Comprehensible Output by Ali Sheadeh

49
Non-Native Speakers’ Production of Modified Comprehensible Output and Second Language Learning Ali Shehadeh University of Aleppo King Saud University This study investigated the ability of NNSs to modify their output toward comprehensibility in the contexts of NS-NNS and NNS-NNS interactions and the degree to which such modified comprehensible output (MCO) was other- or self-initiated. Picture-dictation and opinion- exchange tasks were used to collect data from 8 NSs and 24 NNSs of English representing 13 different L1 back- grounds. The 2 tasks were performed in pairs (NS-NNS and NNS-NNS) and were audiotaped. The results showed that most repairs were self-initiated and that NNS-NNS interactions produced more other-initiations and other- initiated MCOs on the picture-dictation task. The frequen- cies of these MCOs support the importance of modification toward comprehensible output as a process of second lan- guage acquisition. Language Learning 49:4, December 1999, pp. 627–675 627 Ali Shehadeh, Department of English Language, University of Aleppo, Syria (currently at the College of Languages and Translation, King Saud Univer- sity, Saudi Arabia). I acknowledge the help I received from the editors of Language Learning and a number of anonymous reviewers, and I would like to thank them all for their advice and their many helpful suggestions and comments. Correspondence concerning this article may be sent to Ali Shehadeh, College of Languages and Translation, King Saud University, P.O. Box 87907, Riyadh 11652, Saudi Arabia. Internet: [email protected] Copyright © 2000. All rights reserved.

Transcript of Comprehensible Output by Ali Sheadeh

Page 1: Comprehensible Output by Ali Sheadeh

Non-Native Speakers’ Productionof Modified Comprehensible Output and

Second Language Learning

Ali ShehadehUniversity of Aleppo

King Saud University

This study investigated the ability of NNSs to modifytheir output toward comprehensibility in the contexts ofNS-NNS and NNS-NNS interactions and the degree towhich such modified comprehensible output (MCO) wasother- or self-initiated. Picture-dictation and opinion-exchange tasks were used to collect data from 8 NSs and24 NNSs of English representing 13 different L1 back-grounds. The 2 tasks were performed in pairs (NS-NNSand NNS-NNS) and were audiotaped. The results showedthat most repairs were self-initiated and that NNS-NNSinteractions produced more other-initiations and other-initiated MCOs on the picture-dictation task. The frequen-cies of these MCOs support the importance of modificationtoward comprehensible output as a process of second lan-guage acquisition.

Language Learning 49:4, December 1999, pp. 627–675

627

Ali Shehadeh, Department of English Language, University of Aleppo, Syria(currently at the College of Languages and Translation, King Saud Univer-sity, Saudi Arabia).

I acknowledge the help I received from the editors of Language Learningand a number of anonymous reviewers, and I would like to thank them allfor their advice and their many helpful suggestions and comments.

Correspondence concerning this article may be sent to Ali Shehadeh,College of Languages and Translation, King Saud University, P.O. Box 87907,Riyadh 11652, Saudi Arabia. Internet: [email protected]

Copyright © 2000. All rights reserved.

Page 2: Comprehensible Output by Ali Sheadeh

Swain (1985) argued that comprehensible input is not suffi-cient for successful second language acquisition (SLA), but thatopportunities for non-native speakers (NNSs) to produce compre-hensible output are also necessary. She based her conclusions onfindings from studies she conducted on immersion students inCanada. She found that although immersion students were pro-vided with a rich source of comprehensible input, their inter-language (IL) performance was still off-target; that is, they wereclearly identifiable as non-native speakers or writers (Swain,1984, 1985). Thus, Swain claimed that understanding new formsis not enough and that learners must also be given the opportunityto produce them. She, therefore, doubted that interactions andcomprehensible input on their own are sufficient for SLA:

Conversational exchanges . . . are not themselves thesource of acquisition derived from comprehensible input.Rather they are the source of acquisition derived fromcomprehensible output: output that extends the linguisticrepertoire of the learner as he or she attempts to createprecisely and appropriately the meaning desired. (Swain,1985, p. 252)

Swain (1985) argued that the role of learner production ofcomprehensible output is independent in many ways of the role ofcomprehensible input. She proposed a hypothesis relating to thesecond language (L2) learner’s production comparable to thatrelating to comprehensible input. This she termed the “compre-hensible output hypothesis” for SLA (1985, p. 249). Swainacknowledged the role of comprehensible input in SLA, butshe argued that comprehensible output is also a necessarymechanism that aids SLA in many ways:

Its role is, at minimum, to provide opportunities for con-textualized, meaningful use, to test out hypotheses aboutthe target language, and to move the learner from a purelysemantic analysis of the language to a syntactic analysisof it. (Swain, 1985, p. 252)

628 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 3: Comprehensible Output by Ali Sheadeh

Swain and Lapkin (1995) refined the comprehensible outputhypothesis, developed in Swain (1985, 1993), and summarized itas follows:

In producing the L2, a learner will on occasion becomeaware of (i.e., notice) a linguistic problem (brought tohis/her attention either by external feedback (e.g., clarifi-cation requests) or internal feedback). Noticing a problem“pushes” the learner to modify his/her output. In doing so,the learner may sometimes be forced into a more syntacticprocessing mode than might occur in comprehension.Thus,output may set “noticing” in train, triggering mental pro-cesses that lead to modified output. (Swain & Lapkin,1995,pp. 372–373)

Thus, Swain and Lapkin held the view that the activity of produc-ing the target language (TL) is a mechanism that enables learnersto notice a gap in their existing IL capacity. This noticing pushesthem to consciously reprocess their performance in order to pro-duce modified output.

Swain and Lapkin (1995, p. 374) argued that noticing of a gapleads to mental processes that in turn lead to the production ofmodified, reprocessed output, and that this “may represent theinternalization of new linguistic knowledge, or the consolidationof existing knowledge.” Figure 1 is an illustration of second lan-guage learning from an output perspective. Swain and Lapkinargued that “what goes on between the first output and the

‘noticing’ no solution solutionneed to communicate

output 1 (feedback) -internal -external

(analysis)-simple inspection-complex thinking

analyzeinput

output 2

Figure 1. Output and second language learning. From “Problems in Outputand the Cognitive Processes They Generate: A Step Towards SecondLanguage Learning,” by M. Swain and S. Lapkin, 1995, Applied Linguistics,16, p. 388. Reprinted with permission.

Shehadeh 629

Copyright © 2000. All rights reserved.

Page 4: Comprehensible Output by Ali Sheadeh

second . . . is part of the process of second language learning”(p. 386). In light of these arguments, they concluded that “some-times, under some conditions, output facilitates second languagelearning in ways that are different from, or enhance, those ofinput” (p. 371).

More recently, several SLA researchers have also argued thatlearning partly depends on L2 learners’ ability to attend to therelevant language features (Harley, 1998; Schmidt, 1995), to re-structure knowledge (Dekeyser, 1998), to focus on form whenlearners notice a “hole” in their interlanguage (Doughty & Wil-liams, 1998; Wright, 1996), on the extent to which noticing islearner-initiated (Long & Robinson, 1998), and on language pro-duction that gives learners the opportunity to reprocess andmodify their performance toward comprehensible output (Swain,1995,1998).Of particular relevance to the current study is Swain’sargument that language production enables learners to notice thegap between what they can say and what they want to say whenthey formulate and test out hypotheses about the TL, when theyconsciously reflect on the language they are producing (metatalk),and when they move from semantic analysis of the TL to a moresyntactic analysis of it (Swain, 1998, p. 79). According to Swain(1995, p. 126), language production, thus, prompts learners tostretch their current IL capacity in order to fill in the gap, “ena-bling them to control and internalize linguistic knowledge.”

Negotiated interaction and the negotiation of meaning haveuntil recently been taken as the basis for the provision of compre-hensible input only (e.g., Chiang & Dunkel, 1992;Deen,1995;Gass& Varonis, 1985, 1994; Holliday, 1995; Long, 1983a, 1983b, 1996;Pica, Doughty, & Young, 1986; Pica, Young, & Doughty, 1987;Varonis & Gass, 1985). In his (updated) interaction hypothesis, forinstance, Long maintained that negotiation for meaning triggersinteractional adjustments that are the main source of comprehen-sible input, and that comprehensible input in turn facilitates L2acquisition (Long, 1983b, p. 214, 1996, pp. 451–452).

Some SLA researchers have argued for the importance ofnegotiated interaction and the negotiation of meaning for the

630 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 5: Comprehensible Output by Ali Sheadeh

production of comprehensible output as well (e.g., Pica, Holliday,Lewis, Berducci, & Newman, 1991; Pica, Holliday, Lewis, & Mor-genthaler, 1989). For instance, Pica et al. (1989, p. 65) pointed out:

Although . . . research has focused mainly on the ways inwhich negotiated interaction with an interlocutor helps thelearner to understand unfamiliar L2 input, we believe thatit is also through negotiation that learners gain opportu-nities to attempt production of new L2 words and gram-matical structures as well.

Pica et al. (1989,p.84) further argued that through the negotiationof meaning both learners and their interlocutors work together toproduce comprehensible input and comprehensible output.

In her review article, Pica (1994) concluded that negotiationof meaning is important because it provides L2 learners withcomprehensible input, provides feedback on their production, andgives them an opportunity to modify their IL utterances. Otherresearchers also drew attention to the importance of negotiationof meaning to NNSs’ output, particularly the modifications theymake to their IL utterances when interlocutors signal difficulty inunderstanding (e.g., Hatch, Flashner, & Hunt, 1986; Lyster &Ranta, 1997; Sato, 1986; Tarone & Liu, 1995; Van den Branden,1997). For example, Van den Branden (1997, pp. 626–627) arguedthat L2 learners’ enhanced performance is primarily determinedby the frequency of negotiation routines they are engaged in. Heemphasized that during negotiations “learners can be pushed toproduce far more than merely comprehensible output; they can bepushed to the production of output that is more complete andaccurate” (p. 630). Similarly, Lyster and Ranta (1997) maintainedthat negotiation involves the provision of corrective feedback thatencourages self-repair, which in turn leads to more accurate andmore comprehensible output. They stated, “Negotiation involves. . . the provision of corrective feedback that encourages self-repairinvolving accuracy and precision and not merely comprehensibil-ity” (p. 42).

By the same token, Tarone and Liu (1995, p. 118) demon-strated that participation in different situational contexts affects

Shehadeh 631

Copyright © 2000. All rights reserved.

Page 6: Comprehensible Output by Ali Sheadeh

the rate as well as the route of acquisition because this causes thecompetence of the L2 learner to develop differently. They sug-gested that the contexts in which the learner stretches his/hercompetence in the TL to its limits are the contexts in which the ILdevelops faster. Tarone and Liu concluded that the data examinedin their study showed that

it is precisely in those contexts where Bob [the subject oftheir longitudinal case study] has to produce output whichhis IL cannot handle that the IL develops faster, with therichest variety of IL utterances and even possibly withstructures out of developmental sequence. (p. 120)

Studies on Comprehensible Output

The empirical research that investigated Swain’s argumentsfor the comprehensible output hypothesis and the related ILmodification claims will be briefly summarized here for the sakeof exposition.

Pica (1988) examined negotiated interactions between 1 NSand 10 NNSs of English to find out how the NNSs make their ILutterances comprehensible when the NS indicated difficulty inunderstanding them. Data for the study were obtained from Pica’searlier (1982) study. The data consisted of 10, 1 hr, transcribedaudiotapes of 10 first language (L1) Spanish/L2 English NNSs oflow-level proficiency interacting with an English NS who was anESL teacher experienced in talking to foreigners. The topics dis-cussed varied considerably and included issues such as previouseducation, future plans, personal matters, friends, relatives, aca-demic work, and so forth. The findings revealed that NNSs gener-ated their own modifications of initial trigger utterances only 48%of the time, but when they did so, consistently (91%) they showedtargetlike use of English. Pica (1988) concluded that these resultsshow that NNSs can modify their IL utterances in response to anNS signal to achieve output that is both more comprehensible andmore targetlike:

632 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 7: Comprehensible Output by Ali Sheadeh

. . . more comprehensible because . . . during 95% of theinteractions in which the NS signaled comprehension dif-ficulty, the NNS response to the signal led to a successfulresolution; and more target-like because 48% of the NNSs’total number of responses to the NS signal were encoded,through their own initiation, in more target-like form, afigure which represents 91% of the NNSs’ self-initiatedmodification. (Pica, 1988, pp. 59–60).

In a more detailed and controlled study, Pica et al. (1989)transcribed and analyzed audiotapes of 10 adult NSs of Japanese,5 males and 5 females, each interacting in English with 1 of 10female NSs of English in a noninstructional, experimental setting.The NNSs were of low-intermediate and mid-intermediate profi-ciency levels.Three tasks were employed to collect data: information-gap, jigsaw,and discussion (opinion-exchange).The results showedthat the NS signal type had a significant impact on the type ofresponse NNSs made, regardless of task. Across all tasks, NNSstended to modify their output most often when NSs signalled anexplicit need for clarification rather than provided a model utter-ance for confirmation. Pica et al. (1985, pp. 83–84) argued that thefindings of their study provide empirical validation for Swain’s(1985) construct of the comprehensible output hypothesis as wellas the other claims about the IL modification of NNSs.

Nobuyoshi and Ellis (1993) investigated learners’ ability toproduce more accurate output when teachers requested clarifica-tion. Six Japanese students performed the same task twice (tellinga picture-based story to their teachers) a week apart. The re-searchers studied the use of past tense forms. The teachers re-quested clarification either when they noticed an incorrect use ofthe past tense form or when they did not actually understand theverb used. Nobuyoshi and Ellis found that when teachers pushedlearners in the direction of greater accuracy in their production,learners were able not only to make self-repair, but also to achievea higher accuracy level in their output. This improved accuracy ofoutput resulted in improved performance, both immediately andover time.

Shehadeh 633

Copyright © 2000. All rights reserved.

Page 8: Comprehensible Output by Ali Sheadeh

Pica, Lincoln-Porter, Paninos, and Linnell (1996) comparedthe interaction of 10 dyads of English L2 learners with that of10 dyads of learners and English NSs on two jigsaw tasks (a housesequence task and a story task). All NNS participants shared onemother tongue background (Japanese). They were all adult learn-ers enrolled in pre-academic, low-intermediate-level classes. Theresearchers sought to investigate the difference in the learners’further modification of their previous utterances when these ut-terances were modified by NSs or other learners. Pica et al. (1996)predicted that “when learners were given signals that modifiedtheir previous utterances, they would produce a similar amountof modified output in their responses whether the signals werefrom other learners or from NSs” (p. 63). As predicted, the resultsshowed that there was comparable production of modified ver-sions of their previous utterances when learners negotiated withNSs or with each other.

Finally, Van den Branden (1997) investigated the effect, im-mediate and delayed, of various types of negotiation on L2 learn-ers’ output in an instructional setting.He collected data from threegroups of 16 fifth grade primary school pupils whose ages rangedfrom 11 to 12 years. The NNS pupils were each asked to describesix drawings (to solve a murder case) to an NS peer or teacherpartner who was not allowed to see the drawings. The researcherfound that children were able to modify their output interaction-ally when confronted with a negative feedback, and that theseinteractional modifications had delayed effect on their output insubsequent interactions (p. 626). Van den Branden concluded that“children of 11 to 12 years old appear to be quite able to push eachother’s output through negotiating meaning and content, and tolearn enough from these negotiations to produce higher-qualityoutput during subsequent interactions” (p. 627).

This review shows that previous empirical studies on com-prehensible output and IL modification have collected data mostlyfrom NS-NNS interaction (but see Pica et al., 1996) and only wheninterlocutors requested clarification. Past research did not inves-tigate the role of NNS-NNS interaction or, more importantly, the

634 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 9: Comprehensible Output by Ali Sheadeh

role of self-initiation in providing opportunities for the productionof comprehensible output. Examining the effect of these variablesis important to further substantiate the theoretical claims under-lying Swain’s comprehensible output hypothesis.

Purpose

This study therefore examines the ability of NNSs to modifytheir IL utterances in the direction of comprehensible output,termed here modified comprehensible output (MCO), in responseto other-initiation and self-initiation, in both NS-NNS and NNS-NNS interactions on two communication tasks: picture-dictationand opinion-exchange. These objectives were motivated by severalconsiderations.

First, Gass and Varonis (1985, 1986) and Varonis and Gass(1985) argued that NNS-NNS interactions provide greater oppor-tunities than NS-NNS interactions for the negotiation of meaningand therefore for obtaining comprehensible input.The assumptionis that a greater amount of negotiation work takes place inNNS-NNS discourse than in either NS-NS or NS-NNS discourse.Varonis and Gass (1985, pp. 84–86) maintained that in NNS-NNSinteractions, NNSs do not lose face by negotiating meaning in thesame way as they might in NS-NNS interactions.They argued thisis because NNS-NNS interactions provide NNSs with anonthreatening forum within which to practice developing lan-guage skills, including an opportunity to receive input they havemade comprehensible through negotiation (Varonis & Gass, 1985,p.87).Thus NNS-NNS negotiation serves the function of providingNNS participants with a greater amount of comprehensible input.

By extension, one might expect that NNS-NNS interactionwill provide NNS participants with more opportunities to adjusttheir IL utterances in the direction of comprehensible output thanNS-NNS interaction. Apart from Pica et al. (1996), previous em-pirical research on comprehensible output did not take NNS-NNSinteraction into consideration. However, unlike Pica et al. (1996),the present study will investigate NNSs’ ability to encode their

Shehadeh 635

Copyright © 2000. All rights reserved.

Page 10: Comprehensible Output by Ali Sheadeh

own modification toward comprehensible output rather than ex-amining their response to modified versions of their previousutterances supplied to them by their interlocutors, as was done byPica et al. (1996).

Also, there are occasions when NNSs modify their output tomake it more comprehensible when they realize that their currentor previous utterance is insufficient as a means for communicatingthe intended message. These are normally referred to in theliterature as self-initiated self-completed repairs (Kasper, 1985).Ethnomethodological researchers (Schegloff, Jefferson, & Sacks,1977; Kasper, 1985) earlier drew attention to the importance ofself-repair. Specifically, in their seminal paper, Schegloff et al.(1977, pp. 364–365) identified four types of repair in NS-NSinteractions: self-repair issuing from self-initiation, self-repairissuing from other-initiation, other-repair issuing from other-initiation, and other-repair issuing from self-initiation. It wasfurther shown (Færch & Kasper, 1983; Schegloff et al., 1977) thatself-initiations may be placed in three positions: (a) They may beplaced within the same turn; (b) they may be placed in the turn’stransition space, that is, immediately after the end of the turn; or(c) they can be placed in third turn to the trouble-source, that is,in the turn subsequent to that which follows the trouble-sourceturn.On the other hand,other-initiations occupy one position only:the turn immediately subsequent to the trouble-source turn.

Schegloff et al. (1977) claimed that in NS-NS conversations,self-initiated self-completed repairs are preferred to other typesof repairs. They argued that there is ample evidence in support ofthis claim, in particular that (a) opportunities for self-initiationcome before opportunities for other-initiation; (b) for those re-pairables (i.e., trouble-sources) where repair is initiated, same-turn and transitional space opportunities for self-initiation arefrequently taken by speakers of the trouble-source; (c) the courseof same-turn initiated repairs regularly leads to successful self-repairs in same turn; and (d) other-initiation in most cases yieldsself-repair because although the addressee has a hearing or un-derstanding problem and would initiate repair, he/she will normally

636 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 11: Comprehensible Output by Ali Sheadeh

give the speaker of the trouble-source the opportunity to repair forhim/herself. Therefore, Schegloff et al. argued that self- and other-initiation are related to each other and that their relatedness isorganized in favor of self-repair (p. 374). They concluded thatself-repairs are thus more effective than and are highly privilegedover other-repairs: “the organization of repair in conversationprovides centrally for self-correction, which can be arrived at by thealternative routes of self-initiation and other-initiation—routeswhich are themselves so organized as to favor self-initiated self-repair” (p. 377).

These observations were confirmed in a separate study bySchegloff (1979), who concluded that “self-initiation, same turnrepair is the most common and most successful too” (p. 268).Moreover, the predominance of self-repair was also confirmed bystudies of NS-NNS and NNS-NNS interactions, both in educa-tional and in natural contexts (Brock, Crookes, Day, & Long, 1986;Chun, Day,Chenoweth, & Luppescu, 1982; Day, Chenoweth,Chun,& Luppescu, 1984; Gaskill, 1980; Kasper, 1985; Schwartz, 1980).

Previous empirical research that investigated the IL modifica-tion of NNSs looked only at the opportunities that other-initiatedclarifications (termed here other-initiated clarification requests)give NNSs to modify their performance in the direction of compre-hensible output. For this reason, the present study will alsoconsider the opportunities that self-initiations (termed here self-initiated clarification attempts) give NNSs to modify their outputtoward comprehensibility.

Finally, very few studies have examined the effect of type oftask on initiation opportunities and NNSs’ production of compre-hensible output (Pica et al., 1989; Pica et al., 1991; Pica et al.,1996). For example, Pica et al. (1989) investigated the effect ofpicture-drawing, jigsaw, and discussion tasks on initiation oppor-tunities and IL modification toward comprehensible output. Theypredicted that “opportunities given by NSs for NNSs to make theiroutput comprehensible would be greatest in the information-gappicture drawing task, less so in the jigsaw sequencing task, andleast during discussion task” (p. 69). Pica et al. (1989) argued that

Shehadeh 637

Copyright © 2000. All rights reserved.

Page 12: Comprehensible Output by Ali Sheadeh

their hypothesis was believed to reflect the degree of NNS controlover the amount of information needed for successful task comple-tion and the precision with which this information needed to beemployed. As predicted, they found that the picture-drawing task“offered the largest percentage . . . of NS signals of requests forclarification and confirmation” in comparison to jigsaw and dis-cussion tasks (p. 74).

However, previous research on comprehensible output didnot examine the potential effect of type of task on opportunitiesfor self-initiated clarification attempts by NNSs. For this reason,the effect of type of task on opportunities for other-initiation andself-initiation and on the MCOs produced in response to bothinitiation types will be examined in this study.

Research Hypotheses

By extending the arguments by Varonis and Gass (1985) andGass and Varonis (1985, 1986) and the observations by eth-nomethodological researchers like Schegloff (1979) and Schegloffet al. (1977), it is reasonable to expect that NNS-NNS interactionswill also provide NNS participants with more opportunities forother-initiated clarification requests and self-initiated clarifica-tion attempts in the direction of comprehensible output thanNS-NNS interactions. On the basis of these considerations, thefirst hypothesis was generated as follows:

Hypothesis 1. Opportunities for the production of utterancesexhibiting modification toward comprehensibility in NNS outputwould be greater in NNS-NNS interactions than in NS-NNSinteractions, specifically:

(a) There would be a greater proportion of other-initiatedclarification requests.

(b) There would be a greater proportion of self-initiated clari-fication attempts.

(c) There would be more MCOs produced.

638 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 13: Comprehensible Output by Ali Sheadeh

Pica, Kanagy, and Falodun (1993) characterized picture-dictation as a one-way task in which supply of information forsuccessful task completion is required, and opinion-exchange as atwo-way task in which supply of information for successful taskcompletion is optional. This is because successful completion of apicture-dictation task, unlike an opinion-exchange task, dependsheavily on the ability of the NNS participant dictating the con-tents of the picture to his/her task partner to be as comprehensibleand as accurate as possible in supplying information when per-forming the task. Therefore, there will be more frequent clarifica-tion requests on a picture-dictation task than on anopinion-exchange task. In light of these arguments by Pica et al.(1993), the findings of Pica et al. (1989) above,and the ethnometho-dological observations concerning the preference for self-initiatedself-completed repair, there is reason to believe further that therewill be more frequent other-initiations and self-initiations andmore MCOs produced on the picture-dictation task than on theopinion-exchange task employed in this study. These considera-tions led to the formulation of the second hypothesis.

Hypothesis 2. Opportunities for the production of utterancesexhibiting modification towards comprehensibility in NNS output asa result of other- and self-initiations would be greater on the picture-dictation task than on the opinion-exchange task, specifically:

(a) There would be greater frequency of other- and self-initiations, and

(b) There would be more MCOs produced.

Method

Participants

Thirty-two adults, ranging in age from 22 to 37 years, wereselected. There were 8 NSs (4 males and 4 females) and 24 NNSsof English (10 males and 14 females). All NNS participants wereacquainted as ESL classmates in the same course. The NNSs

Shehadeh 639

Copyright © 2000. All rights reserved.

Page 14: Comprehensible Output by Ali Sheadeh

represented 13 different L1 backgrounds. These included Afri-kaans,Arabic,Bengali,Cantonese,French,Greek, Italian,Korean,Mandarin, Farsi, Portuguese, Serbo-Croat, and Spanish. The NSparticipants were university teachers and postgraduate studentsexperienced in talking to foreigners.

The length of the NNSs’ residence in the UK at the time ofthe investigation ranged between 2 and 3 months. All NNS par-ticipants were enrolled in a communicatively oriented EAP (En-glish for Academic Purposes) program at the language center of aBritish university and were receiving 5 hr of instruction weekly.The course was running simultaneously with their respectivepostgraduate studies at the various departments of the university.At the time of the investigation, the program had three levels ofinstruction, ranging from a beginning course, Level 1, to a coursefor advanced students, Level 3. The 24 NNS students were en-rolled in Level 2, the intermediate ESL level.

Communication Tasks

Two communication tasks were selected and developed forthe purpose of this study: a picture-dictation task and an opinion-exchange (discussion) task. The selection of the two tasks wasmotivated in the first place by previous studies (e.g., Crookes &Rulon, 1988; Gass & Varonis, 1985; Hawkins, 1985; Lynch, 1988;Pica et al., 1991; Pica et al., 1993), and in particular by Pica et al.’s(1989, p. 72) remark that “there is a great deal of consensusregarding the value of these tasks in providing data on interactionin general and negotiated interaction in particular.” (For moredetailed exposition on the significance and usefulness of these andother communication tasks in SLA research and instruction, seetwo collections of informative articles edited by Crookes & Gass,1993a, 1993b.)

In the picture-dictation task, an NNS had to describe thecontents of a picture to an NS or NNS partner who had toreproduce the picture as precisely as possible solely on the basisof the NNS’s description. This means that successful completion

640 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 15: Comprehensible Output by Ali Sheadeh

of the task depends in the first place on the NNS’s ability to supplya clear and accurate description of the picture he/she holds. Theprocedure for this task follows that of previous studies (e.g., Vanden Branden, 1997; Varonis & Gass, 1985) in that an NNS wasgiven a picture (a postcard in our case) to describe to an NS orNNS partner. The postcard itself was carefully selected as havingclear, easily recognizable features describable in a simple vocabu-lary to enable the NNS to understand its content and describe itto his/her partner. Original color copies of the same postcard weredistributed to all NNSs whose role was to describe the picture totheir task partners. In addition to the oral explanation, each NNSspeech partner with the picture was given a written set of instruc-tions and guidelines.

The opinion-exchange task employed in the present studywas a discussion in which the interactants were required toexchange opinions about a short newspaper article. The task wasopen-ended and did not require any of the speech partners to useonly the precise information available to them or adhere to anyspecific aspect(s) of the theme of the article. A copy of a controver-sial and stimulating newspaper article entitled “Chauvinist Hus-band Divorced” was photocopied and distributed to everyparticipant. It is worth noting that this particular text was care-fully selected to match the proficiency level of the NNS partici-pants. Characteristics such as the length of the text (200 to 250words), length of the sentence, complexity of content, and famili-arity of vocabulary were all taken into consideration. Participantswere given 10 min reading time to formulate their views, opinions,arguments, and so on.

Administration and Data Collection Procedures

Sixteen NNS participants were each teamed up with an NSor NNS partner and performed the two tasks. Each dyad per-formed the picture-dictation task and then the opinion-exchangetask. The data were collected in two sessions. In the first session,there were 10 dyads. In the second session, 6 new dyads performed

Shehadeh 641

Copyright © 2000. All rights reserved.

Page 16: Comprehensible Output by Ali Sheadeh

the same two tasks that were performed by the 10 dyads earlier.In the picture-dictation task, dyads sat back-to-back to preventany paralinguistic or nonlinguistic interference. All encounterswere audiotape-recorded. The total number of tape-recordings was32. Dyadic distribution of participants over the two tasks was asfollows: 4 were female NS–female NNS pairs, 4 were maleNS–male NNS pairs, 5 were female NNS–female NNS pairs, and3 were male NNS–male NNS pairs. Altogether, there were 16 pairsdistributed among four types of encounters as follows: 8 NS-NNSand 8 NNS-NNS pairs on the picture-dictation task; 8 NS-NNSand 8 NNS-NNS pairs on the opinion-exchange task.

Data Transcription

Systematically selected 5-min samples (between minutes2–6 inclusive) of each audiotaped activity were transcribed foranalysis. Several factors played a part in the selection of thesesamples. First, drawing on earlier studies (e.g., Duff, 1986; Gass& Varonis, 1985, 1986), which selected samples ranging between5 and 10 min, it was believed that sufficient data for analysiswould be obtained from the 5-min sample from each encounterselected in the present study. Second, as one picture-dictationencounter and three opinion-exchange encounters ended after 6–7min, to maintain systematicity only the first 6 min of the other 28dyadic encounters were transcribed for analysis. Third, the 1stmin of each encounter was not selected for analysis because Iwished to minimize the effect of the conversational preliminariesthat generally precede focussing on the specific topic of the taskto be performed (Aston, 1986, p. 132).

Parts of the tapes selected were transcribed and double-checked. To get the closest transcription possible and to maintainthe highest possible quality, a team of 9 NSs of English recheckedand verified the transcriptions to ensure their accuracy. They wereall given an introductory session informing them what to expectand what they were expected to do. In a limited number of caseswhere there were still unresolved transcription difficulties, the

642 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 17: Comprehensible Output by Ali Sheadeh

original participants were invited to interpret.The final transcrip-tion was intended to reflect what was recorded during the taskswith all its linguistic features marked in the transcription. Theseincluded all hesitation fillers, semiaudible murmuring, low voicequality, indications of unintelligible language, comprehensionchecks/signals, confirmation requests/signals, clarification re-quests, repairs, interruptions, overlaps, simultaneous speech,pauses, turn-giving, and turn-taking moves.

A Model for Interlanguage ModificationToward Comprehensibility

Partly based on Varonis and Gass’s (1985) model,1 a newmodel was developed as a framework for describing NNSs’ ILmodification toward comprehensible output and as a guideline forthe analysis of the data (see also Shehadeh, 1991). The model fallsinto four functional primes. The first prime consists of a trouble-source (TS) or trigger. The second prime consists of an initiator (I),which can be either self-initiation (SI) or other-initiation (OI). Thethird prime is the outcome (O) of initiation. And the fourth primeconsists of the interlocutor’s reaction to the outcome (RO).

A trouble-source may be either ignored or reacted to. In caseswhere the TS was ignored, there is often no way for the investiga-tor to recognize that there was a breakdown in comprehension orcommunication, although something later in the discourse mayindicate that in fact the listener had not understood (Aston,1986),2

or that the speaker did run into difficulty but did not initiate repair(Hawkins, 1985; Varonis & Gass, 1985). On the other hand, thetrouble-source may be reacted to either by the interlocutor, inwhich case we have other-initiation, or by the originator of thetrouble-source himself/herself, in which case we have self-initiation.The outcome of initiation can take different forms, includingignoring the TS or other-initiation/signal of TS, failing to repair,appealing for help, switching to a new topic, expressing difficultyin repairing the TS,repeating the TS utterance without modification,or successfully repairing the TS by reprocessing and modifying IL

Shehadeh 643

Copyright © 2000. All rights reserved.

Page 18: Comprehensible Output by Ali Sheadeh

performance in the direction of comprehensible output—in otherwords, supplying modified comprehensible output (MCO). Thereaction to the outcome is an optional unit of the routine that helpsto tie up the routine in some way before the speakers return to themain flow of conversation.

Following Lyster and Ranta (1997) and Spada and Fröhlich(1995), clarification requests indicate to the NNS that his/her utter-ance has not been understood or has been misunderstood or that theutterance is ill-formed in some way. This means that clarificationrequests “can refer to problems in either comprehensibility or accu-racy, or both” (Lyster & Ranta, 1997, p. 47). Similarly, self-initiatedclarification attempts also reflect the NNS’s realization that his/herutterance or part of it was not understood or was misunderstood bythe interlocutor(s), or that the utterance was/is ill-formed in someway. In either case, the NNS realizes that a reformulation toward“greater message comprehensibility is needed” to transmit messagemeaning successfully, to use Pica’s (1994, p. 498) words.

Modified comprehensible output can therefore be operation-ally defined as the output that NNSs achieve to make an initialutterance more accurate and/or more comprehensible to theirinterlocutor(s) in response to (a) other-initiated clarification re-quests and (b) self-initiated clarification attempts.3 On the basisof this definition, the model operates according to two differentbut closely related mechanisms. The first mechanism(TS—OI—O—RO) implies the presence of interpersonal negotia-tion of information units, which may be described as routines/instances in which negotiation of meaning in the direction ofgreater message comprehensibility takes place between interlocu-tors. This mechanism is demonstrated by the following excerpt,which illustrates a simple routine for the negotiation of meaningbetween two NNSs. Here and throughout the excerpts cited,underlined items refer to the trouble-source (TS), italicized itemsto the outcome (O), and items that are both underlined anditalicized refer to an outcome that is still a source of trouble.Numbers in parentheses refer to the length, in seconds, of a pauseor cutoff within a speech turn or between speech turns.

644 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 19: Comprehensible Output by Ali Sheadeh

NNS1: two small bottle TSNNS2: two small what? OINNS1: bot (1.0) small bottles ONNS2: yeah RO

Below is an excerpt of an extended negotiation of information unitthat illustrates complex and embedded layers with a series ofthree signal-response exchanges for the negotiation of meaning:

NNS1: and another side one in that er (1.0)in side of table emm nabikin is hanging TS

NNS2: what? OINNS1: nappikin or towil O/TSNNS2: what? OINNS1: towel or nappkin for er rubbing hand O/TSNNS2: I don’t know what is it (1.0) what

is this for? OINNS1: for men to dry hands after washing

towel towel ONNS2: uhh washing to dry hands yes yes RO

The second mechanism (TS—SI—O—RO) implies the provi-sion of information units that are completed without negotiationof meaning. These may be described as routines/instances in thedirection of greater message comprehensibility in which no actualnegotiation of meaning takes place. This mechanism is demon-strated by the following example, which illustrates a series ofNNS’s self-initiated clarification attempts leading to IL modifica-tion toward comprehensibility.

NNS: yes because if the woman is (0.8) the wife alwaysgo out (0.6) goes out and left his his husband eh(1.0) her husband and her son in the home (0.7)at home it’s it’s not reason able for for . . .

It is worth noting that the reason self-initiated modificationshave not been previously studied within the task-based frame-work used in the present study is that repair and its variousoutcomes (e.g., modified input, feedback, and modified output)

Shehadeh 645

Copyright © 2000. All rights reserved.

Page 20: Comprehensible Output by Ali Sheadeh

have been investigated within the context of negotiation, seen asa mutual activity that occurs during episodes of mutual attemptsto clarify message meaning (see, for example, studies by Gass &Varonis, 1986; Nobuyoshi & Ellis, 1993; Pica, 1988; Pica et al.,1989; Pica et al., 1991; Pica et al., 1996; Van den Branden, 1997;Varonis & Gass, 1985). In this sense, self-initiated clarificationattempts appear to fall slightly out of the operationalized defini-tion of negotiation moves. However, the present study will look atboth initiation types, because, as demonstrated above, both givethe NNS the chance to modify his/her IL utterances towardgreater comprehensibility in order to transmit message meaningsuccessfully (see also Swain, 1998; Swain & Lapkin, 1995, for asimilar position).

Coding of Data

All data were coded on the basis of the model proposed aboveand according to the categories displayed in the Appendix. Thedata were coded/marked for the following categories: (a) other-initiated clarification requests in which interlocutors requestedclarification, made an explicit statement of nonunderstanding, orrequested reformulation, explanation, expansion, paraphrase, orelaboration;4 (b) self-initiated clarification attempts in whichNNSs attempted to clarify or reformulate an utterance or part ofit by lexical means or nonlexical means when they realized theiroutput was not understood or was misunderstood, or that it was/isill-formed in some way; (c) NNSs’ response to (a), including codingfor the particular outcome type (ignoring the signal for repair, failingto repair,appealing for help [direct and indirect],expressingdifficultyin repairing, switching the topic, or successfully producing MCO); (d)NNSs’ response to (b), including coding for the particular outcometype (failing to repair, appealing for help [direct and indirect], ex-pressing difficulty in repairing/communicating the intended mes-sage, switching the topic, or successfully producing MCO).

It is worth mentioning that in the NNS-NNS interactions,just like NS-NNS interactions, the focus was on one NNS

646 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 21: Comprehensible Output by Ali Sheadeh

only—the NNS whose role was to describe the picture to his/hertask partner (on the picture-dictation task)—and on that sameNNS on the opinion-exchange task. More precisely, in all inter-actions, the focus was on one NNS’s responses to (a) self-initiationand (b) an NS or NNS partner initiation.

All data were coded for c-units, as defined by Brock et al.(1986) as utterances (words, phrases and sentences, grammaticaland ungrammatical) that provide linguistic, referential, or prag-matic meaning to NS-NNS and NNS-NNS interactions. Based ona 20% sample of data, interrater reliability for the coding was .92.This sample was coded by three raters, myself and two othertrained raters of the nine NSs who had already double-checkedand verified the transcriptions for accuracy.

Analysis

The data collected for each hypothesis were analyzed in termsof frequencies, proportions, and percentages, and for level of signifi-cance.The first analysis examines the frequencies,proportions,andpercentages of initiations (other and self) in relation to tasks(picture-dictation vs. opinion-exchange), encounters (NS-NNS vs.NNS-NNS), and the proportions of MCO instances provided inthese contexts, including their provision after one-signal orextended-negotiation routines. The second type of analysis relatesto the level of significance of the MCOs produced. An analysis ofvariance (ANOVA) test was used to examine the frequency of MCOsproduced in relation to type of speech partner, type of task,and typeof initiation.5 The level of significance was set at .05.

Results

Hypothesis 1(a)

The first part of Hypothesis 1 predicted that NNS-NNSinteractions would provide a greater proportion of other-initiated

Shehadeh 647

Copyright © 2000. All rights reserved.

Page 22: Comprehensible Output by Ali Sheadeh

clarification requests than NS-NNS interactions. To test thisprediction, proportions of NS other-initiations were compared toproportions of NNS other-initiations on the same task. This pre-diction was partially supported by the results. On the one hand,there were no significant differences between the total proportionof NS other-initiations and NNS other-initiations on the picture-dictation task and the opinion-exchange task when taken together,F(1, 30) = 1.31, ns, nor when the opinion-exchange task was takenalone, F(1, 14) = 0.06, ns. On the other hand, however, a level ofsignificance was achieved on the picture-dictation task whentaken alone, F(1, 14) = 8.46, p < .05, showing that the proportionof NNS other-initiated clarification requests was greater than thatof NS other-initiated clarification requests on this particular task.These results are shown in Table 1.6

As Table 1 shows, a total of 212 other-initiated clarificationrequests were made across both tasks, 88 cases (or 42%) of whichwere initiated by the NS partners and 124 cases (or 58%) by theNNS partners.

Hypothesis 1(b)

The second part of Hypothesis 1 predicted that NNS-NNSinteractions would result in a greater proportion of self-initiatedclarification attempts than NS-NNS interactions. To test this

Table 1

Frequencies and Proportions of NS Other-Initiations and NNSOther-Initiations

NS NNSTask other-initiations other-initiations Total

n % n % nPicture-dictation 74 41 108 59 182Opinion-exchange 14 47 16 53 30Total 88 42 124 58 212

648 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 23: Comprehensible Output by Ali Sheadeh

prediction, frequencies and proportions of self-initiations in NS-NNS interactions were compared to frequencies and proportionsof self-initiations in NNS-NNS interactions. This prediction wasnot supported by the results. There were no significant differencesbetween the total proportion of self-initiations in NS-NNS inter-actions and the total proportion in NNS-NNS interactions on thepicture-dictation task and the opinion-exchange task when takentogether, F(1, 30) = 0.17, ns, or when looked at separately: F(1, 14)= 0.02, ns, for the picture-dictation task, and F(1, 14) = 0.16, ns,for the opinion-exchange task. The results of this prediction areshown in Table 2.

As Table 2 shows, a total of 435 self-initiated clarificationattempts were made across both tasks, 212 cases (or 49%) of whichwere self-initiated by NNSs when their speech partners were NSs,and 223 cases (or 51%) when their speech partners were otherNNSs. Unlike the findings of prediction (a) of Hypothesis 1, thisfinding did not exhibit any significant differences in the perfor-mance of NNSs with regard to self-initiation in both types ofinteraction.

Hypothesis 1(c)

The most important part of Hypothesis 1 relates to the possibleassociation of type of speech partner with NNS production of MCOs.

Table 2

Frequencies and Proportions of Self-Initiations in NS-NNSInteraction and NNS-NNS Interaction

Self-initiations in Self-initiations inin NS-NNS in NS-NNS

Task interaction interaction Total

n % n % nPicture-dictation 115 50 117 50 232Opinion-exchange 97 48 106 52 203Total 212 49 223 51 435

Shehadeh 649

Copyright © 2000. All rights reserved.

Page 24: Comprehensible Output by Ali Sheadeh

This is the motivation underlying the third prediction of thishypothesis. Hypothesis 1 (c) predicted that there would be agreater proportion of MCOs produced in NNS-NNS interactionthan NS-NNS interaction. To test this prediction, proportions ofthe MCOs produced in NS-NNS interactions were compared tothose produced in NNS-NNS interactions (Table 3). Results ofANOVA produced mixed support for this prediction. On the onehand, the first case, which deals with the occurrence of NS-initiatedMCOs versus NNS-initiated MCOs produced on the picture-dictation task, reached significance, F(1, 14) = 20.17, p < .05. Onthe other hand, no significant differences were observed betweenthe MCOs produced across both tasks and both initiation sources(other- and self-) when all these variables were taken together,F(1, 62) = 1.52, ns, or when the other three cases were looked atseparately: F(1, 14) = 0.00, ns, for NS-initiated MCOs versusNNS-initiated MCOs on the opinion-exchange task; F(1, 14) =0.20, ns, for self-initiated MCOs in NS-NNS interactions versusNNS-NNS interactions on the picture-dictation task; F(1, 14) =0.35, ns, for self-initiated MCOs in NS-NNS interactions versusNNS-NNS interactions on the opinion-exchange task.

As displayed in Table 3, a total of 573 instances of MCOs wereproduced across both tasks and all contexts, 258 (or 45%) of whichwere produced by NNSs when their task partners were NSs, and315 (or 55%) when their task partners were other NNSs.

The findings of Hypothesis 1 so far show that there are veryfew differences of effect between NSs and NNSs as speech part-ners. Such differences were observed in the proportion of other-initiated clarification requests (in 1(a)) and other-initiated MCOsproduced (in 1(c)) on the picture-dictation task.

One of the basic assumptions underlying the hypothesisunder consideration was that NNS-NNS interactions would in-volve more negotiation of meaning than NS-NNS interactions.With this in mind, it was believed that NNS-NNS interactionswould exhibit a greater proportion of extended negotiation rou-tines toward comprehensibility than NS-NNS interactions. Basedon this assumption, a further analysis was conducted to see if

650 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 25: Comprehensible Output by Ali Sheadeh

Table 3

Frequency of MCOs Produced in NS-NNS and NNS-NNS Interactionsin Relation to Source of Initiation and Task

Source of MCOs produced in MCOs produced ininitiation Task NS-NNS interactions NNS-NNS interactions Total

n % n % nOther PD 54 37 91 63 145

OE 13 50 13 50 26

Self PD 102 48 109 52 211OE 89 47 102 53 191

Total 258 45 315 55 573

Note. PD: picture-dictation; OE: opinion-exchange.

Copyright © 2000. All rights reserved.

Page 26: Comprehensible Output by Ali Sheadeh

there were any significant differences between the two types ofinteraction in this respect. As displayed in Table 4, this analysisrevealed that there were 53 cases of MCOs produced after ex-tended negotiations in NS-NNS interactions and 84 in NNS-NNSinteractions. ANOVA showed that there were significant differ-ences between the two contexts, F(1, 62) = 5.39, p < .05, with theNNS-NNS interactions providing a significantly greater propor-tion of MCOs produced after extended negotiations than NS-NNSinteractions.

Overall, the results of Hypothesis 1 revealed that there weresome differences between NS-NNS and NNS-NNS interactionswith regard to initiation and MCO. Namely, NNS-NNS inter-actions provided significantly more instances of other-initiationsand other-initiated MCOs on the picture-dictation task. They alsoprovided a significantly higher proportion of the MCOs producedafter extended negotiations than NS-NNS interactions. On theother hand, there were no significant differences between the twotypes of encounters with regard to (a) proportions of self-initiationson either task, (b) proportions of self-initiated MCOs produced oneither task, and (c) proportions of other-initiations and other-initiated MCOs produced on the opinion-exchange task. Thus, thefindings of Hypothesis 1 lend partial support to the assumptionthat NNS-NNS interaction would provide better contexts thanNS-NNS interaction for other-initiation and self-initiation,and forIL modification toward comprehensibility.

Hypothesis 2(a)

The first part of Hypothesis 2 predicted that there would begreater frequency of other-initiations and self-initiations on thepicture-dictation task than the opinion-exchange task. To test thisprediction, the effect of type of task on initiation was examined bycomparing the total frequency of initiations (other- and self-) oneach task. The results showed that there was a total of 647initiation occurrences across both tasks;414 (or 64%) of these weremade on the picture-dictation task and 233 (or 36%) on the

652 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 27: Comprehensible Output by Ali Sheadeh

opinion-exchange task. ANOVA showed that there were signifi-cant differences in opportunities for initiation on each task,F(1, 62) = 16.72, p < .01, such that the picture-dictation taskprovided more initiation instances than the opinion-exchangetask.

Further analyses were conducted to see if the type of initia-tion (other- or self-) played any role in bringing about thesedifferences. Examining the frequency of other-initiated clarifica-tion requests showed that there were significant differences,F(1, 30) = 86.81, p < .001, between the two tasks. There was a totalof 212 cases of other-initiated clarification requests across bothtasks; 182 (or 86%) of these were made on the picture-dictationtask and 30 (or 14%) on the opinion-exchange task. It is worthnoting that although the level of significance adopted for thepresent study was .05, here the difference between the opportuni-ties provided by the two tasks was significant at the level of .001.

Examining the frequency of self-initiated clarification at-tempts showed no difference between the two tasks,F(1,30) = 1.24,ns. A total of 435 cases of self-initiated clarification attempts weremade across both tasks; 232 (or 53%) of these were made on thepicture-dictation task and 203 (or 47%) on the opinion-exchangetask. As can be seen from these figures, the occurrence of self-initiated clarification attempts was almost evenly distributedbetween the two tasks.

Hypothesis 2(b)

The second part of Hypothesis 2 predicted that there wouldbe more MCOs produced on the picture-dictation task than theopinion-exchange task. In light of the results of Hypothesis 2(a),it is reasonable to expect significant differences between the twotasks in terms of the opportunities each task offers NNSs tomodify their IL utterances in the direction of comprehensibleoutput.

To test this prediction, the proportion of MCOs produced onthe picture-dictation task was compared to the proportion of

Shehadeh 653

Copyright © 2000. All rights reserved.

Page 28: Comprehensible Output by Ali Sheadeh

Table 4

Frequencies, Proportions, and Percentages of MCOs Produced After One-Signal and Extended Negotiationsin NS-NNS and NNS-NNS Interactions

Participants Task Initiation source One-signal MCOs Extended-negotiation MCOs Total

n % n % nNS-NNS PD Other 44 81 10 19 54

interactions Self 82 80 20 20 102OE Other 9 69 4 31 13

Self 70 79 19 21 89

Total 205 79 53 21 258

NNS-NNS PD Other 58 64 33 36 91interactions Self 81 74 28 26 109

OE Other 8 62 5 38 13Self 84 82 18 18 102

Total 231 73 84 27 315

Note. PD: picture-dictation; OE: opinion-exchange.

Copyright © 2000. All rights reserved.

Page 29: Comprehensible Output by Ali Sheadeh

MCOs produced on the opinion-exchange task.The results showedthat there was a total of 573 instances of MCOs produced acrossboth tasks; 356 (or 62%) of these were provided by the picture-dictation task and 217 (or 38%) by the opinion-exchange task.ANOVA showed that there were significant differences betweenthe two tasks, F(1, 62) = 10.30, p < .05, such that the picture-dictation task provided a greater proportion of the MCOs producedthan the opinion-exchange task.

Further analyses were also conducted here to see if the typeof MCOs produced (other-initiated or self-initiated) played anyrole in bringing about these differences. Examining the frequencyof other-initiated MCOs showed that there were significant differ-ences, F(1, 30) = 73.47, p < .001, between the two tasks. There wasa total of 171 cases of other-initiated MCOs produced across bothtasks; 145 (or 85%) of these were provided by the picture-dictationtask and 26 (or 15%) by the opinion-exchange task. Again, it isworth noting that although the level of significance adopted forthe present study was .05, here we have another case in which thedifference between the opportunities provided by the two taskswas significant at the level of .001.

The frequency of self-initiated MCOs produced on each taskwas examined. The results found no difference between the twotasks, F(1, 30) = 0.58, ns. There was a total of 402 cases ofself-initiated MCOs produced across both tasks; 211 (or 52%) ofthese were provided by the picture-dictation task and 191 (or 48%)by the opinion-exchange task. It is clear from these figures thatthe occurrence of self-initiated MCOs produced was also almostevenly distributed between the two tasks.

Yet, a further analysis was carried out to examine if therewere differences between the two tasks with regard to extendednegotiations toward the MCO instances produced. As shown inTable 5, there were 91 instances of extended negotiations for theMCOs produced on the picture-dictation task and 46 on theopinion-exchange task. ANOVA showed that the picture-dictationtask offered a significantly greater proportion of extended nego-tiations than the opinion-exchange task, F(1, 62) = 12.57, p < .05.

Shehadeh 655

Copyright © 2000. All rights reserved.

Page 30: Comprehensible Output by Ali Sheadeh

Table 5

MCOs Produced After One-Signal and Extended Negotiations on Picture-Dictation and Opinion-Exchange

Type of MCO One-signal Extended-Task Participants produced MCOs negotiation MCOs Total

n % n % nPicture- NS-NNS Other 44 81 10 19 54

dictation NNS-NNS Other 58 64 33 36 91NS-NNS Self 82 80 20 20 102NNS-NNS Self 81 74 28 26 109

Total 265 74 91 26 356Opinion- NS-NNS Other 9 69 4 31 13

exchange NNS-NNS Other 8 62 5 38 13NS-NNS Self 70 79 19 21 89NNS-NNS Self 84 82 18 18 102

Total 171 79 46 21 217

Copyright © 2000. All rights reserved.

Page 31: Comprehensible Output by Ali Sheadeh

Altogether, the results of analysis provided full support forHypothesis 2. The findings showed that the picture-dictation taskprovided significantly greater opportunities than the opinion-exchange task to initiate clarification and produce MCOs. Furtheranalyses revealed that these differences were primarily due todifferences in the frequency of other-initiated clarification re-quests and the proportions of other-initiated MCOs produced. Theoccurrence of self-initiated clarification attempts and the propor-tions of self-initiated MCOs produced were not significantly dif-ferent between the two tasks. Differences between the two taskswere also found in relation to extended negotiations toward theMCO instances produced, with picture-dictation exhibiting asignificantly higher proportion of extended negotiations thanopinion-exchange.

Discussion

The overall results of hypothesis testing showed that in mostcases trends in the predicted direction were observed, although insome cases not at a significant level. First of all, Hypothesis 1(a)showed that NNS partners provided a greater proportion of other-initiated clarification requests than NS partners, despite the factthat the differences were not statistically significant (124 in-stances or 58% vs. 88 instances or 42%, respectively). This findinglends limited support for Varonis and Gass’s (1985) position thatNNS-NNS interaction provides NNSs with greater opportunitiesthan NS-NNS interaction to negotiate meaning (but see discus-sion of Hypothesis 1(c) below).

Unlike prediction (a), which was partly based on previousresearch, I am not aware of any study to date that has investigatedthe proportion of self-initiated clarification attempts in NS-NNSinteraction versus NNS-NNS interaction. For this reason, it willnot be possible to compare the findings of prediction (b) with thoseof other research. This study has revealed that the occurrence ofself-initiated clarification attempts was almost evenly distributedbetween the two types of interaction (212 instances or 49% in

Shehadeh 657

Copyright © 2000. All rights reserved.

Page 32: Comprehensible Output by Ali Sheadeh

NS-NNS interactions and 223 instances or 51% in NNS-NNSinteractions). This suggests that the occurrence of self-initiatedclarification attempts is less likely to be affected by the type of thespeech partner (be it NS or NNS) than that of other-initiatedclarification requests. Confirmation of this result awaits furtherresearch findings.

With regard to prediction (c), my findings showed that asignificantly greater amount of extended negotiation work tookplace in NNS-NNS interactions than in NS-NNS interactions forthe MCOs produced (84 cases vs. 53 cases, respectively). Thisimplies that NNS-NNS interaction provides NNSs not just withinput that they have made comprehensible through negotiation(Varonis & Gass, 1985), but also with more extended negotiationroutines to modify their IL utterances toward comprehensibleoutput than NS-NNS interaction provides them. This result fur-ther supports observations by Tarone and Liu (1995, pp. 120–121)that the interactional context in which L2 learners need to produceoutput which their current IL capacity cannot handle pushes thelimits of their IL system to make it handle the output, a processthat gives rise to more extended negotiations. At the same time,this reflects the pressure placed on NNSs in the NNS-NNS inter-action to stretch and exploit their IL capacity to the limit in orderto make themselves understood, given the limited linguisticknowledge of their NNS interlocutors.

With regard to task, Pica et al. (1989, p. 74) found thatpicture-drawing offered the highest occurrence of NS signals ofrequests for clarification and confirmation in comparison to jigsawand discussion tasks (over 20% of NS c-units on the picture-drawingtask, 11% of NS c-units on the discussion task, and 9% of NSc-units on the jigsaw task). In this respect, the results of Hypothe-sis 2(a) provide further confirmation for Pica et al.’s (1989) findingthat a picture-dictation task offers a significantly higher occur-rence of other-initiated clarification requests than an opinion-exchange task (182 cases or 86% vs. 30 cases or 14%, respectively).

On the other hand, the findings of Hypothesis 2(b) were notconsistent with Pica et al.’s (1989) study, which investigated the

658 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 33: Comprehensible Output by Ali Sheadeh

amount of opportunity each task offers NNSs for the productionof modified output. Pica et al. (1989) found that “the tasks in whichNNSs participated [picture-drawing, discussion, and jigsaw tasks]did not have a significant effect on their modification of theiroutput” (p. 77). Contrary to Pica et al.’s (1989) findings, the resultshere revealed that the picture-dictation task did provide NNSswith significantly more opportunities to produce other-initiatedMCOs than the opinion-exchange task (145 cases or 85% of theMCO instances produced vs. 26 cases or 15%, respectively).

In fact, the finding of Pica et al.’s (1989) study was unex-pected because, as described above, they also found that thepicture-dictation task offered the largest percentage of NS signalsof requests for clarification and confirmation in comparison todiscussion and jigsaw tasks. Indeed, Pica et al. (1989) themselveswere surprised to get this result:

The absence of significant effect for task was surprisingsince we had predicted that NNS control over informationwould be shared with NS during the jigsaw and discussiontasks but would persist throughout the information-gaptask, and that this would influence the signal-responsepattern of our participants. (Pica et al., 1989, p. 78)

However, the finding of the present investigation supportsPica et al.’s (1996) study, which showed that type of task may playvarying roles in providing learners with opportunities in thedirection of comprehensible output. They found that the story-telling task provided NNSs with higher percentages of modifica-tion of their output in both NS-NNS and NNS-NNS interactionsthan the house sequence task.

Differences between the two tasks were also observed withrespect to the proportion of MCOs produced after extended nego-tiation routines. Picture-dictation exhibited a significantly higherproportion of extended negotiations than opinion-exchange. Thismeans that when successful completion of the task depends pri-marily on the NNS supplying accurate and comprehensible output(as in the case of picture-dictation), a higher proportion of ex-tended negotiation routines is more likely to occur.

Shehadeh 659

Copyright © 2000. All rights reserved.

Page 34: Comprehensible Output by Ali Sheadeh

The results of the present study further revealed that thesignificant differences between the two tasks observed in this studywere due to the frequencies of other-initiations and other-initiatedMCOs produced and not those of self-initiations or self-initiatedMCOs produced. Opportunities for the latter were not affected bytype of task, because both tasks provided equally high proportionsof self-initiations and self-initiated MCO instances.Type of task onlyaffected the proportions of other-initiations and instances of other-initiated MCOs produced in favor of picture-dictation.

It is important to note here that this study has revealed thatthe frequencies of self-initiations and self-initiated MCOs pro-duced, unlike those of other-initiations and other-initiated MCOs,are not affected by type of interlocutor (NS or NNS) or type of task(picture-dictation or opinion-exchange). As shown earlier, highproportions of self-initiations and the self-initiated MCOs pro-duced were prevalent in all the contexts and variables examinedin the study. This result provides empirical validation for earlierobservations about the preference for self-initiated, self-completedrepair over other-initiated repair in NS-NS, NS-NNS, and NNS-NNS interactions (Brock et al., 1986; Chun et al., 1982; Day et al.,1984; Gaskill, 1980; Kasper, 1985; Schegloff, 1979; Schegloff et al.,1977; Schwartz, 1980). Indeed, this study has shown not only thatself-initiated self-completed repairs are more favored than other-initiated other-completed repairs, but also that their high occur-rence is constant across all contexts and encounters. If furtherstudies provide support for these findings, then we may haveimportant pedagogical implications for language learning withregard to the importance of self-initiated self-completed repair inthe L2 classroom, in particular that (a) learner-based adjustmentsshould be encouraged over teacher- or peer-based adjustments,and (b) the final objective of repair must be directed towardapproximating NS-NS interactional norms—namely, the predomi-nance of self-initiated self-completed repair.7

660 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 35: Comprehensible Output by Ali Sheadeh

MCO and L2 Learning

Pica et al. (1989, p. 83) maintained that comprehensibleoutput is an outcome of linguistic demands placed on the NNS bythe interlocutors in the course of negotiated interaction. Previousresearch has shown that NNSs tend to modify and reformulatetheir output when their NS speech partners signal an explicit needfor clarification (Nobuyoshi & Ellis, 1993; Pica, 1988; Pica et al.,1989; Pica et al., 1991; Pica et al., 1996; Van den Branden, 1997).The present study has further shown that NNSs encoded theirown modification toward comprehensible output in response toother-initiation and self-initiation, in both NS-NNS and NNS-NNS contexts. As displayed in Table 6, the present study revealedthat out of 212 cases of other-initiated clarification requests acrossboth tasks, NNSs achieved 171 NNS-based MCO instances (81%);and out of 435 cases of self-initiated clarification attempts, theyachieved 402 NNS-based MCO instances (92%).

These results are important in light of theoretical conclusions(Lyster & Ranta, 1997; Pica, 1988; Pica et al., 1989) that the NNS’sability to accomplish self-adjusted comprehensible output (i.e.,NNS-based MCO) rather than other-adjusted comprehensible output (i.e.,interlocutor-based MCO) is evidence that supports Swain’s (1985,1993) claim that the comprehensible output forces the learner tomove from semantic analysis of the TL to a syntactic analysis of it.And according to Swain (1995) and Swain and Lapkin (1995), whenlearners move from semantic analysis to syntactic analysis of the TL(i.e., when they attempt to produce the L2), they become aware of(notice) a gap in their current IL performance. Swain (1995, 1998)and Swain and Lapkin (1995) have consistently argued that whenlearners reprocess and modify their current performance to make itmore enhanced, they are engaged in mental processes that are partof the process of language learning. For instance, Swain and Lapkin(1995,p.374) concluded that “‘pushing’ learners beyond their currentperformance level can lead to enhanced performance, a step whichmay represent the internalization of new linguistic knowledge,or theconsolidation of existing knowledge.”

Shehadeh 661

Copyright © 2000. All rights reserved.

Page 36: Comprehensible Output by Ali Sheadeh

Swain and Lapkin (1995) based their conclusion on thefindings of a recent introspection study they conducted in whichthey tried to shed more focussed light on the processes andmechanisms that L2 learners follow to reprocess and modify theirIL utterances. The researchers sought “to try to arrive at themental processes . . . reflected in the changes students made totheir output” (p. 381). They examined the ability of 18 eighth gradeimmersion students learning French to consciously reprocesstheir IL output without any sort of external feedback when facedwith a performance problem. The task given to the students wasto write a report on some environmental problem. The students

Table 6

Frequencies, Proportions, and Percentages of Outcome Types inRelation to Other-Initiated Clarification Requests andSelf-Initiated Clarification Attempts

Other-initiated clarification requestsTask n Outcome

MCOs produced Other outcometypesa

n % n %Picture-dictation 182 145 80 37 20Opinion-exchange 30 26 87 4 13

Total 212 171 81 41 19

Self-initiated clarification attemptsTask n Outcome

MCOs produced Other outcometypesa

n % n %Picture-dictation 232 211 91 21 9Opinion-exchange 203 191 94 12 6

Total 435 402 92 33 8

aOther outcome types: outcomes not resulting in NNSs’ production of MCO(failing to repair, appealing for help, expressing difficulty in repairing theintended message, repeating the trouble-source utterance withoutmodification, switching to a new topic).

662 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 37: Comprehensible Output by Ali Sheadeh

were instructed to think aloud while writing, and especially whenthey were faced with a problem.

Swain and Lapkin (1995) found that there were 190 occasions inwhich students encountered a linguistic problem in their output. Ineachcasethestudents forcedthemselvestomodifytheiroutputtowardgreater comprehensibility. That is, communicative needs forced stu-dents to move from semantic analysis of the language to syntacticanalysis of it. Swain and Lapkin argued that “on each occasion, thestudents engaged in mental processing that may have generatedlinguistic knowledge that is new for the learner, or consolidatedexisting knowledge” (p. 384). In other words, it was argued that in theprocess of modifying their IL utterances toward greater messagecomprehensibility, L2 learners are engaged in some restructuring ofthe system that affects their access to the knowledge base, and thatthis restructuring process is part of second language learning.

If these arguments by Swain (1995, 1998) and Swain andLapkin (1995) are valid, the present findings provide empiricalsupport for Swain’s (1985,1993) earlier theoretical conclusion thatcomprehensible output—like comprehensible input—is a mecha-nism that plays a role in promoting SLA. Accordingly, it will bepossible to argue that the function of L2 learners’ production isnot just that of indirectly generating more comprehensible input,as claimed by, for instance, Krashen (1982, 1985, 1994), but alsofacilitating second language development by providing the learnerwith an opportunity to produce comprehensible output achievedby reprocessing and modifying his/her current IL capacity towardsuccessful and more accurate use of the target language. By thesame token, it is possible to argue further that negotiated inter-actions are important not just because they provide NNSs with anopportunity to receive input that they have made comprehensiblethrough negotiation (Deen, 1995; Doughty, 1992; Gass & Varonis,1994; Loschky, 1994; Varonis & Gass, 1985), but also because theyprovide them with an opportunity for IL modification in thedirection of comprehensible output, as shown by this and otherstudies (e.g., Pica et al., 1996; Tarone & Liu, 1995; Van denBranden, 1997). Based on these considerations, Figure 2 shows a

Shehadeh 663

Copyright © 2000. All rights reserved.

Page 38: Comprehensible Output by Ali Sheadeh

possible link between negotiation of meaning, comprehensibleinput, comprehensible output, and language learning.

It is important to note, though, that this model involves somespeculation because no one has yet shown clearly if there is (or ifthere is not) a direct relationship between interaction with L2learners, their comprehension of input data, feedback on their ILutterances,and their (modified) performance,on the one hand,andL2 acquisition on the other (Pica, 1994, p. 507). However, Swainand Lapkin (1995, p. 373) speculated that “although no one hasyet shown directly that these modified, or reprocessed, responsesare maintained in the learner’s interlanguage, the assumption isthat this process of modification contributes to second language

Opportunity for less competent NNSs to providefeedback on their current comprehension abilityin the L2, enabling their interlocutors to make thenecessary simplifications and adjustments totheir production

Opportunity for NNSs to expand their currentinterlanguage capacity in response to other-initiated clarification requests and self-initiatedclarification attempts, to test out hypothesesabout the L2, and to move from a purelysemantic analysis of the L2 to a more syntacticanalysis of it

Comprehensible outputComprehensible input

Modifying input to NNSs(provides)

Modifying output by NNSs(generates)

Language learning

Communication that involvesnegotiation of meaning

Figure 2. A link between negotiation of meaning, comprehensible input,comprehensible output, and language learning.

664 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 39: Comprehensible Output by Ali Sheadeh

acquisition.” The present model is based on the underlying as-sumption that when learners stretch their current IL capacity andmodify their performance to meet communicative demands, theyare engaged in “cognitive processes that have been implicated insecond language learning,” to use Swain’s (1995, p. 130) words.

Conclusion

Perhaps the most important conclusion of the findings of thisstudy is that the role of L2 learners’ output should be extendedbeyond just being a source of obtaining feedback in order to generate(more) comprehensible input. If this conclusion is valid, then thepresent study provides empirical validation for previous theoreticalclaims (e.g., Hatch, 1978; Hatch et al., 1986; Lyster & Ranta, 1997;Sato, 1986; Swain, 1985, 1993; Tarone & Liu, 1995); and it supportsthe findings of existing experimental and quasi-experimental re-search (Nobuyoshi & Ellis, 1993; Pica, 1988; Pica et al., 1989; Pica etal., 1996; Swain & Lapkin, 1995; Van den Branden, 1997) into thesignificance of L2 learners’ output in language learning. The otherconclusion is that if interlanguage modification leading to compre-hensibility is found to be integral to successful L2 learning, it is notonly other-initiated clarification requests that matter, but moreimportantly the extent to which self-initiated self-completedmodification is a normal learning/teaching strategy because itshigh occurrence is more prevalent and more constant across allvariables and contexts, as shown by this study.

However, more focussed empirical research, introspectionstudies, and retrospective interviews are needed into this seem-ingly important area, which may ultimately reveal that successfulsecond language learning depends on ways of comprehendinginput and on ways in which language production relates to acqui-sition and learning. If further studies provide support for thepresent findings and these assumptions, learners’ output shouldbe given a more active role in second language learning.

Revised version accepted 10 May 1999

Shehadeh 665

Copyright © 2000. All rights reserved.

Page 40: Comprehensible Output by Ali Sheadeh

Notes

1Unlike Varonis and Gass’s (1985) model (which considered other-initiationsonly), the present model takes into consideration both other-initiation andself-initiation of repair.2Aston (1986, p. 139) pointed out that sometimes interlocutors do not reactto a trouble-source (i.e., they feign comprehension) in order to keep theconversation going, reaffirm satisfactory communication, and maintain asatisfying rapport.3A Language Learning anonymous reviewer pointed out that comprehensibleoutput may not necessarily always be correct, or vice versa. This is becauseinterlocutors’ reactions are sometimes unreliable indicators of comprehen-sion or noncomprehension (Aston, 1986). To an extent, this is true. However,it is assumed here that the adjustments that NNSs make to their perfor-mance toward comprehensibility will be achieved by their attempts to providemore accurate/correct and more comprehensible output (see also Lyster &Ranta, 1997, p. 47; Spada & Fröhlich, 1995, p. 25, for a similar argument).Furthermore, it is believed that NNSs’ attempts to modify and adjust theiroutput (whether to make it more accurate/correct or more comprehensible)will expand their current IL capacity (Pica et al., 1996); and the underlyingassumption of this study in Swain and Lapkin’s (1995, p. 374) terms is that“‘pushing’ learners beyond their current performance level can lead to en-hanced performance, a step which may represent the internalization of newlinguistic knowledge, or the consolidation of existing knowledge.” Neverthe-less, isolating and examining the proportion of comprehensible but incorrect/lessaccurate output versus correct/more accurate but less comprehensible output,and the proportion of output that is more targetlike versus output that is lesstargetlike, merits an independent and more detailed analysis of NNSs’adjusted/modified output. This will be the subject of a future paper.4It is well established that explicit corrections and confirmation requests tendnot to elicit NNS-generated repair, and thus they are not as conducive tomodification of NNS output as clarification requests (see, for example, Aston,1986; Lyster & Ranta, 1997; Pica et al., 1989). For this reason, these wereexcluded from analysis in this study.5Some initiations (other and self) gave rise to other outcome types (those notresulting in MCO), including failure to repair, appealing for help (direct andindirect), expressing difficulty in repairing the intended message, repeatingthe trouble-source utterance without modification,or switching to a new topic(see the Appendix). Examining the specific proportions, frequencies, andpercentages of these outcome types falls outside the scope of the presentanalysis. These were not reported here, therefore.6I did not consider instances of initiation or the MCOs produced by individualdyads because the current study investigated the overall effect of type ofspeech partner (NS vs. NNS) and type of task (picture-dictation vs. opinion-

666 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 41: Comprehensible Output by Ali Sheadeh

exchange) on NNSs’ ability to adjust their IL performance in response toother- and self-initiations.7It must be noted that to some extent this depends on the proficiency level ofthe learner. For instance, Lyster and Ranta (1997, p. 58) pointed out that this“is . . . feasible in L2 classrooms only where learners already possess anadequate level of proficiency.”

References

Aston, G. (1986). Trouble-shooting in interaction with learners: The more themerrier? Applied Linguistics, 7, 128–143.

Brock, C., Crookes, G., Day, R., & Long M. (1986). The differential effects ofcorrective feedback in native speaker–nonnative speaker conversation. InR. Day (Ed.), Talking to learn (pp. 229–236). Rowley, MA: Newbury House.

Chiang, C., & Dunkel, P. (1992). The effect of speech modification, priorknowledge and listening proficiency on EFL lecture learning. TESOLQuarterly, 26, 345–374.

Chun, A., Day, R., Chenoweth, N., & Luppescu, S. (1982). Types of errorscorrected in native-nonnative conversations. TESOL Quarterly, 16,537–548.

Crookes, G., & Gass, S. (Eds.). (1993a). Tasks and language learning: Integrat-ing theory and practice. Clevedon, UK: Multilingual Matters.

Crookes, G., & Gass, S. (Eds.). (1993b). Tasks in a pedagogical context:Integrating theory and practice. Clevedon, UK: Multilingual Matters.

Crookes, G., & Rulon, K. (1988). Topic and feedback in native speaker/non-native speaker conversation. TESOL Quarterly, 22, 675–681.

Day, R., Chenoweth, N., Chun, A., & Luppescu, S. (1984). Corrective feedbackin native-nonnative discourse. Language Learning, 34, 19–45.

Deen, J. (1995). Dealing with problems in intercultural communication: Astudy of negotiation of meaning in native/nonnative speaker interaction.Unpublished doctoral dissertation, University of Groningen, Groningen,The Netherlands.

Dekeyser,R. (1998).Beyond focus on form:Cognitive perspectives on learningand practicing second language grammar. In C. Doughty & J. Williams(Eds.),Focus on form in classroom second language acquisition (pp.42–63).Cambridge: Cambridge University Press.

Doughty, C. (1992). Second language instruction does make a difference:Evidence from an empirical study of SL relativization. Studies in SecondLanguage Acquisition, 13, 431–470.

Doughty, C., & Williams, J. (1998). Pedagogical choices in focus on form.In C. Doughty & J. Williams (Eds.), Focus on form in classroom second

Shehadeh 667

Copyright © 2000. All rights reserved.

Page 42: Comprehensible Output by Ali Sheadeh

language acquisition (pp. 197–261). Cambridge: Cambridge UniversityPress.

Duff, P. (1986). Another look at interlanguage talk: Taking task to task. InR. Day (Ed.), Talking to learn (pp. 147–181). Rowley, MA: Newbury House.

Færch, C., & Kasper, G. (1983). On identifying communication strategies ininterlanguage production. In C. Færch & G. Kasper (Eds.), Strategies ininterlanguage communication (pp. 210–238). London: Longman.

Gaskill, W. (1980). Correction in native speaker-nonnative speaker conversa-tion. In D. Larsen-Freeman (Ed.), Discourse analysis in second languageresearch (pp. 125–137). Rowley, MA: Newbury House.

Gass, S., & Varonis, E. (1985). Task variation and nonnative/nonnativenegotiation of meaning. In S. Gass & C. Madden (Eds.), Input in secondlanguage acquisition (pp. 149–161). Rowley, MA: Newbury House.

Gass,S.,& Varonis,E. (1986).Sex differences in nonnative speaker-nonnativespeaker interactions. In R. Day (Ed.), Talking to learn (pp. 327–352).Rowley, MA: Newbury House.

Gass, S., & Varonis, E. (1994). Input, interaction and second language produc-tion. Studies in Second Language Acquisition, 16, 283–302.

Harley, B. (1998). The role of focus-on-form tasks in promoting child L2acquisition. In C.Doughty & J. Williams (Eds.),Focus on form in classroomsecond language acquisition (pp. 156–174). Cambridge: Cambridge Uni-versity Press.

Hatch, E. (1978). Discourse analysis and second language acquisition. InE. Hatch (Ed.), Second language acquisition: A book of readings(pp. 401–435). Rowley, MA: Newbury House.

Hatch,E.,Flashner,V.,& Hunt,L. (1986).The experience model and languageteaching. In R.Day (Ed.),Talking to learn (pp.5–22).Rowley,MA:NewburyHouse.

Hawkins, B. (1985). Is an “appropriate response” always so appropriate? InS. Gass & C. Madden (Eds.), Input in second language acquisition(pp. 162–178). Rowley, MA: Newbury House.

Holliday, L. (1995). NS syntactic modifications in NS-NNS negotiations asinput data for second language acquisition of syntax.Unpublished doctoraldissertation, University of Pennsylvania, Philadelphia.

Kasper, G. (1985). Repair in foreign language teaching. Studies in SecondLanguage Acquisition, 7, 200–215.

Krashen, S. (1982). Principles and practice in second language acquisition.Oxford: Pergamon.

Krashen, S. (1985). The input hypothesis: Issues and implications. London:Longman.

668 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 43: Comprehensible Output by Ali Sheadeh

Krashen, S. (1994). The input hypothesis and its rivals. In N. Ellis (Ed.),Implicit and explicit learning of languages (pp. 45–77). London: AcademicPress.

Long, M. (1983a). Native speaker/non-native speaker conversation and thenegotiation of comprehensible input. Applied Linguistics, 4, 126–141.

Long, M. (1983b). Native speaker/non-native speaker conversation in thesecond language classroom. In M. Clarke & J. Handscombe (Eds.), OnTESOL ’82: Pacific perspectives on language learning and teaching(pp. 207–225). Washington, DC: TESOL.

Long, M. (1996). The role of the linguistic environment in second languageacquisition. In W. Ritchie & T. J. Bhatia (Eds.), Handbook of secondlanguage acquisition (pp. 413–468). Orlando, FL: Academic Press.

Long, M., & Robinson, P. (1998).Focus on form: Theory, research, and practice.In C. Doughty & J. Williams (Eds.), Focus on form in classroom secondlanguage acquisition (pp. 15–41). Cambridge: Cambridge UniversityPress.

Loschky, L. (1994). Comprehensible input and second language acquisition:What is the relationship? Studies in Second Language Acquisition, 16,303–323.

Lynch, A. J. (1988). Grading foreign language listening comprehension mate-rials: The use of naturally modified interaction. Unpublished doctoraldissertation, University of Edinburgh, Scotland.

Lyster, R., & Ranta, L. (1997). Corrective feedback and learner uptake:Negotiation of form in communicative classrooms. Studies in SecondLanguage Acquisition, 19, 37–66.

Nobuyoshi, J., & Ellis, R. (1993). Focused communication tasks and secondlanguage acquisition. ELT Journal, 47, 203–210.

Pica, T. (1982). Second language acquisition in different language contexts.Unpublished doctoral dissertation, University of Pennsylvania, Philadel-phia.

Pica, T. (1988). Interlanguage adjustments as an outcome of NS-NNS nego-tiated interaction. Language Learning, 28, 45–73.

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

Pica, T., Doughty, C., & Young, R. (1986). Making input comprehensible: Dointeractional modifications help? I.T.L. Review of Applied Linguistics, 72,1–25.

Pica, T., Holliday, L., Lewis, N., Berducci, D., & Newman, J. (1991). Languagelearning through interaction: What role does gender play? Studies inSecond Language Acquisition, 13, 343–376.

Shehadeh 669

Copyright © 2000. All rights reserved.

Page 44: Comprehensible Output by Ali Sheadeh

Pica, T., Holliday, L., Lewis, N., & Morgenthaler, L. (1989). Comprehensibleoutput as an outcome of linguistic demands on the learner. Studies inSecond Language Acquisition, 11, 63–90.

Pica, T., Kanagy, R., & Falodun, J. (1993). Choosing and using communicationtasks for second language research and instruction. In G. Crookes &S. Gass (Eds.), Tasks and language learning: Integrating theory andpractice (pp. 9–34). Clevedon, UK: Multilingual Matters.

Pica,T.,Lincoln-Porter,F.,Paninos,D.,& Linnell,J. (1996).Language learners’interaction: How does it address the input, output, and feedback needs ofL2 learners? TESOL Quarterly, 30, 59–84.

Pica, T., Young, R., & Doughty, C. (1987). The impact of interaction oncomprehension. TESOL Quarterly, 21, 737–758.

Sato, C. (1986). Conversation and interlanguage development: Rethinkingthe connection. In R. Day (Ed.), Talking to learn (pp. 23–45). Rowley, MA:Newbury House.

Schegloff, E. (1979). The relevance of repair to syntax-for-conversation. InT. Givon (Ed.), Syntax and semantics. Vol. 12: Discourse and syntax(pp. 261–286). New York: Academic Press.

Schegloff, E., Jefferson, G., & Sacks, H. (1977). The preference for self-correction in the organization of repair in conversation. Language, 53,361–382.

Schmidt, R. (1995). Consciousness and foreign language learning: A tutorialon the role of attention and awareness in learning. In R. Schmidt (Ed.),Attention and awareness in foreign language learning (pp. 1–63). Hono-lulu: University of Hawaii Press.

Schwartz, J. (1980). The negotiation of meaning: Repair in conversationsbetween second language learners of English. In D.Larsen-Freeman (Ed.),Discourse analysis in second language research (pp.138–153).Rowley,MA:Newbury House.

Shehadeh, A. (1991). Comprehension and performance in second languageacquisition: A study of second language learners’ production of modifiedcomprehensible output. Unpublished doctoral dissertation, University ofDurham, England.

Spada, N., & Fröhlich, M. (1995). COLT. Communicative orientation of lan-guage teaching observation scheme: Coding conventions and applications.Sydney, Australia: National Center for English Language Teaching andResearch.

Swain, M. (1984). A review of immersion education in Canada: Research andevaluation studies. ELT Documents, 119, 35–51.

Swain, M. (1985). Communicative competence: Some roles of comprehensibleinput and comprehensible output in its development. In S. Gass &

670 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 45: Comprehensible Output by Ali Sheadeh

C. Madden (Eds.), Input in second language acquisition (pp. 235–253).Rowley, MA: Newbury House.

Swain, M. (1993). The output hypothesis: Just speaking and writing aren’tenough. The Canadian Modern Language Review, 50, 158–164.

Swain, M. (1995). Three functions of output in second language learning. InG. Cook & B. Seidlhofer (Eds.), Principle and practice in applied linguis-tics: Studies in honor of H.G. Widdowson (pp. 125–144). Oxford: OxfordUniversity Press.

Swain, M. (1998). Focus on form through conscious reflection. In C. Doughty& J. Williams (Eds.), Focus on form in classroom second language acqui-sition (pp. 64–81). Cambridge: Cambridge University Press.

Swain, M., & Lapkin, S. (1995). Problems in output and the cognitive pro-cesses they generate: A step towards second language learning. AppliedLinguistics, 16, 371–391.

Tarone, E., & Liu, G. Q. (1995). Situational context, variation, and secondlanguage acquisition theory. In G. Cook & B. Seidlhofer (Eds.), Principleand practice in applied linguistics: Studies in honor of H.G. Widdowson(pp. 107–124). Oxford: Oxford University Press.

Van den Branden, K. (1997). Effects of negotiation on language learners’output. Language Learning, 47, 589–636.

Varonis, E., & Gass, S. (1985). Non-native/non-native conversations: A modelfor negotiation of meaning. Applied Linguistics, 6, 71–90.

Wright, R. (1996). A study of the acquisition of verbs of motion by grade 4/5early French immersion students. Canadian Modern Language Review,53, 257–280.

Appendix

The First Mechanism: Trouble-Source—Other-Initiation—Outcome—Reaction to Outcome

Underlined items refer to trouble-source signal and italicizeditems refer to the outcome. PD = picture-dictation task; OE =opinion-exchange task. Numbers in parentheses refer to thelength, in seconds, of a pause or cutoff within a speech turn orbetween speech turns.

1. NNS trouble-source

Shehadeh 671

Copyright © 2000. All rights reserved.

Page 46: Comprehensible Output by Ali Sheadeh

2. Other-initiated clarification requests (NS or NNS speech part-ner signal of total or partial lack of understanding follows NNStrouble-source)a. Indication of partial or complete nonunderstanding(PD) NNS: annd on the table you have two flask

NS: (1.0) flask?b. Partial repetition of the trouble-source plus a question word(PD) NNS: on the table there’s two fask . . .

NS: two two what?c. Request for elaboration, expansion, or paraphrasing(PD) NNS: two fas fas

NS: what is it for?d. Explicit indication of nonunderstanding(OE) NNS: both of them is not hope this thing divorce happen

NS: I didn’t understand sorry3. Outcomea. The NNS may ignore the signal to repair(PD) NNS: (6.0) annd on the back have two pillow

NS: at the back?NNS: (6.0) and between the window and towel above the

tableb. Failing to repair(PD) NNS: annd there is a hannn

NS: a what?NNS: a hannkerNS: what is it for?NNS: which hanging coat and jacket a hannkkerNS: a hanger?NNS: yes (0.8) err also there is two two . . .

c. Expressing difficulty in responding to the signal(PD) NS: which corner?

NNS: err the pigture is not sure of the er which cornerd. Inserting new, but not directly relevant information(PD) NS: can you see the door er full view?

NNS: it’s err rectangular door in the whole

672 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 47: Comprehensible Output by Ali Sheadeh

e. Repeating the original trouble-source utterance withoutmodification(PD) NNS: there emm (1.0) they put it in paile

NS: sorry?NNS: paile (0.8) understand?

f. Confirmation or acknowledgment of signal only(PD) NNS: you have a plaete

NS: a plaete?NNS: yes

g. Modification of TS utterance (producing other-initiated, self-completed MCO)

i. Phonological modification(PD) NNS1: it’s a betroom

NNS2: a bed whom?NNS1: a bedroom a bedroomii. Morphosyntactic modification

(PD) NNS1: two small bottleNNS2: two small what?NNS1: bot (1.0) small bottlesiii. Semantic modification

1. Synonyms and paraphrasing(PD) NNS2: a table?

NNS1: no towel er handkerchief er for washing or bathing2. Substitution

(PD) NNS2: above?NNS1: you have yeah above on the top of the

3. Paraphrase and/or description(PD) NNS2: wha’ is it?

NNS1: desk bedt bed bed bed is used for to sleep in4. Reaction to outcomea. Comprehension signal(PD) NNS1: annnd two small bottle

NNS2: two small what?NNS1: bot (0.6) small bottlesNNS2: yeah

Shehadeh 673

Copyright © 2000. All rights reserved.

Page 48: Comprehensible Output by Ali Sheadeh

b. Continuation move(PD) NS: is the bed at the top or the bottom of the picture?

NNS: emm (2.0) it’s on the right side of the pictureNS: right carry on

The Second Mechanism: Trouble-Source—Self-Initiation—Outcome

1. Trouble-sourcea. Phonological(PD) NNS: there is a dtoor doorb. Lexical(PD) NNS: on the right of the fo (0.8) of the picture there’sc. Morphosyntactic(OE) NNS: but he want wanted her to be at homed. Referential(PD) NNS: (1.0) on the western (0.9) on the eastern wall2. Self-initiated clarification attemptsa. Lexical means(PD) NNS: just in front of me (0.7) I mean behind the doorb. Nonlexical means

i. Self-initiation of repair exhibited by a pause(OE) NNS: the wife always go out (0.6) goes out and left his his

husband eh (1.0) her husbandii. Self-initiation signalled by a cutoff

(PD) NNS: just to describe just the woo the direct walliii. Self-initiation of repair signalled by the use of “eh”

(OE) NNS: they have to be eh (1.0) to get divorcediv. Self-initiation signalled by the use of “er”

(OE) NNS: to take er err to have some independence3. The outcomea. Failing to self-repair(PD) NNS: beside the door there is a there is a cloat clo clo

NS: a coat?b. Indirect appeal for help(PD) NNS: yes a glass (3.0) another thing which name in

English I don’t know but common in cooksc. Direct appeal for help

674 Language Learning Vol. 49, No. 4

Copyright © 2000. All rights reserved.

Page 49: Comprehensible Output by Ali Sheadeh

(PD) NNS: what you call on the wood before putting somethinghanging something?

d. Explicit expression of difficulty in repairing/communicatingthe message(OE) NNS: not not emm I don’t know how to say (laughs)e. Repeating the original trouble-source utterance or part of itwithout modification(PD) NNS: there’s two fas (1.0) two fas and a cupf. Modification of TS utterance (producing self-initiated, self-completed MCO)

i. Phonological modification(PD) NNS: on the left-hand side there is a dtoor door

ii. Syntactic modification(OE) NNS: and her son in the home (0.7) at home it’s it’s not

iii. Morphological modification(OE) NNS: the wife er just only want wanted to to divorce

iv. Semantic modification1. Using synonyms

(PD) NNS: I think it’s also it’s picture or postcard2. Exemplification

(OE) NNS: in our case in Somalia emm always man likesexactly that the way he thinks but eh for the womenemm they do all the activity (0.8) for example if they arefarmers they look after the animals

3. Reordering and/or restructuring(OE) NNS: but the question of independence for me is not

emm (1.0) I don’t agree with the people

Shehadeh 675

Copyright © 2000. All rights reserved.