Collaborative dialogue in learning pragmatics: Pragmatics-related episodes as an opportunity for...

Post on 12-Mar-2023

0 views 0 download

Transcript of Collaborative dialogue in learning pragmatics: Pragmatics-related episodes as an opportunity for...

Applied Linguistics 2014 1ndash23 Oxford University Press 2014

doi101093applinamu039

Collaborative Dialogue in LearningPragmatics Pragmatic-Related Episodesas an Opportunity for LearningRequest-Making

1NAOKO TAGUCHI and 2YOUJIN KIM1Modern Languages Department Carnegie Mellon University Pittsburgh PA USA and2Department of Applied Linguistics and ESL Georgia State University Atlanta GA USA

E-mail taguchiandrewcmuedu

This study examined the effects of collaborative dialogue in learning the speech

act of request Seventy-four second-grade girlsrsquo junior high students were

divided into three groups The lsquocollaborative grouprsquo (n = 25) received explicit

metapragmatic information on request (request head act and modifications)

followed by a dialogue construction task in pairs The lsquoindividual grouprsquo

(n = 25) received the same information but completed the same task individually

while thinking aloud The last group control group (n = 24) did not receive

instruction During-task interaction in the collaborative group and think-aloud

protocols in the individual group were audio-recorded Instructional effect was

measured by a discourse completion task (DCT) Target request head acts in

DCT were scored and request modifications were analyzed for frequency

The collaborative group outperformed the individual group on the production

of the head act at immediate post No group difference was found in request

modifications Analysis of interaction and think-aloud data showed that the

collaborative group produced the target head act more successfully than the

individual group but no group difference was found in the use of modifications

Task-based learning in language classrooms has received much attention

among second language (L2) researchers and practitioners (Van den

Branden et al 2009) Tasks are often used to generate interaction and negoti-

ation of meaning during pairgroup interaction which leads to language learn-

ing Collaborative dialogue in which learners use language as a tool for their L2

development has been the prime topic of investigation in the field of task-

based interaction (Swain and Watanabe 2013) However previous collabora-

tive dialogue studies have mainly focused on grammar and vocabulary and

they have not included pragmatics as a target language area In pragmatics

instructional research has been concentrated in the comparison of explicit and

implicit teaching (Takahashi 2010 Taguchi forthcoming) and very few studies

have examined other instructional approaches including task-based teaching

This study intends to bring these two fields together by investigating the effect

of task-based collaborative dialogue in teaching pragmatics

Applied Linguistics Advance Access published July 29 2014 by guest on July 30 2014

httpapplijoxfordjournalsorgD

ownloaded from

BACKGROUND

Pragmatic competence reflects an intertwined relationship between pragma-

linguistics and sociopragmatics Pragmalinguistics refers to linguistic resources

needed to perform language functions while sociopragmatics refers to the

language userrsquos assessment of the context in which such resources are put

into use (Leech 1983) There has been a steep increase in the body of instruc-

tional pragmatics studies that focuses on the role of instruction in developing

L2 learnersrsquo knowledge of pragmalinguistics and sociopragmatics (for a review

see Jeon and Kaya 2006 Takahashi 2010 Taguchi 2011) The benefits of in-

struction over non-instruction were found consistently across different lan-

guages teaching methods and assessment tasks and presented strong

evidence that teaching leads to increased pragmatic knowledge More recent

studies have compared the impact of different instructional methods including

explicit and implicit teaching input processing instruction and skill acquisi-

tion Theoretical assumptions behind these methods have informed us about

underlying cognitive mechanisms that drive pragmatics learning

Schmidtrsquos (2001) noticing hypothesis and explicit vs implicit teaching moti-

vated by the hypothesis currently dominates the field of instructed pragmatics

The noticing hypothesis claims that learnersrsquo attention to linguistic forms their

functions and relevant contextual factors serve as a necessary condition for

pragmatic input to become intake Following Kasper (2001) most studies

operationalized explicit teaching to involve direct metapragmatic explanation

followed by structured practice while implicit teaching holds back metaprag-

matic information and instead assists learnersrsquo implicit understandings of the

target features Previous research generally confirmed the superiority of the

explicit over implicit method particularly in outcome measures that involve

greater cognitive demand (Taguch forthcoming)

Although the explicit and implicit debate has prevailed in the literature for

quite some time the field has recently expanded the theoretical scope by

adding new frameworks for pragmatic instruction Particularly relevant to

this study is the concept of collaborative dialogue in the construction of lin-

guistic knowledge which is often operationalized as language-related episodes

(LREs) (Swain and Lapkin 1998 Swain 2006) Collaborative dialogue is a form

of output but it is an output used for a cognitive function because language

mediates learnersrsquo process of working together to solve linguistic problems and

jointly construct knowledge (Swain and Lapkin 1998) According to Swain and

Lapkin (1998) LREs refers to lsquoany part of a dialogue where the students talk

about the language they are producing question their language use or correct

themselves or othersrsquo (p 326) and have been found to be beneficial for

learning

Because tasks are considered vehicles that promote interaction and negoti-

ation collaborative dialogue (operationalized as LREs) during task perform-

ance has received an increasing attention in the field of task-based language

teaching (Kim in press) Research in grammar and vocabulary learning has

2 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

investigated to what extent collaborative dialogue during interactive tasks pro-

motes language learning compared with individual task conditions revealing

advantages of the former For instance in Kimrsquos (2008) study of L2 Korean

individual grouprsquos think-aloud data and collaborative grouprsquos interaction data

were analyzed for LREs and their vocabulary learning was compared Results

showed that although the number of LREs was similar between the groups

the latter performed significantly better on vocabulary tests Nassaji and Tian

(2010) also compared collaborative and individual task conditions on phrasal

verbs learning when learners carried out two tasks (editing tasks and cloze

tasks) in a classroom Collaborative tasks resulted in a greater accuracy of

task completion than individual tasks but they did not lead to greater vocabu-

lary gain The authors suggested the task effect editing tasks were more ef-

fective than cloze tasks in promoting negotiation and learning As shown

above the benefit of collaborative dialogue and LREs stemming from task-

based interaction is noteworthy In order to confirm the effectiveness of col-

laborative dialogue during interactive tasks compared with learnersrsquo attention

to language forms during individual tasks more studies are needed in other

language areas such as pragmatics

In pragmatics learning this article considers that collaborative dialogues

(ie LREs) go beyond just language forms They generate moments in which

learners talk about the form and its relevance to the function and context of

language use During interaction pragmalinguistic forms and contextual fac-

tors are constantly emphasized negotiated and recycled for use This process

helps consolidate pragmatic knowledge because it prompts a deeper level of

cognitive processing by requiring learners to think through pragmatic rules

while verbalizing their thoughts Collaborative dialogue enables learners to

negotiate and co-construct pragmatic knowledge learners can discuss prag-

matic forms and contextual features associated with them and develop a

joint understanding of the principles underlying the associations

Several pragmatics studies have used collaborative dialogue as a treatment

method in a format of metapragmatic discussion (eg Nguyen 2013)

However these studies used it in combination with other activities and thus

failed to reveal the sole effect of metapragmatic discussion To our knowledge

Takimoto (2012) is the only study that directly tested the efficacy of metaprag-

matic discussion Takimoto compared two conditions in teaching English re-

quest downgraders consciousness-raising instruction with and without

metapragmatic discussion Both treatment groups analyzed request-making

expressions with contextual features and rated the appropriateness of the

expressions Then they generated a list of ways to make the requests more

appropriate One group made a list collaboratively while the other group did

so individually The metapragmatic discussion group outperformed their indi-

vidual task counterpart on the production of requests which suggests the

benefit of collaborative dialogue in developing pragmatic knowledge

Given the paucity of findings on the role of collaborative dialogue in prag-

matics it is pertinent to examine to what extent learners negotiate and

N TAGUCHI AND Y KIM 3

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

co-construct knowledge of sociopragmatic factors and pragmalinguistic fea-

tures during collaborative tasks Furthermore the question of how collabora-

tive dialogue during interactive tasks promotes such processes and subsequent

learning of pragmatics compared with during individual tasks warrants further

investigation Theoretical interest in collaborative dialogue is about language

being used as a cognitive tool that mediates the process of thinking (Swain

2006 Swain and Watanabe 2013) Languaging contends that verbalization of

target pragmatic forms when guided strategically through a collaborative task

could lead to a deeper-level understanding of the features Although

Takimotorsquos study tapped into this theoretical underpinning of collaborative

dialogue and revealed the benefits of learnerndashlearner interaction in pragmatic

learning actual mechanisms of learning during interaction were not discern-

ible because he did not analyze interaction data for evidence of learning oppor-

tunities He analyzed only the posttest results (appropriateness judgment and

production of request downgraders) To gain a fuller account of the theoretical

framework of collaborative dialogue it is necessary to analyze the quality of

LREs targeting pragmatics which we call pragmatic-related episodes (PREs) and

link the analyses to learning outcomes Furthermore in order to compare the

amount of languaging between individual tasks and collaborative tasks it is

also important to document how learners performing individual tasks con-

struct their knowledge of pragmatics on their own Thus in the current

study we collected learnersrsquo think-aloud protocols in the individual task

group and they were analyzed for PREs as well

In this study we examine the effects of collaborative dialogue on the acqui-

sition of speech act of request in English by comparing three groups the lsquocol-

laborative grouprsquo who completes the dialogue construction tasks targeting a

request in pairs the lsquoindividual grouprsquo who completes the task alone and the

control group who does not receive any task-based instruction In addition we

investigate the characteristics of collaborative dialogue (operationalized as

PREs) as language resources that facilitate learning Following Swain and

Lapkin (1998) we define PREs as any part of language production where

learners talk about the pragmalinguistic forms they are producing and the

sociopragmatic factors they are attending to (eg setting and interlocutor re-

lationship) question their pragmatic language use or correct themselves or

others In this study we specifically focus on the frequency of PREs produced

during collaborative and individual tasks and instances of correctly solved

pragmatic problems to see whether PREs provide opportunities for learning

pragmatics This study asks two research questions

1 What is the effect of task-based pragmatic instruction on L2 pragmatic

development Are there any differences in the learning of the request

speech act between the collaborative and individual task group

2 Are there any differences in the frequency of PREs and quality of task

performance during instruction between the collaborative and individual

task group

4 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

METHODS

Participants

Participants were 74 learners of English in a second-grade girlsrsquo junior high

school in South Korea Their average age was 1374 years (range = 13ndash14

years) They had received 5 years of formal English education at the time of

the study They were receiving 4 hours of English instruction per week taught

mostly by Korean-speaking instructors Participantsrsquo proficiency measured by

Test of English for International Communication (TOEIC) Bridge was between

high beginner and intermediate with a mean score of 13527 (SD = 1442

range = 100ndash172) Participantsrsquo English classes focused on grammar vocabu-

lary and readingwriting with little focus on pragmatics Three intact classes

were randomly assigned to three groups a collaborative (n = 25) individual

(n = 25) and control group (n = 24) Analysis of variance showed no significant

group difference in the TOEIC Bridge scores suggesting that the groups were

comparable in terms of their proficiency level (F = 355 p = 70)

Instructional targets

The instructional target was the speech act of request in a formal situation

which appeared in the participantsrsquo English textbook (Lee et al 2009) Formal

request was operationalized using Brown and Levinsonrsquos (1987) contextual

factors power (P) distance (D) and degree of imposition (R) and was defined

as a request that carries a larger size of imposition and is made to someone in a

greater power and distance (PDR-high) An example is asking a professor to

reschedule a test This PDR-high request contrasts with a request made to

someone in equal power and small distance and degree of imposition (PDR-

low) such as asking your sister to pass you a TV remote

In order to identify appropriate request situations a pilot study was con-

ducted with students of the same grade in the institution who were not

included in the main study (n = 34) This involved asking them to indicate

the degree of psychological ease or difficulty in performing the request on

Likert scale from 1 (easy) to 5 (difficult) The survey included 14 PDR-high

and -low requests adapted from the previous literature (eg Taguchi 2012)

The survey also asked participants to indicate the degree of commonality of

each situation on a scale from 1 (very rare) to 5 (very common) Finally the

survey provided an open question in which students wrote PDR-high and -low

requests that they had personally encountered

Results showed that PDR-low situations received a mean difficulty rating of

178 (SD = 106) while PDR-high situations received a mean rating of 264

(SD = 124) This difference was statistically significant t (237) =900

p = 0001 indicating that PDR-high were perceived as more difficult to perform

than PDR-low requests From the 14 situations we selected four PDR-high

items that received a rating of 25 or above and four PDR-low items with a

rating of below 15 In order to ensure the authenticity of tasks commonality

N TAGUCHI AND Y KIM 5

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

ratings of these items were checked to avoid teaching situations that do not

represent real life in Korea Finally we analyzed participantsrsquo responses to the

open-ended question and used their descriptions when writing situations

Instruction focused on two categories of pragmalinguistic forms request

head acts and modifications (Table 1) Request head acts refer to the minimal

core unit that conveys the illocutionary force of request (Blum-Kulka et al

1989) In this category we taught mitigated preparatory forms Mitigated pre-

paratory involves syntactic forms that make reference to the hearerrsquos ability or

will in an embedded question or bi-clausal structure (eg lsquoI was wondering

ifrsquo + verb and lsquoWould it be possible torsquo + verb) (Takahashi 1996) Request modi-

fications have two categories external and internal External modifications

involve semantic moves that are attached externally to the head act and miti-

gate its force (Blum-Kulka et al 1989) We taught two external modifications

preparators (preparing a hearer for a request) and grounders (giving a reason

for request) Internal modifications include sentence-internal syntactic and

lexical devices used to modify the force of request We taught two forms

hedging that softens the tone of speech and amplifiers that strengthen self-

expression

These pragmalinguistic forms were selected based on baseline data collected

from native English speakers Twenty-four students in a US university

Table 1 Target pragmalinguistic forms

Head acts

Mitigated preparatory forms

Reference to the hearerrsquos ability will and possibility or reference to thespeakerrsquos wish in an embedded question or sentence

Irsquom wondering if you could give me an extension on the assignment

Is there any way that I could get an extension on the assignment

External modifications

Preparators

Semantic moves used to prepare the hearer for the request

eg May I ask you something

Grounders

Reason or explanation used to support the request

eg I caught a cold so I couldnrsquot finish the assignment

Internal modifications

Hedging

Words that minimize self-expression (maybe possibly)

eg Irsquom wondering if I could possibly have an extension on the assignment

Amplifiers

Words that strengthen self-expression (really very)

eg I really need more time to work on the assignment

6 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

completed a discourse completion task (DCT) that contained four PDR-high

and -low requests Mitigated preparatory appeared 73 percent of the time as a

head act in PDR-high requests while it appeared only 4 percent of the time in

PDR-low requests This indicates that mitigated preparatory characterizes PDR-

high requests Grounders and preparators were common supportive moves

appearing 98 and 13 percent of the time respectively while they never ap-

peared in PDR-low situations Hedging and amplifiers were also frequent ap-

pearing at a rate of 52 percent while they were absent in PDR-low requests

Based on these findings along with studies that revealed learnersrsquo difficulty

with these forms (Taguchi 2012) we targeted these forms Although DCT data

can only tell us what people would say in a hypothetical situation the rela-

tively uniform responses were considered to represent normative patterns of

request making

Instructional materials and tasks

Treatment groups received instruction in school during their regular English

class sessions over two consecutive days The instruction started with a 5-min

explanation of target pragmatic forms using a written dialogue There were

two dialogues one featuring a PDR-high and the other featuring a PDR-low

request We introduced pragmalinguistic forms and sociopragmatic variables in

each dialogue After receiving the direct metapragmatic information the two

treatment groups (collaborative and individual) proceeded to the dialogue con-

struction task during which learners were asked to complete drama scripts

based on given scenarios as drama scriptwriters They received two scenarios

(PDR-high and -low) with pictures of main characters and created a dialogue

involving a request based on each scenario1 These tasks were considered au-

thentic and relevant to learner interests because TV dramas are popular among

the target population and all the task scenarios were directly related to their

school life2 Learners constructed three PDR-high and -low situation dialogues

during treatment sessions (see Supplementary material for sample instruc-

tional materials) Although PDR-high requests were the instructional target

we introduced PDR-high and -low requests together so the differences be-

tween the two become salient

The collaborative group created a dialogue in pairs while the individual

group completed the task alone Learners vocalized their thoughts during

task performance in the language of their choice either collaboratively or

individually depending on their treatment condition To ensure learnersrsquo

understanding of the task following Kim (2013) we showed a 2-min pre-

task modeling video prior to the beginning of their task performance In the

modeling video for the collaborative group two teachers demonstrate how to

perform a task collaboratively using a similar scenario whereas in the video for

the individual group one teacher performed the same task while thinking

aloud on her own Interaction and think-aloud protocols during task comple-

tion were audio-recorded using individual MP3 recorders The treatment

N TAGUCHI AND Y KIM 7

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

session took about 45 min We repeated the same procedures in the second

task treatment session with different scenarios The control group received

regular English instruction

Assessment measure

A written DCT instrument was used to measure learning outcomes DCT

has been criticized because of a lack of authenticity and non-interactive

nature (Golato 2003) However DCT has a merit because the data can provide

information about learnersrsquo knowledge of normative conventions of prag-

matic language use (McNamara and Roever 2006) DCT can also control

social factors in scenarios and help us obtain data that are comparable across

learners over time Most importantly DCT in this study conformed to the

principle of lsquotransfer appropriate processingrsquo (DeKeyser 2007) which claims

that transfer of skill from a learned to a novel task occurs if the cognitive

operations involved in the novel task resemble those in the treatment task

The treatment task in this study was a dialogue construction task which

shared the same modality with DCT which elicited participantsrsquo pragmalinguis-

tic forms in writing

The DCT had 15 items 4 each of PDR-high and PDR-low requests and 7

filler items involving non-target speech acts Each DCT item had a situation

written in Korean to ensure participantsrsquo comprehension Participants were

asked to follow the first turn or prompt provided in English and write the

speech act in English In order to minimize the practice effect coming from

administering the same test repeatedly we prepared three versions of DCT

Scenarios were kept constant but minor wording changes were made in the

scenarios (eg changing names and locations) Different filler items were used

each time DCT items were different from the items used in the treatment

sessions Supplementary material contains sample DCT items

Procedure

The study was conducted over 6 weeks (see Table 2) During the first ses-

sion the learners took the pretest and two task groups carried out a practice

task to become familiar with recording devices and think-aloud

procedures Task treatment sessions were provided on Days 7 and 9 On

Days 10 and 14 all learners took the immediate posttest and the TOEFIC

Bridge respectively Four weeks after the immediate posttest the learners

took the delayed posttest

Data analysis

This study asked (i) whether task-based instruction is beneficial for learning

request-making expressions and if so whether there is a difference in learning

outcomes between the collaborative and individual group and (ii) to what

extent learners produce PREs during individual or collaborative tasks In

8 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

response to the first research question we first assigned scores on the request

head acts in the DCTs Because the purpose of this study was to compare

instructional effects between two task conditions (collaborative vs individual)

we focused our analysis on the forms we targeted during instruction Three

points were given if the head act contained one of the target forms (lsquoIrsquom

wondering ifrsquo + clause or lsquoIs there any wayrsquo + clause) and was also grammat-

ically accurate Two points were given if the head act took the target form but

was ungrammatical One point was given for a grammatical non-target form in

the context of this study (request in a form other than the mitigated prepara-

tory such as lsquoCan Irsquo) No point was given for an ungrammatical non-target

form and missing response3 Total scores from four PDR-high requests (scale

0ndash12) were compared across collaborative individual and control groups

using the KruskalndashWallis test In addition to head acts frequency of request

modifications (preparators grounders hedging and amplifiers) was counted

separately and compared across groups using the KruskalndashWallis test Both

authors coded 50 of the data yielding 953 agreement rate The discre-

pancies were solved through a follow-up discussion The non-parametric test

(KruskalndashWallis) was used because the data did not confirm normal

distribution

To answer the second research question task interaction and think-aloud

data were transcribed and analyzed for PREs defined as any part of language

production where learners talk about pragmalinguistics (request-making

forms) and sociopragmatics (contextual factors) they are attending question

their pragmatic language use or correct themselves or others PREs were coded

for context (eg setting interlocutor relationship and size of imposition) head

Table 2 Study procedures

Session Individual group Collaborative group Control group

Session 1(Day 1)

Pretest (DCT 1)Practice taskwith thinking aloud

Pretest (DCT 1)Practice task

Pretest (DCT 1)

Session 2(Day 7)

Explicit information(5 min)

Task modeling video(2 min)

Individual dialogueconstruction (35 min)

Explicit information (5 min)

Task modeling video (2 min)

Collaborative dialogueconstruction (35 min)

Reading activity

Session 3(Day 9)

Same procedure withdifferent scenarios

Same procedure with differentscenarios

Reading activity

Session 4(Day 10)

Immediate posttest (DCT 2) Immediate posttest (DCT 2) Immediateposttest (DCT 2)

Session 5(Day 14)

TOEIC Bridge TOEIC Bridge TOEIC Bridge

Session 6(Day 38)

Delayed posttest (DCT 3) Delayed posttest (DCT 3) Delayedposttest (DCT 3)

N TAGUCHI AND Y KIM 9

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

acts grounders preparators hedging and amplifiers See the following ex-

amples about asking the school principal to change a location of a picnic

Example 1 PRE targeting a request head act during interactive task

1 Learner 1 (making a request) I am wondering if I could pos-

sibly (First we have to make a request) I am won-

dering if you could possibly I was wondering

2 Learner 2 I am

3 Learner 1 I am wondering

4 Learner 2 if you could was could (Since

we used lsquocouldrsquo shouldnrsquot we use lsquowasrsquo) I was wondering if I could go

to picnic in Everland

5 Learner 1 if you could (No

since this person needs to make a decision lsquoif you couldrsquo sounds right) If

you could

Example 2 PRE targeting the request head act during think-aloud

Learner 1 In this fall fall picnic fall picnic picnic is there any is there

any way that I could that I could could is there can could possibly

possibly go to Everland participate possibly go to go to Everland

Once the PREs were coded studentsrsquo task performance data (the completed

written dialogues) was scored following the same rubrics as the DCT data and

the frequency of each modification in the dialogues was also counted The

second rater coded 20 percent of task performance data independently and

94 percent agreement was obtained Any disagreement was resolved through

discussion which was applied for the rest of the data coding

RESULTS

Effects of collaborative dialogue in learning the speech act ofrequest

Analysis of request head act

Table 3 displays descriptive statistics of head act scores Frequency analysis

showed that 295 out of the 296 head acts (74 learners 4 items) were non-

target (not the focus of instruction) Most of the non-target head acts took a

form of permission (lsquoMay Irsquo) or ability inquiry (lsquoCould yoursquo) indicating that all

groups were unfamiliar with the mitigated preparatory forms before the in-

struction The KruskalndashWallis test revealed no group difference at pretest

2 = 344 p = 18

The KruskalndashWallis test found a significant group difference at immediate

posttest 2 = 3190 plt 001 Pair-wise comparisons using the MannndashWhitney

U test showed that both treatment groups outperformed the control group

10 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

z = 536 plt 001 (collaborative group) and z = 405 plt 001 (individual group)

The collaborative group further surpassed the individual group on the produc-

tion of mitigated preparatory z = 240 p = 016 However the benefit of this

condition disappeared at delayed posttest because there was no difference be-

tween the treatment groups z = 38 p = 71 In addition the KruskalndashWallis test

showed that there was no significant difference between the control and treat-

ment groups at delayed posttest 2 = 25 p = 884

In summary task-based instruction resulted in a strong effect on the learn-

ing of appropriate request head acts as found in the treatment groupsrsquo superior

performance at immediate posttest compared with that of non-instructional

condition However the effect did not last long as the treatment groups went

back to the level of the control group at delayed posttest Similarly the col-

laborative task was more effective than the individual task at the immediate

posttest but the effect disappeared 1 month later

Analysis of request modification

Tables 4ndash7 present frequency counts of request modifications that appeared in

DCT Preparators almost never appeared at pretest but they showed a large

increase in the two treatment groups at immediate posttest (Table 4) In con-

trast preparators were absent in the control group Both treatment groups

outperformed the control group at immediate and delayed posttest No differ-

ence was found between the collaborative and individual groups z = 176

Table 3 Request head act scores

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 122 200 400

Immediate posttest 1012 306 200 1200

Delayed posttest 432 236 200 1100

Individual group (n = 25)

Pretest 340 071 200 400

Immediate posttest 836 341 300 1200

Delayed posttest 488 273 100 1200

Control group (n = 24)

Pretest 375 043 300 400

Immediate posttest 383 047 200 400

Delayed posttest 388 044 200 400

Note Each student produced four PDR-high requests Each head act was scored on a 3-point

scale Score range = 0ndash12 Delayed posttest was given 1 month later

N TAGUCHI AND Y KIM 11

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

BACKGROUND

Pragmatic competence reflects an intertwined relationship between pragma-

linguistics and sociopragmatics Pragmalinguistics refers to linguistic resources

needed to perform language functions while sociopragmatics refers to the

language userrsquos assessment of the context in which such resources are put

into use (Leech 1983) There has been a steep increase in the body of instruc-

tional pragmatics studies that focuses on the role of instruction in developing

L2 learnersrsquo knowledge of pragmalinguistics and sociopragmatics (for a review

see Jeon and Kaya 2006 Takahashi 2010 Taguchi 2011) The benefits of in-

struction over non-instruction were found consistently across different lan-

guages teaching methods and assessment tasks and presented strong

evidence that teaching leads to increased pragmatic knowledge More recent

studies have compared the impact of different instructional methods including

explicit and implicit teaching input processing instruction and skill acquisi-

tion Theoretical assumptions behind these methods have informed us about

underlying cognitive mechanisms that drive pragmatics learning

Schmidtrsquos (2001) noticing hypothesis and explicit vs implicit teaching moti-

vated by the hypothesis currently dominates the field of instructed pragmatics

The noticing hypothesis claims that learnersrsquo attention to linguistic forms their

functions and relevant contextual factors serve as a necessary condition for

pragmatic input to become intake Following Kasper (2001) most studies

operationalized explicit teaching to involve direct metapragmatic explanation

followed by structured practice while implicit teaching holds back metaprag-

matic information and instead assists learnersrsquo implicit understandings of the

target features Previous research generally confirmed the superiority of the

explicit over implicit method particularly in outcome measures that involve

greater cognitive demand (Taguch forthcoming)

Although the explicit and implicit debate has prevailed in the literature for

quite some time the field has recently expanded the theoretical scope by

adding new frameworks for pragmatic instruction Particularly relevant to

this study is the concept of collaborative dialogue in the construction of lin-

guistic knowledge which is often operationalized as language-related episodes

(LREs) (Swain and Lapkin 1998 Swain 2006) Collaborative dialogue is a form

of output but it is an output used for a cognitive function because language

mediates learnersrsquo process of working together to solve linguistic problems and

jointly construct knowledge (Swain and Lapkin 1998) According to Swain and

Lapkin (1998) LREs refers to lsquoany part of a dialogue where the students talk

about the language they are producing question their language use or correct

themselves or othersrsquo (p 326) and have been found to be beneficial for

learning

Because tasks are considered vehicles that promote interaction and negoti-

ation collaborative dialogue (operationalized as LREs) during task perform-

ance has received an increasing attention in the field of task-based language

teaching (Kim in press) Research in grammar and vocabulary learning has

2 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

investigated to what extent collaborative dialogue during interactive tasks pro-

motes language learning compared with individual task conditions revealing

advantages of the former For instance in Kimrsquos (2008) study of L2 Korean

individual grouprsquos think-aloud data and collaborative grouprsquos interaction data

were analyzed for LREs and their vocabulary learning was compared Results

showed that although the number of LREs was similar between the groups

the latter performed significantly better on vocabulary tests Nassaji and Tian

(2010) also compared collaborative and individual task conditions on phrasal

verbs learning when learners carried out two tasks (editing tasks and cloze

tasks) in a classroom Collaborative tasks resulted in a greater accuracy of

task completion than individual tasks but they did not lead to greater vocabu-

lary gain The authors suggested the task effect editing tasks were more ef-

fective than cloze tasks in promoting negotiation and learning As shown

above the benefit of collaborative dialogue and LREs stemming from task-

based interaction is noteworthy In order to confirm the effectiveness of col-

laborative dialogue during interactive tasks compared with learnersrsquo attention

to language forms during individual tasks more studies are needed in other

language areas such as pragmatics

In pragmatics learning this article considers that collaborative dialogues

(ie LREs) go beyond just language forms They generate moments in which

learners talk about the form and its relevance to the function and context of

language use During interaction pragmalinguistic forms and contextual fac-

tors are constantly emphasized negotiated and recycled for use This process

helps consolidate pragmatic knowledge because it prompts a deeper level of

cognitive processing by requiring learners to think through pragmatic rules

while verbalizing their thoughts Collaborative dialogue enables learners to

negotiate and co-construct pragmatic knowledge learners can discuss prag-

matic forms and contextual features associated with them and develop a

joint understanding of the principles underlying the associations

Several pragmatics studies have used collaborative dialogue as a treatment

method in a format of metapragmatic discussion (eg Nguyen 2013)

However these studies used it in combination with other activities and thus

failed to reveal the sole effect of metapragmatic discussion To our knowledge

Takimoto (2012) is the only study that directly tested the efficacy of metaprag-

matic discussion Takimoto compared two conditions in teaching English re-

quest downgraders consciousness-raising instruction with and without

metapragmatic discussion Both treatment groups analyzed request-making

expressions with contextual features and rated the appropriateness of the

expressions Then they generated a list of ways to make the requests more

appropriate One group made a list collaboratively while the other group did

so individually The metapragmatic discussion group outperformed their indi-

vidual task counterpart on the production of requests which suggests the

benefit of collaborative dialogue in developing pragmatic knowledge

Given the paucity of findings on the role of collaborative dialogue in prag-

matics it is pertinent to examine to what extent learners negotiate and

N TAGUCHI AND Y KIM 3

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

co-construct knowledge of sociopragmatic factors and pragmalinguistic fea-

tures during collaborative tasks Furthermore the question of how collabora-

tive dialogue during interactive tasks promotes such processes and subsequent

learning of pragmatics compared with during individual tasks warrants further

investigation Theoretical interest in collaborative dialogue is about language

being used as a cognitive tool that mediates the process of thinking (Swain

2006 Swain and Watanabe 2013) Languaging contends that verbalization of

target pragmatic forms when guided strategically through a collaborative task

could lead to a deeper-level understanding of the features Although

Takimotorsquos study tapped into this theoretical underpinning of collaborative

dialogue and revealed the benefits of learnerndashlearner interaction in pragmatic

learning actual mechanisms of learning during interaction were not discern-

ible because he did not analyze interaction data for evidence of learning oppor-

tunities He analyzed only the posttest results (appropriateness judgment and

production of request downgraders) To gain a fuller account of the theoretical

framework of collaborative dialogue it is necessary to analyze the quality of

LREs targeting pragmatics which we call pragmatic-related episodes (PREs) and

link the analyses to learning outcomes Furthermore in order to compare the

amount of languaging between individual tasks and collaborative tasks it is

also important to document how learners performing individual tasks con-

struct their knowledge of pragmatics on their own Thus in the current

study we collected learnersrsquo think-aloud protocols in the individual task

group and they were analyzed for PREs as well

In this study we examine the effects of collaborative dialogue on the acqui-

sition of speech act of request in English by comparing three groups the lsquocol-

laborative grouprsquo who completes the dialogue construction tasks targeting a

request in pairs the lsquoindividual grouprsquo who completes the task alone and the

control group who does not receive any task-based instruction In addition we

investigate the characteristics of collaborative dialogue (operationalized as

PREs) as language resources that facilitate learning Following Swain and

Lapkin (1998) we define PREs as any part of language production where

learners talk about the pragmalinguistic forms they are producing and the

sociopragmatic factors they are attending to (eg setting and interlocutor re-

lationship) question their pragmatic language use or correct themselves or

others In this study we specifically focus on the frequency of PREs produced

during collaborative and individual tasks and instances of correctly solved

pragmatic problems to see whether PREs provide opportunities for learning

pragmatics This study asks two research questions

1 What is the effect of task-based pragmatic instruction on L2 pragmatic

development Are there any differences in the learning of the request

speech act between the collaborative and individual task group

2 Are there any differences in the frequency of PREs and quality of task

performance during instruction between the collaborative and individual

task group

4 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

METHODS

Participants

Participants were 74 learners of English in a second-grade girlsrsquo junior high

school in South Korea Their average age was 1374 years (range = 13ndash14

years) They had received 5 years of formal English education at the time of

the study They were receiving 4 hours of English instruction per week taught

mostly by Korean-speaking instructors Participantsrsquo proficiency measured by

Test of English for International Communication (TOEIC) Bridge was between

high beginner and intermediate with a mean score of 13527 (SD = 1442

range = 100ndash172) Participantsrsquo English classes focused on grammar vocabu-

lary and readingwriting with little focus on pragmatics Three intact classes

were randomly assigned to three groups a collaborative (n = 25) individual

(n = 25) and control group (n = 24) Analysis of variance showed no significant

group difference in the TOEIC Bridge scores suggesting that the groups were

comparable in terms of their proficiency level (F = 355 p = 70)

Instructional targets

The instructional target was the speech act of request in a formal situation

which appeared in the participantsrsquo English textbook (Lee et al 2009) Formal

request was operationalized using Brown and Levinsonrsquos (1987) contextual

factors power (P) distance (D) and degree of imposition (R) and was defined

as a request that carries a larger size of imposition and is made to someone in a

greater power and distance (PDR-high) An example is asking a professor to

reschedule a test This PDR-high request contrasts with a request made to

someone in equal power and small distance and degree of imposition (PDR-

low) such as asking your sister to pass you a TV remote

In order to identify appropriate request situations a pilot study was con-

ducted with students of the same grade in the institution who were not

included in the main study (n = 34) This involved asking them to indicate

the degree of psychological ease or difficulty in performing the request on

Likert scale from 1 (easy) to 5 (difficult) The survey included 14 PDR-high

and -low requests adapted from the previous literature (eg Taguchi 2012)

The survey also asked participants to indicate the degree of commonality of

each situation on a scale from 1 (very rare) to 5 (very common) Finally the

survey provided an open question in which students wrote PDR-high and -low

requests that they had personally encountered

Results showed that PDR-low situations received a mean difficulty rating of

178 (SD = 106) while PDR-high situations received a mean rating of 264

(SD = 124) This difference was statistically significant t (237) =900

p = 0001 indicating that PDR-high were perceived as more difficult to perform

than PDR-low requests From the 14 situations we selected four PDR-high

items that received a rating of 25 or above and four PDR-low items with a

rating of below 15 In order to ensure the authenticity of tasks commonality

N TAGUCHI AND Y KIM 5

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

ratings of these items were checked to avoid teaching situations that do not

represent real life in Korea Finally we analyzed participantsrsquo responses to the

open-ended question and used their descriptions when writing situations

Instruction focused on two categories of pragmalinguistic forms request

head acts and modifications (Table 1) Request head acts refer to the minimal

core unit that conveys the illocutionary force of request (Blum-Kulka et al

1989) In this category we taught mitigated preparatory forms Mitigated pre-

paratory involves syntactic forms that make reference to the hearerrsquos ability or

will in an embedded question or bi-clausal structure (eg lsquoI was wondering

ifrsquo + verb and lsquoWould it be possible torsquo + verb) (Takahashi 1996) Request modi-

fications have two categories external and internal External modifications

involve semantic moves that are attached externally to the head act and miti-

gate its force (Blum-Kulka et al 1989) We taught two external modifications

preparators (preparing a hearer for a request) and grounders (giving a reason

for request) Internal modifications include sentence-internal syntactic and

lexical devices used to modify the force of request We taught two forms

hedging that softens the tone of speech and amplifiers that strengthen self-

expression

These pragmalinguistic forms were selected based on baseline data collected

from native English speakers Twenty-four students in a US university

Table 1 Target pragmalinguistic forms

Head acts

Mitigated preparatory forms

Reference to the hearerrsquos ability will and possibility or reference to thespeakerrsquos wish in an embedded question or sentence

Irsquom wondering if you could give me an extension on the assignment

Is there any way that I could get an extension on the assignment

External modifications

Preparators

Semantic moves used to prepare the hearer for the request

eg May I ask you something

Grounders

Reason or explanation used to support the request

eg I caught a cold so I couldnrsquot finish the assignment

Internal modifications

Hedging

Words that minimize self-expression (maybe possibly)

eg Irsquom wondering if I could possibly have an extension on the assignment

Amplifiers

Words that strengthen self-expression (really very)

eg I really need more time to work on the assignment

6 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

completed a discourse completion task (DCT) that contained four PDR-high

and -low requests Mitigated preparatory appeared 73 percent of the time as a

head act in PDR-high requests while it appeared only 4 percent of the time in

PDR-low requests This indicates that mitigated preparatory characterizes PDR-

high requests Grounders and preparators were common supportive moves

appearing 98 and 13 percent of the time respectively while they never ap-

peared in PDR-low situations Hedging and amplifiers were also frequent ap-

pearing at a rate of 52 percent while they were absent in PDR-low requests

Based on these findings along with studies that revealed learnersrsquo difficulty

with these forms (Taguchi 2012) we targeted these forms Although DCT data

can only tell us what people would say in a hypothetical situation the rela-

tively uniform responses were considered to represent normative patterns of

request making

Instructional materials and tasks

Treatment groups received instruction in school during their regular English

class sessions over two consecutive days The instruction started with a 5-min

explanation of target pragmatic forms using a written dialogue There were

two dialogues one featuring a PDR-high and the other featuring a PDR-low

request We introduced pragmalinguistic forms and sociopragmatic variables in

each dialogue After receiving the direct metapragmatic information the two

treatment groups (collaborative and individual) proceeded to the dialogue con-

struction task during which learners were asked to complete drama scripts

based on given scenarios as drama scriptwriters They received two scenarios

(PDR-high and -low) with pictures of main characters and created a dialogue

involving a request based on each scenario1 These tasks were considered au-

thentic and relevant to learner interests because TV dramas are popular among

the target population and all the task scenarios were directly related to their

school life2 Learners constructed three PDR-high and -low situation dialogues

during treatment sessions (see Supplementary material for sample instruc-

tional materials) Although PDR-high requests were the instructional target

we introduced PDR-high and -low requests together so the differences be-

tween the two become salient

The collaborative group created a dialogue in pairs while the individual

group completed the task alone Learners vocalized their thoughts during

task performance in the language of their choice either collaboratively or

individually depending on their treatment condition To ensure learnersrsquo

understanding of the task following Kim (2013) we showed a 2-min pre-

task modeling video prior to the beginning of their task performance In the

modeling video for the collaborative group two teachers demonstrate how to

perform a task collaboratively using a similar scenario whereas in the video for

the individual group one teacher performed the same task while thinking

aloud on her own Interaction and think-aloud protocols during task comple-

tion were audio-recorded using individual MP3 recorders The treatment

N TAGUCHI AND Y KIM 7

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

session took about 45 min We repeated the same procedures in the second

task treatment session with different scenarios The control group received

regular English instruction

Assessment measure

A written DCT instrument was used to measure learning outcomes DCT

has been criticized because of a lack of authenticity and non-interactive

nature (Golato 2003) However DCT has a merit because the data can provide

information about learnersrsquo knowledge of normative conventions of prag-

matic language use (McNamara and Roever 2006) DCT can also control

social factors in scenarios and help us obtain data that are comparable across

learners over time Most importantly DCT in this study conformed to the

principle of lsquotransfer appropriate processingrsquo (DeKeyser 2007) which claims

that transfer of skill from a learned to a novel task occurs if the cognitive

operations involved in the novel task resemble those in the treatment task

The treatment task in this study was a dialogue construction task which

shared the same modality with DCT which elicited participantsrsquo pragmalinguis-

tic forms in writing

The DCT had 15 items 4 each of PDR-high and PDR-low requests and 7

filler items involving non-target speech acts Each DCT item had a situation

written in Korean to ensure participantsrsquo comprehension Participants were

asked to follow the first turn or prompt provided in English and write the

speech act in English In order to minimize the practice effect coming from

administering the same test repeatedly we prepared three versions of DCT

Scenarios were kept constant but minor wording changes were made in the

scenarios (eg changing names and locations) Different filler items were used

each time DCT items were different from the items used in the treatment

sessions Supplementary material contains sample DCT items

Procedure

The study was conducted over 6 weeks (see Table 2) During the first ses-

sion the learners took the pretest and two task groups carried out a practice

task to become familiar with recording devices and think-aloud

procedures Task treatment sessions were provided on Days 7 and 9 On

Days 10 and 14 all learners took the immediate posttest and the TOEFIC

Bridge respectively Four weeks after the immediate posttest the learners

took the delayed posttest

Data analysis

This study asked (i) whether task-based instruction is beneficial for learning

request-making expressions and if so whether there is a difference in learning

outcomes between the collaborative and individual group and (ii) to what

extent learners produce PREs during individual or collaborative tasks In

8 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

response to the first research question we first assigned scores on the request

head acts in the DCTs Because the purpose of this study was to compare

instructional effects between two task conditions (collaborative vs individual)

we focused our analysis on the forms we targeted during instruction Three

points were given if the head act contained one of the target forms (lsquoIrsquom

wondering ifrsquo + clause or lsquoIs there any wayrsquo + clause) and was also grammat-

ically accurate Two points were given if the head act took the target form but

was ungrammatical One point was given for a grammatical non-target form in

the context of this study (request in a form other than the mitigated prepara-

tory such as lsquoCan Irsquo) No point was given for an ungrammatical non-target

form and missing response3 Total scores from four PDR-high requests (scale

0ndash12) were compared across collaborative individual and control groups

using the KruskalndashWallis test In addition to head acts frequency of request

modifications (preparators grounders hedging and amplifiers) was counted

separately and compared across groups using the KruskalndashWallis test Both

authors coded 50 of the data yielding 953 agreement rate The discre-

pancies were solved through a follow-up discussion The non-parametric test

(KruskalndashWallis) was used because the data did not confirm normal

distribution

To answer the second research question task interaction and think-aloud

data were transcribed and analyzed for PREs defined as any part of language

production where learners talk about pragmalinguistics (request-making

forms) and sociopragmatics (contextual factors) they are attending question

their pragmatic language use or correct themselves or others PREs were coded

for context (eg setting interlocutor relationship and size of imposition) head

Table 2 Study procedures

Session Individual group Collaborative group Control group

Session 1(Day 1)

Pretest (DCT 1)Practice taskwith thinking aloud

Pretest (DCT 1)Practice task

Pretest (DCT 1)

Session 2(Day 7)

Explicit information(5 min)

Task modeling video(2 min)

Individual dialogueconstruction (35 min)

Explicit information (5 min)

Task modeling video (2 min)

Collaborative dialogueconstruction (35 min)

Reading activity

Session 3(Day 9)

Same procedure withdifferent scenarios

Same procedure with differentscenarios

Reading activity

Session 4(Day 10)

Immediate posttest (DCT 2) Immediate posttest (DCT 2) Immediateposttest (DCT 2)

Session 5(Day 14)

TOEIC Bridge TOEIC Bridge TOEIC Bridge

Session 6(Day 38)

Delayed posttest (DCT 3) Delayed posttest (DCT 3) Delayedposttest (DCT 3)

N TAGUCHI AND Y KIM 9

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

acts grounders preparators hedging and amplifiers See the following ex-

amples about asking the school principal to change a location of a picnic

Example 1 PRE targeting a request head act during interactive task

1 Learner 1 (making a request) I am wondering if I could pos-

sibly (First we have to make a request) I am won-

dering if you could possibly I was wondering

2 Learner 2 I am

3 Learner 1 I am wondering

4 Learner 2 if you could was could (Since

we used lsquocouldrsquo shouldnrsquot we use lsquowasrsquo) I was wondering if I could go

to picnic in Everland

5 Learner 1 if you could (No

since this person needs to make a decision lsquoif you couldrsquo sounds right) If

you could

Example 2 PRE targeting the request head act during think-aloud

Learner 1 In this fall fall picnic fall picnic picnic is there any is there

any way that I could that I could could is there can could possibly

possibly go to Everland participate possibly go to go to Everland

Once the PREs were coded studentsrsquo task performance data (the completed

written dialogues) was scored following the same rubrics as the DCT data and

the frequency of each modification in the dialogues was also counted The

second rater coded 20 percent of task performance data independently and

94 percent agreement was obtained Any disagreement was resolved through

discussion which was applied for the rest of the data coding

RESULTS

Effects of collaborative dialogue in learning the speech act ofrequest

Analysis of request head act

Table 3 displays descriptive statistics of head act scores Frequency analysis

showed that 295 out of the 296 head acts (74 learners 4 items) were non-

target (not the focus of instruction) Most of the non-target head acts took a

form of permission (lsquoMay Irsquo) or ability inquiry (lsquoCould yoursquo) indicating that all

groups were unfamiliar with the mitigated preparatory forms before the in-

struction The KruskalndashWallis test revealed no group difference at pretest

2 = 344 p = 18

The KruskalndashWallis test found a significant group difference at immediate

posttest 2 = 3190 plt 001 Pair-wise comparisons using the MannndashWhitney

U test showed that both treatment groups outperformed the control group

10 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

z = 536 plt 001 (collaborative group) and z = 405 plt 001 (individual group)

The collaborative group further surpassed the individual group on the produc-

tion of mitigated preparatory z = 240 p = 016 However the benefit of this

condition disappeared at delayed posttest because there was no difference be-

tween the treatment groups z = 38 p = 71 In addition the KruskalndashWallis test

showed that there was no significant difference between the control and treat-

ment groups at delayed posttest 2 = 25 p = 884

In summary task-based instruction resulted in a strong effect on the learn-

ing of appropriate request head acts as found in the treatment groupsrsquo superior

performance at immediate posttest compared with that of non-instructional

condition However the effect did not last long as the treatment groups went

back to the level of the control group at delayed posttest Similarly the col-

laborative task was more effective than the individual task at the immediate

posttest but the effect disappeared 1 month later

Analysis of request modification

Tables 4ndash7 present frequency counts of request modifications that appeared in

DCT Preparators almost never appeared at pretest but they showed a large

increase in the two treatment groups at immediate posttest (Table 4) In con-

trast preparators were absent in the control group Both treatment groups

outperformed the control group at immediate and delayed posttest No differ-

ence was found between the collaborative and individual groups z = 176

Table 3 Request head act scores

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 122 200 400

Immediate posttest 1012 306 200 1200

Delayed posttest 432 236 200 1100

Individual group (n = 25)

Pretest 340 071 200 400

Immediate posttest 836 341 300 1200

Delayed posttest 488 273 100 1200

Control group (n = 24)

Pretest 375 043 300 400

Immediate posttest 383 047 200 400

Delayed posttest 388 044 200 400

Note Each student produced four PDR-high requests Each head act was scored on a 3-point

scale Score range = 0ndash12 Delayed posttest was given 1 month later

N TAGUCHI AND Y KIM 11

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

investigated to what extent collaborative dialogue during interactive tasks pro-

motes language learning compared with individual task conditions revealing

advantages of the former For instance in Kimrsquos (2008) study of L2 Korean

individual grouprsquos think-aloud data and collaborative grouprsquos interaction data

were analyzed for LREs and their vocabulary learning was compared Results

showed that although the number of LREs was similar between the groups

the latter performed significantly better on vocabulary tests Nassaji and Tian

(2010) also compared collaborative and individual task conditions on phrasal

verbs learning when learners carried out two tasks (editing tasks and cloze

tasks) in a classroom Collaborative tasks resulted in a greater accuracy of

task completion than individual tasks but they did not lead to greater vocabu-

lary gain The authors suggested the task effect editing tasks were more ef-

fective than cloze tasks in promoting negotiation and learning As shown

above the benefit of collaborative dialogue and LREs stemming from task-

based interaction is noteworthy In order to confirm the effectiveness of col-

laborative dialogue during interactive tasks compared with learnersrsquo attention

to language forms during individual tasks more studies are needed in other

language areas such as pragmatics

In pragmatics learning this article considers that collaborative dialogues

(ie LREs) go beyond just language forms They generate moments in which

learners talk about the form and its relevance to the function and context of

language use During interaction pragmalinguistic forms and contextual fac-

tors are constantly emphasized negotiated and recycled for use This process

helps consolidate pragmatic knowledge because it prompts a deeper level of

cognitive processing by requiring learners to think through pragmatic rules

while verbalizing their thoughts Collaborative dialogue enables learners to

negotiate and co-construct pragmatic knowledge learners can discuss prag-

matic forms and contextual features associated with them and develop a

joint understanding of the principles underlying the associations

Several pragmatics studies have used collaborative dialogue as a treatment

method in a format of metapragmatic discussion (eg Nguyen 2013)

However these studies used it in combination with other activities and thus

failed to reveal the sole effect of metapragmatic discussion To our knowledge

Takimoto (2012) is the only study that directly tested the efficacy of metaprag-

matic discussion Takimoto compared two conditions in teaching English re-

quest downgraders consciousness-raising instruction with and without

metapragmatic discussion Both treatment groups analyzed request-making

expressions with contextual features and rated the appropriateness of the

expressions Then they generated a list of ways to make the requests more

appropriate One group made a list collaboratively while the other group did

so individually The metapragmatic discussion group outperformed their indi-

vidual task counterpart on the production of requests which suggests the

benefit of collaborative dialogue in developing pragmatic knowledge

Given the paucity of findings on the role of collaborative dialogue in prag-

matics it is pertinent to examine to what extent learners negotiate and

N TAGUCHI AND Y KIM 3

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

co-construct knowledge of sociopragmatic factors and pragmalinguistic fea-

tures during collaborative tasks Furthermore the question of how collabora-

tive dialogue during interactive tasks promotes such processes and subsequent

learning of pragmatics compared with during individual tasks warrants further

investigation Theoretical interest in collaborative dialogue is about language

being used as a cognitive tool that mediates the process of thinking (Swain

2006 Swain and Watanabe 2013) Languaging contends that verbalization of

target pragmatic forms when guided strategically through a collaborative task

could lead to a deeper-level understanding of the features Although

Takimotorsquos study tapped into this theoretical underpinning of collaborative

dialogue and revealed the benefits of learnerndashlearner interaction in pragmatic

learning actual mechanisms of learning during interaction were not discern-

ible because he did not analyze interaction data for evidence of learning oppor-

tunities He analyzed only the posttest results (appropriateness judgment and

production of request downgraders) To gain a fuller account of the theoretical

framework of collaborative dialogue it is necessary to analyze the quality of

LREs targeting pragmatics which we call pragmatic-related episodes (PREs) and

link the analyses to learning outcomes Furthermore in order to compare the

amount of languaging between individual tasks and collaborative tasks it is

also important to document how learners performing individual tasks con-

struct their knowledge of pragmatics on their own Thus in the current

study we collected learnersrsquo think-aloud protocols in the individual task

group and they were analyzed for PREs as well

In this study we examine the effects of collaborative dialogue on the acqui-

sition of speech act of request in English by comparing three groups the lsquocol-

laborative grouprsquo who completes the dialogue construction tasks targeting a

request in pairs the lsquoindividual grouprsquo who completes the task alone and the

control group who does not receive any task-based instruction In addition we

investigate the characteristics of collaborative dialogue (operationalized as

PREs) as language resources that facilitate learning Following Swain and

Lapkin (1998) we define PREs as any part of language production where

learners talk about the pragmalinguistic forms they are producing and the

sociopragmatic factors they are attending to (eg setting and interlocutor re-

lationship) question their pragmatic language use or correct themselves or

others In this study we specifically focus on the frequency of PREs produced

during collaborative and individual tasks and instances of correctly solved

pragmatic problems to see whether PREs provide opportunities for learning

pragmatics This study asks two research questions

1 What is the effect of task-based pragmatic instruction on L2 pragmatic

development Are there any differences in the learning of the request

speech act between the collaborative and individual task group

2 Are there any differences in the frequency of PREs and quality of task

performance during instruction between the collaborative and individual

task group

4 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

METHODS

Participants

Participants were 74 learners of English in a second-grade girlsrsquo junior high

school in South Korea Their average age was 1374 years (range = 13ndash14

years) They had received 5 years of formal English education at the time of

the study They were receiving 4 hours of English instruction per week taught

mostly by Korean-speaking instructors Participantsrsquo proficiency measured by

Test of English for International Communication (TOEIC) Bridge was between

high beginner and intermediate with a mean score of 13527 (SD = 1442

range = 100ndash172) Participantsrsquo English classes focused on grammar vocabu-

lary and readingwriting with little focus on pragmatics Three intact classes

were randomly assigned to three groups a collaborative (n = 25) individual

(n = 25) and control group (n = 24) Analysis of variance showed no significant

group difference in the TOEIC Bridge scores suggesting that the groups were

comparable in terms of their proficiency level (F = 355 p = 70)

Instructional targets

The instructional target was the speech act of request in a formal situation

which appeared in the participantsrsquo English textbook (Lee et al 2009) Formal

request was operationalized using Brown and Levinsonrsquos (1987) contextual

factors power (P) distance (D) and degree of imposition (R) and was defined

as a request that carries a larger size of imposition and is made to someone in a

greater power and distance (PDR-high) An example is asking a professor to

reschedule a test This PDR-high request contrasts with a request made to

someone in equal power and small distance and degree of imposition (PDR-

low) such as asking your sister to pass you a TV remote

In order to identify appropriate request situations a pilot study was con-

ducted with students of the same grade in the institution who were not

included in the main study (n = 34) This involved asking them to indicate

the degree of psychological ease or difficulty in performing the request on

Likert scale from 1 (easy) to 5 (difficult) The survey included 14 PDR-high

and -low requests adapted from the previous literature (eg Taguchi 2012)

The survey also asked participants to indicate the degree of commonality of

each situation on a scale from 1 (very rare) to 5 (very common) Finally the

survey provided an open question in which students wrote PDR-high and -low

requests that they had personally encountered

Results showed that PDR-low situations received a mean difficulty rating of

178 (SD = 106) while PDR-high situations received a mean rating of 264

(SD = 124) This difference was statistically significant t (237) =900

p = 0001 indicating that PDR-high were perceived as more difficult to perform

than PDR-low requests From the 14 situations we selected four PDR-high

items that received a rating of 25 or above and four PDR-low items with a

rating of below 15 In order to ensure the authenticity of tasks commonality

N TAGUCHI AND Y KIM 5

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

ratings of these items were checked to avoid teaching situations that do not

represent real life in Korea Finally we analyzed participantsrsquo responses to the

open-ended question and used their descriptions when writing situations

Instruction focused on two categories of pragmalinguistic forms request

head acts and modifications (Table 1) Request head acts refer to the minimal

core unit that conveys the illocutionary force of request (Blum-Kulka et al

1989) In this category we taught mitigated preparatory forms Mitigated pre-

paratory involves syntactic forms that make reference to the hearerrsquos ability or

will in an embedded question or bi-clausal structure (eg lsquoI was wondering

ifrsquo + verb and lsquoWould it be possible torsquo + verb) (Takahashi 1996) Request modi-

fications have two categories external and internal External modifications

involve semantic moves that are attached externally to the head act and miti-

gate its force (Blum-Kulka et al 1989) We taught two external modifications

preparators (preparing a hearer for a request) and grounders (giving a reason

for request) Internal modifications include sentence-internal syntactic and

lexical devices used to modify the force of request We taught two forms

hedging that softens the tone of speech and amplifiers that strengthen self-

expression

These pragmalinguistic forms were selected based on baseline data collected

from native English speakers Twenty-four students in a US university

Table 1 Target pragmalinguistic forms

Head acts

Mitigated preparatory forms

Reference to the hearerrsquos ability will and possibility or reference to thespeakerrsquos wish in an embedded question or sentence

Irsquom wondering if you could give me an extension on the assignment

Is there any way that I could get an extension on the assignment

External modifications

Preparators

Semantic moves used to prepare the hearer for the request

eg May I ask you something

Grounders

Reason or explanation used to support the request

eg I caught a cold so I couldnrsquot finish the assignment

Internal modifications

Hedging

Words that minimize self-expression (maybe possibly)

eg Irsquom wondering if I could possibly have an extension on the assignment

Amplifiers

Words that strengthen self-expression (really very)

eg I really need more time to work on the assignment

6 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

completed a discourse completion task (DCT) that contained four PDR-high

and -low requests Mitigated preparatory appeared 73 percent of the time as a

head act in PDR-high requests while it appeared only 4 percent of the time in

PDR-low requests This indicates that mitigated preparatory characterizes PDR-

high requests Grounders and preparators were common supportive moves

appearing 98 and 13 percent of the time respectively while they never ap-

peared in PDR-low situations Hedging and amplifiers were also frequent ap-

pearing at a rate of 52 percent while they were absent in PDR-low requests

Based on these findings along with studies that revealed learnersrsquo difficulty

with these forms (Taguchi 2012) we targeted these forms Although DCT data

can only tell us what people would say in a hypothetical situation the rela-

tively uniform responses were considered to represent normative patterns of

request making

Instructional materials and tasks

Treatment groups received instruction in school during their regular English

class sessions over two consecutive days The instruction started with a 5-min

explanation of target pragmatic forms using a written dialogue There were

two dialogues one featuring a PDR-high and the other featuring a PDR-low

request We introduced pragmalinguistic forms and sociopragmatic variables in

each dialogue After receiving the direct metapragmatic information the two

treatment groups (collaborative and individual) proceeded to the dialogue con-

struction task during which learners were asked to complete drama scripts

based on given scenarios as drama scriptwriters They received two scenarios

(PDR-high and -low) with pictures of main characters and created a dialogue

involving a request based on each scenario1 These tasks were considered au-

thentic and relevant to learner interests because TV dramas are popular among

the target population and all the task scenarios were directly related to their

school life2 Learners constructed three PDR-high and -low situation dialogues

during treatment sessions (see Supplementary material for sample instruc-

tional materials) Although PDR-high requests were the instructional target

we introduced PDR-high and -low requests together so the differences be-

tween the two become salient

The collaborative group created a dialogue in pairs while the individual

group completed the task alone Learners vocalized their thoughts during

task performance in the language of their choice either collaboratively or

individually depending on their treatment condition To ensure learnersrsquo

understanding of the task following Kim (2013) we showed a 2-min pre-

task modeling video prior to the beginning of their task performance In the

modeling video for the collaborative group two teachers demonstrate how to

perform a task collaboratively using a similar scenario whereas in the video for

the individual group one teacher performed the same task while thinking

aloud on her own Interaction and think-aloud protocols during task comple-

tion were audio-recorded using individual MP3 recorders The treatment

N TAGUCHI AND Y KIM 7

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

session took about 45 min We repeated the same procedures in the second

task treatment session with different scenarios The control group received

regular English instruction

Assessment measure

A written DCT instrument was used to measure learning outcomes DCT

has been criticized because of a lack of authenticity and non-interactive

nature (Golato 2003) However DCT has a merit because the data can provide

information about learnersrsquo knowledge of normative conventions of prag-

matic language use (McNamara and Roever 2006) DCT can also control

social factors in scenarios and help us obtain data that are comparable across

learners over time Most importantly DCT in this study conformed to the

principle of lsquotransfer appropriate processingrsquo (DeKeyser 2007) which claims

that transfer of skill from a learned to a novel task occurs if the cognitive

operations involved in the novel task resemble those in the treatment task

The treatment task in this study was a dialogue construction task which

shared the same modality with DCT which elicited participantsrsquo pragmalinguis-

tic forms in writing

The DCT had 15 items 4 each of PDR-high and PDR-low requests and 7

filler items involving non-target speech acts Each DCT item had a situation

written in Korean to ensure participantsrsquo comprehension Participants were

asked to follow the first turn or prompt provided in English and write the

speech act in English In order to minimize the practice effect coming from

administering the same test repeatedly we prepared three versions of DCT

Scenarios were kept constant but minor wording changes were made in the

scenarios (eg changing names and locations) Different filler items were used

each time DCT items were different from the items used in the treatment

sessions Supplementary material contains sample DCT items

Procedure

The study was conducted over 6 weeks (see Table 2) During the first ses-

sion the learners took the pretest and two task groups carried out a practice

task to become familiar with recording devices and think-aloud

procedures Task treatment sessions were provided on Days 7 and 9 On

Days 10 and 14 all learners took the immediate posttest and the TOEFIC

Bridge respectively Four weeks after the immediate posttest the learners

took the delayed posttest

Data analysis

This study asked (i) whether task-based instruction is beneficial for learning

request-making expressions and if so whether there is a difference in learning

outcomes between the collaborative and individual group and (ii) to what

extent learners produce PREs during individual or collaborative tasks In

8 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

response to the first research question we first assigned scores on the request

head acts in the DCTs Because the purpose of this study was to compare

instructional effects between two task conditions (collaborative vs individual)

we focused our analysis on the forms we targeted during instruction Three

points were given if the head act contained one of the target forms (lsquoIrsquom

wondering ifrsquo + clause or lsquoIs there any wayrsquo + clause) and was also grammat-

ically accurate Two points were given if the head act took the target form but

was ungrammatical One point was given for a grammatical non-target form in

the context of this study (request in a form other than the mitigated prepara-

tory such as lsquoCan Irsquo) No point was given for an ungrammatical non-target

form and missing response3 Total scores from four PDR-high requests (scale

0ndash12) were compared across collaborative individual and control groups

using the KruskalndashWallis test In addition to head acts frequency of request

modifications (preparators grounders hedging and amplifiers) was counted

separately and compared across groups using the KruskalndashWallis test Both

authors coded 50 of the data yielding 953 agreement rate The discre-

pancies were solved through a follow-up discussion The non-parametric test

(KruskalndashWallis) was used because the data did not confirm normal

distribution

To answer the second research question task interaction and think-aloud

data were transcribed and analyzed for PREs defined as any part of language

production where learners talk about pragmalinguistics (request-making

forms) and sociopragmatics (contextual factors) they are attending question

their pragmatic language use or correct themselves or others PREs were coded

for context (eg setting interlocutor relationship and size of imposition) head

Table 2 Study procedures

Session Individual group Collaborative group Control group

Session 1(Day 1)

Pretest (DCT 1)Practice taskwith thinking aloud

Pretest (DCT 1)Practice task

Pretest (DCT 1)

Session 2(Day 7)

Explicit information(5 min)

Task modeling video(2 min)

Individual dialogueconstruction (35 min)

Explicit information (5 min)

Task modeling video (2 min)

Collaborative dialogueconstruction (35 min)

Reading activity

Session 3(Day 9)

Same procedure withdifferent scenarios

Same procedure with differentscenarios

Reading activity

Session 4(Day 10)

Immediate posttest (DCT 2) Immediate posttest (DCT 2) Immediateposttest (DCT 2)

Session 5(Day 14)

TOEIC Bridge TOEIC Bridge TOEIC Bridge

Session 6(Day 38)

Delayed posttest (DCT 3) Delayed posttest (DCT 3) Delayedposttest (DCT 3)

N TAGUCHI AND Y KIM 9

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

acts grounders preparators hedging and amplifiers See the following ex-

amples about asking the school principal to change a location of a picnic

Example 1 PRE targeting a request head act during interactive task

1 Learner 1 (making a request) I am wondering if I could pos-

sibly (First we have to make a request) I am won-

dering if you could possibly I was wondering

2 Learner 2 I am

3 Learner 1 I am wondering

4 Learner 2 if you could was could (Since

we used lsquocouldrsquo shouldnrsquot we use lsquowasrsquo) I was wondering if I could go

to picnic in Everland

5 Learner 1 if you could (No

since this person needs to make a decision lsquoif you couldrsquo sounds right) If

you could

Example 2 PRE targeting the request head act during think-aloud

Learner 1 In this fall fall picnic fall picnic picnic is there any is there

any way that I could that I could could is there can could possibly

possibly go to Everland participate possibly go to go to Everland

Once the PREs were coded studentsrsquo task performance data (the completed

written dialogues) was scored following the same rubrics as the DCT data and

the frequency of each modification in the dialogues was also counted The

second rater coded 20 percent of task performance data independently and

94 percent agreement was obtained Any disagreement was resolved through

discussion which was applied for the rest of the data coding

RESULTS

Effects of collaborative dialogue in learning the speech act ofrequest

Analysis of request head act

Table 3 displays descriptive statistics of head act scores Frequency analysis

showed that 295 out of the 296 head acts (74 learners 4 items) were non-

target (not the focus of instruction) Most of the non-target head acts took a

form of permission (lsquoMay Irsquo) or ability inquiry (lsquoCould yoursquo) indicating that all

groups were unfamiliar with the mitigated preparatory forms before the in-

struction The KruskalndashWallis test revealed no group difference at pretest

2 = 344 p = 18

The KruskalndashWallis test found a significant group difference at immediate

posttest 2 = 3190 plt 001 Pair-wise comparisons using the MannndashWhitney

U test showed that both treatment groups outperformed the control group

10 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

z = 536 plt 001 (collaborative group) and z = 405 plt 001 (individual group)

The collaborative group further surpassed the individual group on the produc-

tion of mitigated preparatory z = 240 p = 016 However the benefit of this

condition disappeared at delayed posttest because there was no difference be-

tween the treatment groups z = 38 p = 71 In addition the KruskalndashWallis test

showed that there was no significant difference between the control and treat-

ment groups at delayed posttest 2 = 25 p = 884

In summary task-based instruction resulted in a strong effect on the learn-

ing of appropriate request head acts as found in the treatment groupsrsquo superior

performance at immediate posttest compared with that of non-instructional

condition However the effect did not last long as the treatment groups went

back to the level of the control group at delayed posttest Similarly the col-

laborative task was more effective than the individual task at the immediate

posttest but the effect disappeared 1 month later

Analysis of request modification

Tables 4ndash7 present frequency counts of request modifications that appeared in

DCT Preparators almost never appeared at pretest but they showed a large

increase in the two treatment groups at immediate posttest (Table 4) In con-

trast preparators were absent in the control group Both treatment groups

outperformed the control group at immediate and delayed posttest No differ-

ence was found between the collaborative and individual groups z = 176

Table 3 Request head act scores

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 122 200 400

Immediate posttest 1012 306 200 1200

Delayed posttest 432 236 200 1100

Individual group (n = 25)

Pretest 340 071 200 400

Immediate posttest 836 341 300 1200

Delayed posttest 488 273 100 1200

Control group (n = 24)

Pretest 375 043 300 400

Immediate posttest 383 047 200 400

Delayed posttest 388 044 200 400

Note Each student produced four PDR-high requests Each head act was scored on a 3-point

scale Score range = 0ndash12 Delayed posttest was given 1 month later

N TAGUCHI AND Y KIM 11

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

co-construct knowledge of sociopragmatic factors and pragmalinguistic fea-

tures during collaborative tasks Furthermore the question of how collabora-

tive dialogue during interactive tasks promotes such processes and subsequent

learning of pragmatics compared with during individual tasks warrants further

investigation Theoretical interest in collaborative dialogue is about language

being used as a cognitive tool that mediates the process of thinking (Swain

2006 Swain and Watanabe 2013) Languaging contends that verbalization of

target pragmatic forms when guided strategically through a collaborative task

could lead to a deeper-level understanding of the features Although

Takimotorsquos study tapped into this theoretical underpinning of collaborative

dialogue and revealed the benefits of learnerndashlearner interaction in pragmatic

learning actual mechanisms of learning during interaction were not discern-

ible because he did not analyze interaction data for evidence of learning oppor-

tunities He analyzed only the posttest results (appropriateness judgment and

production of request downgraders) To gain a fuller account of the theoretical

framework of collaborative dialogue it is necessary to analyze the quality of

LREs targeting pragmatics which we call pragmatic-related episodes (PREs) and

link the analyses to learning outcomes Furthermore in order to compare the

amount of languaging between individual tasks and collaborative tasks it is

also important to document how learners performing individual tasks con-

struct their knowledge of pragmatics on their own Thus in the current

study we collected learnersrsquo think-aloud protocols in the individual task

group and they were analyzed for PREs as well

In this study we examine the effects of collaborative dialogue on the acqui-

sition of speech act of request in English by comparing three groups the lsquocol-

laborative grouprsquo who completes the dialogue construction tasks targeting a

request in pairs the lsquoindividual grouprsquo who completes the task alone and the

control group who does not receive any task-based instruction In addition we

investigate the characteristics of collaborative dialogue (operationalized as

PREs) as language resources that facilitate learning Following Swain and

Lapkin (1998) we define PREs as any part of language production where

learners talk about the pragmalinguistic forms they are producing and the

sociopragmatic factors they are attending to (eg setting and interlocutor re-

lationship) question their pragmatic language use or correct themselves or

others In this study we specifically focus on the frequency of PREs produced

during collaborative and individual tasks and instances of correctly solved

pragmatic problems to see whether PREs provide opportunities for learning

pragmatics This study asks two research questions

1 What is the effect of task-based pragmatic instruction on L2 pragmatic

development Are there any differences in the learning of the request

speech act between the collaborative and individual task group

2 Are there any differences in the frequency of PREs and quality of task

performance during instruction between the collaborative and individual

task group

4 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

METHODS

Participants

Participants were 74 learners of English in a second-grade girlsrsquo junior high

school in South Korea Their average age was 1374 years (range = 13ndash14

years) They had received 5 years of formal English education at the time of

the study They were receiving 4 hours of English instruction per week taught

mostly by Korean-speaking instructors Participantsrsquo proficiency measured by

Test of English for International Communication (TOEIC) Bridge was between

high beginner and intermediate with a mean score of 13527 (SD = 1442

range = 100ndash172) Participantsrsquo English classes focused on grammar vocabu-

lary and readingwriting with little focus on pragmatics Three intact classes

were randomly assigned to three groups a collaborative (n = 25) individual

(n = 25) and control group (n = 24) Analysis of variance showed no significant

group difference in the TOEIC Bridge scores suggesting that the groups were

comparable in terms of their proficiency level (F = 355 p = 70)

Instructional targets

The instructional target was the speech act of request in a formal situation

which appeared in the participantsrsquo English textbook (Lee et al 2009) Formal

request was operationalized using Brown and Levinsonrsquos (1987) contextual

factors power (P) distance (D) and degree of imposition (R) and was defined

as a request that carries a larger size of imposition and is made to someone in a

greater power and distance (PDR-high) An example is asking a professor to

reschedule a test This PDR-high request contrasts with a request made to

someone in equal power and small distance and degree of imposition (PDR-

low) such as asking your sister to pass you a TV remote

In order to identify appropriate request situations a pilot study was con-

ducted with students of the same grade in the institution who were not

included in the main study (n = 34) This involved asking them to indicate

the degree of psychological ease or difficulty in performing the request on

Likert scale from 1 (easy) to 5 (difficult) The survey included 14 PDR-high

and -low requests adapted from the previous literature (eg Taguchi 2012)

The survey also asked participants to indicate the degree of commonality of

each situation on a scale from 1 (very rare) to 5 (very common) Finally the

survey provided an open question in which students wrote PDR-high and -low

requests that they had personally encountered

Results showed that PDR-low situations received a mean difficulty rating of

178 (SD = 106) while PDR-high situations received a mean rating of 264

(SD = 124) This difference was statistically significant t (237) =900

p = 0001 indicating that PDR-high were perceived as more difficult to perform

than PDR-low requests From the 14 situations we selected four PDR-high

items that received a rating of 25 or above and four PDR-low items with a

rating of below 15 In order to ensure the authenticity of tasks commonality

N TAGUCHI AND Y KIM 5

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

ratings of these items were checked to avoid teaching situations that do not

represent real life in Korea Finally we analyzed participantsrsquo responses to the

open-ended question and used their descriptions when writing situations

Instruction focused on two categories of pragmalinguistic forms request

head acts and modifications (Table 1) Request head acts refer to the minimal

core unit that conveys the illocutionary force of request (Blum-Kulka et al

1989) In this category we taught mitigated preparatory forms Mitigated pre-

paratory involves syntactic forms that make reference to the hearerrsquos ability or

will in an embedded question or bi-clausal structure (eg lsquoI was wondering

ifrsquo + verb and lsquoWould it be possible torsquo + verb) (Takahashi 1996) Request modi-

fications have two categories external and internal External modifications

involve semantic moves that are attached externally to the head act and miti-

gate its force (Blum-Kulka et al 1989) We taught two external modifications

preparators (preparing a hearer for a request) and grounders (giving a reason

for request) Internal modifications include sentence-internal syntactic and

lexical devices used to modify the force of request We taught two forms

hedging that softens the tone of speech and amplifiers that strengthen self-

expression

These pragmalinguistic forms were selected based on baseline data collected

from native English speakers Twenty-four students in a US university

Table 1 Target pragmalinguistic forms

Head acts

Mitigated preparatory forms

Reference to the hearerrsquos ability will and possibility or reference to thespeakerrsquos wish in an embedded question or sentence

Irsquom wondering if you could give me an extension on the assignment

Is there any way that I could get an extension on the assignment

External modifications

Preparators

Semantic moves used to prepare the hearer for the request

eg May I ask you something

Grounders

Reason or explanation used to support the request

eg I caught a cold so I couldnrsquot finish the assignment

Internal modifications

Hedging

Words that minimize self-expression (maybe possibly)

eg Irsquom wondering if I could possibly have an extension on the assignment

Amplifiers

Words that strengthen self-expression (really very)

eg I really need more time to work on the assignment

6 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

completed a discourse completion task (DCT) that contained four PDR-high

and -low requests Mitigated preparatory appeared 73 percent of the time as a

head act in PDR-high requests while it appeared only 4 percent of the time in

PDR-low requests This indicates that mitigated preparatory characterizes PDR-

high requests Grounders and preparators were common supportive moves

appearing 98 and 13 percent of the time respectively while they never ap-

peared in PDR-low situations Hedging and amplifiers were also frequent ap-

pearing at a rate of 52 percent while they were absent in PDR-low requests

Based on these findings along with studies that revealed learnersrsquo difficulty

with these forms (Taguchi 2012) we targeted these forms Although DCT data

can only tell us what people would say in a hypothetical situation the rela-

tively uniform responses were considered to represent normative patterns of

request making

Instructional materials and tasks

Treatment groups received instruction in school during their regular English

class sessions over two consecutive days The instruction started with a 5-min

explanation of target pragmatic forms using a written dialogue There were

two dialogues one featuring a PDR-high and the other featuring a PDR-low

request We introduced pragmalinguistic forms and sociopragmatic variables in

each dialogue After receiving the direct metapragmatic information the two

treatment groups (collaborative and individual) proceeded to the dialogue con-

struction task during which learners were asked to complete drama scripts

based on given scenarios as drama scriptwriters They received two scenarios

(PDR-high and -low) with pictures of main characters and created a dialogue

involving a request based on each scenario1 These tasks were considered au-

thentic and relevant to learner interests because TV dramas are popular among

the target population and all the task scenarios were directly related to their

school life2 Learners constructed three PDR-high and -low situation dialogues

during treatment sessions (see Supplementary material for sample instruc-

tional materials) Although PDR-high requests were the instructional target

we introduced PDR-high and -low requests together so the differences be-

tween the two become salient

The collaborative group created a dialogue in pairs while the individual

group completed the task alone Learners vocalized their thoughts during

task performance in the language of their choice either collaboratively or

individually depending on their treatment condition To ensure learnersrsquo

understanding of the task following Kim (2013) we showed a 2-min pre-

task modeling video prior to the beginning of their task performance In the

modeling video for the collaborative group two teachers demonstrate how to

perform a task collaboratively using a similar scenario whereas in the video for

the individual group one teacher performed the same task while thinking

aloud on her own Interaction and think-aloud protocols during task comple-

tion were audio-recorded using individual MP3 recorders The treatment

N TAGUCHI AND Y KIM 7

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

session took about 45 min We repeated the same procedures in the second

task treatment session with different scenarios The control group received

regular English instruction

Assessment measure

A written DCT instrument was used to measure learning outcomes DCT

has been criticized because of a lack of authenticity and non-interactive

nature (Golato 2003) However DCT has a merit because the data can provide

information about learnersrsquo knowledge of normative conventions of prag-

matic language use (McNamara and Roever 2006) DCT can also control

social factors in scenarios and help us obtain data that are comparable across

learners over time Most importantly DCT in this study conformed to the

principle of lsquotransfer appropriate processingrsquo (DeKeyser 2007) which claims

that transfer of skill from a learned to a novel task occurs if the cognitive

operations involved in the novel task resemble those in the treatment task

The treatment task in this study was a dialogue construction task which

shared the same modality with DCT which elicited participantsrsquo pragmalinguis-

tic forms in writing

The DCT had 15 items 4 each of PDR-high and PDR-low requests and 7

filler items involving non-target speech acts Each DCT item had a situation

written in Korean to ensure participantsrsquo comprehension Participants were

asked to follow the first turn or prompt provided in English and write the

speech act in English In order to minimize the practice effect coming from

administering the same test repeatedly we prepared three versions of DCT

Scenarios were kept constant but minor wording changes were made in the

scenarios (eg changing names and locations) Different filler items were used

each time DCT items were different from the items used in the treatment

sessions Supplementary material contains sample DCT items

Procedure

The study was conducted over 6 weeks (see Table 2) During the first ses-

sion the learners took the pretest and two task groups carried out a practice

task to become familiar with recording devices and think-aloud

procedures Task treatment sessions were provided on Days 7 and 9 On

Days 10 and 14 all learners took the immediate posttest and the TOEFIC

Bridge respectively Four weeks after the immediate posttest the learners

took the delayed posttest

Data analysis

This study asked (i) whether task-based instruction is beneficial for learning

request-making expressions and if so whether there is a difference in learning

outcomes between the collaborative and individual group and (ii) to what

extent learners produce PREs during individual or collaborative tasks In

8 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

response to the first research question we first assigned scores on the request

head acts in the DCTs Because the purpose of this study was to compare

instructional effects between two task conditions (collaborative vs individual)

we focused our analysis on the forms we targeted during instruction Three

points were given if the head act contained one of the target forms (lsquoIrsquom

wondering ifrsquo + clause or lsquoIs there any wayrsquo + clause) and was also grammat-

ically accurate Two points were given if the head act took the target form but

was ungrammatical One point was given for a grammatical non-target form in

the context of this study (request in a form other than the mitigated prepara-

tory such as lsquoCan Irsquo) No point was given for an ungrammatical non-target

form and missing response3 Total scores from four PDR-high requests (scale

0ndash12) were compared across collaborative individual and control groups

using the KruskalndashWallis test In addition to head acts frequency of request

modifications (preparators grounders hedging and amplifiers) was counted

separately and compared across groups using the KruskalndashWallis test Both

authors coded 50 of the data yielding 953 agreement rate The discre-

pancies were solved through a follow-up discussion The non-parametric test

(KruskalndashWallis) was used because the data did not confirm normal

distribution

To answer the second research question task interaction and think-aloud

data were transcribed and analyzed for PREs defined as any part of language

production where learners talk about pragmalinguistics (request-making

forms) and sociopragmatics (contextual factors) they are attending question

their pragmatic language use or correct themselves or others PREs were coded

for context (eg setting interlocutor relationship and size of imposition) head

Table 2 Study procedures

Session Individual group Collaborative group Control group

Session 1(Day 1)

Pretest (DCT 1)Practice taskwith thinking aloud

Pretest (DCT 1)Practice task

Pretest (DCT 1)

Session 2(Day 7)

Explicit information(5 min)

Task modeling video(2 min)

Individual dialogueconstruction (35 min)

Explicit information (5 min)

Task modeling video (2 min)

Collaborative dialogueconstruction (35 min)

Reading activity

Session 3(Day 9)

Same procedure withdifferent scenarios

Same procedure with differentscenarios

Reading activity

Session 4(Day 10)

Immediate posttest (DCT 2) Immediate posttest (DCT 2) Immediateposttest (DCT 2)

Session 5(Day 14)

TOEIC Bridge TOEIC Bridge TOEIC Bridge

Session 6(Day 38)

Delayed posttest (DCT 3) Delayed posttest (DCT 3) Delayedposttest (DCT 3)

N TAGUCHI AND Y KIM 9

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

acts grounders preparators hedging and amplifiers See the following ex-

amples about asking the school principal to change a location of a picnic

Example 1 PRE targeting a request head act during interactive task

1 Learner 1 (making a request) I am wondering if I could pos-

sibly (First we have to make a request) I am won-

dering if you could possibly I was wondering

2 Learner 2 I am

3 Learner 1 I am wondering

4 Learner 2 if you could was could (Since

we used lsquocouldrsquo shouldnrsquot we use lsquowasrsquo) I was wondering if I could go

to picnic in Everland

5 Learner 1 if you could (No

since this person needs to make a decision lsquoif you couldrsquo sounds right) If

you could

Example 2 PRE targeting the request head act during think-aloud

Learner 1 In this fall fall picnic fall picnic picnic is there any is there

any way that I could that I could could is there can could possibly

possibly go to Everland participate possibly go to go to Everland

Once the PREs were coded studentsrsquo task performance data (the completed

written dialogues) was scored following the same rubrics as the DCT data and

the frequency of each modification in the dialogues was also counted The

second rater coded 20 percent of task performance data independently and

94 percent agreement was obtained Any disagreement was resolved through

discussion which was applied for the rest of the data coding

RESULTS

Effects of collaborative dialogue in learning the speech act ofrequest

Analysis of request head act

Table 3 displays descriptive statistics of head act scores Frequency analysis

showed that 295 out of the 296 head acts (74 learners 4 items) were non-

target (not the focus of instruction) Most of the non-target head acts took a

form of permission (lsquoMay Irsquo) or ability inquiry (lsquoCould yoursquo) indicating that all

groups were unfamiliar with the mitigated preparatory forms before the in-

struction The KruskalndashWallis test revealed no group difference at pretest

2 = 344 p = 18

The KruskalndashWallis test found a significant group difference at immediate

posttest 2 = 3190 plt 001 Pair-wise comparisons using the MannndashWhitney

U test showed that both treatment groups outperformed the control group

10 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

z = 536 plt 001 (collaborative group) and z = 405 plt 001 (individual group)

The collaborative group further surpassed the individual group on the produc-

tion of mitigated preparatory z = 240 p = 016 However the benefit of this

condition disappeared at delayed posttest because there was no difference be-

tween the treatment groups z = 38 p = 71 In addition the KruskalndashWallis test

showed that there was no significant difference between the control and treat-

ment groups at delayed posttest 2 = 25 p = 884

In summary task-based instruction resulted in a strong effect on the learn-

ing of appropriate request head acts as found in the treatment groupsrsquo superior

performance at immediate posttest compared with that of non-instructional

condition However the effect did not last long as the treatment groups went

back to the level of the control group at delayed posttest Similarly the col-

laborative task was more effective than the individual task at the immediate

posttest but the effect disappeared 1 month later

Analysis of request modification

Tables 4ndash7 present frequency counts of request modifications that appeared in

DCT Preparators almost never appeared at pretest but they showed a large

increase in the two treatment groups at immediate posttest (Table 4) In con-

trast preparators were absent in the control group Both treatment groups

outperformed the control group at immediate and delayed posttest No differ-

ence was found between the collaborative and individual groups z = 176

Table 3 Request head act scores

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 122 200 400

Immediate posttest 1012 306 200 1200

Delayed posttest 432 236 200 1100

Individual group (n = 25)

Pretest 340 071 200 400

Immediate posttest 836 341 300 1200

Delayed posttest 488 273 100 1200

Control group (n = 24)

Pretest 375 043 300 400

Immediate posttest 383 047 200 400

Delayed posttest 388 044 200 400

Note Each student produced four PDR-high requests Each head act was scored on a 3-point

scale Score range = 0ndash12 Delayed posttest was given 1 month later

N TAGUCHI AND Y KIM 11

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

METHODS

Participants

Participants were 74 learners of English in a second-grade girlsrsquo junior high

school in South Korea Their average age was 1374 years (range = 13ndash14

years) They had received 5 years of formal English education at the time of

the study They were receiving 4 hours of English instruction per week taught

mostly by Korean-speaking instructors Participantsrsquo proficiency measured by

Test of English for International Communication (TOEIC) Bridge was between

high beginner and intermediate with a mean score of 13527 (SD = 1442

range = 100ndash172) Participantsrsquo English classes focused on grammar vocabu-

lary and readingwriting with little focus on pragmatics Three intact classes

were randomly assigned to three groups a collaborative (n = 25) individual

(n = 25) and control group (n = 24) Analysis of variance showed no significant

group difference in the TOEIC Bridge scores suggesting that the groups were

comparable in terms of their proficiency level (F = 355 p = 70)

Instructional targets

The instructional target was the speech act of request in a formal situation

which appeared in the participantsrsquo English textbook (Lee et al 2009) Formal

request was operationalized using Brown and Levinsonrsquos (1987) contextual

factors power (P) distance (D) and degree of imposition (R) and was defined

as a request that carries a larger size of imposition and is made to someone in a

greater power and distance (PDR-high) An example is asking a professor to

reschedule a test This PDR-high request contrasts with a request made to

someone in equal power and small distance and degree of imposition (PDR-

low) such as asking your sister to pass you a TV remote

In order to identify appropriate request situations a pilot study was con-

ducted with students of the same grade in the institution who were not

included in the main study (n = 34) This involved asking them to indicate

the degree of psychological ease or difficulty in performing the request on

Likert scale from 1 (easy) to 5 (difficult) The survey included 14 PDR-high

and -low requests adapted from the previous literature (eg Taguchi 2012)

The survey also asked participants to indicate the degree of commonality of

each situation on a scale from 1 (very rare) to 5 (very common) Finally the

survey provided an open question in which students wrote PDR-high and -low

requests that they had personally encountered

Results showed that PDR-low situations received a mean difficulty rating of

178 (SD = 106) while PDR-high situations received a mean rating of 264

(SD = 124) This difference was statistically significant t (237) =900

p = 0001 indicating that PDR-high were perceived as more difficult to perform

than PDR-low requests From the 14 situations we selected four PDR-high

items that received a rating of 25 or above and four PDR-low items with a

rating of below 15 In order to ensure the authenticity of tasks commonality

N TAGUCHI AND Y KIM 5

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

ratings of these items were checked to avoid teaching situations that do not

represent real life in Korea Finally we analyzed participantsrsquo responses to the

open-ended question and used their descriptions when writing situations

Instruction focused on two categories of pragmalinguistic forms request

head acts and modifications (Table 1) Request head acts refer to the minimal

core unit that conveys the illocutionary force of request (Blum-Kulka et al

1989) In this category we taught mitigated preparatory forms Mitigated pre-

paratory involves syntactic forms that make reference to the hearerrsquos ability or

will in an embedded question or bi-clausal structure (eg lsquoI was wondering

ifrsquo + verb and lsquoWould it be possible torsquo + verb) (Takahashi 1996) Request modi-

fications have two categories external and internal External modifications

involve semantic moves that are attached externally to the head act and miti-

gate its force (Blum-Kulka et al 1989) We taught two external modifications

preparators (preparing a hearer for a request) and grounders (giving a reason

for request) Internal modifications include sentence-internal syntactic and

lexical devices used to modify the force of request We taught two forms

hedging that softens the tone of speech and amplifiers that strengthen self-

expression

These pragmalinguistic forms were selected based on baseline data collected

from native English speakers Twenty-four students in a US university

Table 1 Target pragmalinguistic forms

Head acts

Mitigated preparatory forms

Reference to the hearerrsquos ability will and possibility or reference to thespeakerrsquos wish in an embedded question or sentence

Irsquom wondering if you could give me an extension on the assignment

Is there any way that I could get an extension on the assignment

External modifications

Preparators

Semantic moves used to prepare the hearer for the request

eg May I ask you something

Grounders

Reason or explanation used to support the request

eg I caught a cold so I couldnrsquot finish the assignment

Internal modifications

Hedging

Words that minimize self-expression (maybe possibly)

eg Irsquom wondering if I could possibly have an extension on the assignment

Amplifiers

Words that strengthen self-expression (really very)

eg I really need more time to work on the assignment

6 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

completed a discourse completion task (DCT) that contained four PDR-high

and -low requests Mitigated preparatory appeared 73 percent of the time as a

head act in PDR-high requests while it appeared only 4 percent of the time in

PDR-low requests This indicates that mitigated preparatory characterizes PDR-

high requests Grounders and preparators were common supportive moves

appearing 98 and 13 percent of the time respectively while they never ap-

peared in PDR-low situations Hedging and amplifiers were also frequent ap-

pearing at a rate of 52 percent while they were absent in PDR-low requests

Based on these findings along with studies that revealed learnersrsquo difficulty

with these forms (Taguchi 2012) we targeted these forms Although DCT data

can only tell us what people would say in a hypothetical situation the rela-

tively uniform responses were considered to represent normative patterns of

request making

Instructional materials and tasks

Treatment groups received instruction in school during their regular English

class sessions over two consecutive days The instruction started with a 5-min

explanation of target pragmatic forms using a written dialogue There were

two dialogues one featuring a PDR-high and the other featuring a PDR-low

request We introduced pragmalinguistic forms and sociopragmatic variables in

each dialogue After receiving the direct metapragmatic information the two

treatment groups (collaborative and individual) proceeded to the dialogue con-

struction task during which learners were asked to complete drama scripts

based on given scenarios as drama scriptwriters They received two scenarios

(PDR-high and -low) with pictures of main characters and created a dialogue

involving a request based on each scenario1 These tasks were considered au-

thentic and relevant to learner interests because TV dramas are popular among

the target population and all the task scenarios were directly related to their

school life2 Learners constructed three PDR-high and -low situation dialogues

during treatment sessions (see Supplementary material for sample instruc-

tional materials) Although PDR-high requests were the instructional target

we introduced PDR-high and -low requests together so the differences be-

tween the two become salient

The collaborative group created a dialogue in pairs while the individual

group completed the task alone Learners vocalized their thoughts during

task performance in the language of their choice either collaboratively or

individually depending on their treatment condition To ensure learnersrsquo

understanding of the task following Kim (2013) we showed a 2-min pre-

task modeling video prior to the beginning of their task performance In the

modeling video for the collaborative group two teachers demonstrate how to

perform a task collaboratively using a similar scenario whereas in the video for

the individual group one teacher performed the same task while thinking

aloud on her own Interaction and think-aloud protocols during task comple-

tion were audio-recorded using individual MP3 recorders The treatment

N TAGUCHI AND Y KIM 7

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

session took about 45 min We repeated the same procedures in the second

task treatment session with different scenarios The control group received

regular English instruction

Assessment measure

A written DCT instrument was used to measure learning outcomes DCT

has been criticized because of a lack of authenticity and non-interactive

nature (Golato 2003) However DCT has a merit because the data can provide

information about learnersrsquo knowledge of normative conventions of prag-

matic language use (McNamara and Roever 2006) DCT can also control

social factors in scenarios and help us obtain data that are comparable across

learners over time Most importantly DCT in this study conformed to the

principle of lsquotransfer appropriate processingrsquo (DeKeyser 2007) which claims

that transfer of skill from a learned to a novel task occurs if the cognitive

operations involved in the novel task resemble those in the treatment task

The treatment task in this study was a dialogue construction task which

shared the same modality with DCT which elicited participantsrsquo pragmalinguis-

tic forms in writing

The DCT had 15 items 4 each of PDR-high and PDR-low requests and 7

filler items involving non-target speech acts Each DCT item had a situation

written in Korean to ensure participantsrsquo comprehension Participants were

asked to follow the first turn or prompt provided in English and write the

speech act in English In order to minimize the practice effect coming from

administering the same test repeatedly we prepared three versions of DCT

Scenarios were kept constant but minor wording changes were made in the

scenarios (eg changing names and locations) Different filler items were used

each time DCT items were different from the items used in the treatment

sessions Supplementary material contains sample DCT items

Procedure

The study was conducted over 6 weeks (see Table 2) During the first ses-

sion the learners took the pretest and two task groups carried out a practice

task to become familiar with recording devices and think-aloud

procedures Task treatment sessions were provided on Days 7 and 9 On

Days 10 and 14 all learners took the immediate posttest and the TOEFIC

Bridge respectively Four weeks after the immediate posttest the learners

took the delayed posttest

Data analysis

This study asked (i) whether task-based instruction is beneficial for learning

request-making expressions and if so whether there is a difference in learning

outcomes between the collaborative and individual group and (ii) to what

extent learners produce PREs during individual or collaborative tasks In

8 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

response to the first research question we first assigned scores on the request

head acts in the DCTs Because the purpose of this study was to compare

instructional effects between two task conditions (collaborative vs individual)

we focused our analysis on the forms we targeted during instruction Three

points were given if the head act contained one of the target forms (lsquoIrsquom

wondering ifrsquo + clause or lsquoIs there any wayrsquo + clause) and was also grammat-

ically accurate Two points were given if the head act took the target form but

was ungrammatical One point was given for a grammatical non-target form in

the context of this study (request in a form other than the mitigated prepara-

tory such as lsquoCan Irsquo) No point was given for an ungrammatical non-target

form and missing response3 Total scores from four PDR-high requests (scale

0ndash12) were compared across collaborative individual and control groups

using the KruskalndashWallis test In addition to head acts frequency of request

modifications (preparators grounders hedging and amplifiers) was counted

separately and compared across groups using the KruskalndashWallis test Both

authors coded 50 of the data yielding 953 agreement rate The discre-

pancies were solved through a follow-up discussion The non-parametric test

(KruskalndashWallis) was used because the data did not confirm normal

distribution

To answer the second research question task interaction and think-aloud

data were transcribed and analyzed for PREs defined as any part of language

production where learners talk about pragmalinguistics (request-making

forms) and sociopragmatics (contextual factors) they are attending question

their pragmatic language use or correct themselves or others PREs were coded

for context (eg setting interlocutor relationship and size of imposition) head

Table 2 Study procedures

Session Individual group Collaborative group Control group

Session 1(Day 1)

Pretest (DCT 1)Practice taskwith thinking aloud

Pretest (DCT 1)Practice task

Pretest (DCT 1)

Session 2(Day 7)

Explicit information(5 min)

Task modeling video(2 min)

Individual dialogueconstruction (35 min)

Explicit information (5 min)

Task modeling video (2 min)

Collaborative dialogueconstruction (35 min)

Reading activity

Session 3(Day 9)

Same procedure withdifferent scenarios

Same procedure with differentscenarios

Reading activity

Session 4(Day 10)

Immediate posttest (DCT 2) Immediate posttest (DCT 2) Immediateposttest (DCT 2)

Session 5(Day 14)

TOEIC Bridge TOEIC Bridge TOEIC Bridge

Session 6(Day 38)

Delayed posttest (DCT 3) Delayed posttest (DCT 3) Delayedposttest (DCT 3)

N TAGUCHI AND Y KIM 9

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

acts grounders preparators hedging and amplifiers See the following ex-

amples about asking the school principal to change a location of a picnic

Example 1 PRE targeting a request head act during interactive task

1 Learner 1 (making a request) I am wondering if I could pos-

sibly (First we have to make a request) I am won-

dering if you could possibly I was wondering

2 Learner 2 I am

3 Learner 1 I am wondering

4 Learner 2 if you could was could (Since

we used lsquocouldrsquo shouldnrsquot we use lsquowasrsquo) I was wondering if I could go

to picnic in Everland

5 Learner 1 if you could (No

since this person needs to make a decision lsquoif you couldrsquo sounds right) If

you could

Example 2 PRE targeting the request head act during think-aloud

Learner 1 In this fall fall picnic fall picnic picnic is there any is there

any way that I could that I could could is there can could possibly

possibly go to Everland participate possibly go to go to Everland

Once the PREs were coded studentsrsquo task performance data (the completed

written dialogues) was scored following the same rubrics as the DCT data and

the frequency of each modification in the dialogues was also counted The

second rater coded 20 percent of task performance data independently and

94 percent agreement was obtained Any disagreement was resolved through

discussion which was applied for the rest of the data coding

RESULTS

Effects of collaborative dialogue in learning the speech act ofrequest

Analysis of request head act

Table 3 displays descriptive statistics of head act scores Frequency analysis

showed that 295 out of the 296 head acts (74 learners 4 items) were non-

target (not the focus of instruction) Most of the non-target head acts took a

form of permission (lsquoMay Irsquo) or ability inquiry (lsquoCould yoursquo) indicating that all

groups were unfamiliar with the mitigated preparatory forms before the in-

struction The KruskalndashWallis test revealed no group difference at pretest

2 = 344 p = 18

The KruskalndashWallis test found a significant group difference at immediate

posttest 2 = 3190 plt 001 Pair-wise comparisons using the MannndashWhitney

U test showed that both treatment groups outperformed the control group

10 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

z = 536 plt 001 (collaborative group) and z = 405 plt 001 (individual group)

The collaborative group further surpassed the individual group on the produc-

tion of mitigated preparatory z = 240 p = 016 However the benefit of this

condition disappeared at delayed posttest because there was no difference be-

tween the treatment groups z = 38 p = 71 In addition the KruskalndashWallis test

showed that there was no significant difference between the control and treat-

ment groups at delayed posttest 2 = 25 p = 884

In summary task-based instruction resulted in a strong effect on the learn-

ing of appropriate request head acts as found in the treatment groupsrsquo superior

performance at immediate posttest compared with that of non-instructional

condition However the effect did not last long as the treatment groups went

back to the level of the control group at delayed posttest Similarly the col-

laborative task was more effective than the individual task at the immediate

posttest but the effect disappeared 1 month later

Analysis of request modification

Tables 4ndash7 present frequency counts of request modifications that appeared in

DCT Preparators almost never appeared at pretest but they showed a large

increase in the two treatment groups at immediate posttest (Table 4) In con-

trast preparators were absent in the control group Both treatment groups

outperformed the control group at immediate and delayed posttest No differ-

ence was found between the collaborative and individual groups z = 176

Table 3 Request head act scores

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 122 200 400

Immediate posttest 1012 306 200 1200

Delayed posttest 432 236 200 1100

Individual group (n = 25)

Pretest 340 071 200 400

Immediate posttest 836 341 300 1200

Delayed posttest 488 273 100 1200

Control group (n = 24)

Pretest 375 043 300 400

Immediate posttest 383 047 200 400

Delayed posttest 388 044 200 400

Note Each student produced four PDR-high requests Each head act was scored on a 3-point

scale Score range = 0ndash12 Delayed posttest was given 1 month later

N TAGUCHI AND Y KIM 11

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

ratings of these items were checked to avoid teaching situations that do not

represent real life in Korea Finally we analyzed participantsrsquo responses to the

open-ended question and used their descriptions when writing situations

Instruction focused on two categories of pragmalinguistic forms request

head acts and modifications (Table 1) Request head acts refer to the minimal

core unit that conveys the illocutionary force of request (Blum-Kulka et al

1989) In this category we taught mitigated preparatory forms Mitigated pre-

paratory involves syntactic forms that make reference to the hearerrsquos ability or

will in an embedded question or bi-clausal structure (eg lsquoI was wondering

ifrsquo + verb and lsquoWould it be possible torsquo + verb) (Takahashi 1996) Request modi-

fications have two categories external and internal External modifications

involve semantic moves that are attached externally to the head act and miti-

gate its force (Blum-Kulka et al 1989) We taught two external modifications

preparators (preparing a hearer for a request) and grounders (giving a reason

for request) Internal modifications include sentence-internal syntactic and

lexical devices used to modify the force of request We taught two forms

hedging that softens the tone of speech and amplifiers that strengthen self-

expression

These pragmalinguistic forms were selected based on baseline data collected

from native English speakers Twenty-four students in a US university

Table 1 Target pragmalinguistic forms

Head acts

Mitigated preparatory forms

Reference to the hearerrsquos ability will and possibility or reference to thespeakerrsquos wish in an embedded question or sentence

Irsquom wondering if you could give me an extension on the assignment

Is there any way that I could get an extension on the assignment

External modifications

Preparators

Semantic moves used to prepare the hearer for the request

eg May I ask you something

Grounders

Reason or explanation used to support the request

eg I caught a cold so I couldnrsquot finish the assignment

Internal modifications

Hedging

Words that minimize self-expression (maybe possibly)

eg Irsquom wondering if I could possibly have an extension on the assignment

Amplifiers

Words that strengthen self-expression (really very)

eg I really need more time to work on the assignment

6 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

completed a discourse completion task (DCT) that contained four PDR-high

and -low requests Mitigated preparatory appeared 73 percent of the time as a

head act in PDR-high requests while it appeared only 4 percent of the time in

PDR-low requests This indicates that mitigated preparatory characterizes PDR-

high requests Grounders and preparators were common supportive moves

appearing 98 and 13 percent of the time respectively while they never ap-

peared in PDR-low situations Hedging and amplifiers were also frequent ap-

pearing at a rate of 52 percent while they were absent in PDR-low requests

Based on these findings along with studies that revealed learnersrsquo difficulty

with these forms (Taguchi 2012) we targeted these forms Although DCT data

can only tell us what people would say in a hypothetical situation the rela-

tively uniform responses were considered to represent normative patterns of

request making

Instructional materials and tasks

Treatment groups received instruction in school during their regular English

class sessions over two consecutive days The instruction started with a 5-min

explanation of target pragmatic forms using a written dialogue There were

two dialogues one featuring a PDR-high and the other featuring a PDR-low

request We introduced pragmalinguistic forms and sociopragmatic variables in

each dialogue After receiving the direct metapragmatic information the two

treatment groups (collaborative and individual) proceeded to the dialogue con-

struction task during which learners were asked to complete drama scripts

based on given scenarios as drama scriptwriters They received two scenarios

(PDR-high and -low) with pictures of main characters and created a dialogue

involving a request based on each scenario1 These tasks were considered au-

thentic and relevant to learner interests because TV dramas are popular among

the target population and all the task scenarios were directly related to their

school life2 Learners constructed three PDR-high and -low situation dialogues

during treatment sessions (see Supplementary material for sample instruc-

tional materials) Although PDR-high requests were the instructional target

we introduced PDR-high and -low requests together so the differences be-

tween the two become salient

The collaborative group created a dialogue in pairs while the individual

group completed the task alone Learners vocalized their thoughts during

task performance in the language of their choice either collaboratively or

individually depending on their treatment condition To ensure learnersrsquo

understanding of the task following Kim (2013) we showed a 2-min pre-

task modeling video prior to the beginning of their task performance In the

modeling video for the collaborative group two teachers demonstrate how to

perform a task collaboratively using a similar scenario whereas in the video for

the individual group one teacher performed the same task while thinking

aloud on her own Interaction and think-aloud protocols during task comple-

tion were audio-recorded using individual MP3 recorders The treatment

N TAGUCHI AND Y KIM 7

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

session took about 45 min We repeated the same procedures in the second

task treatment session with different scenarios The control group received

regular English instruction

Assessment measure

A written DCT instrument was used to measure learning outcomes DCT

has been criticized because of a lack of authenticity and non-interactive

nature (Golato 2003) However DCT has a merit because the data can provide

information about learnersrsquo knowledge of normative conventions of prag-

matic language use (McNamara and Roever 2006) DCT can also control

social factors in scenarios and help us obtain data that are comparable across

learners over time Most importantly DCT in this study conformed to the

principle of lsquotransfer appropriate processingrsquo (DeKeyser 2007) which claims

that transfer of skill from a learned to a novel task occurs if the cognitive

operations involved in the novel task resemble those in the treatment task

The treatment task in this study was a dialogue construction task which

shared the same modality with DCT which elicited participantsrsquo pragmalinguis-

tic forms in writing

The DCT had 15 items 4 each of PDR-high and PDR-low requests and 7

filler items involving non-target speech acts Each DCT item had a situation

written in Korean to ensure participantsrsquo comprehension Participants were

asked to follow the first turn or prompt provided in English and write the

speech act in English In order to minimize the practice effect coming from

administering the same test repeatedly we prepared three versions of DCT

Scenarios were kept constant but minor wording changes were made in the

scenarios (eg changing names and locations) Different filler items were used

each time DCT items were different from the items used in the treatment

sessions Supplementary material contains sample DCT items

Procedure

The study was conducted over 6 weeks (see Table 2) During the first ses-

sion the learners took the pretest and two task groups carried out a practice

task to become familiar with recording devices and think-aloud

procedures Task treatment sessions were provided on Days 7 and 9 On

Days 10 and 14 all learners took the immediate posttest and the TOEFIC

Bridge respectively Four weeks after the immediate posttest the learners

took the delayed posttest

Data analysis

This study asked (i) whether task-based instruction is beneficial for learning

request-making expressions and if so whether there is a difference in learning

outcomes between the collaborative and individual group and (ii) to what

extent learners produce PREs during individual or collaborative tasks In

8 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

response to the first research question we first assigned scores on the request

head acts in the DCTs Because the purpose of this study was to compare

instructional effects between two task conditions (collaborative vs individual)

we focused our analysis on the forms we targeted during instruction Three

points were given if the head act contained one of the target forms (lsquoIrsquom

wondering ifrsquo + clause or lsquoIs there any wayrsquo + clause) and was also grammat-

ically accurate Two points were given if the head act took the target form but

was ungrammatical One point was given for a grammatical non-target form in

the context of this study (request in a form other than the mitigated prepara-

tory such as lsquoCan Irsquo) No point was given for an ungrammatical non-target

form and missing response3 Total scores from four PDR-high requests (scale

0ndash12) were compared across collaborative individual and control groups

using the KruskalndashWallis test In addition to head acts frequency of request

modifications (preparators grounders hedging and amplifiers) was counted

separately and compared across groups using the KruskalndashWallis test Both

authors coded 50 of the data yielding 953 agreement rate The discre-

pancies were solved through a follow-up discussion The non-parametric test

(KruskalndashWallis) was used because the data did not confirm normal

distribution

To answer the second research question task interaction and think-aloud

data were transcribed and analyzed for PREs defined as any part of language

production where learners talk about pragmalinguistics (request-making

forms) and sociopragmatics (contextual factors) they are attending question

their pragmatic language use or correct themselves or others PREs were coded

for context (eg setting interlocutor relationship and size of imposition) head

Table 2 Study procedures

Session Individual group Collaborative group Control group

Session 1(Day 1)

Pretest (DCT 1)Practice taskwith thinking aloud

Pretest (DCT 1)Practice task

Pretest (DCT 1)

Session 2(Day 7)

Explicit information(5 min)

Task modeling video(2 min)

Individual dialogueconstruction (35 min)

Explicit information (5 min)

Task modeling video (2 min)

Collaborative dialogueconstruction (35 min)

Reading activity

Session 3(Day 9)

Same procedure withdifferent scenarios

Same procedure with differentscenarios

Reading activity

Session 4(Day 10)

Immediate posttest (DCT 2) Immediate posttest (DCT 2) Immediateposttest (DCT 2)

Session 5(Day 14)

TOEIC Bridge TOEIC Bridge TOEIC Bridge

Session 6(Day 38)

Delayed posttest (DCT 3) Delayed posttest (DCT 3) Delayedposttest (DCT 3)

N TAGUCHI AND Y KIM 9

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

acts grounders preparators hedging and amplifiers See the following ex-

amples about asking the school principal to change a location of a picnic

Example 1 PRE targeting a request head act during interactive task

1 Learner 1 (making a request) I am wondering if I could pos-

sibly (First we have to make a request) I am won-

dering if you could possibly I was wondering

2 Learner 2 I am

3 Learner 1 I am wondering

4 Learner 2 if you could was could (Since

we used lsquocouldrsquo shouldnrsquot we use lsquowasrsquo) I was wondering if I could go

to picnic in Everland

5 Learner 1 if you could (No

since this person needs to make a decision lsquoif you couldrsquo sounds right) If

you could

Example 2 PRE targeting the request head act during think-aloud

Learner 1 In this fall fall picnic fall picnic picnic is there any is there

any way that I could that I could could is there can could possibly

possibly go to Everland participate possibly go to go to Everland

Once the PREs were coded studentsrsquo task performance data (the completed

written dialogues) was scored following the same rubrics as the DCT data and

the frequency of each modification in the dialogues was also counted The

second rater coded 20 percent of task performance data independently and

94 percent agreement was obtained Any disagreement was resolved through

discussion which was applied for the rest of the data coding

RESULTS

Effects of collaborative dialogue in learning the speech act ofrequest

Analysis of request head act

Table 3 displays descriptive statistics of head act scores Frequency analysis

showed that 295 out of the 296 head acts (74 learners 4 items) were non-

target (not the focus of instruction) Most of the non-target head acts took a

form of permission (lsquoMay Irsquo) or ability inquiry (lsquoCould yoursquo) indicating that all

groups were unfamiliar with the mitigated preparatory forms before the in-

struction The KruskalndashWallis test revealed no group difference at pretest

2 = 344 p = 18

The KruskalndashWallis test found a significant group difference at immediate

posttest 2 = 3190 plt 001 Pair-wise comparisons using the MannndashWhitney

U test showed that both treatment groups outperformed the control group

10 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

z = 536 plt 001 (collaborative group) and z = 405 plt 001 (individual group)

The collaborative group further surpassed the individual group on the produc-

tion of mitigated preparatory z = 240 p = 016 However the benefit of this

condition disappeared at delayed posttest because there was no difference be-

tween the treatment groups z = 38 p = 71 In addition the KruskalndashWallis test

showed that there was no significant difference between the control and treat-

ment groups at delayed posttest 2 = 25 p = 884

In summary task-based instruction resulted in a strong effect on the learn-

ing of appropriate request head acts as found in the treatment groupsrsquo superior

performance at immediate posttest compared with that of non-instructional

condition However the effect did not last long as the treatment groups went

back to the level of the control group at delayed posttest Similarly the col-

laborative task was more effective than the individual task at the immediate

posttest but the effect disappeared 1 month later

Analysis of request modification

Tables 4ndash7 present frequency counts of request modifications that appeared in

DCT Preparators almost never appeared at pretest but they showed a large

increase in the two treatment groups at immediate posttest (Table 4) In con-

trast preparators were absent in the control group Both treatment groups

outperformed the control group at immediate and delayed posttest No differ-

ence was found between the collaborative and individual groups z = 176

Table 3 Request head act scores

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 122 200 400

Immediate posttest 1012 306 200 1200

Delayed posttest 432 236 200 1100

Individual group (n = 25)

Pretest 340 071 200 400

Immediate posttest 836 341 300 1200

Delayed posttest 488 273 100 1200

Control group (n = 24)

Pretest 375 043 300 400

Immediate posttest 383 047 200 400

Delayed posttest 388 044 200 400

Note Each student produced four PDR-high requests Each head act was scored on a 3-point

scale Score range = 0ndash12 Delayed posttest was given 1 month later

N TAGUCHI AND Y KIM 11

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

completed a discourse completion task (DCT) that contained four PDR-high

and -low requests Mitigated preparatory appeared 73 percent of the time as a

head act in PDR-high requests while it appeared only 4 percent of the time in

PDR-low requests This indicates that mitigated preparatory characterizes PDR-

high requests Grounders and preparators were common supportive moves

appearing 98 and 13 percent of the time respectively while they never ap-

peared in PDR-low situations Hedging and amplifiers were also frequent ap-

pearing at a rate of 52 percent while they were absent in PDR-low requests

Based on these findings along with studies that revealed learnersrsquo difficulty

with these forms (Taguchi 2012) we targeted these forms Although DCT data

can only tell us what people would say in a hypothetical situation the rela-

tively uniform responses were considered to represent normative patterns of

request making

Instructional materials and tasks

Treatment groups received instruction in school during their regular English

class sessions over two consecutive days The instruction started with a 5-min

explanation of target pragmatic forms using a written dialogue There were

two dialogues one featuring a PDR-high and the other featuring a PDR-low

request We introduced pragmalinguistic forms and sociopragmatic variables in

each dialogue After receiving the direct metapragmatic information the two

treatment groups (collaborative and individual) proceeded to the dialogue con-

struction task during which learners were asked to complete drama scripts

based on given scenarios as drama scriptwriters They received two scenarios

(PDR-high and -low) with pictures of main characters and created a dialogue

involving a request based on each scenario1 These tasks were considered au-

thentic and relevant to learner interests because TV dramas are popular among

the target population and all the task scenarios were directly related to their

school life2 Learners constructed three PDR-high and -low situation dialogues

during treatment sessions (see Supplementary material for sample instruc-

tional materials) Although PDR-high requests were the instructional target

we introduced PDR-high and -low requests together so the differences be-

tween the two become salient

The collaborative group created a dialogue in pairs while the individual

group completed the task alone Learners vocalized their thoughts during

task performance in the language of their choice either collaboratively or

individually depending on their treatment condition To ensure learnersrsquo

understanding of the task following Kim (2013) we showed a 2-min pre-

task modeling video prior to the beginning of their task performance In the

modeling video for the collaborative group two teachers demonstrate how to

perform a task collaboratively using a similar scenario whereas in the video for

the individual group one teacher performed the same task while thinking

aloud on her own Interaction and think-aloud protocols during task comple-

tion were audio-recorded using individual MP3 recorders The treatment

N TAGUCHI AND Y KIM 7

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

session took about 45 min We repeated the same procedures in the second

task treatment session with different scenarios The control group received

regular English instruction

Assessment measure

A written DCT instrument was used to measure learning outcomes DCT

has been criticized because of a lack of authenticity and non-interactive

nature (Golato 2003) However DCT has a merit because the data can provide

information about learnersrsquo knowledge of normative conventions of prag-

matic language use (McNamara and Roever 2006) DCT can also control

social factors in scenarios and help us obtain data that are comparable across

learners over time Most importantly DCT in this study conformed to the

principle of lsquotransfer appropriate processingrsquo (DeKeyser 2007) which claims

that transfer of skill from a learned to a novel task occurs if the cognitive

operations involved in the novel task resemble those in the treatment task

The treatment task in this study was a dialogue construction task which

shared the same modality with DCT which elicited participantsrsquo pragmalinguis-

tic forms in writing

The DCT had 15 items 4 each of PDR-high and PDR-low requests and 7

filler items involving non-target speech acts Each DCT item had a situation

written in Korean to ensure participantsrsquo comprehension Participants were

asked to follow the first turn or prompt provided in English and write the

speech act in English In order to minimize the practice effect coming from

administering the same test repeatedly we prepared three versions of DCT

Scenarios were kept constant but minor wording changes were made in the

scenarios (eg changing names and locations) Different filler items were used

each time DCT items were different from the items used in the treatment

sessions Supplementary material contains sample DCT items

Procedure

The study was conducted over 6 weeks (see Table 2) During the first ses-

sion the learners took the pretest and two task groups carried out a practice

task to become familiar with recording devices and think-aloud

procedures Task treatment sessions were provided on Days 7 and 9 On

Days 10 and 14 all learners took the immediate posttest and the TOEFIC

Bridge respectively Four weeks after the immediate posttest the learners

took the delayed posttest

Data analysis

This study asked (i) whether task-based instruction is beneficial for learning

request-making expressions and if so whether there is a difference in learning

outcomes between the collaborative and individual group and (ii) to what

extent learners produce PREs during individual or collaborative tasks In

8 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

response to the first research question we first assigned scores on the request

head acts in the DCTs Because the purpose of this study was to compare

instructional effects between two task conditions (collaborative vs individual)

we focused our analysis on the forms we targeted during instruction Three

points were given if the head act contained one of the target forms (lsquoIrsquom

wondering ifrsquo + clause or lsquoIs there any wayrsquo + clause) and was also grammat-

ically accurate Two points were given if the head act took the target form but

was ungrammatical One point was given for a grammatical non-target form in

the context of this study (request in a form other than the mitigated prepara-

tory such as lsquoCan Irsquo) No point was given for an ungrammatical non-target

form and missing response3 Total scores from four PDR-high requests (scale

0ndash12) were compared across collaborative individual and control groups

using the KruskalndashWallis test In addition to head acts frequency of request

modifications (preparators grounders hedging and amplifiers) was counted

separately and compared across groups using the KruskalndashWallis test Both

authors coded 50 of the data yielding 953 agreement rate The discre-

pancies were solved through a follow-up discussion The non-parametric test

(KruskalndashWallis) was used because the data did not confirm normal

distribution

To answer the second research question task interaction and think-aloud

data were transcribed and analyzed for PREs defined as any part of language

production where learners talk about pragmalinguistics (request-making

forms) and sociopragmatics (contextual factors) they are attending question

their pragmatic language use or correct themselves or others PREs were coded

for context (eg setting interlocutor relationship and size of imposition) head

Table 2 Study procedures

Session Individual group Collaborative group Control group

Session 1(Day 1)

Pretest (DCT 1)Practice taskwith thinking aloud

Pretest (DCT 1)Practice task

Pretest (DCT 1)

Session 2(Day 7)

Explicit information(5 min)

Task modeling video(2 min)

Individual dialogueconstruction (35 min)

Explicit information (5 min)

Task modeling video (2 min)

Collaborative dialogueconstruction (35 min)

Reading activity

Session 3(Day 9)

Same procedure withdifferent scenarios

Same procedure with differentscenarios

Reading activity

Session 4(Day 10)

Immediate posttest (DCT 2) Immediate posttest (DCT 2) Immediateposttest (DCT 2)

Session 5(Day 14)

TOEIC Bridge TOEIC Bridge TOEIC Bridge

Session 6(Day 38)

Delayed posttest (DCT 3) Delayed posttest (DCT 3) Delayedposttest (DCT 3)

N TAGUCHI AND Y KIM 9

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

acts grounders preparators hedging and amplifiers See the following ex-

amples about asking the school principal to change a location of a picnic

Example 1 PRE targeting a request head act during interactive task

1 Learner 1 (making a request) I am wondering if I could pos-

sibly (First we have to make a request) I am won-

dering if you could possibly I was wondering

2 Learner 2 I am

3 Learner 1 I am wondering

4 Learner 2 if you could was could (Since

we used lsquocouldrsquo shouldnrsquot we use lsquowasrsquo) I was wondering if I could go

to picnic in Everland

5 Learner 1 if you could (No

since this person needs to make a decision lsquoif you couldrsquo sounds right) If

you could

Example 2 PRE targeting the request head act during think-aloud

Learner 1 In this fall fall picnic fall picnic picnic is there any is there

any way that I could that I could could is there can could possibly

possibly go to Everland participate possibly go to go to Everland

Once the PREs were coded studentsrsquo task performance data (the completed

written dialogues) was scored following the same rubrics as the DCT data and

the frequency of each modification in the dialogues was also counted The

second rater coded 20 percent of task performance data independently and

94 percent agreement was obtained Any disagreement was resolved through

discussion which was applied for the rest of the data coding

RESULTS

Effects of collaborative dialogue in learning the speech act ofrequest

Analysis of request head act

Table 3 displays descriptive statistics of head act scores Frequency analysis

showed that 295 out of the 296 head acts (74 learners 4 items) were non-

target (not the focus of instruction) Most of the non-target head acts took a

form of permission (lsquoMay Irsquo) or ability inquiry (lsquoCould yoursquo) indicating that all

groups were unfamiliar with the mitigated preparatory forms before the in-

struction The KruskalndashWallis test revealed no group difference at pretest

2 = 344 p = 18

The KruskalndashWallis test found a significant group difference at immediate

posttest 2 = 3190 plt 001 Pair-wise comparisons using the MannndashWhitney

U test showed that both treatment groups outperformed the control group

10 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

z = 536 plt 001 (collaborative group) and z = 405 plt 001 (individual group)

The collaborative group further surpassed the individual group on the produc-

tion of mitigated preparatory z = 240 p = 016 However the benefit of this

condition disappeared at delayed posttest because there was no difference be-

tween the treatment groups z = 38 p = 71 In addition the KruskalndashWallis test

showed that there was no significant difference between the control and treat-

ment groups at delayed posttest 2 = 25 p = 884

In summary task-based instruction resulted in a strong effect on the learn-

ing of appropriate request head acts as found in the treatment groupsrsquo superior

performance at immediate posttest compared with that of non-instructional

condition However the effect did not last long as the treatment groups went

back to the level of the control group at delayed posttest Similarly the col-

laborative task was more effective than the individual task at the immediate

posttest but the effect disappeared 1 month later

Analysis of request modification

Tables 4ndash7 present frequency counts of request modifications that appeared in

DCT Preparators almost never appeared at pretest but they showed a large

increase in the two treatment groups at immediate posttest (Table 4) In con-

trast preparators were absent in the control group Both treatment groups

outperformed the control group at immediate and delayed posttest No differ-

ence was found between the collaborative and individual groups z = 176

Table 3 Request head act scores

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 122 200 400

Immediate posttest 1012 306 200 1200

Delayed posttest 432 236 200 1100

Individual group (n = 25)

Pretest 340 071 200 400

Immediate posttest 836 341 300 1200

Delayed posttest 488 273 100 1200

Control group (n = 24)

Pretest 375 043 300 400

Immediate posttest 383 047 200 400

Delayed posttest 388 044 200 400

Note Each student produced four PDR-high requests Each head act was scored on a 3-point

scale Score range = 0ndash12 Delayed posttest was given 1 month later

N TAGUCHI AND Y KIM 11

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

session took about 45 min We repeated the same procedures in the second

task treatment session with different scenarios The control group received

regular English instruction

Assessment measure

A written DCT instrument was used to measure learning outcomes DCT

has been criticized because of a lack of authenticity and non-interactive

nature (Golato 2003) However DCT has a merit because the data can provide

information about learnersrsquo knowledge of normative conventions of prag-

matic language use (McNamara and Roever 2006) DCT can also control

social factors in scenarios and help us obtain data that are comparable across

learners over time Most importantly DCT in this study conformed to the

principle of lsquotransfer appropriate processingrsquo (DeKeyser 2007) which claims

that transfer of skill from a learned to a novel task occurs if the cognitive

operations involved in the novel task resemble those in the treatment task

The treatment task in this study was a dialogue construction task which

shared the same modality with DCT which elicited participantsrsquo pragmalinguis-

tic forms in writing

The DCT had 15 items 4 each of PDR-high and PDR-low requests and 7

filler items involving non-target speech acts Each DCT item had a situation

written in Korean to ensure participantsrsquo comprehension Participants were

asked to follow the first turn or prompt provided in English and write the

speech act in English In order to minimize the practice effect coming from

administering the same test repeatedly we prepared three versions of DCT

Scenarios were kept constant but minor wording changes were made in the

scenarios (eg changing names and locations) Different filler items were used

each time DCT items were different from the items used in the treatment

sessions Supplementary material contains sample DCT items

Procedure

The study was conducted over 6 weeks (see Table 2) During the first ses-

sion the learners took the pretest and two task groups carried out a practice

task to become familiar with recording devices and think-aloud

procedures Task treatment sessions were provided on Days 7 and 9 On

Days 10 and 14 all learners took the immediate posttest and the TOEFIC

Bridge respectively Four weeks after the immediate posttest the learners

took the delayed posttest

Data analysis

This study asked (i) whether task-based instruction is beneficial for learning

request-making expressions and if so whether there is a difference in learning

outcomes between the collaborative and individual group and (ii) to what

extent learners produce PREs during individual or collaborative tasks In

8 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

response to the first research question we first assigned scores on the request

head acts in the DCTs Because the purpose of this study was to compare

instructional effects between two task conditions (collaborative vs individual)

we focused our analysis on the forms we targeted during instruction Three

points were given if the head act contained one of the target forms (lsquoIrsquom

wondering ifrsquo + clause or lsquoIs there any wayrsquo + clause) and was also grammat-

ically accurate Two points were given if the head act took the target form but

was ungrammatical One point was given for a grammatical non-target form in

the context of this study (request in a form other than the mitigated prepara-

tory such as lsquoCan Irsquo) No point was given for an ungrammatical non-target

form and missing response3 Total scores from four PDR-high requests (scale

0ndash12) were compared across collaborative individual and control groups

using the KruskalndashWallis test In addition to head acts frequency of request

modifications (preparators grounders hedging and amplifiers) was counted

separately and compared across groups using the KruskalndashWallis test Both

authors coded 50 of the data yielding 953 agreement rate The discre-

pancies were solved through a follow-up discussion The non-parametric test

(KruskalndashWallis) was used because the data did not confirm normal

distribution

To answer the second research question task interaction and think-aloud

data were transcribed and analyzed for PREs defined as any part of language

production where learners talk about pragmalinguistics (request-making

forms) and sociopragmatics (contextual factors) they are attending question

their pragmatic language use or correct themselves or others PREs were coded

for context (eg setting interlocutor relationship and size of imposition) head

Table 2 Study procedures

Session Individual group Collaborative group Control group

Session 1(Day 1)

Pretest (DCT 1)Practice taskwith thinking aloud

Pretest (DCT 1)Practice task

Pretest (DCT 1)

Session 2(Day 7)

Explicit information(5 min)

Task modeling video(2 min)

Individual dialogueconstruction (35 min)

Explicit information (5 min)

Task modeling video (2 min)

Collaborative dialogueconstruction (35 min)

Reading activity

Session 3(Day 9)

Same procedure withdifferent scenarios

Same procedure with differentscenarios

Reading activity

Session 4(Day 10)

Immediate posttest (DCT 2) Immediate posttest (DCT 2) Immediateposttest (DCT 2)

Session 5(Day 14)

TOEIC Bridge TOEIC Bridge TOEIC Bridge

Session 6(Day 38)

Delayed posttest (DCT 3) Delayed posttest (DCT 3) Delayedposttest (DCT 3)

N TAGUCHI AND Y KIM 9

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

acts grounders preparators hedging and amplifiers See the following ex-

amples about asking the school principal to change a location of a picnic

Example 1 PRE targeting a request head act during interactive task

1 Learner 1 (making a request) I am wondering if I could pos-

sibly (First we have to make a request) I am won-

dering if you could possibly I was wondering

2 Learner 2 I am

3 Learner 1 I am wondering

4 Learner 2 if you could was could (Since

we used lsquocouldrsquo shouldnrsquot we use lsquowasrsquo) I was wondering if I could go

to picnic in Everland

5 Learner 1 if you could (No

since this person needs to make a decision lsquoif you couldrsquo sounds right) If

you could

Example 2 PRE targeting the request head act during think-aloud

Learner 1 In this fall fall picnic fall picnic picnic is there any is there

any way that I could that I could could is there can could possibly

possibly go to Everland participate possibly go to go to Everland

Once the PREs were coded studentsrsquo task performance data (the completed

written dialogues) was scored following the same rubrics as the DCT data and

the frequency of each modification in the dialogues was also counted The

second rater coded 20 percent of task performance data independently and

94 percent agreement was obtained Any disagreement was resolved through

discussion which was applied for the rest of the data coding

RESULTS

Effects of collaborative dialogue in learning the speech act ofrequest

Analysis of request head act

Table 3 displays descriptive statistics of head act scores Frequency analysis

showed that 295 out of the 296 head acts (74 learners 4 items) were non-

target (not the focus of instruction) Most of the non-target head acts took a

form of permission (lsquoMay Irsquo) or ability inquiry (lsquoCould yoursquo) indicating that all

groups were unfamiliar with the mitigated preparatory forms before the in-

struction The KruskalndashWallis test revealed no group difference at pretest

2 = 344 p = 18

The KruskalndashWallis test found a significant group difference at immediate

posttest 2 = 3190 plt 001 Pair-wise comparisons using the MannndashWhitney

U test showed that both treatment groups outperformed the control group

10 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

z = 536 plt 001 (collaborative group) and z = 405 plt 001 (individual group)

The collaborative group further surpassed the individual group on the produc-

tion of mitigated preparatory z = 240 p = 016 However the benefit of this

condition disappeared at delayed posttest because there was no difference be-

tween the treatment groups z = 38 p = 71 In addition the KruskalndashWallis test

showed that there was no significant difference between the control and treat-

ment groups at delayed posttest 2 = 25 p = 884

In summary task-based instruction resulted in a strong effect on the learn-

ing of appropriate request head acts as found in the treatment groupsrsquo superior

performance at immediate posttest compared with that of non-instructional

condition However the effect did not last long as the treatment groups went

back to the level of the control group at delayed posttest Similarly the col-

laborative task was more effective than the individual task at the immediate

posttest but the effect disappeared 1 month later

Analysis of request modification

Tables 4ndash7 present frequency counts of request modifications that appeared in

DCT Preparators almost never appeared at pretest but they showed a large

increase in the two treatment groups at immediate posttest (Table 4) In con-

trast preparators were absent in the control group Both treatment groups

outperformed the control group at immediate and delayed posttest No differ-

ence was found between the collaborative and individual groups z = 176

Table 3 Request head act scores

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 122 200 400

Immediate posttest 1012 306 200 1200

Delayed posttest 432 236 200 1100

Individual group (n = 25)

Pretest 340 071 200 400

Immediate posttest 836 341 300 1200

Delayed posttest 488 273 100 1200

Control group (n = 24)

Pretest 375 043 300 400

Immediate posttest 383 047 200 400

Delayed posttest 388 044 200 400

Note Each student produced four PDR-high requests Each head act was scored on a 3-point

scale Score range = 0ndash12 Delayed posttest was given 1 month later

N TAGUCHI AND Y KIM 11

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

response to the first research question we first assigned scores on the request

head acts in the DCTs Because the purpose of this study was to compare

instructional effects between two task conditions (collaborative vs individual)

we focused our analysis on the forms we targeted during instruction Three

points were given if the head act contained one of the target forms (lsquoIrsquom

wondering ifrsquo + clause or lsquoIs there any wayrsquo + clause) and was also grammat-

ically accurate Two points were given if the head act took the target form but

was ungrammatical One point was given for a grammatical non-target form in

the context of this study (request in a form other than the mitigated prepara-

tory such as lsquoCan Irsquo) No point was given for an ungrammatical non-target

form and missing response3 Total scores from four PDR-high requests (scale

0ndash12) were compared across collaborative individual and control groups

using the KruskalndashWallis test In addition to head acts frequency of request

modifications (preparators grounders hedging and amplifiers) was counted

separately and compared across groups using the KruskalndashWallis test Both

authors coded 50 of the data yielding 953 agreement rate The discre-

pancies were solved through a follow-up discussion The non-parametric test

(KruskalndashWallis) was used because the data did not confirm normal

distribution

To answer the second research question task interaction and think-aloud

data were transcribed and analyzed for PREs defined as any part of language

production where learners talk about pragmalinguistics (request-making

forms) and sociopragmatics (contextual factors) they are attending question

their pragmatic language use or correct themselves or others PREs were coded

for context (eg setting interlocutor relationship and size of imposition) head

Table 2 Study procedures

Session Individual group Collaborative group Control group

Session 1(Day 1)

Pretest (DCT 1)Practice taskwith thinking aloud

Pretest (DCT 1)Practice task

Pretest (DCT 1)

Session 2(Day 7)

Explicit information(5 min)

Task modeling video(2 min)

Individual dialogueconstruction (35 min)

Explicit information (5 min)

Task modeling video (2 min)

Collaborative dialogueconstruction (35 min)

Reading activity

Session 3(Day 9)

Same procedure withdifferent scenarios

Same procedure with differentscenarios

Reading activity

Session 4(Day 10)

Immediate posttest (DCT 2) Immediate posttest (DCT 2) Immediateposttest (DCT 2)

Session 5(Day 14)

TOEIC Bridge TOEIC Bridge TOEIC Bridge

Session 6(Day 38)

Delayed posttest (DCT 3) Delayed posttest (DCT 3) Delayedposttest (DCT 3)

N TAGUCHI AND Y KIM 9

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

acts grounders preparators hedging and amplifiers See the following ex-

amples about asking the school principal to change a location of a picnic

Example 1 PRE targeting a request head act during interactive task

1 Learner 1 (making a request) I am wondering if I could pos-

sibly (First we have to make a request) I am won-

dering if you could possibly I was wondering

2 Learner 2 I am

3 Learner 1 I am wondering

4 Learner 2 if you could was could (Since

we used lsquocouldrsquo shouldnrsquot we use lsquowasrsquo) I was wondering if I could go

to picnic in Everland

5 Learner 1 if you could (No

since this person needs to make a decision lsquoif you couldrsquo sounds right) If

you could

Example 2 PRE targeting the request head act during think-aloud

Learner 1 In this fall fall picnic fall picnic picnic is there any is there

any way that I could that I could could is there can could possibly

possibly go to Everland participate possibly go to go to Everland

Once the PREs were coded studentsrsquo task performance data (the completed

written dialogues) was scored following the same rubrics as the DCT data and

the frequency of each modification in the dialogues was also counted The

second rater coded 20 percent of task performance data independently and

94 percent agreement was obtained Any disagreement was resolved through

discussion which was applied for the rest of the data coding

RESULTS

Effects of collaborative dialogue in learning the speech act ofrequest

Analysis of request head act

Table 3 displays descriptive statistics of head act scores Frequency analysis

showed that 295 out of the 296 head acts (74 learners 4 items) were non-

target (not the focus of instruction) Most of the non-target head acts took a

form of permission (lsquoMay Irsquo) or ability inquiry (lsquoCould yoursquo) indicating that all

groups were unfamiliar with the mitigated preparatory forms before the in-

struction The KruskalndashWallis test revealed no group difference at pretest

2 = 344 p = 18

The KruskalndashWallis test found a significant group difference at immediate

posttest 2 = 3190 plt 001 Pair-wise comparisons using the MannndashWhitney

U test showed that both treatment groups outperformed the control group

10 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

z = 536 plt 001 (collaborative group) and z = 405 plt 001 (individual group)

The collaborative group further surpassed the individual group on the produc-

tion of mitigated preparatory z = 240 p = 016 However the benefit of this

condition disappeared at delayed posttest because there was no difference be-

tween the treatment groups z = 38 p = 71 In addition the KruskalndashWallis test

showed that there was no significant difference between the control and treat-

ment groups at delayed posttest 2 = 25 p = 884

In summary task-based instruction resulted in a strong effect on the learn-

ing of appropriate request head acts as found in the treatment groupsrsquo superior

performance at immediate posttest compared with that of non-instructional

condition However the effect did not last long as the treatment groups went

back to the level of the control group at delayed posttest Similarly the col-

laborative task was more effective than the individual task at the immediate

posttest but the effect disappeared 1 month later

Analysis of request modification

Tables 4ndash7 present frequency counts of request modifications that appeared in

DCT Preparators almost never appeared at pretest but they showed a large

increase in the two treatment groups at immediate posttest (Table 4) In con-

trast preparators were absent in the control group Both treatment groups

outperformed the control group at immediate and delayed posttest No differ-

ence was found between the collaborative and individual groups z = 176

Table 3 Request head act scores

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 122 200 400

Immediate posttest 1012 306 200 1200

Delayed posttest 432 236 200 1100

Individual group (n = 25)

Pretest 340 071 200 400

Immediate posttest 836 341 300 1200

Delayed posttest 488 273 100 1200

Control group (n = 24)

Pretest 375 043 300 400

Immediate posttest 383 047 200 400

Delayed posttest 388 044 200 400

Note Each student produced four PDR-high requests Each head act was scored on a 3-point

scale Score range = 0ndash12 Delayed posttest was given 1 month later

N TAGUCHI AND Y KIM 11

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

acts grounders preparators hedging and amplifiers See the following ex-

amples about asking the school principal to change a location of a picnic

Example 1 PRE targeting a request head act during interactive task

1 Learner 1 (making a request) I am wondering if I could pos-

sibly (First we have to make a request) I am won-

dering if you could possibly I was wondering

2 Learner 2 I am

3 Learner 1 I am wondering

4 Learner 2 if you could was could (Since

we used lsquocouldrsquo shouldnrsquot we use lsquowasrsquo) I was wondering if I could go

to picnic in Everland

5 Learner 1 if you could (No

since this person needs to make a decision lsquoif you couldrsquo sounds right) If

you could

Example 2 PRE targeting the request head act during think-aloud

Learner 1 In this fall fall picnic fall picnic picnic is there any is there

any way that I could that I could could is there can could possibly

possibly go to Everland participate possibly go to go to Everland

Once the PREs were coded studentsrsquo task performance data (the completed

written dialogues) was scored following the same rubrics as the DCT data and

the frequency of each modification in the dialogues was also counted The

second rater coded 20 percent of task performance data independently and

94 percent agreement was obtained Any disagreement was resolved through

discussion which was applied for the rest of the data coding

RESULTS

Effects of collaborative dialogue in learning the speech act ofrequest

Analysis of request head act

Table 3 displays descriptive statistics of head act scores Frequency analysis

showed that 295 out of the 296 head acts (74 learners 4 items) were non-

target (not the focus of instruction) Most of the non-target head acts took a

form of permission (lsquoMay Irsquo) or ability inquiry (lsquoCould yoursquo) indicating that all

groups were unfamiliar with the mitigated preparatory forms before the in-

struction The KruskalndashWallis test revealed no group difference at pretest

2 = 344 p = 18

The KruskalndashWallis test found a significant group difference at immediate

posttest 2 = 3190 plt 001 Pair-wise comparisons using the MannndashWhitney

U test showed that both treatment groups outperformed the control group

10 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

z = 536 plt 001 (collaborative group) and z = 405 plt 001 (individual group)

The collaborative group further surpassed the individual group on the produc-

tion of mitigated preparatory z = 240 p = 016 However the benefit of this

condition disappeared at delayed posttest because there was no difference be-

tween the treatment groups z = 38 p = 71 In addition the KruskalndashWallis test

showed that there was no significant difference between the control and treat-

ment groups at delayed posttest 2 = 25 p = 884

In summary task-based instruction resulted in a strong effect on the learn-

ing of appropriate request head acts as found in the treatment groupsrsquo superior

performance at immediate posttest compared with that of non-instructional

condition However the effect did not last long as the treatment groups went

back to the level of the control group at delayed posttest Similarly the col-

laborative task was more effective than the individual task at the immediate

posttest but the effect disappeared 1 month later

Analysis of request modification

Tables 4ndash7 present frequency counts of request modifications that appeared in

DCT Preparators almost never appeared at pretest but they showed a large

increase in the two treatment groups at immediate posttest (Table 4) In con-

trast preparators were absent in the control group Both treatment groups

outperformed the control group at immediate and delayed posttest No differ-

ence was found between the collaborative and individual groups z = 176

Table 3 Request head act scores

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 122 200 400

Immediate posttest 1012 306 200 1200

Delayed posttest 432 236 200 1100

Individual group (n = 25)

Pretest 340 071 200 400

Immediate posttest 836 341 300 1200

Delayed posttest 488 273 100 1200

Control group (n = 24)

Pretest 375 043 300 400

Immediate posttest 383 047 200 400

Delayed posttest 388 044 200 400

Note Each student produced four PDR-high requests Each head act was scored on a 3-point

scale Score range = 0ndash12 Delayed posttest was given 1 month later

N TAGUCHI AND Y KIM 11

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

z = 536 plt 001 (collaborative group) and z = 405 plt 001 (individual group)

The collaborative group further surpassed the individual group on the produc-

tion of mitigated preparatory z = 240 p = 016 However the benefit of this

condition disappeared at delayed posttest because there was no difference be-

tween the treatment groups z = 38 p = 71 In addition the KruskalndashWallis test

showed that there was no significant difference between the control and treat-

ment groups at delayed posttest 2 = 25 p = 884

In summary task-based instruction resulted in a strong effect on the learn-

ing of appropriate request head acts as found in the treatment groupsrsquo superior

performance at immediate posttest compared with that of non-instructional

condition However the effect did not last long as the treatment groups went

back to the level of the control group at delayed posttest Similarly the col-

laborative task was more effective than the individual task at the immediate

posttest but the effect disappeared 1 month later

Analysis of request modification

Tables 4ndash7 present frequency counts of request modifications that appeared in

DCT Preparators almost never appeared at pretest but they showed a large

increase in the two treatment groups at immediate posttest (Table 4) In con-

trast preparators were absent in the control group Both treatment groups

outperformed the control group at immediate and delayed posttest No differ-

ence was found between the collaborative and individual groups z = 176

Table 3 Request head act scores

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 122 200 400

Immediate posttest 1012 306 200 1200

Delayed posttest 432 236 200 1100

Individual group (n = 25)

Pretest 340 071 200 400

Immediate posttest 836 341 300 1200

Delayed posttest 488 273 100 1200

Control group (n = 24)

Pretest 375 043 300 400

Immediate posttest 383 047 200 400

Delayed posttest 388 044 200 400

Note Each student produced four PDR-high requests Each head act was scored on a 3-point

scale Score range = 0ndash12 Delayed posttest was given 1 month later

N TAGUCHI AND Y KIM 11

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

p = 079 at pretest z = 097 p = 33 at immediate posttest and z = 012 p = 90 at

delayed posttest

The grounders showed different patterns they appeared in majority of the

items at pretest (see Table 5) indicating that the learners were already familiar

with this supportive move before the instruction because grounders appear in

Table 4 Frequency of preparators

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 028 061 000 200

Immediate posttest 276 151 000 400

Delayed posttest 228 172 000 400

Individual group (n = 25)

Pretest 004 020 000 100

Immediate posttest 316 131 000 400

Delayed posttest 236 173 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 004 020 000 100

Delayed posttest 013 045 000 200

Note Mean = mean frequency of preparators in four PDR-high request items

Table 5 Frequency of grounders

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 336 081 100 400

Immediate posttest 336 095 100 400

Delayed posttest 356 096 000 400

Individual group (n = 25)

Pretest 324 088 200 400

Immediate posttest 332 075 200 400

Delayed posttest 340 087 100 400

Control group (n = 24)

Pretest 350 114 000 400

Immediate posttest 346 098 000 400

Delayed posttest 379 051 200 400

Note Mean = mean frequency of grounders in four PDR-high request items

12 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Korean requests and the learners were able to transfer their L1-based strategy

Because of this ceiling effect there was no instructional effect The results of

the KruskalndashWallis test showed no significant difference in any of the test

sessions 2 = 274 p = 25 at pretest 2 = 121 p = 55 at immediate posttest

and 2= 355 p = 17 at delayed posttest5

Table 7 Frequency of amplifiers

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 016 047 000 200

Immediate posttest 092 119 000 400

Delayed posttest 028 054 000 200

Individual group (n = 25)

Pretest 036 064 000 200

Immediate posttest 124 109 000 400

Delayed posttest 044 077 000 200

Control group (n = 24)

Pretest 063 097 000 300

Immediate posttest 071 096 000 400

Delayed posttest 033 076 000 300

Note Mean = mean frequency of amplifiers in four PDR-high request items

Table 6 Frequency of hedging

Group Mean SD Minimum Maximum

Collaborative group (n = 25)

Pretest 000 000 000 000

Immediate posttest 168 149 000 400

Delayed posttest 012 060 000 300

Individual group (n = 25)

Pretest 000 000 000 000

Immediate posttest 140 153 000 400

Delayed posttest 028 098 000 400

Control group (n = 24)

Pretest 000 000 000 000

Immediate posttest 000 000 000 000

Delayed posttest 008 028 000 100

Note Mean = mean frequency of hedging in four PDR-high request items

N TAGUCHI AND Y KIM 13

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Compared with external modifications internal modifications (hedging and

amplifiers) were relatively underused It was absent in the pretest data There

was a noticeable increase at immediate posttest in the treatment groups but

the mean frequency was only moderate Although the frequency was not

particularly large we still note the instructional effect because the treatment

groups outperformed the control group at the immediate posttest 2 = 481

plt 0001 for the collaborative group and 2 = 422 plt 0001 for the individual

group However the effect was short-lived because no group difference was

found at delayed posttest (2 = 43 p = 81) When the two treatment groups

were compared there was no significant difference z = 73 p = 47 at immedi-

ate posttest or z = 61 p = 54 at delayed posttest

Amplifiers also showed very small occurrence at pretest but different from

hedging the frequency remained low throughout the time No instructional

effect was observed in the case of amplifiers as there was no significant group

difference at pretest (2 = 449 p = 11) immediate posttest (2 = 393 p = 14)

or delayed posttest (2 = 45 p = 80)

In summary frequency comparison of request modification across groups

revealed a mixed picture At pretest no group difference was found in any of

these modifications At immediate posttest both treatment groups surpassed

the control group on the production of preparators and hedging but not on the

use of grounders and amplifiers At delayed posttest the treatment groups

outperformed the control group on the production of the preparators only

Most importantly there was no significant difference between the collabora-

tive and individual group in any of these modifications These findings mark a

sharp contrast to those of request head acts in which the collaborative condi-

tion was more advantageous than the individual task condition in learnersrsquo

production of mitigated preparatory forms

Frequency of PREs and quality of treatment task performance

Table 8 reports descriptive statistics for the frequency of PREs Over the two

treatment sessions the individual group produced 1160 PREs on average

while the mean for the collaborative group was 1632 The MannndashWhitney

U test showed that the difference was significant (z =276 p = 006)

Regarding PREs on specific features there was no group difference on PREs

targeting the contextual features of request (z =92 p = 36) head acts

(z = 28 p = 78) or grounders (z = 70 p = 48) but a difference was found in

preparators (z = 465 plt 0001) amplifiers (z = 413 plt 0001) and hedging

(z = 296 p = 003) in favor of the collaborative group

We will turn to the analysis of learnersrsquo task outcome by examining their

accurate resolution of PREs We analyzed the outcome of the dialogue con-

struction tasks using the same scoring rubrics from the DCT analysis 0ndash3 point

scale for head acts and frequency count for modifications (9 maximum points

for three treatment tasks) The individual group scored a mean of 724

(SD = 145) whereas the collaborative group scored a mean of 828

14 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

(SD = 84) This difference was significant z =289 plt 0001 indicating that

the collaborative group resolved PREs more successfully than the individual

group and produced more accurate target head acts as task outcome In con-

trast frequency of other request modifications revealed no group difference

preparators z =143 p = 15 grounders z =182 p = 07 hedging z =95

p = 34 and amplifiers z =147 p = 14

Example 3 demonstrates a sample PRE by a pair in the collaborative group

Here the learners are negotiating the target head act form lsquoI was wondering ifrsquo

by attending to each otherrsquos form and providing feedback and confirmation

During this process the target form receives prominence through recycled use

over multiple turns The form is jointly constructed with both participants

contributing to the knowledge building This collaborative work eventually

leads to the production of the accurate target form as task outcome A by-

product of this process is likely the deeper level of processing of the form and

consolidation of pragmatic knowledge as suggested in the collaborative

grouprsquos superior performance at immediate posttest

Example 3 Correctly resolved PRE in the collaborative group

1 Learner 1 (Jaemyong left his Korean home-

work at home) (He

was asking whether he could go home to bring his homework during his

self-study session) (Then he should make a

request politely)

2 Learner 2 (Yes)

----several turns later----------------------------------------------

3 Learner 1 (We need to make a long

polite request) I was wondering

4 Learner 2 if I could go to

5 Learner 1 Go to the home (Is that lsquogo to the homersquo) Go to home

(Shouldnrsquot it be lsquogo to homersquo)

Table 8 Frequency of PREs

PREs targets Collaborative group (n = 25) Individual group (n = 25)

Mean SD Mean SD

Context 476 296 404 302

Head acts 296 1306 288 120

Preparators 252 087 088 105

Grounders 272 124 276 101

Amplifiers 184 111 052 077

Hedging 152 130 052 083

Total 1632 618 1160 548

N TAGUCHI AND Y KIM 15

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

6 Learner 2 I am

7 Learner 1 I was wondering if I could go to go to

8 Learner 2 Go home

9 Learner 1 go home

The nature of the collaborative PREs contrasts with that of individual PREs

In Example 4 the learner is working by herself to resolve the PRE targeting lsquois

there any way that I couldrsquo She recognizes that this is the PDR-high request

and that she should use the mitigated preparatory form She produces the

target form once but without a peer attending to her form and giving feed-

back she goes back to the form that she is used to (lsquoCan Irsquo) resulting in the

production of non-target form as task outcome These instances of incorrectly

resolved PREs in the individual task group probably led to their underperform-

ance at immediate posttest compared with the collaborative group

Example 4 Incorrectly resolved PRE in the individual group

Learner 1 (I need to ask

whether I can go home to bring the assignment during self-study ses-

sion) uh may uh big request is there any way that I could uh

(then) Is there any way that I could possible homework my home-

work Can I can I go to my house to get my homework And can I go to

my house to get my homework (Is this right)

Summary of the findings

The learners who completed tasks collaboratively produced more PREs and

target-like request head acts than those who completed the tasks individually

during the task and again at immediate posttest However the benefit of the

collaborative group was short-lived because their performance was the same

with that of the individual group 1 month after the instruction The benefit of

collaborative tasks was also partial because there was no group difference in

the production of request modifications during the task or at immediate

posttest

DISCUSSION

Our findings contribute to the task-based literature by demonstrating the

benefits of collaborative tasks in improving pragmatics knowledge In the lit-

erature of instructed pragmatics our findings support previous studies on

metapragmatic discussion Takimoto (2012) found that collaborative dialogue

in a form of metapragmatic discussion gave learners access to information

about pragmatic features which eventually led to better control of pragmatic

knowledge than the condition where learners completed the task alone with

or without verbalization of the target features This study adds to these findings

16 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

because it documented the nature of collaborative dialogue operationalized as

PREsmdashhow often learners discussed pragmatics features in interaction and

whether interaction led to better task performance and linked these with

the development of pragmatic knowledge It became clear that collaborative

tasks generated more PREs than individual tasks PREs also contributed to

learnersrsquo accurate use of request head acts during treatment tasks which led

to their superior performance at immediate posttest Because the quantity of

PREs targeting head acts was similar between the collaborative and individual

groups the quality of PREs namely the extent to which learners resolved

PREs correctly was crucial for learning The present results support previous

findings of LREs studies (eg Payant 2012) and extended them to pragmatics

This study implemented collaborative dialogue through a form of PREs in

which learners interacted with their peers and jointly constructed a dialogue

that involved the target speech act During this process learners attended the

pragmalinguistic forms and semantic moves and their correspondence with

contextual features (relationship power and degree of imposition) Learners

in the individual task group also attended the target pragmatic forms and

verbalized them on their own as observed in PREs during thinking aloud

However negotiation with others was absent Without negotiation working

alone did not generate as many opportunities for attention to pragmatic forms

in context or recycled use of them as did the collaborative dialogue during

interactive tasks This finding was supported by the fact that the individual

group produced fewer PREs than the collaborative group during the treatment

tasks Metapragmatic talk around the target head acts modifications and con-

textual features occurred more frequently in the collaborative group

Negotiation of pragmatic forms and their meaning during the process of joint

task completion prompts a deeper level of cognitive processing that helps con-

solidate learnersrsquo pragmatic knowledge and their subsequent use of the know-

ledge Critically when learners produce language in a collaborative

environment they do not just produce output but they also have an oppor-

tunity to receive feedback from peers while trying to make their language

precise (Swain and Lapkin 1998 Swain 2006) During interaction learners

attend to each otherrsquos forms evaluate their accuracy and appropriateness

and offer feedback and corrections to each other In this process two levels

of monitoring self-monitoring and other-monitoring are occurring simultan-

eously which essentially promote more accurate high-quality task outcomes

Analyses of treatment task data confirmed this The collaborative grouprsquos re-

quest head acts (mitigated preparatory) in the dialogue construction task were

significantly more accurate than those of the individual group which was the

result of monitoring scaffolding and peer feedback that featured the process of

joint task completion

In sum learners completed tasks more successfully when they worked col-

laboratively and this successful task performance led to significantly greater

gains of the knowledge of appropriate request head acts The findings are

consistent with the results of previous findings that joint activities improved

N TAGUCHI AND Y KIM 17

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

task performance in terms of the accurate production of the target linguistic

forms and subsequent learning of the forms (eg Storch 2007 Kim 2008) The

present study has advanced this body of literature by demonstrating that ef-

fects of collaborative tasks and PREs extend beyond grammar and vocabulary

PREs provide opportunities for pragmatic learning that involves a complex

mapping of formndashfunctionndashcontext

Although the efficacy of collaborative dialogue became clear in this study

questions remain as to why the effects were restricted to the request head acts

and did not extend to request modifications One possible interpretation of the

findings relates to the different assessment methods used for the head act and

modification Head acts were assessed using a scoring rubric that reflected both

appropriateness (choosing the target mitigated preparatory forms over other

request-making forms) and grammaticality (accuracy of the forms)

Modifications on the other hand were analyzed mainly on frequency of pro-

duction We made this decision because different from the head act that

occurs only once in a single speech act modifications typically co-occur over

multiple times In addition internal modifications used in this study were all

lexical involving only one word which made the analysis of grammaticality

irrelevant

It is possible that the collaborative dialogue made a difference when the

quality of pragmatic forms (accuracy and appropriateness) was considered

compared with simple quantity (frequency) As shown in the treatment task

data strength of the collaborative dialogue was observed in how learners scaf-

folded each other and provided feedback to each other while jointly carrying

out tasks It might be the case that these instances of negotiation and inter-

action occurring in the collaborative task pushed learners to a deeper level of

processing of the request head act which eventually helped them consolidate

their knowledge of both accuracy and appropriateness of the forms Because of

the linguistic complexity of the head act forms (bi-clausal structure) and the

more strenuous scoring method used to evaluate the forms learnersrsquo know-

ledge base had to be solid enough to allow retrieval and use of the knowledge

The collaborative condition that involved intensive attention negotiation

monitoring and peer feedback probably assisted the formation of this robust

knowledge Nevertheless both treatment conditions were equally effective on

the level of request modifications This might be due to the fact that the groups

were compared on the frequency of use (whether they were able to produce

them in an intelligible manner) not on the quality of use (absolute accuracy of

the forms)

These interpretations also corroborate with Takimotorsquos (2012) findings that

revealed the advantage of metapragmatic over non-metapragmatic discussion

on the acquisition of request Positive effects were found in the production-

based test (DCT) but not in the receptive-skills test (appropriateness judg-

ment) Takimoto attributed the findings to the relative ease of the receptive

task Participants without metapragmatic discussion were still able to cope with

judging appropriateness because the test posed lower demands than a

18 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

production task however they were not able to deal with the demand

coming from the DCT that required them to produce target forms accurately

and appropriately Because of the precise syntactic processing required in pro-

duction and resulting task demand learners benefitted more from metaprag-

matic discussion that assisted the consolidation of knowledge

Although the two treatment groups differed in their use of the request head

acts their performance was the same at the level of modifications Both groups

outperformed the control group in the production of preparators and hedging

right after the instruction as well as 1 month later in the case of preparators

These findings indicate a strong effect of instruction over non-instruction but

also reveal variation within the effect across modification types The strong

effect found in the preparators is probably due to the saliency of the prepara-

tors as semantic strategies Different from hedging or amplifiers that involve

precise pragmalinguistics preparators are part of the discourse moves that

characterize the structure of a request reflecting sociopragmatic information

Logistics and rituals of how to structure a request using preparators were prob-

ably easier to learn than the precise linguistics forms of a request and once

learned the knowledge stayed in the memory because of its saliency These

findings conform to the previous longitudinal findings that revealed that ac-

quisition of sociopragmatic rules (eg use of semantic strategies) is faster than

that of pragmalinguistic forms (Bardovi-Harlig and Hartford 1993 Taguchi

2012) The present findings confirmed this in an instructional study

LIMITATIONS OF THE STUDY AND IMPLICATIONS FORFUTURE RESEARCH

This study is limited to a small number of individuals in a single L2 group a

single measure and a narrowly defined pragmatic target (request) Future

studies should involve a greater participant pool with wider target languages

and pragmatic measures and examine the development of different aspects of

pragmatic competence in relation to collaborative dialogue

Equally important is an investigation into individual differences and lear-

nersrsquo perceptions of instruction It is possible that not all learners had equal

amount of participation or contribution to the collaborative tasks which con-

sequently restricted their knowledge-building process It is also possible that

learners were not familiar with the type of instructional activities used in this

study and needed more time to get accustomed to the tasks in order to gain the

full benefit from them Two class periods were probably not enough time to

boost their robust learning Because this study did not address individual lear-

ner variation and learnersrsquo perceptions of their learning processes future stu-

dies are called for in this direction

Additionally although there has been a surge amount of research on task

design and task implementation in instructed SLA the learning of pragmatics

using a variety of collaborative tasks has not been explored Because this study

N TAGUCHI AND Y KIM 19

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

used only one type of writing task future studies should explore task-related

variables to find ultimate task-based learning contexts for pragmatic

development

In the current study we used PREs as an analysis unit to determine the

amount of learnersrsquo attention to target pragmatic forms for both individual and

collaborative groups The results need to be interpreted with acknowledging

that the characteristics of PREs in these two groups might not be exactly

the same due to the different nature of task performance As a first step of

using PREs we focused on the overall frequency of PREs Future studies are

warranted to explore the quality of PREs (eg number of turnssentences per

PRE) to further our understanding of collaborative dialogue around prag-

matics Another limitation is that this study did not collect Supplementary

data such as stimulated verbal recall or interviews to understand actual cog-

nitive processing involved in PREs which should be explored in the future

research

Finally different effects of collaborative dialogue between the request head

acts and modifications found here need to be addressed more explicitly in the

future Learnability and teachability of pragmatic features have been discussed

in the literature pointing to the need for exploring the interaction among

instructional methods pragmatic targets and learner profiles (eg Roever

2009 Takahashi 2010 Taguhi 2011) Roever (2009) suggests that future re-

search should investigate what kinds of pragmatic features (eg syntactic miti-

gations and lexical downgraders) should be taught to what types of learners

(in terms of proficiency or other individual characteristics) using what kinds of

intervention methods (eg interaction vs non-interaction) Taguchi (forth-

coming) adds characteristics of assessment methods to this call because instruc-

tional effects diverge substantially depending on the demands involved

in the assessment measures Collaborative dialogue and PREs could serve

as a fruitful area to investigate this interaction among pragmatic targets in-

structional methods learnersrsquo instructional readiness and assessment

measures

SUPPLEMENTARY DATA

Supplementary material is available at Applied Linguistics online

ACKNOWLEDGEMENTS

We are grateful for three anonymous reviewers for their helpful comments on earlier versions of

the manuscript Special thanks also go to our research assistants Feng Xiao Franklin Bradfield

Sharon Kim and YeonJoo Jung for their help with data analysis We are responsible for all the

errors that may remain

Conflict of interest statement None declared

20 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES

1 We acknowledge the limitation of not

using video-recorded authentic prag-

matic materials

2 One of the reviewers stated that the au-

thenticity of tasks should be based on

whether situations happen in the

target culture We agree with this con-

cern However being familiar with

given contexts is crucial for the partici-

pants to be able to analyze the contexts

and discuss the pragmatic-related issues

Because making requests in both

Korean and English follow similar

discourse patterns we decided that it

was appropriate to use context-specific

scenarios (situations in Korea) in the

study

3 Missing responses occupied about 3

percent of the data

4 We conducted additional analysis with

analysis of covariance with proficiency

measured by TOEIC Bridge test as cov-

ariate Results were the same

5 This study did not analyze content of

the grounders because content was

fixed provided in the DCT scenarios

REFERENCES

Bardovi-Harlig K and B Hartford 1993

lsquoLearning the rules of academic talk A

longitudinal study of pragmatic changersquo

Studies in Second Language Acquisition 15

279ndash304

Blum-Kulka S J House and G Kasper

1989 Cross-Cultural Pragmatics Requests and

Apologies Ablex

Brown P and S Levinson 1987 Politeness

Cambridge University Press

Golato A 2003 lsquoStudying complement re-

sponses A comparison of DCTs and recordings

of naturally occurring talkrsquo Applied Linguistics

24 90ndash121

DeKeyser R 2007 lsquoSkill acquisition theoryrsquo

in B VanPatten and J Williams (eds)

Theories in Second Language Acquisition An

Introduction Lawrence Erlbaum Associates

Publishers pp 94ndash113

Jeon E -H and T Kaya 2006 lsquoEffects of L2

instruction on interlanguage pragmatic devel-

opmentrsquo in N John and L Ortega (eds)

Synthesizing Research on Language Learning and

Teaching John Benjamins pp 165ndash211

Kasper G 2001 lsquoClassroom research on inter-

language pragmaticsrsquo in K Rose and G Kasper

(eds) Pragmatics in Language Teaching

Cambridge University Press pp 33ndash62

Kim Y 2008 lsquoThe contribution of collaborative

and individual tasks to the acquisition of L2

vocabularyrsquo Modern Language Journal 92

113ndash40

Kim Y 2013 lsquoEffects of pretask modeling on

attention to form and question developmentrsquo

TESOL Quarterly 47 8ndash35

Kim Y in press lsquoThe role of tasks as vehicles for

learning in classroom interactionrsquo in N Markee

(ed) Handbook of Classroom Discourse and

Interaction Wiley-Blackwell

Lee I S Kim S Yoon J Nam S Park

Y Hong and T Orr 2009 Middle School

English 2 Chun-Jae-Gyo-Yuk

Leech G 1983 Principles of Pragmatics Longman

McNamara T and C Roever 2006 Language

Testing The Social Dimension Blackwell

Nassaji H and J Tian 2010 lsquoCollaborative and

individual output tasks and their effects on

learning English phrasal verbsrsquo Language

Teaching Research 14 397ndash419

Nguyen TTM 2013 lsquoInstructional effects on

the acquisition of modifiers in constructive

criticisms by EFL learnersrsquo Language

Awareness 22 76ndash94

Payant C 2012 lsquoLearner-learner interaction

An exploration of the mediating functions of

multilingual learnersrsquo languages in an L3 for-

eign language classroomrsquo Unpublished disser-

tation Georgia State University

Roever C 2009 lsquoTeaching and testing pragmaticsrsquo

in M Long and C Doughty (eds) Handbook of

Language Teaching Wiley-Blackwell pp 560ndash77

Schmidt R 2001 lsquoAttentionrsquo in P Robinson

(ed) Cognition and Second Language Instruction

Cambridge University Press pp 3ndash32

N TAGUCHI AND Y KIM 21

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

Storch N 2007 lsquoInvestigating the merits of pair

work on a text editing task in ESL classesrsquo

Language Teaching Research 2 143ndash59

Swain M 2006 lsquoLanguaging agency and col-

laboration in advanced second language

Proficiencyrsquo in H Byrnes (ed) Advanced

Language Learning The Contribution of Halliday

and Vygotsky Continuum pp 95ndash108

Swain M and S Lapkin 1998 lsquoInteraction

and second language learning Two adolescent

French immersion students working togetherrsquo

Modern Language Journal 82 320ndash37

Swain M and Y Watanabe 2013

lsquoCollaborative dialogue as a source of learningrsquo

in C Chapelle (ed) The Encyclopedia of Applied

Linguistics Wiley Blackwell

Taguchi N 2011 lsquoTeaching pragmatics Trends

and issuesrsquo Annual Review of Applied Linguistics

31 289ndash310

Taguchi N 2012 Context Individual

Differences and Pragmatic Competence

Multilingual Matter

Taguchi N forthcoming lsquoInstructed pragmatics

at a glance Where instructional studies were

are and should be going State-of-the-art art-

iclersquo Language Teaching

Takahashi S 1996 lsquoPragmatic transferabilityrsquo

Studies in Second Language Acquisition 18 189ndash223

Takahashi S 2010 lsquoAssessing learnability in

second language pragmaticsrsquo in A Trosborg

(ed) Handbook of Pragmatics VII Mouton de

Gruyter pp 391ndash421

Takimoto M 2012 lsquoMetapragmatic discussion

in interlanguage pragmaticsrsquo Journal of

Pragmatics 44 1240ndash53

Van den Branden K M Bygate and

J Norris (eds) 2009 Task-Based Language

Teaching A Reader John Benjamins Publishers

22 COLLABORATIVE DIALOGUE AND PRAGMATICS

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from

NOTES ON CONTRIBUTORS

Naoko Taguchi is an associate professor in the Modern Languages Department at

Carnegie Mellon University where she teaches courses in SLA and Japanese language

and culture Her research interests include pragmatics classroom-based research and

English-medium instruction Address for correspondence Naoko Taguchi Carnegie Mellon

University Pittsburgh PA USA lttaguchiandrewcmuedugt

YouJin Kim is an assistant professor in the Department of Applied Linguistics and ESL at

Georgia State University Her primary research interests include instructed second lan-

guage acquisition task-based language teaching and assessment and structural prim-

ing Her research focuses on adolescent and adult English and Korean language

learners Address for correspondence YouJin Kim Georgia State University Atlanta

GA USA ltykim39gsuedugt

N TAGUCHI AND Y KIM 23

by guest on July 30 2014httpapplijoxfordjournalsorg

Dow

nloaded from