.w... vu.’ in... IE x. . H. 3:? arm.» I. vu {:5}... v. ' .4): it. tox .. 3. L5XE?Y, ‘ 1. tr A.. a... . a . a. 5 )3 l . L.‘I.zhfi: . v I it Sure :23! .J , 45“....22" ;nw..lu........mr E! A! I... . a I‘ltav’lvtl 1 N...atfb.5|IY0..l\ ._ .1; a»: vu. ‘til . 3:3. 3: 43.11. 2.3!: 1 ‘ $5.4? .. v :33. 2. 1...: A . . .51.th , b- 1.. L. 7... , .....Lr tum»... ' 1. O. (3.3.. u 1 I. .u..? Vin..- ’ HES‘S SITY LIBRARIES lllllllllmlllWlW“lllllllllllHIWl‘ H Hill 3 1293 01555 This is to certify that the thesis entitled NEGOTIATION OF MEANING IN NON-NATIVE/NON-NATIVE INTERACTIONS presented by Sungsoo Jang has been accepted towards fulfillment of the requirements for M.A. degree in TESOL Major professor 08/23/96 Date 0-7639 MS U is an Affirmative Action/Equal Opportuniiy Institution LIBRARY Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE pus MAG} C 9. <7 “- DATE DUE DATE DUE MSUleAnNflnnltlvertlon/E omlmwon ‘ M Wane-9.1 Negotiation of Meaning in Non-native/Non-native Interactions By Sungsoo Jang A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of English 1996 AB STRACT NEGOTIATION OF MEANING IN NON-NATIVE/ NON-NATIVE INTERACTIONS By Sungsoo Jang This thesis attempts to expand the insights into the role of negotiation of meaning in second language acquisition (SLA) on two dimensions. First, such variables as task type, Ll background and cultural distance between interlocutors were tested for their effects on negotiation in conversation between non-native speakers (NNS) in order to see how negotiation works in NNS/NNS conversation, compared to native-speaker(NS) involving conversation. Second, it was examined whether and how negotiation facilitates SLA in terms of not only comprehension but feedback and production by classifying negotiation into two types of self- and other-motivated according to its initiator. The results revealed that (l) NNS/NNS contexts allow negotiation to contribute more favorably to SLA than NS/NNS settings and (2) the role of negotiation is not limited to enhancement of comprehension but varies according to its type. Copyright by SUNGSOO JANG 1996 ACKNOWLEDGMENTS I would like to thank Dr. Susan Gass, Dr. Dennis Preston and Dr. Alison Mackey for their comments on the contents of this thesis and suggestion on editing. I would also like to thank the twenty-four Michigan State University ELC students who participated in the research as subjects for their cooperation. I would also give thanks to Jiyoung Kang for her help in the process of transcribing and coding the data. TABLE OF CONTENTS Comprehension and Second Language Acquisition ................. 1 Comprehension and Negotiation of Meaning ......................... 5 Review of Studies on Negotiation of meaning ....................... 12 Definition and Categorization of Negotiation of Meaning ......................................................................... 12 Negotiation of Meaning across Variables ........................ 17 Two Issues of Concern .................................................... 24 Negotiation of Meaning within NS/NNS vs NNS/NNS Conversational Settings .............................................. 3O Negotiation of Meaning as Facilitative to Other Necessary Conditions for SLA .................................................... 39 The Purpose of Research .................................................. 43 Research Design ............................................................. 46 Subjects .................................................................. 46 Procedure ................................................................ 48 Operational Categorization of Negotiation ..................... 50 Data Coding ............................................................. 58 Variables and Predictions ............................................ 6O (1) Task type ..................................................... 6O (2) L1 background ............................................... 62 (3) Task type and L1 background ............................ 64 (4) Cultural distance ............................................ 64 Results and Discussion ................................................... 69 Hypothesis 1 ...................................................... ' ...... 69 Hypothesis 2 ............................................................ 71 Hypothesis 3 ............................................................ 76 Hypothesis 4 ............................................................ 76 Response ................................................................. 79 vii Summary of Results ........................................................ 83 Conclusion and Implications for Classroom and Further Research ............................................................................. 85 Appendix ...................................................................... 91 List of References .......................................................... 92 viii Table Table Table Table Table Table Table Table LIST OF TABLES Beneficiaries from the Use of Indicators ............... 55 Beneficiaries from Response to Indicators ............. 58 Paired t-test Comparing Numbers of Indicators in P and F .............................................................. 70 Two way t-test comparing Numbers of Indicators occurring in S and D .......................................... 72 Two way t-test Comparing Numbers of Indicators Used in K and non-K Dyads ........................... , ............ 76 Two way t-test Comparing Indicators by Koreans in K/K, K/A and K/N Dyads ...................... , ............. 77 Percentages of Responses in Indicators .................. 80 Percentages of ‘R’s to ‘I’s and of CR/VM in ‘I’s used by Korean ........................................................ 82 ix LIST OF FIGURES FIGURE 1 - Pictures for Information-exchange Task ............. 92 Comprehension and Second Language Acquisition Although every approach to second language acquisition (SLA) has agreed on the necessity of input for SLA, there has been some controversy on how critical a role input plays on SLA among these approaches. From the perspective of behaviorists, input is seen as a target itself to be acquired and the mechanism for acquisition is imitation of the input. In their view, the extent of a learner’s acquisition depends upon how much input the learner is exposed to and imitates. Thus, what is important to language learning is to make sure that the learner is exposed to as much input as possible. However, for those who believe , in the working of Universal Grammar (UG) as mechanism for language acquisition, input has been viewed only as necessary to trigger the very UG system, not as target for language learning. According to the UG approach, everyone is born with the UG system which contains limited possibilities of ways languages work. Each language represents only one of them. Input is only needed to the extent that it helps identify which of the possibilities the target language (TL) represents. Once the kind of possibility is identified for the target language, then UG does the rest of the laborious task of language acquisition. What still remains controversial within this frame of theory is 2 whether UG is available to adult learners learning a second language (L2) and if it is, whether it works in the same way it does in children acquiring their first language (L1). If UG is available but works differently across L1 and L2 learning, the role of input would go beyond triggering UG in adult L2 learning. (For discussions of availability of UG to L2 Learning see Bley—Broman, 1989; Schacter, 1988. 1989. 1990; White, 1989) In spite of these different views on the role of input (i.e., as target versus as means for acquisition), the difference is basically on how much input, not on what kind of input, is necessary for acquisition. This is the motive for this thesis to seek to find the best type of input for language learning and the contexts in which the type of input is provided. Corder (1967) introduced the term ‘intake’ to elaborate the notion of ‘input.’ He defined ‘input’ as ‘what is available for going in’ and ‘intake’ as ‘what goes in.’ That is, input is what is available to the learner whereas intake is what is internalized in the learner’s mind as a part of TL knowledge. What is important about his distinction is that it implies that only internalized input (i.e., intake) can be integrated into the learner’s language system and thus input must be comprehended into intake in order 3 for it to become a part of the learner’s acquired language. In other words, the value of input for SLA lies not in the provision of it but in the comprehension of it. The approach which stresses the significance of comprehension shed a new light on SLA in that it suggested a possibility that language learning is controlled by the learner, not by the input provider because it is the learners who comprehend input. The importance of comprehension was theoretically supported by Krashen’s Input Hypothesis (1980). He explicitly argued that if input is to be facilitative to SLA, it must be comprehensible to the learner. He defined ‘comprehensible input’ as a bit of language at one stage above the learner’s current stage of L2 knowledge. In his view, if a learner receives enough comprehensible input, the learner’s grammatical knowledge develops automatically. Thus comprehensible input is all that learners need for acquisition. Later (1983) he specified how comprehensible input automatically leads to the development of the learner’s L2 grammar. First, the learner understands the meaning of comprehensible input by using contexts in which the input is provided - the learner can understand because by definition the input is comprehensible even though it is not within the 4 learner’s current language system - and tries to connect the meaning of the new input with his or her current interlanguage (IL) grammar. Second, the learner notices the gap between the meaning and the form of the new input, and recognizes the form as a new L2 grammatical form. Third, if this newly recognized form occurs with frequency, then this new form is incorporated into the learner’s IL system. The claims in Krashen’s Hypothesis, however, have not been tested for their validity and remain as nothing but theoretical inference, causing a lot of criticism. Among the critics, Chaudron (1985) pointed out that it is unclear how to determine the learner’s current state. White (1987) stated that the context is not the only path to changes in the learner’s grammar - the changes can be brought about by explicit instructions on the learner’s incorrect IL forms in some cases. The biggest challenge has been against Krashen’s claim that comprehensible input is a necessary and sufficient condition for SLA. To argue against this, conditions other than comprehensible input have been suggested as necessary for SLA. In spite of this criticism, Krashen’s theory has been highly influential in the field in that it has brought to light the importance of comprehension to SLA. All other necessary 5 conditions claimed after his theory were based upon the recognition of the role of comprehension and comprehensible input in SLA. And at least the necessity of comprehensible input has not been denied in any other study, though its sufficiency has been questioned by later critics. Comprehension and Negotiation of Meaning Given its solid facilitative role, the concern for comprehensible input has centered around the question of in what context the learner increases chances of being exposed to more of it. Input from native speakers (NS) of TL as the source for comprehensible input is not necessarily limited to that directed toward the learner. All kinds of input, directed to the learner or not, can be a source for comprehensible input to the learner. However, in the situation where NS speech is directed to the learner and where the learner is involved in conversation with the NS, chances are greater that the learner is exposed to more comprehensible input. There are two reasons for this. First, NSs tend to modify their speech to their NNS interlocutor’s proficiency level in order to enhance the comprehensibility. Ferguson (1971) termed this kind of modified speech ‘foreigner talk.’ The features of foreigner talk include avoidance of slang, idioms and low frequency words, 6 repetition and elaboration of what was previously said, slow and louder speech, etceteras (Gass & Varonis,1985; Hatch,l983; Parker and Chaudron,l987). Krashen (1980) pointed out that, since the NS cannot know precisely what the learner’s current state is, the modified input is not always fine-tuned to the learner’s proficiency and comprehensible to the learner. But, even if roughly tuned, chances are that those modifications made under consideration of the learner’s deficiency in L2 may increase the possibility of supplying more comprehensible input to the learner. In fact, Krashen considered foreigner talk as an ideal source for comprehensible input in that it is easy to understand due to its features and it is not intended to be fine- tuned. He believed input that is assumed to be fine-tuned for comprehensible input most likely ends up being irrelevant to the learner’s current state because only the learner knows his or her current state exactly and can judge whether input is comprehensible; therefore, it is pointless for the input provider to make input fine-tuned. The other reason for conversation being a good setting for comprehensible input involves the possibility that the learner can take actions on his or her own to obtain comprehensible input if the input is way beyond his or her current state. The 7 actions can be taken only by the learner, not NS, because it is only the learner who can judge whether the input is comprehensible and whether actions need to be taken for comprehensible input. An action of this sort is exemplified by the learner’s signal of failing to understand NS’s previous speech (i.e., input). The signal leads NS to recognize that his or her earlier speech was not understood and to modify the earlier speech in an attempt to make it comprehensible to the learner. This modification is different from foreigner talk in that the modification results from the learner’s initiation instead of NS’s. This process of modification is widely called ‘negotiation of meaning.’ That is, the learner initiates negotiation of meaning of what was not understood with the speaker in order to make it comprehensible. So negotiation may be viewed as an active action to turn input into comprehensible input on the part of the learner, although NS can also initiate negotiation when the learner’s speech is not understandable. Several studies have been conducted to empirically show the value of modifications for comprehension and further to determine which of the two kinds of modifications is more 8 effective for comprehension. The results of the studies showed the superiority of negotiated modifications to premodifications. Long (1980) called modified speech through negotiation ‘modified interaction’ and distinguished it from what he termed 9 ‘modified input, the input which is already modified by the speaker by the time it is uttered. He compared four learning situations according to which of the two kinds of input is involved and showed that SLA is possible and facilitated only in the situations with modified interaction regardless of the existence of modified input, evidencing that negotiation is critical to SLA. Pica, Young and Doughty (1987) made similar distinctions as Long did - ‘premodified input’ and ‘interactionally modified input’ - and empirically showed that negotiated input (i.e., interactionally modified input) is of more help to comprehension than premodified input. Using directions to a task as material for the study, they compared NNSs’ comprehension of the directions when the NNSs were given in the two different types of input. NNSs better comprehended the directions when they were given unmodified directions but were allowed to interact and negotiate the meaning with NS presenter of the directions than when they were given directions 9 premodified by the researcher to the learners’ proficiency level and not allowed interaction with NS presenter. Whether premodifications or modifications through negotiation, both are efforts to make adjustments to NS speech for enhancing the comprehension of what is being said. As claimed above in the previous section, comprehension is necessary for acquisition. Thus, the relationship of adjustments, comprehension and acquisition can be simply summarized as Long (1983) argued in three steps: Step 1 : adjustments promote comprehension Step 2 : comprehension promotes acquisition Step 3 : adjustments promote acquisition The linear relationship of adjustments-to-acquisition stresses the importance of adjustments (i.e., modifications) for acquisition. Given that negotiation is an effective means for adjustment, Long’s argument may imply that negotiation is of considerable benefit to L2 acquisition. But this simple-looking argument needs empirical testing for its true value. In fact, Loschky (1994) tested the validity of this relation with Japanese as TL by measuring NNSs’ retention of vocabulary items and two locative structures. Some of the findings in this work went against Long’s linear relation hypothesis. While negotiated 10 input led to more comprehension, premodified input did not facilitate comprehension, invalidating step 1 and implying that not all adjustments help comprehension. And it was also found that more comprehension did not necessarily lead to more acquisition. The difference shown in comprehension was reduced to insignificant levels on acquisition measures. However, what is to be noted is that the results do not reject the importance of adjustments and comprehension but only indicate the inadequacy of their sufficiency to SLA. So far I have pointed out that input which learners have access to for comprehensible input has been categorized into three types: (1) input not directed to but surrounding the learner, (2) input directed to the learner and premodified by NS, and (3) input directed to the learner and modified through negotiation. Modified input - (2) and (3) - is more desirable as the source for comprehensible input. In other words, ‘Interaction’ plays an important role in comprehension as interaction is involved in (2) and (3). With the support of theoretical and empirical studies on modifications (Long, 1980; Pica et al., 1987; Loschky, 1994), it was verified that, of modified input, negotiated input can help more with learner’s comprehension than premodified input. ‘Negotiation of ll meaning’ leads to more comprehension in the way that it allows learners’ initiatives to make input way beyond their current state comprehensible by signaling failure to understand to NS. Due to its powerful effects on comprehension, negotiation of meaning has been investigated a great deal across various interactional conditions especially in terms of effects of learning variables on it. However, there is much to be explored for the full view of its function and value in SLA. The rest of this thesis is committed to giving a clearer picture of the role of negotiation of meaning in SLA by touching on some unexplored areas. Main issues, such as how to define negotiation and what learning variables are affecting it, and studies relating to those issues are first reviewed in detail to locate the issues in need of further investigation. Hypotheses and predictions are presented in an attempt to address some of the newly identified issues. An experiment is conducted to test the hypotheses. Results and discussion of the tested hypotheses follow along with suggestions for future research and implications for classroom teaching. Review of Studies on Negotiation of Meaning DefinitionJand Categorizjtion of Negotiation of Meaning It seems that there has been some confusion in the distinction between ‘negotiation of meaning’ and ‘interaction.’ In order to isolate the role of negotiation on comprehension - or more broadly acquisition - in the course of interaction, the two notions must be clearly distinguished even though they are closely related. Interaction refers to a situation in which speakers direct their speech to each other. And, as Pica (1992b) well defines, negotiation of meaning is a type of interaction with a particular purpose: “negotiation of meaning is interactions in which learners and their interlocutors adjust their speech phonologically, lexically, and morphosyntactically to resolve difficulties in mutual understanding that impede the course of their communication.” (p. 200) In order for negotiation to occur, (1) non-understanding of the whole or part of the utterance should take place and (2) an action should be taken to signal non-understanding. Varonis and Gass (1985b) pointed out that when communication breaks down, the interlocutors deviate from the main flow of conversation and engage in negotiation to get over 12 13 non-understanding. If understanding is achieved through the negotiation, then they go back to the main flow of the conversation. They called the deviation from the flow ‘push down’ and returning to the flow ‘pop up.’ Thus, using their terms, negotiation can be viewed as an effort to pop up from the push-down. The moves for signaling, or for initiation of negotiation, were first listed by Long (1980): clarification requests, confirmation checks, comprehension checks, conversational frames, self-repetitions, other-repetitions and expansions. Since his categorization, no well-defined and conclusive set of negotiation moves has yet been proposed to apply across all studies on negotiation. Each research has created and/or employed different definitions and kinds of moves according to the purpose of the research at hand. However, the first three moves in Long’s list are relatively well defined and widely adopted across studies in the field as essential moves for the initiation of negotiation. Pica, Young and Doughty gave good definitions for the three moves and used them in their study (1987) as measures for negotiation. Below are the definitions made by them and examples presented in their other study (Doughty and Pica, 1986): l4 Confirmation checks : moves by which one speaker seeks confirmation of the other’s preceding utterance through repetition, with rising intonation of what was perceived to be all or part of the preceding utterance. A: Mexican food have a lot of ulcers. B: Mexican food have a lot of ulcers? Clarification requests : moves by which one speaker seeks assistance in understanding the other speaker’s preceding utterance through questions (including wh-, disjunctive, ex.or, uninverted with rising intonation, or tag), statements such as ‘I don’t understand’ or imperatives such as ‘please repeat? A: She is on welfare. B: What do you mean by welfare? Comprehension checks : moves by which one speaker attempts to determine whether the other speaker has understood a preceding message. A: Do you know what I mean? (Pica et al., 1987, p. 740; Doughty and Young, 1986, p.313) As mentioned above, Varonis and Gass (1985b) saw negotiation as a sequence of utterances both apart from the main 15 flow of conversation and attempted to return to it. They called this sequence ‘non-understanding routine’ and proposed it as a model for negotiation of meaning. The routine consists of the following four parts: Trigger (T) : speech which causes non-understanding Indicator (1) : response to trigger (T), indicating non- understanding. This is a point where conversation deviates from a main flow. Response (R) : what is spoken in response to indicator (1) Reaction to response (RR) : what is spoken, responding to response (R) Those elements occur T,I,R, and RR, in order, within a routine as in an example below: Nobue: uh...uh, there’s uh two people. (T) Hiro: Two people? (I) Nobue: umm (R) Hiro: Uh-hmm (RR) (Gass and Varonis, 1986, p. 329) According to this model, negotiation of meaning is initiated by the utterance of I. Since the moves listed above are all for initiation of negotiation, they belong to I within this model. R is the move for providing modified input in response to the 16 signal of non-understanding (i.e., 1). RR is the move to show whether the non-understanding is resolved by R. The resolution of non-understanding through negotiation can be identified or achieved by the occurrence of R and RR. One thing to note, however, is that T and I are obligatory elements for forming a negotiation routine while the occurrence of R and RR is not guaranteed during the routine. Thus, what this model shows is that the occurrence of one indicator does notalways guarantee the completion of the whole negotiation routine or resolve the non-understanding. For this reason, in studies on negotiation, occurrences of negotiation should ideally be counted only when R and RR also take place in the non-understanding routine in order to measure its contribution to comprehension with accuracy . However, determining whether R and RR occur is, in some cases, troublesome because they do not always take verbal forms. For example, physical signals can also serve as R and RR. This may be the reason many studies relying on audio- taped conversations counted only indicators into data base on the assumption that they represent the whole process of the non- understanding routine. Indicators, however, can also take physical forms. Thus for a fuller investigation into the role of negotiation, other ways of data-gathering(e.g., videotaping) are needed which can capture more than verbal forms. NegotiatioLof Meanimcross Variables Given that negotiation of meaning was believed to be a powerful predictor of comprehensible input, many studies on interaction have been conducted to see which learning variables are more conducive over others to the occurring of negotiation. Long (1980) compared tasks requiring two-way versus one-way information exchange. He found that both modified interaction and modified input occur more frequently in two-way tasks over one-way. Demands of exchanging information in two-way tasks proved to be a driving force for eliciting more modified interaction (i.e., negotiation) along with modified input. Crookes and Rulon (1985) examined free conversation and two- way communication tasks. What was found is that more interaction in the form of feedback by NSs was spotted in communication task than in free conversation task. In that study, feedback referred to correct usage by NS immediately following NNS’s incorrect utterance. Pica (1987) examined relative effects of decision-making discussion and information- exchange task. Both are considered to be two-way tasks but different in that the latter requires exchanging information whereas the former does not. Again, the requirement of information-exchange makes the latter more profitable than the 17 18 former in terms of elicitation of negotiation. She also differentiated the information gap task from the information exchange task. They are different in that one participant holds the whole information to complete the task in the former while the information is equally held by each participant in the latter. Thus, the participant is required to both seek and provide information in the information-exchange task while the other task does not necessarily require the bi-directional interaction. She argued that the information-exchange task is ideal for negotiation elicitation in the classroom setting for its collaborative and equalizing nature. The participation pattern was investigated by Doughty and Pica (1986). They found that more modified interaction occurred in group or dyad interaction patterns than in the teacher-fronted situation. These two variables (i.e., task type and participation pattern) were compared by Pica and Doughty (1985,1986) for their relative effects on negotiation. In their studies, task type turned out to be more powerful indicator of negotiation over participation pattern. In the decision-making task, participants seldom engaged in negotiation across participation patterns of teacher-fronted and group work activities. This is because the decision-making task did not 19 fully require all participants to be actively involved in the interaction in spite of its participatory nature. For the task requiring information-exchange, the negotiation considerably increased in both participation patterns. As for variables relating to individual characteristics of the participants, the variable of age was studied by Scarcella and Higa (1981). They found that NSs provided more simplified input to younger learners than to older learners. However, on the part of learners, older learners used more interactional modifications than younger learners. Thus, younger learners received more of the input which is simplified by NS whereas older learners received more of the input modified through negotiation than younger learners. They also agreed that the simplified input more likely received by younger learners is not as optimal as the kind of input the older learners are more likely to receive: Native speakers do not always guess accurately: they are not always able to correctly ascertain the nonnative speaker’s English proficiency level and adjust their speech accordingly. Thus, nonnative speakers who do not negotiate may receive input which is too simple or too difficult. Through the work of negotiation, the nonnative 20 speakers can receive challenging input, in advance of their linguistic competence, and make this input comprehensible. (p. 430) A recent study looking at gender difference in negotiational interaction was conducted by Pica et al. (1991). One important finding of the study was that, in NS/NNS interaction, the role of gender is shown more clearly in NS gender rather than in NNS gender. When NNS was in interaction with NS female, there was a comparable amount of negotiation opportunities across NNS genders. However, during interaction with NS male, the opportunities were smaller in number for female NNS than for male NNS. And it was also found that NS females were more supportive and consistent than NS males with both NNS males and females in promoting negotiation. Gass and Varonis (1985) looked at the variable of learners’ proficiency level. NSs initiated more negotiation with the lower-level NNSs than with higher-level NNSs. This is, they argued, the evidence that the amount of NS-initiated negotiation reflects comprehensibility of NNS interlocutor. However, NNS- initiated negotiations rarely occurred across their proficiency levels. Thus, NNSs’ negotiation did not reflect comprehensibility of their NS interlocutor to them. 21 Rost and Ross (1991) studied the relationship between proficiency and a particular negotiation move - clarification requests. They classified this move into three different levels (global, local and inferential) according to the degree of the learner’s understanding of the NS previous speech. An example for each level of the clarification requests is presented below: Global level : NNS does not understand the whole message of NS. NS : He wrote a letter to Diane, who was in Europe, and told her that he was planning to take a trip, with Carol, the woman he had just met. NNS : I don’t understand. Local level : NNS partially understands the message of NS and indicates the part that was not understood. NS : So he went on a trip to some tropical countries with Carol. NNS : Tropical countries? Inferential level : NNS understands the whole message of NS and use a question to supplement the message. NS : So he went on a holiday trip to some tropical countries with Carol. 22 NNS : Mm. Maybe Hawaii or a place like that? (Rost and Ross, 1991, pp. 241 - 242) They found that the use of a certain level of clarification requests is correlated with the learner’s proficiency level. The more proficient the learner is, the more the learner tends to use local or inferential levels of clarification requests. It should be noted, however, that the local level of clarification is, in fact, widely considered as confirmation checks in other studies. And inferential levels of clarification requests are not actually moves for recovering from communication breakdowns because NNS fully understands what NS said in previous speech in using them. So this level of clarification requests cannot be taken as a negotiation move since negotiation presupposes communication breakdown. Therefore, I think what was really found in this study is not the relation of subcategories of clarification requests but that of main negotiation moves (i.e., clarification requests and confirmation checks) with proficiency. In other words, the less proficient learners are, the more clarification requests they tend to rely on over confirmation checks. This may be attributable to the reasoning that less proficient learners are more likely to not understand the whole message of NS. 23 The effects of personality on interaction was explored by Seliger (1977) where learners were categorized into high input generators (HIG) and low input generators (LIG). Learners classified into HIG were ones who are active in practicing and using L2 and try to direct input to them. LIG are the kind of learners who do not usually use or practice L2 unless asked to do so (e.g., asked by the teacher in the class). The results said that HIG engaged more actively in interaction and thus got more benefit from it for their development of L2. What has been discussed in this section is the effects of several variables on negotiated interaction. As the studies reviewed above have shown, such variables as task type, participation pattern, age, gender, proficiency, and personality, turned out to have effects on the learners’ interactive behaviors. However, all such variables do not seem to contribute to the occurrence of negotiation of meaning to the same extent. Task type is certainly a very powerful predictor of negotiation, depending heavily on whether it requires learners to exchange information for the completion of the task. But this is only an inference from the study comparing task type with participation pattern. Interaction occurs in the way that all effective variables relate to each other. However, most studies on 24 interaction so far have considered only one particular variable and investigated its effects on negotiation. In order to obtain a more complete picture for the variables’ relationships with negotiation, further studies are needed, studies which consider more than one variable and identify relative weights of each of the variables for elicitation of negotiation. Two Issues of Concern In the earlier section, such issues as how negotiation of meaning is defined and contributes to SLA, and what variables are affecting initiation of negotiation, have been addressed. There are at least two more issues yet to be more fully addressed concerning negotiation. First, all studies discussed above dealt with issues of negotiation within NS/NNS interactional settings. In this setting, NS assumes the role of input provider and presents comprehensible input through premodification on his or her own, or negotiation with NNS learner. However, as Pica et al. (1996) point out, the reality is that most learners, if not all, have very restrictive opportunities to get comprehensible input from interaction with NS: For many L2 learners, however, opportunities for either extensive or wide-ranging interaction with NSs is all too infrequent and often simply impossible. This is 25 especially so for learners in foreign language contexts, where classrooms of nonnative-speaking teachers and other L2 learners are the basis for most of their interaction. ...... across a wide range of settings, ...... , language learners are frequently and increasingly each other’s resource for language learning. (p. 60) For the most part, NNS learner is engaged more frequently in interaction with other NNSs than with NSs. Given this fact, it has become worthwhile to investigate the amount and role of negotiation of meaning occurring in NNS/NNS conversation in comparison with those in NS/NNS interaction. Second, it was mentioned earlier that Krashen claimed both necessity and sufficiency of comprehensible input to acquiring L2 forms, and that its sufficiency has been challenged by several studies while its necessity to SLA remains unquestionable. In an attempt to discredit its sufficiency, some other necessary conditions have been proposed which could account for internalization of L2 forms. One striking proposal was made by Swain (1985). She used students in a French immersion program as evidence against the sufficiency of comprehensible input. In that program, students were exposed to enough comprehensible input for many years, yet they did not produce 26 native-like competence. She asserted that this phenomenon is attributed to the fact that they did not have opportunities to use L2 in a productive way. Learners need to be in the situation where they are “pushed toward the delivery of a message that is not only conveyed, but that is conveyed precisely, coherently, and appropriately” (p. 249). In other words, learners need to have opportunities to produce ‘comprehensible output.’ She argued that comprehensible output is a necessary condition along with comprehensible input and presented main roles that the kind of output has in language learning as follows: (1) output provides learners with opportunities for ‘contextualized’ and ‘pushed’ language use; (2) output allows learners to test hypotheses about their IL; and (3) output forces learners to move their attention from semantic processing to syntactic processing. White (1987) put forth another necessary condition - what she named ‘incomprehensible input.’ This kind of input is needed especially for the acquisition of grammar because, as she put it, “the driving factor for grammar change is that the input is incomprehensible” (p. 98). She argued that input is given not in the form of structure but in the form of data which asks for a structural reanalysis. To change the current grammar of 27 IL, the learner has to detect the gap between his or her current knowledge and that represented by new input data. Incomprehensible input makes the gap stand out and promotes a reanalysis of the grammar. However, she stressed that not all incomprehensible input highlights the gap and results in reanalysis. In some instances, incomprehensible input is ignored, instead of being marked, by being mistakenly recognized and interpreted within the learner’s current grammar. She took an example of a learner whose grammar contained no rule for the passive. The learner may not have any problem in interpreting input like ‘John was hit by Fred’ because that could be interpreted as ‘John hit Fred’ in the learner’s current grammar, thus the passive structure in the input would not be recognized and no reanalysis of the learner’s current grammar would result. Because of this, she pointed out the importance of negative input or specific structural teaching. In her view, such ignored structure can effectively be brought back to light by direct supply of negative evidence or instruction. This action also helps eliminate non-target intermediate form residing in IL in addition to learning new but ignored structures. Gass (1988) distinguished ‘comprehended input’ from Krashen’s comprehensible input and considered the former as a 28 better necessary condition for acquisition. According to her, there are two differences between the two notions of input. First, comprehensible input is controlled by input provider, whereas comprehended input is controlled by the learner. Krashen assumed that all comprehensible input is comprehended by the learner. Here, Gass implied, from the distinction and against Krashen’s assumption, that all comprehensible input can be but not necessarily is comprehended. Considering that both the process of acquiring L2 forms and comprehended input is ultimately controlled by the learner, comprehended input is the better predictor of success of acquisition than comprehensible input which is heavily controlled by the input provider. Second, comprehensible input is dichotomous, that is, input is either comprehensible or not. But comprehended input represents a continuum of multi-staged comprehension. In other words, there are different levels in the learner’s comprehension. For example, when input is considered to be comprehended, that comprehension may be at the semantic level or syntactic level or both. This implies that, in investigating the relation of comprehension to acquisition, we need to look at not only whether comprehension takes place but also how 29 closely or differently a certain level of comprehension is related to acquisition. To emphasize the importance of consciously focusing on linguistic forms as necessary behavior for acquisition of the very forms, such concepts as consciousness, awareness, noticing (Schmidt, 1990), attention (Tomlin and Villa, 1994) and apperception (Gass, 1988) have been treated in studies on the role of that behavior. The concepts are not identical in meaning and there were some studies (Schmidt, 1990; Tomlin and Villa, 1994) trying to differentiate the concepts from each other and further elaborate each concept into subconcepts to investigate the role of each on acquisition. In spite of the use of different concepts, what those studies all agree on is that it is important to draw one’s cognitive sense to conscious level and focus it on the TL forms to acquire them. Given these conditions along with comprehensible input as necessary for acquiring L2 forms, it would be interesting to see whether negotiation of meaning contributes to these conditions as well and, if it does, how. In fact, the two issues that Ijust put forth - negotiation of meaning in NNS/NNS interaction and its relation to necessary conditions for SLA other than comprehensible input - have 30 already been touched on in some studies. However, compared with research dealing with negotiation in relation to NS/NNS conversation and to comprehensible input, the findings about the two issues are relatively restrictive in scope and depth, leaving much room for further research. I discuss below some achievements of those few studies on the issues, and based upon them, I present research questions for the current study which attempts to enlarge them. liagotiation of Meaning within NS/NNS vs NNS/NNS Qwersatiml Setting_s_ Varonis and Gass (1985b) is one of the main studies exploring NNS/NNS conversational settings with regard to negotiation of meaning. They compared NS/NS, NS/NNS and NNS/NNS conversations, and found that negotiation of meaning is' more prevalent among NNS/NNS pairs than NS/NNS and NS/NS pairs, thus implying that NNS/NNS conversation is no less valuable setting for SLA than NS/NNS settings in terms of bring about negotiation of meaning. They attributed this result to a non-threatening atmosphere that NNS/NNS interaction offers. In this interaction, neither of the interlocutors is competent in TL, so neither of them feels blame when communication breaks down, which may allow them to practice 31 developing their IL more freely, thus resulting in more interaction and negotiation. Doughty and Pica (1984) looked at classroom NNS/NNS activity in the form of group work. They found that group work elicits more negotiation than teacher- fronted activity which represents NS/NNS interaction in the study. The value of group work as a NNS/NNS interaction type, was detailed in Long and Porter (1985). They enumerated the merits for group work compared with teacher-fronted in SLA: increasing language practice opportunities, individualizing instruction through face-to-face interaction, promoting positive affective climate, and motivating learners. They concluded that group work is a good context for stimulating interaction among learners within classroom setting. In spite of the value of NNS/NNS interaction as context for negotiation of meaning, what could be a problem is the quality of negotiation of meaning produced in it. That is, since unlike NS/NNS settings, interlocutors are incompetent in TL, the changes made as the result of negotiation in NNS/NNS conversation may not be in the direction of target language. Pointing this out, Porter (1983) argued that learners cannot provide each other with the accurate grammatical and sociolinguistic input that NSs can. Pica et al. (1996) also 32 found that learners in NNS/NNS interaction do not provide as much modified input as NSs in NS/NNS interaction when they respond to their interlocutor’s signal of non-understanding. However, some other studies offered strong evidence that even NNS/NNS conversation serves as a forum for correct L2 forms. Gass and Varonis (1989) showed that the changes made even as a result of the interaction in NNS/NNS discourse are in the direction of the target language. The following is part of the script presented as an example in their study: Excerpt (1) a. Ana : Can you tell me where is the train station? b. Keiko : Can you tell me where the train station is? 0. Ana : Can you tell me where is the train station? (1. Keiko : Can you tell me where the train station is? e. Ana : Can you tell me where the train station is? f. Keiko : Can you tell me where the train station is? The interlocutors in the script were both NNSs and told to ask for directions. As can be seen, Ana made a change in the form of her utterance from incorrect to correct while Keiko made no change. Gass and Varonis noticed even more surprising phenomena in their data. Even when NNS received corrective input as a result of interaction, if the form was not correct, the 33 NNS did not accept the repair and maintained the original form. The excerpt for this case is below: Excerpt (2) a? Hiroko : A man is uh drinking c-coffee or ten uh with uh the saucer of the uh uh coffee set is uh in his uh knee b. Izumi : in him knee 0. Hiroko : uh on his knee (1. Izumi : Yeah e. Hiroko : on his knee f. Izumi : so sorry. On his knee Hiroko maintained her form ‘his,’ ignoring Izumi’s corrective response of ‘him.’ Rather, Izumi changed her form from ‘him’ to ‘his.’ They explained these phenomena by introducing the notion of ‘strength of knowledge.’ Theyargued that learners have different degrees of knowledge strength over different parts of their L2 grammar forms. If the grammar forms in their IL knowledge are correct, the strength of knowledge over those forms tends to be stronger than that over incorrect forms. Due to this strength of knowledge, the correct forms are less susceptible to external influence than incorrect forms in their IL grammar. In the above excerpt (1), Keiko’s knowledge strength 34 over indirect question form is stronger than Ana’s because Keiko’s knowledge is correct. Thus, change was made in Ana’s knowledge which is weaker and more susceptive to external input due to its incorrectness. Likewise, in the excerpt (2), Hiroko’s knowledge of the pronominal is stronger than Izumi’s in its strength, so Hiroko’s form ‘his’ was not sensitive to incorrect input ‘him’. Gass and Varonis stated in a later article (1994) that 89% of all modifications as a consequence of NNS/NNS interaction in this study were in the direction of target language, evidence that errors of NNS speaker are not generally incorporated into the learner’s grammar. Similar evidence was shown in Bruton and Samuda (1980). They encountered only one instance of a change from correct to incorrect in 10 hours of taped conversations between nonnatives. There is another issue with regard to aspects of negotiation of meaning in NNS/NNS interaction in comparison with NS/NNS interaction. In the same study mentioned above, Gass and Varonis (1989) encountered another interesting finding: the repairs made within NNS/NNS interaction were generally in the correct form but all at the lexical level. There was no incidence of either discourse or pragmatic repairs. Porter 35 (1986) stated, indicating the learner’s lack of appropriate language use, “only native speakers can provide truly appropriate input that will build sociolinguistic competence” (p. 218). Lack of pragmatically appropriate use of language by the learner in their response to non-understanding signals implies that, in qualitative perspective, NNS/NNS negotiation can contribute to development of one aspect of communicative competence (i.e., linguistic competence) but has limits on its contribution to development of another (i.e., sociolinguistic competence). Despite this qualitative limit, the recognition of NNS/NNS interaction as a forum for negotiation, which is comparable to NS/NNS interaction in the amount of negotiation initiated and in grammaticality of resulting repairs, motivated several studies to see whether variables having effects in NS/NNS settings would equally have effects on negotiation in NNS/NNS settings. Gass and Varonis (1985) investigated the effects of task type on negotiation within NNS/NNS interaction. They classified tasks into one-way and two-way according to the distribution of information that is necessary for the completion of the task: an one-way task is one where one interlocutor holds all information whereas the information is equally distributed 36 among interlocutors in a two-way task. Based upon the model for negotiation of meaning proposed in their other article (1985), occurrences of indicators were compared across task types. No significant difference was found between them. As they pointed out, this result did not agree with Long’s study using similar task types (one-way versus two-way) for NS/NNS settings where two-way tasks elicited significantly more negotiation. They attributed this different result to the effect of shared background information in the two-way task. In the task of their study, the topic was robbery. Even though interlocutors were given different portions of information, they shared general background knowledge about crime to some extent, which would reduce the need for negotiation, compared to the one-way task of replicating the other’s picture where shared background knowledge is not feasible. This implies that shared background knowledge would be a major variable affecting negotiation negatively and that the difference in Long’s study was reduced by a factor of background knowledge in the study of Gass and Varonis. Doughty and Pica (1984) sought differential effects of tasks according to whether information-exchange was required. A significant difference was noted, proving that the requirement of information exchange is an affecting variable across NS/NNS and NNS/NNS settings, 37 given that this variable turned out to have effects in NS/NNS setting in another study (Pica, 1987). Gass and Varonis (1986) looked at gender difference. In the study, learners were assigned to one of three types of group formation according to their gender: male/male, female/female and mixed-gender pairs. Men used more indicators than women when they were in the same-sex pairs. But these effects were reversed when learners of different gender interacted within the same pairs: women initiated more negotiation than men in the mixed-gender dyads. It was also found that more negotiation was initiated in mixed-gender pairs than in same-gender pairs. Porter (1986) compared students of advanced levels and those ‘of intermediate levels across their interactions with interlocutors of same and different proficiency. For comparison, six subcategories of repairs were built, which are usually considered negotiation moves in other studies: confirmation checks, clarification requests, comprehension checks, verifications of meaning, definition requests and lexical uncertainties. The results showed no significant difference between two learner proficiency levels for the initiation of negotiation regardless of interlocutor’s proficiency. What was found to be different between the proficiency levels was in the 38 quantity of input each level provided. That is, advanced learners generally provided more input than intermediates and as much input as NS did when they interacted with intermediates. She also found that there is less communication breakdown in interaction of learners from the same native language background than that of learners from different language backgrounds. This is because if learners were from the same language backgrounds, pronunciation errors would not make language incomprehensible to each other. No communication breakdowns owing to phonological problems were detected in interactions of learners from the same L1. The other side of this indication is that there would be less negotiation in interaction of learners of the same L1 backgrounds than those of the different Ll backgrounds. This, along with effects of proficiency, was empirically tested by Varonis and Gass (1985b). They compared 14 NNS dyads grouped by whether they shared L1 and proficiency: group 1 was composed of learners sharing same L1 and proficiency, group 2 of learners sharing one of them, and group 3 of learners sharing neither. Non-understanding routines were found most in the dyads of group 3 and the next highest incidences of routine were in group 2. Group 1 evidenced the lowest incidences. They argued on 39 the basis of this result that the less learners share in their backgrounds, the more chances they bring about negotiation in their interactions. But this conclusion seems a little too general in that only two aspects of learners’ backgrounds (i.e., L1 and proficiency) were considered. More variables such as knowledge of t0pic should be incorporated and tested to validate the relationship of negotiation with degree of difference between interlocutors’ backgrounds. Negotiation of Meaning as Facilitative to Other Necessary Conditions for SLA A possibility that the value for negotiation of meaning goes beyond its role in comprehensible input was suggested by Pica (1994): The perspective that negotiation modifies the L2 in ways that help learners comprehend its meaning is highly restricted, and places negotiation in a secondary role in L2 learning, because it sees comprehension of meaning as the principal way to access and internalize L2 form, and negotiation as simply a way, albeit it a very good way, to bring comprehension about. (p. 506) 40 She argued that negotiation can also be the way to direct access to L2 forms because modifications made during negotiation can be structural as well as lexical ones which, according to her, are mostly used to make input comprehensible. Structural ‘6 modifications take place during negotiation when utterances are segmented and relocated to another constituent position - for example, from object in an initial utterance to subject in a follow-up utterance.”(l994;510). Considering that conditions emerging after comprehensible input are proposed to more directly account for the process of acquiring L2 forms, seeing how negotiation is related to those conditions may help make it clear how negotiation contributes to acquisition of L2 forms. To do so, it is important to note that both NS and NNS can initiate negotiation when either of them does not understand the interlocutor in NS/NNS interaction, and both sides can also modify their speech to the interlocutor’s signal of non-understanding. Within the view that considers negotiation only as facilitator to comprehensible input, the underlying assumption was that negotiation is initiated by NNS and modifications are made by NS. But that in real interaction neither of them is limited to either NS or NNS, makes it possible to think that negotiation can be a source for feedback 41 on IL when NS gives non-understanding signal to NNS. Receiving the signal from NS, NNS gets feedback on his or her previous utterances which leads to recognition that something was wrong in them. In this sense, feedback offered by NS corresponds to negative evidence which White suggested as one condition for SLA. When NNS responds to the NS signal, negotiation becomes a source for production opportunity. This is where negotiation serves to meet another condition of ‘comprehensible output’ proposed by Swain (1985). For, when responding to NS’s non-understanding signal, NNS gets pushed toward using language precisely, coherently or appropriately to get the message across. As Pica (1992b) puts it, “NS utterances of feedback during negotiation provided NNSs with opportunities to produce lexical and structural adjustments in their interlanguage” (p. 220). Thus, what is worth noting is that negotiation gives learners opportunity for comprehensible input if it is initiated by the learners on one hand, but that it provides negative evidence and context for comprehensible output when initiated by NS on the other hand. In the follow-up study, Pica et al. (1996) investigated the expanded value of negotiation in NNS/NNS settings in comparison with NS/NNS settings. Findings included: (1) NNS 42 in NNS/NNS interaction made fewer lexically and structurally modified utterances than NS in NS/NNS interaction, and thus the learner would get less modified input or comprehensible input in NNS/NNS dyads; (2) both NNS and NS made comparable input in structural complexity; (3) the learner in NNS/NNS dyad gets feedback which is comparable in amount but less complex in form (mostly segmentation) than one in NS/NNS dyads; and (4) NNS does not modify his or her utterance in negotiation with - other NNS to a greater extent than in negotiation with NS. In short, the learner gets comparable feedback in NNS/NNS interaction but does not benefit from negotiation as much in NNS/NNS interaction as in NS/NNS setting in terms of obtaining comprehensible input and opportunities for production. Stevick (l976,1980,1981) claimed that active involvement in the discourse is a facilitator of acquisition in that it ‘charges’ the input and allows it to ‘penetrate’ deeply. Given that comprehension is essential to SLA, the learner should be involved in the discourse in order to comprehend his or her interlocutor. One may also think the other way: without understanding the interlocutor’s speech, one cannot keep one’s involvement in the discourse with the interlocutor. So involvement and comprehension are in cyclical, not linear, 43 relation to each other. Negotiation may be viewed as an attempt to keep involved in the discourse in that it is an act to return to the main flow of conversation by signaling non- understanding which could depart the learner from constant involvement. Attention or related concepts were introduced and studied as necessary for acquisition of L2 forms. When one gives a signal to the interlocutor during negotiation, it may be seen, in a sense, as an act to focus the interlocutor on a part of the speech that was not understood. In response to the signal, the responder naturally attends more intensively to the part of his or her previous utterances targeted for negotiation. Thus it seems that negotiation of meaning may serve to meet all main necessary conditions proposed above in one way or another. The Purposes of Research In order to get a better picture of the role of negotiation in NNS/NNS interaction to language learning and to be able to better help learners with their learning within the context of interaction with other learners, the following issues should be addressed, along with and based on the achievements of the studies mentioned above. 44 First, although various variables have been tested for their own effects on negotiation, there has seldom been comparison of one variable with another for relative weight in effects of each whether in NNS/NNS settings or NS/NNS settings. The efforts should be pursued to determine how much of the variation in negotiation each of the variables accounts for. This is as important as the work of determining whether the variables have effects on negotiation because it would help teachers and learners create more beneficial learning environments for negotiation by informing which variables should be considered more over others. Second, in NS/NNS conversations, the interlocutor of a particular NNS is more likely to be homogeneous in terms of cultural background than in NNS/NNs/ conversations because all NSs share the same culture. There is less variation in cultural background of NS. Therefore, cultural distance is more likely intrinsically controlled and little affects NNS’s initiation of negotiation in NS/NNS settings. But in NNS/NNS conversations where both interlocutors are NNS, it is not controlled in itself. There may exist great variation in the cultural distance between interlocutors. For example, when a Korean learner of English talks with a native speaker of 45 English, the NS interlocutor has the same cultural background (i.e., American culture) whoever he or she is.(As American culture is composed of various distinctive subcultures, it may sound too simplistic to consider it as one homogeneous culture. I will discuss this problem later.) However, when talking with a non-native, the interlocutor may be another Korean or a Japanese or Arabic speaker. They are all different in cultural backgrounds thus causing variation in cultural distance between the Korean learner and the interlocutor. This unique factor may surface in NNS/NNS conversation and have impact on negotiation of meaning. Thus the cultural distance is worth testing for its effects in NNS/NNS settings. Third, although it is for sure that negotiation of meaning contributes to the fulfillment of various necessary conditions for SLA, it has not been investigated whether all negotiations have equal potential for its contribution across all conditions discussed. We already saw that there is difference in the effects of negotiation across NS/NNS and NNS/NNS settings. Likewise, there may be some different types of negotiation and negotiation may contribute to the conditions differently according to its type. 46 In the current study, an experiment was designed to touch on those issues, though to a limited extent. The study first focuses on the variables of task type and L1 background and investigates whether they still have effects on negotiation within NNS/NNS interactional context as in NS/NNS. Second, if both are proven to be effective, they are compared for their relative weight in effects on negotiation. Third, the variable of cultural distance is tested for its effects on negotiation with Korean learners being at the center. Lastly, the contribution of different types of negotiations to each of necessary conditions for SLA which were introduced and discussed earlier (i.e., comprehensible input, comprehensible output, negative evidence and so on), is investigated. To do so, negotiation is classified into two types. Research Design Subjects Twenty four nonnative speakers of English participated in the study. They were all enrolled in an intensive language program at Michigan State University. The subjects included 12 Koreans, 3 Japanese, 2 Thais, 2 Taiwanese, 2 Chinese, 1 Ecuadorian, 1 Arab Emiratian, and 1 Brazilian. Each subject was assigned to a dyad, forming 12 dyads for the 47 research. Six of the twelve dyads consisted of subjects who were of different Ll backgrounds and the rest of them were of the same L1 background. The composition of the dyads by L1 is displayed below: Dyads of same Ll : K/K(3), J/J(1), Th/Th(1), Tw/Tw(l) Dyads of different Ll : K/J(1), K/Ch(2), K/A(l), K/B(l), K/E(1) (K-Korean, J-Japanese, Th-Thai, Tw-Taiwanese, Ch-Chinese, A-Arab Emiratian, B-Brazilian, E-Ecuadorian) Subjects were controlled for sex, proficiency, age and familiarity with interlocutor as follows. As for sex, subjects were not limited to one sex. Fourteen females and ten males participated. But all dyads were composed of same-sex subjects to control for effects of sex difference. This arrangement is based upon the finding by Gass and Varonis (1986) that there is no significant difference in amount of negotiation between same-sex dyads of male and female while there is significant difference between same-sex dyads and mixed-sex dyads. Therefore, the caution was used that no dyad is composed of subjects of different sex regardless of which sex the dyad represents. 48 With regard to proficiency, subject selection was limited to those in the most advanced levels. All the subjects were pulled out of English for Academic Purposes (EAP) classes, whose students are comparable to the advanced level in proficiency. I was not be able to get the exact ages of all subjects because some of them did not tell their ages. Rough figures of ages of those who refused were obtained, for example, by asking “Are you under 30?”. All subjects turned out to be aged Ir somewhere between 20 and 30. Familiarity was controlled in that both subjects in pairs belong to the same class. Since the experiments were conducted in the middle of the semester, all students in the same class had been together for more than two months. Thus, there was no chance that subjects in the same pair were strangers to each other. Procedure All dyads engaged in two rounds of tasks. One of them was a free conversation and the other was an information-exchange task. In free conversation, the dyads chose whatever topic they wanted and talked about it. They were allowed to freely change or shift to another topic when the flow of the conversation lagged. A picture difference task was employed for the 49 information-exchange task. Two pictures (see appendix) with eight slight differences were prepared. Each subject in the dyad was given one of the two. They were asked to try to find the differences between those pictures. Subjects were asked to do the task only by talking with each other and were not allowed to see their partner’s picture. Six dyads conducted free conversation first while the rest of dyads did the picture difference task before free conversation. This arrangement was made to control for possible influence on results by task ordering. The number of negotiations occurring in six free conversations that were conducted as the first task were compared with that in the other six conducted as the second task. Picture difference tasks were also compared the same way. The occurrences of negotiation turned out not to be affected by the task ordering in either of the tasks (t= 2.225 df =10 for picture difference, t=l.47 df=10 for free conversation) although the difference for picture difference almost reaches the significant level. Subjects conducted their second task one week after they did first. Due to different class schedules of participants, it was impossible to gather all participants together at the same time at one place and conduct the experiments. Experiments took place 50 right after participants’ EAP classes and in the classrooms where they took classes. All dyads from the same class participated in the tasks in the same order. This was designed to especially ensure that the effects of information-exchange task may not be contaminated. For example, if the dyads in the same class had engaged in different tasks as the first task at the same time, since they were all in the same class and there was an interval of one week between the conductings of tasks, there would have been great chances that subjects would tell each other about what to do on their first task before the second task is conducted, which would ruin particularly the validity for results of the picture difference task. Before engaging in the tasks, participants were given instructions on what to do to complete tasks. Fifteen minutes were allowed for each task. A tape recorder for each dyad was positioned in the full view of participants. All conversations were tape-recorded and the first 10 minute sections of them were transcribed for analysis. Operational Categorization of Negotiation Recall that a model for negotiation of meaning introduced by Varonis and Gass (1985) which is composed of four elements - Trigger, Indicator, Response and Reaction to Response. And also remember the three widely adopted indicators as defined by 51 Pica et al. (1987), which I presented earlier: clarification requests, confirmation checks and comprehension checks. The current study uses those indicators for indicating incidences of negotiation. Among them, however, the comprehension check is distinctive from the two others in that its trigger lies in the utterance of the speaker, not in the utterance of the previous interlocutor. That is, the comprehension check and its trigger exist in the same interlocutor’s speech. This means that unlike the clarification request and the confirmation check, the comprehension check does not depend on the other interlocutor’s speech for its trigger. Therefore, the use of this move is completely in the control of the speaker. This makes it possible for interlocutors to use the move whenever they want during their speech turn. Based upon this difference, indicators are divided into two categories called ‘self—motivated’ and ‘other-motivated’ in this study depending on where the trigger lies. Other-motivated negotiation is defined as that which is initiated by the bearer. The trigger lies in the speaker’s earlier speech. This type includes clarification requests and confirmation checks. Self- motivated negotiation is defined as that which is initiated by the 52 speaker when he or she has difficulty in or is unsure of success in getting the message across during his or her speech. The trigger lies in the speaker’s on-going or just completed speech. The study included verification of meaning as well as comprehension checks into the latter. Verification of meaning is common with comprehension checks in that it also is a move for checking if the speaker’s message is getting across. What is different is that while speakers using comprehension checks feel unsure of the comprehensibility of their completed utterances, speakers using verification of meaning are not sure of the comprehensibility of their on-going utterances. Thus they tend to raise intonation at the end of utterances to check the comprehensibility as substitute for comprehension checks. A: And he he wears tight, tights? B: Tights, tights, yes Is there any paper like uh like ball? A: Yes As seen in the example, this move is particularly used when the speaker is not confident whether the words he or she chose are appropriate in the context. This distinction of moves according to where their trigger lies is worth consideration because it may make clear that, in 53 NNS/NNS interactions, not all types of negotiation moves equally contribute to the supply of modified input, feedback and chance to produce output, which I discussed earlier as some of necessary elements for acquisition of L2 forms. Imagine two NNS learners engaging in negotiation of meaning. We can think of different effects of the negotiation according to which type of indicators used for initiating the negotiation. Specifically, if a learner initiates negotiation by using other-motivated moves (i.e., clarification requests or confirmation checks), the learner is seeking modified input. And from the learner’s initiation, his or her interlocutor is not only receiving feedback on his or her previous utterance, but getting a chance to produce comprehensible output by getting those negotiation moves. Thus, other-motivated moves give the move user a possibility of getting modified input and at the same time provide his or her partner both with feedback on his or her previous output and with a chance to produce comprehensible output. An example is below: A : End of sleeve two dot and right-hand there is three dot. B : What do you mean the dot? (Clarification Request) 54 B would probably get modified input if A responds, but it is not guaranteed since A’s response is optional. A gets feedback that the utterance was not understood (i.e., it was not comprehensible output) and also gets an opportunity to respond and to modify and produce utterances which would be comprehensible to B. If a learner used self-motivated moves (i.e., comprehension checks or verification of meaning), since those are moves seeking feedback, the learner gets him or herself probability of getting feedback from his or her interlocutor. A : Does he have shirt collar in his shirt? Do you know what I am saying? (Comprehension check) B : (No response) In the above example, A 'is using a comprehension check in an attempt to induce response from B that would be feedback on the comprehensibility on A’s message. Unlike other-motivated moves, it benefits only the move user, not his or her partner. This difference probably comes from the fact that in other- motivated negotiation, the trigger and the indicator are uttered by different interlocutors whereas in self- motivated negotiation, both uttered by one same interlocutor. From A’s comprehension check, B cannot expect modified input, 55 feedback or a chance to modify his or her own speech . Only A is a potential beneficiary, building chances to get feedback on the comprehensibility of his or her utterances. A model is displayed in Table 1, which shows the beneficiaries with regard to the relationships of other- and self- motivated moves with their contributions to the three elements. Table l. Beneficiaries from the Use of Indicators Beneficial Areas Use of Modified Feedback Production of Input Output Other-motivated User” Responder# Responder“ (CR/CF) Self-motivated User“ (CP/VM) * opportunity given for beneficial area ‘# opportunity realized for beneficial area ‘User’ indicates ‘user of indicator’ ‘Responder’ indicates ‘provider of trigger’ in that he is a potential responder to indicator CR: Clarification Requests CF: Confirmation Checks CP: Comprehension Checks VM: Verification of Meaning It is important to note that negotiation moves (i.e., indicators) generally ‘offer’ opportunities for the necessary elements but they do not guarantee ‘realization’ of those opportunities. In other words, through indicators, as seen in examples above, the move user only gets ‘opportunities to receive’ modified input 56 and feedback, and the responder gets ‘opportunities to produce’ comprehensible output. It is, however, not ensured by the occurrence of indicators that the move user ‘really receives’ modified input and feedback, and the responder ‘really produces’ comprehensible output. Only the feedback given to responder is really realized through indicators. Then, it should be noted that it is through ‘responses’ that the opportunities may be realized in the course of negotiation of meaning. Consider the following example: A: Ijust saw the from here up, there is B: What do you mean? (Clarification Request) A: I mean, his collar come like this B: Oh, yeah like that By responding to B’s clarification request, A actually modifies and produces output which could be comprehensible. B then would get modified input which would be comprehensible as is the case in the example. If B gives a reaction-to-response like above, it becomes certain that the modified input given was comprehensible to B and in turn comprehensible output was uttered by A. If B did not give a reaction-to-response, then only B would know if the modified output by A is 57 comprehensible and A would not know without any non-verbal signal. A: I think that’s the one difference. Just mine is like twisted. Do you know what I am saying? (Comprehension Check) B: Um, no. From B’s response, A finally gets feedback on whether his or her message was understood. In this case, B’s ‘no’ serves as evidence that something is wrong with A’s message (e.g., grammatically wrong or too complicated, or contextually inappropriate) and that it needs to be modified to make it understood. The point is that the response is optional but not less important than the indicator in appreciation of contribution of negotiation to SLA. Table 2 shows the model that displays the beneficiaries from getting response to each indicator. 58 Table 2. Beneficiaries from Responses to Indicators Beneficial Areas Responds to Modified Feedback Production of Input Output Other-motivated User# Responder# CR/CF Self-motivated User# CP/VM # opportunity realized for beneficial area ‘User’ indicates ‘user of indicator’ ‘Responder’ indicates ‘provider of trigger’ in that he is a potential responder to indicator CR: Clarification Requests CF: Confirmation Checks CP: Comprehension Checks VM: Verification of Meaning As shown, all the opportunities seen in Table l have been realized in Table 2 where it is assumed that the response was made in the non-understanding routine. Data Coding All incidences of each indicator were counted. A graduate student in the English Department participated in counting moves. I instructed her in the operational definitions and classifications of moves employed in the study, and then she and I counted moves in the whole transcribed data separately. The number of moves we both classified into the same category was divided by the number of all moves I counted for interrater reliability, which turned out to be 95.51%. 59 In the course of counting, some caution was used in dealing with confirmation checks and interjections as moves. Varonis and Gass(l985b) pointed out that confirmation checks may serve not only as indicators but as conversational continuants as in the following example: A: I’m from Nagoya. B: Are you from Nagoya? A: Yeah B: I’m from. eh. suburb Nagoya. The first utterance of B may not be intended to show non- understanding of A’s previous utterance but rather may be an act to encourage A to keep saying. In some cases, it is ambiguous which intention the utterance has, and in other cases it may have both intentions. In the present study, the kind of confirmation checks clearly with only the former intention were included. Clarification requests may take the form of interjection as below: A: He has triangle like this. B: Uh? A: Triangle like this. ‘Uh’ in the example is a move to request clarification of the interlocutor’s previous speech 60 and is not the same as that'which functions as a simple filler in the conversation. The interjections of this sort usually have rising intonation. The current study included this kind of interjections into the category of clarification requests. Variablesgand Predictions (1) Task type It has been pointed out earlier that if a task is to elicit many negotiations, it should require participants to exchange information. Including the requiring of information-exchange, Pica et al. (1992) proposed four features of the task that can best generate conversational modifications: (1) each interactant holds a different portion of information and supplies and requests this information; (2) requesting and supplying information is required; (3) interactants have same or convergent goals; and (4) only one acceptable outcome is possible. Applied to these criteria, the two tasks which are contrastive are used in the study in order to see the maximum effects of task type. While the picture difference task has all of the four features, the free conversation does not guarantee any of them as necessary. In the picture difference task, the participants (1) had slightly different pictures representing different portions of information; (2) had to ask the partner for 61 information and supply it by describing their picture; (3) had the same goal of finding eight differences between the two pictures; and (4) only tried to find differences because it was the only acceptable outcome that the task requires. On the other hand, when participants engaged in the free conversation, they (1) did not have set information to exchange; (2) were not required to give or request information; (3) did not have a specific goal but, if any, it could be to keep conversation going; and (4) did not produce any particular outcome because they did not have a specific goal. But these four differences do not seem to be sufficient to account for the superiority of the information- exchange task. One other feature that makes the information- exchange task a strong impetus for negotiation is that both interlocutors engaging in the task must both understand information given by the partner and get their own information across. Just requiring information to be exchanged is not enough. Without understanding and getting understood on the information exchanged on one or both parts, the completion of the task is not plausible. In fact, when non-understanding takes place, the interlocutors in this task are forced to engage in negotiation while those in other types may advance the tasks with non-understanding unsolved. This may make a big difference in initiation of negotiation in favor ofthe 62 information-exchange task. From these distinctive contrasts, it seemed logical to assume that the picture difference task is more effective than the free conversation with regard to generation of negotiation of meaning. This reasoning led to the first hypothesis of the study: Hypothesis 1 NNS dyads use more negotiation in the information-exchange task than in the free conversation. (2) L1 background It is beyond question that the native language influences second language learning at various linguistic levels (e.g., phonological, lexical or syntactic level). Lado (1957) clearly stated on NL influence as follows: Individuals tend to transfer the forms and meaning, and the distribution of forms and meanings of their native language and culture to the foreign language and culture - both productively when attempting to speak the language and to act in the culture, and receptively when attempting to grasp and understand the language and the culture as practiced by natives. (p. 2) However, how strong an influence it has still remains controversial. For adult learners, NL influence is most readily 63 detected at phonological level. They cannot help carrying their NL accents on to a second language. The learners having the same NL carry the same accents. This sharing of particular accents may partly account for the occasions where in talking in a second language, learners can more readily understand someone from the same NL than one with different NL backgrounds from theirs. There are also cases that even incorrect forms of TL are understandable among learners of the same NL. An example is ‘sharp pencil’. This phrase passes on for ‘mechanical pencil’ among Korean learners of English. Many of them even think that ‘sharp pencil’ is a correct form and use it in conversation. If they interact in English with another Korean and use it, there would not be communication break-down because it is understandable to all Korean learners. However, when a Korean uses it to a learner from a different NL, it is most likely that the interlocutor will ask for clarification of meaning of ‘sharp pencil.’ It seems that the sharing of NL linguistic features (especially at phonological and lexical level) may reduce the need for negotiation of meaning. Thus, the second hypothesis was made as follows: Hypothesis 2 NNS dyads of different Ll backgrounds use more negotiation than dyads of the same Ll 64 background. (3) Task type and L1 background There is no basis yet found on which to reason that one variable is more critical than the other in stimulating negotiation of meaning. Thus, with regard to relative weight in importance, a null hypothesis was made: Hypothesis 3 Whether interlocutors in NNS dyads share the same L1 background is as crucial a factor in eliciting negotiation as task type. (4) Cultural distance The importance of considering a learner’s cultural background with regard to negotiation lies in the fact that a culture regulates general interractional styles of the members belonging to that culture. Everyone under the same culture shares social values and those shared values restrict the members’ behavioral modes including linguistic behaviors such as interactional styles. And the very interactional styles specific to a culture may have influence on its members’ initiation of negotiation. Sato (1982) is one that shows how different cultural values of learners affect their styles of interaction with their teacher in the classroom. She divided subjects into Asian and non-Asian 65 groups, and found that Asian students were more reluctant to self-initiate their speech, and more dependent on the teacher’s ‘go ahead’ sign for opportunities to speak in the class. She argued that it is not attributed to their personality of being shy but to different notions of teacher-student relation across the two groups. That is, Asians saw the teacher as an authority figure and thought it was inappropriate to initiate speech without the teacher’s permission in the classroom. She concluded that different cultures represent different social values, which bring out different views on and acceptable attitudes towards the teacher. And they result in different styles of interaction with the teacher among learners. Similar findings were revealed in Findlay (1995) where Hmong, students who are South Asian people immigrating to U.S. 20 years ago, were investigated. This community had quite different values from those of American people, which include cooperation, modesty, collective effort and formal respect for those of higher social rank and status. Within this culture, the deference to authority should be demonstrated through quiet, reserved behavior. As the teacher was seen as a higher status figure by Hmong students, this cultural norm led American students to see Hmong students as being shy in the 66 classroom. What is important is that Hmong students’ style - Findlay termed it ‘taciturn style’- was not a result of their personality, but from the influence of their cultural norms. He well summarized the relationship between the culture of the learner and his or her interactional behavior as follows: As an individual member of a society acquires a language and a set of cultural rules for the organization of a particular world view, that individual will emphasize certain values, attitudes, and perceptions shared by most other members of that society. Hence, language reflects the general orientation of a group of people to the world in which they live. As a result of differential cultural emphases, individuals from differing cultural backgrounds may arrive at conclusions to common problems that differ significantly from those of others ...... People learn how to interact with others according to cultural rules that are specific to the society in which they are socialized. (p.28) Since cultural norms regulate the members’ interactional styles, then it seems reasonable to think that differences in cultural norms and the resulting different interactional styles between interlocutors may affect negotiation of meaning in their interaction in its amount and type, given that negotiation itself 67 is a kind of interaction occurring when communication breaks down. However, in exploring this issue, there is a problem to be settled. The boundaries of culture applicable to research should first be defined because the notion of culture may be broadly applied in scope. A learner belongs to many layers and/or kinds of communities and their culture. For example, a learner may belong to a school, a family and a private group (e.g., tennis club) at the same time. Or he or she may belong to Paris, France and Europe. All these communities have their own culture which determines the learner’s interactional styles. In other words, the learner’s interactional style is not the influence of an isolated culture but combination of all layers and kinds of culture that the learner is under. In a sense, all people are under the same culture - the culture of the earth. Thus, it is necessary to decide at what layer of culture you want to see its effects on interactional styles and negotiation. In the current study, the comparison of interactional styles is made at the national level. That is, the assumption is that learners from the same nation share certain interactional styles under the influence of that nation’s culture. This would be problematic with a nation like the U.S. , which is composed of communities with many distinctive cultures (e.g., the culture of African Americans and whites). To avoid this problem, the study is CG CI It It ‘1 68 conducted with Koreans being at the center, who are relatively culturally homogeneous at the national level. To determine and describe exactly what interactional styles that each culture of the subjects involved in the study represents, is beyond the scope of this study. For what this study is trying to get at is whether the interactional styles of a learner regulated by his or her culture are affected by contact with other styles of a learner with a different cultural backgrounds and thus whether the contact results in changes in amount of the learner’s initiation of negotiation. This study does not go as far as to see what culture represents what interactional styles, but only considers the degree of differences in interactional styles measured by cultural distance. Koreans are assumed to be culturally closer to Japanese and Chinese subjects than other non-Koreans in that their nations have long been under the influence of Chinese culture, which is believed to have had a strong impact on their general lifestyle including interactional styles. Since particular interactional styles and/or their features of the national cultures considered here have not been defined or identified, there is no basis for prediction on the relationship of 69 cultural distance between interlocutors and negotiation in their interaction, so a null hypothesis is set up: Hypothesis 4 NNS initiates equal amount of negotiation regardless of the cultural distance between the NNS and the NNS’s interlocutor. Results and Discussion A paired t-test was conducted to test Hypothesis 1 and two- way t-tests were done for Hypotheses 2,3, and 4. The criterion level for significance was set at p<0.05. Hypothesis 1 The first hypothesis predicted that there would be more negotiation in the information-exchange task than in the free conversation. 70 Table 3 Paired t-test Comparing Numbers of Indicators in P and F Measure Mean SD T value DF P Value All Types . (a) P 9.83 5.47 44.71 11 p<.05 F 3.08 3.06 CR/CF (Other-motivated) (b) P 4.67 3.03 18.84 11 p<.05 F 2.42 2.84 CP/VM (Self-motivated) (c) P 4.92 0.67 28.33 11 p<.05 F 3.82 0.65 P: Picture Difference P: Free Conversation CR: Clarification Requests CF: Confirmation Checks CP: Comprehension Checks VM: Verification of Meaning As shown in Table 3, this was found to be the case (t=44.7l df=ll p<0.05). The information-exchange task elicits more negotiation than the other task. The big gap between means and high t-value (44.71) indicates that the task type is a very strong predictor of amount of negotiation occurring in NNS/NNS interaction. When analyzed by subcategories of negotiation (i.e., self- versus other-motivated, namely, comprehension checks/verification of meaning versus clarification requests/confirmation checks), the effects of task type remain strong across those types (for clarification requests/confirmation checks df=ll t=l8.84 p<0.05, for 71 comprehension checks/verification of meaning t=28.33 df=ll p<0.05). Thus, it turned out that task type considerably affects negotiation occurring in NNS/NNS interaction as well as NS/NNS interaction, which has been shown in some other studies (Duff, 1986; Gass and Varonis, 1985). Hypothesis 2 Table 4 (a) shows that whether interlocutors share the same Ll background does not have effects on negotiation, rejecting hypothesis 2, which predicted that NNS dyads of different L1 backgrounds rely more on negotiation than those of the same Ll background (t=l.28 df=10). 72 Table 4. Two way t-test Comparing Numbers of Indicators Occurring in S and D Measure Mean SD T value DF P Value All Types (a) S 14.33 5.16 1.28 10 ns D 11.55 5.68 CR/CF (Other-motivated) (b) S 7.00 2.76 -0.38 10 ns D 7.67 5.50 CP/VM (Self-motivated) (c) S 7.33 4.18 2.49 10 p<.05 D 3.83 2.48 All Types in P (d) S 12.33 5.24 2.43 10 p<.05 D 7.33 4.84 All Types in F (e) S 2.00 1.26 1.78 10 ns D 4.17 4.02 S: dyads of same Ll ns: non-significant F: Free Conversation CF: Confirmation Checks VM: Verification of Meaning D: dyads of different L1 P: Picture Difference CR: Clarification Requests CP: Comprehension Checks 73 This seems to be against the finding of Varonis and Gass (1985b), who found that learners of different L1 use more negotiation than those of the same L1 background. However, if we take a closer look at results by, for example, negotiation type and task type, it becomes evident that there are some areas which link L1 background to negotiation. First, as shown in table 4 (b) and (c), while there is not significant difference in using clarification requests/confirmation checks between same L1 and different L1 dyads, the difference is significant in the use of comprehension checks/verification of meaning. Same L1 dyads used more comprehension checks/verification of meaning than different Ll dyads. This means that, since comprehension checks/verification of meaning are moves for feedback on the utterances, learners are more active in asking feedback on their comprehensibility of the same Ll learner than of a different L1 learner. Table 4 (d) and (e) show that while there is no difference in amount of negotiation for the free conversation, a significant difference is detected for the picture difference task. Dyads of the same L1 used more negotiations than those of different L1. Thus effects of L1 backgrounds emerge only in the picture difference task and on comprehension checks/verification of meaning. 74 In spite of evidencing some effects of L1 background, these results, for two reasons,still go against findings of Varonis and Gass (1985b) which evidenced more negotiation by different L1 learners. First, the effects of L1 background seen in this study are in the opposite direction to those in their study. While the sharing of same L1 had a negative effect on negotiation in their finding, a positive relationship is found in this study. Second, they treated L1 background as an independent variable for negotiation. But as seen above, effects of L1 background appears only for the picture difference task. This means that L1 background is an interacting variable depending on task type for its effects on negotiation. Varonis and Gass (1985b) compared the effects of two variables of proficiency and L1 background and found no significant difference. Based upon this finding, they assumed that all variables are independent, and concluded that the amount of negotiation occurring in NNS/NNS conversation is determined by the number of variables in which interlocutors differ rather than the kind of variables. However, the results of this study show that L1 background interacts at least with task type, evidencing that not all variables are independent of all others in their effects on negotiation. 75 Table 5 presents another perspective with regard to the variable of L1 background. Among 6 dyads of the same Ll backgrounds, three of them were of Koreans and the others of non-Koreans(one J/J, one Th/Th, and one Ch/Ch). The table compares Korean dyads with non-Korean dyads. Although the difference in using all indicators between two kinds of groups is not statistically significant, it is big enough not to be ignored (t=2.34 df=4 p<.10). As far as clarification requests/confirmation checks are concerned, Korean dyads use them significantly more than the other same Ll dyads. But no difference is found in the area of comprehension checks/verification of meaning. This implies that a particular L1 group probably tends to use more negotiation than the other groups or has preference to a particular type of indicators. This implication is, however, not so conclusive because (1) the. comparison was made with incomparable numbers of each L1 and (2) amount of negotiation from each non-Korean group was mixed into one category and neutralized for comparison with negotiation occurring from Korean groups . But if studies of a larger size were conducted and their results favored the implication, then it could be that what is more important is what L1 background the interlocutors in the same L1 pairs are from, rather than whether interlocutors share the same L1. 76 Table 5. Two way t-test Comparing Numbers of Indicators Used in K and Non-K Dyads Measure Mean SD T value DF P Value All Types (3) K/K 17.33 5.77 2.34 4 p<.10 non- 11.33 2.52 K/non-K CR/CF (Other- motivated) (b) K/K 9.00 1.73 3.70 4 p<.05 non- 5.00 2.00 K/non-K CP/VM (Self-motivated) (c) K/K 7.33 5.13 0.62 4 ns non- 5.67 4.16 K/non-K CR: Clarification Requests CF: Confirmation Checks CP: Comprehension Checks VM: Verification of Meaning ns: non-significant K: Korean non-K: non-Korean Hypothesis 3 The hypothesis predicted that task type and L1 background are equally effective in eliciting negotiation. As discussed for Hypothesis 2, since L1 background is dependent on task type for its effects, task type turned out to be of more weight. Thus this hypothesis is not supported. Hypothesis 4 The hypothesis that NNS initiates comparable amount of negotiation regardless of cultural distance from his or her interlocutor’s, is supported as shown in Table 6. Three t-tests 77 were done to compare the numbers of negotiation initiated by Korean learners in each of K/K, K/Asian, K/non-Asian dyads. Table 6. Two way t-test Comparing Indicators by Koreans in K/K, K/A and K/N Dyads Measure Mean SD T value Signific ance All Types ofI (a) K/K 8.67 2.89 0.22 ns K/A 9.00 2.00 All Types ofI (b) K/A 9.00 2.00 1.99 ns K/N 6.33 6.86 All Types ofI (c) K/K 8.67 2.89 1.53 ns K/N 6.33 6.86 ns: non—significant K: Korean A: Asian N: non-Asian No significant difference is found in any compared groups, indicating that Korean learners are constant in initiating negotiation regardless of the cultural background of their interlocutor. Thus, at least to Korean learners, their interactional styles regulated by Korean culture are so strong that they do not change according to cultural difference from the interlocutor’s. But I must admit that this result is not so highly complete due to not a few limitations that this study contains concerning 78 the variable of cultural difference. First, in order to see the effects of interactional styles represented by cultures, it is necessary to clearly define the features of interactional styles that each cultural members share. This task was reserved because it is not within direct focus of the present study. However, in order to make sure that the outcome is purely from strong effects of interactional styles of Korean culture, it is needed to clarify what interactional styles the culture represents and regulates and to track the process in which the interactional styles work on negotiation of meaning. Second, the study did not control a possible intervention of another variable. The participants in dyads are not strangers to each other and had been in the same class at least two months. It seems that interactional styles of cultures are believed to be most purely represented when the members interact with strangers or people of higher status. As personal interaction accumulates, a learner’s interactional styles may be more influenced by the learner’s personality and less by the learner’s culture. This means that, among learners who are personally acquainted or known to each other, their interactional styles might be more influenced by their personal characters rather than by their culture. It may be especially so when 79 interlocutors are similar in status linguistically and socially (i.e., same proficiency and same status as learners). Thus, the outcome may not be strong enough because the experiment did not control for the influence of personality, something that would be really hard to do. However, more problematic is the small number of subjects used in this study. Twelve Koreans, three non-Korean Asians, and three non-Asians participated. These numbers were too small to ensure that those subjects are representative of their culture. And in order to make three types of groups (i.e., K/K, K/Asian, K/non-Asian) comparable in effects of cultural distance, three Koreans should have participated across all types of groups. These limitations make it hard to generalize the result of this study even to Korean learners. Reliable evaluation of this variable for its effects definitely deserves more elaborate studies covering all limitations discussed here. Response It was pointed out that the response is important because actual contributions that negotiation makes to SLA are realized through it. Given the importance of the response, it would be also worthwhile to consider which variable elicits more 80 responses to given indicators. Table 7 displays effects of each variable discussed thus far as a driving force for the response. Table 7. Percentages of Responses in Indicators Variable No. of No. of Percentage Indicators Responses of ‘R’s , in ‘I’s Task Type Picture Difference 118 85 72.03 Free Conversation 37 27 72.97 Ll Same L1 Dyads 86 61 70.93 Background Different L1 69 51 73.91 Dyads Type of ‘1’ CR (Other- 36 29 80.56 CF motivated) 52 32 61.54 CP (Self- 6 4 66.67 VM motivated) 62 47 75.81 I: Indicator R: Response CR: Clarification Requests CF: Confirmation Checks CP: Comprehension Checks VM: Verification of Meaning The elicitation of responses is found to have nothing to do with task type and L1 background. The differences between the picture difference task and the free conversation (72.03% versus 72.97%), and between the same L1 and different Ll dyads (70.93% versus 73.91%) are not significantly big. However, the kind of indicators turns out to have an effect. Of other motivated moves, the clarification request is over the confirmation check (80.56% versus 61.54%). The clarification 81 request is a clear form for asking for a response. But the confirmation check can be perceived as a conversational continuant by the interlocutor even when the user employs it to ask for modified input as a substitute for the clarification request. For the confirmation check can function both ways and does not change its form for both functions. This may reduce the chance of getting a response on the part of confirmation check users. Among self-motivated moves, the comparison in terms of percentage was not plausible because the incidences of comprehension checks are relatively too small compared with verification of meaning (i.e., 6 versus 62). However the high percentage of getting a response for verification of meaning (75.81%) shows that it is a strong driving force for eliciting a response. Thus, it is found that chances are high of getting a response when a learner uses clarification requests and verification of meaning over confirmation checks and perhaps comprehension checks. Finally, Table 8 shows the difference in responses Koreans got to indicators that they used across the three types of groups. 82 Table 8. Percentage of ‘R’s to ‘1’s and of CR/VM in ‘I’s Used by Korean ‘R’s in ‘I’s CR/VM in ‘1’s K/K 69.23%(18/26)+ 65.38%(17/26)++ K/A 62.96%(17/27) 40.74%(11/27) K/N 78.95%(15/19) 68.42%(13/19) I: Indicator R: Response CR: Clarification Requests VM: Verification of Meaning + (a/b) : a-number of ‘R’s, b-number of ‘1’s ++(a/b) : a-number of CR/VM, b-number of ‘1’s K: Korean A: Asian N: non-Asian As shown, Korean learners got responses to their indicators most frequently from non-Asian partners (15 out of 19, or 78.95%), and then other Koreans (18 our of 26, or 69.23%), and the least from Asian learners (17 out of 27, or 62.96%). This difference has to do with variation in their use of each type of indicators rather than influence of cultural distance. As indicated above, clarification requests and verification of meaning are most likely to get responses. And Koreans used these moves in frequency differently according to their partners’ cultural background. That is, they used these moves 68.42% to non-Asian partners and 65.38% to Korean partners while they only used 40.74% to Asian partners. (It is an interesting fact that Koreans used similar percentages of clarification requests/verification of meaning to non-Asian partners and Korean partners but a much lower percentage to Asian partners, 83 but a reason for this could not be inferred from the transcribed data. Interviews with subjects might have given a clue to this phenomenon.) They received the highest percentage of responses from the group to which they used the highest percentage of clarification requests/verification of meaning. And they received the least percentage of responses from the group to which they used the least percentage of clarification requests/verification of meaning. Thus, it turns out that regardless of the interlocutor’s cultural background, the higher percentage of clarification requests/verification of meaning they used, the higher percentage of responses to their indicators they got. This indicates that, as with indicators, cultural distance is not an affecting variable for eliciting responses. Summary of Results Task type turned out to be a strong factor in elicitation of negotiation in NNS/NNS interaction as in NS/NNS interaction. Whether to require participants to understand each other’s information or message was suggested as one of main factors to make a task effective with regard to negotiation elicitation. L1 background was dependent on another variable(i.e. task type) in whether and how it affects negotiation. When it becomes affecting factor, against expectation in hypothesis 2 84 and findings in a previous study (Varonis and Gass, 1985b), the same L1 dyads initiated more negotiation than different L1 dyads. More importantly, which L1 the interlocutors share within same L1 dyads may be a stronger variable than whether they share same L1. Cultural distance from the interlocutor and resulting difference in interactional styles do not affect amount of negotiation initiated by Korean learners. However, due to various defects residing in this study - lack of clarity in features of interactional styles belonging to a certain culture, failure to control a possible intervening variable, and small number of subjects, the result is barely conclusive. Among indicators, clarification requests and verification of meaning are effective in inducing responses, through which negotiation virtually facilitates SLA. This explains the positive relationship between the rate of using clarification requests and verification of meaning and the rate of getting responses which is evidenced in comparison of Koreans’ interactions with different cultural background partners. L1 background, task type, and cultural distance does little to eliciting responses. Conclusion and Implications for Classroom and Further Research A majority of studies on negotiation of meaning has focused on one side of each of the following two dimensions. The first dimension involves the kind of ‘setting’ where negotiation takes place. In early studies, most of findings were based upon data extracted from NS/NNS interactions. Varonis and Gass (1985b) convincingly argued that a NNS/NNS interactional setting is no less valuable as a context for negotiation. And they further presented a model for negotiation called ‘non-understanding routine’ to show how negotiation typically proceeds. The other dimension is concerned with the ‘role’ of meaning negotiation. When comprehensible input was suggested as necessary and sufficient for SLA, many studies sought the value of negotiation in making input comprehensible, thus allowing a learner to be exposed to more comprehensible input. As more conditions (e.g., comprehensible output, awareness etc.) were suggested by other studies against or based upon Input Hypothesis, efforts followed to explore the possibility that negotiation may contribute to these newly suggested conditions. Pica (1994) was one who evidenced the role of negotiation beyond making input comprehensible. She suggested that 85 86 negotiation not only assists L2 comprehension but allows direct access to acquisition of L2 forms by serving as context for feedback and production of output. She also showed that this twofold role of negotiation holds in NNS/NNS interaction (1996) as well as in NS/NNS interaction (1994). I presented a model in order to elaborate on the role of negotiation in SLA process within NNS/NNS interaction. The model employed the negotiation model proposed by Varonis and Gass(l985b) to consider different parts that indicators and responses play in SLA. In the model, indicators were subcategorized according to where the trigger lies (i.e., speaker’s on-going or listener’s previous speech) into other- (clarification requests/ confirmation checks) and self- (comprehension checks/verification of meaning) motivated moves. Recall Tables 1 and 2 in which the model is displayed. This model is showing several new perspectives regarding the role of negotiation. First, it reveals that negotiation of meaning certainly contributes to things other than modified input for comprehension needs, that is, feedback and comprehensible output. For example, as seen in Figurel, by initiation of a clarification request by his or her interlocutor, the learner gets 87 to receive feedback on his earlier utterances. Second, it shows that the elements of negotiation model (i.e., indicators and responses) contribute to SLA at different levels. As seen in Figure 1, indicators provide opportunities for SLA conditions when initiated, while responses make it possible that opportunities given by indicators are realized. Thus the model stresses that responses are no less important in negotiation’s facilitative role in SLA, even if they are optional, unlike indicators. With regard to variables affecting incidences of indicators and responses, the current study empirically confirmed that task type is a strong factor in eliciting negotiation, or more specifically and strictly, indicators. The study also found that L1 background could be a factor, if not always. However, neither L1 background nor task type was affecting factors in drawing responses to indicators. Instead, the kind of indicators proved to be the factor with clarification requests/verification of meaning being very effective in this matter. Third, it is shown in the model that not all indicators equally contribute across modified input, feedback and comprehensible output. Through their initiation and response to it, clarification requests/confirmation checks serve opportunities and contexts for realization of all the three 88 conditions while comprehension checks/verification of meaning only feedback on the move user’s utterances. Finally, this model shows that negotiation in NNS/NNS setting benefits both the move user and responder. Exchanging indicators and responses, the user of indicators gets benefits in the areas of modified input and feedback at the same time that the responder in feedback and production of output. This validates the use of pair or group work in ESL classrooms. However, the results of this study suggest some consideration in order for the pair work to be maximized in its effects in classroom contexts. Choosing an effective type of task is critical in negotiation elicitation even in NNS/NNS interaction, given the powerful influence of task type. From the perspective of negotiation, the teacher needs to develop the kind of task which requires information-exchange and understanding of the exchanged information, which are viewed as the driving force for stimulating negotiation. And it was found that only in that kind of task, the variable of L1 background works. That is, same Ll dyads tend to use more negotiation than different L1 dyads. Thus, it would be a desirable context for eliciting a maximum of negotiation of meaning, if learners of the same L1 engage in the kind of 89 information task suggested above. But as further analysis of same Ll dyads revealed, the effects of same L1 composition may vary from L1 to L1. And since the comparisons among the same Ll dyads were based upon data from only two incidences of task, lasting effects of same L1 composition remain in doubt. The study also discovered that clarification requests/verification of meaning are more effective moves than the others in that they induce more responses. However, participants in the study used more confirmation checks than clarification requests (52 versus 36). Given the significant role of the response stressed in the current study, it would be worthwhile for teachers to attempt to lead learners to use more clarification requests/verification of meaning by, for example, using them more frequently during their interactions with learners. This study leaves room for elaboration so that more complete confidence in the assertions and findings of the study can be made by further research. The study attempted to touch on the tricky variable of cultural distance, and the findings about it are only tentative and far from conclusive due to various limitations which could have contaminated the findings. As indicated earlier, for more complete research on this issue, such matters as fine set-up of typologies of interactional styles 90 for each culture being studied and the ensuring of enough numbers of subjects should be taken into consideration and clearly settled. With regard to the model presented, it deals with only four negotiation moves. It surely is not the case that those are all the moves that exist. There may be other moves which could fit into this model. For instance, completion with rising intonation of the interlocutor’s incomplete utterances could be a candidate for other-motivated moves, if the study extended the scope of the trigger into an intended but unspoken utterance. The finer and clearer definition of elements for negotiation routine would result in more elaborate and correct analysis of negotiation role in SLA. The model is also incomplete in that it does not consider such important conditions as involvement and awareness. It would be an interesting area for future research to investigate whether the role of negotiation in fulfillment of these conditions can also be explained within the model presented here. APPENDIX 91 APPENDIX Figure 1. Pictures for Information-exchange Task LIST OF REFERENCES LIST OF REFERENCES Bley-Vroman,R. (1989). What is the logical problem of foreign language learning? In S.Gass & J.Schacter(Eds.), Linguistic perspectives on second language acquisition, 41-68. Cambridge: Cambridge University Press. Bruton,P., & Samuda,V. (1980). Learner and teacher roles in the treatment of error in group work. RELC Journal, 11, 49- 63. Chaudron, C. (1985). Intake: On models and methods for discovering learners’ processing of input. SSLA, 7, 1-14 Corder,S.P. (1967). The significance of learner’s errors. International Review of Applied Linguistics, 5,161-170. Crookes,G., & Rulon,K. (1985). Incorporation of corrective feedback in native speaker/non-native speaker conversation (Tech. Rep. No. 3). The Center for Second Language Classroom Research/Social Science Research Institute, University of Hawaii at Manoa. Doughty,C. & Pica,T. (1986). “Information Gap” Tasks: Do They Facilitate Second Language Acquisition? TESOL Quarterly, 20, 305-325. 92 93 Duff,P. (1986). Another Look at Interlanguage Talk: Taking Task to Task. In R. Day (ed.), Talking to learn: _ Conversation in second language acquisition, 147-181. Cambridge, MA: Newbury House. Pica,T. (1987). Second- Language Acquisition, Social Interaction, and the Classroom. Applied Linguistics, 8, 3-20. Ehrlich,S., Avery,P., & Yorio,C. (1989). Discourse Structure and the Negotiation of Comprehensible Input. SSLA, 11, 397-414. Ferguson, C. (1971). Absence of copula and the notion of simplicity: A study of normal speech, baby talk, foreign talk and pidgins. In D. Hymes (Ed.), Pidginization and creolization of languages, 141-150. Findlay,M. (1995). Who Has the Right Answer? Differential Cultural Emphasis in Question/Answer Structures and the Case of Hmong Students at a Northern California High School. Issues in Applied Linguistics, 6, 23-38. Gass,S. (1988). Integrating research areas: A framework for second language studies. Applied Linguistics, 9, 198-217. Gass,M. & Varonis,E. (1985). Variation in Native Speaker Speech Modification to Non-native Speakers. SSLA, 7, 37- 58. Gass,S. & Varonis,E. (1985). Task Variation and Nonnative/Nonnative Negotiation of Meaning. In S. Gass and C. Madden (eds.). Input in Second Language Acquisition. Rowley, Mass.: Newbury House. Gass,S. & Varonis,E. (1986). Sex Differences in NNS/NNS Interactions. In R. Day (ed.), Talking to learn: Conversation in Second Language Acquisition, 327-351. Cambridge, MA: Newbury House. 94 Gass,S. & Varonis,E. (1989). Incorporated Repairs in Nonnative Discourse. In M. Eisenstein (ed.), The dynamic interlanguage: Empirical studies in second language variation, 71-86. New York: Plenum Press. Gass,S. & Varonis,E. (1994). Input, Interaction, and Second Language Production. SSLA, 16, 283-302. Hatch,E. (1083). Psycholinguistics: A second language perspectives. Rowley, MA: Newbury House. Krashen,S. (1980). The Input Hypothesis. In J. Alatis(Ed.), Eurrent Issues in Bilingual Education, 144-158. Krashen,S. (1983). Newmark’s “Ignorance Hypothesis” and Current Second Language Acquisition Theory. In S.Gass & L.Selinker (Eds.), Language transfer in language learning, 135-153. Lado,R. (1957). Linguistics across cultures. Ann Arbor: University of Michigan Press. Long,M. (1980). Input, interaction and second language acquisition. Unpublished doctoral dissertation, University of California, Los Angeles. Long,M. (1983). Linguistic and Conversational Adjustments to Non-Native Speakers. SSLA, 5, 177-193. Long,M., & Porter,P. (1985). Group work, interlanguage and second language acquisition. TESOL Quarterly, 19, 207- 228. 95 Loschky, L. (1994). Comprehensible Input and Second Language Acquisition. SSLA, 16, 303-323 Parker,K., & Chaudron,C. (1987). The effects of linguistic simplications and elaborative modifications on L2 comprehension. University of Hawaii working Papers in English as a Second Language, b, 107-133. Pica,T. (1987). Second language acquisition, social interaction, and the classroom. Applied Linguistics,8, 3-21. Pica,T. (1988). Interlanguage Adjustments as an Outcome of NS-NNS Negotiated Interaction. Language Learning, 38, 45-73. Pica,T. (1992b) The Textual Outcomes of Native Speaker - Nonnative Speaker Negotiation: What Do They Reveal About Second Language Learning? In C.Kramsch & S. McCommell- Ginet(Eds.), Text and context: Cross-disciplinary perspectives on language study, 198-237. Lexington, MA:D.C. Heath. Pica,T. (1994). Research on Negotiation: What Does It Reveal About Second-Language Learning Conditions, Processes, and Outcomes? Language Learning, 44, 493-527. Pica,T. & Doughty,C. (1983). The Rloe of Group Work in Classroom Second Language Acquisition. SSLA, 17, 233-248. Pica,T., Holliday,L., Lewis,N., Berducci,D., & Newman,J. (1991). Language Learning through Interaction. SSLA, 13, 343-376. Pica,T., Holliday,L., Lewis,N., & Morgenthaler,L. (1989). Compreehnsible Output as an Outcome of Linguistic Demands on the Learner. SSLA, 11, 63-90. 96 Pica,T., Kanagy,R., & Falodun,J. (1992) Choosing and Using Communication Tasks for Second Language Instruction and Research. In S.Gass & G. Crookes (Eds.), Task-based learning in a second language, 9-34. London: Multilingual hdatters. Pica,T., Lincoln-Porter,P., Paninow,D., & Linnell,J. (1996). Language Learners’ Interaction: How Does It Address the Input, Output, and Feedback Needs of L2 Learners? TESOL Quarterly, 30, 59-84. Pica,T., Young, R., & Doughty,C. (1987). The Impact of Interaction on Comprehension. TESOL QUARTERLY, 21, 737-758Varonis,E. & Gass,S. (1982). The Comprehensibility of Non-Native Speech. SSLA, 4, 114-136. Porter,P. (1983). Variations in the conversations of adult learners of English as afunction ofproficiency level of the participants. Unpublished doctoral dissertation, Stanford University. Porter,P. (1986). How Learners Talk to Each Other: Input and Interaction in Task-Centered Discussions. In R. Day (ed.), Talking to learn: Conversation in second language acquisition, 200-224. Cambridge, MA: Newbury House. Rost,M. & Ross,S. (1991). Learner Use of Strategies in Interaction: Typology and Teachability. Language Learning, 41, 235-273. Sato,C. (1982). Ethnic Styles in Classroom Discourse. In M. Hines & W. Rutherford (Eds.),On TESOL ‘81, 11-24. Sato, C. (1986). Conversation and interlanguage development: Rethinking the connection. In R. Day (ed.), Talking to learn: Conversation in second language acquisition, 23-45. Cambridge, MA: Newbury House. 97 Scarcella,R. (1983). Discourse Accent in Second Language Performance. In S. Gass & L. Selinker (Eds.), Language transfer in language learning, 306-326. Roeley, MA: Newbury House. Scarcella,R. & Higa, C. (1981). Input, Negotiation, and Age Differences in Second Language Acquisition. Language Learning, 31, 409-438. Schacter,J. (1988). Second language acquisition and its relationship to Universial Grammar. Applied Linguistics, 9, 219-235. Schacter,J. (1989). testing a proposed universal. In S.Gass & J.Schacter (Eds.), Linguistic perspectives on second language acquisition, 73-88. New York: Cambridge University Press. Schacter,J. (1990). On the issue of completeness in second language acquisition. Second Language Research, 6, 93- 124. Schmidt,R. (1990). The Role of Consciousness in Second Language Learning. Applied Linguistics, 11,129-158 Seliger, H. (1977). Does Practice Make Perfect?: A Study of Interaction Patterns and L2 Competence. Language Learning, 27, 263-278. Stevick,E. (1976). Memory, meaning and method. Rowley, MA: Newbury House. Stevick,E. (1980). Teaching lnguages: A way and ways. Rowley, MA: Newbury House. 98 Stevick,E. (1981). The Levertov Machine. In R. Scarcella & S. Krashen (Eds.), Research in second language acquisition, 28-35. Rowley, MA: Newbury House. Svanes,B. (1987). Motivation and Cultural Distance in Second Language Acquisition. Language Learning, 37, 341-359. Swain,M. (1985). Communicative competence: Some roles of comprehensible input and comprehensible output in its development. In S. Gass & C. Madden (eds.), Input in second language acquisition, 235-253. Rowley, MA: Newbury House Tomlin,R. & Villa,V. (1994). Attention in Cognitive Science and second Language Acquisition. SSLA,]6, 183-203. Varonis,E. & Gass,S. (1982). The comprehensibility of non- native speech. SSLA, 4, 114-136. Varonis,E. & Gass,S. (1985a). Miscommunication in native/nonnative conversation. Language in society, 14, 327-343. Varonis, E. & Gass, S. (1985b). Non-native/Non-native Conversations: A Model for Negotiation of Meaning. Applied Linguistics, 6, 71-90. White,L. (1987). Against Comprehensible Input: the Input Hypothesis and the development of Second-Language Competence. Applied Linguistics, 8, 95-110. White,L. (1989), Universal Grammar and Second language acquisition. Amsterdam: John Benjamins. Young,R. (1988). Input and Interaction. Annual Review of Applied Linguistics, 9, 122-134