’t- they don’t pay a lot ((laughter)) so which one do you

think? a:h which you prefer?(2)

In the first 16 turns John clearly emerges as the more dominating speakeras he speaks more, takes longer turns, makes all the initiation moves (Turns2, 8, 14), and is the only speaker to expand on his ideas. What is significant,however, is that he is mostly put in this position by the lack of initiative ofhis partner who does not take up opportunities for uptake at points of speakertransition (at the arrows). The gaps, which come at transition relevanceplaces, are points of speaker transition and invitations for uptake by the otherspeaker. In this extract, the pauses/gaps are followed not by the expectedtopic expansion move from Dana but only by a listener acknowledgmenttoken from her in the form of a backchannel (“yes”) or a filler (“hmm”).Dana’s minimal responses (Turns 3, 7, and 13) function as turn-passingmoves. After being forced to develop the discourse more or less single-handedly for 16 turns, John finally uses two questions (Turn 16) and forcesDana to take the floor. This behaviour is reminiscent of Storch’s (2002)“expert/novice” pattern of interaction where the more dominant speakerseems to take on the role of teacher who encourages the other participant toparticipate.

So far I have discussed the asymmetric discourse oriented to in a low conver-sational dominance dyad; I now turn to the discourse co-constructed by Franco,an Italian man, and Carina, an Italian woman, which is illustrative of asymmetricinteraction with high conversational dominance.

The task at hand involves a discussion of new shops and a public area to beincluded in a shopping centre.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 109

Excerpt [6] F = Franco; C = CarinaTask: New shopping center

(11 turns have been omitted)1 C [I thi]nk they are the the best to::: to put in this area (.4) because there

are there is an area for children s[o::::]F [hm ]

2 C a toy shop maybe (.4) [easier]3 F [a::h for] me:: (.3) I- I think I am agree with

you for the the public area because a:hm is a good idea to put a (.3)a:h play a- a::h (.7) ( ) a place where the (.) children can pl[ay]

C [uhm]F a::h BUT a::h about the ehm (.8) the store I think it’s better to put

a:hm bakery (.5) and bookshop and a ch- chemi[st]4 C [ye]s yeah yes but

ah I: chose this ah this are three (.) shops because (.) a::hm there is ahalready a::h mini market(.5)

5 F a ha6 C so if you need the:: some bread you can:: (.5) ah go in the mini

mar[ket]F [yes]C there i:s ah for example a newsagent so if you need a:::h some books

(.3) you c[a:n a:h ]7 F [you can] ok

(.4)8 C take THEM to the newsagent. so::: i chose the (.) the chemist the

to[y::: ] shopF [a ha]C et ce te ra for ah

(.4)9 F a:h ok I thi- [yes ]

10 C [may]be11 F a:h could be a good idea because a:h ((cough)) a store (bread) a:nd (1)

a (cloth) a:h

As can be seen, Franco’s contributions to this task are often tentative, short (oftenjust minimal responses) and usually overlapped by the other speaker. Carina, onthe other hand, dominates the interaction by initiating more topic developmentmoves and by taking longer turns. Turn 3 in the excerpt (which is, in fact, Turn 11in the whole exchange) is Franco’s only chance for a full turn. As we can see, afterhis disagreement with Carina in this turn, he quickly loses speakership, and what

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

110 GALACZI

follows his move is a series of turns (4–10) where Carina is mostly the floorholder, starting with a rather impatient sounding “yes yeah yes” in Turn 4. In thisstretch of the interaction Franco takes on a passive role, the result of both Carina’sdominating behavior and his own lack of initiative to take the floor. Franco’s pas-sive role is clearly seen in the minimal responses that he supplies in the responseslots, instead of more fully expanded contributions (Turns 5, 7, 9). Instead of tak-ing more initiative, he provides weak listener support while Carina expounds herideas. She seems to appropriate the task from the beginning by focusing only onher own opinions and not giving the other speaker many chances to participate, asseen in the absence of pauses serving as possible speaker-change points.

The analysis of asymmetric dyads revealed a range of complex reasons behindorienting to a passive or dominant interactional role. It could be argued that inaddition to second-language (L2) proficiency, cultural influence regarding con-versational roles could also play a role, which is a contention proposed by Young(1995) and Ross (1998). Following this line of thought, the more dominantspeaker in a dyad may be interactionally passive because of lack of L2 profi-ciency or simply an expectation to be invited to speak. Similarly, the more domi-nant speaker could be dominating because of inappropriate interactional behavior(as would be the case in a high conversational dominance style of speaking), orhe or she may be put in a more dominant interactional position as a result of thelack of initiative of the other interlocutor. There is a marked difference between adyad member who dominates the task in a domineering manner and a participantwho dominates in a facilitative manner. These different participant roles haveimportant implications for performance testing. The fundamental issue here isone of fairness: How can we be sure that paired test takers receive equal opportu-nities to display their speaking skills? Critics of the paired format (Foot, 1999;Norton, 2005) have voiced similar concerns in the literature, noting that theremay not be equal opportunities for both candidates to perform to the best of theirability. The issue of fairness is perhaps most clearly seen in asymmetric dyads.One of the means of resolving this dilemma is through the design of multiple-task speaking tests, which allow for potential inequalities in participation in apaired task to be redressed by providing adequate opportunities for contributionin the other tasks, which are much more tightly led by the examiner. This is thecase with the design of the FCE speaking test, which has a multitask format; itsdesign works to minimize any interactional imbalance between the two test tak-ers because interaction with the second candidate is generally only part of thetest. As such, if the interaction in the two-way discussion task is unbalancedbecause of the dominant/passive interactional roles the speakers orient to, the testas a whole is balanced as it allows for more balanced participation in three othertasks. The examiner is trained to “afford equal opportunities to both test takersand to elicit the best possible performance from each” (Taylor, 2001). A discourse-based study such as the one presented here, which has highlighted and described

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 111

the different roles dyad members orient to, could serve as a consciousness-raisingendeavour for test developers and rater trainers by explicitly bringing attention tothe issue of unbalanced participation in some dyads and the need to deal consis-tently with asymmetric dyads and achieve the most adequate and fair means ofscoring test takers.

RELATIONSHIP BETWEEN PATTERNS OF INTERACTION AND IC SCORES

Hymes (1974) noted that “it is fascinating to discover the richness of speech butit is also a bit absurd to treat transcribed tapes of interaction as if they were theDead Sea Scrolls” (p. 81). He is aptly referring to the necessity of a link betweenfine-tuned analyses of conversational behavior and the practical applications ofthe gained knowledge. I believe that this study has the potential to provide such alink as it holds implications for assessment because it has the potential to providevalidity evidence for some of the descriptors used in the FCE marking schemefor IC and provide further recommendations about expanding or changing someof the existing performance descriptors.

As a first step in investigating the relationship between IC scores and interac-tional patterns, median IC scores6 were computed for the collaborative andparallel patterns of interaction. In the case of the asymmetric group, twoseparate median scores were computed, one for the dominant and one for thepassive speaker. This was necessary because their interactional roles withinthe dyad were quite different and it was expected that their scores wouldreflect that.

The descriptive statistics for the median scores for each test taker are pre-sented in Table 4. As can be seen, the two collaborative subgroups had the high-est median IC scores (4.50 and 4.25), with the low conversational subgroupsscoring slightly higher than the high conversational dominance subgroup. Thetwo parallel subgroups had the lowest means (3.00 for both subgroups). Theblended and asymmetric patterns of interaction had a median that fell betweenthe collaborative and parallel group medians. This distribution of median val-ues is in keeping with what the CA analysis projected, that is, that collabora-tive pairs, which oriented to a discourse high in both mutuality and equality,as well as supportive conversational dominance features, performed best inthis task and were rated higher. In contrast, parallel and asymmetric dyads

6The median was chosen as a more appropriate measure of central tendency because of the smalldata set.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

112 GALACZI

exhibiting low mutuality or low equality, respectively, performed less well inthis task, as predicted by the CA analysis.

It is important to note that in this part of the analysis claims can be made only interms of tendencies largely because the IC scores are based on performance of fourtasks, whereas the focus here is on the relationship between IC scores and analysisof one task only (Part 3 of the test). This limitation of the study notwithstanding, itwas still deemed beneficial to investigate the relationship between IC score and theinteraction in Part 3 of the speaking exam, as this task is the most interactive task,allowing test takers to initiate turns and topics, exercise speaker selection, and dis-play turn-taking behaviour. As such, this task is potentially the richest in terms offeatures of “interactive communication” and has the closest link with the IC score.

The relationship between higher IC score and pattern of interaction has pro-vided some validity evidence for the IC scores on this test as it has shown anagreement between the CA findings and the score data. I believe this to be animportant finding that can be taken one step further, and the relationship betweenCA findings and IC score data can be used to help inform the performancedescriptors used for IC in the FCE speaking test marking scheme. The analysishas highlighted discourse features that can discriminate between the differentscore bands in terms of IC. The power of such descriptors is that they arelinked to actual L2 test performances, and as such form an empirical basis for themarking scheme. More specifically, the descriptors for the higher bands (i.e., 4.5

TABLE 4Descriptive Statistics for IC Test Scores and Patterns of Interaction

IC Test Score

GroupTest Takers

(N) Mediana SDMin-Max

Score

Collaborative interactionLow conversational dominance 12 4.50 .657 3.0–5.0High conversational dominance 6 4.25 1.05 2.5–5.0

Parallel interactionLow conversational dominance 12 3.00 .537 2.5–4.0High conversational dominance 6 3.00 .418 3.0–4.0

Blended interactionLow conversational dominance 18 4.00 .616 2.5–4.5

Asymmetric interactionDominant speaker 3 4.00 1.00 3.0–5.0Passive speaker 3 3.50 .764 3.0–4.5

Note. Intercoder agreement 91%. IC = Interactive Communication.aMaximum score = 5.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 113

and 5.0) could be partially derived from the discourse features characterizing thecollaborative pattern of interaction; those for bands 3.5 and 4.0 could be partiallyinformed by the discourse features characterizing the blended pattern of interac-tion, and those for bands 2.5 and 3.0 could be informed by the discourse featurescharacterizing the parallel pattern of interaction.

Figure 3 provides a summary of the relationship between patterns of interac-tion, salient discourse features, and IC score. The suggestions are just theoreticaland have not yet been adopted in practice. When that is attempted, the challengewill be to combine the proposed empirically derived descriptors with a theoreti-cal definition of the construct and expert judgment and produce an assessmentscale that covers all five bands. As such, the suggestions in Figure 3 are just aninitial step of a further project.

IMPLICATIONS OF THE STUDY

The significance of the study lies in the more accurate understanding it has pro-vided of the paired oral test interaction generated during the two-way collaborative

FIGURE 3 Operationalisation of descriptors for Interactive Communication (IC).

Discourse characteristics

more self-expansion of

topics

through topic extension

moves

more other-expansion of

topics

through topic extension

moves and follow-up

questions

smoother transition between

turns

through cooperative latches

and overlaps

more listener support

more responsibility for

speaker nomination

through questions

more equality in topic

development

through balance in topic

initiations, topic extensions

and amount of talk

more cohesion between turns

tokens and lexical and

syntactic cohesive links

IC

SCORE

Operationalisation

through backchannels

through acknowledgement

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

114 GALACZI

task in the FCE. It has also provided insights that could aid rater training and leadto more accurate and consistent assessment of FCE candidate output.

A further contribution of the study is the relevance of its findings for FCErating scale construction. Concerns have been voiced in the literature regardingthe empirical basis of the assessment criteria used in performance testing. In thisrespect, Fulcher (1996) criticized scoring rubrics as being frequently based on“armchair notions of language development and structure” (p. 209) and advo-cated for a data-driven approach to developing rating scales. The need for empir-ically and theoretically sound rating scales is a fundamental concern because, asO’Sullivan and Weir (2002) noted, “without efficient, accessible, operationalis-able and construct valid scales the whole rating operation is suspect” (p. 45). Ihope that my study has contributed to this research concern, as one of the goals ininvestigating the nature of talk in the paired oral test format has been to informthe descriptors employed in the FCE scoring rubric for IC.

The findings of this study have implications for second language acquisitionas well. The participants in this study are both L2 test takers and L2 learners.Swain (2001) rightly noted that examining their dialogues as they jointlyconstruct their performance is very much at the interface between the fields ofdiscourse analysis, L2 assessment, and L2 learning. It is possible, I believe, toextrapolate the present findings of dyadic interaction and their alignment withtest scores, and argue for a lower level and higher level conversational manage-ment ability. The results reported here seem to suggest that the conversationalmanagement ability of L2 learners has a higher and a lower level, with collabo-rative dyadic interaction being the higher level skill and parallel dyadic interac-tion the lower-level skill. It may be possible to claim, therefore, that theconversation management abilities of learners at a lower level of languagedevelopment resemble a “parallel” style of interaction where they displaylimited engagement with their interlocutor. At a higher level of conversationmanagement ability, learners develop the ability to work with other conversa-tionalists and shift more successfully between the role of listener and speaker.This is only a tentative claim that needs to become the subject of furtherresearch. It would be informative to see, for example, whether conversationmanagement ability develops in tandem with other aspects of language devel-opment. This question is beyond the scope of this study but would be a naturalnext step investigation.

Finally, the study holds pedagogical insights. It has shown that nonnativespeakers tend to orient to different patterns of interaction. As such, learners mightbenefit from explicit teaching of collaborative dyadic interaction, with a focus onstrategies of extending the prior speaker’s turn, of signaling involvement in theinteraction through follow-up questions, and starting and terminating turns, apoint also advocated by Sayer (2005) in his study on direct instruction of conver-sational skills and its impact on students’ performance.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 115

The compilation of profiles of interactional patterns is clearly a complex undertak-ing, and the results of this analysis can not provide definitive answers. I would notwish to claim that the analysis given here offers a definitive answer to the investigationof the collaborative task in FCE and the IC assessment criterion. Rather, the analysisis an exploratory step toward a more comprehensive picture of test-taker talk in apaired task and is intrinsically incomplete. As Geertz (1973) wrote,

There is an Indian story—at least I heard it as an Indian story—about an English-man who, having been told that the world rested on a platform which rested on theback of an elephant which rested in turn on the back of a turtle, asked . . . what didthe turtle rest on? Another turtle. And that turtle? “Ah, Sahib, after that it’s turtlesall the way down.” (p. 28)

Just like Geertz’s (1973) “turtles” story, my analysis is just the starting point, asthere are many further issues to explore regarding dyadic test-taker talk. Yet,despite its limited scope, this study has highlighted some important strengths andcaveats associated with the FCE paired test format. As such, this study has madeits contribution to understanding the interaction in a high-stakes test. I hope thatthis will have a positive effect on the test stakeholders involved—test developers,test takers, teachers, and researchers—and lead the way to the fairest possibleassessment of speaking ability in the FCE paired speaking test.

ACKNOWLEDGMENTS

This article has its origins in my doctoral thesis, and my sincere thanks go toJim Purpura and Leslie Beebe, who guided me through its completion. Mythanks also go to Cambridge ESOL for giving me access to their data and makingthis study possible, and to Lynda Taylor and the two anonymous reviewers forthe Language Assessment Quarterly for their insightful comments on an earlierdraft of this article.

REFERENCES

Ainsworth-Vaughn, N. (1992). Topic transitions in physician-patient interviews: Power, gender, anddiscourse change. Language in Society, 21, 409–426.

Atkinson, J. M., & Heritage, J. (Eds.). (1984). Structures of social action: Studies in conversationanalysis. Cambridge, UK: Cambridge University Press.

Bachman, L. (1990). Fundamental consideration in language testing. Oxford, UK: Oxford UniversityPress.

Bachman, L. F., & Palmer, A. S. (1996). Language testing in practice. Oxford, UK: Oxford Univer-sity Press.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

116 GALACZI

Brown, A. (2003). Interviewer variation and the co-construction of speaking proficiency. LanguageTesting, 20(1), 1–25.

Brown, A. (2005). Interviewer variability in oral proficiency interviews. Frankfurt, Germany:Peter Lang.

Brown, A. (2006). Candidate discourse in the revised IELTS speaking test. IELTS Research Reports,6, 71–90.

Brown, G., & Yule, G. (1983). Discourse analysis. Cambridge: Cambridge University Press.Button, G. (1991). Conversation-in-a-series. In D. Boden & Zimmerman (Eds.), Talk in social struc-

ture (pp. 251–277).Button, G., & Casey, N. (1984). Generating the topic: The use of topic initial elicitors. In J. M.

Atkinson & J. Heritage, Eds, Structures of social action: Studies in Conversation analysis.Cambridge: Cambridge University Press, 167–190.

Coates, J. (1996). Women Talk. Oxford: Blackwell.Damon, W., & Phelps, E. (1989). Critical distinctions among three approaches to peer education.

International Journal of Educational Research, 58(2), 9–19.Egbert, M. (1998). Miscommunication in language proficiency interviews of first-year German stu-

dents: A comparison with natural conversation. In R. H. Young* A. He (Eds.), Talking and testing(pp. 147–169). Amsterdam: John Benjamins.

Egyud, G., & Glover, P. (2001). Oral testing in pairs—A secondary school perspective. ELT Journal,55(1), 70–76.

FCE: First Certificate in English Handbook. (2002). Cambridge, UK: University of Cambridge LocalExaminations Syndicate.

ffrench, A. (1999). Study of qualitative differences between CPE individual and paired test formats(Internal UCLES EFL report). Cambridge, UK: University of Cambridge Local ExaminationsSyndicate.

Foot, M. C. (1999). Relaxing in pairs. ELT Journal, 53(1), 36–41.Fulcher, G. (1996). Does thick description lead to smart tests? A data-based approach to rating scale

construction. Language Testing, 13(2), 208–238.Galaczi, E. D. (2004). Peer-peer interaction in a paired speaking test: The case of the First Certifi-

cate in English. Unpublished dissertation, Teachers College, Columbia University.Geertz, C. (1973). The interpretation of cultures. New York: Basic Books.Grice, H. (1975). Logic and conversation. In P. Cole & J. Morgan (Eds.), Syntax and semantics. New

York: Academic (p.41–58).Hatch, E., & Lazaraton, A. (1991). The research manual: Design and statistics for applied linguistics.

Boston: Heinle & Heinle.Heritage, J. (1995). Conversation Analysis: Methodological aspects. In U. M. Quasthoff (Ed.),

Aspects of Oral Communication (pp. 391–418). Berlin: Walter deGruyter.He, A. (1998). Answering questions in LPIs: A case study. In R. E. Young & A. He (Eds.), Talking

and testing (pp. 101–116). Amsterdam: John Benjamins.Heritage, J., & Roth, A. L. (1995). Grammar and institution: Questions and questioning in the broad-

cast news interview. Research on Language and Social Interaction, 28(1), 1–60.Hymes, D. (1974). Directions in sociolinguistics. London: Tavistock Publications.Ikeda, K. (1998). The paired learner interview: A preliminary investigation applying Vygotskian

insights. Language, Culture, and Curriculum, 11(1), 71–96.Itakura, H. (2001). Describing conversational dominance. Journal of Pragmatics, 33, 1859–1880.Iwashita, N. (1998). The validity of the paired interview format in oral performance assessment.

Melbourne Papers in Language Testing, 5(2), 51–65.Jefferson, G. (1984). On stepwise transition from talk about a trouble to inappropriately next-position

matters. In J. M. Atkinson & J. Heritage, Eds, Structures of social action: Studies in Conversationanalysis. Cambridge: Cambridge University Press, (191–222).

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 117

Jefferson, G. (1993). Caveat speaker: Preliminary notes on recipient topic-shift implicature. Researchon Language and Social Interaction, 26(1), 1–30.

Johnson, M. (2001). The art of nonconversation. New Haven, CT: Yale University Press.Jones, E. E., & Gerard, H. B. (1967). Foundations of social psychology. New York: Wiley.Kormos, J. (1999). Simulating conversations in oral-proficiency assessment: A conversation analysis

of role plays and non-scripted interviews in language exams. Language Testing, 16(2), 163–188.Lazaraton, A. (1991). A conversation analysis of structure and interaction in the language interview.

Unpublished doctoral dissertation, UCLA.Lazaraton, A. (1992). The structural organization of a language interview: A conversation analytic

perspective. System, 20(3), 373–386.Lazaraton, A. (1996). Interlocutor support in oral proficiency interviews: The case of CASE. Lan-

guage Testing, 13(2), 151–172.Lazaraton, A. (1997). Preference organization in Oral Proficiency Interviews. Research on Language

and Social Interaction, 30, 53–72.Lazaraton, A. (2002). A qualitative approach to the validation of oral language tests. Cambridge,

UK: Cambridge University Press.Lazaraton, A., & Frantz, R. (1996). An analysis of the relationship between task features and candi-

date output for the revised FCE speaking examination. Cambridge, UK: EFL Division, Universityof Cambridge Local Examinations Syndicate.

Linell, P. (1998). Approaching Dialogue. Amsterdam: John Benjamins.Maynard, D. W. (1980). Placement of topic changes in conversation. Semiotica, 30, 263–290.McNamara, T. F. (1997). “Interaction” in second language performance assessment: Whose perfor-

mance? Applied Linguistics, 16(2), 159–179.McNamara, T., Hill, K., & May, L. (2002). Discourse and assessment. Annual Review of Applied Lin-

guistics, 22, 221–242.Moder, C. L., & Halleck, G. B. (1998). Framing the language proficiency interview as a speech event:

Native and non-native speakers’ questions. In R. H. Young & A. He (Eds.), Talking and testing(pp. 117–146). Amsterdam: John Benjamins.

Nakatsuhara, F. (2006). The impact of proficiency-level on conversational styles in paired speakingtests. Cambridge ESOL Research Notes, 25.

Norton, J. (2005). The paired format in the Cambridge speaking tests. ELT Journal, 59(4), 287–296.Ochs, E. (1979). Transcription as theory. In E. Ochs & B. Schieffelin (Eds.), Developmental prag-

matics (pp. 43–72). New York: Academic.O’Loughlin, K. (2002). The impact of gender in oral proficiency testing. Language Testing, 19(2),

169–192.O’Sullivan, B. (2002). Learner acquaintanceship and oral proficiency test pair-task performance.

Language Testing, 19(3), 277–295.O’Sullivan, B., & Weir, C. J. (2002). Research issues in testing spoken language. Cambridge, UK:

University of Cambridge Local Examinations Syndicate.Pomerantz, A., & Fehr, B. J. (1997). Conversation analysis: An approach to the study of social

action as sense making practices. In T. A. van Dijk (Ed.), Discourse as social interaction.London: Sage (Vol. 3, pp. 64–91).

Psathas, G. (1995). Conversation Analysis: The study of talk-in-interaction. Thousand Oaks, CA: Sage.Ross, S. (1992). Accommodative questions in oral proficiency interviews. Language Testing, 9(2),

173–186.Ross, S. (1998). Divergent frame interpretations in language proficiency interview interaction. In

R. Young & A. He (Eds.), Talking and testing (pp. 333–353). Amsterdam: John Benjamins.Sacks, E. A., Schegloff, E. A., & Jefferson, G. (1974). A simplest systematics for the organization of

turn-taking for conversation. Language, 50, 696–735.Sayer, P. (2005). An intensive approach to building conversation skills. ELT Journal, 59(1), 14–22.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

118 GALACZI

Schegloff, E. A. (1993). Reflections on quantification in the study of conversation. Research onLanguage and Social Interaction, 26, 99–128.

Skehan, P. (2001). Tasks and language performance assessment. In M. Bygate, P. Skehan, &M. Swain (Eds.), Researching pedagogic tasks (pp. 167–185). London: Longman.

Some guidelines for conducting quantitative and qualitative research in TESOL. (2003). TESOLQuarterly, 37(1), 157–178.

Storch, N. (2002). Patterns of interaction in ESL pair work. Language Learning, 52(1), 119–158.Swain, M. (2001). Examining dialogue: another approach to content specification and to validating

inferences drawn from test scores. Language Testing, 18(3), 275–302.Tannen, D. (1981). New York Jewish conversational style. International Journal of the Sociology of

Language, 30, 133–149.Taylor, L. (2001). The paired speaking test format: Recent studies. Cambridge ESOL Research Notes, 6.Teasley, S., & Roschelle, J. (1993). Constructing a joint problem space: The computer as a tool

for sharing knowledge. In S. P. Lajoie & S. J. Derry (Eds.), Computers as cognitive tools(pp. 229–258). Hillsdale, NJ: Lawrence Erlbaum Associates.

ten Have, P. (1999). Doing conversation analysis. London: Sage.Tracy, K., & Moran, J. P. (1983). Conversational relevance in multi-goal settings. In T. Craig &

K. Tracy (Eds.), Conversational coherence (pp. 116–135). London: Sage.van Lier, L. (1989). Reeling, writhing, drawling, stretching and fainting in coils: Oral proficiency

interviews as conversations. TESOL Quarterly, 23, 480–508.Yngve, V. H. (1970). On getting a word in edgewise. Papers from the Sixth Regional Meeting of the

Chicago Linguistics Society, 6, 657–677.Yoshida-Morise, Y. (1998). The use of communication strategies in language proficiency interviews.

In R. H. Young & A. He (Eds.), Talking and testing (pp. 205–238). Amsterdam: John Benjamins.Young, R. (1995). Conversational styles in language proficiency interviews. Language Learning,

45(1), 3–42.Young, R., & Milanovic, M. (1992). Discourse variation in oral proficiency interviews. Studies in

Second Language Acquisition, 14, 403–424.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

PEER–PEER INTERACTION 119

APPENDIX

Transcription Conventions

[ Beginning of overlapping utterances] End of overlapping utterances= A turn latched immediately to a previous speaker’s turn with no overlap(.) Pause shorter than .2 seconds(.8) The length of a pauseUnderline StressCAPS Very emphatic stress% % Softly spoken sounds< > Decreased speed> < Increased speed, Falling intonation within the turn. Falling intonation at the end of the turn? Rising intonation (not necessarily a question)- Abrupt cutting off of sound: lengthened vowel sound (extra colons indicate greater lengthening)(( )) Non-verbal action, e.g. laughter( ) Doubtful transcription→ Highlights point of analysis

Note. Adapted from Atkinson and Heritage (1984).

Dow

nloa

ded

by [

Uni

vers

ity o

f C

ambr

idge

] at

02:

32 1

9 A

ugus

t 201

5

View publication statsView publication stats