computer-mediated negotiated interaction: an …plaza.ufl.edu/tedbow/study4/computer-mediated...

Computer-Mediated NegotiatedInteraction: An Expanded ModelBRYAN SMITHCollege of EducationTexas Tech UniversityPO Box 41071Lubbock, TX 79409-1071Email: [email protected]

This study examines task-based, synchronous computer-mediated communication (CMC)among intermediate-level learners of English. The research specifically explores (a) whetherlearners engage in negotiated interaction when they encounter new lexical items, (b) whethertask type has an effect on the amount of negotiation that transpires, and (c) how thiscomputer-mediated negotiation compares to that noted in the face-to-face literature. Fourteennonnative–nonnative dyads collaboratively completed 4 communicative tasks using ChatNet,a browser-based chat program. Each dyad completed 2 jigsaw and 2 decision-making tasks,which were each “seeded” with 8 target lexical items. The chatscripts reveal that learners doin fact negotiate for meaning in the CMC environment when nonunderstanding occurs.Furthermore, task type was found to have a definite influence on the extent to which learnersengaged in negotiation, but not necessarily in the same way that has been observed in theface-to-face literature. Though the negotiation that occurs in the CMC environment proceedsin ways that are roughly similar to face-to-face negotiation, the observed differences call for anew model of computer-mediated negotiation. This new model is presented as a more accuratetool for describing computer-mediated negotiated interaction than those offered to chartface-to-face negotiation episodes.

AS THE PROLIFERATION OF COMPUTERS INthe language learning classroom continues, it isimportant for language teachers embracing theuse of computer technology to understand thenorms of language use during computer-medi-ated interaction and their potential relationshipto second language acquisition (SLA). The use ofsynchronous computer-mediated communica-tion (CMC) in particular has recently increasedin the communicative language classroomthrough freeware and readily-available Web-based “chat” programs such as AOL Instant Mes-senger and Yahoo Messenger, among countlessothers. In general, CMC appears to be a poten-tially useful tool for language teaching and learn-ing as well as for research into both second lan-guage use and acquisition. Research suggests thatCMC may elicit more (and more equitable)

learner participation (Beauvois, 1992; Kelm,1992; Kern, 1995; Kim, 1998; Warschauer, 1996),as well as better quality language (Chun, 1994;Kelm, 1992; Kern, 1995; Warschauer, 1996) thanthat found in face-to-face interaction. Computer-mediated communication may also help create aless stressful environment for second languagepractice (Chun, 1998). Furthermore, because ofthe “logging” capabilities of most CMC programs,we may capture and readily access this interactionfor both research and pedagogical purposes. In-deed, the computer as a data collection instru-ment seems to be less intrusive in many ways thantraditional procedures for recording face-to-facestudent interaction.

Though the use of well-crafted communicativeactivities, which promote learner-learner interac-tion, is generally considered sound pedagogicalpractice, the theoretical and empirical supportfor the efficacy of such activities for facilitatingSLA is less than conclusive. Nevertheless, it iswidely held that communicative interaction

The Modern Language Journal, 87, i, (2003)0026-7902/02/38–57 $1.50/0©2003 The Modern Language Journal

among nonnative speakers (NNS–NNS), espe-cially that which promotes negotiated interac-tion, is especially facilitative for SLA. Negotiationepisodes have been shown to abound in face-to-face learner interaction (Brock, Crookes, Day, &Long, 1986; Ellis, Tanaka, & Yamazaki, 1994;Gass, 1998; Gass & Varonis, 1985b; Long, 1985;Loschky, 1994; Mackey, 1999; Pica, 1994; Pica,Doughty, & Young, 1986; Pica, Holliday, Lewis,Berducci, & Newman, 1991; Pica, Young, &Doughty, 1987; Sato, 1986), especially whenlearners are engaged in certain types of tasks(Doughty & Pica, 1986; Duff, 1986; Gass &Varonis, 1985a; Long, 1981, 1983; Nobuyoshi &Ellis, 1993; Pica & Doughty, 1985; Pica, Kanagy, &Falodun, 1993). Perhaps seduced by the increas-ing presence and popularity of computers in sec-ond language classrooms, we may be tempted toassume that computer-mediated (negotiated) in-teraction among learners occurs to the same de-gree and in the same fashion as that found in aface-to-face environment. However, such an as-sumption requires a leap of faith, to be sure,given the differences reported between face-to-face and CMC discourse and interactional pat-terns (see Blake, 2000; Fernández-García &Martínez-Arbelaiz, 2002; Pellettieri, 1999; Smith,2001).

Research has found CMC discourse to exhibitfeatures of both oral and written language.Among those characteristics similar to spokenlanguage is the real-time nature of the communi-cation, the ability to provide stress to words andphrases (via italics or bolding), the use of the firstperson, and the clear informality found in CMCdiscourse. Characteristics of CMC resemblingwriting include the lack of intonation, the perma-nent record of the discourse, the lexical density,and the use of punctuation and textual format-ting in messages. Computer-mediated communi-cation possesses many unique characteristics aswell. For example, learners, when communicat-ing in a CMC environment, make use of simpli-fied registers, including shorter sentences, abbre-viations, simplified syntax, the acceptance ofsurface errors, and the use of symbols and emoti-cons to express emotion. Furthermore, openingsand closings in discourse have been reported tobe largely optional in CMC (Murray, 2000).Moreover, turn-taking includes many more over-laps than in face-to-face exchanges. This overlap-ping is largely due to a short time delay (even insynchronous CMC) between the actual initiationof the message and its receipt by the addressee.These overlaps are also due in part to the fact thatonly one message at a time may traverse the CMC

interface. A heightened degree of learner up-take, large amounts of learner self-correction,and clear lexical and syntactic development havealso been observed during computer-mediatednegotiated interaction as has a perceived sense ofcommunicative urgency and frequent, yet incon-sequential misspellings (Smith, 2001).

Indeed, synchronous CMC may provide anideal medium for students to benefit from inter-action primarily because the written nature ofcomputer-based discussions allows a greater op-portunity to attend to and reflect upon the formand content of the message, while retaining theconversational feel and flow as well as the interac-tional nature of verbal discussions. Beauvois(1992) has described such “chatting” as conversa-tion in slow motion because the CMC interfaceslows down the communicative interaction whilelargely retaining its real-time interactive nature.From an interactionist perspective on SLA, this isconsidered one of the most beneficial aspects ofsynchronous CMC—that learners are affordedmore processing time while reading and typingmessages, though the “feel” of the interactionremains similar to that of face-to-face oral interac-tion. From this theoretical perspective, negoti-ated interaction in particular is viewed as benefi-cial for SLA as learners elicit modified input fromone another, are pushed to modify their ownlinguistic output, and receive important feedbackon their target language use, thus potentially fo-cusing their attention on their problematic utter-ances.

Determining the nature of computer-mediatednegotiated interaction and establishing the de-gree to which such negotiation is similar to thatreported in traditional interactionist studies is es-sential if we are to use a theoretical rationalebased on face-to-face interaction to justify the useof similar activities in a CMC environment. Thepresent study was exploratory in nature in that itsought to determine how far computer-mediatednegotiation resembles face-to-face negotiation.To this end, the most widely used model of nego-tiation (Varonis & Gass, 1985) was employed toevaluate these computer-mediated negotiationroutines. This study also attempted to determinewhether the role of task type has a similar effecton computer-mediated NNS–NNS interaction ashas been noted in the face-to-face literature.

ESTABLISHING NEGOTIATEDINTERACTION IN CMC

The potential benefits of meaning negotiationinclude making input more comprehensible

Bryan Smith 39

through input modifications, eliciting pushedoutput, providing feedback, and forcing learnersto focus attention on certain aspects of theirspeech. Evidence suggesting a positive relation-ship between negotiation and SLA continues toemerge in areas of lexical acquisition (Ellis et al.,1994; Smith, 2001), morphosyntax (Loschky,1994), and question formation (Mackey, 1999).Despite the plentiful evidence that shows thatlearners negotiate for meaning in the languagelearning classroom, there are some studies thatquestion whether negotiated interaction is aliveand well at all (Foster, 1998). With regard tocomputer-mediated communication, the modestamount of research to date suggests that learnersnegotiate for meaning in some ways that are simi-lar to the traditional classroom and in other waysthat are quite different (Blake, 2000; Darhower,2002; Fernández-García & Martínez-Arbelaiz,2002; Pellettieri, 1999; Smith, 2001).

RESEARCH QUESTIONS

The present study was designed to answer thefollowing research questions:

1. How do learners negotiate for meaning dur-ing task-based CMC?

2. Does task type affect how learners negotiatefor meaning during CMC?

3. How do these negotiation routines compareto those found in the face-to-face negotiation lit-erature?

METHOD

Participants

Online synchronous chat conversations from14 nonnative-nonnative dyads form the data basefor this study. Participants consisted of two intactclasses of nonmatriculated, intermediate-levelstudents enrolled in intensive English classes at alarge Midwestern university. They ranged in agefrom 19 to 28 (M � 23.36, SD � 3.25) and repre-sented five countries and four languages (Ja-pan/Japanese, Korea/Korean, Taiwan/Chinese,People’s Republic of China/Chinese, andQatar/Arabic).1 Based on a pretreatment back-ground questionnaire, learners were determinedto have similar previous experience working withcomputers, ranging from 1 to 4 years (M � 2.17,SD � .87).

Procedures

The participants met once a week for 5 weeksin a campus computer lab during regularly sched-uled class meetings. All computer lab sessionswere part of the regular class syllabus. During thefirst meeting, the students participated in a train-ing session where they received an introductionto the ChatNet Internet Relay Chat (IRC) pro-gram, which was to be used in the study. Duringthis session, the students were able to practiceinteracting with an unidentified partner online inreal time. ChatNet is a basic IRC program thatallows users to type messages in one window andread messages in another window. It was chosenas the interface because it resembles various freechat software and Web-based programs availabletoday. After this introductory session, the partici-pants were given two practice tasks to completethat corresponded to the two task types to be usedin the data collection. These tasks were shortenedversions of similar tasks used in the treatment.The learners completed four tasks in total overthe duration of the study, two jigsaw tasks and twodecision-making tasks. During each of the sub-sequent meetings (meetings 2 through 5), thelearners completed a short warm-up task wherethey had to log on to the program and chat on-line about the day’s activities or plans for the restof the day with one other student. This warm-upwas followed by the day’s task, which was limitedto 30 minutes. All student discourse was loggedand collected later by the researcher. The com-puter lab was set up in such a way that the learn-ers could not easily view the screens of their class-mates or talk with one another without drawingattention to themselves. Specifically, each com-puter station was outfitted with a privacy guard oneither side of the station. Furthermore, the termi-nals were located either along the outer perime-ter of the classroom facing the wall or in an innercircle facing toward the center of the classroom.

Tasks

Pica et al. (1993) cited two recurrent featurescommon to virtually all discussions of task in theliterature. The first feature is that tasks are ori-ented toward goals. Participants are expected toarrive at an outcome accomplished through theirtalk. The second feature is activity, which suggeststhat the participants take an active role in carry-ing out tasks. Pica et al.’s task typology is perhapsthe most informative of typologies within the in-teractionist framework because the researchersdelineated task type along the two categories—

40 The Modern Language Journal 87 (2003)

goals and activities—in relation to the task’s im-pact on opportunities for learner comprehensionof input, feedback on production, and interlan-guage modification. The two task types chosenfrom this typology for use and comparison in thisstudy were the jigsaw task and the decision-mak-ing task. While keeping to the specifications out-lined in Pica et al. (1993), appropriate and rele-vant original tasks were sculpted for each task type(see Appendixes A and B for examples)—two jig-saw tasks and two decision-making tasks. Thoughthe task design was essentially pedagogical innature (Nunan, 1989), the researcher strove tomaintain a certain degree of real-world authentic-ity in the tasks. These tasks were tested in a pilotstudy with several native and nonnative speakersof English who were consistently able to accom-plish all four tasks in the intended manner.

Jigsaw Task

According to Pica et al. (1993), the jigsaw tasktype should elicit a high degree of negotiation ofmeaning. In the jigsaw task, the participants wereto arrange a series of six pictures in the correctsequence (see Appendix A for example). Basedon an earlier pilot study, a sequence of six pic-tures was determined to be the optimal numberfor computer-mediated dyadic interaction withintermediate level English as a Second Language(ESL) students. Under this structure, each par-ticipant worked with a task sheet (A or B). Theinstructions on sheets A and B were identical; thedifference between the two sheets was that theparticipants had different parts of a six-part pic-torial story. The three pictures held by each stu-dent were arranged in random order and werelabeled A, B, C or D, E, F, respectively. The taskswere designed to reflect a similar “problem-solu-tion” framework, while being moderately humor-ous. It was also important that the learners beable to proceed through the tasks by drawing ontheir existing English language competence. Thetopics sought for each task were neutral in natureand common to the general life experience of theparticipants in the study.

The structure of both jigsaw tasks was essen-tially the same in that they initially showed theprotagonist(s) facing a mild problem, thenshowed them taking action to solve that problem,and culminated in a humorous ending. In aneffort to ensure that negotiation would occur, thetasks were seeded with low-frequency, unknownitems (objects) with each participant receivingfour different target lexical items. For example,in the messy garage task, the pictorial sequence

included objects or items such as a rake, a ther-mos, overalls, and so forth. These lexical items(concrete nouns) were selected based on the re-sults of a pretest that was administered 2 weeksbefore the treatment and that followed themethod reported in Ellis and He (1999). Thoseitems that were least known by the students wereselected for inclusion in the tasks.

The two jigsaw tasks employed in this study fitunambiguously into Pica et al.’s (1993) typologyin terms of the two chief defining features of tasktype, interactional activity and communicationgoal. In each of these tasks each interactant helda different portion of information and suppliedor requested this information as needed to com-plete the task. Each interactant was also requiredto request and supply information because with-out such an exchange, there was no reasonableexpectation that a dyad could put the six picturesin a logical sequential order. The goal of arrang-ing the pictures in the proper sequence of eventswas common to both participants and thereforeconvergent in nature. Finally, great caution wastaken during the preparation of the jigsaw tasksto ensure that only one reasonable solution oroutcome was possible for each task. As men-tioned, the series of pictures was pretested onseveral native speakers as well as on learners ofEnglish in order to validate this assumption. Foreach of the four areas described, learner compre-hension of input, feedback on production, andinterlanguage modification were expected (Picaet al., 1993).

Decision-Making Task

The decision-making task type was chosen tocontrast against the jigsaw task because it liesnear the opposite end of Pica et al.’s (1993) tasktypology and should, thus, elicit a much smalleramount of negotiation. Like the jigsaw tasks,these decision-making tasks attempted to drawon the learners’ life experiences, including theirrecent experience in the United States. Thestructure was similar for both decision-makingtasks. The learners were faced with a situationwhere a larger list of items (8 items) needed tobe culled to a list half its size. Based on the taskscenario, the participants were required to cometo a mutually agreeable joint decision regardingthe appropriateness or usefulness of the targetlexical items. The decision-making tasks con-sisted of version A and version B, with each par-ticipant (dyad half) possessing one or the otherversion. That is, the participants had exactly thesame task, with the exception of four target lexi-

Bryan Smith 41

cal items listed at the bottom of their task sheet(see Appendix B for example).

In the two decision-making tasks, both interac-tants had convergent goals given that they had toselect and agree upon four items from a larger listof eight. In contrast to the jigsaw tasks, though,multiple outcomes were possible. That is, eachdyad could have theoretically come up with aunique, yet perfectly valid, solution to the task.Unlike the jigsaw tasks, which required interac-tants to request and supply information, the deci-sion-making tasks allowed the participants toseek, withhold, or exchange information as theysaw fit. That is, it was technically possible for thelearners to complete these tasks successfully with-out actually sharing information.

ANALYSIS

All turns involving negotiated interaction werecalculated from the chatscripts. As mentioned,the model outlined in Varonis and Gass (1985)was used to identify negotiation routines.Varonis and Gass (1985) defined instances of ne-gotiation as exchanges that begin with an ex-plicit indication of nonunderstanding and thatresult in a temporary “push down” in the con-versation, away from the main line of discourse.This difficulty can be real or perceived by theparticipants and can be caused by any numberof factors. For example, a learner may use thewrong word, which causes the interlocutor aproblem in understanding the intended mes-sage. In contrast, the same learner may correctlyuse a word that is beyond the lexicon of his orher interlocutor. This usage of an unknownword may also result in an indication of nonun-derstanding. Finally, an utterance may be prob-lematic because of task-related issues or ambi-guities, which arise during the co-constructionof the task discourse.

In addition, a ratio of negotiated turns to totalturns was calculated in order to make the datafrom all dyads comparable. It is important to es-tablish a ratio of negotiated turns to total turnsbecause the comparison of the instances of nego-tiation around target lexical items alone may besensitive to the quantity of discourse produced bythe dyads.

Turns

A turn was counted each time there was a trans-fer of the “floor” from one participant to the other.In Excerpt 1,2 there are a total of seven turns.

Excerpt 11. C: F) He is in the garage and has an ax.

C: that’s all i have2. A: I dont know as.

A: ax3. C: you use it,when you cut down trees

C: it’s made of steal, i guess4. A: ok.. I got it5. C: alright6. A: let’s start to make a order7. C: ok

This somewhat conservative approach was takenrather than simply counting each line of text as aturn because the researcher believed it would bepresumptuous to claim to be able to distinguishaccurately when a participant intended a new lineto be a genuinely new turn and when the new linesimply reflected a highly individualized tech-nique of keyboarding. Furthermore, the struc-ture of CMC discourse is often quite different inmany ways from face-to-face interaction. Asidefrom the unique structure of participation and itseffect on the way topics are explored, CMC dis-course often develops in a multilinear and asso-ciative fashion. Indeed, turn-taking in the senseconceived of by Sacks, Schegloff, and Jefferson(1974) is often not present in CMC largely be-cause Sacks et al. assume that all potentialspeakers have access to the same channel(s) ofcommunication at once. Most CMC systems, in-cluding the one used in this study (ChatNet) areone-way in nature meaning that only one interac-tant at a time can use a given channel. Thus,disrupted turn adjacency is the rule rather thanthe exception.

Negotiation Routines

Varonis and Gass (1985) suggested that nego-tiation episodes occur during NNS–NNS interac-tion largely due to a lack of shared background aswell as a “shared incompetence” in the targetlanguage. These routines essentially serve to helpinteractants maintain an equal footing in the con-versation once explicit problems in communi-cation occur and are explicitly acknowledged. Ac-cording to this model, negotiation episodes areresponses to instances of nonunderstanding, asopposed to misunderstandings, and are com-prised of three obligatory phases and one op-tional phase. Figure 1 illustrates this sequence.

Given that the research questions askedwhether learners negotiate for meaning whenproblems in communication arise and sought to


characterize the nature of such negotiation, eachof the four component parts of the negotiationroutines was broken down into predeterminedcategories (see Figure 2) that were based on theexisting interactionist research (e.g., Bremmer,Broeder, Roberts, Simonot, & Vasseur, 1988;Pellettieri, 1999; Pica et al., 1991; Rost & Ross,1991; Varonis & Gass, 1985).

Triggers. A trigger <T> is the catalyst of a negotia-tion routine. Many types of triggers have beennoted in the interactionist literature includinglexical/semantic, structural (morphological/syn-tactic), content- and task-related, discourse, andpragmatic. Research consistently shows that mosttriggers of negotiation routines are lexical in na-ture. However, in an effort to account for non-lexical item negotiation, additional catego-ries—syntactic, discourse, and content—wereused to classify the additional triggers. Lexical trig-gers are those cases where the problematic utter-ance can be clearly linked to a specific lexicalitem. Syntactic triggers are those cases where theproblematic utterance can be clearly attributed toa structural or grammatical construction. Dis-course triggers are related to the general coherenceof the discourse or conversation. For example,noncommunication caused by an inability to ref-erence correctly the antecedent of a pronounduring interaction would be categorized as a dis-course trigger. Content triggers are those instanceswhere the entire content of a previous message is

in some way problematic, including cases whenthe preceding message was vague. That is, whenthe problem could not be attributed to one of theother trigger types listed above, it was classified asa content trigger.

Indicators (Signals). The second part of the ne-gotiation routine is called the indicator orsignal. According to Varonis and Gass (1985),these signals can be explicit or implicit. In thismodel, signals can also take the form of confirma-tion checks or clarification requests that repeatthe problematic part of the previous phrase (thetrigger). Indicators or signals are executed by theinitiator of the negotiation routine. FollowingRost and Ross (1991), each indicator was classi-fied here as global, local, or inferential. Globalstrategies are those in which the respondent indi-cates nonunderstanding in a way that does notidentify the trigger specifically. An example of aglobal indicator is the question (clarification re-quest) “What?” or the statement “I don’t under-stand.” Local strategies are those in which the re-spondent explicitly identifies the trigger orindicates its precise location in the preceding dis-course. Examples of local strategies are “Whatdoes wrench mean?” (clarification request) and“What was his name again?” Local strategies canalso include confirmation checks such as “Do youmean machine?” (after the interlocutor wrote“mascien”). Inferential strategies occur when learn-ers test out hypotheses and in doing so indicate

Note. From “Non-Native/Non-Native Conversations: A Model for Negotiation of Meaning,” by E. Varonis andS. Gass, 1985, Applied Linguistics, 6, p. 74. Copyright 1985 by Applied Linguistics. Reprinted with permission ofOxford University Press.

FIGURE 1Model of Negotiated Interaction

Note. RT � Lexical � Repeat Trigger with Lexical Modification.

FIGURE 2Subcategories of Negotiation Routine Stages

Bryan Smith 43

noncomprehension. An example of this type ofstrategy is when a participant says or writes, “OK,so that means he is tired?” after his interlocutorhas attempted to describe a man that is bored.

Response. The third component of a negotiationroutine is the response <R>. A response, quite sim-ply, is any utterance by the respondent that re-plies to a signal or indicator of nonunderstand-ing. Various types of responses have been notedin the literature, including (a) minimal re-sponses, (b) simply repeating the trigger with orwithout lexical or syntactic modifications, (c) stat-ing an inability to respond, and (d) rephrasing orelaborating (expansion of) the problematic ele-ment. Essentially these types of responses can becategorized as minimal responses, modificationresponses, and elaborative responses.

Based on pilot study data, only the three differ-ent response types elaborated below were antici-pated in the data. A minimal response by definitionprovides little new input to the initiator of thenegotiation routine. This response may consist ofmerely repeating the trigger or simply respond-ing “yes” to the indication of nonunderstanding.Therefore, in this study, any response that pro-vided no new information to the interlocutor andwas a short, one- or two-word response was con-sidered minimal. Repeating the trigger (in mostcases a lexical item) with lexical modification to thesurrounding text is a learner’s attempt to clarifyhis or her intended meaning, but the respondentdoes not address the fundamental problem sig-naled in the indicator phase. In this category, thelength of the response utterance is about thesame as the trigger phrase. The final alternative,rephrasing and elaborating, seems intuitively themost helpful to learners who have signalednonunderstanding. By rephrasing the prior utter-ance, the respondent may better illustrate thenature of the problematic lexical item, and byelaborating on the previous discourse, more con-text may be provided.

Reaction to the Response. The fourth componentof the negotiation routine, the reaction to the re-sponse <RR>, is optional. As Varonis and Gass(1985) noted, this phase serves to signal thatlearners are ready to resume the main line ofdiscourse. This phase normally takes the form ofan explicit statement of understanding such as“OK,” “Good,” or “I understand.” These shortreactions to the response are called minimal re-sponses in the present study. Another possibilitynoted in the literature is metalinguistic in nature.In these types of utterances, learners comment

explicitly on what the cause of the problem hadbeen.

Two new categories emerged from the presentdata. Task appropriate responses <TAR> are “utter-ances” that are contextually relevant to the pre-ceding stretch of discourse and that implicitlyshow a degree of understanding of the targetelement. The second type of reaction to the re-sponse found in the data was testing deductions<TD>. Testing deductions strategies are similarin many ways to inferential indications of nonun-derstanding during the signal phase of negotia-tion and may result in a confirmation or non-confirmation of correctness by the respondent.Testing deductions occurs when, during the re-action to the response, learners, reacting to therecent input provided in the response phase,make certain inferences, testing out their cur-rent state of understanding regarding the origi-nal problematic utterance. The term hypothesistesting, though fitting in some respects, is inten-tionally not used here because it has been tra-ditionally associated with learners’ building of ahypothetical grammar (Corder, 1981), whichseems inappropriate for the current context.

RESULTS AND DISCUSSION

Table 1 shows the raw number of turns, negoti-ated turns, and relative number of turns negoti-ated to total turns for all dyads across all tasks. Inthis way, we are able to determine the relativeamount of negotiation that occurred while thelearners were engaged in task-based CMC. FromTable 1 we can see that the learners were involvedin negotiated interaction about one-third of thetime. This result also supports the finding ofPellettieri (1999), who reported that negotiationroutines accounted for 34% of the total turnsgenerated by all dyads engaged in task-basedCMC in her data. This result suggests that learn-ers, when engaged in CMC tasks designed tofacilitate negotiation, engage in negotiated inter-action in about one-third of their total turns.Therefore, a full two-thirds of their discourse isfocused on collaborative progression toward taskcompletion. It appears, then, that even whentasks are designed to elicit negotiation aroundnew vocabulary, not an excessive amount of theoverall discourse is spent on negotiation, norshould it be. This finding may allay some of theconcerns about tasks that promote too much ne-gotiation (Aston, 1986).

Task type did indeed seem to have an effect onhow much learners negotiated for meaning. Ta-

ble 2 shows the results of a paired groups t testwith percentage of turns negotiated and task typeas the dependent and independent variables, re-spectively. From this table, we see that the learn-ers negotiated a significantly higher percentageof turns when they were engaged in the decision-making tasks than when they worked on the jig-saw tasks. The effect size statistic of 1.37 (Cohen’sd), which measures the magnitude of the treat-ment effect, suggests a large effect.

Table 3 provides a breakdown by task type ofthe target lexical items that triggered negotiatedinteraction among learner dyads. From this tablewe can see that 78% of these items were negoti-ated during the decision-making tasks whereasonly 22% were negotiated during jigsaw tasks.These findings are consistent with the data pre-sented in Tables 1 and 2.

The findings above seem to run contrary to theexpectations outlined in Pica et al.’s (1993) study,which posited that jigsaw tasks should facilitatenegotiation over information gap, problem-solv-ing, decision-making, and opinion exchangetasks. However, when we consider only the non-target lexical item triggers (n � 20), we see thatthe results, though modest in numbers, are con-sistent with the expectations of Pica et al. (1993)as well as those of Robinson (2001). Table 4 showsthe detailed breakdown of all nontarget item trig-gers in the data by trigger type for both task types.From this table we see that lexical items made up60% of all nontarget triggers. Furthermore, thejigsaw tasks elicited more negotiation aroundnontarget items than the decision-making tasks,though the modest numbers preclude further sta-tistical analysis.

Thus the current data are not entirely at oddswith previous research predicting more negotia-tion during jigsaw tasks because there is some evi-dence that jigsaw tasks may elicit more “inciden-tal” negotiation as predicted by earlier studies.However, when target lexical items are infusedinto the task, the scale tips toward the decision-making task type. That is, when performing lexi-cally seeded jigsaw tasks, the participants mayoften relegate the target lexical items to a level ofsecondary importance. This result may be due toa learner perception of these target items as lesssalient for task completion. Though the jigsawtask required an information exchange for com-pletion, it seems that the degree of target itemsaliency elicited by the decision-making tasks maysupercede this task parameter of interaction re-quirement (Pica et al., 1993) when it comes togenerating negotiated interaction around spe-cific target lexical items. Indeed, there may verywell be a task-induced saliency (or nonsaliency inthis case) at work here regarding the seeded lexi-cal items. The notion of saliency has been ex-plored to some degree in regard to morphosyntax(Doughty, 1991; Loschky & Bley-Vroman, 1993;Mackey, Gass, & McDonough, 2000) but not inconnection with lexis. Though the learners wereencouraged to use the lexical items provided inorder to help them complete each task speedily, itseems that these items were viewed as secondaryduring the picture-sequencing tasks ( jigsaw), de-spite the fact that these tasks were designed so thatthe use and understanding of the target lexicalitems would help determine a definitive chrono-logical sequence of events and thus facilitate thecorrect ordering of the picture sequence.

TABLE 1Total Turns and Negotiated Turns during CMC

Task Type Negotiated Turns Turns Total Mean Percentage of Turns Negotiated

Jigsaw 157 676 23%Decision-Making 335 779 44%Total 492 1455 34%

TABLE 2Comparison of Mean Percentage of Negotiated Turns to Total Turns across Task Type

N Sig. EffectTask Type (Dyads) M SD t df (2-tailed) Size

Jigsaw 14 .23 .12 –3.63 13 .003 1.37Decision-Making 14 .44 .18

Bryan Smith 45

The notion of task-induced saliency of lexicalitems is similar in some respects to what Lauferand Hulstijn (2001) have referred to as task-in-duced involvement. The basic assumption inLaufer and Hulstijn’s (2001) study is that theretention of unknown vocabulary words is de-pendent on the degree of involvement in process-ing these words. Though their model specificallyaddresses incidental vocabulary learning, itseems that many of the concepts are appropriatefor task-based, form-focused intentional vocabu-lary learning as well. According to this model,words processed with higher involvement loads willbe retained better than those processed withlower involvement loads. The involvement load isdetermined by the presence or absence of threeinvolvement factors: need, or the drive to complywith the task requirements; search, or the attemptto find the meaning of an unknown target lan-guage word by looking in a dictionary or consult-ing another authority; and evaluation, or the com-paring or combining of a given word with otherwords and meanings. According to this model,teacher-/researcher-designed tasks with a higherinvolvement load will be more effective for vo-cabulary retention than tasks with a lower involve-ment load.

The findings in the present study can be par-tially explained in terms of Laufer and Hulstijn’smodel in that the target lexical items in the jigsawtasks may have had a perceived lower need. Thevarious messages sent back and forth among the

dyads may have been understandable without em-ploying many of the target lexical items. Thisrelatively low need also seems to affect negativelythe degree of search and evaluation necessary.The decision-making tasks yielded quite differentresults, with the bulk of negotiation occurringhere. This same notion of item saliency may helpexplain this occurrence as well. Unlike the par-ticipants performing the sequential orderingtasks, the learners working on the decision-mak-ing tasks seemed to view their list of lexical items(objects) as instrumental for task completion.This perception led to an explicit introduction ofmany of the target lexical items, resulting often-times in nonunderstanding and negotiation.

Establishing the Nature of Task-BasedComputer-Mediated Negotiation

Thus far it has been demonstrated that negoti-ated interaction does, in fact, occur during task-based CMC among learners of English and thattask type seems to influence whether the learnersengage in negotiated interaction in a computer-mediated environment. But what exactly is thenature of this negotiation and how do existingmodels of charting negotiation hold up in anelectronic environment?

The chatscripts examined in this study yieldeda total of 79 explicit indicators of nonunderstand-ing surrounding target lexical items. Table 5 pre-sents the various stages of negotiation reflected inthe chatscripts. The raw number of instances ofeach stage, as well as the relative percentage ofnegotiation episodes this number represents, ispresented.

Table 5 shows that overall, close to 94% of allsignals of nonunderstanding were followedthrough by a complete negotiation routine,based on Varonis and Gass’s (1985) model. Thatis, they consisted of at least a trigger, an indicator,and a response. Also, 82% of all negotiation rou-tines begun around target lexical items eventuallyculminated in a reaction to the response, theoptional fourth part of Varonis and Gass’s model.

TABLE 3Breakdown of Target Lexical Items Negotiated byTask Type

Target PercentageTask Type Item Triggers of Total

Jigsaw 15 22%Decision-Making 54 78%(Total) 69 100%

Note. All lexical triggers.

TABLE 4Breakdown of All Nontarget Item Triggers by Task Type (n = 20)

Trigger Type Jigsaw Decision-Making Total %

Lexical 7 5 12 60%Discourse 3 1 4 20%Content 2 2 4 20%(Total) 12 8 20 100%


In contrast, only 11% of negotiation routines in-itiated end at the response phase. It seems, then,that in a CMC environment learners feel highlycompelled to bring the routine to some explicitclosure, perhaps even more so than during face-to-face interaction. Though the present author isunaware of any NNS–NNS face-to-face data thatwould allow for a direct test of this assertion,Foster (1998) and Pica, Holliday, Lewis, and Mor-genthaler (1989) permit us to make some tenta-tive inferences in this regard. Foster’s (1998)study of task-based classroom interaction amongNNSs suggested that less than 23% of negotiationmoves are followed by modified responses. Like-wise, Pica et al. (1989), in a study of nativespeaker (NS)–NNS interaction, found that NNSinterlocutors made modified responses to NS-initiated negotiation moves about 35% of thetime. If we assume that an optional reaction tothe response will occur less often than the re-sponse itself, then we may reasonably assume thatthe percentage of negotiation sequences thatreach this optional phase would fall somewherebelow 23% and 35% respectively in the studies byFoster (1998) and Pica et al. (1989), discussedabove. In perhaps the only other CMC study toaddress this issue explicitly, Fidalgo-Eick’s (2001)exploration of NS–NNS interaction found thatlearners completed all four stages of the negotia-tion sequence 70% of the time.

One possible explanation for the high occur-rence of reaction to the response phases in com-puter-mediated negotiated interaction may bethat CMC removes, or at least reduces, many ofthe para- and nonlinguistic aspects of face-to-facespeech that facilitate verbal communication.Thus in CMC, a certain degree of support isstripped away, concentrating the entire burden ofcommunication on written characters. As a result,

a more explicit marking of understanding andnonunderstanding, as well as turn boundaries, isrequired in CMC than in face-to-face interaction.

Components of the Negotiation Routine

Although various trigger types have been docu-mented in both traditional and CMC interaction-ist research, only lexical triggers are addressedfrom this point forward because the tasks in thisstudy were seeded with largely unknown lexicalitems in an effort to elicit negotiated interaction.Table 6 shows the breakdown of each phase ofnegotiation based on the chatscript data for tar-get lexical items. The response and reaction tothe response phases take into account those ex-tended negotiation routines that resulted in twoand, in some cases, three passes through the re-sponse and reaction to the response phases.

It is interesting to note the clear tendency bythe learners to indicate nonunderstanding (of alexical trigger) through a local indicator, support-ing recent research by Fernández-García andMartínez-Arbelaiz (2002). The reason for thistendency is likely related to the need for thelearners to be very explicit during communica-tion in a CMC environment due to the absence ofnonlinguistic cues. A global indicator couldprove ambiguous in relation to the surroundingtext, which occupies the screen. The learnersoverwhelmingly opted for rephrasal/ elaborationin the response phase. A minimal response in theresponse phase, by definition, provides little newinput to the initiator of the negotiation routine.Similarly, by simply repeating the trigger, in thiscase a lexical item, and lexically modifying thesurrounding lexical items, the respondent maynot address the fundamental problem signaled inthe indicator phase. By rephrasing the prior ut-terance, the respondent may illustrate the natureof the problematic lexical item better, and byelaborating on the previous discourse, may pro-vide more context. By contrast, little more than aminimal response is required in the reaction tothe response phase because the function of thisoptional phase is, in large part, simply to acknow-ledge understanding. This phase is essentially asignal that the initiator is ready to “pop back up”to the main topic of the conversation. It is, then,not surprising that learners (initiators) over-whelmingly opt for a minimal response in thefinal phase of the negotiation routine. Thoughmetalinguistic talk may prove helpful in uncover-ing the root of the problem, it is not essential fortask completion and may divert valuable time

TABLE 5Stages of Negotiation Routines Completed byDyads

Total Number (RelativeNegotiation Percentages) of RoutinesSequence Terminating at This Stage

T → I 5 (6%)T → I → R 9 (11%)T → I → R → RR 65 (82%)Total 79 (100%)

Note. T � Trigger; I � Indicator; R � Response;RR � Reaction to the Response.

Bryan Smith 47

away from the task. Task-appropriate responsesand testing deductions strategies were also pre-sent in the data. These strategies show a height-ened degree of student involvement, even in thisoptional phase of negotiation, and will be dis-cussed in detail below.

An Expanded Model of Negotiation of Meaning forTask-Based CMC

In broad terms, the negotiation patterns in thisCMC study were similar to those observed in face-to-face communication, fitting loosely into theVaronis and Gass (1985) model. It appears,though, that this model is insufficient to dealadequately with negotiation in a CMC environ-ment in a detailed manner and must be ex-panded. The inadequacy of the Varonis and Gassmodel when applied to CMC has also been notedelsewhere. Pellettieri (1999), for example, foundthat the negotiation routines among learners ofSpanish could occur in slightly different ways.Specifically, she argued that the Varonis and Gassmodel does not account for the use of (spontane-ous) appeals for assistance.

The data from the current study suggest thatany CMC model of negotiation must allow for adelay, sometimes a long delay, between the initialtrigger <T> and the indicator. This allowance fordelay is needed largely because of the lack ofstrict turn adjacency in CMC compared to thatfound in face-to-face communication. The lack ofturn adjacency causes many triggers to go unan-swered initially and often results in episodes get-

ting sidetracked. However, as mentioned above, itis rare for a trigger to be ignored permanently.The proposed model allows for the regular occur-rence of split negotiation routines, which often be-gin with a trigger and are followed by an indicatorof nonunderstanding, whose response may onlyoccur after a second (or third) repeat indicatorsome time later in the discourse. In these cases,the trigger remains the same, the nonacknowl-edged signal(s) or indicator(s) are referred to asIndicator 1 <Ii> and Indicator 2 <Iii> respectively,with the eventual response signified by <R>. Thefollowing excerpt shows an example of a splitnegotiation routine. This episode differs from anormal negotiation routine in that the initial in-dicator of nonunderstanding of the word axcomes long after the trigger. The permanence ofthe CMC discourse in the form of a scrolling chatlog allows for such split routines, which were verycommon in the present data.

Excerpt 2Example of Split Negotiation Routine

J: There are Ax, Rake, and so on.[11 lines of text]

<T> J: He hold ax in a clean garage, andeverything is in order in everywhere.

[43 lines of text] B: ax mean is hammer?

J: noJ: That’s differentB: what is it?

<R> J: Ax is used to cut treeJ: or wood

TABLE 6Subcategories and Relative Occurrence of Each Element of Negotiation Phase

Negotiation Element Subcategory Percentage of Total

Trigger Lexical 100%

Indicator Global 18%Local 76%Inferential 5%

Response Minimal 1%Repeat Trigger with Lexical

Modification 3%Rephrasing/Elaboration 96%

Reaction to Response Minimal 70%Metalinguistic Talk 1%Task Appropriate Response 4%Testing Deductions 24%

In addition, careful examination of the chat-scripts suggests that the reaction to the responsephase of computer-mediated negotiation rou-tines appears to be more dynamic than previouslyreported. Furthermore, the data reveal a strongtendency for learners to carry on negotiationroutines well past this reaction to the responsestage. Two additional phases were characteristicof the CMC negotiation routines in the presentstudy. These will be referred to as the confirmation<C> and reconfirmation <RC> phases.

The optional reaction to the response <RR>phase largely serves the purpose of closing outthe negotiation routine through an overt indica-tion of understanding. It is at this point that theinteractants can return to the main trajectory ofthe conversation. Indeed, the present data haveshown that in 82% of the negotiation routinesthe learners did in fact follow through with areaction to the response. Clearly, though, not allreactions to the response bring the routine to aclean and appropriate finale. That is, as is thecase in face-to-face interaction, in CMC we mayexpect the occasional reaction to the response toserve as an indicator of continued lack ofunderstanding or incomplete understanding(Varonis & Gass, 1985). Excerpt 3 illustrates thispoint.

Excerpt 3<T> 1. C: and he hold the

dust-pan 2. O: I don’t under-

stand what it is?3. O: how look like?

<R> 4. C: dustpad os collect-ing the trash.

5. C: is6. C: dustpan is collect-

ing the trash<RR–><TD> 7. O: m . . . it . . . looks

like finger?<C–> 8. C: no

9. O: ?<R²> 10. C: it was invented

before baccom.<RR²+> 11. O: ok . . .<C²> Reaffirmation 12. C: old people used

the Dustpan

In Excerpt 3, we see that in line 7, the student,O, makes an implicit showing of her incompleteunderstanding of the target item dustpan. Thisstrategy in line 7, aside from filling the reactionto the response slot in the negotiation routine,

serves to test a working hypothesis of what a dust-pan is, a step labeled here as testing deductions<TD>. This testing deductions strategy employedby O also indicates or signals to C that she has anincomplete understanding of the target item,hence the negative reaction to the response<RR–> categorization. Reactions to the responsethat indicate a proper understanding (or those inwhich there is no evidence of nonunderstanding)of the negotiated item(s) are represented by<RR+> in the new model. This is an importantand often overlooked distinction to make be-cause it directly influences the subsequent dis-course. What follows is a model of computer-me-diated negotiated interaction (see Figure 3). Thismodel is essentially based on that proposed byVaronis and Gass (1985), but it expands the for-mer model in order to incorporate the patternsobserved during computer-mediated negotiatedinteraction.

As mentioned, a reaction to the response canbe either positive or negative. That is, it can indi-cate understanding or continued non- or partialunderstanding. In either case, the reaction to theresponse can be either explicit or implicit. Anexample of an explicit positive reaction to theresponse is “OK” or “I understand.” Similarly, anexplicit negative reaction to the response is some-thing like “I don’t understand” or “Can you ex-plain more?” In the latter case, the flow of thenegotiation routine tends to jump immediatelyback to a second round of the response phase<R²>, where the respondent provides more infor-mation. Alternatively, initiators may use an im-plicit reaction to the response. In the currentdata, there were two types of implicit reactions tothe response. These consisted of testing deduc-tions and task appropriate responses <TAR>. Thetesting deductions strategy was found in bothpositive and negative reaction to the responsephases, but task appropriate responses were pre-sent only in the former.

Testing deductions occurs in the reaction tothe response phase of the negotiation routinewhen learners (initiators) believe they have someidea regarding the nature of the element undernegotiation. This idea may range from a fairlycertain notion of the target element to a vague“stab in the dark” based on the available inputprovided by the interlocutor. Most important, thisstrategy shows a heightened degree of learnerinvolvement that the minimal response andmetalinguistic talk lack. It is quite easy simply toacknowledge one’s understanding with a simple“OK,” as witnessed in the high percentage of

Bryan Smith 49

minimal reactions to the response, but it is quiteanother matter to insist on an explicit confirma-tion of one’s hunch, regardless of how strong, byopting for the testing deductions strategy.

In contrast, task appropriate responses are “ut-terances” that are contextually relevant to thepreceding stretch of discourse and that implicitlyshow a degree of understanding (or, theoretically,nonunderstanding, though there were no casesin the present data) of the target element. Exam-ples of the testing deductions and task appropri-ate response strategies are listed in Excerpts 4and 5.

Excerpt 4Testing Deductions

 P: ok what is razor<R> C: Razor? This is very useful

for guys.<RR–> Explicit P: can describ it more<R²> C: If the guy want to cut his

hair, he can cut use Razor.C: Most of guys use it in the

morning.<TD+> <RR+> P: you mean for shaving

C: That’s right!P: ok . . .

Note. Adapted and expanded from Varonis and Gass (1985).

FIGURE 3Model of Computer-Mediated Negotiated Interaction

Excerpt 5Task Appropriate Response

<T> C: if u like the tree, you needthe chainsaw.

 O: what is chainsaw?<R> C: chainsaw is cutting the

tree.<TAR><RR+> O: i hope to protect tree

C: ok . . .

Testing deductions strategies may also be char-acterized as negative if they implicitly show a con-tinued lack of full understanding on the part ofthe initiator. Of course they are only negative inthe sense that they elicit a negative confirmation<C–>. By employing a testing deductions strategy,the initiator grapples with a certain degree ofuncertainty but puts forth his or her best guess.In the case of a faulty testing deduction <TD–>,the respondent will normally respond with anexplicit nonconfirmation (such as using the wordno) followed by a return to the response phase.Occasionally, however, the respondent abandonsthe routine at this juncture. An example of theformer follows in Excerpt 6 below.3

Excerpt 6 A: what is comb?<R> B: after shower, I get my hair

straight by this<RR–><TD–> A: it’s like hair drier?<C–> B: no

A: oh, I see<R²> B: like fish bone<RR²+> A: ok

A: ^ ^

The confirmation phase (optional) of the ne-gotiation routine is where the respondent eitherconfirms <C+> or disconfirms the degree of un-derstanding by the initiator based on the latter’sreaction to the response. As mentioned above, inthe case of a negative confirmation, the respon-dent reinitiates the response phase with furtherinput, or in very rare cases, may simply abandonthe negotiation routine. In contrast, positive con-firmation affords three possibilities for the re-spondent, simple confirmation, reaffirmation, andcomprehension check. Simple confirmation consistsof a “minimal” response of some sort such as“OK,” “Good,” or “Right.” It can also take theform of praise such as “Good job!” or “Great!”Alternatively, respondents can opt for reaffirma-tion whereby, in addition to a minimal confirma-tion, they provide a bit more information to their

interlocutor. In these cases, it often seems thatthere is some level of doubt by the respondent asto whether the initiator has fully grasped the ne-gotiated element or not. Finally, a simple compre-hension check may occur largely, it seems, for thesame reasons as the reaffirmation. Examples ofeach are listed in Excerpts 7 through 11.

Excerpt 7Simple Confirmation

<R> C: . . . when you open wine bottle orsomething like that, you use it

<RR+> A: A . . . Ok!<C+> C: ok

Excerpt 8Simple Confirmation

<R> J: . . . and have a red ribbon on thebottom of the green circle

<RR+> B: I got it<C+> J: Good job, B.<RC> B: Thanks . . .

Excerpt 9Reaffirmation

<R> C: corkboard is similar blackboardC: do u understand?

<RR+> E: I see<C+> C: but corkboard have a pin

Excerpt 10Reaffirmation

 O: what is bongos?<R> C: bongos is similar to drum<RR+><TD+> O: it is play music

O: oh,,,,<C+> C: but it is traditional drum

Excerpt 11Comprehension Check

 B: what is razor? can you explain?A: razor is . . .

<R> A: when you want to cut your chinhair, you use it.

A: it’s kind of knife.<RR+> B: I see<C+> A: got it?<RC> B: ok

The final phase in this model is the optionalreconfirmation phase. The reconfirmation by theinitiator follows the respondent’s confirmationand is essentially the same as the positive explicit(minimal) reaction to the response. It normallyconsists of single words like “OK,” “Good,”“Right,” and “Yes,” or, if the reconfirmation fol-

Bryan Smith 51

lows, an instance of praise in the simple confirma-tion, “Thanks.” Like the confirmation phase, thereconfirmation is very common in task-basedCMC because of the apparent demand for ex-plicit acknowledgments of the understanding/nonunderstanding that CMC interaction elicits.Thus, rather than simply acknowledging under-standing of the negotiated element like the ex-plicit (minimal), positive reaction to the re-sponse, the reconfirmation serves as a definitivesignal that the negotiation detour is now over andthat the conversation (task completion in thiscase) may resume. Examples of reconfirmationsare found in Excerpts 8 and 11 above.

CONCLUSION

Results from this research show that learnersdo negotiate meaning when problems in commu-nication arise during task-based CMC. Indeed,one-third of the total turns were spent negotiat-ing. Given that the tasks employed in this studywere seeded with new lexical items, it is no sur-prise that most of the negotiated interaction wasaround these items. In addition, those negotia-tion routines around nontarget items were mostoften triggered by lexical difficulty as well, whichconfirms previous interactionist research (Blake,2000; Brock et al., 1986; Fernández-García &Martínez-Arbelaiz, 2002; Pellettieri, 1999; Pica,1994; Sato, 1986). The data also suggest that tasktype does indeed influence the amount of nego-tiation that learners engage in during task-basedCMC. This influence was evident in the signifi-cantly higher number of negotiated turns foundin the decision-making tasks as well as the highernumber of target lexical items negotiated in thesesame tasks. There was some evidence that jigsawtasks may elicit more incidental negotiation, aspredicted by Pica et al. (1993), but when targetlexical items are infused into the task, decision-making tasks yield more negotiation sequencesthan jigsaw tasks. The notion of task-induced sa-liency was offered as a possible explanation forthis finding. However, more research specificallyaddressing this possibility is needed before anyfirm conclusions may be drawn.

The results of this study are also consistent withexisting face-to-face and CMC research in that thelearners were found to use local indicators ofnonunderstanding most often. Responses tothese indicators were comprised mostly of re-phrasals or elaboration, thus confirming someprevious research (Pellettieri, 1999) while contra-dicting other studies (Pica, 1988a, 1988b, 1992;

Pica, et al., 1989; Pica et al., 1991). These re-sponses were normally followed by a minimal re-action to the response or a testing deductionsstrategy. Presence of the latter strategy indicates aheightened degree of active involvement, which Iargue here is facilitative for SLA, though it neednot necessarily be limited to explicit instances oftesting deductions or manifest itself in any ob-servable way, a point supported by Pica (1992),Reiss (1985), and Slimani (1989). Finally, thoughthe most widely espoused model for charting ne-gotiation routines (Varonis & Gass, 1985) wasfound to be largely applicable to CMC, the pres-ent data require an expansion of this model inorder to incorporate better the observed featuresof negotiation episodes during task-based CMC.This new model of computer-mediated negoti-ated interaction is presented as a more accurateinstrument for charting negotiation routines in aCMC environment.

ACKNOWLEDGMENTS

I wish to thank Robert Ariew, Sue Gass, Jun Liu,and Mary Wildner-Bassett for comments on vari-ous aspects of this work. I would also like to thankthe four anonymous Modern Language Journal re-viewers whose insightful comments certainlyhelped strengthen this article. Finally, manythanks are also due the participants in this study.

NOTES

1 These categorizations are taken directly from par-ticipant-completed background questionnaires.

2 Excerpts are reproduced exactly as they appeared inthe chat. No spelling or other errors have been cor-rected.

3 ^ ^ is an emoticon reported by various Asian stu-dents to signify smiling.

REFERENCES

Aston, G. (1986). Trouble-shooting in interaction withlearners: The more the merrier? Applied Linguis-tics, 7, 128–143.

Beauvois, M. H. (1992). Computer-assisted classroomdiscussion in the foreign language classroom:Conversation in slow motion. Foreign Language An-nals, 25, 455–464.

Blake, R. (2000). Computer mediated communication:A window on L2 Spanish interlanguage. LanguageLearning and Technology 4(1), 120–136.


Bremer, K., Broeder, P., Roberts, C., Simonot, M., &Vasseur, M. (Eds.). (1988). Ways of achieving under-standing: Communicating to learn in a L2. Stras-bourg, France: European Science Foundation.

Brock, C., Crookes, G., Day, R., & Long, M. (1986). Thedifferential effects of corrective feedback in nativespeaker conversations. In R. Day (Ed.), Talking tolearn: Conversations in second language acquisition(pp. 229–236). Cambridge, MA: Newbury House.

Chun, D. M. (1994). Using computer networking tofacilitate the acquisition of interactive compe-tence. System, 22(1), 17–31.

Chun, D. M. (1998). Using computer-assisted class dis-cussion to facilitate the acquisition of interactivecompetence. In J. Swaffar, S. Romano, P. Markley,& K. Arens (Eds.), Language learning online: Theoryand practice in the ESL and L2 computer classroom(pp. 57–80). Austin, TX: Labyrinth Publications.

Corder, S. P. (1981). Error analysis and interlanguage. Ox-ford: Oxford University Press.

Darhower, M. (2002). Instructional features of synchro-nous computer-mediated communication in theL2 class: A sociocultural case study. CALICO Jour-nal, 19, 249–277.

Doughty, C. (1991). Second language instruction doesmake a difference: Evidence from an empiricalstudy of SL relativization. Studies in Second Lan-guage Acquisition 13, 431–469.

Doughty, C., & Pica, T. (1986). Information gap tasks:An aid to second language acquisition? TESOLQuarterly, 20, 305–325.

Duff, P. (1986). Another look at interlanguage talk: Tak-ing task to task. In R. Day (Ed.), Talking to learn:Conversation in second language acquisition (pp.121–145). Rowley, MA: Newbury House.

Ellis, R., & He, X. (1999). The roles of modified inputand output in the incidental acquisition of wordmeanings. Studies in Second Language Acquisition,21, 285– 301.

Ellis, R., Tanaka, Y., & Yamazaki, A. (1994). Classroominteraction, comprehension, and the acquisitionof L2 word meanings. Language Learning, 44,449–491.

Fernández-García, M., & Martínez-Arbelaiz, A. (2002).Negotiation of meaning in non-nativespeaker–non-native speaker synchronous discus-sions. CALICO Journal, 19, 279–294.

Fidalgo-Eick, M. (2001, February). Synchronous on-linenegotiation of meaning by intermediate learners of Span-ish. Paper presented at the annual conference ofthe American Association for Applied Linguistics,St. Louis, MO.

Foster, P. (1998). A classroom perspective on the nego-tiation of meaning. Applied Linguistics, 19, 1–23.

Gass, S. (1998). Apples and oranges: Or, why apples arenot oranges and don’t need to be. Modern Lan-guage Journal, 82, 83–90.

Gass, S., & Varonis, E. (1985a). Task variation and non-native/nonnative negotiation of meaning. In S.Gass & C. Madden (Eds.), Input and second lan-

guage acquisition (pp. 149–161). Rowley, MA: New-bury House.

Gass, S., & Varonis, E. (1985b). Variation in nativespeaker speech modification to non-native speak-ers. Studies in Second Language Acquisition 7, 35–57.

Kelm, O. (1992). The use of synchronous computernetworks in second language instruction: A pre-liminary report. Foreign Language Annals, 25,441–454.

Kern, R. (1995). Restructuring classroom interactionwith networked computers: Effects on quantityand characteristics of language production. Mod-ern Language Journal, 79, 457–476.

Kim, Y. (1998). The effect of a networked computer-mediateddiscussion on subsequent oral discussion in the ESLclassroom. Unpublished doctoral dissertation, Uni-versity of Texas at Austin.

Laufer, B., & Hulstijn, J. (2001). Incidental vocabularyacquisition in a second language: The construct oftask-induced involvement. Applied Linguistics, 22,1–26.

Long, M. (1983). Native speaker/non-native speakerconversations in the second language classroom.In M. A. Clarke & J. Handscombe (Eds.), OnTESOL ‘82 (pp. 207–225). Washington, DC:TESOL.

Long, M. (1985). Input and second language acquisi-tion theory. In S. Gass & C. Madden (Eds.), Inputand second language acquisition (pp. 377–393).Rowley, MA: Newbury House.

Loschky, L. (1994). Comprehensible input and secondlanguage acquisition: What is the relationship?Studies in Second Language Acquisition, 19, 37–66.

Loschky, L., & Bley-Vroman, R. (1993). Grammar andtask-based methodology. In G. Crookes & S. Gass(Eds.), Tasks and language learning: Integrating the-ory and practice (pp. 123–167). Bristol, PA: Multi-lingual Matters.

Mackey, A. (1999). Input, interaction, and second lan-guage development: An empirical study of input,interaction, and second language development.Studies in Second Language Acquisition, 21, 557–588.

Mackey, A., Gass, S., & McDonough, K. (2000). How dolearners perceive interactional feedback? Studiesin Second Language Acquisition, 22, 471–497.

Murray, D. E. (2000). Protean communication: The lan-guage of computer-mediated communication.TESOL Quarterly, 34, 397–421.

Nobuyoshi, J., & Ellis, R. (1993). Focused communica-tion tasks and second language acquisition. ELTJournal, 47, 203–210.

Nunan, D. (1989). Toward a collaborative approach tocurriculum development: A case study. TESOLQuarterly, 23, 9–25.

Pellettieri, J. (1999). Why-Talk?: Investigating the role oftask-based interaction through synchronous network-based communication among classroom learners ofSpanish. Unpublished doctoral dissertation, Uni-versity of California at Davis.

Pica, T. (1988a). Interlanguage adjustments as an out-

Bryan Smith 53

come of NS–NNS negotiated interaction. Lan-guage Learning, 38, 45–73.

Pica, T. (1988b). Negotiated interaction as an aid tolearner output. Language Learning, 37, 471–493.

Pica, T. (1992). The textual outcomes of nativespeaker–non-native speaker negotiation: What dothey reveal about second language learning? In C.J. Kramsch & S. McConnell-Ginet (Eds.), Text andcontext: Cross-disciplinary perspectives on languagestudy (pp. 198–237). Rowley, MA: Newbury House.

Pica, T. (1994). Research on negotiation: What does itreveal about second-language learning condi-tions, processes, and outcomes? Language Learn-ing, 44, 493–527.

Pica, T., & Doughty, C. (1985). The role of group workin classroom second language acquisition. Studiesin Second Language Acquisition, 7, 233–248.

Pica, T., Doughty, C., & Young, R. (1986). Making inputcomprehensible: Do interactional modificationshelp? I.T.L. Review of Applied Linguistics, 72, 1–25.

Pica, T., Holliday, L., Lewis, N., Berducci, D., & New-man, J. (1991). Language learning through inter-action: What role does gender play? Studies in Sec-ond Language Acquisition, 13, 343–372.

Pica, T., Holliday, L., Lewis, N., & Morgenthaler, L.(1989). Comprehensible output as an outcome oflinguistic demands on the learner. Studies in SecondLanguage Acquisition, 11, 63–90.

Pica, T., Kanagy, R., & Falodun, J. (1993). Choosing andusing communication tasks for second languageresearch and instruction. In G. Crookes & S. Gass(Eds.), Tasks and second language learning (pp.9–34). Clevedon, UK: Multilingual Matters.

Pica, T., Young, R., & Doughty, C. (1987). The impact ofinteraction on comprehension. TESOL Quarterly,21, 737–758.

Reiss, M. (1985). The good language learner: Anotherlook. Canadian Modern Language Review, 41,511–523.

Robinson, P. (2001). Task complexity, task difficulty,and task production: Exploring interactions in acomponential framework. Applied Linguistics, 22,27–57.

Rost, M., & Ross, S. (1991). Learner use of strategies ininteraction: Typology and teachability. LanguageLearning, 41, 231–249.

Sacks, H. E., Schegloff, E., & Jefferson, G. (1974). Asimplest systematics for the organization of turn-taking for conversation. Language, 50, 696–735.

Sato, C. (1986). Conversation and interlanguage devel-opment: Rethinking the connection. In R. Day(Ed.), Talking to learn: Conversation in second lan-guage acquisition (pp. 2–45). Rowley, MA: NewburyHouse.

Slimani, A. (1989). The role of topicalization in class-room language learning. System, 17, 223–234.

Smith, B. (2001). Taking students to task: Task-based com-puter-mediated communication and negotiated interac-tion in the ESL classroom. Unpublished doctoral dis-sertation, University of Arizona, Tucson.

Varonis, E. M., & Gass, S. (1985). Non-native/non-nativeconversations: A model for negotiation of mean-ing. Applied Linguistics, 6, 71–90.

Warschauer, M. (1996). Comparing face-to-face andelectronic communication in the second languageclassroom. CALICO Journal, 13, 7–25.


APPENDIX AJigsaw Task

Student A Messy Garage

Part 1: Look at the series of pictures about a messy garage. You have three scenes (pictures) and your partner hasthree different scenes. Together with your partner put the scenes in the correct order. To do this, you will need to describeeach of your scenes to your partner since he/she cannot see your pictures. You may use the words below to help youdescribe your pictures. Your partner will do the same for you.

The scenes are marked A, B, C, D, E, F. When you finish, please type the correct order. For example – “The correctorder is C, B, F, A, D, E”

MAKE SURE YOU HAVE THE CORRECT ORDER!!!!!

Part 2: When you are SURE that you have found the correct order discuss the following question with your partner:What chores (jobs) around the house do/did you have? Do parents expect their children to do jobs around the houseto help out? Is there a difference in the KINDS of chores boys and girls are expected to do while living at home?Do/did you and your partner have similar experiences? If not, what are the differences?

When you are finished raise your hand!

Tricycle Snow shovel Broom Thermos

Student A – Picture Sequence

Bryan Smith 55

Jigsaw TaskStudent B Messy Garage

Part 1: Look at the series of pictures about a messy garage. You have three scenes (pictures) and your partner hasthree different scenes. Together with your partner put the scenes in the correct order. To do this, you will need to describeeach of your scenes to your partner since he/she cannot see your pictures. You may use the words below to help youdescribe your pictures. Your partner will do the same for you.

The scenes are marked A, B, C, D, E, F. When you finish, please type the correct order. For example – “The correctorder is C, B, F, A, D, E”

MAKE SURE YOU HAVE THE CORRECT ORDER!!!!!

Part 2: When you are SURE that you have found the correct order discuss the following question with your partner:What chores (jobs) around the house do/did you have? Do parents expect their children to do jobs around the houseto help out? Is there a difference in the KINDS of chores boys and girls are expected to do while living at home?Do/did you and your partner have similar experiences? If not, what are the differences?

When you are finished raise your hand!

Ax Rake Dustpan Overalls

Student B - Picture Sequence


APPENDIX BDecision-Making Task

Student A Shopping for a Gift

Part 1: You and your roommate/friend are trying to decide on some gifts for your home stay family here in the UnitedStates. Your host family has four (4) members; Mr. Jones (father), Mrs. Jones (mother), Billy Jones (son 15 years old),and Mary Jones (daughter 14 years old).

Below are some items you have noticed while shopping at the Mall, which may make good presents. Yourroommate/friend has been shopping at the Mall and has also seen some (different) things that he/she thinks mightmake good presents. Since the presents will be from both of you, you must decide together on one present for eachfamily member (four total).

Razor Corkboard Wreath Corkscrew

Part 2: After you have decided on the four gifts you will buy, discuss gift-giving customs in your countries! Is there anydifference in gift giving practices between your country and your chat partner’s country? If not, or if you come fromthe same country, discuss similarities or differences you have noticed in gift-giving practices between your countryand the United States. When you are finished, raise your hand!

Student B Shopping for a Gift

Part 1: You and your roommate/friend are trying to decide on some gifts for your home stay family here in the UnitedStates. Your host family has four (4) members; Mr. Jones (father), Mrs. Jones (mother), Billy Jones (son 15 years old),and Mary Jones (daughter 14 years old).

Below are some items you have noticed while shopping at the Meridian Mall, which may make good presents. Yourroommate/friend has been shopping at the Lansing Mall and has also seen some (different) things that he/she thinksmight make good presents. Since the presents will be from both of you, you must decide together on one present foreach family member (four total).

Bouquet Extension cord Magnifying glass Comb

Part 2: After you have decided on the four gifts you will buy, discuss gift-giving customs in your countries! Is there anydifference in gift giving practices between your country and your chat partner’s country? If not, or if you come fromthe same country, discuss similarities or differences you have noticed in gift-giving practices between your countryand the United States. When you are finished, raise your hand!

Bryan Smith 57

computer-mediated negotiated interaction: an …plaza.ufl.edu/tedbow/study4/computer-mediated...

Documents