Tilburg University Dealing with problems in intercultural communication Deen, Jeanine Yolanda Publication date: 1995 Link to publication in Tilburg University Research Portal Citation for published version (APA): Deen, J. Y. (1995). Dealing with problems in intercultural communication: a study of negotiation of meaning in native-nonnative speaker interaction. [s.n.]. General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Download date: 21. Jan. 2022 ~eaiing . Jeanine Deen with P rob lems A Study of Negotiation of Meaning in Native-Nonnative Speaker Interaction Intercultural Communication i , Dealing with Problems in Intercultural Communication I„ Dit werk terug te bezorgen uiterlijk op: ~ í~-iA~i~,-!."' Bibliotheek - Katholieke Universiteit Brabant Postbus 90153 5000 LE Tilburg Telefoon: 013 - 466 21 24 Dealing with Problems in Intercultural Communication A Study of Negotiation of Meaning in Native-Nonnative Speaker Interaction Proefschrift ter verkrijging van de graad van doctor aan de Katholieke Universiteit Brabant, op gezag van de rector magnificus, prof. dr. L.F.W. de Klerk, in het openbaar te verdedigen ten overstaan van een door het college van dekanen aangewezen commissie in de aula van de Universiteit op vrijdag 8 december 1995 om ]4.15 uur door Jeani~ie Yolanda Uee~i geboren op 13 augustus 1958 te Groningen Promotores: Prof. Dr. G. Extra Prof. Dr. R. van Hout K.U.B. B,BLEQTHEEK TiLB~1~G To the ntentory of Michael Canale, nzy dearfriend and inspiring teacher CIP-gegevens Koninklijke Bibliotheek. Uen Haag Deen. leanine Yolanda Dealing with problems in intercultural communication ~ a study of negotiation of rneaning in native - nonnauve speaker mieracuon 1 leanine Yolanda Deen. -[S.I. s.n.] IGroningen Universiteitsdrukkenj). - III. Prcefschrift Katholieke Universiteit Brabant. Tilburg. Met lit. opg. - Met samenvatting in het Nederlands. ISBN 90-9009037-1 NUG1471 Trefw. interculturele communicatie I tweedetaalverwerving. Lay out and Cover design: Paul Noort Typesetting: Ruurd van der Weij Printed by Universitei[sdrukkerij, Groningen O leanine Deen A(1 rigltt.r re.ten~ed. Nn purt of lhis publicution mu~ be reproduced, stnred rn u retriernl.rystent, or trun.wnitted, in cut}~,jnnn nr by ruty meuns, elecrrorur, nrechunicul, phutocapi~in,e, recording, or orherx~ise, without dre prrarx ritten penmssinn ot rhe ~ utlrnr. ISBN 90-9009037-1 Preface The book in your hands is the result of a child's dream, the dream of becoming a professor at a university. I quietly and steadíly worked at this dream; however, I hesitated before I embarked on the last part of the trip, the writing of this dissertation. I went to the United States to prepare myself further academically, but also to gain more teaching experience so that I would not lose contact with the world out there. The topic of the dissertation may show you that I did not want to become a resident of the ivory tower only, but also wanted my work to be relevant to the surrounding world. Intercultural communication has become a topic of much interest in recent decades. The globalization of the world is resulting in increasing contacts between people from different cultures and language backgrounds. These contacts create challenges in communication, particularly if the participants' cultural backgrounds differ greatly and language proficiency is limited. My dissertation describes how native speakers of Dutch and nonnative speakers with limited proficiency in this language deal with problems in communication. Now I have reached my destination. However, I did not do it on my own. When a dissertation project comes to an end, it becomes clear how many people have supported its realization. Firstly, I'd like to thank my professors at UCLA Evelyn Hatch and Mary McGroarty, who gave me the confidence to start my PhD. Secondly, I'm indebted to the researchers and participants of the European Science Foundation Project `Second Language Acquisition by Adult Immigrants', of which this dissertation is a spin-off. I would specifically like to thank Guus Extra and Roeland van Hout for their overall guidance, and Rceland in particular for the many hours spent on the data analyses, which taught me to work meticulously with figures. I would like to thank my former colleagues and roommates in Tilburg, Gertrud Lemmens and Saskia Schenning, for sharing the ups and downs in the first phase of the research project. Titia Meier and Martin Reddeman have made my later trips from Groningen to Tilburg unforgettable by providing me with a much-needed bed and with many enjoyable conversations over a good glass of wine. Furthermore, I would like to thank everyone who helped in the last stressful phase with giving feedback on unfinished versions of chapters and with inputting corrections: Maaike Dautzenberg, Eveline Deen, Hilde Hacquebord, Harrie Mazeland, Henk van Essen, Karen Kukurenda, Annediet Smit and Areke van der Sluis. I would like to thank Elizabeth Wall for cor- recting my English. I thank Paul Noort for the layout and cover design and Ruurd van der Weij for the typesetting. Both showed to me a new world beyond the content of the book which influences the reception by the reader. I learned that writing a book is not the same as making a book. Finally, I would like to thank my parents for their general support, and particularly for the house cleaning in the last phase, and my friends for showing interest and preparing meals when I did not have time to cook them myself. Contents Chapter 1 1.0 1.1 1.2 1.3 1.4 2.3.4 2.4 2.4J 1 Introduction 1 Mediterranean Minorities in the Netherlands 2 Negotiation of Meaning in NS-NNS Interaction 4 Objectives 7 Contents of the Book 8 Chapter 2 2.0 2.1 2.2 2.2.1 2.2.2 2.3 2.3.1 2.3.2 2.3.3 Aims and Goals of the Study Negotiation of Meaning in Intercultural Communication Introduction 11 Intercultura) Communication 12 Concepts of Competence 14 Interpersonal and Ititercultural Communication Competence 14 Linguistic Concepts of Competence 15 Interaction and Second Language Acquisition 18 ForeignerDiscourse 19 Comprehensible Input 20 Comprehensible Output 21 NNS Communication Strategies 22 Problematic Communication 25 Understandi~tg 25 2.4.2 Miscommunication 2.5 2.5.1 Negotiation of Meanin~ in Clarification Sequences The Micro-Analysis of Interaction 28 27 2.5.2 Repair 29 2.5.3 2.5.4 2.6 2.6.1 2.6.2 30 31 AsymmetricalInteraction 34 Power. Dominance and Asynunetn~ Institutionallnteraction 35 2.6.3 2.7 Negotiation of Meaning Clarification Seguences 35 Native Speaker-Nonnative Speaker Interaetion 36 Conclusions 39 28 11 Chapter 3 Researeh Questions and Method 41 3.0 3.1 3.2 Introduction 41 Research Questions 41 Informants 44 3.2.1 ~3.2.2 The Noruiative Speakers The Native Speakers 46 3.3.1 3.~3.2 The Institutional Conversations 49 The lnformal Conversations 50 3.4 3.4.1 3.4.2 Data Analysis 50 The Selection of Clarification Sequences 50 The Codbag Process 51 3.3 Data Chapter 4 48 The Organization of Clarifieation Sequences 4.0 4.1 4. I J Introduction 55 The Negotiation Moves Main Line Moves 57 4.1.2 Side Line Moves 4.2 4.3 4.4 55 55 58 NS and NNS Clarification Sequences 63 Some Aspects of Sequential Organization 66 Conclusi~~ns 69 Chapter 5 5.0 5.1 45 The Amount of Negotiatian of Meaning 71 5.2 Introduction 71 The Influence of NNS Communicative Competence and Interaction Type 71 Results 73 5.2.1 5.2.2. 5.3 The hiformal Conversations 76 The lnstitutional Conversations 79 Conclusions 85 Chapter 6 6.0 6.1 6.2 6.2. I ó.2.2 Asymmetry in Negotiation of Meaning 87 Introduction 87 The Intluence of NNS Communicative Competence and Interaction Type 87 Results 90 Trouble Indicators 91 Trouble Clarifications 94 6.2.3 6.3 Confirntutinn Checks 95 Conclusions 101 Chapter 7.0 7.1 7.1.1 7J.2 7.1.3 7.2 7.2.1 7.2.2 7.3 7.3.1 7.3.2 7.4 7.5 7 Indieating Trouóle in Understanding 103 Introduction 103 A Classification of Trouble Indicator Types and Forms Trouble Indicator Tj~pes 107 Trouble Indicator Fonns 109 E.rpectations 114 Native Speaker Trouble Indicators 115 The Influence of Interaction Type 117 htdividual Portraits of the Native Speakers 119 Nonnative Speaker Trouble Indicators 122 The Influence of fnteraction T~~pe 123 Individual Portraits of the Nnnnative Speakers 125 Comparing NS and NNS Trouble Indicators 132 Conclusions 135 Chapter 8 Clarifying Trouble in Understanding 104 137 8.0 8.1 8.IJ Introduction 137 A Classification of Trouble Clarification Types and Forms Trouble Clarification Types 138 8. L2 8J.3 Trouble Clarification Forms Expectations 145 145 8.2 8.2.1 Native Speaker Trouble Claritïcations 147 The lnflue~tce of hneraction Tjpe 150 8.2.2 8.2.3 Individual Portraitsof the Native Speaker.s Ungramntatica! Foreigner Talk 157 8.3 8.3.1 8.3.2 8.4 8.5 154 Nonnative Speaker Trouble Clarifications 159 The Influence of Interactiort Type 161 Individual Portraitsof the Nonnative Speakers Comparing NS and NNS Trouble Clarifications Conclusions 170 Chapter 9 Requesting Confrmation 138 162 168 173 9.0 Introduction 9.1 9.1.1 A Classitïcation of Confirmation Check Types and Forms CorTfirmation Check Types 174 173 9.1.2 9.1.3 Cortfirmation Check Fonns 175 Expectations I76 174 9.2 9.2.1 Native Speaker Confirmation Checks 177 The hifluence of Interaction Tj~pe 18l 9.2.2 9.3 9.3.1 Individual Portraits of the Natii~e Speakers 184 Nonnative Speaker Confirmation Checks 186 The Influence of Interaction Type 188 9.3.2. 9.4 9.5 Individual Portraits of the Nonnative Speakers Comparing NS and NNS Confirmation Checks Conclusions 197 Chapter 10 10.0 10.1 ] 0.2 10.2.1 10.2.2 10.3 10.4 Conclusion 190 193 201 Introduction 201 Summary and Discussion of the Results 201 Theoretical Implications of the Findings 212 Intercultural Communication Theory 212 Seeond Language Acquisition Theory 2l3 Practical Implications of the Findings 216 Scope of the Study and Suggestions for Further Research 217 List of References Appendices Appendix I: Appendix 2: Appendix 3: 221 231 List of abbrevations 231 Transcription conventions 232 Tables 233 Samenvatting 243 Resume I Curriculum vitae 250 Chapter 1 Aims and Goals of the Study 1.0 Introduction Communication between people of different linguistic and cultural backgrounds has increased enormously in recent decades. This sharp increase creates challenges to effectively deal with intercultural communication in the modern world. The influx of migrant groups in modern post-industrialized societies and regular contacts between members of majority and minority groups create the need for intercultural and multilingual competence on the side of the receiving society as well as the newcomers. In addition, the increased possibilities for international communication lead to the same need in contexts such as international diplomacy and commerce. Many communication problems arise when nonnative speakers (NNS) have to communicate in a second or foreign language with native speakers (NS) or with nonnative speakers with other cultural and language backgrounds. These problems may be caused by cultural differences which may lead to diverging expectations concerning the content, goals and process of the interaction. However, another important source of communication problems may be the NNS's limited second language proficiency. In other words, such communication is not only intercultural but also interlingual. The extract which follows is an example of communication problems caused by limited language proficiency. A Turkish NNS of Dutch does not understand the question asked by a Dutch housing official about the kind of house he~ is looking for. The Turk reacts by repeating words used by the NS. The NS, in turn, solves the problem by repeating part of the original question, by suggesting a possible answer ("apartment"), and by checking the NNS's answers.2 I 2 Often only 'he' is used for practical reasons. However, both 'he' and 'she' are intended. Transcription conventions can be found in Appendix I. The translations of the examples in English are made in such a way that they approach the Dutch original ac closely as possible, also in degree of grammaticality. 1.0 Irttroduction 1 (1) Institutional conversation NS: ja t uh tt even kijken hoor t Iwat voor huis wil je hebben? NNS: huis? NS: wat voor huis? NNS: voor NS: uh t een flat? NNS: nee (x) normaal thuis he'~[Ix)1 NS `[jij] wilt een normaal huis? NNS: ja normaal niet duur he NS: en niet duur? aughs NNS: beetje gceie he NS: aughs t ja t aa dat heb ik met f ik heb geen gewoon huis voorjou yes t uh tt just a minute t Iwhat kind of house do you want? house? what kind of house? kind uh t an apartrmnt? no Ix) normal home huh `[(x)j `[you] want a normal house? yes normal not expensive huh and not expensive? aughs bit good huh aughs t yes t ah ( don't have that f 1 don't have a normal house for you Problems in understanding may be local, e.g., concentrated around specific lexical items as above, or they may be so extensive that only limited verbal interaction is possible. Gass 8c Varonis (1991) state that a common language between speaker and hearer is a bridge to understanding. However, in NS-NNS communication this bridge can be very shaky. Especially if the speaker and hearer share little common socio-cultural background and the second language knowledge of the nonnative speaker is limited, "conversation is likely to be peppered with interruptions for clarifications of content and language form" (Gass 8c Varonis, 1991: 122). Such interruptions often lead to side sequences in which trouble is clarified, as is exemplified in Example (]). In such clarification sequences the interlocutors attempt to come to an acceptable level of mutual understanding. This process is commonly referred to by the term negotiation of ineaning (Pica, 1994). More research is necessary on the way NS and NNS deal with such problems, which roles they play in negotiating, and how negotiation is influenced by factors such as level of NNS communicative competence and institutional context. In this chapter the object of study, negotiation of ineaning in NS-NNS interaction, is introduced. In section l.l the Dutch societal context of minorities is described from which this study derives its ecological validity. The theoretical framework is described in section 1.2. Following this are the objectives of the study in section 1.3. The chapter ends with an outline of the book in section 1.4. 1.1 Mediterranean Minorities in the Netherlands Since World War II, the Netherlands has, like many other Western-European countries, received many migrants from different parts of the world, such as its former colonies and Mediterranean countries. In the 1960s, many male migrants were re- 2 Clmpter 1 Ainis and Goals ajthe Stud~~ cruited for unskilled jobs because of a labor shortage. In the Netherlands, the majority of these Mediterranean men came from rural areas in Turkey and Morocco. Generally, these men's educational background was limited. Most of them intended to work in the Netherlands for a limited period of time in order to save enough money to build a better life in their home country. They frequently lived and worked with compatriots and did not have much social contact with Dutch native speakers. In the Netherlands, these migrants were confronted with many challenges and problems arising from having to live and work in unfamiliar surroundings. Communication with Dutch speakers was a problem. Often no common language for intercultural communication was available. The Dutch are commonly not familiar with Turkish, Moroccan Arabic or Berber languages, and interpreters for these languages were often not available. Therefore, these migrants needed to learn some Dutch to take care of their needs. However, because of their work load and short-term perspective, investing time and energy in learning Dutch often did not have priority. When it became clear that the economic situation in the home countries was not improving, the migrants' perspective changed. Many male migrants decided to stay in the Netherlands longer and had their families join them. Recent Central Bureau of Statistics data indicate that out of a population of approximately 15 million about 190,000 Turks and 150,000 Moroccans were living in the Netherlands (Roelandt et al., 1992). About 40qc live in four big cities: Amsterdam, The Hague, Rotterdam and Utrecht. Now with a perspective of more permanence, the incentive became greater to develop communicative competence in Dutch in order to be able to take care of themselves and their families. In the past, Dutch native speakers hardly ever communica[ed in Dutch with foreigners. Most foreigners who came to the Netherlands spoke one of three languages taught at Dutch secondary schools: English, German or French. Thus these foreigners were addressed in one of these languages. For the Dutch now living in the big cities, it has become habitual to use their own native language to communicate with migrants. Only on some occasions were Moroccans able to use a modern foreign language, that is, French, in their contacts with the Dutch, but often competence in this language was minimal on one or other side. Gumperz (1990: 223) states that in modern post-industrial individualistic societies control over one's life depends on the individual's capacity to communicate effectively. Most Mediterranean migrants come from countries with collectivistic rather than individualistic norms and values (Hofstede, 1991: 68-103). In these countries, one can often take care of external contacts [hrough intermediaries. However, in modern Western societies this is not possible: "bureaucratization of public life in the modern welfare state has brought about the elimination of the stratum of informal leaders and intermediaries who formerly played a key role in communication between local people and official organizations" (Gumperz, 1990: 225). Dealing with governmental agencies for immigration, employment, and housing, with superiors and colleagues at work, with service personnel in stores, with the physician or the plumber cannot generally be left to others. l. I Mediterranean Minorities in the Netherlands 3 Thus, many migrants had to learn more Dutch to gain control over their lives. Most of them received little or no instruction in Dutch as a second language. They mainly learned Dutch through daily interaction with iheir Dutch speaking bosses, colleagues, and with service personnel in stores and governmental institutions. These daily encounters are not very conducive to language learning. In addition to discrimination, migrants frequently have to face the embarrassment and frustration of communication problems. Such problems may be negotiated depending on time pressure so that understanding is achieved and goals can be achieved. However, learning Dutch as a second language under such conditions is a complex task for migrants with a limited educational background and little or no experience in learning foreign languages (Bremer et al., 1993). 1.2 Negotiation of Meaning in NS-NNS Interaction Conversation analysts have studied the process of negotiation of ineaning and repair strategies in NS-NS interaction in order to discover the conventions of conversational structure. Studying the points where communication fails in some way or other has led to insight into the NS rules and structure behind it. Schegloff et al. (1977: 381) stated that: "An adequate theory of the organization of natural language will need to depict how a natural language handles its intrinsic trouble sources. Such a theory will, then, need an account of the organization of repair". Looking at NS-NNS interaction is more complicated because the NNS has not yet fully acquired the conversational rules and structure in the target language. However, negotiation of ineaning in such interaction is of crucial interest because it can create insight into the process of natural language acquisition in interaction. In interaction with NSs, NNSs find themselves in the paradoxical situation that they have to communicate to learn, while they are learning to communicate. In other words, the nonnative speaker has to face two tasks at the same time: - to communicate using his limited L2 repertoire in an optimal way, receptively as well as productively - to acquire the target language (Klein, 1986: 17). Typical of language acquisition in interaction is that the focus is on communication rather than on learning. There is little time for metalinguistic reflection. According to Bingham Wesche (1994: 229) three types of features are typical for NS-NNS interaction. First, there are features which sustain the conversation, such as the restriction to "here-and-now" topics. Secondly, there are features which prevent problems occutring. On the one hand, a NS may use yeslno questions to which a NNS only has to give a one-word answer. On the other hand, a NNS may employ avoidance strategies (of words, topics) which are effective for communication purposes but not contributory to language learning. Thirdly, there are features which repair problems. This last type of feature concerns how NSs and NNSs negotiate meaning when they have trouble understanding each other, and is the focus of the current study. 4 Chapter 1 Aims and Goals of the Study The foremost goal of negotiation of ineaning in clarification sequences is to increase communicative effectiveness. However, some researchers claim that negotiation of meaning also stimulates language acquisition. Linguistic and conversational adjustments resulting from negotiation of ineaning are claimed to be important in creating comprehensible input, which is taken to be a necessary condition for second language acquisition (Long, 1983a). Pica (1987) remarks that no coherent picture has emerged which accounts for successful and unsuccessful second language acquisition. However, there is agreement that the learning environment must include the following: - it should create opportunities for learners to engage in meaningful social interaction with NSs to discover (socio)linguistic rules (Hatch, 1978, 1983, Krashen, 1981, 1982, Long 1981, 1983b); - NNSs and NSs should share an equal need or desire to communicate with each other; - mutual understanding should be reached through restructuring the interaction (negotiation of ineaning) and not through prestructured adaption that impoverishes the input (Long, 1983c). Pica et al. (1989) conclude that negotiation of ineaning is the most vital source of data for second language acquisition because it leads to reswcturing of the input. They state that "through the study of negotiation, what is emerging is an understanding and appreciation of what both learners and interlocutors contribute to the SLA process" (Pica et al., 1989: 84). If this is the case, in spite of the paradox, communication and learning may indeed go hand in hand. Klein (1986: 18-19) concluded that a great deal more research is done on guided second language learning than on spontaneous acquisition. He regrets this because spontaneous language acquisition through interaction is a facility which has evolved in human beings over millennia. Language instruction, on the other hand, is a form of intervention in this natural process. Therefore, he argues that the study of second language use and acquisition in communication should anticipate the study of second language learning through systematic and intentional intervention. In addition, Faerch 8c Kasper (1985: 22) state that negotiation of ineaning should be investigated in spontaneous second language communication in order to gain insight into how it can be manipulated in the classroom to stimulate language acquisition. From 1982 to 1987, a longitudinal and cross-linguistic multiple case study on spontaneous second language acquisition by adult immigrants was carried out under the auspices of the European Science Foundation (ESF) (Perdue, 1984 and 1993a1b, and Feldweg, 1991). Three research yuestions were posed. The first one concerned major factors which influence second language acquisition: cognitive disposition (source language), propensity factors (needs, attitudes), and exposure to the language. The second question concerned the general structure of second language acquisition with respect to the order of acquisition, the speed and success of the acquisition process. The third yuestion concerned the characteristics of communication between NSs and NNSs (Perdue, 1993a: 2). Perdue claims that untutored second language acyuisition cannot be completely understood, unless one studies how a learner communicates in the target language using a restricted 1.2 Negntiation of Mea~:ing in NS-NNS Interaction 5 repertoire of grammar and lexicon in combination with general nonverbal and discursive communication skills. ]n the ESF project, conversational and experimental data were collected of nonnative speakers with limited proficiency from different source language backgrounds in five countries of the European Union. The source languages were paired with target languages in the following way. Target Language Source Language Swedish i~., Finnish Dutch ~~. French i ~~~ . Spanish German ,' ~ ~ Arabic English ~~ ~ ~~~ Turkish Italian Punjabi Figure l. l: Lang~uiges of the ESF project The NNS informants all belonged to major immigrant groups in the countries under investigation. Some of the source languages chosen are closely related to the target language (TL) in question, whereas others are widely disparate. The two source languages of the Dutch-as-a-second-language speakers, Arabic and Turkish, were widely disparate from Dutch as well as from each other. Data of the NNSs interacting and performing tasks were collected over a ninemonth period with intervals of about a month. This data collection was repeated twice over time, so that eventually comparable data of three cycles of nine months became available for longitudinal analyses (Perdue, 1993a). The longitudinal dimension of this project was of major importance. Klein 8c Perdue (1988: 5) have stated that this dimension has been present in very few studies, probably due to time and money related considerations. The scope of this project, subsidized by the European Science Foundation, made it possible to collect data over a longer period of time. Part of the Dutch data were used for the current study. At the end of the project, six research reports were written, which are summarized in two books (Perdue. 1993 alb). Four reports focused on aspects of language production: utterance structure, word formation processes in talking about entities, the acquisition of temporality, and reference to space in learner varieties. Two reports dealt with native-nonnative speaker interaction: feedback in adult language acquisition and ways of achieving understanding. The first study of feedback (Allwood, 1993) focussed on how NNSs give and elicit feedback. Fourteen categories of feedback units and words were distinguished. Allwood concludes ihat the NNS's use of feedback is most crucial in the earlier stáges of acquisition, when they use most single feedback words in a multifunctional way. Words such as yeah and trihnt do not require much linguistic competence but they are particularly useful when understanding is difiicult to achieve. The second interaction study (Wa~~s nf achies~ing totderstmtdin~q, Bremer et al. 1988I1993) dealt with those instances of interaction in which problems in under- 6 Chnpter l Airrts aud Goals o(the Su~d~ standing arise. Bremer et al. studied three aspects of the understanding process in institutional conversations. Firstly, they studied causes of problems in understanding by analyzing trouble sources with which clarification sequences start. These were classified as mishearings, relative diftïculty of an utterance, and pragmatic lack of understanding. This classificatíon appears to be quite global. Relative difficulty of an utterance máy be caused by morphology, lexicon, syntactic complexity, reduced forms, lack of orientation on the "here-and-now", etc. Nor does the classification seem to be systematic or exhaustive. Secondly, they studied the ways NNSs (but not NSs) indicate problems in understanding, which is an important part of the process of negotiating meaning, (see also the discussion in Chapter 7). And thirdly, they described joint NS and NNS negotiation of understanding. Although understanding is seen as an interactive process, the analyses focussed mainly on either the NNS or the NS (Bremer et al., 1993: 153). Finally, the methodology used for the Ways ofachieving understatrding project was purely quali[ative. Evidence given is anecdotal and thus gives no insight into its representativeness for the rest of the data. Bremer et al. (1993) conclude that the analysis of problems in understanding can give insight into the process of understanding only if the interaction behavior of NS and NNS are studied in relationship to each other. Although their work gave a start to a more interactive perspec[ive on negotiation of ineaning, it did not fully succeed in doing that. Only some of the negotiation moves in clarification sequences were analyzed and the organization of the moves in interactional sequences was not described. The current study follows up on the Ways of achieving understanding study with a more extensive and in-depth analysis of the Dutch data. The focus will be on the interactional process of achieving understanding through negotiation of meaning. Therefore, clarification sequences were studied for the type of moves the NS and NNS use in negotiating meaning. Not only [rouble sources and indicators are part of such sequences. Other moves such as clarifications and confirmation checks were also studied (see also Pica, 1994). In addition, several feedback items are described which have important functions in negotiating meaning. A description of the sequential organization of the negotiation moves is also given. Furthermore, the analyses of both informal and institutional interaction make possible comparison of NS and NNS behavior in different situational contexts. The present study is mainly qualitative but also contains basic quantitative analyses to gain insight into relative frequencies of the phenomena studied. 1.3 Objectives In this longitudinal study, negotiation of ineaning is described in interaction between native and nonnative speakers of Dutch. Research on second language acquisition and interaction has had a strong Anglo-Saxon bias (Ellis, 1987: 74; Levinson, 1983: 284 8c 368). The ESF project, however, gives a wealth of data in !.3 Objectives 7 five different target languages used by nonnative speakers from widely disparate language backgrounds. The analysis of the Dutch data add to our knowledge about universalities and specificities of language learning and interaction. Several aspects of the study need specific attention. In this study, NS and NNS behavior are not analyzed separately but in interactional sequence. The first aim of the study is to analyze both NS and NNS rnoves in negotiation~f ineariing with the same coding system to describe "all productions of learners and their interlocutors as they negotiate the meaning of their messages to each other" (Pica et al., 1989: 84). This makes comparison on both speakers' contributions to the negotiation process possible. The second aim is to determine the amount of interaction spent on negotiation of méaning. This amount has consequences for communicative effectiveness as well as for the opportunities for language acquisition. The third aim is to analyze the main negotiation moves, trouble indicators, trouble clarifications, and confirmation checks as to their forms and strategy types. These analyses may provide insight into the means NSs and NNSs use to realize their negotiation moves. This type of analysis was propagated by Pica et aL (1989). In addition, the longitudinal research design makes it possible to investigate the iníluence of NNS communicative competence on the quantity and quality of negotiation of ineaning over time. There has been no longitudinal research done so far on interaction with less proficient adult NNSs. The comparison of the NNS's negotiation of ineaning behavior at different points in time may provide new insights into the relationship between level of communicative competence and ways of negotiating meaning. Two different types of interaction are analyzed: informal and institutional conversations. Typically, in institutional conversations, one speaker is a representative of an institution, and as such has an authoritative role and more or less controls the interaction. The interaction is goal-oriented. In informal conversations, speakers have no roles or script. The interaction is not goal-oriented. Thus the fourth aim was to investigate the influence of a script and institutional roles on nego[iation of ineaning. Both interaction type and NNS communicative proficiency may lead to asymmetry in interaction. 1.4 Contents of the Book This thesis consists of ten chapters. In Chapter l, the aims and goals of the study were described. Chapter 2 contains a discussion of the relevant literature on intercultural and problematic communication. on concepts of competence, on interaction and second language acquisition, on negotiation of ineaning, and asymrnetric interaction. Chapter 3 presents the research questions and method, including a description of the data and informants. Chapter 4 presents a model for the analysis of clarification sequences in NS-NNS interaction and describes the moves found to be present. Chapter 5 contains quantitative results on the influence of NNS communicative proficiency and interaction type on the amount of negotiation of 8 Chupter ! Aints and Gouls ~~jthe Smdi meaning. Chapter 6 focusses on the respective roles of the NS and NNS in the interaction. It discusses the influence of NNS communicative proficiency and interaction type on interactional asymmetry. In Chapters 7, S and 9 the three main moves in the negotiation process are described in more detail. These are trouble indicators (Chapter 7), trouble clarifications (Chapter 8) and confirmation checks (Chapter 9). NSs and NNSs are compared according to the strategy types and linguistic forms they use for these three moves. NNS communicative proficiency and interaction type are considered for their influence on the realization of these negotiation moves. Chapter 10 contains a summary of the results, conclusions for intercultural communication and second language acquisition theory, practical implications of the findings, an evaluation ~f the study, and suggestions for further research. 1.4 Contents of the Book 9 Chapter 2 Negotiation of Meaning in Intercultural Communication 2.0 Introduction This chapter describes social science and linguistic studies in intercultural nativenonnative (NS-NNS) communication. It is argued that social science studies on intercultural communication largely ignore language (proficiency) issues (see section 2.1). To be effective in intercultural communication, both a speaker who speaks in the mother tongue as well as a nonnative speaker of that language need certain general and language~culture specific knowledge and skills. The concept of such competence in intercultural communication is discussed in section 2.2. In [he social sciences, intercultural communication competence has for the most part been viewed as general interpersonal knowledge and skills. The role of language and culture specific knowledge and skills has received little attention whereas the lack of such skills distinguishes intercultural communication from other forms of interpersonal communication (see section 2.2.1). In linguistics on the other hand, competence was deiined by Chomsky as knowledge of language structure. Later this concept was broadened to communicative competence which entails language~culture specific knowledge and skills for language use in interaction. However, interpersonal and intercultural skills have rarely been considered (section 2.2.2). In the social sciences, the study of competence is mainly motivated by a desire to improve communicative effectiveness. In structural linguistics, this study was generally motivated by an interest in its determining role in language acquisition. There was little interest in language performance in communication, which contained too much `noise', that is false starts, slips of the tongue, repetitions, etc. However, in second language acquisition (SLA) studies, researchers claimed that it is not only innate competence that determines language acquisition but that language use in communication is also important for acquisition to take place (section 2.3). Initially, NS (adjusted) input was considered a necessary and sufficient condition for language acquisition (section 2.3.1). Such adjustments would lead to comprehensible input which would support language acquisition (see section 2.3.2). When it became clear that NS input could not completely explain the learners' language acquisition, the role of learners's output and strategies were con- 2.0 lntroduction 11 sidered too (see sections 2.3.3 8z 2.3.4). It was concluded that in order to explain variation in second language acquisition (SLA), both NS and NNS communication behavior should be studied in interaction with each other. An interactional process which has been claimed to influence acquisition positively is negotiation of ineaning, which occurs when there are problems in communication. Problematic communication has been studied extensively in intercultural communication with the focus of increasing communicative effectiveness. Section 2.4 will deal with two important concepts in the study of problematic communication, i.e., understanding (section 2.4.1) and miscommunication (section 2.4.2). The relationship of negotiation of ineaning with communicative effectiveness as well as SLA is reviewed in section 2.5. Negotiation of ineaning may lead to the creation of comprehensible input and output, and thus enhance communicative effectiveness in intercultural communication. It may also make the learner pay attentíon to linguistic form, which may be an additional condition for language acquisition to take place. Whether negotiation of ineaning takes place and whether it is effective in NSNNS intercultural communication and language acquisition may depend on several factors. Limited NNS communicative competence may lead to asymmetry in NS and NNS negotiation of ineaning behavior, limiting the opportunities for language acquisition. In addition, institutional roles may lead to such asymmetry as well. The last part of this chapter is devoted to factors which may lead to asymmetry in interaction (section 2.6). The concepts of power, dominance and asymmetry are discussed in section 2.6.1. The influence of institutional roles on interactional asymmetry is discussed in section 2.6.2. In addition, the influence of NNS communicative competence in NS-NNS interaction is discussed in section 2.6.3. In section 2.7, the chapter is concluded with the implications of the state of the art on intercultural communication and SLAstudies for the focus and design of the current study. 2.1 Intercultural Communication In the social sciences, different concepts are used to describe the process of communication between speakers of different cultural and linguistic backgrounds. Intercultural, cross-cultural, interracial, interethnic and international communication are some of those used. These terms are often used as Knapp 8z Knapp-Pothoff (1987: 3) stated `in free variation'. In this study, intercultural communication will be used for the communication between people of different cultural and language backgrounds. The study of intercultural communication is a relatively recent interdisciplinary field with contributions from several disciplines. Harman 8t Briggs (1991) surveyed members of SIETAR (Society for Intercultural Education, Training and Research) on their opinions about these contributions. They viewed anthropology to 12 Chapter 2 Negotiation of Menning in Intercultural Communlcation be the most important contributing discipline, followed by sociology, linguistics, psychology, speech communication and political science, respectively. Whereas linguistics and speech communication are recognized as important disciplines, none of the SIETAR members outside of linguistics mentioned a linguist as an important author having had an impact on the field. Books written on intercultural communication often contain collections of papers dealing with a wide variety of issues. However, few comprehensive approaches have yet been developed, in particular, those which account for the specific role of language (proficiency) as part of intercultural communicative competence. Knapp 8z Knapp-Pothoff (1987) convincingly argue that the crucial difference between intracultural and intercultural communication is related to language. In intercultural communication at least one of the speakers does not speak the language of the interaction as her or his native language and this may have a great impact on communicative effectiveness. Whereas native speakers focus on communication per se, nonnative speakers also need to focus on the language used. Language (proficiency) should therefore be an important factor in studies on intercultural communication; however, it does not appear to be a prominent issue at all in the social science literature. If language is discussed it is generally done from a socio-psychological perspective. These studies deal among other things with attitudes towards language and use macro-level survey research methods. For example, in the collection edited by Asante 8t Gudykunst (1989), the first contribution by Gudykunst 8z Níshida (1989) on theoretical perspectives is about culture and communication but does not mention language (proficiency) as an important part of intercultural communication competence. There are only two papers out of twenty-three in the book which deal with linguistic issues. One is about translation (Gonzalez, 1989). The other (Gudykunst et al., 1989) concerns the social psychology of language. On the one hand, it discusses socio-psychological factors influencing group formation and perception: stereotypes, language attitudes, and communication accommodation. On the other hand, it deals with sociolinguistic issues of how people use language to communicate: group vitality, second language competence and code-switching. On each of these issues less than one page is written. Another recent volume on conceptual issues, theoretical and empirical research perspectives concerning intercultural communication competence (Wiseman 8c Koester, 1993) deals with issues such as uncertainty reduction, face negotiation and management. It contains only one contribution from a linguistic perspective. This is one on cultural pragmatics (Carbaugh, 1993). This paper describes whether and how Soviet citizens and Americans discuss personal matters in public speech situations. Part of the data appear to be notes taken through participant observation. The only excerpt of (videotaped) interaction included is from an American talk show broadcasted from the Soviet Union in which an American host interviews Soviet citizens. From the excerpt presented one gets the impression the Russians interviewed spoke native-like American-English. Although this may coincidentally have been the case, it is more ]ikely that it was an artefact created in transcription. In any case, it is doubtful that all Russians Carbaugh met spoke such 2.1 Intercultural Communication 13 native-like English. Nevertheless, Carbaugh does not deal with their language proficiency, although it may just as well have affected their talk about sensitive personal issues. In other words, his concept of cultural pragmatics is not linked to language competence issues. 2.2 Concepts of Competence In intercultural communication, communicatíon skills are viewed as one of three main dimensions which are important for functioning in a foreign culture (next to the abilities to deal with intercultura] stress and with interpersonal relationships) ~(Hammer, 1989). To be effective in intercultural communication one needs to l possess general communication knowledge and skills, and specific knowledge and ! skills in the language~culture of communication. In the social sciences, the term communication competence is generally used to refer to general knowledge and skills. In linguistics, the term communicative competence is used to refer to languagelculture specific knowledge and skills. The social science of interpersonallintercultural communication competence will be discussed in section 2.2.1 and the linguistic concept of communicative competence will be discussed in section 2.2.2 in order to create some clarity on what intercultural communicative competence may entail. 2.2.1 Interpersonal and Intercultural Communication Competence Social science researchers as well as linguists state that the concept of competence is ill-defined and not sufficiently tested empirically. One problem with the concept of competence is whether, along with the unobservable phenomenon of knowledge, skill is included, which is directly observable in communication. It is important to clarify what intercultural communicative competence is, while doing justice to social science as well as linguistic theory. Webster's Dictionary (1989) defines competence as: "possession of required skills, knowledge, qualification or capacity". Thus, the common sense use of the term competence includes knowledge as well as skill. However, Taylor (1988: 160) quotes definitions in psychology which are solely focused on skills, whereas in education, the common definition contains a combination of knowledge and skill. Probably skills are sometimes stressed more than knowledge because social research has generally been outcome-oriented towards improving communicative effectiveness. There is a considerable body of literature on intracultural communication competence and skills. In their book on interpersonal communication competence, Spitzberg 8z Cupach (1984) reviewed the literature from several disciplines on interpersonal communication competence. On the basis of this review they present a model for what they call relational competence consisting of five major compo- 14 Gtupter 2 Negotintion of Mea~znig in Interciiltival Comnuinication nents: motivation, knowledge, skills, context and outcomes. They take an interactive perspective when they stress that communicative competence should not be conceived as something which an individual possesses but that competent interaction is manifested in the behavior of both interactants. In other words, an individual's competence can only be measured through his performance in a certain context. This performance may vary from context to context and thus it cannot be objectively determined for the individual. In Hammer's review (1989) of communication competence, it appears that the concept of competence is generally equated with skill. He mentions several communication skills which appear to be central in interpersonal communication in general: immediacy, interaction management, social relaxation, expressiveness, and altercentrism. In reviewing the intercultural communication literature, he found that the same main communication skills are supposed to be effective in the intercultural context as we1L In addition, he mentions some other skills such as problem-solving skills, innovativeness and task-orientedness which appear to be specifically related to success in intercultural communication. Apparently in the intercultural context certain other skitls are necessary to compensate for the lack of a shared language and culture. Hammer refers to a few studies which mention foreignlsecond language skills as part of intercultural competence. These skills he classifies under the communication skill `expressiveness' in which the expression of respect is also included. Thus he does not distinguish general knowledge and skills from specific ones. In addition, he does not appreciate the fact that in intercultural communication (lack of) language specific skills may have a much greater impact on the effectiveness and appropriateness of the interaction than in intracultural communication where all speakers use their native language. The general lack of recognition that some knowledge and skills may be more or less general or universal while others are languagelculture specific is particularly problematic because the research is very much biased by a European, and in particular, an Anglo-Saxon perspective. In addition, it is remarkable that socio-cultural knowledge of norms and values is not mentioned as a crucial element in intercultural communication. Like language proficiency, lack of knowledge of the target culture may influence communication negatively. It can be concluded that the social science concept of intercultural communication competence is mainly geared towards general skills and knowledge. The roles of culture and language specific knowledge and skills are not sufficiently taken into account, although lack of these may seriously impede communication. In the next section linguistic concepts of communicative competence will be discussed with regard to these specific knowledge and skills. 2.2.2 Linguistic Concepts of Competence Formerly, linguistics was firmly rooted in the social sciences through the work of Boas, Sapir, Malinowski and others on language use in communication. However in the 1960's, the work of Chomsky strongly stimulated the theoretical interest of 2.2 Concepts of Competence 15 linguists in the structure of language. The goal was to describe grammar rules of the `ideal native speaker' which would cover every occurrence of linguistic behavior. Chomsky used the term `competence' solely to refer to the ideal speaker-hearer's knowledge of a language. In addition, he used the term `Qerformance' to refer to the actual use of language in communication, which wóuld include skills (Chomsky, 1965: 4). Performance could be ill-formed and was not seen as rulebased. Chomsky was primarily interested in competence, an underlying, not directly observable phenomenon, rather than in performance. Interest in language use returned when it was found that many syntactic rules could not be properly defined without referring to the context in which language was used and its function in communication. Hymes' criticism of Chomsky was that his concepts of competence and performance did not account for rule-based language use (1971, 1972). This was later acknowledged by Chomsky when he used the concept of pragmatic competence as knowledge of the conditions and manner of appropriate use (Chomsky, 1980: 59). Hymes developed a broader concept to include social dimensions next to linguistic competence. He called this concept communicative competence. By adding the social element of use to the concept, competence also became associated with skills in linguistics. The concept of communicative competence has been further developed for application in language teaching and testing, the area of applied linguistics. In foreign language teaching, the object which had to be taught was initially equated with Chomsky's linguistic competence. Teachers and researchers viewed problems in interlingual communication as problems originating from lack of competence in pronunciation, grammar or lexicon, and they underestimated pragmalinguistic áspects and sociocultural knowledge (Knapp 8c Knapp-Pothoff, 1987: 2). In foreign language teaching, socio-cultural knowledge mainly came about through readi~ the literature of the country or through what in German is specifically called Landeskunde. This type of teaching may have given many students of foreign languages sufficient competence to pass exams. However, in the intercultural context, with second language learners who need the second language for daily interaction, this type of teaching was found to be insufficient. For these learners new teaching methods were developed which were based on the acquisition of the broader concept of communicative competence. Learning the structure of a second language must coincide with learning how to use language in a new cultural context (see Gunthner, 1989). Canale 8z Swain (1980) developed a model of communicative competence for second language teaching and testing purposes. They distinguish three levels: 1. linguistic competence (knowledge of grammar) 2. sociolinguistic competence (knowledge of how to use language appropriately) 3. strategic competence (knowledge of how to deal with trouble and breakdown) (Swain, 1984: 189). The first two types competence are obviously language~culture specific, although it is not entirely clear whether and how socio-cultural knowledge of, for instance, norms and values is included in the concept of sociolinguistic competence. However, it is not clear whether strategic competence is languagelculture specific. 16 Chapter 2 Negotiation of Meaning in Intercultural Communication Strategic competence appears to be related to what Hammer (1989) called intercultural problem-solving skills. Canale and Swain (1980) distinguish communicative competence from performance. They acknowledge that both knowledge and skills are to be acquired in the second language, which together lead to second language proficiency. In particular, the relationship and differences between communicative competence, performance and proficiency is unclear. As competence is primarily defined as an unobservable phenomenon, this creates problems in empirical research and in language testing. Therefore, the concept proficiency is used instead, which is directly related to performance, that is, the demonstration of skill. However, some applied linguists equate competence with proficiency (Stern, 1983: 341). Taylor tries to clarify the conceptual ambiguities by introducing the concept of communicative proficiency which consists of both competence and proficiency, which he defines as "the ability to make use of competence" (Taylor, 1988: 166-167). This communicative proficiency can be determined on the basis of what a speaker dces, which is observable. However, in this definition the concept of competence is still intertwined with abilitylskill. Dirven 8z Putz (1994) in their contribution to the Proceedings of the LAUD (Linguistic Agency of the University of Duisburg) Symposium on Intercultural Communication, perceive a shift of focus in applied linguistics from linguistic competence to communicative competence to intercultural competence (1994: 1617). They define `intercultural communicative competence' as not only the nonnative speaker's competence to deal with his limited proficiency but also as the native speaker's ability and willingness to accommodate to them, which in German is called Fremdverstehen. This means a shift of focus to both NS and NNS behavior and the interaction between them, similar to Spitzberg á Cupach's (1984) definition of communicative competence. It can be concluded that the concept of competence is still not clearly defined. More research is needed on what intercultural communicative competence entails and how it can be measured. Clearly, with respect to the concept of competence an integration of social science and linguistic concepts is needed. Wiemann (1977) has tried to integrate the social science and linguistic concepts of competence in the following definition: "the ability of an interactant to choose among available communicative behaviors in order that he [she] may successfully accomplish his [her] own interpersonal goals ... while maintaining face and line of his [her] fellow interactants within the constraints of the situation" (1977: 198). The above definition of Wiemann's could be applicable to both native and nonnative speakers in intercultural communication. The communication behavior he refers to should include general as well as languagelculture specific knowledge and skills in the target language. Intercultural communicative competence is thus perceived as having two components: a general one and a specific one. The general component consists of interpersonal and intercultural knowledge and skills. The specific one consists of language and culture specific knowledge and skills. The NS and NNS both may have different shortcomings in their intercultural communicative competence. F'or the NNS, ihere is often a gap in the language and culture specific knowledge and 2.2 Concepts of Competence 17 skills in the target language and culture. He will also need to develop some intercultural skills to deal with the problems his limited specific competence creates. The NS, on the other hand, may play an important role in supporting the NNS wiih a limited communicative competence in the task of communication. He can use intercultural problem-solving skills such as clarifying problems or checking meaning. He may use interpersonal skills such as altercentrism which may lead to allotment of extra time that may be needed in interaction with limited second language speakers. Using language specific skills, he may adjust his language to the NNS's level of understanding (Hatch, 1983: 183-184). Therefore, the success of communication depends as much on the cooperation of the NS as on the communicative competence of the NNS. In addition, the degree of cooperation of the NS influences the NNS's opportunities to acquire the second language. For practical purposes, in this study communicative competence will be used to refer to knowledge and skills concerning a specific language and culture. Language proficiency will be used when the focus is on observable levels of performance. Linguistic competence will be defined as knowledge of syntax, morphology, lexicon, etc. 2.3 Interaction and Second Language Acquisition In the field of second language acquisition (SLA) a great deal of work has been done which pertains to intercultural communication, or as it is generally called in this field, native speaker-nonnative speaker (NS-NNS) interaction. As could be seen from section 2.2.2 a theoretical and applied impulse led to interest in verbal interaction and NSINNS roles in SLA. As a reaction to Chomsky's theory that innate competence determines language acquisition, studies were initiated to show the intluence of such learner-external factors as NS input in interaction on SLA (Hatch et al. 1978). SLA research on foreigner discourse will be discussed in section 2.3.1, and NS comprehensible input in section 2.3.2. However, NS input appeared not to be the only factor influencing SLA. A reaction to the input studies was brought about by Swain's work (1985) on comprehensible output which focussed on the role of the NNS in interaction and the effect of NNS output on SLA. The research on NNS output will be discussed in section 2.3.3. Other research on NNS output which is also of interest concerns NNS communication strategies which may or may not support language acquisition. This research will be discussed in secuon 2.3.4. 2.3.1 Foreigner Discourse First language acquisition research shows that the input of caretaker speech plays an important role in language acquisition. It does not contain as many errors as was 18 Chapter 2 Negotiation of Meaning in Intercultural Conununication hypothesized by Chomsky in his concept of performance: it is a simplified but well-formed register (see Snow, 1976). Because of the initial concern with ill-formed input, early studies on NS adjustments in NS-NNS interaction drew attention towards simplified speech which contained ungrammatical adjustments such as in the frequently quoted example "me Tarzan, you Janè'. Ferguson (1975) coined the term 'foreigner talk' for this register of ungrammatical simplified speech as a parallel to baby talk. He claims there are three processes of NS adjustment: deletion, substitution and addition. These adjustments lead to shorter utterances, more overt marking of grammatical or semantic relations, consistent use of the canonical word order, frequent use of questions, paraphrases, slower pace, etc. (Larsen-Freeman 8t Long, 1991: 119). Hatch (1983a) suggests that foreigner talk can have the same functions as caretaker speech or motherese: 1. promoting communication 2. affective bonding 3. implicit language teaching. Some earlier studies confirmed the fear of ill-formed input and showed that ungrammatical language was indeed used to foreigners (Clyne, 1977, 1978; Snow, Van Eeden 8z Muysken, 1981). However, Long (1983b) shows that there are two other types of adjustments that may occur which do not have to lead to ungrammatical input. Firstly, there are grammatical linguistic adjustments (e.g. simplification, paraphrasing). Secondly, there are adjustments which concern the structure of the interaction, also called conversational adjustments. They result in an interruption of the conversation which leads to changes in the structure of the interaction. Extra turns are added, such as repetition of the speaker or interlocutor's former utterance, more comprehension and confirmation checks, and more expansion. These conversational adjustments will be discussed extensively in Chapters 8 and 9. In his review of foreigner talk studies, Ellis (1985) distinguishes three accumulative types of foreigner talk: 1. with conversational adjustments 2. with conversational and grammatical linguistic adjustments 3. with conversational, grammatical and ungrammatical linguistic adjustments, resulting in a simplified register that resembles caretaker speech, pidgins and learner interlanguages. Long (1983b) claims that NSs use conversational adjustments most frequently and ungrammatical adjustments least frequently. Although ungrammatical adjustments do occur in NS-NNS conversations, research appears to indicate that in most cases the input is grammatically well-formed (Bingham Wesche, 1994: 221-23 I; Long 8c Larsen-Freeman, 1991: 117-128; Ellis, 1985: 133-134). Ungrammatical adjustments are most likely to occur when (i) the NNS has a very low second language proficiency, (ii) when the NS has perceived higher status, (iii) when the NS has prior foreigner talk experience, (iv) when the conversation occurs spontaneously (Long, 1983b: 179). It is not clear whether the NS adjustments are accumulative as Ellis proposes. It may be that linguistic adjustments can also occur without conversational adjust- 2.3 Interaction and Secoruf Language Acyuisition 19 ments. It may also be that linguistic and conversational adjustments are not always easy to distinguish. Long (1983b) classifies expansion as a conversational adjustment because the use of it resu(ts in an increase in the number of turns; however, it could just as easily be called a linguistic adjustment because usually such expansion also leads to the use ~f new linguistic material. More research is needed to determine whether the distinction between linguistic and conversational adjustments can be applied consistently and when different types of NS adjustments occur. 2.3.2 Comprehensible Input Research on NS adjustments per se was followed up by research which focussed on the effects of such input on comprehension and language acquisition. Krashen posited in the Input Hypothesis that a necessary and even sufficient condition for language acquisition is that input should be comprehensible (Krashen, 1981, 1982). He used Vygotsky's idea of `zone of proximal development' to define this type of input: it should be at the i t 1 level, just one step beyond the NNS's current level of protíciency. This idea was generally supported (see, for instance, LarsenFreeman 8z Long, 1991: 142), yet it is hard to prove a direct causal relationship. Long (1983b, 1985b) gives an overview of studies which deal with the issue. These studies try to prove that NS adjustments lead to comprehensible input which in turn leads to NNS language acquisition. That is: A(adjustments) ~ B(comprehensible input) ~ C(language acquisition) (Long, 1983b). Then Long discusses studies which either tried to prove A~ C, A-~ B or B--~ C. There are few studies which show A-~ C. However, he finds several studies which proved A-~ B or B-~ C which gives indirect proof of A~ C. He concludes that the Input hypothesis is plausible but needs to be further investigated. Some studies showed that comprehensible input resulting from conversational adjustments more consistently improves comprehension than linguistic adjustments. Pica (1987) found that conversational adjustments such as repetitions lead to improved comprehension. Chaudron ~4c Parker's review (1987) quotes research which showed elaboration improved comprehension which may in turn lead to acquisition. There is research which shows that linguistic adjustments also lead to better comprehension but the results are less consistent than those for conversational adjustments (Larsen-Freeman 8c Long, 1991). Long (1983c) and White (1987) even argue that grammatical input if simplified linguistically may lead to impoverished input and thus may not enhance acquisition. Linguistic adjustments which result in ungrammatical input may or may not improve comprehension, but it probably is not conducive for the acquisition of native-like competence in the second language. However, little research has been done on this issue. In addition, research has not been able to find a dírect relationship between comprehension and internalization of second language forms by the learner. Pica (1994: 507) therefore concludes that "the role of comprehension in SLA has become increasingly controversial". It has been shown that in negotiation of ineaning more NS input 20 Chapter 2 Negotiation oJMeaning iii l~ueretdtura! Communicatíon modifications occur than in the rest of the NS-NNS interaction (Holliday, 1993; Long, 1980 in Pica, 1994). Thus negotiation of ineaning may play a crucial role in stimulating NS adjustments. Pica claims that negotiation of ineaning may lead to manipulation of form which may be noticed by the NNS because it interrupts the conversation. NNS attention is hypothesiaed to be important for acquisition. However, NNS output may be better evidence of acquisition than comprehension, because using second language forms reyuires more competence than comprehending them. Therefore, comprehensible input may not be the only factor enhancing acquisition. Swain (1985) claims that language acquisition also depends on NNS opportunities to produce `comprehensible output'. In the next section, the role of NNS output will be discussed. 2.3.3 Comprehensible Output Swain (1985) tested the language proficiency of English speaking immersion students in Canada who acquired French through instruction of school subjects in this language. She found that, although these students had had seven years of comprehensible input, their output did not become nativelike as far as their linguistic competence was concerned. This might be explained by the fact that these immersion students only used French in the classroom where, generally, the teacher talks and the students listen. Thus they had little opportunity to negotiate meaning. Swain states that a function of negotiation of ineaning may be that the learner is pushed toward the delivery of a message that is conveyed precisely, coherently, and appropriately. Therefore, she coins this type of output `comprehensible output' as a parallel concept to that of comprehensible input (Swain, 1985: 249). Earlier studies on interaction between NNSs in the classroom have shown that learners can profit from interaction with each other within certain conditions, such as working in small groups (Long 8c Porter, 1985; Pica 8i Doughty, 1985), and in particular, cooperative learning environments (Bejarano, 1987; Deen, 1991). It is probable that in such environments much more negotiation of ineaning takes place than in teacher-centered interaction. Shehadeh (1991) found more extended negotiation of ineaning sequences in NNS-NNS than in NS-NNS interaction. Pica et al. (1989) carried out a study to test the effect of negotiation of ineaning on NNS comprehensible output. They found that in some cases the NNSs appeared to be `pushed' through the NS signals of trouble. This was specifically the case when the NSs used clarification requests which required an open answer. It was much less the case when the NSs used confirmation checks which could be answered with one-word utterances like `yes'. It was this lasi signal type, however, which the NSs used most frequently. Therefore, although negotiation of ineaning may lead to the adjustment of the interactional structure, the effect on opportunities for NNS comprehensible output is not yet clear. We need to know when NSs use clarification requests rather than confirmation checks. 2..3 Interaction and Second L.anguage Acyuisitiat 21 Shehadeh (1991) investigated in which situations NNSs modify their output. He concludes NNSs produce more comprehensible output after self-initiated clarification attempts than after NS clarification requests. Therefore, even if NSs use clarification requests, this still may not be the most conducive situation for the production of comprehensible output. Swain (1985) mentions some additional functions of output in SLA. A NNS can test hypotheses and may be forced to go beyond semantic processing towards syntactic processing. For merely understanding input the latter is not always necessary. In addition, the NS may provide models for correct and appropriate language geared to the NNS's intended meaning which the NNS may adopt afterwards. However, Pica et aL (1989) did not give evidence of this occurring. In conclusion, the NNS may not only need comprehensible input but also opportunities to produce comprehensible output for language acquisition. It may be in negotiating meaning, in particular, that the NNS is pushed towards comprehen-sible output. However, the relationship between cornprehensible output and negotiation of ineaning is not clear. In addition, whether negotiation of ineaning leads to acquisition may not only depend on the amount of comprehensible input and comprehensible output but also on the NNS's motivation to focus on acquisition in interaction. Some NNS strategies show evidence of such attention to form and thus promote language acquisition whereas others do not. The NNS may choose to use strategies which promote communication only or those which create opportunities for language acquisition. The next section thus concerns NNS strategies for communication. 2.3.4 NNS Communication Strategies There is a substantial body of research on NNS communication strategies. Communication strategies are used by the NNS to deal with the demands of interaction when his command of the language is limited. NNS strategies may be contrasted with NS adjustment strategies. NNS strategies aim ai compensating for his own lack of communicative competence, whereas NS strategies (adjustments) aim at compensating for NNS gaps in competence. The most influential studies on NNS strategies have been described in Bialystok's book (1990) on communication strategies. She discusses the taxonomies proposed by Tarone, Faerch 8c Kasper, Paribakht and Poulisse et al. as well as her own work in this area. It is remarkable that this body of research is relatively separate from the second language interaction studies described above, although they both share the focus of NNSs dealing with problematic communication. There appears to be less overlap than expected in the strategies NNSs use in this research compared to language behavior described in second language studies. Several characteristics of communication strategy research may have led to this lack of overlap. One characteristic of the communicatíon strategy research is that the NNS participants are mainly foreign language learners who are learning a typologically related language, whereas SLA studies often deal with second language learners with limited proficiency who come from much more divergent language and cultu- 22 Chapter 2 Negotiarion oJMecnting in Intercultura! Comntunicnlion ral backgrounds. For instance, in Faerch 8c Kasper (1983a, 1983b), Danish high school students use English. in Poulisse (1989) Dutch NNSs of English are studied, and in Bialystok 8c Frohlich (1980). French NNSs of English. This language-relatedness may influence choices in strategies. It is not surprising that strategies such as lexical borrowings and word coinage are found to be so prominent because these languages have so many cognates that such strategies have a good chance of being effective. This would not be the case with languages from entirely different language families. In addition, the language learners in communication strategy studies generally possessed a high level of education (high school or more) and had experience in language learning. These factors may also have intluenced the type and quantity of strategies found. Another characteristic of communication strategy research is that most studies are experimental and aim at uncovering the NNS's cognitive processes. This psycholinguistic perspective has led to the use of data collection techniques in which NNSs had to solve certain artificial tasks, usually without interaction with, or cooperation from, a NS. The tasks were generally oriented towards solving lexical production problems. Many strategies found to be used result in linguistic adjustments related to the lexicon, such as borrowings, word coinage, paraphrase and circumlocution which could be realized within one utterance or sometimes with only one word. It is remarkable that conversational adjustments are generally not reported, although these adjustments are more common in NS-NNS interaction than linguistic ones (see section 2.3.1). Only Tarone's taxonomy includes interactional strategies which may occur in negotiation of ineaning. In addition, Poulisse (1989), in her study on compensatory strategies, included one task (interview) out of four in which a NS participated actively. Poulisse concludes that this participation makes the task easier because the NS would provide possible models for the NNS problems (confirmation checks). She also concludes that the immediate NS feedback often leads to further negotiation if the compensatory strategy used did not clarify the problem. Then it would result in a side-sequence. Communication strategy research has been modest in its goals of describing NNS behavior. It can be stated that this research, because of its learner-internal focus, has neglected to analyze the interactional process which may support communication and acquisition. It has tocused on a priori strategies as a result of speech planning and self-monitoring. The main aím was to uncover the NNS's potential in dealing with communication problems, and not to investigate the collaborative process of achieving understanding. However, the effectiveness of NNS communication strategies can only be determined by the interlocutor's acceptance of NNS contributions (see also section 2.4.1). Thus it does not suffice to look at NNS strategies in isolation. Yule 8c Tarone (1991) argue for an interactional perspective in communication strategy research. Tarone's definition of a communication strategy in her earlier work already showed this perspective: "mutual attempts of two interlocutors to agree on a meaning in situations where the requisite meaning structures do not seem to be shared" (Tarone, 1980: 420). However, Yule 8t Tarone (1991) also criticize the `input analysts' for ignoring the communication strategy research. 2.3 liileraction nnd Secnnd La~iKuage Acq~eisi~ioii 23 They refer to work by Pica (1987) in which only the NS contribution in a clarification seyuence is analyzed while the NNS is actually doing most of the clarification work using communication strategies. In the research on communication strategies a point of discussion has also been whether these strategies are unique to language learners or appear in NS-NS interaction too. Bialystok (1990) and Poulisse (1989) both show that the latter is the case. Communication strategies also occur in certain specific situations with NSs such as a professional explaining to a layperson or when a NS has trouble retrieving a word. In addition, these strategies appear in child language. Therefore, communication strategies might be used in all situations where there is a gap in communicative competence between the interlocutors. NNSs just have to use these strategies more often because they lack more linguistic competence in the target language. However, this is no reason to assume that the cognitive processes in dealing with trouble in NS-NNS interaction differ for both types of speakers. If we look at the types of processes described by NNS strategy researchers and NS input researchers similar descriptions occur. Bialystok (1990: 34) calls the NNS meaning and form adjustments: reduction, substitution or expansion. With NNS production one cannot speak of deletion of forms from the NNS's point of view because a NNS does not possess certain linguistic means; there may, however, be meaning reduction (compared to intended meaning). In addition, deletion may occur from an interactional perspective when a NNS repeats part of a former utterance. Similar process descriptions are mentioned by Ferguson (1975: 4) for NSs' foreigner talk: omission, replacement and expansion. This points up a resemblance between NNS and NS strategies in dealing with lack of shared language and culture in problematic communication. Researchers on NS input as well as on NNS output acknowledge that NS and NNS strategies and language use in building common ground in interaction resemble each other. Therefore, Yule 8z Tarone (1991) argue for an analysis of both NS and NNS moves in negotiation of ineaning by using one analytic framework. They claim researchers from both strands might benefit from each other's work in two ways: "(1) by making use of an analytic framework which encourages the analyst to look at both sides of the conversational exchange, and enables the analyst to better identify key moves made by both participants in the negotiation and resolution of communicatíon problems; and (2) by returning to the 'more humble approach' of describing both input and learner performance in interaction, and refraining from making claims about acquisition which are based upon untested assumptions" (1991: 169-170). They furthetmore add that such an analytic framework applied in longitudinal studies may give insight into the question whether negotiation of ineaning indeed leads to increased SLA. In the current study, both NS and NNS behavior in negotiation of ineaning will be described with one analytic framework from such a longitudinal perspective. In particular, yualitative studies which clarify the relationship between input, interaction and output are needed (Larsen-Freeman, 1985). Therefore, NS and NNS input and output will be analyzed, and it will be investigated how input and output are organized in negotiation of ineaning. 24 Chup~er 2 Negoliatio~i ~~Mearting i~i h~terculuirul Commtutlcalion 2.4 Problematic Communication This section deals with the context in which negotiation of ineaning occurs, that is, problematic communication. Within intercultural communication, a great deal of research has been done on miscommunication. This is not surprising because people who do not share the same linguistic and cultural background often have problems interpreting each other's messages. Scollon 8r. Scollon (1995) base their approach to improving professiona] (intercultural) communication on the assumption that misunderstandings are more likely to occur than not in such circumstances. In particular, if the NNS's communicative competence in the second language is limited, there is a greater chance of trouble in unders[anding and communication breakdown. First the concept of understanding will be explored brietly, because this is the basic goal of communication which is disturbed when miscommunication occurs. An interactionist perspecdve on understanding will be taken. Then miscommunication will be defined as a type of misunderstanding recognized by at least one of the speakers which may be followed by a negotiation of ineaning sequence. 2.4.1 Understanding In interaction, a basic goal of the participants is to understand and to be understood. Understanding may be defined as the attribution of the same meaning by the hearer as the one intended by the speaker (Allwood, 1986). Language is a crucial tool in achieving understanding in human communication. Taylor (1986) investigated the his[ory of the study of understanding. In the 17th century, Locke defined understanding as a mental event: the hearer inferences from an utterance the idea the speaker intends to convey (Taylor, 1986: 171). With this concept of `telementation' Locke expressed a concern with the imperfection of words, that is, because understanding is a mental event the speaker can never know whether the hearer has understood the intended meaning. Thus language use is by definition ambiguous. De Saussure (in Taylor, 1986), however, claimed that mutual understanding is possible because of the conventionalized relationship between the signified and the signifiec In discourse studies the concept of common ground refers to such conventions. It includes not only the speaker's knowledge of the relationship between language and the world but also other information such as his prior beliefs and assumptions concerning the world which are often shared by speakers of the same speech community. Speakers presuppose there is some common ground and try to add to it through interaction. In this view, understanding is achieved when an utterance presented by speaker A is accepted by speaker B. Clark 8c Schaeffer (1989: 267) give five types of evidence for such acceptance. They range from the weakest form showing continued attention which may not indicate understanding at all, to the strongest `display', which includes repetitions of all or part of A's 2.4 Proble~natic Commt~nicntion 25 presentation. If this evidence is accepted by A. this means A believes that B understood him well enough for current purposes. In intercultural communication. there is often less shared linguistic and cultural common ground than in intracultural communication. As Liberman (1990) formulated it, the `horizon' is more open than in communication between speakers of the same linguistic and cultural group. This influences understanding. If the speakers are unfamiliar with each other, they do not know how much common ground they can presuppose because they do not know how much linguistic and socio-cultural knowledge they share. Thus there is ample chance of miscommunication when wrong assumptions about shared common ground are made. According to Scollon 8c Scollon (1995: 52-53), understanding is achieved through coherence in interaction. They mention four sources which influence coherence in oral discourse. The first source is lexical and grammatical cohesive devices. This level may cause many problems, particularly with less proficient nonnative speakers. The second are schemata, which organize utterances in `adjacency pairs' which in turn regulate expectation patterns. Schemata and scripts for e.g. institutional conversations may differ between cultures or even be nonexistent. The third are prosodic patterns, such as intonation and timing. Nonnative speakers are often still influenced by their tïrst language patterns even if they speak the second language fluently. Work by Gumperz (1982, 1990, 1991) and others has shown the major impact, in particular on the affective level, problems with the latter source of coherence can have. The fourth source is metacommunication through conversational inference (Scollon 8r Scollon, 1995: 68), which deals with how messages in a certain context should be interpreted. This is probably a type of coherence with which even advanced speakers of a second language may continue to have trouble. The above discussion on coherence suggests that understanding is a state which can be achieved by creating coherence in the speaker's talk. However, some researchers stress that understanding should be seen as a collaborative process. Bremer et al. (1993) prefer to view understanding as an active process achieved in negotiation between the speakers. The participants collaborate in achieving mutual understanding. This view agrees with that of Clark 8t Schaeffer (1989). Understanding is expressed in terms of mutual acceptance of continuation of the interaction rather than in cognitive processes of speech processing. In the current study, understanding will be viewed as a collaborative process in which speakers work on establishing common ground which can be realized through the negotiation of meaning when trouble occurs with either of the four sources of coherence mentioned. 2.4.2 Miscommunication Miscommunication occurs in both intracultural and intercultural encounters. It may have its cause in all aspects of the communication process. Researchers use several descriptions of the communication process to distinguish sources of 26 Chapter 2 Negotiatio~t of Meaning in Intercultara! Cnfn~nunication miscommunication. Coupland et al. (1991: 13) distinguish six levels of analysis: L"discourse and meaning transfer are inherently flawed 2. minor misunderstandings or misreadíngs 3. presumed personal deficiencies 4. goal-referenced, identity and instrumentality 5. grouplcultural differences in linguisticlcommunication norms 6. ideological framings of talk". The first two levels are directly linked to occurrences in language use. Levels three, four and five refer to speaker characteristics such as (perceived) identity and goals. The deepest level, the sixth, refers, for example, to `political' gender studies in which certain interactions are defined as `miscommunication' because they disadvantage women and advantage men. Ensink (]993) critiques work on miscommunication such as that of Coupland et aL (1991). He states that they do not use an adequate model of verbal interaction to base their theory on. Ensink's communication model distinguishes eleven aspects of the communication process which may cause trouble. These range from the sender's intention, the encoding of the message, the medium of communication, and the context, to the interpretation of the receiver. He shows that all these aspects of communication are vulnerable to miscommunicaUon. Research not only differs as to the levels of analysis or the aspects of communication which are considered vulnerable to miscomrtiunication, but also as to the definition of the concept miscommunication itself. One main difference concerns whether participants realize that there is misunders[anding. Banks, Ge 8z Baker (1991) define miscommunication as "a particular kind of misunderstanding, one that is unintended yet is recognized as a problem by one or more of the persons involved" (1991: ]06). This type of miscommunication surfaces when one speaker indicates trouble in understanding, clarifies trouble or checks intended meaning. This leads to a side-sequence in which meaning is negotiated. Misunderstandings which do not surface and thus cannot directly be resolved nor studied easily are not included in this definition. The current study deals with negotiation of ineaning which occurs when there is surfaced miscommunication as defined above by Banks, Ge 8z Baker (1991). 2.5 Negotiation of Meaning in Clarification Sequences In this section, two approaches for the micro-analysis of face-to-face interaction will be discussed: discourse analysis and conversation analysis. In particular, the concepts repair, negotiation of ineaning, and clarification seguence, which relate to problematic communication, will be discussed extensively because they are basic to the curren[ study. It is argued that research on intercultural communicatiort~NS-NNS interaction could profit from the methodology that conversation analysts in particular have developed. 2.5 Negotiation of Meaning in Clarification Sequences 27 2.5.1 The Micro-Analysis of Interaction Face-to-face interaction is the most important context of language use. Through intensive, recursive micro-analysis (of transcripts) of audio or videotapes, this ephemeral type of interaction, which normally tends to go unnoticed, can be investigated. There are two types of interaction analysis, which differ in goals and methodologies, but which are both pertinent to the current study. These are discourse analysis and conversation analysis. In contrast to grammarians. a discourse analyst looks for regularities in interaction rather than categorical rules which work 100qo of the time and which distinguish grammatical from ungrammatical utterances. The discourse analyst uses the methodology of descriptive linguistics by describing linguistic forms, but relative to the environment in which they occur. The data used are typically elicited in some way or constructed by the researchers to gain comparable data, for example, on a certain speech act. However, naturally occurring interaction is also sometimes used. Discourse analysis differs from traditional descriptive linguistics in that linguistic forms (sentences or texts) are not seen as products, but as dynamic means for expressing intended meaning (Brown 8r. Yule. 1983). The focus is on the communication process, or as Brown 8e Yule state ir. "We shall be particularly interested in discussing how a recipient might come to comprehend the producer's intended message on a particular occasion, and how the requirements of the particular recipient(s), in definable circumstances, influence the organisation of the producer's discourse" (1983: 24). The focus of interest is not language as an object but communication as a process. Another approach for the micro-analysis of interaction is conversation analysis, which originates from ethnomethodology in sociology. Conversation analysis, as developed by Sacks, Schegloff, and Jefferson (1973. 1974, 1977) and practised by many others uses a methodology which is quite different from discourse analysis. Transcripts and audiotapes (sometimes also videotapes) of naturally occurring interaction are used as data to ensure a maximum of authenticity and a minimum of disturbing factors influencing the nature of conversation. Conversation analysis has revealed many aspects of the structure of conversation, such as the turn-taking system and preference organization. According to Levinson (1983: 364) conversation analysis also contributes to the understanding of utterance meaning and to the study of linguistic form, because the ]atter are often dependent on the conversatíonal organization. Conversation analysis has been developed for the analysis of native-speaker interaction by researchers from the same speech community. It has been applied to interaction in several speech communities. In addition to the work on North-American interaction from Sacks, Schegloff, Jefferson and others, work has been done on languages such as Swedish (Lindstrám, 1994), Dutch (Mazeland, 1992; Houtkoop, 1987), and Italian (Mueller, 1991). Conversation analysis has also been applied by nonnative researchers to speech communities other than IndoEuropean ones, such as Thai (Bilmes, 1986; Moerman, 1988). In addition, more recently it has been used for the analysis of NS-NNS data in German by native 28 Chap~er 2 Negotiatian of Meai~ing i~i Iniercultural Cornmuiticntio~i speaker researchers (Rost, 1989; Dietrich. 1982), and in Swedish by a researcher who is a member of the nonnative Finnish speech community (Kalin, 1995). These studies show that conversation analysis can be applied to interaction with nonnative speakers as we1L However, the interpretation of behavior is undoubtedly less easy and more liable to faults when the researcher is not from the same speech community as the interactants or when speech is analyzed of nonnative speakers who are not yet full members of a speech community. In the next sections, some major concepts referring to trouble in interaction developed by conversation analysts will be discussed. 2.5.2 Repair Conversation analysts have done research on the resolution of trouble in~interaction. Schegloff, Jefferson 8c Sacks (1977: 363) called this problem solving activity `repair'. Repair can be defined as a speech activity during which speakers locate and replace a prior information unit where there may have been problems of speaking, hearing, or understanding (Stalpers, 1993: 60). Four types of repair can be distinguished, depending on who initiated the repair and who consecutively did the actual repair, the person who caused the problem (the self) or the other: 1. self-initiated self-repair 2. other-initíated self-repair 3. self-initiated other-repair 4. other-initiated other-repair (Schegloff, Jefferson 8c Sacks, 1977). Schegloff, Jefferson 8r Sacks state that there is a strong preference for self-initiated self-repair in conversations between native speakers. Self-initiated self-repair generally occurs within one and the same turn. Other-initiated self-repairs occur less frequently. Other-initiated other-repair are claimed to be the least preferred type of repair. Such `other-correction' may occur most frequently in interaction with learners of a target language, e.g. children or nonnative speakers (Schegloff, Jefferson 8z Sacks, 1977: 381). Nofsinger (1991) states that in NS-NNS interaction almost no NNS self-initiated self-repair has been found. Faerch 8r Kasper (1982) found that NSs do use self-initiated self-repair in interaction with NNSs, but mainly with NNSs with limited communicative competence. This indicates that NSs are sensitive to NNS limitations of linguistic knowledge and processing capacity and, therefore, adapi their input to the NNS level (see sections 2.3.1 8c 2.3.2). However, Kasper (1985) also found that in NS-NNS interaction other-initiated self-repair and other-repair occur more often than in NS-NS interaction. Difference in linguistic abilities and socio-cultural knowledge of the NS and NNS may lead to such higher frequencies of other-initiated repairs. Both speakers may have trouble in understanding the other because they are not able to estimate the common ground, and so they may request a repair. The NNS may initiate a repair because the NS's utterance could not be processed. In addition, the NS may initiate a repair because the NNS did not use a nativelike construction which could be conventionally interpreted, and the 2.5 Negotiation of Meaning in C[arification Seguences 29 NNS did not attempt a self-repair due to limited linguistic competence. Wíth the help of the NS, the NNS might be able to repair the utterance in such a way that it becomes interpretable. Other-initiations and other-repairs lead to an interruptíon of the main line of the conversation. A side-sequence is started to negotiate meaning. Such clarification sequences for the negotiation of ineaning may thus be typical for NS-NNS interaction (see Stalpers, 1993). The collaborative process of negotiating meaning and its conversational structure in clarification sequences wilÍ be discussed in the following sections. 2.5.3 Negotiation of Meaning Descriptions of the interaction processes occurring when there is some trouble in understanding have been named with several terms, such as alignment action (Nofsinger, 1991) and metalinguistic phase (Glahn. 1985). In SLA literature the term negotiation of (for) meaning is generally used, and it was introduced in this field by Schwartz (1980). Negotiation of ineaning in SLA is defined by Pica et al. (1989: 65) as "exchanges between learners and their interlocutors as they attempt to resolve communication breakdowns and to work to mutual comprehension". Pica (1987) claims that initiative of the NNS in negotiation of ineaning is important for language acquisition. Based on second language acquisition theory she states the belief that "learners can advance their receptive and expressive capacities in a second language if they obtain their interlocutor's assistance in understanding linguistic material not currently within their L2 reper[oire" (Pica, 1987: 5). Van den Branden (1995) found some empirical results which show that negotiation of ineaning has a positive effect on comprehension in classroom interaction with NNSs and NSs of Dutch. Negotiation of ineaning may serve more than one goal at a time. While creating communicative effectiveness may be the foremost goal, language acyuisition may be a second one. Long (1983c) and Varonis 8z Gass (1985a) have suggested that it is the process of negotiating meaning rather than comprehensible input per se which stimulates second language acquisition. Ellis (1985) mentions as the most plausible explanation that the NS decides on the basis of NNS feedback in negotiating meaning how to adjust the input to a comprehensible level. Pica (1994) claims that research on negotiation of ineaning indicates that the creation of comprehensible input and the opportunities for comprehensible output do not automatically positively affect language acquisition. Based on other research studies Píca suggests as a third learner condition NNS attention to second language form, which may be activated by the negotiation process. However, Aston (1986) criticizes these claims. He states that a third goal of negotiation of ineaning may be avoiding communication breakdown, and maintaining and enhancing support (Aston, 1986). When negotiation of ineaning is used for these reasons, it may not lead to better understanding, and may not have any positive effect on language acquisition. Research by Ehrlich, Avery 8c Yorio (1989) seems to support this view. They showed that a higher frequency of negotiation of meaning procedures does not necessarily lead to more comprehensible input. 30 Chapter 2 Negotiation ojMeanii~g iii Inrercidtura! Communication Thus negotiation of ineanir.g may create learner conditions which support language acquisition, but this depends on the goals of the NS and NNS in the negotiation. As Van den Branden 11992) states there are high expectations about the effectiveness of negotiation of ineaning for language acquisition, but many claims have not yet been empirically tested. 2.5.4 Claritication Sequences Conversation and discourse analysts have shown that conversation is organized in units at several levels. On a local level, conversation is organized in turns and turn constructional units (Sacks, Schegloff 8r Jefferson, 1974; see also Chapter 3). Alongside turn organization on the ]ocal level, some organization around functional units also appears to operate. In one turn, a speaker can make more than one move (Levinson, 1983: 289); for instance, he may indicate a problem and suggest a possible repair. One type of unit on the discourse level are sequences. Such sequences consist of turn pairs. Such turn pairs are called adjacency pairs, e.g. an answer following a question or a greeting following another greeting (Schegloff 1972; Schegloff 8z Sacks, 1973). Often there is one type of second part which is preferred over others (such as a yes answer to a yeslno question). Some pairs are followed by a third part such as an acknowledgement of the reception of the preferred second part (Houtkoop, 1987). As stated above, other-repair typically results in a collaborative process of negotiation of ineaning which occurs in side-sequences interrupting the topic progression in the ongoing conversation (Levinson, 1983: 304, 340; Jefferson 1972: 294). Such repair sequences have also been called completion sequences (Stalpers, 1993: 61) or non-understanding routines (Gass 8c Varonis, 1985a). In the current study they will be called clarification sequences, because the goal of these sequences is clarification through negotiation of ineaning. Gass 8c Varonis (1985a: 151) describe a four-move model for clarification sequences in interaction between nonnative speakers and exemplify it with the following example: NNSI NNS2 My father now is retire retire? NNSI NNS2 yes oh yeah 7rik~;er lnrl~~utur Respunse Reucrion to u response Apart from the fact that the example sequence ts not very illuminative because it is not clear whether the problem is really clarified, the model is very global. It is not possible to distinguish different types of moves which may follow different types of indicators and may show either weak or strong evidence of acceptance (see Clark 8c Schaeffer, 1989). Please, compare the above sequence with another imaginable one which would receive the same classification of moves while there is much stronger evidence of understanding: 2.5 Negotintion of Menntng tn Chrificulinn Seyuences 31 NNS I NNS2 NNS 1 NNS2 NNS1 NNS2 My father now is retire what? my father no work anymorc oh he is retired yes o yeah Ti,,~Rer Inrircurnr Respunse Reuction tu u respnnse Re.cponse Renctinn tn u re.cpon.ce The indicator in the original example reyuires a different response than the one in the second example and thus could be considered of a different type. In the second example, with the first response after the indicator NNS l clarifies the trouble source and in the first reaction to a response, NNS2 checks his understanding by rephrasing this trouble source. These two moves differ considerably from the second response and reaction to a response, which are both minimal feedback signals. They respectively confirm the correctness of NNS2's rephrasing of the trouble source and acknowledge that the trouble is sufficiently resolved. The first two moves require the NNSs to manipulate the content and form of the original trouble source and thus may be considered `comprehensible output'. The second two moves have the form of minimal feedback signals and thus require little linguistic competence. Therefore, the categories indicator, response and reaction to a response appear to be too general to differentiate adequately between negotiation moves which manipulate contentlform and feedback signals which give or elicit feedback (Allwood, 1993: 197). Allwood states that linguistic feedback may play an important role in language acquisition in interaction. Of the feedback categories he found, many could be used for the negotiation of ineaning (1993: 211). Furthermore it is not clear whether the moves of Gass 8c Varonis' model for interaction between nonnative speakers will be sufficient to describe interaction between native and nonnative speakers. The following conversational modifications are found to occur when there is trouble in NS-NNS interaction (see Pica 8z Doughty, 1988 and Long, 1983a): 1. seeking clarification 2. seeking confirmation 3. verifying NNS comprehension 4. repeatíng, paraphrasing and expanding. Seeking clarification is comparable to what in conversation analysis is called repair initiation and seeking confirmation to other-correction (Schegloff, Jefferson 8t Sacks, 1977). More recent research from Pica et al. (1989) has shown that these modifications aze also used by NNSs. Gass 8c Varonis (1985a) do not describe the NNS-NNS sequences with these categories. Their category indicator may be a clazification or confirmation request ( l and 2). Comprehension checks do not occur in their model (3). Reactions to a response may be equal to category 4. Especially with longer and more complex sequences, their categorizations may lack descriptive and explanatory power. Not only Vazonis 8t Gass's categories for NNS-NNS interaction are problematic but so are the categories for com~ersatíonal mcxlifications in NS-NNS interaction described above by Pica 8z Doughty (1988). The list suggests a completeness and order which is not there. The modifications under four are of a different order than the ones under one to three. For example, a repetition (4) may be used to request clarifica[ion (1) 32 Chapter 2 Negotiation of Mecrning in !nlerctrltttraf Comnntnication of a trouble source or it may function as a confirmation check (2). However, it may also function as a clarification itself (which is not mentioned as a category but should parallel 1 to 3 as the move of giving clarification). Claritïcations, on the other hand, may also take the form of paraphrasing and expansion as stated under 3. Therefore, we need a descripdon mode] which is more extensive and complex and which can be used in the analysis of NS-NS, NNS-NNS and NS-NNS interaction. It should cover the kinds of negotiation moves (see Yule 8z Tarone, 1991) e.g. clarification request, confirmation check and their realizations, e.g. repetition or expansion. In the current study, a proposal will be made for a revision of the model for negotiation of ineaning and the classification of negotiation moves (see Chapter 4). From a language acquisition perspective it is also informative to describe the forms used to realize certain moves. Gass 8c Varonis (1985a) do discuss some forms to indicate non-understanding but do not describe these forms systematically. Whereas NSs may use conventionalized forms, for instance, to indicate trouble, NNSs may use other fornis depending on their stage of acquisition and [heir first language background. In addition, face considerations may influence requests such as those for clarification (Brown 8c Levinson, 1989: 76). Trouble indicating is an act threatening the negative face (desire of freedom from imposition) of the other by requesting help. Insofar as trouble indicators reveal lack of understanding, ihey also threaten the speaker's positive face (desire of recognition of positive self-image, see also Chapter 7). Therefore, the choice of trouble indicator forms may be a subtle one. Another point of interest may be the amount of negotiation of ineaning occurring in a conversation. If clarification sequences occur often and last several turns, they may have a negative impact on the interactional climate. Such sequences interrupt the main line of the conversation and threaten the face of the participants, and are therefore avoided. As Stalpers (1993) stated, at the ]east they can slow dow~ conversation considerably. There may be several factors which lead to a high relative share of ciarification sequences. Limited NNS communicative competence may be one factor and institutional context may be another; however, no systematic research into these factors has been done. It can be concluded that more research is needed on negotiation of ineaning in NS-NNS interaction. Attention should be paid to the types of moves of both speakers in interaction with each other. Therefore, empirical research of face-to-face interaction is necessary. The micro-analysis of interaction may contribute to the study of intercultural communication. Liberman (1990) argues that social science studies in intercultural communication should incorporate more of the methodology of ethnomethodological studies. He considers as an example of such a combined approach the analysis of interaction between native-speaking majority and nonnative-speaking minority members in gatekeeping encounters (see Gumperz, 1990). The current study of NS-NNS interaction uses concepts and methods of conversation analysis as well as SLA theory and it applies it to the domain of interaction with nonnative speakers. It looks not only at the structure of interaction but also at the verbal means used to realize certain moves, which is closer to discourse analytic practices. 2.5 Negotiation of Meaning in Clarificntiai Sequences 33 It can be concluded that the micro-analysis of interaction with second-language speakers may add explanatory power to a comprehensive theory of intercultural communication which accounts not only for the relationship between culture and communication, but views language as the most important tool for human communication. 2.6 AsymmetricalInteraction One of the aspects of ínteraction which has received a considerable amount of attention deals with unequal contributions of the different speakers depending on such factors as gender, proficiency and institutional roles. The NNS's limited communicative competence may lead to asymmetry in NS and NNS interaction behavior. Processes of asymmetry and dominance are of particular interest in creating insight into intercultural communication, for instance, in interaction between members of ethnic minority groups and institutional representatives which control resources such as employment or housing. In these contexts asymmetry may be created not only by differences in communicative competence but also by institutional roles. Section 2.6.1 deals with the concepts of power, dominance and asymmetry. The factors which may lead to asymmetry-communicative competence and institutional roles-are discussed in sections 2.6.2 and 2.6.3. 2.6.1 Power, Dominance and Asymmetry In the recent decades, a great deal of research has been done on power, dominance, and asymmetry in interaction. This research has been popular because these phenomena oppose the Western cultural ideal that participants in conversations should have equal rights and duties, and thus that the interaction should be symmetrical (Faerch 8c Kasper. 1985; Linell et al., 1988). In other words, it is preferred that speakers "... have the same option in turn-taking, the same obligations towards speaking, the same right to abstain from this, the same choice of speech acts, etc." (Holmen, 1985: 80-81). A great deal of research on dominance has concerned mixed-gender conversations (see also Brouwer 8t De Haan, 1987). Under influence of the feminist movement, the male dominance hypothesis was formulated, which stated that the social power which men tend to have in many societies is reflected in differences in the speech of inen and women (Thorne éc Henley, 1975). Several studies investigated the male and female use of certain speech indices, such as interruptions, topic initiation or speech productivity. Some of these studies found differences in such speech indices and attributed these to male dominance (Zimmerman 8z West, 1975; Mulac, 1989). This picture, however. turns out to be too one-sided. Later studies show that variables such as `expert status' may interact with gender and result 34 Cknpter 2 Negotiation o(MeuitinR in btterctrltura! Conimunicatton in diverse patterns of dominance (Leet-Pelligrini, 1980). In addition, some research shows that verbal behaviors have different functions depending on participant roles and context characteristics: for instance, a high occurrence of interruptions does not always indicate dominance but can show high involvement as well (Aries, 1987 ) In this study the focus will be on asymmetry rather than power or dominance. Asymmetry is a more neutral term than power or dominance, and refers solely to differences in certain aspects ot behavior between the partners. Such asymmetry, however, may be a reflection of power differences (Leet-Pellegrini, 1990). Linell et aL (1988) contrast the ideal of equality with two other types of interaction that seem by nature asymmetrical: institutional interaction and NS-NNS interaction. These two types of interaction will be described next. 2.6.2 Institutional Interaction Institutional interactions often have the form of gatekeeping interviews between institutional representatives and clients. Such institutional interaction with clients can be divided into several segments (Agar, 1985: 149-156). The first segment is usually the diagnosis. In this phase the institutional representative tries to fit the client's problem in a particular institutional frame and requests certain information to fill into appropriate slots. In this phase, the representative generally asks many questions and initiates topics, thus controlling the flow of information. According to Shuy (1983) client participation occurs mainly through responses and agreement but also through requests for clarification, interruption, expressing hesitation and uncertainty in responses. The second optional phase concerns directives. Usually something needs to be done about the client's problem either by the institutional representative or by the clíent. In the third segment, consequentially, a report is made for the files, which can be done either ïn the client's presence or afterwards. Efficiency, economy, time pressure and background knowledge are aspects of the discourse ecology as Agar (1985) calls it. The first three aspects limit the amount of time available to the participants. These restrictions are present in all forms of business talk (see Stalpers, 1993). The fourth aspect, background knowledge, can lead to even more interactional dominance by the institutional representative than had been expected on the basis of his authoritative role (Berenst, 1994: 177). Linell et al. (1988) point out that institutional interaction is generally asymmetrical. Many studies support this claim for different types of institutional interaction such as doctor-patient interaction and courtroom interaction (see Markova 8z Foppa, 1991; Adelswaerd, Aronsson, Jánsson, 8t Linell, 1987; AinsworthVaughn, 1992; Davis, 1988). Through the institutional role and background knowledge the representative acquires expert status, which has been shown to result into dominance. The factor expert status will further be discussed in the next section. On the other hand, the interactive behavior of a client who is put in a subordinate 2.6 As~'mmetricnllrtteraclio~t 35 role will likely be influenced by face considerations (Brown 8c Levinson, 1989: 74). This may influence negotiation of ineaning. To ask an official to clarify trouble may be too great an imposition. Therefore. the NNS may feel inhibited from explicitly indicating trouble in understanding. This may affect communication and language acquisition opportunities negatively. 2.6.3 Native Speaker-Nonnative Speaker Interaction Zuengler 8c Bent (1991) and Zuengler (1989) state that in the SLA literature there appears to be a general assumption that in NS-NNS interaction, the NS has more control because of havmg greater communicative competence. Woken 8z Swales (1989: 212) referred to several studies which found that NSs in interaction with NNSs "tend to take the lead in negotiating meaning, in nominating and terminating topics, in setting repairs in motion, and in offering assistance with syntax, lexis and pronunciation". Although this asymmetry seems very plausible, little empirical research has been done on this assumption. For example. Koole 8c Ten Thije (1994) observe that in team meetings between Dutch NSs and their Moroccan and Turkish colleagues, the NSs used their fluency to direct the conversation. A pilot analysis described in Deen (1994) showed similar results. Concerning NS adjustments in NS-NNS interaction, Larsen-Freeman 8t Long (1991) claim that there are only quantitative differences with NS-NS interaction but no qualitative ones (see also Long, 1983c). This is a strong a priori claim which needs to be empirically proven, although earlier research found many quantitative differences. For example. Pica (1987) finds that a NNS who has a subordinate role tends to participate less in negotiating meaning. It should be noted, however, that in this study the teacher is the NS; therefore, the asymmetry may also be caused by expert status instead of differences in communicative competence. Pica et al. (1989) show that a NS in negotiating meaning uses confirmation requests much more often than clarification requests. When a NNS requests clarification in interaction, the NS usually reacts by clarifying. When, on the other hand, a NS requests clarification, the NNS may not have the language proficiency to do the interactional work of repairing. Therefore, by requesting confirmation instead, NSs may take on the clarification work by providing hypotheses and linguistic models for the NNS's intended meaning. In both cases, independent of who caused the trouble, the NS manipulates the content of the trouble source. This may indicate that the NS also has more control over the content and the outcome of the interaction, and thus that there is asymmetry. Pica (1994) finds that NNSs, when they clarify, use simpler lexical modification than the NSs (1994: 515). This indicates that the NNSs if they perform the same role may use different means, which may show the influence of proficiency on negotiation. It is not clear whether no qualitative differences could also be found. Communication strategy research has shown that learners use strategies which compensate for gaps in competence. Probably cooperative NSs will use com- 36 Chapter 2 Negotiation o(hfeanin,q in lntereultural Communication plementary strategies to support the NNS in this compensation. In any case, linguistically, NS ungrammatical input is a type of adjustment that does not appear in NS-NS interaction but is typical for NS-NNS interaction. There may also be other factors which lead to asymmetry in NS-NNS interaction. Zuengler 8c Bent (1991) quote Beebe 8t Giles (1984) who claim NS ethnolinguistic status per se leads to asymmetry. This claim was refuted by Zuengler (1989) who shows in her study that the higher ethnolinguistic status of the NS does not automatically cause NS dominance in NS-NNS interaction. On the contrary, in her study, the NNSs dominate on several measures whereas on others no differences were found. Topic knowledge appears to be the main variable in explaining dominance. Zuengler 8t Bent (1991) for their study of NS-NNS interaction used beginning and advanced students of studies such as engineering and dairy science. They relate their results to relative knowledge of the topics discussed. However, they find that when the NS and NNS have equal knowledge there is no clear pattern of asymmetry: rather. on some measures, the NS and NNS score equally (topic moves, interruptions), whereas on others, either party may score higher. The NS talks more and the NNS uses more fillers and backchannels (feedback signals). Zuengler 8c Bent state that the NS and NNS appear to have different complementary roles rather than the NS dominating the NNS. The NS is the active speaker and the NNS is the active listener even though both have equal knowledge. 1'hus communicative competence may play a role. On the other hand, Zuengïer (1989) shows that the NNS dominates the conversation in amount of talk, fillers and backchannels but not on some other measures. Therefore, the outcomes of these studies do not show a consistent pattern. However, Zuengler 8c Bent view their outcomes as supportive of the Discourse Domain Model (Selinker á Douglas, in Zuengler 8z Bent 1991) which claims NNSs develop their interlanguages through various content domains. This knowledge makes somebody an informal `expert'. Therefore, different aspects can give a person expert status. First it may be his authoritative or institutional role. This may be real or perceived by assigning somebody an expert role in a task (Zuengler, 1987). Secondly, it may be background or topic knowledge. If this knowledge is combined with an institutional role, it is often acquired through special training and experience on the job, which makes somebody a `formal expert'. A study by Woken 8c Swales (1989) supports this `expert status' view based on topic knowledge and skill. They found that NNSs are not automatically dominated by NSs if they are the expert on the topics discussed. In their study on the influence of expertise and authority on NS-NNS interaction, Woken 8c Swales (1989) used NNS computer science students who taught native speakers to work with a word processing program. Thus, expert status was related to skill as well as knowledge. An interesting outcome is that the NSs never offered verbal help such as offering vocabulary explanations and the NNSs never asked for it. NNS expert status may thus inhibit negotiation of ineaning because it would threaten the NNS's status. There is probably an interaction between expert status and language proficiency. Yule 8~ MacDonald (1990) study the effects of type of task and participant arrangements on negotiation of ineaning in conversations between high-level and 2.6 Asymmetrica! ~nteraction 37 low-level NNSs. They find that when a higher proficiency participant has a dominant role, that is instructing the other to draw a map, there is little negotiation. However, when a lower protïciency participant has a dominant role there is substantial negotiation. It may be that as soon as more advanced speakers, native or not, notice a low-level of language proficiency of the other speaker, they will take control over the interaction (Zuengler, 1989). Therefore, there may be an interaction between expert status and proficiency. Linell et aL (1988) report initiative-response analyses in NS-NNS interaction between a 12-year-old immigrant boy with a native-speaking boy and of the same NNS boy in a language lesson with his NS teacher. The NS-NNS interaction between the two boys is balanced whereas the lesson with the teacher shows great asymmetry (1988: 433). They therefore claim that the expert status in institutional interaction has a greater effect on dominance than language proficiency. Young 8r Milanovic (1991) find that in oral proficiency interviews the NS examiner and NNS candidate make different contributions to the discourse. NS dominance is expressed through much greater NNS reactiveness and more NNS persistence of NS topics. NS dominance is not found in quantity of t-units. In this study, the NNSs talk twice as much as the NSs but the NSs do control topics and the tloor. However, here also the NS has expert status based on authority and knowledge; therefore, it is not clear whether differences in status or in communicative competence or both are responsible for the results. Thus, the relationship between dominance, social power, and verbal behavior appears to be less obvious than was supposed. More research is needed to detennine which (combination of) variables may influence dominance and how dominance is expressed in interaction. Questions remaining concern the role of the NNS's communicative competence on his participation in negotiating meaning. When the NNS becomes more proficient in the second language, does the interaction become more symmetrical? Zuengler 8c Bent's own study and the others reviewed do appear to find more symmetry in conversations with highly proficient NNSs; however, there are no longitudinal studies on this matter which could more convincingly show the influence of language proficiency. More research is needed on the interaction of expert status and level of communicative competence on asymmetry in interaction. In particular, more research is needed on this interaction in negotiation of ineaning. For example, the time pressure in institutional conversations may seriously influence the room for negotiating meaning, whereas the goal-orientedness of this type of conversation requires mutual understanding. The point of supposed NS dominance in NS-NNS interaction is even more important if Pica's claim (1987, 1994) is correct that active participation of the NNS in negotiating meaning supports language acquisition. 38 Chapter 2 Negotiation nf hleaning in Iruerctiltural Co~wrtunication 2.7 Conclusions This chapter leads to several conclusions. First, there is a need for integration of linguistic and social sciences studies in intercultural communication. In particular, the concept of intercultural communicative competence is diffuse because social scientists have mainly concerned themselves with the general aspects of interpersonal competence, whereas linguists have been concerned with language specific competence. SLA researchers have done a great deal of work on the influence of native and nonnative speaker behavior in interaction on the development of language specific communicative competence. Initially this research focussed on either the NS (foreigner talk, comprehensible input) or the NNS (comprehensible output, learner strategies). However, a tendency is seen to a combined approach on NS and NNS language use in interaction with each other. One particular area of interest in SLA research concerns negotiation of meaning, which is hypothesized to enhance not only comprehension and thus communicative effectiveness but also language acquisition. Communication between speakers of different first languages and cultures may frequently result in clarification sequences in which meaning is negotiated. However, more empirical research is needed on NS and NNS behavior in negotiation of ineaning. There are several factors which may influence the NNS's opportunities to negotiate meaning, such as communicative competence and expert status (institutional role and topic knowledge). These factors may lead to asymmetry in interaction. Such asymmetry may inhibit a NNS's active role in negotiating meaning, which in turn may influence ]anguage acquisition negatively. However, earlier research studies are not consistent in their results on the influence of communicative competence and institutional roles and do not specifically concern negotiation of meaning. Factors such as gender, role and proficiency appear to interac[ with each other in influencing NS and NNS interaction behavior. Therefore, more research on negotiation of ineaning in NS-NNS institutional and informal interaction is necessary. In the current study, NS and NNS negotiation of ineaning moves in clarification sequences will be longitudinally studied through micro-analyses of interaction. Self-initiated self-repair is excluded because this does not lead to negotiation of meaning between speakers. The NS and NNS behavior will be described per move with regard to linguistic form and strategy type using the same analytic framework for both NS and NNS participants. In addition, the influence of NNSs' level of communicative competence and NSMNS institutional roles on negotiation of meaning is studied. 2.7 Condusions 39 Chapter 3 Research Questions and Method 3.0 Introduction In this chapter, first the research questions will be presented, which are derived from the objectives described in Chapter 1 and the theoretical and empirical findings described in Chapter 2(section 3.1). Secondly, protïles of the nonnativespeaker and native-speaker informants selected from the European Science Foundation project (ESF) are presented (section 3.2). Then the institutional and informal conversation data selected from the ESF data base will be described (section 3.3). Consequently, the data analysis process based on the grounded theory approach will be described (section 3.4). 3.1 Research Questions If we want to gain insight in how NSs and NNSs negotiate meaning in clarification sequences, we first need to know which moves constitute clarification sequences. A model for negotiation of ineaning in interaction between NNSs was presented by Gass 8c Varonis (1985a). However, in Chapter 2 it was argued on the basis of other research on NS-NNS negotiation that this model does not appear to differentiate sufficiently between the different possible moves. Thus the first research question IS: la. Which moves do claritïcation sequences in NS-NNS interaction consist of? Schegloff, Jefferson 8c Sacks (1977) showed that repair moves in NS-NS interaction are organized into adjacency pairs of initiations and completions. There may also be a third part, such as a reaction which closes the side-sequence (Houtkoop, 1987). It is expected that the negotiation moves in NS-NNS interaction are either initiations, completions of, or reactions to, a move of another speaker, and that they are organized in adjacency pairs. The second part of the first research question will be: 3.0 Introduction 41 lb. How are negotiation moves organized in clarification sequences'? ln Chapter 4 the results on questions la and Ib will be presented and discussed. Once we know which moves clarification sequences in NS-NNS interaction consist of and how such sequences are organized, the next question concerns the amount of the interaction which is dedicated to negotiation of ineaning. If there is a great deal of negotiation of ineaning, many opportunities for language acquisition may occur; at the same time, however, communicative effectiveness may be hindered. Two factors which may influence the quantity and quality of NS and NNS negotiation moves will be investigated systematically: the level of NNS communicative competence and the type of interaction (institutional versus informal conversation). These factors may influence the amount of talk which is spent on negotiation of ineaning. It is not clear how NNS communicative competence or interaction type will influence the amount of negotiation of ineaning. It might be expected that as the NNS's communicative competence develops, fewer problems will arise. In addition, those that do arise will take fewer turns to be resolved. The amount of negotiation would thus decrease over time. However, there may also be a certain limit to the amount of negotiation interaction partners can handle in the beginning stages of acquisition without the communication breaking down. It may be that NNSs need a certain level of communicative competence to profit from negotiation of ineaning and thus more clarification sequences will occur when the communicative competence increases. Faerch 8z Kasper (1985: 19) found that learners at the intermediate level (grade 10) more frequendy initiated negotiation of ineaning than lower-level learners. Though they still experienced considerable perception and production problems, they were more able to make use of the NS's help. Furthermore, the time which is allowed for negotiation of ineaning may be influenced by the interaction type, by the presence or absence of a script and clearly defined roles. In institutional conversations there may be a greater need for the exact interpretation of inessages and thus there may be a substantial amount of negotiation of ineaning. On the other hand, as soon as the script and terminology are acquired less trouble may occur because of the predictability of these conversations, whereas for informal conversations unpredictability is characteristic and thus negotiation of ineaning may be stable. The next questions concern the relative number of turns spent on negotiation of ineaning. 2a. 2b. What is the amount of interaction spent on negotiating meaning? How do NNS communicative competence and interaction type influence the amount of interaction spent on negotiating meaning? The results on question 2a and 2b will be presented in Chapter 5. The next question concerns NS and NNS roles in negotiating meaning. Due to limited communicative competence in Dutch and interactional role, a NNS may 42 Chnpter 3 Research Questions and Method display different preferences in types of moves used and may onty be able to make a limited contribution to the negotiation process. A NS may clarify trouble and propose models for the NNS's intended meaning more often because he has the linguistic means to do so. The NNS may indicate trouble with minimal linguistic means and confïrm the models of intended meaning more frequently. This would lead to asymmetric participation in negotiating meaning and the NS doing most of the linguistic work. This may limit NNS opportunities for language acquisition, because trouble indicators and confirmations require little communicative competence and thus may not push the NNS to create comprehensible output. The next question concerns only those negotiation behaviors which can be characterized as central to the negotiation of ineaning process and which can be realized in different ways with a variety of linguistic forms. These are trouble indicators, trouble clarifications and confirmation checks (see Chapter 4). Trouble clarifications and confirmation checks in particular may be strongly influenced by gaps in NNS communicative competence because they manipulate content and form. 3a. Is there asymmetry in NS and NNS participation in negotiating meaning? The second part of this question deals with the factors NNS communicative competence and interaction type, which may explain any asymmetry found. When the NNS's communicative competence is limited, he may not be able to clarify trouble himself but leave it to the NS to provide models for intended meaning. When the NS creates a problem, the NNS may not be able to provide such models or even indicate specifically what the problem is. As the NNS's communicative competence increases, he may become more capable of clarifying trouble and of checking the NS's intended meanings himself, which may enhance language acquisition. The distribution of negotiation of ineaning moves over the NSs and NNSs may not only be influenced by the NNSs' communicative competence bu[ also by the interaction type. Institutional interaction may differ from other forms of conversational activity, such as informal conversations. In informal conversations, verbal activity is often more geared towards defining the relationship between the speakers and the communication situation than towards exact information transfer (Watzlawick et al., 1967). On the one hand, the result in such a context may be that problems in understanding will often no[ be attended to because this may threaten face, and is therefore, on affective grounds, avoided. On the other hand, however, when the interlocutors become familiar with each other and nothing is at stake, a safe atmosphere is created which might stimulate speakers to negotiate meaning. In institutional conversations, the communication is usually informationoriented rather than relation-oriented. In addition, there is frequently some script confining speakers to certain interactional roles. The participants set up specific (sometimes conflicting) goals to be achieved, such as the exchange of information or certain actions to be taken. The interactional roles played in negotiating meaning may be restricted by the institutional roles. The latter may create asymmetry in the interactional contributions. 3.1 Researeh Questions 43 3b. How do NNS communicative competence and interaction type influence the distribution of negotiation moves over the NS and NNS? The results of questions 3a and 3b are reported for NS and NNS trouble indicators, trouble clarifications and confirmation checks. The distribution of these moves over the NSs and NNSs is described in Chapter 6. From an acquisitional point of view not only is the organization of moves in clarification sequences itself of interest but so is how these moves are realized in language use, that is, which strategy types can be distinguished and which linguistic forms are used for the moves. Such analysis may create insight into how NSs of Dutch indicate trouble, adjust their input in clarifications and present models for NNS meanings, and how NNSs develop forms and strategy types for these negotiation moves in their second-language use. As stated before, it is not clear whether differences in negotiation moves will only be quantitative as Larsen-Freeman 8c Long (1991) claim or also qualitative. This analysis will again be restricted to NS and NNS main negotiation moves. These moves are analyzed for quantitative and qualitative differences of form and strategy type. In addition, they are compared with each other over the points of ineasurement to determine whether there is some development in the NNS moves over time. In addition, the results will be compared for the interaction types in order to determine the influence of a script and institutional roles on the strategy types and linguistic forms of the main negotiation moves. 4a. 4b. How can NS and NNS trouble indicators, trouble clarifications and confirmation checks be described in terms of their types and fonns? What is the influence of NNS communicative competence and interaction type on the variety and quantity of trouble indicating, clarifying and checking behavior of the NS and NNS? 'I'he results of these analyses will be presented and discussed for trouble indicators in Chapter 7, for irouble clarifications in Chapter 8, and for confirmation checks in Chapter 9. 3.2 Informants In this section, first the ideal nonnative-speaker informant profile will be presented. Consequently, individual portraits will be given of the nonnative speakers who participated in the study to show how they suited and sometimes diverged from the ideal profile. Then some information on the native speakers with whom the nonnative speakers interacted will be given because they differ considerably on some personal characteristics which may influence the interaction. ~ Clrupter 3 Resenrclt Questiott.c atu! Metftod The NNS informants who participated in the ESF project were selected to closely fit an `ideal' informant profile. This profile was formulated to control characteristics which are supposed to be important for the acquisition process, such as age and educational background (Perdue, 1984: 253). At the time when the data collection started, the NNS informants were supposed to have lived in the Netherlands no more than one year. They were to be young adults and to have no native target-language-speaking spouse or children of school age. Their level of formal education was supposed to be limited (elementary school only). At the time of the first recordings, the NNS informants' proficiency in Dutch was to be very low. Finally, they were supposed to be spontaneous acquirers of the target language, and thus should not have received formal instruction in Dutch. Four key informants were selected out of eight NNS informants all of whose data sessions were transcribed: two Moroccans and two Turks to represent both source languages, Moroccan-Arabic and Turkish, selected for the Dutch part of the ESF project. Two relatively slow and two relatively fast learners were selected. From their sociobiographic profiles and their performance on language tests taken in the first sessions, the Moroccan-Arabic woman Fatima and the Turkish man Mahmut were estimated to be relatively slow learners, and the Moroccan-Arabic man Mohamed and the Turkish man Ergun as relatively fast learners (Broeder, 1991). 3.2.1 The Nonnative Speakers The following impressionistic portraits of the four nonnative speakers are derived from Broeder (1991: 14-15). Apart from general background information; in particular, characteristics which deviate from the ideal nonnative speaker profile are discussed. Fatima After only two years of elementary school in a town in Western Morocco, Fatima had lessons in sewing and knitting, which she made her profession. She later had a little shop and gave lessons to other women. At the age of twenty four, she married a Moroccan who had been living in the Netherlands and moved to a city in the Netherlands, Tilburg. There she had a part-time job as a cleaning woman. During part of the data collection period, Fatima received very limited instruction in Dutch (one hour a week). Her son was born two years after she moved to the Netherlands. She had little contact with Dutch native speakers except for a period in which her husband was out of town and she had to take care of daily business herself. She knew a small amount of French. Mohatned Mohamed was the only NNS informant who after primary school attended two years of secondary school in Casablanca, a big city in Morocco. In addition, he received training as a mechanic but did not complete it. He had limited proficiency 3.2 Informants 45 in French. His family joined his father in the Netherlands when he was nineteen years old. There he worked as a factory worker most of the time. He had many formal and informal contacts with Dutch native speakers. His relationship with his parents was problematic. During the data collection he therefore moved out. Part of the time he stayed with his uncle, part with his first Dutch girlfriend's parents, and later with another Dutch girl. Mahmut After finishing primary school in a small town in the middle of Turkey, Mahmut worked as a mechanic. He moved to the Netherlands to join his wife when he was nineteen years old. First, he lived with his wife's parents but later he rented a separate house for his family on the same street. After a year of unemployment he found a job in a factory which he held during the data collection period. He had limited formal contacts with Dutch native speakers. Mahmut frequently reflected on his language proficiency but had no time to attend courses because of his family responsibilities. Ergiin Ergun also started working as a mechanic in Turkey after he attended primary school. He left Turkey to join his parents when he was seventeen. After his arrival, he irregularly attended a language course of two hours a week at a vocational school over a period of five months. However, at the time the project started, his language proficiency in Dutch was minimal. He worked off and on as a factory worker. Like Mohamed he had no (family) responsibilities during the project and so he was able to spend much time socializing with Turkish and Dutch friends, going to discotheques and playing soccer. After having had problems with his parents, he moved away to Groningen, a city in the Northern part of the Netherlands, where he worked as a car-wreckec He had a couple of relatives there, but he had even more contact with Dutch people because there were fewer Turks in this area. 3.2.2 The Native Speakers The focus of the ESF project was mainly on the language use of the nonnative speakers from an acquisitional perspective. Therefore, the NSs were, unlike the NNSs, not selected according to certain profile criteria. Thus the NSs ditfer from each other not only in their roles in the interaction (institutional versus informal) but also, among other things, in their familiarity with communicating with NNSs in different contexts. Because of the interactional focus of this study, the portraits of the NSs will be given too. Their language behavior will serve as baseline data with which to compare the NNS behavior. The NS negotíation of ineaning behavior in particular may give some insight into the language behavior in Dutch that is to be acquired by the NNSs. One may ask whether data of NS-NNS ínteraction is appropriate to come to such insight. The question whether NS-NS data would be better to achieve this goal cannot be answered because no comparable NS-NS data 46 Ciwpter 3 Researclr Questions and Method were available. However, a problem with NS-NS data is that, generally, few comparable problems in understanding arise which may lead to the negotiation of meaning. Selting (1988) describes three types of local problems in understanding which are repaired in NS-NS interaction: "problems on the.level of acoustic decoding or formulating an utterance, problems on the level of semantic organization (relating objectslexpressions to each other or to referents and meanings), or problems on the level of expectations (compatibility of inessages or activities with own knowledge and expectatíons)" (Selting, 1988: 296). These types of problems are signalled with intonation. She adds a fourth level for global problems in understanding which, according to her, are signalled by syntactic cues such as repetition. This is the level of problem that is very present in NS-NNS interaction. Therefore, the advantage of the NS-NNS interaction data used is that sufiicient NS negotiation of ineaning behavior was available for study. In the first cycle, the role of the housing official was assumed by a social worker who had had experience in working with ethnic minorities. In the second and third cycles, a housing official participated who had had professional contact with ethnic minorities because he worked at one of the housing corporations in Tilburg that had many migrants as clients. The interviewer in the first cycle will be referred to as Paul, the interviewer in the other two cycles as Bart. In the informal conversations, the NNSs spoke wíth a female researcher (Janet), except for one conversation in cycle 3 with Mohamed which was taped with a male researcher (Peter). Janet was a minorities consultant at an Educational Center and had been a teacher of Dutch as a Second Language. Therefore, she had had experience in talking to NNSs. Peter also had some experience because he had taught Dutch as a second language to foreign workers on a voluntary basis. He participated in the last informal conversation with Mohamed because the relationship between Janet and Mohamed had deteriorated because of some misunderstandings, in such a way that normal conversation was no longer possible. Figure 3.1 shows which NS participated in which conversations. Informal conversations Data Cycle Participating NS 1 Janct Jane[ Institutional conversations 3 1 Janed Paul 2 I Bart 3 ~ Bart Petcr Figure 3J: Native speaker participant.r in the informal and institutional conversations 3.2 Informants 47 3.3 Data For the European Science Foundation project, conversations of nonnative speakers interacting with native speakers of Dutch were tape-recorded andlor videotaped monthly in three repeating cycles of nine monthly sessions with a total range of two-and-a-half years (see also Feldweg, 1991). The data contained several types of activi[ies ranging from informal conversations, institutional conversations and film-retellings to more experimental data, such as language tests. The collection in cycles made longitudinal analyses of the data possible. For this study, only conversational data are used. In these conversations, it can be expected that there is a two-way flow of information and some interchangeability of roles, which may lead to initiation of negotiation of ineaning by the NNS as well as the NS. Two types of conversations were selected for each NNS informant, which recurred in the three data collection cycles: an informal conversation with a NS researcher with a more or less free flow of topics, and an institutional conversation with a fixed script in which the NS held the role of an institutional representative and the NNS that of a client. All but one conversation were taped at the university studio. This one informal conversation wi[h Fatima was tape-recorded at her home because of her domestic situation at the time. For the institutional conversation, a registration interview for housing was selected. The institutional conversations were both tape-recorded and videotaped in the óth session of a cycle, except for the second housing office conversation with Fatima, which was tape-recorded only. The sixth session did not contain comparable informal conversation; therefore, informal conversations from the seventh session were used that were tape-recorded approximately one month after the role play. Successive sessions were chosen to create the greatest comparability of conversation types, thus minimizing differences in NNSs' level of communicative competence between the two interaction types. The first analyzed sessions were taped approximately a year to a year-and-a-half after the NNSs' arrival in the Netherlands. The transcrip[s made for the ESF project were used for the analysis. At some points some transcribed data were added. These consisted of a few excerpts which were not transcribed by the ESF project researchers because these excerpts contained extended NS speech and were thus less interesting from a NNS acquisitional perspective. The approximate length of the conversations was timed in minutes and the number of turns counted (see Table 3.1). A turn was defined as any utterance of a speaker which ends another speaker's prior turn, including back-channelling (McLaughlin, 1984). Because of the fixed script, the institutiona) conversations were analyzed completely regardless of their length. Therefore, the number of turns in the institutional conversations is higher than that of the others. For the informal conversations, only parts were analyzed because these were quite long. The parts analyzed were more or less comparable as to topic treatment. In Fatima's first institutional conversation. the number of turns is quite low due to her low level of communicative competence. In the last informal conversation the number of 48 Chahler 3 Researcla Queslions and Metltotl turns is also low, particularly when compared to the length in minutes. This may have been caused by the fact that this conversation was tape-recorded at her home while her husband and child were present. Her husband sometimes participated in the conversation. His turns were not further analyzed. However, no more comparable data were available in this session to extend the number of turns for analysis. Conversation type Cycle Institutional conversations Informal conversations 1 rnin. tutns 2 min. turns Fatima 17 409 20 406 10 165 8 Mohamed 11 12 254 17 415 Mahmut 10 287 21 507 Ergun 10 217 ll 310 212 3 min. tums 1 min . turns 2 min. tutns 3 min. tums 181 IS 374 23 547 16 363 22 547 9 305 20 424 15 344 45 10 269 11 298 20 477 1294 29 902 20 524 Table 3.1: Length of conversations appro.ximated in miiiutes and turns res~ectively 3.3.1 The Institutional Conversations The housing office conversation is a typical `gate-keeping' encounter (Erickson, 1976) in which the institutional representative controls excess over affordable housing. In the Netherlands, a large proportion of the rental homes consist of rentcontrolled social housing. To be eligible for such housing one has to apply at the municipal housing office andlor housing corporations, where personal infotmation such as marital status, number of children and income is recorded on an application form, in addition to specific wishes and needs concerning the new housing. Generally, these wishes and needs will be discussed in a conversation with a housing official who at that time either checks the information already filled in or fills in the application form for the applicant (the first segment of institutional interaction: Agar, 1985). The rules and regulations concerning rent-controlled housing are quite complex. There are several types and degrees of urgency taken into account (e.g. medical, social, and economic) which may lead to receiving a housing assignment more quickly. This type of housing office interview was staged in a role play. The main reason why role plays were used instead of authentic interviews is the longitudinal design of the project. It would have been impossible otherwise to obtain interaction data with comparable content from four NNS informants at three different points in time. In addition, it would have been more difficult to get access to that type of data for privacy reasons. For the housing office conversation, the NNS infotmants were instructed (in their mother tongue) to take up the role of a fiance(e) who, like hisltter partner, lives at home with hislher parents but who wants to marry soon, and therefore 3.3 Data 49 needs a house or apartment. All events were fictional, although the NNS informants added real facts and experiences from their own living arrangements and contacts with housing corporations. The interviewers used an authentic application form for housing accommodation from the housing corporations in Tilburg. They discussed all topics relevant for the completion of this form with the NNS `applicant'. These we.re divided into six sections: 1. personal data of applicant 2. data on current living arrangements 3. data on household applying for new housing 4. housing preferences 5. special circumstances (for indications of urgency) 6. choice of contact address 3.3.2 The Informal Conversations In the 7th session, the conversations could be divided into episodes which covered different topics. The guiding topic of the episode used for analysis concerned cultural differences between the Netherlands and Morocco or Turkey respectively. This general topic was initiated by the NS project researcher but further freely developed by both speakers. Because of the informal character of the conversations, the NNS sometimes also developed other topics such as work or family related ones. Kalin (1995), who also analyzed comparable conversations from the ESF-project with Finnish speakers of Swedish, rightfully argues ihat the informal conversations used also have some features of institutional conversations because of the context of the recordings in the university studio and the participatíon of the researchers. However, these conversations will be refen-ed to as informal conversations because it is clear that these conversations differ from the housing office conversations, particularly in the free development of topic and the lack of a fixed script. 3.4 Data Analysis This section wil] describe the selection process of the clarification sequences and their subsequent coding with a computer program for qualitative data analysis. 3.4.1 The Selection of Clarification Sequences From the informal and institutional conversations, those parts of the interaction in which miscommunication surfaced and meaning was negotiated in clarification sequences were selected for further quantitaUve and qualitative analysis. Clarification sequences generally start with a turn which causes the other speaker trouble 50 Ch~pter .~ Research Questions nnd Method in understanding. This speaker does not react with a preferred turn but often indicates that there is a problem in understanding or checks intended meaning. Generally, these indicators occur right after the trouble source (Nofsinger, 1991: 126). In Example 1, the NS asked Mohamed where his girlfriend lived. Mohamed did not answer the question but indicated trouble in understanding. (1) Institutional conversation with Mohamed, Cycle 3 Bart: waar is dat? where is it? Mohamed: hm? mhm? Bart: waar is dat? where is it? Mohamed: f hier in Tilburg f here in Tilburg Bart: hier in Tilburg ja here in Tilburg yes In some other cases, a trouble source is not followed by a request for clazification or confirmation of the intended meaning, rather, the first speaker concludes from the second speaker's (lack of appropriate) reaction that there is a problem. Thus this first speaker, who caused the trouble, may initiate a clarification sequence. This is the case in Example 2. (2) Informal conversation with Fatima, Cycle 2 lanet: slik je veel medicijnen? Fatima: ja gced Janet: ja?jij? ja neem je medicijnen? Fatima: ik? lanet ja Fatima: nee ik niet veel do you take a lot of inedicine? yes fine yes? you? yes do you take medicine? me? yes no I not much A clarification sequence is thus identified by the orientation of at least one of the interlocutors towards a trouble source. By taking the subsequent reactions of the interlocutors after a trouble source as a signal, it becomes less speculative to identify a piece of interaction as a clarification sequence. Otherwise the decision would only be based on the researcher's interpretation based on reviewing the whole of the interaction. 3.4.2 The Coding Process This study was an explorative journey. The qualitative analysis is therefore the most important. Some quantitative analyses of frequencies and distribution of negotiation moves, types and forms were added after the qualitative analyses led to a descriptive coding scheme representative for the data. These quantitative analyses are not included to show any statistical representation for a certain population, but to use as indications for the systematic occurrence of the phenomena described. The qualitative analyses will be described below, whereas the quantitative analyses will be described further in chapters 5, 6 and 7. 3.4 Data Analysis 51 The purpose of the qualitative analyses was to develop a coding system with which NS and NNS negotiation of ineaning behavior could be described. After the clarification sequences were selected, they were coded on several characteristics. Because there was no complete structural descriptive scheme available for moves in NS-NNS clarification sequences, the turns in the transcripts which were part of clarification sequences were initially openly coded with a computer program for qualitative data analysis (Peters, Wester 8t Richardson, 1989). The computer program Kwalitan is based on grounded theory (Glaser 8t Strauss, 1979). The goal of this approach is to derive theoretical concepts through a qualitative analysis of empirical data. The program Kwalitan aims at supporting the cyclical conceptualization process that leads to systematic coding and description of naturalistic data. It gives sufficient room to the necessarily stepwise interpretation and re-interpretation of the data. Audiotapes and videotapes (the latter if available) of the data were used to support coding decisions. Relevant coding categories of other sources were incorporated where applicable (see Chapters 4, 7 and 8). The codings contained three dimensions: (i) negotiation move, (ii) specification of strategy used (type), and (iii) linguistic form. The three levels will be described in more detail. On the most global level, the turn contains one or more turn constructional units with which a negotiation move can be realized, such as a trouble clarification or a confirmation check (Sacks et al., 1974). A turn constructional unit can be identified as an intonational unit andlor by syntactic criteria. It can be an independent clause or a non-restrictive dependent clause, but also smaller and less complete units, such as an elliptical sentence or a one-word affirmation. Such incomplete units are also called satellite units (Bygate, 1988). In the data analyzed, one turn generally consisted of no more than one negotiation move. This indicates that the turns were rather short, which may be due to the relatively low level of communicative competence of the NNSs. However, when occasionally two moves were realized in one turn, they were both coded. More than two moves were rarely found and if they did occur, for practical reasons, they were left out of the analysis. One move generally consisted of one turn contructional unit. The only moves which more frequently contained more than one turn constructional unit were trouble clarifications, if they elaborated on the trouble source. In Example 3, Janet wanted to know how Ergun could pass the theoretical part of the driving exam in the Netherlands without enough language proficiency in Dutch. (3) Informa! conversatíon with Ergun, Cycle 3 Janet en al die vragen die je mcet invullen? Ergiin: Janet Ergun: Janet: 52 ik weet ik niet - ~fotTttulaic expression~ weet je niet? nee ik weet ik niet van dit bordbetekent van t dan krijg je vier vragen en mcet je aanktuisen welke delgcede and all those questions you have to fill out? [ don't 1 know - ~formulaic expression~ you don't know? no 1 don't 1 know (ike rhis sign means fike t then you get four questions and you have to mark which thelright Chapter 3 Research Questions and Method lhet gceie antwoord Iwat het gceie antwoord is lthe right answer Iwhat the right answer is The second level of coding concerned a further specification of [he strategy types used for a particular move. For example, the move trouble clarification could be realized by a repetition of the trouble source or by an elaboration as in Example 3. All moves were coded for such strategy types. However, only trouble indicators, trouble clarifications and confirmation checks were eventually described for their types, because these moves are most essential to the negotiating of ineaning process (see Chapters 7, 8 and 9 respectively). The third level of coding concerned the linguistic forms of the negotiation moves. Again, all moves were coded for their linguistic form but only the results of the main negotiation moves are included in this study. These codings contained, for instance, information about whether a move was a certain type of question (such as a yes~no or Wh-question) or a declarative. The form categories are further discussed in Chapter 7. On the basis of a pilot analysis of the first institutional conversation with Mohamed, the coding scheme was developed for all three levels of coding. When the rest of the conversations contained new phenomena, or led to new insights, new categories were developed and added or old categories subdivided where necessary. With such changes to the categorization, the data already coded were checked and codings adapted as needed, thus making the data analysis a cyclical process. Success of the clarification and trouble source types were also coded but not included in this study, because the interpretation appeared to be very subjective and no second rater was available to determine the reliability of the interpretations. It should be mentioned, however, that all coding work is by definition subjective. It was done by the researcher using her own knowledge of interaction as a member of the Dutch speech community. The coding of the NNS data is more problematic, because the NNSs are members of other speech communities, whose knowledge and skill in the Dutch language and culture is in development. However, in most cases NNS behavior also could be interpreted. The researcher could compare moves to instances in interaction occurring earlier and later and could use the development of a certain sequence after the occurrence. In addition, some other information was available. There were self-confrontation intervíews in which the NNSs were confronted with tapes of earlier conversations and asked for intended meaning. In addition, there were still some former ESF project researchers present who had participated in the data collection and could give background information about certain conversations. Taken together, these conditions increased [he reliability of the codings. 3.4 Dara Aitalysis 53 Chapter 4 The Organization of Clari~cation Sequences 4.0 Introduction In this chapter, the first research yuestion will be answered: la. Which moves do clarification sequences in NS-NNS interaction consist of? First a description will be given of the negotiation moves NSs and NNSs used in the clarification seyuences (section 4.1). Secondly, the relative frequency of NS sequences in which a NS caused trouble will be compared to NNS sequences in which a NNS caused trouble (sectíon 4.2). In addition, the second part of research question 1 will be answered: lb. How are negotiation moves organized in clarification sequences? The organization of negotiation moves in the NS and NNS sequences will be discussed in section 4.3. The chapter ends with conclusions in section 4.4. 4.1 The Negotiation Moves After selecting the clarification sequences from the conversation data, the negotiation moves were coded for the type of contribution they made to the negotiation of meaning. Although the coding categories emerged from an open coding process, many of Long's categories for NS adjustments (1983a) and NS repair categories from Schegloff, Jefferson 8i Sacks (1977) could be applied to NS as well as to NNS negotiation moves. The coding resulted in a taxonomy of twelve coding categories which can be applied to both NS and NNS negotiation moves. These categories are thus much more sophisticated than the four-part Varonis 8t Gass model for NNS-NNS sequences (1985a; see also Chapter 2). Example 1 is a clarification sequence which contains many of the negotiation moves which will be described below. In this sequence with Fatima and the housing official, the French word `jardin' (- Dutch `tuin') causes trouble. At the beginning of each line the initials of the negotiation move presented are given (see also Figure 4.1). 4.0 lntroduction 55 (I) Instilutional conversation with Fatima, Cycle 3 ts Fatima: ja ook eh mijn zoon nou wil eh loop in die jardin ti tc tc cc co Bart: Fatima: waar loopt ie? in die jardinl buiten die~in die huis in de tuin? ac Bart: Bart: Fatima: yes also uh my son now wants uh walk in that jardin ('jardin' is French for'garden'1 where dces he walk? in that jardin outside thad in that house in the garden? yes garden yes ja tuin ja This clarification seyuence starts with a main line turn that caused trouble. Fatima used the French word for garden and did not self-initiate a repair because she probably could not retrieve the Dutch word. Therefore. Bart starts an other-initiation by requesting repetition. Fatima clarifies by repeating the reyuested part of the trouble source and by elaborating on it. Bart requests confirmation by rephrasing the trouble source with the correct Dutch word for garden. Fatima confirms that this is the word she intended with an affirmatíve feedback signal and a repetition. Bart acknowledges that the trouble is resolved sufficiently with another feedback signal. Several moves were found with which the above described actions are performed. They can be distinguished as being either main line moves which start and finish the sequence or side line moves which are the heart of the negotiation ( see Figure 4.1). ts - trouble sourcc Main line moves mr - main line reaction tr - thematic reaction - escape NEGOTIATION MOVES tc - trouble clarification cc - confirmation check ti - trouble indicator Sidc line moves co - confirmation gc - gratuitous concurrence ac - acknowledgement cm - comprehension check mc - meta comment Figure 4J: The negotiation moves 56 Chapter 4 The Organi~ation of Clarification Sequences The side line moves can be differentiated into three types. The first type manipulates the content of the trouble source (trouble clarification, confirmation check). The second type elicits or gives feedback (trouble indicator, confirmation, ete.). The third type comments on the problematic character of the interaction (comprehension check, meta comment). First the main line moves will be discussed and then the side line moves. 4.1.1 Main Line Moves Trouble source (- ts) When a problem in understanding arises one of the speakers may interrupt the main line of the conversation by indicating, clarifying or checking understanding. Thus, the second turn displays an analysis of the first one as a problem (Levinson, 1989: 321). The first turn which triggers a clarification sequence will be called trouble source (Schegloff, Jefferson 8c Sacks, 197?). This definition excludes problems that are not attended to by the speakers, such as miscommunication and non-understandings that do not surface. Trouble sources may signal local problems in, e.g., pronunciation or lexicon (see Example l, 4), or they may signal more global problems concerning the referent (Example 3), or pragmatic function of the utterance (Example 2). Main line reaction (- mr) If a problem seems to have been satisfactorily resolved, often a reaction follows similar to the preferred reaction which was expected after the trouble source. This move brings the conversation back to the main line and thus is called main line reaction. It often forms a pair with the trouble source, even though the second part is not adjacent. However, the orientation to the second part can stay all through an insertion sequence (Nofsinger, 1991: 75-77). In Example 2, Janet asked what Mohamed did after some fireworks exploded in his hand when he was a child. (2) Informal conversation with Mohamed, Cycle 2 ts Junet aa en toen? ti Mohamed: hm? tc ti tc ti tc gc tc mr Janet: Mohamed: JaneC Mohamed: Janet Mohamed: Janet: Muhunted: en toen? `[en daatna?J ` [en tcen] wat heb je toen gedaan'? t je was helemaal alleen ja en dan zie je dat en ja ik ;.ie Lzx) huilen naar huis ~ ~ -daughs~ nvaaa~ - ~imita[es crymg~ -daughs~ 4. I The Negotiation Moves nh and then? mhm? and then? `[and afterwards?] `[and then] what did you do then? t you were all alone yec and then you see that and ves [ see (zx) cry home~ ~ -daughs~ ~wa.tiv - ~imitates crying~ -daughs~ 57 Thematic reaction (- tr) Sometimes the interlocutor reacts in a way he sees as appropriate and which is content-related, but which is not a preferred reaction. From this reaction, the other speaker concludes he was not sufficiently understood. This is called a thematic reaction. It could also be considered to be a second pair part to the trouble source. In Example 3. Mohamed reacted to the question of the housing official whether he could not live with his parents that he had lived with them for ten years. He might have understood the word ~olmtg (literally 'so long') as the question word hoelang (how long). (3) Institutional conversation with Mohamed, Cycle ts Paul: kun jelkun je niet zolang bij je ouders wonen ? cc Mohamed: t ik? colte Paul: ja 1[in de c.~straat] - ~name~ tr Mohumed: `l(.r)J ik uh nnu uh t~en jnur 1 can't youlcan't you live with your parents in the meantíme? t me'? yes `[on c.~ street] - ~name~ `(Ix)J I uh nox~ uh ren yeur.c Escape (- es) When a speaker has no faith in a positive outcome of the negotiation of ineaning he may cut out of a clarification sequence by changing the topic. This move is cal]ed escape. It may occur when a speaker doubts whether the problem can be resolved. An escape may indicate lack of cooperation as well as a cooperative attitude (saving one's own or the other's face). For example, on the question whether it was an in-law's wedding Mahmut was referring to, Janet got no preferred reaction (see Example 4). After a repetition, she apparently had no other means to explain what she meant, and when coffee was brought in she reacted to this interruption and left the seyuence unended. After a pause Mahmut took up the conversation with a different topic. Apparently they both decided not to attempt to resolve the problem any further. (4) Informal conversation with Mahmut, Cycle 3 Janet schoonfamilie? tS ti tc ti Mahmut Janet: tc laner e.c Mahmut family-in-law? what? wat? family-in-law schoonfamilie in-law? schoon'? - ~imitates~ - ~imitates~ family-in-law isn't it t schoonfamilie toch t `your in-law jouw schoon` uh jctntastic ua Jantustisch -~somebody enters and serves coffee~ 4.1.2 Side Line Moves Trouble clarifrcntion (- tc) Trouble clarifications are moves with which a speaker tries to repair trouble. A trouble indicator by the interlocutor may stimulate this speaker to react with such a move. A trouble clarification manipulates the content and~or form of the [rouble 58 Chupter 4 The Orguni;ation of Clariftcntion Seqtrences source. A speaker may repeat what was said before with a similar or adjusted utterance, or elaborate on what has been said in order to achieve understanding (See Example 1). In the terms used by Schegloff, Jefferson 8c Sacks (1977) trouble clarifications are instances of other-initiated self-repair. Confirnuttion check (- cc) When a trouble source occurs, and the iirst speaker does not initiate a self-repair, the other speaker may react to it in two ways. He may request clarification. This negotiation move is called a trouble indicator. These indicators will be described below. In addition, if the other speaker thinks he has understood all or part of the first speaker's trouble source but wants to check understanding, he may use a confirmation check. Sometimes the listener is not certain he understands the speaker sufficiently and wants to double-check. If he has understood enough, he may be able to manipulate the other's content and form, and formulate some hypothesis concerning the intended meaning. This move is called a confirmation check (See Examples 1 and 3). Long (1983a) defines confirmation checks as expressions "immediately following an utterance by the interlocutor which are designed to elicit confirmation that the utterance has been correctly heard or understood by the speaker" (Long, 1983a: 137). Confirmation checks are often characterized by repetition or rephrasing, with rising intonation of all or part of the other speaker's preceding utterance. Confirmation checks may also check the other's intended (implied) meaning or underlying presuppositions (Pica, 1988: 53). In the latter case, the confirmation check is an hypothesis rather than a repetition or rephrasing. Confirmation checks are answerable by a confirmation if the utterance is correctly understood. Confirmation checks overlap with what Schegloff, Jefferson 8c Sacks (1977: 378) call other-correction. However, the term cotrection implies a judgement which is not necessarily intended. In unclear cases, decisions on whether an utterance is a trouble indicator or a confirmation check were often based on the next turn. If the next turn was in the affirmative (`yes') or the negative (`no'), it was coded as a confirmation check. If it was in the negative, there might be a trouble clarification following along with or instead of a simple negation. Then the answerability with a`yes' answer was taken to signal the earlier utterance was a confirmation check. If no `yes' answer could follow, it was coded as a trouble indicator. Trouble indicator (- ti) A trouble indicator is defined as a move which signals to the other speaker that there is a problem in understanding. This phenomenon is similar to what Long (1983: 137) calls NS clarification requests: "any expression designed to elicit clarification of the interlocutor's preceding utterance(s)". The first speaker is implicitly or explicitly requested to clarify the original trouble source (See example 1). Schegloff, Jefferson 8r Sacks (1977) call this other-initiation of repair. By using a trouble indicator a speaker may request an explanation about the meaning and intention of a preceding turn with questions such as`mhm?', `what do you mean by X?' or declaratives such as`I don't understand'. Allwood (1993) 4.1 The Negotiation Moves 59 sees such indicators as forms of feedback and calls them feedback elicitors. In some cases, a speaker may attempt to repeat (part of) a word in such a way that it is clear it is not understood. Then it is not a check. Such keyword repetition was also classified as trouble indicator. It is followed by a clarification rather than a confirmation. Sometimes a speaker does not react verbally at all at a transition relevant place. Then there is a selected speaker's significant silence which may function as a trouble indicator ( Levinson, 1983: 299). Schegloff, Jefferson 8c Sacks (1977: 374) state that often at such a point there is a slight gap to provide extra room for selfrepair. One could, in fact, reason two ways. After such a silence, the other speaker self-initiates in the transition space, or the silence is interpreted as an other-initiation. This last interpretation is most logical if nonverbal signals by the other speaker accompany such silence. These may be facial expressions or body and hand movements. Conversation analysts prefer to work with telephone data in which nonverbals signals do not play a role. However, for the interpretation of silence, these signals may be crucial. The interpretation whether the silence was a trouble indicator depended on such nonverbal signals and the other speaker's next turn. If the other speaker reacted with a clarification, the silence was coded as a trouble indicator. Trouble indicators often form an adjacency pair with trouble clarifications, as in Example 1. In the institutional conversations there was another type of pause which was not a trouble indicator. This type occurred, for instance, when the housing official was writing on the registration form. These pauses were not coded ( see Example 5). (5) Iastitutional conversation with Mahmut, Cycle 2 ac gc Bart: [weel twol of tweehonderdvijfenzestig ja or two hundred sixty-five yes Bart: writes~ ~writes ~ Malunut: Bart: Mahmut t ~pau.te~ ja ja die gas he .... t ~pau.ce~ yes yes that gas huh ... Feedback signals which do not elicit, but give, feedback can have several functions in clarification sequences. The next three moves, confirmation, acknowledgement and gratuitous concurrence can all be realized with a minimal feedback signal but they operate at different places in the sequence. Confirmation (- co) A confirmation check is generally followed by a second part, an affirmative feedback signal (`yes') or sometimes a negative ('no') with or without partial repetition of content words found in the confirmation check (See example 1). These feedback signals will be called confirmations. Gratuitous concurrence (- gc) A frequently occurring category, which is mainly used to stimulate the other to continue, is called gratuitous concurrence (Liberman, 1984). This category is 60 Chapter 4 The Organimtion of C{arifcation Sequences generally realized by the feedback signal ja (yes). It does not necessarily indicate agreement or even understanding but just displays attention. In terms of Clark 8c Schaefer (1989) this would be the lowest level of acceptance. Nofsinger (1991) calls similar items continuers because they signal acceptance that the other speaker is the main producer of substantive turns. Gratuitous concurrence moves are, for example, used when one speaker is clarifying something. He breaks up the information into small pieces. This strategy is called decomposition (Long, 1983a: 135136). As can be seen in Example 6, after each piece there is a pause in which the other speaker shows acceptance with a`yes'. (6) Informal conversation with Mahmut, Cycle 3 [s Janet-. ben je lid van een vakbond? ti Mahmut: wat? tc lanet: ben je lid van een vakbond? ti Mahmut vakbond cimitates~ wat is dat? cc vakbond ~imitates~ co Janer. ja gc Mahmut: ja tc Janet [t3t] dat is tt als uh t zeg maar iedereen die werkt gc Mahmul: ja tc Janet: en f dan betaal je bepaald geld per maand paar gulden gc Mahmut: ja tc Janet: gc tc Mahmut: Janet: gc cm te cm Mahmut lanet en dan zijn (er) die mensen t die zorgen dan voor dat jullie allemaal hetzelfde geld krijgen t dat je zoveel vakantiedagen ~JBt ja dat je niet zomaaz ontslagen kan worden ja hè? en dat soort dingen t vakbond hè? are you a member of a union? what? are you a member of a union? union ~imitates~ what is that? union ~imitates~ yes ye.r [t3t] that is tt if uh t say everyone who works yes and t then you pay certain money per month few guilders yes and then (there) are people t who then make sure for that you all get the same money t that you get a certain number of vacation days yes that you can't just be fired ye.r you see? and that sort of Ihing t union you see? Acknowledgement (- ac) As a concluding move of a clarification sequence often feedback signals appear. They function as an acknowledgement with which a speaker expresses that the trouble is sufiiciently resolved to move back to the main line (Jefferson, 1972; Houtkoop, 1987; Stalpers, 1993: 64). Usually such acknowledgements have the fotm of affirmative feedback signals such as ja (yes) or goed (okay). Houtkoop (1987) mentions many funcuons of acknowledgements in interaction: appreciation, affirmation, Oh-receipt, and sequence closing. In clarification sequences the last function of sequence closing might be most prominent. The speaker who 4.1 The Negotiation Moves 61 started the negotiation of ineaning by indicating trouble or checking may explicitly signal the closure of the insertion sequence and the contínuation of the main line with an acknowledgement. Acknowledgements can often be considered a third part to an adjacency pair (see Example 1). Acknowledgements also occur at other places in a sequence. They then signal strong acceptance of what the other speaker is explaining. Often a double jaja or oja is used (as can be seen in Exatnple 7). Allwood found in his analyses of ESF NNS data that all the participating language groups except for speakers of agglutinative source languages such as Turkish appear to use reduplication (Allwood, 1993: 221). Turkish speakers appear to use repetitions of part of the other speaker's utterance rather than reduplicated feedback signals. An acknowledgement is a stronger form of evidence of understanding than gratuitous concurrence. Example 7 shows the occurrence of midsequence and closing acknowledgements. (7) Informal conversation with Ergun, Cycle 3 Ergun: zeven uud[zeven] lanet: Ergun: Le cc co tc lane[: Ergiin: Janet: ac Ic ac tc Ergiin: Janet Ergun: ac Ergiin: lanet [wij hebben er twee] lzeven uur begmnen of acht uur we hebben er twee twee? ja skychannel t dat is dus een commerciele zender en wij hebben een zender uit londen ~biebiecie~ - ~meaning b.b.c.~ oja ~biebíecie~twee ~biehiecie~twee jaja dus net zoals nederland twee zeg maar ja seven o'clockl[seven] [we have two] Jseven o'clock begin or eight o'clock we have two two? yes skychannel that's a commercial station and we have a station from london bbc rit;ht bbC two óbc two yeah yeah sojust like the Netherlands two say veah The next two moves can occur at different places in the sequences and refer to the problematic character of the interaction. These are comprehension checks and me[a comments. Comprehension check (- em) Pica ~ Doughty phrase Long's (1980) category of NS comprehension checks in the following way: "expressions designed to establish whether the speaker's own preceding utterance has been understood by the addressee" (1988: 48). They usually have the form of questions such as hè? (`you see?') or begrijpje? (`do you understand?'). Example 6 ends with two such comprehension checks. 62 Chapler 4 The Orgnni~ation of Clnrification Sequences Pica 8c Doughty (1988) also added repetitions with rising intonation of all or part of the utterance to the category of comprehension checks. However, then the category overlaps with confirmation checks which are also defined by them as containing repetitions and rising intonation. In the current study, those moves were all coded separately as confirmation checks, because the repetition of content words often indicates a hypothesis of the former speaker's utterance which was not sufficiently understood. Comprehension checks, on the other hand, are used, for example, when a speaker stays silent at a point where his interlocutor wants to yield the turn. They may refer to language problems as well as problems concerning socio-cultural knowledge. In the case of a perceived language problem often verbs such as `understand' and `mean' are used: `do you understand what I mean?'. In the case of a perceived socio-cultural knowledge problem, verbs such as 'to know' and 'to hear of' are used: `do you know?' or `have you heard of this?'. Meta comment (- mc) When the conversation appears to breakdown, a speaker may try to get back on track by explicitly referring to the problematic character of the conversation and encouraging the interlocutor to continue in spite of the problems. Such moves are called meta comments. In Example 8, two such meta comments occurred halfway into a very long clarification sequence. (8) Institutional rnnversation with Fatima, Cycle 2 mc Bart: maerlrjk hè? di~cult, isn't it? co Fatima: daughs~ daughs~ mc Bart ja toch proberen yes try anyway 4.2 NS and NNS Clarif;cation Sequences The overall picture of the analysis showed the appearance of several adjacency pairs. The main adjacency pairs that clarification sequences consist of are: trouble indicator-trouble clarification, confirmation check-confirmation, and trouble source-main linelthematic reaction. The above described moves can now be put into a diagram (See Figure 4.2). A clarification sequence starts with a main line turn of a speaker A which is perceived as a trouble source by speaker B. If speaker A did not self-initiate a repair, speaker B will do this. He can do this with a trouble indicator or a confirmation check. This initiation will be followed by a completion which will either clarify the trouble or confirm the correctness of the confirmation check. Thus, two adjacency pairs form the heart of the negotiation process. If the trouble is resolved, this may be explicitly acknowledged by speaker B andlor a main line response may end the sequence. 4.2 NS and NNS Clarification Seguences 63 Speaker A main line tum (ts) other-initiation of repair (ti, cc) clarificazion (tc) confirmation (co) main line turn (mr) acknowledgement of resolution (ac) Figure 4.2: Prototypical representatio~t of the negotiation of ineaning process This diagram represents a problem smoothly resolved through the negotiation of meaning. However, in many cases solving the problem is not done so quickly. Instead of one clarification, there may be many. Speaker B may then use gratuitous concurrence to stimulate speaker A to continue with the clarification. Clarifications may also follow a thematic response rather than a trouble indicator. This thematic response functions as a symptom of non-understanding and thus triggers a repair. Both speakers may also check each other's comprehension at any time in the negotiation process. They may metacommunicate to prevent breakdown or they may switch topics to escape the trouble altogether. All these moves, which do not occur in the prototypical representation of Figure 4.2, may occur incidentally. Differences in NS and NNS communicative competence may influence the organization of sequences, depending on whether it is the NS or the NNS who causes the trouble. Faerch 8z Kasper (1985: 15) state that in asymmetric interaction, such as between NS and NNS it is important to distinguish NNS from NS trouble sources and repairs. They thus distinguished eight types of repair, rather than the four combinations of selflother-initiations and repair mentioned above. They investigated the frequencies of NS and NNS trouble sources, repair-initiation and completions in interaction between NSs of English and Danish óth and lOth grade learners and Danish college students of English. In the current study, the clarification sequences are divided into two groups. The sequences starting with a NS trouble source are called NS sequences. The sequences in which the NNS caused the trouble are called NNS sequences. Faerch 8c Kasper (1985) showed that there was different preferences of organization in repair, depending on whether it was the NS or the NNS who caused trouble. Preference for self-repair gave way to NS repair. The NS's role is to self-repair (in particular with low proficiency NNSs) and to initiate other-repairs. The NNSs 64 Chapter 4 The Organization of Clarification Sequences were found to be more passive. They often relied on the NS to resolve the trouble they caused. Face protection appeared to reduce the NNS's willingness to initiate other-repair of their own and self-repair of the NS's utterances. In Tables 4.1 and 4.2, the numbers of NS and NNS sequences per interaction type, cycle and NNS infotmant are presented to give an indication of their relative frequency of occurrence. Cycle 2 Cycle 1 ~ .v ~ s ~ t ~v ~ c L ~ ~ .~ ~ o F E ' ~ E ï ~ tit ri ~ ~ rn` ri ~ i w NS clarification sequences 6 14 8 8 4 21 8 11 6 20 8 12 136 NNS clarification sequences 10 1 9 7 11 4 1I 3 6 6 12 4 84 Total clarification sequences 16 15 17 15 IS 25 19 14 12 26 20 16 210 r~ ~ ~ ~ É r Informant 5 ~ .~ I `~ Cycle 3 Table 4. I: Number of NS and NNS clarification sequences in the informal conversations Cycle 2 Cycle I ~ b ~ É ~ É r ~ ~ ~ '~ w ' ri 8 15 11 20 4 4 15 Total clarification sequences 12 19 26 Informant -~ ti NS clarification sequences NNS clarification sequences ~ ~É É Cycle 3 ~ ~ ~ ~É ~ ~ ~ F s ~ ~ ~ `~ w ri s ~ ~ ~ -~ w 15 14 17 20 19 9 11 I8 177 9 4 1 19 9 10 2 15 8 100 29 19 15 36 29 29 11 26 26 277 Table 4.2: Number of NS and NNS clarification sequences in the institutional conversations The number of turns per conversation varies considerably. Therefore, no direct quantitative comparisons can be made on the absolute number of NS and NNS sequences. Only relative frequency of NS versus NNS sequences will be discussed. As can be seen from Tables 4.1 and 4.2 in both the infotrrtal and institutional conversations, there are more NS than NNS sequences indicating that more NS than NNS trouble sources are negotiated. These findings agree with those of Faerch 8z Kasper (1985), who also found more NS trouble sources which led to repair with beginning learners. These beginning learners are most comparable to the ESF informants. However they found more NNS trouble sources with the intermediate and advanced learners. In particular, in the informal and institutional conversations with Mohamed and Ergun there are more NS sequences. By indicat- 4.2 NS and NNS Clarification Sequences 65 ing trouble, they show their motivation to achieve understanding. Thus they may receive more comprehensible input which may increase opportunities for language acquisition (Pica, 1994). With the slower learners Fatima and Mahmut there are different patterns. Consistent with the pattern of Mohamed and Ergiin, in the institutional conversations with Fatima there are more NS sequences. However in the informal conversations, there aze proportionally more NNS sequences in the first and second cycles showing that Fatima caused most problems in understanding which are negotiated. In other words, Janet initiated more clarification sequences asking Fatima for clarification than the other way around. In the third cycle, the number of NNS sequences equals the number of NS sequences. In the conversations with Mahmut, there are more NNS sequences in both conversation types. The NSs more often found it necessary to indicate trouble or check than the other way around. Faerch 8L Kasper (1985) found this to be typical for more advanced learners. This is doubtful, because the same can be said for Mahmut even though he was one of the slow learners. A more likely explanation is that Mahmut took more initiative than the other NNSs. Therefore, the NSs had more trouble understanding him. When a NNS replies to a question, the NS has expectations about the content of the NNS's turn. When a NNS initiates a new topic the NS may have fewer expectations. Mahmut's communicative style will be further discussed in Chapter 6. 4.3 Some Aspects of Sequential Organization Differing roles and communicative competence may lead to a different organization of moves in clarification sequences in the two interaction types. Faerch 8r. Kasper (1985) assume different preferences in repair types of NSs and NNSs in interaction with each other. The sequences starting with a NS trouble source will be compared with those starting with a NNS trouble source. It is expected that in the sequences starting with a NS trouble source, the NNS will indicate trouble. This can be done with minimal linguistic means (feedback signals). The NS will manipulate content in clarifying the trouble. The NS may want to relieve the NNS of doing such linguistically demanding clariFcation work as far as possible. Therefore, in NNS sequences, the NS may opt for using confirmation checks after a trouble source occurs, rather than indicating trouble. To a confirmation check a NNS can react with minimal means (`yes') which would lead to a different sequential organization. The actual organization can be determined on the basis of the second turn in a clarification sequence. The second turns were calculated per type and the frequencies are presented in Table 4.3. The figures in Table 4.3 show the frequencies of trouble indicators and confirmation checks in the second turns. These frequencies diverge somewhat from the total 66 Chapter 4 Tlte Organization of Clarification Sequences Conversation type Informal conversation Institutional conversa[ion trouble indicator confirmation check other trouble indicator confirmation check other NNS 98 23 7 119 35 28 NS 20 52 17 37 63 8 Type of move Table 4.3: Number of trouble indicators and confirmation checks as a second turn number of clarification sequences mentioned in Table 4.1 and 4.2 because more than one move could be realized within a second turn. In the NS sequences, the NNS usually reacted in the second turn with a trouble indicator regardless of the conversation type. This can be seen in Figure 4.3. trouble source trouble indicator trouble clarification main line reaction Figure 4.3: Model of a native speaker sequence In Example 9, Ergun told Janet that there would be a new television channel in the town where he was living. Janet asked whether that day was the first broadcasting day. (9) Informal conversation with Ergiin, Cycle 3 ts Ergun: ]anet: daar komt vandaag die franse~ `voor het eerst? there the French one will corne today` `for the first time? ti tc mr Ergiin: Janet Ergun: ac Janet hè? is dat voor het eerst? ja eerste gisteren beginnen gisterenavond (xx) jaja huh? is that for the first time? yes first yesterday begin yesterday evening (xx) yeah yeah In contrast, the NS as expected usually reacted in the NNS sequences with a confirmation check rather than a trouble indicator. This confirms the results of Faerch á Kasper (1985: 18), who also found after a NNS trouble source, there is a clear preference for other-cotrection (which is comparable to confirmation checks). 4.3 Some Aspects of Sequentia[ Organization 67 As can be seen in Figure 4.4 the prototype of a NNS sequence looks different. trouble source confirmation check confirmation ~ acknowledgement Figure 4.4: Model of a nonnative speaker sequence In Example 10, Mohamed wanted to know how many municipalities the town of Tilburg has and he tried to explain that Casablanca is built up out of forty-five districts and five municipalities. The Dutch municipal structure is very different from the Moroccan one. The town Tilburg belongs to one greater municipality. Therefore, Peter checked whether he correctly understood what Mohamed was relating. (10) Informal conversation with Mohamed, Cycle 3 mhm we in morocco I think ts Mohamed: hm wij in mazokko ik denk lin casablanca huh t tin casablanca hè t ik detilc nou t 1 think now f forty-five or t municipality vijfenveertig of t gemeent cc Peter: in casablanca? in casablanca? co Mohamed: ja yes Differences in communicative competence or roles may lead to these different patterns. By using confirmation checks instead of trouble clarification, NSs present models for the meaning the NNSs intend to convey and thus relieve them of the sometimes linguistically demanding work of clarification (see also Example 1). For the NNS a`yes' response often suffices. The two Turkish informants, however, also reacted to confirmation checks with partial repetition of the check, often in combination with a simple `yes' answer (see also Allwood, 1993). Therefore, in both NS and NNS sequences, the NS typically manipulates the content and the NNS uses feedback signals. There aze some other moves that were used in the second position too (see category `other' in Table 4.3). These were, in the majority of cases, trouble clarifications which followed a combined NNS trouble source and trouble indicator (see Example 11). (11) Informal conversation with Fatima, Cycle 2 ts 68 Fatima: (kenirra niet] rabat alles met uh Ciuzpter 4 Ikenitra notJ rabat all with uh The Organi~ation of Clarification Sequences ti tc cc mr 7anet Fatima: niet politie ander ~casken - ~-m-soldier~ leget dege~ - ~imitates~ moet uh kijk maar politie ook met die mens van ruzie maak not police other ~casken - ~-m-soldier~ army - ~imitates~ has to uh look but police also with the people like quarrel make Sometimes, the second turns were thematic or inappropriate main line responses which led the first speaker to clarify or check in the third turn. Although several examples of the prototype NS and NNS sequences can be found in the data, most sequences are longer than the ones above, averaging between 5 and 10 turns. If at any point one of the speakers is not satisfied with the resolution of the problem, pairs of negotiation moves may be repeated as often as desired: e.g. the pair trouble indicator-trouble clarification, the pair trouble clarification-gratuitous concurrence, andlor the pair confirmation check-confirmation. About a third of all the sequences in both conversation types are between 10 and 30 turns. Only 7 sequences out of the almost 500 total number are longer than 30 turns, with one extremely long sequence of 67 turns in which a complex housing question is segmented into several subquestions which led to almost complete resolution of the trouble (see Chapter 5, Example 9). Five of the 71ong sequences occurred in the institutional conversations. Such long sequences generally occurred when the negotiation of ineaning did not clarify the trouble, but created even more misunderstanding. The special jargon used by the housing official and the complicated social housing system led to such misunderstandings in some cases. 4.4 Conclusions In this chapter, the taxonomy of twelve nego[iation of ineaning moves found in the NS-NNS conversations was described. These moves can be differentiated into main line moves such as the trouble source and side line moves. Two side line moves manipulate the content of the trouble source in order to solve the trouble. These are trouble clarifications and confirmation checks. Several feedback moves were found, such as confirmations and acknowledgements. In addition, some side line moves relate to the problematic character of the interaction, such as comprehension checks. The number of different moves found was much larger than the number of moves described in the Varonis and Gass model for interaction between NNSs (1985a). The presented taxonomy appears to describe more effectively NSNNS interaction and differentiates between negotiation behaviors which manipulate the content and feedback signals. It is not entirely clear how generalizable the moves found are to other types of interaction. However, the moves overlap with categories described in other studies on NS-NNS interaction (Long, 1983a; Pica 8z Doughty, 1988) and NS-NS interaction (Schegloff, Jefferson 8c Sacks, 1977) in 4.4 Conclusions 69 which English is the language of communication. Therefore, some universality may be attributed to these moves. However, studies on other languages from different language families should confirm whether this is actually the case. The clarification sequences were divided into two types, the ones starting with a NS trouble source and the ones started by the NNS. The analyses show that NS sequences occur more frequently. This way the NNSs may receive more comprehensible input and thus create opportunities for language acquisition. With Mahmut, there are more NNS sequences, which may have to do with his communicative style. Both NS and NNS sequences aze generally built up from adjacency pairs: trouble indicator-trouble clarification, confirmation-check confirmation, trouble source-main line response. Often an additional third member is used to close the sequence: the acknowledgement. However, the organization of NS sequences differs from that of the NNS sequences. The NS sequences most frequently contain a NNS trouble indicator as a second turn. The NNS sequences, on the other hand, more frequently contain confirmation checks from the NS as a second turn. This indicates that the NS most frequently manipulates the content of the trouble source either by clazifying his own utterance or by presenting a model for the NNS's intended meaning, thus relieving the NNS from the linguistically most demanding work of clazification. 70 Chapter 4 The Organization of Clarification Sequences Chapter 5 The Amount of Negotiation of Meaning 5.0 Introduction In Chapter 4, we saw that clarification sequences in NS-NNS interaction may consist of a number of different moves and may have different organizations depending on whether the NS or the NNS causes the trouble. In Chapter 3, the following research questions were formulated concerning the amount of conversation time spent on negotiating meaning. 2a. How much of the interaction is spent negotiating meaning? 2b. How do NNS communicative competence and interaction type influence the amount of interaction spent on negotiating meaning? A high amount of negotiation may, on the one hand, create opportunities for language learning, but, on the other hand it may also influence communication effectiveness negatively. In section 5.1 some expectations will be formulated on the influence of NNS communicative competence and interaction type. T'he results follow in section 5.2. There appears to be an interaction between NNS communicative competence and interaction type. Therefore, the influence of NNS communicative competence will not be discussed separately but will be discussed in section 5.2.1 where the focus is on the informal conversations and in section 5.2.2 (the institutional conversations). Conclusions follow in section 5.3. 5.1 The Influence of NNS Communicative Competence and Interaction Type As we saw in Chapter 2, results from studies on repair have shown a preference for self-initiated self-repair. Mazeland (1992: 140) analyzed a form of institutíonal NS-NS interaction, that is, interviews with open-ended questions which were held for some social science research projects. He found that one third of the questionanswer sequences analyzed contained some form of modification, repair or completiorJsupplement. Most of those were self-initiated self-repairs. Schegloff, 5.0 Introduction 71 Jefferson 8t Sacks (1977) claim that other-initiated repair, which results in clarification sequences, is not preferred and will thus generally occur much less frequently than self-initiated repair. They also noted that other-repair (othercorrection) may occur more frequently in interaction with learners. It then is "a device for dealing with those who are still learning or being taught to operate with a system which requires, for its routine operation, that they be adequate self-monitors and self-correctors as a condition of competence" (Schegloff, Jefferson 8c Sacks, 1977: 183). In other words, NNSs may not be able to clarify their own trouble sources and thus other-initiation may occur. In addition, the NS may not be able to adjust hisltter language behavior to the NNS's level of communicative competence beforehand. Therefore, in NS-NNS interaction with NNSs with limited proficiency a great deal of negotiation of ineaning may be expected, the type of repair two speakers engage in together. In the current study self-initiated self-repair which is done by only one speaker was excluded from the analysis. The amount of negotation of ineaning in NS-NNS interaction may be influenced by several factors. When the communicative competence of the NNSs increases, it would seem obvious that fewer problems would arise, and that these problems could be resolved more quickly. A logical expectation, then, would be that the percentage of turns spent negotiating meaning decreases over time. Another possible outcome is that the amount of clarification sequences in NSNNS interaction remains stable over time. In the case of NNSs with limited L2 proficiency, problems in understanding may continue to occur, but, the time spent on negotiation may be limited because it is not a preferred method of repair (Schegloff, Jefferson 8z Sacks, 1977). Self-initiated self-repairs appear to be much more successful than other-initiated repairs, because they resolve the trouble usually within the turn in which they are initiated (Schegloff, Jefferson 8c Sacks, 1977: 369). Other-initiated repairs, as seen in Chapter 4, result in side-sequences which interrupt the main line of the conversation. Too much negotiation of ineaning may lead to a conversational breakdown and loss of face. Only a certain number of trouble sources may be negotiated because the potential of solving problems through negotiation is limited as clarifications cause new problems. Therefore, the participants may initially limit the amount of negotiation and be satisfied with a minimal mutual understanding or even with ihe illusion of undersianding. In later stages, problems with increasing levels of difficulty may be negotiated. Thus the amount of negotiation stays constant or increases in spite of the fact that the degree of mutual understanding is increasing too. There may be a certain optimal balance between the level of negotiation and the level of understanding. It may be that only when the NNS proficiency has reached a near-native level a decrease in negotiation will be found. Faerch 8c Kasper (1985: 19) found that learners at the intermediate level (grade 10) more frequently initiated negotation of ineaning than lower-level learners because there were considerable perception and production problems, but also because they were more able to make use of the NS's help. Another factor which may influence the amount of negotiation is the interaction type. The institutional conversations may differ from informal conversations because of the prescribed roles of housing official and client, and the presence of a 72 Chapter S The Amount of Negotiation of Meaning housing office script. However, it is not clear beforehand whether these characteristics will increase or decrease the amount of negotiation. On the one hand, the structure of institutional conversations and its goal-directedness may limit the room for side-tracking from the main line of the conversation. In addition, the asymmetric roles of an official and a client may create a need for face-saving which may also impede repair activity. Pica (19ó7) quotes a few studies which found a relative absence of negotiation of ineaning in institutional interaction in the classroom. She tries to explain this outcome by stating that in the classroom, there is hardly any room for negotiation of ineaning because the interaction is structured and pitched to the learner's level of competence. In addition, if a NNS student were to initiate a negotiation of ineaning sequence, this might be felt as a challenge to the teacher's authority. The institutional role of the teacher and the subordinate roles of the students may restrict opportunities to negotiate meaning. In the cutrent study, the interaction may not be pitched as much to the NNS's level but some inhibition to negotiate meaning may certainly be felt. In informal conversations, participants may take more time and feel freer to come to mutual understanding through negotiation of ineaning. Thus, one might expect more negotiation in the informal conversations than in the institutional conversations. On the other hand, the structure and goal-directedness of institutional conversations also forces the particípants to deal with certain topics and transfer information. As described in Chapter 3, in the institutional conversations, the housing official needs a great deal of client information to meet the client's desires within the limitations of housing supply. In addition, the play-acting scene was set up in such a way that the NNS participants had to attempt to convince the NS of the `urgency' of their situation. This put pressure on the NNSs to make themselves understood. In the informal conversations, there was a guiding topic, `cultural differences'. However, as is normal in conversation, the participants side-tracked to other topics. Thus, they could more easily opt out of a difficult topic or avoid trouble by initiating a new topic. As a result, there may be more negotiation in the institutional conversations than in the informal conversations. 5.2 Results The amount of interaction that is spent negotiating meaning was calculated on the basis of turns. The number of clarification sequence turns was divided by the total number of turns. Percentage scores were calculated to determine the relative share of negotiation of ineaning turns to the total number of turns. These percentage scores ace presented per cycle in Table 5.1 for both types of conversations (see also Appendix 3, Table 1 and 2 for the raw scores). The results are presented for every NNS informant separately. 5.2 Results 73 Informant Fatima Mohamed Mahmut Ergun Cycle 1 2 3 1 2 3 1 2 3 1 2 3 Informal conversation 42 30 48 54 60 50 43 27 49 71 47 55 [nsti[utional conversation ~11 ~ FS ~ ~~ 3~ ~`' ~~ SR ~~ ~~ h' ' ~F~ - ---1 ---- --- - - L-- -- --- - -- -- l-t ~ - Table 5.1: Percentage of clarification sequence turns in the informal and institutional conversations Table 5.1 shows that in the analyzed NS-NNS conversations, generally, a large share of the turns are part of a clarification sequence. However, there is a great deal of variation because the lowest score is 22qo and the highest 71 qo. As Figure 5.1 depicts, there is no consistent pattern of decrease. With Ergiin there is a decrease in the second cycle, but a small increase again in the third cycle. With Fatima and Mahmut there is also a decrease in the second cycle with an increase to an even higher level in cycle 3. In the conversations with Mohamed, a constant percentage of SOqo or more turns are part of a clarification sequence, with the highest scores in the second cycle. In general, there seems to be a movement towards a high average of SOqo for all informants. l00 , Fatima 90 ti, - Mohamed ~~ Mahmut 80 ,' ~ -M...,~,. Ergun 70 60 ~ 50 40 -~ 30 -~ 20 -, 10 -~ 0 1 2 3 cycle Figure 5.1: Percentage ojclariftcation sequence turns in the informal conversations 74 Chapter 5 The Amount of Negotiation of Meaning 100 Fatima 90 ~ - Mohamed SO ~ ~ Mahmut ~,....m.,..~ Ergun 70 ~ 50 ~ 40 30 20 10 -I 0 1 2 3 cycle FiRure 5.2: Percentage ofclarification sequence turns in the institutional conversations The average first cycle scores are more or less comparable over both interaction types. As can be seen in Figure 5.2, in the third cycle in the institutional conversations, however, the percentage of clarification sequence turns is much lower than in the informal conversations for the three male informants Mohamed, Mahmut and Ergun. With Mohamed, from the beginning, fewer turns were spent on clarification but there too a decrease of more than lOolo was found. With Fatima, the percentage of clarification sequence turns is around 45qo in the first and third cycle, with a high of 65qo in second cycle. Therefore, there appears to be a firm decrease for most of the NNS informants in the institutional but not in the informal conversations. Apparently the factors NNS communicative competence and interaction type do not operate independently but interact with each other. However, there is also a great deal of individual variation, which can only be explained by other factors, such as personality, gender and culture. Because interaction type and NNS communicative competence interact with each other, the results will not be presented and discussed separately. In section 5.2.1, the influence of the informal conversations, NNS communicative competence and other factors will be discussed. In section 5.2.2 the results for the institutional conversations in combination with the other factors follow. 5.2 Results 75 5.2.1 The Informal Conversations As can be seen from Table 5.1 and Figure 5.1, there is no one consístent pattern of development in the informal conversations. The four NNS informants show different patterns but none shows a downward line over the three cycles. The conversations with three of the informants (except Mohamed) show a decrease in cycle 2 followed by an increase in cycle 3. Thus, it might be the case ihat in the informal conversations, problems in the first cycle were not negotiated as much as in the later cycles. In other words, the participants were satisfied with minimal mutual understanding or pretended understanding, which happens in many conversations which are not goal-directed. There was more negotiation with the fast learners Mohamed and Ergun than with the slower learners Fatima and Mahmut. As seen in Chapter 4, this is partly due to their own initiative because they both initiated more clarification sequences than Fatima and Mahmut. This confirms findings by Faerch 8c Kasper (]985: 19) that learners at the intermediate level in their study more frequently initiated negotation of ineaning than lower-level learners because they were able to make better use of the NS's help than [he beginning learners. In this way, NNSs create opportunities for language learning for themselves. In particular, because the conversations were recorded in an experimental setting, the obligation to keep the conversation going was probably stronger than in other settings, even if there was a great dea] of misunderstanding. This may have influenced the outcome also. The NS Janet did appear to despair at times in her first conversation with Mohamed. He was aware of his limited proficiency in Dutch and protected his face by speaking as little as possible. This happened, for instances, when she started with the topic of cultural differences between the Netherlands and Morocco surrounding `giving presents'. In Example 1, Janet tried to explain to Mohamed that she was really surprised that a Moroccan family in Morocco did not open her present when she gave it to them but merely thanked her. She initially thought they did not like receiving the present. She tried to find out whether Mohamed knew that there was a cultural difference in present giving and how he dealt with it. Mohamed appeared to understand some of it but he reacted only minimally or turned to a Moroccan researcher who was present for translation. (1) Informal conversation with Mohamed, Cycle 1 ts Janet in het begin dacht ik t ik dce iets fout of uh f heb jij dat ook? ti Mohamed: hè? tc es ]aner. Moharned: in the beginning I thought t I do something wrong or uh t do you have that too? huh? heb jij dat ook t gehad? have you had that t too? t ~'..`~ t~'..'~ - ~in Moroccan Arabic to Moroccan researchen After some more unsuccessful attempts by Janet to stimulate Mohamed to talk, Mohamed commented that he understood what she wanted from him but that he was not able to express himself. This can be seen in Example 2. 76 Chapter 5 The Amount of Negotiation of Meaning (2) Informal conversation with Mohamed, Cycle I lanet: uh je weet niks geen verschil t tussen nededand en marokko? Mohamed: ja ik weet maar kan uh ]anet ~kan niets zeggen~ ~completes~ Mohamed: mceilijk voor mij lanet: ja het is ook mceilijk uh you don't know any differences t between the netherlands and morocco? yes I know but can uh ccan say nothing~ ~completes~ difficult for me yes it is difficult At a certain point after having tried several other areas of differences, Janet commented that this topic was not working. This whole conversation was filled with clarification sequences, which did not have the desired effect of stimulating Mohamed to talk. In a natural situation, a NS's motivation to negotiate meaning and continue the conversation might be have been less. Another explanation of the relative constancy of negotiation of ineaning may be that, because in infotmal conversation topic initiation is less controlled, new problems kept occurring. In the first cycle of the informal conversations, Janet controlled the topic quite definitely by continuously guiding it back to cultural differences between Morocco or Turkey, and the Netherlands, respectively. With Mahmut, in contrast to the conversation with Mohamed, Janet spent the whole conversation exploring the topic of giving presents. As can be seen in Example 3, they negotiated meaning frequently, starting with the introduction of the topic. (3) Infotmal convetsation with Mahmut, Cycle 1 ts Janer en als je kadootjes hebt [t] ec co ac Mahmut: lanet: Mahmut: Janer. MahmuC Janet: [~kado~] ~een kado (x) ja kado en je gaat naar een vriend? ja in nederlandl weet je hce ze het in nederland dcen? uh kado ja t ts cc co ti tc gc tc gc tc gc tc tc tc mr Mahmut: Janet: Mahmut: lanet: Mahmut: Janet: Mahmut: ]aneC Mahmut: lanet: Mahmut: ik kom bij jou [ja] [nee] jij komt bij mij ja jij geeft een kadotje ja dan maak ikl dan dce ik zol maak ik meteen open nee turkse mensen nee die kadootjel jij uh mij kadootje halen 5.2 Results and if you have ~presents~[t] ~dimunitive forrru [dcado~] - vepeats~ la present (x) yes present and you go to a friend? yes in the netherlandsl do you know how they do it in the netherlands? uh present yes t 1 cotne to you [yes] [no] you come to me yes you give a presen[ yes then 1 makel then I do like thisl make it open immediately no turkish people no that presentl you uh me present ~geb 77 Mahmut: lanet-. Mahmut 1aneC Mahmut: Janet: Mahmut ]anet. Mahmut: lanet aneaning geven~ hè [t] [ja] l~die~ die hè - ~points to cup~ ja ~meaning give~ huh [t] [yes] en dan kadootje (x) kijken [t] [ja] and then watch (x) present [t] [yes] Iturkse mensen niet kadootje hierl zij hier zitten ik hier" niet kadootje hier" niet kijken niet kijken t waarom niet? nee mag niet mag niet Iturkish people not present herel they here sit 1 here" not present here" not watch not watch t zij ~weg"~ - r-P-wEb l~that~ that huh - ~points to cup~ yes they gone why not? no not allowed not allowed It took several clarification sequences before Janet had explained her experiences in Morocco. Mahmut then tried to explain that in Turkey it is impolite to open a present immediately. They would open the present later. Opening a present immediately may embatrass the giver, if it happened to be a small inexpensive present. In the last part of the conversation Janet and Mahmut discussed the kind of presents appropriate for a Turkish wedding. In contrast to Mohamed, Mahmut did not let his minimal competence hinder him. As can be seen in Example 4, he frequently requested Janet's help if he did not know the words to express his intended meaning. (4) Informal conversation with Mahmut, Cycle 1 tslti Mahmut die ~zo~ - ~gestures of necklace~ tc Janet jaja ketting co Mahmut: t ~ketting~ - ~imitates~ Mahmut: die ~deze~ - ~points~ tc Janet: oorbellen cc Mahrnut ~oorbellen?~ - ~imitates~ tc Janet: oor [oorJ cc MahmuC [oor] oor [c Janet bellen cc Mahmut: bellen tc lanet oorbel bellen cc Mahmut: oorbel (...) that ~like this~ - ~gestures of necklace~ yeah yeah necklace gt ~necklace~ - ~imitates~ that vhis~ - ~points~ eamngs ~earrings?~ - ~imitates~ ear [earJ [ear] ear rings rings earrings rings earring (...) In the second and third cycle, with all NNS informants the topics became more diverse, because the infotmants appeared to be able to initiate topics themselves. Differences in culture still came into the discussion, but personal topics such as work and family were also favored. In the second as well as the third cycle with 78 Chapter S The Amoatnt of Negotiation of Meaning (2) Informal convetsation with Mohamed, Cycle 1 Janet: uh je weet niks geen verschil t tussen nederland en marokko? Mohamed: ja ik weet maar kan uh Janet ~kan niets zeggen~ ~completes~ Mohamed: mceilijk voor mij lanet: ja het is ook mceilijk uh you don't know any differences t between the netherlands and morocco? yes I know but can uh ~can say no[hing~ ~completes~ difficult for me yes it is difficult At a certain point after having tried several other areas of differences, Janet commented that this topic was not working. This whole conversation was filled with clarification sequences, which did not have the desired effect of stimulating Mohamed to talk. In a natural situation, a NS's motivation to negotiate meaning and continue the conversation might be have been less. Another explanation of the relative constancy of negotiation of ineaning may be that, because in informal conversation topic initiation is less controlled, new problems kept occurring. In the first cycle of the informal conversations, Janet controlled the topic quite definitely by continuously guiding it back to cultural differences between Morocco or Turkey, and the Netherlands, respectively. With Mahmut, in contrast to the conversation with Mohamed, Janet spent the whole conversation exploring the topic of giving presents. As can be seen in Example 3, they negotiated meaning frequently, starting with the introduction of the topic. (3) Infor~ttal conversation with Mahmut, Cycle I ts lanet: en als je kadootjes hebt [t] cc co ac ts cc co ti tc gc [c gc tc gc tc tc tc mr Mahmut: Janet MahmuC lanet: Mahmut: JaneC Mahmut: ]anet Mahmut: Janet: Mahmut Janet Mahmut: Janet: Mahmut: Janet Mahmut [~kado~] leen kado(x) ja kado en je gaat naar een vriend? ja in nederlandl weet je hce ze het in nederland dcen? uh kado ja t ik kom bij jou [ja] [nee] jij komt bij mij ja jij geeft een kadotje ja dan maak ikl dan doe ik zol maak ik meteen open nee turkse mensen nee die kadootjel jij uh mij kadootje halen 5.2 Results and if you have ~presents~[t] ~dimunitive forrru [dcado~] - vepeats~ la present (x) yes present and you go to a friend? yes in the netherlandsl do you know how they do it in the netherlands? uh present yes t ! come to you [Yesl [no] you come to me yes you give a present yes then I makel then 1 do like thisl make it open immediately no turkish people no that presend you uh me present ~get~ 77 Mahmut: Janet: Mahmut JaneC MahmuC ]anet-. Mahmut: Janet: MahmuC ]anet: ~meaning geven~ hè [t] [ja] ~meaning give~ huh [t] l~die~ die hè - ~points to cup~ ja [yes] 1~thab that huh - ~points to cup~ yes en dan kadootje (x) kijken [t] [ja] hurkse mensen niet kadootje hierl zij hier zitten ik hier" niet kadootje hier" niet kijken niet kijken t waarom niet? nee mag niet mag niet and then watch (x) present [t] [yes] Iturkish people not present herel they here sit 1 here" not present here" not watch not watch t why not? no not allowed not allowed zij ~weg"~ - ~-P-wEk~ they gone It took several clarification sequences before Janet had explained her experiences in Morocco. Mahmut then tried to explain that in Turkey it is impolite to open a present immediately. They would open the present later. Opening a present immediately may embarrass the giver, if it happened to be a small inexpensive present. In the last part of the conversation Janet and Mahmut discussed the kind of presents appropriate for a Turkish wedding. In contrast to Mohamed, Mahmut did not let his minimal competence hinder him. As can be seen in Example 4, he frequently requested Janet's help if he did not know the words to express his intended meaning. (4) Informal conversation with Mahmut, Cycle 1 ts~ti Mahmut die ~zo~ - ~gestures of necklace~ tc Janer. jaja ketting co Mahmut t ~ketting~ - ~imitates~ rslti Mahmut: die ~deze~ - ~points~ tc lanet: oorbellen cc Mahmut: ~oorbellen?~ - ~imitates~ tc Janet: oor [oor] cc Mahmut: [oor] oor tc lanet: bellen cc Mahmut: bellen tc Janet oorbel bellen cc MahmuC oorbel (...) that ~like this~ - ~gestures of necklace~ yeah yeah necklace gt ~necklace~ - ~imitates~ that uhis~ - ~points~ earrings ~earrings?~ - ~imitates~ ear [ear] [ear] ear rings rings earrings rings earring (...) In the second and third cycle, with al] NNS infotmants the topics became more diverse, because the informants appeazed to be able to initiate topics themselves. Differences in culture still came into the discussion, but personal topics such as work and family were also favored. In the second as well as the third cycle with 78 Chapter 5 The Amount ofNegotiation ofMeaning Mahmut, a variety of topics were discussed: work, illness, social security, housing conditions, insurance, vacation, unions, coffee prices, etc. Mahmut appeared to be more able to expand on these topics. However, these new topics may also have created a variety of new difficulties in understanding. Because of the absence of a script which could have supported the guessing of ineaning, negotiation remained necessary. Thus it is likely that the NNSs had continuous opportunity to acquire new linguistic items and structures through negotiation of ineaning. Glahn (1985) investigated what she called the metalinguistic phases in informal NS-NNS conversations in English with Danish students, whi~h "aim at establishing or reestablishing shared understanding". These phases are more or less comparable with clarification sequences. She found that the number of NS repairs was four times as high with grade 6 learners than with grade 121earners, indicating there was more negotiation of ineaning with grade 6 learners. In addition, Glahn (1985: 24) found that Danish grade 12 learners of English compared to grade 6 learners were able to communicate more fluently. They produced twice as many words and half as many communication strategies in interaction with a NS of English. In contrast to what is found in the current study, the amount of negotiation of meaning appeared to decrease as the NNS communicative competence increased. Still, comparison of the results is difficult because Glahn's study was cross-sectional and the current one is longitudinal. In addition, the operationalization of metalinguistic phase excludes feedback turns such as contirmatíons and gratuitous concurrence, which makes quantitative comparison difficult. 5.2.2. The Institutional Conversations As discussed in section 5.2, there is more of a decrease over time in the relative share of clarification sequence turns in the institutional conversations. The average percentage spent on negotiation in the first cycle was 51 qo, in the second 45qo, and in the third cycle 33010 (see also Figure 5.2). This decrease of 20qo may show a learning effect resulting from the fact that these conversations follow a script. The content of the institutional conversations was quite similar over time because a registration form was used as a guide for the topics to be discussed. The NNS informants were at iirst not familiar with the script of a Dutch housing office interview. In Morocco and Turkey there is no government-controlled social housing as there is in the Netherlands. Therefore, the script for this type of institutional conversation had to be acquired over time. The NNS informants learned that they were supposed to explain their housing needs and then the official would request or check information from them about, among other things, their cunent living arrangements to determine the urgency of the situation (see also Chapter 3). In addition, they would be asked to express such specific desires as to where the house should be located or what the maximum rent should be. At the third point of ineasurement, the NNSs will have had some idea of which topics are to be discussed. They will have acquired some of the socio-cultural knowledge, specific lexicon and idiomatic structure. This can be shown by comparing the Examples 5 and 6 5.2 Results ~9 below. In cycle 2, the housing official wanted to know how much rent Fatima was willing to pay. Rent can include or exclude utilities. The housing official used the concept kale huur (`bare' rent) in Dutch to refer to rent without utilities. Fatima initially appeared to think that he meant rent inctuding utilities and therefore mentioned quite a high amount she was willing to pay. (5) Institutional conversation with Fatima, Cycle 2 ts Bart: en uh de huur hce hoog mag hij zijn? ti Fatima: t wat? kale huur of ook gas en licht? t ja t gal o(t] - daughs~ [ja] l~'dow we lema'~ - ~-m-water and IighU - daughs~ kale huur geen gas en licht? want dat wordt duurder hè f dan wordt het duurder alleen huur vierhonderd gulden ja? [en] (ja] dan gas en licht erbij misschien zeshonderd gulden [zeshonderd vijftigj 'bare' rent no gas and light? because that is more expensive huh then it becomes more expensive only rent four hundred guilders yes?[and] [yes] then gas and light with it maybe síx hundred guilders [six hundred fifty] dan wordt het te duur f ja - daughs~ then it becomes too expensive t yeah - almost inaudible Bart: tc tc mr cc co and uh the rent how high can it be? what? - almost maudible de huur betalen pay the rent how much are you ready to pay? yes maybe four hundred rent four hundred guilders? yes you can pay that? yes yeah 'bare' rent or also gas and light? yes t gal o[t] - daughs~ [Yeah1 l~~`dow we lema'~ - ~-m-water and light~ - daughs~ Fatima: Bart: Fatima: Bart: Fatima: Bart: Fatima: Bart: Fatima: Fatima: Bart: Fatima: Bart: Fatima: Bart: Fatima: Bart: hceveel bent u bereid te betalen? ja misschien vierhonderd huurvierhonderd gulden? ja dat kunt u betalen? ja ja [zeshonderd] ja dan wordt duur oo veel duur o - daughs~ [six hundred] yes then becomes expensive oh much expensive In the third cycle, Fatima prevented the same misunderstanding from occurring again by immediately checking whether Bart meant rent without utilities. She then mentioned a lower maximum amount of rent (see Example 6). (~ Institutional conversation with Fatima, Cycle 3 Bart: 80 ja en hceveel huur wil je betalen? yes and how much rent do you want to pay? Chapter 5 The Amotrnt of Negotiation of Meaning Fatima: Ban: Fatima: Bart: Fatima: Bart: Fatima: t (x) uhlhuur zonder uh gas of licht~ [t] [hmhm] misschien t tweeenvijftig of driehonderd [tweehonderdvijftig f of driehonderd] [tweehonderdvijftig t of driehonderd] [ja] [hm] wij heeft niet veel uh f[t] t (x) uhlrent without uh gas or light~ [tJ [mhmhmJ maybe t fifty-two or three hundred [two hundred fifty t or three hundred] [two hundred fifry t or three hundred] [yes] [mhm] we ha.e not much uh t[t] In the continuation Fatima tried to convince the housing official that her family income was too low to afford more rent. She appeared to have acquired the concept of rent used by the housing official and thus was better able to achieve her goals. On other occasions, the decrease in negotiation of ineaning may have originated from the housing official's adjus[ed goals and strategies. In the second cycle, Bart attempted to explain the whole system of urgency credits to Ergun. He asked if there were any financial, social or medical problems for which Ergun could receive extra credits (see Example 7). (~ Institutional conversation with Ergun, Cycle 2 ts Bart: en zijn er omstandigheden op medisch gebied [t] [t] ti Ergun: waarop urgentiepunten tc Bart: te krijgen zijn hè? t bent u of uw aanstaande vrouw ziek? Ergun: ja gc tc Bart: om gezondheidsredenen urgentiepunten ja of nee? een van de twee ziek of niet? [c problemen med met uw gezondheid of niet? Ergun: jawel t-~ die wat betekent die? mrltc Bart: dat is medisch he cc co tc Ergun: Bart: ti tc mr cc cc co ac mc co Ergun: Bart: Ergiin: Bart: Ergun: Ban: Ergiin: 5.2 Results ~medische~ medisch dat is voor uw gezondheid [als] [~gezondheid~] - ~imitates~ u gezond bent ja ja allebei gezond? geen problemen? (x) geen problemen geen problemen duidelijk? ja and are there circumstances in the medical area (tJ [t] for which you would get urgency credits huh? or aze you or your wife-to-be ill? yes because of inedical reasons urgency credits yes or no? one of the two ill or not? problems withl with your health or not? yes tt that what dces [hat mean? that is medical huh ~medicab medical that is for your health [ifJ [health] -~imitates~ you aze healthy yes yes both healthy? no problems? (x) no problems no problems clear? yes g1 In the third cycle Bart did not explicitly refer to any aspect of the urgency system again. He probably decided not to get into the system again because it caused so many problems before. In the third cycle, he only tried to find out whether any credits could be given in a more implicit way. He inquired about Ergiin's health when Ergun brought up the topic himself (see Example 8). (S) Institutional conversation with Ergiin, Cycle 3 Ergun: ik heb vijf maandl drie maand werk ik heb twee maand ziek geweesdziektewet Bart: hm Ergun: noul ik ben nou helemaal niks dcen Bart: hm ts en wat heb je (.rx)I Iwat ziek zijn? wat is dat? cc co rtv Ergun: Bart: Ergun: waarvoor die ziek van mijn? ja ik heb operatie gehad nou (x) knie ac Bart: Ergun: Bart: knie wat? meniscus qjwat? (xx) hm ja Ergun: met vcetballen of zo (xx)? ja met vcetballen ja Bart: Ergun: ja - ~laughs~ cc ac 1 have five months three months work I have two months been sicklhealth insurattce act mhm nowl 1 am now totally nothing to do mhmhm anA what do you haro~e (z.r)I Iwhat to be ill? what is that? what for that ill of mine? yes 1 had surgery now (x) knee knee what? meniscus or what? (xx) mhmhm yeah with soccer or something (xx)? yes with soccer yes yes - daughs~ However, a comparison of certain typical idioms over the three cycles also showed continuous semantic problems. Expressions such as `ingeschreven stamt' (to be registered) remained problematic for most of the NNS informants because of the opaque combination of a common verb, 'staan' (to stand) and a particle-verb combination of inschrijven (`to write in'). Three conversations over two years were probably not enough to acquire an understanding of such a specific idiom. Attention should also be given to individual differences between the NNS informants because they are considerable, in particular in the institutional conversations. As could be seen in Table 5.1, the results of the second cycle are most divergent for Fatima and Ergun, for whom the amount of negotiation is much higher in the second cycle than in the third. They may still have been developing towards an optimal level of negotiation. Fatima was the least proficient NNS when the recordings started and her language proficiency where syn[actic development is concerned did not increase much over the two year period (Van de Craats, 1994). This might explain the lack of decrease in negotiating meaning in both interaction types. In the second cycle of the institutional conversations, the percentage is especially high. The results here, however, are also influenced by one clarification sequence of 67 turns in which [he housing official was only partly successful in 82 Chapter 5 The Amount of Negotiatfon of Meaning finding out whether Fatima had been living in a rental home from one of the housing corporations (see Example 9). Fatima fell silent quite often and appeared to become very confused and discouraged after she had answered the first two questions correctly but the NS continued to ask for more detailed infotmation. (9) Institutional conversation with Fatima, Cycle 2 Ban: waar u woont he bij familie [t] Fatima: [ja] Bart: is dat eigen huis of niet? Fat: niet Bart: eigen huis van familie of niet? ts ti tc Fatima: Bart: Fatima: Bart: nee niet van wie is dat huis? t4t waar u nu woon[ t hè t u woont nu bij familie ti cc Fatima: Bart: t zei u straks hè t wonen bij familie t van wie is dat huis t waar u woont? hm ~?C~straat - ~address of informant~ tc tr Fatima: ac2c co gc ti tc cc co Bart: Fatima: Bart: Fatima: Bart: Fatima: Bart: cc co cc co ac tc tc Fatima: Bart: Fatima: Bart: ti tc tr ac tc Fatima: Bart: Fatima: Bart: tr Fatima: ac Bart: cc co Bart: Fatima: 5.2 Results ja ~X~straat ja ja [t] [~] van wie is die woning tt nummer? [hè uh] [nou] nummer niet t mag ook t where you live huh with family [t] [yes] is that own house or not? not own house of family or not? no not whose is that house? t4t where you live now t huh t you now live with family t you said a whíle ago t live with family t who's the house from t where you live? mhmhm ~7{~street - ~address of informant ~ yeah ~J~C~s[teet yes yes [t] [tl whose is the residence tt number? [hm uh] [well] not number t that's okay t maar dat is wat anders t but that's something different uh the house where you live now uh het huis waar u nu woont in the ~?C~ the ~3C~ street t in de ~3C~ de ~X~ straat t that house is not yours dat huis is niet van jullie no nee that is family's dat is van familie yes ja yes t ja t van wie is dat huis? whose is that house? is that [heir own? is dat van hun zelt? or that of somebody else? of dat van iemand anders? t t is dat een eigen woning of niet? is that an owned residence or not? t not t niet no nee did these people buy hebben die mensen dat it t [or rent]? gekocht t [of huren]? [no] only rent [nee] alleen huur also rent? ook huren? yes ja yeah yeah jaja g3 v ac tc Fatima: Bart: tr cc ti mc Fatima: Bart: Fatima: Bart: co mc ti cc Fatima: Bart: Fatima: Bart: ti cc co Fatima: Bart: Fatima: alleen huur ja van een bouwvereniging? of niet? nee niet van een bouwvereniging? ~ mceilijk he? ja - daughs~ - daughs~ ja t toch proberen [t] [t] only rent yes from a housing corporation? or not? no not from a housing corporation? ~ difficult huh? yes - daughs~ -daughs~ yes t try anyway [t] [tl you live in de street t yes? yes (...) u woont in de straat f ja? ja (...) At this point, Bart started all over again, checking the already acquired information and trying to get further from there. Eventually he did establish that Fatima lived in a house owned by a housing corporation by asking her yeslno questions with the acronyms of the corporations such as "is it from the SVW or TBV?", the two housing corporations in Tilburg. She, however, answered this question with something like `FWW', so he did not get the final piece of information. This sequence shows how difficult conversations which require exact information transfer in particular can be, because face considerations also seem to limit the number of opportunities to negotiate meaning. Fatima became very discouraged by the extensive `crossexamination' so it must have influenced the whole conversation negatively. This sequence may remind readers familiar with John Gumperz's work of the social security interview in the film `Cross Talk' (1991), which also shows the defeatedness of the NNS client after neither he nor the off'icial had been able to achieve their communicative goals because they were at cross-purposes. If we look at the relative share of negotiation for Mohamed and Mahmut, we see that they are both more or less equal in cycles 2 and 3 while they were higher in cycle 1. This may have been caused by the change of interviewer (Paul in the first cycle, Bart in the other two), or it may be due to a learning effect which was established between cycles 1 and 2. It might be that the latter informants had already reached an optimal level of negotiation in cycle 2, because their percentages stayed more or less constant at a low level in cycle 3. However, it may also show that Bart was more effective in communicating with these NNSs. This latter hypothesis will be explored further in Chapter 8. Comparison of the results with other research indicates that the average of 33qo of clarifcation sequence turns in the third cycle is still high. Pica (1987) found that in classrooms with low-intermediate English-as-a-Second-Language students, the percentage of comprehensionlconfirmation checks and clarification requests (other-initiations) varies from 6qo to 24010. These percentages are not completely comparable because Pica does not include negotiation moves such as feedback moves in her analyses, but they do give an indication. In addition, it is not clear 84 Chapter 5 The Amount of Negotiation of Meaning whether the communicative competence of the NNSs in this study was compazable to that of the low-intermediate learners in Pica's study. Pica explains that certain aspects of the interactional structure (such as little genuine information exchange) in the classroom lead to such low percentages. The highest percentage of 24qo was reached in a small-group task with a two-way information exchange, while the lowest percentage occurred during a decision-making discussion. Therefore, the task appeared to have influenced the amount of negotiation of ineaning. The conversations of the current study appear to be most comparable to the two-way smallgroup task. The degree of negotiation of ineaning reached is high compared to what Gumperz (1990) found in institutional interaction between NSs of British English, and Indian or Pakistani speakers who are generally quite fluent in English because of the colonial influence of this language in their home countries. Gumperz found that about one third of the conversations was spent on repair. It is not clear how far his definition of repair overlaps with what in the current study is called a clarification sequence. Most likely Gumperz's definition is more limited and did not include feedback turns, and that self-initiated self-repairs made up the bulk of the repairs. Mazeland (1992: 140) found a similar percentage in NS-NS research interviews which were mainly self-initiated self-repairs. Therefore, the 33qo average relative share of clarification sequences turns in the current study is high considering that self-initiated self-repair is excluded. Long (1983a: 136-137) refers to research of his own which showed that clarification requests (trouble indicators), trouble clarifications (self-repetitions, other-repetitions, and expansions), confirmation checks, and comprehension checks were used significantly more often in NS-NNS interaction than in NS-NS interaction; however, he did not give percentages which could be compared with the current study. Nevertheless, as would be expected, NS-NNS interaction contained more negotiation moves than NS-NS interaction. 5.3 Conclusions In the analyzed data, the amount of turns spent on negotiation of ineaning is high but variable in most conversations (from around 20"Io up to 70qo). These findings agree with those reported in Long ( 1983a) that there is more negotiation of meaning in NS-NNS interaction than in NS-NS interaction. Gumperz (1990) and Mazeland ( 1992) found that about a third of the institutional conversations is spent on repair in interaction between ( near-)native speakers. The major share is probably self-initiated self-repair, which is not included in the current study. Glahn (1985) also found lower levels of NS repairs and NNS communication strategies in informal conversations between Danish students and NSs of English. The high levels of negotiation of ineaning in the current study may, on the one hand, create opportunities for language learning. On the other hand, they may also threaten 5.3 Conclusions 85 communicative success. It appeared that the factors NNS competence and interaction type influence the amount of negotiation of ineaning in interaction with each othec For most NNS informants, it is only in the institutional conversations that there appeazs to be a decrease over time from an average of SOqo to 33qo. Fatima showed a somewhat different pattern with the largest percentage of clarification sequence turns occurring in the second cycle. Her relatively low language proficiency or gender may have influenced the results. In general, there was considerable individual variation, which may be caused by personality and gender differences. These factors, however, were not systematically investigated. In the informal conversations, the level of negotiation of ineaning over the three points of ineasurement was also high but more or less stable around the SOolo. No relationship between level of NNS communicative competence and amount of negotiation was found. This may indicate that in less goal-directed conversation more continuous negotiation seems to take place. At the beginning, although there may be minimal understanding or even only an illusion of understanding, the level of negotiation if controlled by face factors and the threat of breakdown. In addition, negotiation of ineaning may only increase problems rather than solve them. In the subsequent cycles, new topics may have created new problems, especially because a script that could support guesswork was lacking. This may have kept the amount of negotiation of ineaning more or less constant. In addition, the fact that the conversations were held to collect NNS language data may have motivated the NSs to continue interaction through negotiation of ineaning at points where in real life they would have given up. The level of negotiation in NS-NNS interaction would probably remain higher than in NS-NS interaction, until the NNS reaches near-native competence and shares more cultural background with the NS. None of the four NNSs, however, were anywhere near this point even by the end of the data collection. gb C{wpter 5 The Amount ofNegotiation of Meaning Chapter 6 Asymmetry in Negotiation of Meaning 6.0 Introduction Now that we know from the outcomes of Chapter 5 that a large amount of the conversations is spent on negotiating meaning, the question of what the role of the NNS is in the process of negotiating meaning becomes even more relevant. In Chapter 4, it was found that the NNSs mainly used trouble indicators to initiate repair after a trouble source, while the NSs preferred confirmation checks.lfiis may reflect an asymmetrical distribution of roles in negotiation of ineaning, which may influence NNS communicative effectiveness and opportunities for language acquisition. As formulated in Chapter 3, the questions discussed in this chapter concern the interactional asymmetry and two factors influencing it, NNS communicative competence and interaction type. 3a. Is there asymmetry in NS and NNS participation in negotiating meaning? 36. How do NNS communicative competence and interaction type influence the distribution of negotiation moves over the NS and NNS? In section 6.1, expectations based on the literature are formulated about the influence of NNS communicative competence and expert status (institutional role and topic knowledge) on asymmetry in negotiation of ineaning. In section 6.2, the procedure for investigating asymmetry is described. The results are presented for the three main negotiation moves, i.e., trouble indicators (section 6.2.1), trouble clarifications (section 6.2.2), and confirmation checks (section 6.2.3), followed by a discussion of the results. The chapter ends with conclusions (section 6.3). 6.1 The Influence of NNS Communicative Competence and Interaction Type In clarification sequences, three turn types were identified which may play an important role in negotiating meaning: - trouble indicators, which initiate the negotiation process, - trouble clarifications with which repair work is done, 6.O lntroduction 87 - confirmation checks with which repair is initiated and repair work is done (see Chapter 4). By indicating trouble, the NNS may receive comprehensible input from the NS who will clarify. If the NNS uses a confirmation check instead of a trouble indicator, he creates an opportunity for producing comprehensible output, but he may also risk threatening the NS's face. Lacking communicative competence and having a dependent role, the NNS may be viewed as challenging the NS's expertise in rephrasing the NS's words. If the NS indicates trouble, the NNS may create comprehensible output by clarifying himself. Pica et al. (1989: 83) found that NNSs modified their output most often when the NS requested clarification instead of confirmation. Therefore, it might be conducive for language learning if the NNS is pushed to clarify; it will, however, take time, and the NNS might have to face the frustration of not being able to clarify himself. Thus the NNS may feel that the NS is being uncooperative. It may be for this reason that Pica et al. (1989) find that a NS in negotiating meaning uses confirmation requests much more often than trouble indicators. In this case a simple `yes' from the NNS often suffices as response to a confirmation check. Using a confirmation check, however, may also be a communicative risk in that a NS may confirm the check too quickly, even though he may not have exactly understood what was said. At worst, this may lead to misunderstanding, but in any case the NNS loses control over the outcome of the interaction. Trouble clarifications and confirmation checks may be typical NS moves because they manipulate content and form, and therefore require a considerable amount of language. The NNS is expected to make more frequent use of feedback moves, in particular trouble indicators and confirmations. This would confirm results from Zuengler 8z Bent (1991) which showed [hat NS and NNS have complementary roles, the NS talking more and the NNS using more feedback signals. The influence of two factors central to this study on asymmetry in negotiation of ineaning are investigated, NNS communicative competence and interaction type. Concerning the NNS's communicative competence, the following expectations can be formulated. As discussed in Chapter 2, Woken 8c Swales (1989: 212) referred to several studies which found that NSs in interaction with NNSs dominate in negotiating meaning, that is, in initiating repairs and in offering assistance with syntax, lexicon and pronunciation. These last two actions are similar to what is here called trouble indicating and trouble clarification, or using confirmation checks. They then define dominance in negotiation of ineaning as taking an initiative as well as a reactive position. This is curious because negotíation is an interactive process. If one speaker indicates trouble, the other is requested to clarify. One speaker does not do both if these roles are complementary as expected. Therefore, one would expect that either indicating trouble or clarifying and checking would show dominance. In the current study, dominance will be defined as the latter. A speaker who uses confirmation checks and trouble clarifications manipulates the content of the conversation and thus is expected to have more control over the outcomes. 88 Chapter 6 Asymmetrv in Negotiation of Meaning As long as the NNS does not have sufficient competence to direct the negotiation process himself, he may indicate trouble but leave the manipulation of content to the NS because it requires a great deal of linguistic competence. Faerch 8z Kasper (1985: 19) found that learners at the intermediate level in their study (grade 10) more frequently initiated negotation of ineaning than lower-level (grade 6) learners not only because there were considerable perception and production problems, but also because they were more able to make use of the NS's help. It is not clear whether the NNSs in this study are comparable in competence. It might, however, be that NNS trouble indicating will increase over time, rather than decrease. Then the NS's use of clarification moves will probably also increase. However, if the NNS's communicative competence increases over time, we could also expect a movement towards a more equal role distribution in initiating repair, clarifying and checking. T'hus, the NS will use more trouble indicators as the capacity of the NNS to clarify trouble increases. In addition, the NNS will use more trouble clarifications and confirmation checks as his communicative competence increases. This would agree with findings from Gass 8c Varonis (1985b) that NSs initiate more negotiations with low-level than with high-level learners, and in contrast to Faerch 8z Kasper (1985) that NNS low-level learners initiate more negotiations than highlevel learners. Thus it is not clear how the distribution of trouble indicating behavior will change over time. Concerning the influence of the interaction type, in Chapter 2 two aspects of expert status were mentioned which may influence the asymmetry in the informal and institutional conversations. These are institutionallauthoritative role and topic knowledge. Both will be discussed. In the informal conversations, the NSs, as project researchers, are in a sense representatives of an organization, i.e. the university where the conversations were recorded. However, they have no authoritative role in the interaction because their university status is irrelevant. Thus their institutional role may or may not lead to asymmetry in negotiation of ineaning. In the institutional conversations, the NS has the role of housing official and thus plays a`gatekeeper' who represents an organization which controls access to affordable housing (see also Chapter 2.6). The interaction is structured and NNS initiative may challenge the NS's authority, in particular, through the use of confirmation checks. Therefore, asymmetry is expected. The NS is expected to be dominant through the use of confirmation checks and trouble clarifications. The secund characteristic of expert status which is found in earlier research to influence asymmetry is topic knowledge. In the informal conversations, there was no formal expert. However, the NS and NNS could both be considered informal experts on their own cultures and experiences. Therefore, no asymmetry is expected based on topic knowledge. In the institutional conversations, the housing official was the expert on the content domain of social housing. He knew the procedures and regulations. For the NS Bart, this knowledge was based on training and experience. The social worker Paul may have had less knowledge because he only play-acted the part of housing official. However, the NNSs were informal experts in a sense, too, on their own hous- 6.1 The Influence of NNS Communicative Competence and Interaction Type 89 ing conditions and desires. Still, asymmetry is expected because the NS is the formal expert combining authoritative role with knowledge resulting from training (see Chapter 2.6.3). The NNSs may indicate trouble more frequently in the institutional conversations because the interview format presented them regularly with questions they could not answer otherwise. This would lead to more NS clarification work. The housing official had to explain housing regulations and procedures and justify the need for certain informa[ion to be conveyed. In addition, the role of the housing official and the script may lead to more NS checking because the information received had to be correctly reflected in the registration form. Therefore, any ambiguity had to be resolved. In conclusion, it can be expected that there is more asymmetry in the institutional interaction, which is sustained when the NNS communicative competence increases because of the asymmetry in expert status (institutional role and topic knowledge). Thus the NS will manipulate content by clarifying and checking most frequently whereas the NNS may be restricted to feedback moves such as trouble indicators. Linell et al. (1988) claim that the expert status in institutional interaction has a greater effect on dominance than linguistic competence. If that is the case, the asymmetry should be considerable in the institutional conversations, while there should be more symmetry in the informal conversations. Therefore, in the informal conversations, less asymmetry is expected at the first point of measurement. This asymmetry is expected to decrease over time in the direction of equal distribution of trouble indicator, clarifying and checking moves. 6.2 Results To determine the relative distribution of trouble indicators, clarifications and confirmation checks over the NS and NNS, the natural logarithms of the NS and NNS frequencies were calculated, showing the relative distribution over the two types of participants (Rietveld 8c Van Hout, 1993: 328-331). The outcomes are so-called logits (logit - ln ((frequency NS moves)~frequency NNS moves)); see also Rietveld 8z van Hout, 1993: 328-331). Because there were zero scores for a few NS or NNS moves, 0.5 was added to all figures, so that the logit could be calculated for these cases also. If the NS and NNS distribute the negotiation moves between them equally, the logit will be zero. If the NS uses a particular type of move more frequently than the NNS, the logit is greater than zero. If, however, the NNS has a greater share, the logit is less than zero. In Appendix 3, Tables 3 and 4, overviews are given of all frequencies of NS and NNS trouble indicators, clarifications and confirmation checks per NNS informant, cycle and interaction type, with their respective logits (Ln NS~NNS). The logits are also presented in Figures 6.1 to 6.6 in the following sections, in which the results will be discussed per negotiation move. The influence of NNS communicative competence and interaction type will be discussed together per move because, like in Chapter 5, they interact with each other. 90 Chapter 6 As~mmetn~ in Negotiation of Meaning 6.2.1 T~-ouble Indicators As can be seen in Figures 6.1 and 6.2, it appears that the NNSs did most of the trouble indicating at all three points of ineasurement because all the scores except for one are negative. 1 2 cycle 3 Figure 6.1: Logits of trouble indicators in the informal conversations ---w-, Fafírna [ ~~Iohamed ~~ : Mahmut ~ Ergiín É -~.~:~ 1 2 3 cycle Figure 6.2: Logits of trouble indicators in the institutiona! conversations 6.2 Results 91 This outcome confirms the results of Chapter 4. There it was found that NNSs prefer to initiate a repair sequence with a trouble indicator. Trouble indicating is the task which requires the least linguistic competence, because it can be done with non-verbal and minimal verbal means, such as raising eyebrows or minimal feedback questions (`mhm?'). It may therefore be the favored way of initiating repair for the NNSs. From the analysis of the forms of NNSs' trouble indicators in Chapter 7, it will become clear whether the NNSs most frequently used such minimal means. Gass 8L Vazonis (1985a: 160) found that recipients of information used more trouble indicators than the speakers who provided it. In the current study such a division cannot be so easily made because both NSs and NNSs give and receive information to and from each other, although in the institutional conversations the NNS may have given more information than they received. This might explain high frequencies of NNS trouble indicators in this interaction type. However, the asymmetry in the informal conversations cannot be explained this way. There was a tendency towards more symmetry for the male NNSs in the informal conversations. With Ergun, there was a development towards more symmetry towards the third cycle in the informal conversations. For Mohamed and Mahmut, there was also a movement towards more symmetry in the informal conversations. Only with Mahmut is symmetry almost reached in the third cycle in both interaction types. Mahmut's conversational style may have led to these results. He was very talkative, even though his linguistic abilities were at first very limited. He lacked morphology and prepositions, which created problems in the institutional conversations in particulaz, because it was necessary to convey precise information on uncommon topics, and the NSs seemed to have difficulty comprehending him. They often asked for clarification when they were not able to guess Mahmut's intended meaning. Fatima showed a deviating pattern compared to the three male informants in both interaction types. With Fatima, there is development towazds more asymmetry in the informal conversations, while there is no consistent pattern in the institutional conversations. As seen in Chapter 4, in the informal conversations Fatima increasingly initiated more NS sequences in the second and third cycle. Only from the second cycle on might she have been able to make use of the NS's help. This agrees with Faerch 8c Kasper's findings (1985: 19) that intermediate learners more frequently initiated negotation of ineaning than lower-level learners. Fatima's communicative competence was so low at first that she did not take much initiative in indicating trouble but tried to conceal non-understanding as much as she could by keeping the conversation going with minimal linguistic means. In the second and third cycle, she may have felt more confident about indicating trouble (see also section 6.2.2). In addition, there may have been a gender effect. Fatima interacted with a male in the institutional conversations and with a female in the informal conversations. Gender has been shown to influence the distribution of conversational behavior (Mulac, 1989; Ainsworth-Vaughn. 1992). Pica et al. (1989) concluded that gender and ethnicity may have influenced the amount of negotiation of ineaning in their study but did not investigate it systematically. Gass 8z Varonis (1985a) found that 92 Chapter 6 Asymmetr}~ in Negotiatioa of Meaning women use fewer trouble indicators. In the institutional conversations, Fatima interacted with males with a higher status. Especially for NNSs who come from a country with clear gender role division, such as Arabic countries (Hofstede, 1980), cross-gender conversations may differ from same-gender ones. Culturally, in Morocco a woman would be expected not to take initiative or contradict someone of the other sex and with a higher social status. Therefore, Fatima might have performed according to a somewhat different conversational role in the institutional conversations than the three male informants. In Example 1, Fatima contradicted the housing official only after several affirmations which were not accepted by him. Then when she did contradict him, she at first did not give him an acceptable reason by using the empty phrase daarom (that's why). This may show her reticence in having to be assertive and argue her case. Stimulated further, she gave an acceptable reason by stating that she wanted to marry soon. The hesitations in this utterance show her discomfort. With the next question, she resorted again to the phrase daarom (that's why) and she laughed, which may again show her discomfort. (1) Institutiona) conversation with Fatima, Cycle 1 ts PauL kun je nied kun je niet bij je ouders gaan wonen? Fatima: f ti bij jouw vader en mceder? tc Paul: gc Fatima: ja cc Paul: ja Fatima: ja gc tc Paul: kun je daar niet zolang blijven? gc Fatinw: ja tc PauL totdat er een huis is mr Fatima: nee Paul: waarom? Fatima: ik wil mijn huis Paul: waarom? waarom kun je niet naarje ouders? Fatima: daarom Paul: aa datislnee~ dat is geen mooi antwoord t waarom niet? Futima: t ik uh wil uh t vlu~ troux7 Paul: jij wil vlug trouwen? Futima: ja Paul: waarom? Fudmu: duuram ~~ - ~fau~hs~ can't youl can't you go and live with your paren[s? t with your father and mother? yes yes yes can't you stay there in the meantime? yes until there is a house no why? I want my house why? why can't you go to your parents? that'.s why ah that islno~ that is not a nice answer t why not? t I uh x~ant uh t yuickly marries you want to many quickly? ye.c why? that's why ~~ - ~faughs~ In the informal conversations, Fatima appeared to feel more intimate with Janet, who was also a female, which may have made the conversations more symmetric. However, the asymmetry increased over time. Thus gender role distribution is not the best explanatory variable here. The increase may be caused by her very low 6.2 Results 93 communicative competence. In the first cycle she could not make much use of the NS's help, but later on when her communicative competence developed she did request more help. 6.2.2 Trouble Clarifications As can be seen in Figures 6.3 and 6.4, clarifying trouble was mainly the task of the NSs. 1 2 3 cycle Figure 6..i: Logits of trouble clarifications in the informal conversations In the first cycle in particular, there was a great deal of asymmetry in both interaction types for most NNS informants. This is especially true of the informal conversations with Mohamed, who was reluctant to speak when he was not sure he had understood things well (see Chapter 5, Example 1 and 2). There is quite a range in the distribution of clarification moves over the NSs and NNSs. With Fatima there is no consistent pattern in either of the interaction types. In the informal conversations, there is almost a perfect balance in the distribution between her and Janet in the first two cycles. In general, the scores for the three male informants approach each other more closely, and move towards symmetry, which might indicate that a certain optimal balance had been reached by the third cycle. Mahmut shows a deviating pattern in the institutional conversations: he clarified more in cycles 1 and 2 than the NS. Mahmut's conversational style may have influenced the outcomes which caused the NS trouble. 94 Chapter 6 As~~mmetry in Negotiation of Meaning 1 2 3 cycle Figure 6.4: Logits of trouble clarifications in the institutional conversations As can be seen in Example 2, Mahmut's clarification turns sometimes only intensified the confusion (Deen 1991). Here he tried to convey the message that his fiancee was delivering a baby in five months. However, in his learner language the verb `komen' (to come) was often replaced by one of its common particles `terug' (back). It seemed to mean something like `appear' in Mahmut's learner language and this created substantial misunderstanding. (2) Institutional conversation with Mahmut, Cycle 1 fS ti cc tc gc tc cc ti Mahmut: Paul: Mahmut.. Paul: Mahmut: Paul: babyl mijn meisje baby die twee f vijflvijf maanden he baby terug ja t ovedzeg het nog eens lover vijf maanden? ja t vijf maanden baby terug komen he ja die moet getrouw en dan baby terug moet die baby dan temg? hce bedcel je? babyl my girl baby that two t fivelfive months huh baby back yes t irJsay it once again !in five months? yes t five months baby come back huh yeah that must marrv and then baby back the baby must be back? what do you mean? 6.2.3 Confirmation Checks As can be seen in Figures 6.5 and 6.6, checking whether something was understood correctly was mainly the task of the NS. This confirms the results of Chapter 4 which showed that the NSs prefer confirmation checks to initiate a repair se- á.2 Results 95 quence. There are no consistent patterns of development towards more symmetry for any of the four NNSs. a 2 1 -2 -3 -4 1 2 3 cycle Figure 6.5: Logits of confirmation checks in the infonnal conversations 1 2 3 cycle Figure 6.6: Logits of confirmation checks in the institutional conversations 96 Chapter 6 Asymmetry in Negotiation of Meaning In the informal conversations, the scores for Ergun are stable; the NS checked more. For Mohamed the same can be said; except in cycle two he checked slightly more. Mahmut, first checked more himself but in the second and third cycle, the NS used more confirmation checks. In the institutional conversations, the scores for Ergun are quite stable, like in the informal conversations. For Mohamed and Mahmut, there is a movement towards more symmetry in the second cycle, but more asymmetry again in the third cycle. With Fatima there was no consistent pattern over the two interaction types. In the institutional conversations there was more asymmetry from the beginning, and there was a movement to substantial asymmetry in the second cycle of the institutional conversations. The asymmetry may have resulted from the NS's task to double-check all the information given to make sure the housing form was filled out correctly. After some common ground was established, the NS Bart also used confirmation checks to create a common starting point, for getting to a new piece of infortnation, as can be seen in Example 3. (3) Institutional conversation with Fatima, Cycle 3 Bart: je woont er nu drie jaar zoiets he? a - ~writes down ~ Fatima: ja drie jaar Bart: drie jaar wonen t je hebt een man t en een zoon o t ja? t en je wil graag verhuizen - ~writes dowro Bart: je wil graag ergens anders naartce wat wil je dan? ts Fatima: t-~ die huis cc Bart.~ je woont nu op de flat cc Fatima: nee co Bart: nu op de flat cc vind je het niet leuk co ce co tc ti tc ti Fatima: nee Bart: Fatima: Bart: Fatima: Bart: Fatima: je wil ieLC atulers ja die ander wat is dat? f wat wil je dan? a-t (...) you've lived there for three years something like that huh? o - ~writes down~ yes [hree years three years live t you've got a man t and a son ~~ t yes? t and you would like to move - ~writes down~ you would like to move somewhere else what do you want? ~ that house you now live in an apartment no nnw in an upartment you don't like it no ynu wcuit sntnething else yes that other what is that? f what do you want? tt (...) If NSs frequently model what they believe NNSs are trying to convey, it is hypothesized they might have more control over the content of the interaction and a greater chance of achieving their goals than the NNSs. Confirmation checks then 6.2 Results 97 could be an indicator of dominance. In part of the institutional conversation in cycle 1 about Mohamed's housing preferences, the NS controlled the outcome through confirmation checks to which Mohamed responded in the affirmative even though they did not reflect his desires. In Example 4, it remained implicit that an apartment (instead of a house with a garden) was the only realistic option for a young man like Mohamed. Houses are quite rare on the social housing market in the Netherlands, and not commonly assigned to young childless couples or singles. The NS guided the interaction to the most realistic solution, an apartment, without explicitly justifying it. Mohamed lacked the communicative competence to keep the other option, a house, open, although he did try at several points. The NS confitmation checks and clarifications rephrased Mohamed's utterances in the direction of the NS's desired outcome. Paul even asked for Mohamed's explicit consent, which was grudgingly given. (4) Institutional conversation with Mohamed, Cycle 1 ts Paul: wat voor huis zcek je? what kind of house are you looking for? ti Mohamed: ~ tt tc Paul: wil je een flat do you want an aparttnent of wil je een huis or do you want a house met een tuin? with a garden? ~ ti Mohamed: t tc Paul: tr Mohamed: of uhl t wat voor soort huis zcek je? t uh een flat of eh` cc colmr tc Paul: Mohamed: Paul: nee flat of uh ~` `zt~u je een fIat willen hebben? mr ts ti Mohamed: Paul: tc ti tc Mohamed: Paul: Mohamed: Paul: tr cc tr cc tr cc co Mohamed: Paul. Mohamed: Paul: Mohamed: Paul: Mohamed: - ~knocks on table~ ~een jlat ? flat of eh huis of eh maak je niet uit? hm? maakt dat niks uit? t wat je zcekl wat je wil nee wat uh` ~een flat is altijdgoed? ja? ja? mag wel mag dat we!? ja or uhl t what kind of house are you looking for? uh an apartment or uh` - dcnocks on table~ ~art apartment? no apartment or uh ~` `would you like tn have an apartment? apartment or uh house or uh dcesn't matter to you? mhm? dcesn't that matter? t what you're looking fod what you want no what uh` ~nn apartment wil[ always be fine? yes? yes? okay is rhar akay? yes Mohamed's last ja (yes) appears not to indicate agreement as much as willingness to cooperate after he was not able to achieve his goal (Liberman, 1984). In cycle 3, Mohamed was able to express his wishes more clearly. Then the NS also made explicit why a house was not an option, as can be seen in Example 5. 98 Chapter 6 As~.nunetry in Negotiation of Meaning (5) Institutional conversation wíth Mohamed, Cycle 3 Bart: en je wil een flat? ... and you want an apartment? Moharned: ja flat of f gewoon huis yes apartment or t normal house Batt: ja yes small Mohamed: klein Bart: ja een huis kan niet t yes a house is not possible t with the two of you met zijn tweetjes t dan kun je niet naar een huis t then you can't get a house It partly depended on the NNSs' linguistic abilities how much remained implicit and unsaid. In such a situation, NS interactional control may be perceived as manipulation by the NNS even if that is not the intention of the NS. Gumperz in his work on `contextualization conventions' (Gumperz 1990) states that such occurrences of miscommunication may be perceived as intentional discrimination although they may not be intended as such. Participants in the communication process arrive at different interpretations because of their different linguistic and cultural knowledge. Gumperz suggests that explicitness (about assumptions and interpretations) may prevent such miscommunication. Such explicitness is, however, not possible with less proficient NNSs because it would require more verbal exchange, which could in turn lead to more trouble. The findings on the three negotation moves will now be related together to the expectations formulated in section 6.1 concerning the influence of NNS communicative competence and expert status (interaction type). If we relate the results to the expectations formulated about the influence of NNS communicative competence, we can see that the general assumption that there is asymmetry in negotiation of meaning in NS-NNS interaction is confirmed. It was found that there appeared to be a complementary role distribution between the NS and NNS as was found by Zuengler 8c Bent (1991). The NNSs at all points of ineasurement indicated trouble more frequently than the NSs. The NSs, on the other hand, clarified more. Trouble clarifications aze often second parts of an adjacency pair with trouble indicators. Thus it is logical that when the NNSs indicated trouble more, the NSs clarified more. There is, as expected, a tendency over time towazds more symmetry between the NS and NNS for the trouble indicators and trouble clazifications. This is most clearly the case for the faster learners Mohamed and Ergun. This confirms the expectation that there will be more symmetry in negotiation moves as NNS communicative competence increases. However, for the confirmation checks, there is less development over time towards symmetry. The NS used confirmation checks more frequently. The adjacency pair confirmation check-confirmation, was not part of Gass 8t Varonis's model for NNS-NNS interaction (1985a). It might be the case that this pair is quite typical for conversations of NSs with NNSs and thus the use of confirmation checks may be indicative of dominance. As seen in Chapter 4, the NSs often appeared to relieve the NNSs of the manipulation of content for clarification by using confirmation checks instead of trouble indicators. Hence a NS can also direct the outcome of the conversation as was seen in Example 4. Hol- 6.2 Results 99 liday (1988) described the function of NS confirmation-like checks in NS-NNS interaction as compensatory and mentioned the negative effect this behavior may have on NNS language acquisition because they take away opportunity for the NNS to produce comprehensible output. The other factor which may influence asymmetry was expert status. It was expected that there would be more asymmetry in the institutional conversations, because differences in communicative competence as well as the authoritative role of the housing official and expert knowledge would lead to asymmetry. As we have seen, there is asymmetry in both interaction types. The difference between the interaction types is less than expected. Only for the trouble clarifications does there appeaz to be a sustained asymmetry in the institutional conversations, while there is a movement towards symmetry in the informal conversations. Apparently, differences in communicative competence exert a great influence on negotiation of meaning roles. These outcomes do not confirm the results of Linell et al. (1988) that the type of interaction has more influence on interactional symmetry than the communicative competence of the NNS. The results for trouble indicating and checking moves can be explained in three ways. Firstly, the factor topic knowledge may have influenced the results. As was discussed in section 6.1, research has shown that topic knowledge can have a powerful impact on participation in interaction. NNSs are not automatically dominated if they have expert knowledge. In institutional conversations, the housing official was the formal expert because he knew the procedures and regulations. This was particularly true for Bart, the real-life housing ofiicial. It may have been less true for Paul, the social worker, who had to make do with the information he received for the play-acting scene, and most likely his own experiences as a client (and from hearsay). However, in both ihe institutional as well as the informal conversations, the NNSs were informal experts because they had some specific knowledge about the topic (their housing needs versus cultural differences between their home country and the Netherlands). This may have created some symmetry and diffused the results. The results agree with Zuengler 8r. Bent (1991) who found that when the NS and NNS have equal knowledge there is no clear pattern of asymmetry. Rather, on some measures, the NS and NNS score equally (topic moves, interruptions), while on others, one or other party scores higher. They state that NSs talk more and NNSs use more fillers and backchannels, in other words, the NS is the active speaker and the NNS is the active listenec In the current study, this role distribution manifests itself in that the NS manipulated content more frequently and the NNS used trouble indicators, which are feedback elicitors, more frequently. It may be necessary to more clearly define what topic knowledge consists of and how it influences role distribution. Topic knowledge obtained through formal study and training should be distinguished from informal expert knowledge based on, e.g., cultural background. In addition, it is important to determine the relative impact of authoritative role and topic knowledge on expert status. Secondly, the semi-authenticity of the interaction may have influenced the results. Because there was nothing really at stake in the institutional conversations. the asymmetry might not have materialized as much as it would have in an authen- 1~ Chapter 6 Asi~mmetry in Negotiation of Meaning tic situation. In the informal conversations, institutional role and social status might have had an influence too. The NSs were university researchers who as such also represented an institution. Even though the topics were informal, there might have been an influence of a majority higher-status researcher speaking with a minority lower-status informant. This may agree with the theory of status characteristics of Berger, Cohen 8c Zelditch (1972). They claim that status characteristics can become relevant to an immediate (task) situation even when they are not directly relevant. Once they are associated with the immediate situation they influence performance and the exercise of influence (Berger, Cohen 8z Zelditch, 1972: 254). This may have been the case in the informal conversations. A third explanation for the lack of difference in asymmetry between the two interaction types may be that negotiation behavior is not indicative of conversational dominance. Trouble clarifications and confirmation checks were expected to be indicative of dominance. The results suggest that the negotiation move confirmation check may be an indicator of dominance. These checks were used most frequently by NSs and can influence the outcome of the conversation considerably because they rephrase the other's words. They may be a reflection of NS dominance and control. For trouble clarifications it is not clear; neither is it clear for the trouble indicators. An increase of NNS trouble indicators may also lead to an increase of the influence on what is negotiated. 6.3 Conclusions The results of this chapter show a complementary role distribution in negotiation of ineaning between NSs and NNSs. NNSs indicated trouble most frequently and NSs clarified it and used confirmation checks. Thus the NS manipulated content and form most frequently, independent of who caused the trouble. The most sustained asymmetry was found for confirmation checks. When the NSs had trouble understanding the NNSs, they used confirmation checks more often than trouble indicators. Such checks may give a speaker control over the content and direction of the interaction and thus may be indicative of NS dominance. The role distribution can best be explained by the different levels of communicatíve competence. Trouble indicating can be done with linguistically simple means. Pica (1987) claimed that such initiative on the part of the NNS in negotiation of ineaning is important for language acquisition. She stated the belief that learners increase their receptive and expressive capacities in a second language if they initiate a negotiation through which they may obtain comprehensible input. There is some development over time towards more symmetry for trouble indicating and clarifying behavior, particularly in the informal conversations. However, even in the third cycle, the NNSs' competence appeared to be so low that there was still role asymmetry. There is considerable individual variation. The results for Fatima, in particular, deviate from those of the other NNS informants. This may be 6.3 Conclusions 101 explained by the fact that she was the least competent speaker. In addition, there may have been a gender effect. Interaction type had some influence on symmetry in NS and NNS interactional behavior, particularly on trouble clarifications and confirmation checks, but less than expected. In the institutional conversations there appeared to be somewhat more sustained asymmetry, in particular, for the trouble clarifications. A factor which may have diffused the results is expert status. In the institutional conversations, the housing official Bart was a formal expert because he combined authoritative role with topic knowledge based on training and experience on the job. Paul, the other native speaker who acted as housing official in the first data session, had less expert knowledge because he was a social worker in real life. However, the NNSs were also informal experts on their own housing conditions and preferences. In addition, all conversations were semi-authentic; therefore, less asymmetry may have occurred than in authentic conversations in which both participants try to achieve their goals within a certain time frame. In the informal conversations, the NSs and NNSs could also be considered informal experts on topics concerning their own culture and life experiences. Therefore, in this study, institutional role may have been less clearly connected to authoritative role and topic knowledge than might have been the case otherwise. 102 Chapter 6 Asynunetry in Negotiation of Meanrng Chapter 7 Indicating Trouble in Understanding 7.0 Introduction In the preceding chapter the distribution of the main negotiation moves-trouble indicators, trouble clarifications, and confirmation checks-over the NS and NNS was discussed. It was concluded that NSs use trouble clarifications and, in particular, confirmation checks more frequently than the NNSs, whereas the NNSs use trouble indicators more frequently. In the following chapters a description of the forms and (strategy) types of the main negotiation moves will be given. In addition, the influence of communicative competence and interaction type on the realization of these negotiation moves will be described. Trouble indicators, trouble clarifications and confirmation checks are not homogenous categories. There are different ways in which a speaker may indicate or clarify trouble. These different types of indicators, clarifications and confirmation checks may also be realized with a variety of linguistic forms, the choices of which are probably not arbitrary but depend on the kind of trouble and may differ for the NS and NNS (Long, 1983b: 183). The research questions formulated in Chapter 3 concerning the three main negotiation moves are: 4a. How can NS and NNS trouble indicators, trouble clarifications and confirmation checks be described in terms of their types and forms? 4b. What is the influence of the NNS communicative competence and interaction type on the variety and quantity of trouble indicating, clarifying and checking behavior of the NS and NNS? These questions will be answered for each of the three main negotiation moves separately in the following chapters: trouble indicators in this chapter, trouble clarifications in Chapter 8, and confirmation checks in Chapter 9. For all three chapters, the structure will be similar. Firstly, a classification of (strategy) types and forms will be given. These classifications are developed to describe differences between NS and NNS moves based on differences in communicative competence, interaction type and other factors. The classification of forms is more or less equal for all three moves. Therefore, this classification is extensively discussed in Chapter 7(section 7.1.2) and only some additional information and examples are given in Chapters 8 and 9. A separate classification is given for trouble indicators types on the one hand in Chapter 7(section 7.1.1) and for trouble 7.0 lntroduction 103 clarifications and confirmation checks types on the other hand in Chapter 8(section 8.1.1). Secondly, the NS moves will be described. This gives some baseline information on how NSs use the main negotiation moves in Dutch. The NS moves are also compared per interaction type and individual portraits are presented to determine which factors influence NS trouble indicating behavior. Thirdly, the NNS moves will be described and compared for the two interaction types. In contrast to the section on the NSs, individual portraits will be based on the scores per cycle to show whether individual variation can be explained by NNS communicative competence. In other words, we wíll be investigating whether there is a development of the variety and frequency of types and forms of NNS negotiation moves over time in the direction of NS behavior. Fourthly, the NS types and forms will be compared to those of the NNS group to see if there are differences between both groups which could support the expectation that NNS communicative competence influences negotiation of ineaning behavior. In this chapter, the following research questions 4a and 4b will be answered for the part which concerns trouble indicators: 4a. How can NS and NNS trouble ituíicators be described in terms of their types and forms? 4b. What is the influence of the NNS communicative competence and interaction type on the variety and quantity of trouble indicating behavior of the NS and NNS? A trouble indicator was defined in Chapter 3 as a move which signals to the other speaker that there is a problem in understanding. 7.1 A Classification of 1~ouble Indicator Types and Forms Presented in this section is the classification of trouble indicator types, which emerged from the open coding. First some relevant literature will be discussed. Then the classification used for coding is presented (section 7.1.1). Section 7.1.2 contains a description of form characteristics with examples of trouble indicators; however, it will be applied to the other negotiation moves as well (Chapters 8 and 9). In section 7.1.3, some expectations are formulated about the outcomes concerning the use of NS and NNS trouble indicator types and forms in the different conditions. As a starting point for coding of trouble indicator types, a classification by Bremer et al. (1988 8c 1993: 168) was used. They distinguished explicit indications of non-understanding from more indirect symptoms of non-understanding. As indicators of non-understanding were classified: - metalinguistic comments (I don't understand) - minimal questions (what?) - reprise of trouble source (repetition) 104 Chapter 7 lndicating Trouble in Understanding - lack of uptake ( silence at transition relevant place). As symptoms of non-understanding were classified: - hypothesis forming ( best guessing) - minimal feedback (yes, mhm) - other markers such as topic switching and code-switching. This classification distinguishes explicit from implicit indicators. This is a relevant dimension because it refers to the degree of face-keeping which is necessary when doing such an intrinsically face-threatening act as indicating trouble. Gass 8e Varonis ( 1985a) made the same distinction but used the terms direct versus indirect indicators. Brown 8t Levinson ( 1989) have shown that all competent adult members of a speech community act rationally in that they choose certain means to achieve their goals and in that they have `face', which consists of two desires. Firstly, they desire the freedom of action and from imposition, which is called negative face. Secondly, they desire a positive consistent self-image through the appreciation and approval of others, which is called positive face. Face must be attended to in interaction. Face wants have to be satisfied by the other and thus a speaker will give face to the other in expectation that the other will do the same in exchange. According to Brown 8c Levinson (1989: 76), requests are likely to threaten the face of both speakers. Indicating trouble is a certain type of request, namely for clarification. It is an act threatening the negative face of the other by requesting help. In addition, with such requests a person admits lack of understanding which may make a person look incompetent, and they also threaten the speaker's positive face. By going on record (explicit indicator) a speaker can avoid being misunderstood in his intentions and thus pressure the other speaker to clarify the trouble. Therefore he is being efficient and non-manipulative. By going off record ( implicit indicator), a speaker can avoid face damage and avoid responsibility for his action (Brown 8c Levinson, 1989: 72). As Schegloff, Jefferson 8i Sacks (1977) stated self-initiated self-repairs are preferred in interaction. According to Faerch 8c Kasper (1985: 14), it seems logical that this preference relates to the face wants formulated by Brown 8r Levinson (1989). However, Faerch 8z Kasper (1985: 21) conclude that in NS-NNS discourse there may be less NNS self-initiated self-repair and more NS other-initiations and other-repairs, because the NNS simply lacks the linguistic competence to do selfrepaic Whether lack of communicative competence always leads to face threats is not clear. It may be that NNSs, like children, in certain contexts are not held responsible for not maintaining face. However, Gumperz ( 1990) has shown that particularly in institutional conversations between NSs and NNSs, many negative inferences are drawn when certain politeness markers are missing. Thus the classification proposed by Bremer et al. (1988 8t 1993) is relevant as far as the dimension explicitness versus implicitness of indicating trouble. However, several dimensions of differences which are also relevant for classification are not systematically distinguished. The reasons for this may be that Bremer et al. looked at NNS indicators only and were interested in interaction rather than language acquisition. 7.1 A Classification of Trouble Indicator Types and Forms 105 For the current study, which focusses on NS and NNS behavior from an interactive as well as an acquisitional point of view, some other distinctions are necessary. One dimension which emerged through the coding, and which is not present in Bremer et al.'s study, is the degree of specificity with which a problem is indicated. This dimension is relevant because it requires more linguistic competence to identify which part of a turn caused trouble than to react globally with `What?'. A specific indicator makes it relatively easy to locate the element which caused trouble. For the verbal indicators the dimension of explicitness and specificity lead to a basic distinction of categories which will be further discussed below. Secondly, in Bremer et al. (1993) nonverbal indicators are incorporated in the indications of non-undersianding. From an acquisitional point of view, it is logical to make a separate category for nonverbal indicators because they show no linguistic form development. This results in three dimensions on which trouble indicator types can be described; as can be seen in Figure 7.1, the classification leads to five types. i trouble indicators nonverbal global implicit explicit implicit explicit F'igure 7.1: Dimensions of trauble indicator tt~pes Thirdly, from an acyuisitional point of view, it is relevant to distinguish linguistic form from trouble indicating types in this study. For example, in Bremer at al.'s categorization, `minimal question' refers to form while the category `reprise of trouble source' refers to a strategy (repetition). In the current siudy all verbal negotiation moves will be classified for type and form. Fourthly, categories such as topic switching may function as a symptom of non-understanding. However, it does not initiate a clarification seyuence as trouble indicators do. On the contrary, its goal is to escape from the trouble. Topic switching (escape) in the current study is therefore categorized as a separate negotiation move. Fifthly, on negotiation 106 Chapter 7 htdicating Trouble in Understanding move level a distinction between trouble indicators and confirmation checks will be made based on Long (1983c). In Bremer et al. (1988 8t 1993) confirmation checks, which are guesses of the other speaker's intended meaning, are classified as a symptom (hypothesis forming). This distinction may have been less relevant in Bremer et al.'s study because they only looked at NNS indicators. However, as was seen in Chapters 4 and 6, NSs have different preferences in dealing with trouble source because they use confirmation checks more regularly than trouble indicators. Therefore, this distinction is relevant in the current study. 7.1.1 ZYouble Indicator Types In this section, the classification for trouble indicator types will be described. Firstly, the nonverbal indicators will be described. Then the four verbal categories based on a combination of the two dimensions implicit-explicit and global-specific will be described. Nonverbal Indicators of Trouble Trouble indicating may be done by nonverbal means. It can be done with gestures (open hands), body movements ( raising shoulders), and facial expressions such a raising eyebrows, moving lips, blank looks, or by looking away. In addition, the use of hesitation markers such as `uh', or a silence at a transition relevant place are taken into account. When no visual information was available only the latter signals could be depended on. Usually silences were coded as trouble indicators if they occurred instead of responses which should have followed a question. Such silences were often followed by clarifications, indicating the interlocutor interpreted the silence as trouble indicator. According to Schegloff, Jefferson 8t Sacks (1977: 374) often extra time is allowed for self-repair before the interlocutor starts clarification. Nonverbal indicators are not categorized any further because the focus was on linguistic form. In addition, complete visual information was not available because videotapes were made of the institutional conversations only (except for the second conversation with Fatima). However, especially in the earlier conversations, nonverbal indicators occur quite regularly at transition relevant places for the NNS because no linguistic competence in the target language is needed to convey non-understanding. Therefore, nonverbal trouble indicators are included in the quantitative analysis. specificity global specific explicitness implicit explicit global á implicit global 8t explicit specific á implicit specific á explicit Figure 7.2: The verbal trouble indicator t~~pes 7.1 A Classification oJTrouble Indicator Tvpes and Forms 107 As stated before, the types of the verbal indicators could be categorized on two dimensions (see Figure 7.2). The first dimension concerns explicit versus implicit indication of trouble. This distinction refers to on-record and off-record behavior for face threatening acts (Brown 8z Levinson, 1989: 68). Going on record, a speaker explicitly indicates that there is a problem in understanding through a direct reference to trouble, such as 1 don't understand. The communicative intention of the speaker is in this case clear. A speaker can also go off record and indicate trouble implicitly through questions such as mhm?. These minimal feedback questions request a repetition but do not state the reason why a repetition is needed. In this case more possible communicative intentions can be attributed to the speaker. He might not have been listening because he drifted away with his [houghts. Such implicit indicators proteei the face of the speaker who dces not have to admit that his lack of communicative competence causes problems in understanding. The second dimension concerns the degree of specificity with which the trouble is indicated. A trouble source `X' may be specified through an indicator such as What does X mean?. However, the indicator What? does not give the interlocutor a clue as to what the problem is. Generally, trouble indicators follow directly after trouble sources, so it is at least clear which turn causes the trouble. However, within a turn a specific element X, a word, morpheme or phoneme may cause trouble. A specific indicator refers directly to this element X. Thus, the division on two dimensions resulted in four categories of verbal trouble indicators, as can be seen in Figure 7.2. Global á Implicit Indicators of Trouble Into the category of global 8c implicit indicators fall all minimal feedback questions such as: What? Mhm? Huh?. These questions are ambiguous as to the reason of the request because they may indicate non-understanding or a lack of attentiveness, and thus they protect face. They do not give the interlocutor an idea of which element of the input is problematic and leave it to the other speaker to decide what needs to be clarified. Specific á lmplicit Indicators of Trouble The category of specific 8c implicit indicators encompasses all requests for explanation (why?), specification (who?), repetition (what did you say?), fill-inthe-blank questions, etc. These requests give an indication to the interlocutor of what kind of information is needed to achieve understanding, but they protect the face of the interlocutors because no explicit mention of non-understanding is made. In Example 1, Ergun used an implicit request for specification. (1) Informal conversation with Ergun, Cycle 2 ts Janet: is er verschil tussen die mensen en zoals ik praat? ti ErRun: t wie? 108 is there a difference between those people and how 1 talk? . who? Chapter 7 Indicating Trouble in Understanding In addition, utterances which are used to request a specific word from the interlocutor were coded as specific 8c implicit indicators. In Example 2, Fatima used a word request when the NS and NNS were talking about medicines. (2) Informal conversation with Fatima, Cycle 2 tslti Fatima: met uh ~dcea~ -~m-medicine~ ~ash smitce~ -~m-how do you call it?~ tablet of with uh ~dcea~ -~m-medicine~ ~a.eh smitce~ -~m-how do you call it?~ tablet or These word requests are self-initiated other-repairs which are a trouble source and trouble indicator in one, while all the other indicators are other-initiations of repairs following the trouble source in the next turn. Such requests are also called word searches (Day et al., 1984: 30) or appeals to authority (Tarone, 1980). Global óc Explicit Indicator of Trouble Indicators which explicitly indicate non-understanding but do not give the interlocutor a clue regarding what was not understood were categorized as global 8z explicit indicators. Utterances such as I don't understand belong to this category. The speaker goes on record. These are what Bremer et al. (1993) called metalinguistic comments. Specifec á Explicit Indicator of Trouble Utterances which not only explicitly indicate non-understanding but also tell the interlocutor what the trouble source might be were categorized as specific 8c explicit indicators. For example, ! don't understand X. Of course, it may not always be the case that the problem X indicated by the speaker is the only problem, but such indication does give the interlocutor a specific starting point in clarifying the trouble. With these coding categories the type of almost all trouble indicators could be coded by type. Other utterances which could not be coded as one of the five categories were coded as `Other'. These were utterances which were partly unintelligible, so they could not be coded on the two dimensions of verbal types of indicators. This `Other' group is included in the quantitative results (see Tables) but were not further analyzed. 7.1.2 Trouble Indicator Forms In this section, the form aspects of the three negotiation moves, trouble indicators, trouble clarifications, and confirmation checks, are classified. For the forms of the negotiation moves a first simple distinction on communicative function was made in declaratives, interrogatives, and imperatives (Geerts et al., 1984; Quirk 8c Greenbaum, 1974). Quirk 8c Greenbaum (1974) use a fourth category, exclama[ions. These were not present in the analyzed clarification sequences. The other 7.1 A Classification of Trouble Indicator Types and Fonns 109 three categories were subdivided where relevant. In particular, for the intetrogatives a subdivision is necessary because there are many different question types which can be used in negotiating meaning. Some additional categories were added for unclassifiable utterances which aze partly unintelligible or which are in the source languages. The categories described below (see also Figure 7.3) are applicable to all negotiation moves but mainly exemplified by trouble indicator moves. In Chapters 8 and 9 examples for the classification of the other moves, trouble clarifications and confirmation checks, are given where relevant. Full declaratives Elliptic declaratives Declaratives Linguistic fortn Question word Wh-questions Inverse word order Or-choice questions Yes~no questions Question intonation Declamtive questions Interrogatives Incomplete sentence tuzicturc Elliptic questions ~ Minimal feedback questiorts Fill-in-the-blank questions Negazive questions Other Tag questions Indirect questions fmperatives Nonverbal Zero category Other Figure 7.3: List ofform categories 1 10 Chapter 7 Indicating Trouble in Understanding Declaratives A declarative utterance is any statement with a verb and subject such as `Ik begrijp het niet' (I don't understand). Declaratives can also be elliptic. When subject andlor verb were absent, such declaratives were coded as elliptic declarative. Bygate (1988) argues that oral language tolerates ellipsis more than written language because the speakers in face-to-face interaction share knowledge of the situation and have the ability to solve problems together. Therefore, he states that oral syntax is negotiable. He calls utterances which are not full finite sentences `satellite units' (Bygate, 1988: 64). In interaction with NNSs use of such units may occur more frequently than in NS interaction because smaller units can be more easily produced and processed. Interrogatives Interrogatives are utterances to which an answer is expected. They aze generally characterized by a question word, inverted word order or rising intonation (Quirk 8z Greenbaum, 1974). An additional category which was added from an acquisitional perspective is whether the question has a full form or not. Trouble indicators are usually requests for clarification. Therefore, many different questions were used. Through the open coding, the interrogatives were further classified according to several relevant characteristics. The first is the presence or absence of a question word. An instance of the former are Wh-questions, such as Wat bedoel je met X? (What do you mean by X?). The Wh-questions are the most complex question type and may contain several different question words which locate references in trouble sources. These are for example, wat (what), wie (who), waar (where), welke or wat voor (which), waarom (why) and hoe (how). The minimal Wh-question what? was excluded from the Wh-question category because it functions as a minimal feedback question (like huh? see below) and little linguistic competence is needed to use it. All other questions which contained a question word, even if they were elliptic, were coded as Wh-question because of the specific character of the answer it requires. The second group of questions have inverted word order and no question word. First, there are or-choice questions Ben je getrouwd of verloofd? (Are you married or engaged?). These questions present the interlocutor with two alternative response models. Furthermore yeslno-questions such as Benje getrouwd? (Are you married?) have inverted word order. They are generally reacted to with a`yes' or `no' answer. The third group of questions are characterized by normal word order and (sometimes) by rising question intonation. These are called declarative questions, e.g. Je bent getrouwd? (You are married?). Quirk 8r Greenbaum (1974) state that rising intonation (without inversion) is characteristic of declarative questions. However, Beun (1990) did not include question intonation in his definition of declarative questions in Dutch because he found in his study of flight information dialogues that about SOcIo of the declarative questions had no question intonation. He therefore defines declarative questions in the following way: the sentence type is declarative; a. b. the utterance is uttered by a speaker about a topic on which he believes that the hearer is the expert; 7.1 A Classification of Trouble Indicator Ty~pes and Forms I 11 c. the speaker believes that the speaker and heazer mutually believe that the hearer is the expert. This definition is also used in the cutrent study. Often the characteristic `expertise' is indirectly expressed by the second person pronoun (Beun, 1990). This can be seen in Example 3 in which Mahmut used the incotrect form eenduizend (one thousand). The correct form for `one thousand' in Dutch is duizend (thousand). Paul was not sure whether Mahmut referred to his income. T'herefore, he checked and used a second person pronoun. He rephrased Mahmut's intended meaning and modelled the correct form. (3) Institutional conversation with Mahmut, Cycle 1 ts Mahmut ik eenduizend gulden hè (...) tc Mahmut: een maand eenduizend gulden Paul: cc t en je verdient duizend Rulden? co Mahmut ja ~eero duizend gulden he - ~emphatic~ ac Paul: ja duizend gulden ja ac Mahmut: duizend hè 1 one thousand guilders huh one month one thousand guilders t and you earn a tlurusand guilders? yes ~one~ thousand guilders huh - ~emphatic~ yes a thousand guilders yes thousand huh As stated above, declarative questions do not always have question intonation. Those without question intonation are harder to distinguish from declaratives. In general, utterances concerning knowledge on which only the hearer was an expert were coded as declarative questions. Other utterances without question intonation, although sometimes followed by a confirmation, were coded as declaratives. In Example 4, Fatima explained to Janet that acquiring a house in Morocco was not easy because the people were fleeing from the Sahara. When Janet asked why, she did not get a clear answer and hypothesized, based on her own knowledge, that it may have to do with the conflicts with the independence movement Polisario in the area. This Fatima confirmed with her paraphrase. (4) Informal conversation with Fatima, Cycle 2 Fatima: lanet Fatima: cc Janet Fatima: alles betaald weg van sahara waarom? daazom van veel te geld kapot van sahara marokkaans sahara jajaja rnnlnmdat daar oorlog is met ruzie maken met uh polisario everything paid gone from sahara why? that's why from much too money kaput from sahara moroccan sahara yeah yeah yeah be~because titere is war with quarrel with uh polisario If the question did not contain a subject and verb, it could not be determined whether it was a yesJno or declazative question. Therefore, such utterances were coded as elliptic questions or declaratives (see Example 5). (~ Informal conversation with Ergiin, Cycle 1 lanet: duurt het feest een dag? Ergun: een dag en t uh t twee Idrie uur beginnen en [t] cc Janet: ['s middags?J Ergiin: ja middags 1 12 dces the party last one day? one day and t uh t two Ithree o' clock begin and [t] [in the afrerna~m?J yes aftemoon Chapter 7 Indicating Trouble in Understanding In this case, Ergun by repeating middags explicitly showed that he has understood Janet. The question intonation of Janet's check also shows some uncertainty about Ergiin's intention. The second and third group of questions all manipulate content and form and present the interlocutor with a model for an extended answer (Ja), ik ben getrouwd ((Yes), I am married). Thus the interlocutor does not have to manipulate content and form himself. He can just repeat the model. However, generally a short ja (yes) or nee (no) will do. A fourth group of questions could be distinguished in the data which do not adhere to any of Quirk 8c Greenbaum's three conditions (1974). They are characterized by incompleteness of the sentence structure. They often consist of one word. These are: - elliptic questions Getrouwd? (Married?) (see above) - minimal feedback questions Hm? (Mhm?), Wat? (What?), Ja? (Yes?). - fill-in-the-blank questions Je bent ... ?(You are ...?) With this last type of question, the speaker requests a repetition of a specific part of the interlocutor's contribution. In Example 6, Ergun stated the acronym of a TV station SKV which in Dutch would be pronounced as [Es ka. ve.] (Van Gussenhoven, 1992). However, he had the wrong pronunciation for all three letters, `S' `K' and `V'. The way he pronounced `S' caused confusion because it is the same as the pronunciation of `C' in Dutch. (6) Informal conversation with Ergiin, Cycle 3 tc ErgGn: ce ke fe ti Janet ce [t) tc Ergun: ke fe ~p- se. ke. fe.~ ~p- se.~ ~p- ke. fe.~ The fifth type of question is negative questions such as Ben je niet getrouwd? (Are you not married?) or Je bent niet getrouwd? (You are not married?). These are coded separately because it was expected that the negative element would cause problems for NNSs. In particular for negative yes~no questions this is expected to be the case because it is not always clear what a`yes' or `no' answer entails after such a question (Celce-Murcia, 1983: 110). In Dutch a`no' answer to the question `Are you not married?' would mean `I am not mamed'. A plain `yes' answer is not possible. Some other question types occurred very rarely. These were: - tag questions Nou maarje wilt alles hè? (Well, but you want everything, don't you?l - indirect questions Weet u hoelang dat u ingeschreven stond? (Do you know how long you've been registered?). Imperatives An imperative or command can be distinguished by two characteristics (Quirk 8c Greenbaum, 1974): 1. it has no subject; 2. it has an imperative verb form. 7.1 A Classifcation of Trouble Indicator Types and Forms 113 Imperatives may be used to indicate trouble, for instance, for commands like Zeg dat nog eens ( Say that again, please). Nonverbal Furthermore, a zero category was used for the nonverbal trouble indicators which could not be coded for linguistic form. This category did not occur for the trouble clarifications and confirmation checks. Otherforms In addition, an `Other' category was created for negotiation moves in languages other than Dutch and other partly unintelligible indicators (Other). This category was only included in the quantitative analysis. 7.1.3 Expectations Several expectations about the outcomes can be formulated. Trouble indicating moves may be influenced by face concerns and goal realization in opposing ways. For face reasons, one might expect trouble indicating to be done ambiguously and covertly. However, for the effective clarification of the trouble it may be necessary for the speakers to use more explicit and specific means to indicate trouble. Several factors may influence the choice between types of trouble indicators. Firstly, the NNS communicative competence may play a role. The NS may display a broader range of indicator types and forms appropriate for the type of trouble source and the NNS's competence, because the NS has the communicative competence to adjust indicators. The NS generally may opt for implicit indicators because trouble indicating is an intrinsically face threatening act. However, he may also use explicit indicators if he expects that the NNS will misunderstand an implicit signal or has earlier not reacted to such a signal. The NS will use specific indicators which will help the NNS to identify the trouble source. To realize such specific indicators, the NS may use several types of questions, in particular, Wh-questions which help ïn identifying the type of missing information (location, time, manner, etc.). When his competence is low, a NNS may only be capable of indicating trouble through nonverba] or global means. The NNS in the initial stages of language development may use implicit and global means for indicating trouble. He does not have the linguistic means to be specific and is vulnerable to losing face because of his lack of competence. He may not want to openly admit non-understanding. For this reason, he may also use nonverbal indicators regularly. Over time, however, as the NNS's communicative competence increases, he may use more specific indicators. He is expected to develop in the direction of the same range of trouble indicators the NS uses. He may also start using linguistically more demanding types and forms, such as Wh-questions. Secondly, interaction type may play a role. In an institutional conversation the roles of the NS and NNS may intluence trouble indicating behavior. The NS may be less face protective in ins[itutional conversations because of having a higher 114 Chapter 7 Indicating Trouble in Understanding status (Brown 8c Levinson, 1989: 77) and thus the housing officials may use more explicit indicators. The NNSs may be more face protective because of status differences and thus use fewer explicit means in the institutional conversations. In the informal conversations, in which there is less social distance, the NNS may be more explicit (Brown 8c Levinson, 1989: 76). This may be particularly true in the third cycle, where the NNS had gotten to know the NS Janet. 7.2 Native Speaker T~-ouble Indicators In this section, first an overview of frequencies of NS trouble indicator type and form combinations is given. Next, the frequencies of types and forms will be compared for the interaction types (section 7.2.1). Then portraits of the individual NSs are given to describe individual variation (section 7.2.2). No statistics were calculated to determine the significance of the frequency scores. There are several reasons why it was decided not to use statistical tests, such as the chi-square test. Firstly, the cell frequencies are often very low. Secondly, the cell scores do not always represent independent observations, because the negotiation moves are organized in sequences in which one move determines the preferences for the next one. Thirdly, scores which are collapsed from four different speakers cause a problem for statistics because an individual can strongly influence the average. Therefore, only raw frequencies and percentages are presented to give some insight into the distribution of types and forms over the different speakers. In Chapter 6, it became clear that the NSs did not use trouble indicators frequently. Therefore, the total score is quite low (n-110) compared to the NNSs' total of 435 trouble indicators. However, as can be seen in Table 7.1, the NSs used quite a range of trouble indicator types and forms. The question categories are ordered from left to right presenting the linguistically least to the most demanding question type. Because of the small number of declaratives, the figures for full and elliptic declarative are not split out here. In addition, categories which did not occur are not presented in the tables. In almost SOolo of all cases the trouble indicators were specific 8c implicit indicators, such as requests for repetition [wat Zeg je? (what do you say?)], for specification [wie? (who?)] and explana[ion [wat bedoel je? (what do you mean?)]. These almost all had the form of a question. Several question types were used for these requests but most frequently used were Wh-questions. The Wh-questions were usually short, containing solely a question word and sometimes a verb andlor subject. However, elliptic questions, minimal feedback, and fill-in-the-blank questions were also used by the NSs as specific 8z implicit indicators. This last type of question was used to give the NNS a model in which a word not understood could be repeated. In Example 7, Bart used such a fill-in-the-blank question. 7.2 Native Speaker Trouble Indicators 115 Trouble indica[or type Linguistic form Nonverbal indicator .~ ~ ó z tn C ~ ~ o ~ ~ ~ ï ii u' g .;, .o ~ ~ ~~ ~ "~ ~ ,4 ~ ~ ~ 3 ~ É ~ F ~ t2 tp.9 ts 16.a St 46.a t2 Global 8t implicit indicator t7 Specific 8z implicit indicator B t 5 8 3 Global 8c explicit indicator SpeciÏtc át explicit indicator 1 Other Total ol0 C ~ ~ 22 t 2 2 3 4 7 6.a l6 2 19 17.3 1 3 2.7 Ito I 2 Iz 2s 7 9 3 42 7 z 3 1 p.9 22.7 6.4 8.2 2.7 38.2 6.4 1.8 2.7 100.0 Table 7J: Tj~pes andforms of NS trouble indicators (7) Iastitutional conversation with Mahmut, Cycle 2 ts Mahmut: ja t die gas hè yes t that gas huh hele duur hè very expensive huh per maand per month tweehonderdzestientwi ntig two hundred and sixteen twenty gulden guilders ti Bart: lun tweehonderd (t J? mhm tx~o hundred [f J? tc Mahmut: zesentwintig twenty six If the NSs indicated trouble, they apparently tried to help the NNSs by identifying the trouble source and by protecting their faces by generally doing it off record. Global 8c implicit indicators were used regularly (18c1o). Most had the form of minimal feedback questions [e.g. wat? (what?) or hè? (huh'?)]. Specific 8c explicit indicators were used about as often as global 8c implicit ones. Janet used simple Wh-questions for these indicators in cycles 1 and 2(see Example 10). She also used longer questions in cycle 3 with Mahmut, as can be seen in Example 8. (8) Informal conversation with Mahmut, Cyclc 3 ts Mahmut hij zegt elke week vijfenveertig lachtenvee rtigd uizend of vijftigduizend beginnen niet zegt ri Junet: wut bedne! je met beginnen? - ~NNS means winst (profit)~ he says every week forty-five forty-eightthousand or fifty thousand begin not says wlw! dn cnu ntean by begin? - ~NNS means winst (profit)~ When Bart fotmulated specific and explicit requests in the institutional conversations, he sometimes used reduced non-grammatical forms for a complex construction such as `what do you memt b}~ x'?' ( in Dutch: `wat bedoel je met x?'). This ungrammatical use can be found in Example 9. 1 16 Chapter 7 lndicnting Trouble in Understanding (9) Institutional conversations ti Bart: wat dun? or Bart: wat driehnnderd af? ti wluit thin? what three hundred off? Global 8c explicit indicators were used only Solo of the time. These were most freyuently declaratives, e.g.: nee f dat snap ik niet zo goed (no f that I don't quite get) (see Example 11). In 12qo of the cases the NSs used nonverbal means (see Example 11). This is quite high score considering the fact that the NSs have the linguistic competence to ina dicate trouble verbally. This may show a general preference for minimal means of trouble indicating which can be done with little effort, causing minor interruptions. In conclusion, the NSs, as expected, used a variety of trouble indicators. There was a strong preference for specific implicit indicators, and global explicit indicators were used the least. The next section will show whether this distribution is found in both interaction types. 7.2.1 The Influence of Interaction Type As can be seen in Table 7.2, both in the informal and institutional conversations, about SOqo of the trouble indicators were specific 8t implicit. In addition, the distribution of global and explicit indicators is equally minimal. These equal outcomes are remarkable because different speakers are compared in different interaction types. However, as expected, there are different patterns of distribution for the other NS trouble indicator types. In the informal conversations, in 210l0 of the cases specific 8c explicit indicators were used. In addition, nonverbal indicators were used relatively frequently. Global 8c implicit and global 8z explicit indicators were each used only 7qo of the time. Informal conversations NS trouble indicator types Institutional conversations ~ ~o ~ aIo Nonverbal 6 14.3 6 8.8 Global 8t. implicit 3 7.1 15 22.1 Specific 8c implicit 20 47.6 31 45.6 Global 8c explicit 3 7.1 4 5.8 Specific 8c explicit 9 21.4 10 14.7 Other 1 2.4 Total 42 1On.0 2 68 2.9 100.0 Table 7.2: Frequencies ofNS trouble indicator types per ínteraction h~pe 7.2 Native Speaker Trouble Indicators 1 17 In the institutional conversations, in contrast, quite a few global 8t implicit indicators (22010) were used. Specific 8c explicit indicators were used less than in the informal conversations. In general, the NSs used fewer explicit indicators than in the informal conversations. It was expected that the NSs in the institutional conversations would go on record and use more explicit indicators because their expert status would make them less face protective. This turns out not to be the case. On the contrary, trouble indicating by the NS appears to be more of a face threatening act when there are status differences. Nonverbal indicators were used less in the institutional conversations. Informal conversations NS trou ble indicator forms Institutional conversations tt ~Ic ik Nonverbal 6 14.3 6 8.8 Minimal feedback question 6 14.3 19 27.9 Fill-in-the-blank question 4 9.5 3 4.4 Elliptic question 6 14.3 3 4.4 Or-choice question 1 2.4 2 2.9 Wh-question 11 26.2 31 45.6 Declarative 5 11.9 2 2.9 Imperative 0 0.0 2 2.9 Other Forms 3 7.1 0 0.0 fi4 1(1~L(1 , T~~t.~l ,1? ~ lI)q i qn Table 7..3: Frequencies of NS trouble indicators fonns per interaction type As can be seen in Table 7.3, it was found that NS trouble indicators are most frequently questions. However, in the informal conversations, this was `only' the case for ó7qo of the NS trouble indicators, whereas in the institutional conversations 85~1o were questions. This may have been related to the interview format, which made the housing official request detailed information on issues such as housing locations, length of stay, or family structure. In the informal conversations, in 25qo of the cases, Wh-questions were used. In addition, elliptic and minimal feedback questions were used respectively about lOqo and 15qo of the time. Furthermore, fill-in-the blank questions were used almost lOqo of the time. In the informal conversations, other forms were also used in about 30qo of the cases to indicate trouble. In 15qc of the cases nonverbal cues were used. Declaratives were used about lOqc of the time. These may have been used for specific and explicit indicators. Some uncodable trouble indicators in 118 Chupter 7 Indicating Trouble in Understanding ~ other languages were used. However, imperatives were not used. In general, the NSs used a greater variety of forms in the informal conversations. In the institutional conversations, the NSs used Wh-questions even more frequently than in the informal conversations (50c1o). Furthermore, in about 30qo of the cases minimal feedback questions were used. Other questions were used rarely. Nonverbal trouble indicators were used in about lOqo of the cases. Declaratives and imperatives were rarely used. The role of the NSs in the institutional conversations and the goal of having to determine the urgency of a case, may have led the NSs to use so many Wh-questions. If there was no need for Wh-questions, the housing official appeared to use minimal feedback questions. It is remarkable that there is little variety in question use in these conversations. Maybe the interview format led to such homogeneity in form. This contrasts with the informal conversations, where the NSs used a variety of (question) forms, such as elliptic, fill-inthe-blank, and declarative questions. In conclusion, the distribution of specific 8r. implicit and global á explicit indicators was equal over the two interaction types. In the informal conversations, nonverbal indicators and specific 8c explicit indicators were regularly used. In addition, more fill-in-the-blank questions were used. There was more variety in forms in the informal conversations. In the institutional conversations, by contrast, many global 8i implicit indicators were used. The forms used were mainly Whquestions or minimal feedback questions. 7.2.2 Individual Portraits of the Native Speakers In this section, some of the individual characteristics of the NSs are described to investigate whether the NSs have different personal preferences for certain trouble indicator types and forms. The scores for Peter were not analyzed separately because he only substituted for Janet in one session. The results for the individual native speakers showed different accents due to personal style and their role in the interaction. Their scores are not presented in tables but are added to the text in parentheses where relevant. Janet The female researcher, Janet, participated in all but one of the informal conversations. She had experience with NNSs working as a minorities consultant for an Educational Center. She used several types of requests but mainly specific 8z implicit indicators (54qo). She used echo questions quite regularly. These questions repeat a keyword of the other speaker's trouble source (Brceder, 1993: 84). In Example 10, Ergun used a Turkish word biskuvi (biscuit) for the color `beige' because that is what he thought he heard his colleagues use in the factory. Both the Turkish word bej and the Dutch word beige are French borrowings and are pronounced almost the same in both languages. Therefore, it is curious Ergun came up with the term biskuvi, which is also a French borrowing in both languages (Dutch - biscuit) but pronounced slightly differently. It 7.2 Native Speaker Trouble Indicators 1 19 might be that the factory workers indeed used the word biscuit for the color but this is not common usage. Janet was not sure what he meant, so she used the keyword strategy (specific 8c implicit indicator). (10) Infotmal Conversation with Ergun, Cycle 2 ts Ergun: of wit hier t zwart hier t biskuvi hier ti lanet waz? tc Ergun: biskuvi kleur biskuvi ti Janet biskiivi? Ergiin: ja ti Janet: wat is biskiivi? tc Ergun: die kleur hè of (xx) mc ]anet: is dat turks? mc Ergun: nee dat is nederlands ti Janet: biskiiví? tc Ergun: ja kleur zo t zelfde zo papier t zelfde beetje licht 1aneC beige? cc ti Ergun: ~beige~ - ~imitates~ ti ik weet ik niet tc maar daar ook die alle lalle mensen biskuvi or white here f black here biskiivi here what? biskuvi color biskiivi biskuvi? yes what is biskiivi? that color huh or (xx) is that turkish? no that is Dutch biskiivi? yes color like this t like this paper t like this a bit light beige? ~beige~ - ~imitates~ 1 ! don't know but there too all the lall the people biskuvi The keyword strategy is one of the minimal linguistic strategies to specifically indicate a problem. Janet's strategy helped the NNS by identifying the trouble source. She also used many specific 8c explicit indicators (2ó010) such as in the fotmer example. She only rarely used global Bc implicit (8qo) and global 8t explicit indicators (óqo). Her very frequent use of specific indicators may show her cooperative attitude. This cooperative attitude may be based on her `teacher-like' attitude. Yet there may also be a gender influence, if there is truth in the belief that women are more supportive (Thorne 8z Henley, 1975). As far as trouble indicator forms are concerned, Janet used mainly Whquestions but also elliptic, minimal feedback and fill-in-the blank questions. The last type of question may also be a typical teacher-like strategy. Paul Paul was not a housing official but a social worker in real life, who only participated in the first cycle of the housing office conversations. He had had experience with NNSs as a social worker and thus may have been somewhat familiar with their housing problems. He, like Janet, mainly used specific and implicit indicators, which support the NNS in identifying the trouble source and protecting face (48qo). However, Paul also used nonverbal (14qo) and global 8r explicit (19qo) indicators regularly, which do not help the NNS in identifying the trouble source. These categories can be seen in Example 11. Within one sequence Paul first used a nonverbal (implicit) indicator but when Mahmut's clarification did not help, he resorted to an explicit indicator. This order of moving from implicit to explicit indicators was found frequently within sequences for the NNSs as well as for the NSs. (11) Institutional conversation with Mahmut, Cycle 1 ts 120 Mahmut: die uh veel duizend die tweeduizend uh f that uh many thousand that two thousand uh t Chapter 7 Indicating Trouble in Understanding ti tc ti ti Paul: Mahmut: Paul: een duizend vijfhonderd gulden" normaal flat hè t3t betalen niet nee t dar snap ik niet zo goed wat bedoe! je nou? one thousand five hundred guilders" normal apartment huh t3t pay not no t that 1 don'r ger now what do you mean? In Paul's global 8c explicit trouble indicator, the word nou (now) shows Paul's impatience, which threatens Mahmut's negative face even more. Paul also used global 8c implicit indicators IOolo of the time. In using global indicators, Paul put the linguistic task of determining what the trouble source is on the shoulders of the NNSs. Paul is also the only NS who twice indicated trouble through imperative fotms. As can be seen in Example 12, he baldly went on record because he used the most direct unambiguous and concise way to indicate trouble. He did not try to compensate by giving face to repair potential face damage. In addition, he did not indicate specifically what the problem was. (12) Institutional convetsation with Mahmut, Cycle 1 ts Mahmut: ti Paul: babylmijn meisje baby die twee t vijf~ vijf maanden hè baby terug ja t over~ieg het nog eens babylmy girl baby that two t fivel five months huh baby back yes t inlsay it once again Imperative forms are common in requests in Dutch if they are softened by the use of understatement particles such as tnaar (just) and even (a moment) (Van der Wijst, 1991). However, such softened imperatives may still be experienced as too direct in asymmetric conversations. Still, it remains a question whether less proficient speakers pick up such subtleties through linguistic form. They may, however, be able to interpret accompanying intonation, which in this case may have showed impatience. These behaviors may show that Paul is the least co-operative and face protecting NS. Bart The housing official Bart is a goal-oriented interviewer who is experienced in interviewing NNSs at his job. Of all the types of trouble indicators, he most frequently used specific 8c implicit indicators, but in relatively fewer cases than Janet and Paul (45010). He also used specific 8~ explicit indicators (21 qo). These were generally Wh-questions but with minimal forms (see Example 6 wat dun? - what thin?). Thus, contrary to Paul, he was very supportive in identifying trouble sources. In addition, he used global Rc implicit indicators more than the other NSs (25qo of the cases). These were mainly minimal feedback questions. He appeared to be quite co-operative. He did not once use a global 8c explicit indicator. In conclusion, the native speakers in indicating trouble generally stayed off record and used implicit indicators. Their choice of implicit indicators shows a sensitivity for face matters. The NSs used more explicit indicators in the informal conversations but the difference is not so great. However, the smaller social distance in these conversations may have led to such directness. Most indicators the NSs used were also specific. They identified the exact trouble source for the NNSs and thus helped them with the clarifi- 7.2 Native Speaker Trouble Indicators 121 cation. Global á:: explicit indicators were used the least, so the NSs were generally cooperative and face protecting. They preferred specific and implicit indicators. However, they also employed other types when appropriate to the context. Paul appeared to be the least co-operative and face protecting NS because he used global8c explicit indicators even with imperative forms. The NSs generally used simple questions to achieve their goals. They quite regularly used complex question types such as Wh-questions, but the transcripts showed that these were kept as short as possible. The housing official, Bart, is the only one who regularly realized these questions by using reduced ungrammatical forms. Furthermore, the NSs used fill-in-the-blank questions to alleviate the NNS's linguistic task of clarification. This may indicate that they judged the NNS competence to be quite low. Therefore, the NSs took on the linguistic load themselves as much as possible. 7.3 Nonnative Speaker Trouble Indicators In this section, first an overview of frequencies of NNS trouble indicator type and form combinations is given. Next, the frequencies of types and forms will be compared for the interaction types (section 7.3.1). Then portraits of the individual NNSs are given to describe individual development and variation (section 7.3.2). As can be seen in Table 7.4, on most occasions when a NNS had a problem in understanding, this problem was indicated by nonverbal means (38o10). Trouble Linguistic form indicator type Nonverbal indicator ~ L ~ ó z ~ ~ ~ Q ~ :-n. m ~ O. ~ ~ a~ ó ' c ,, o ~ ~ o N ~ ; ~~ ~ ~~ ~ ~ ~ ~ .o ~~ 167 Global 8t implicit indicator Speciilc óC imphci[ indicator ~ Q~ ~ t t7 I IS t 32 3 1 15 1 38 2 ~ F s t67 38.z I ls 27.1 I08 24.8 22 5.t G1oba18c explícit indicator 3 19 Specific 8t explicit indicator 12 3 1 16 3.7 1 2 a o.9 435 Other 1 Total 168 132 33 3 1 31 I 61 5 cIc 38.6 30.3 7.6 0.7 0.2 7.1 0.2 14.1 1.2 100.0 Table 7.4: T}nes and forms of NNS trouble indicators 122 Chapter 7 lndicating Trouble in Understanding Often nonverbal means were used when a NS would ask a question and the NNS did not know how to respond to it. The NNS often remained silent at such a transition relevant place. In 27qo of the cases a global 8c implicit indicator was used in the form of a minimal feedback question such as hè? (huh?) which is linguistically the simplest way to indicate trouble. In 25qo of the cases, specific 8z implicit indicators were used with several forms. Most of them were questions, but word requests which are part of the trouble source's turn were mainly declaratives. Declaratives were also used frequently for global 8c explicit indicators such as ik begrijp niet zo goed (I don't quite get). Specific 8c explicit indicators were used in only 4010 of the cases. This low score may be a result of the level of difficulty such indicators present. A trouble source has to be repeated and trouble indicated explicitly, as can be seen in Example 13. (13) Institutional conversation with Mahmut, Cycle 2 ts Bart: ri tc mr Mahmut: Bart: Mahmut: plcegendienst ploeg niet weten ik ploeg plcegendienst ja ik niet plcegendienst hè shiftwork shift not know 1 shift shiftwork yes I not shiftwork huh In conclusion, the NNSs most frequently indicated trouble in nonverbal and global ways which may be caused by their lower linguistic competence. The global indicators were realized by minimal feedback questions. Specific ~ implicit indicators were also used regularly. Most were word requests. 7.3.1 The Influence of Interaction Type In this section, the NNS trouble indicators are compared for the informal and institutional conversations. As can be seen in Table 7.5 the NNSs most frequently used global 8r implicit indicators in the informal conversations (33oIo). Informal conversations NNS trouble indicator types Institutional conversations ~ ~o ~ ~o Nonverbal 54 26.7 113 48.5 Global8z implicit 67 33.2 51 21.9 Specific 8z implicit 57 28.2 51 21.9 Global8z explicit 12 5.9 10 4.3 Specific 8c explicit 11 5.4 5 2.1 Other 1 0.5 3 1.3 Total 202 100.0 233 100.0 Table 7.5: Distribution of NNS trouble indicalor types per interaction rype 7.3 Nonnative Speaker Trouble Indicators 123 The NNSs also frequently used nonverbal indicators. In addition, they used explicit indicators about llolo of the time. Specific 8c implicit indicators were used more regularly. These were mainly word requests which show an active attitude on the part of the NNS to acquire words to convey intended meaning. In Example 14, Ergun's first trouble indicator used such a word request when he sought the Dutch word for wedding gown. The search for a word can be signalled by a pause, `uh', an unfinished utterance, nonverbal signals, explicit questions such as 'what's it called', a word in another language etc. (14) Infortnal convetsation with Ergun, Cycle 1 ts ti tc cc Ergun: coltc tc Janet: JaneC Ergiin: ja beetje witte uh t kleren weet ik niet ja jurk jurk - vepeats, imitates~ ja dat is anders t jaja dat is in het wit yes a bir white uh t clothes 1 don't know yes dress dress - ~repeats, imitates~ yes that's different t yeah yeah that is in white Such word requests did not occur in Woken 8z Swales's study (1989) when the NNSs had expert status. The NSs never offered verbal help such as offering vocabulary explanations and the NNSs never asked for it. NNS expert status may thus inhibit negotiation of ineaning because it would threaten the NNS's status. In this study, in the informal conversations there was no asymmetry in topic knowledge which is one of the aspects of expert status (see also Chapter 6), and there appeared to be no inhibitions on the NNS's side to requesting assistance in a word search. In the institutional conversations, in which the NNS had a lower status, word requests rarely occurred. Apparently, any asymmetry in expert status, regardless of whether the NNS has a higher or lower status, inhibits requests for words. In the institutional conversations, the NNSs in almost SOqo of the cases resorted to using nonverbal indicators. Furthermore, they used global 8t implicit and specific 8t implicit indicators equally often, but less than in the informal conversations. The specific 8z implicit indicators were mainly keyword repetitions. Only óqo of the time did the NNSs use explicit indicators in these conversations. Therefore, like the NSs, the NNSs appeared to be more face protecting in the institutional conversations due to the asymmetry caused by status differences. As can be seen in Table 7.6, there is little variety in trouble indicator forms in the informal conversations. The forms used most frequently by the NNSs were minimal feedback questions (Mhm?) which are linguistically the simplest means to indicate trouble. Nonverbal cues and declaratives were used regularly. Wh-questions and elliptic questions were used in a few cases. Yeslno, or-choice, and declarative questions were not used at all. In the institutional conversations, the NNSs used nonverbal cues in SOqo of the cases. Minimal feedback questions were used 25qo of the time. Declarative and elliptic questions were used in IOolo of the cases. Wh-questions were used somewhat less than in the informal conversations.In addition, other question types such as yeslno and or-choice questions were used a couple of times. 124 Chapter 7 Indicating Trouble in Understanding Informal conversations NNS trouble indicator forms ~ `~o Institutional conversations ~ qo Nonverbal 55 27.2 113 48.5 Question minimal feedback 78 38.6 54 23.2 Question elleptic 10 5.0 23 9.9 Question yeslno 0 0.0 3 1.3 Question or 0 0.0 1 0.4 Wh-question 17 8.4 14 6.0 0 0.0 1 0.4 39 19.3 22 9.4 3 1.5 2 0.9 202 100.0 233 100.0 Declarative question Declarative Other Forms Total Table 7.6: Distribution of NNS trouble indicators forms per interaction type In conclusion, these results indicate that there were great differences between the two interaction types in NNS trouble indicator types. In the informal conversations, implicit indicators were used most regularly. In the institutional conversations, also many implicit indicators but even more nonverbal cues were used. This may show face protection is a major issue for the NNSs in this context. Explicit indicators were used more in the informal conversations. The greater use of specific 8z explicit and specific 8z implicit indicators in these conversations, may indicate that the NNSs were more able to identify the trouble source or were less hesitant to do so because face was less an issue. No great NNS form differences were found between the two interaction types, although there was somewhat less variety in forms in the informal conversations. Declaratives were used more regularly in this condition, because of the many word requests used. The lack of variety may indicate that the NNSs generally did not have sufficient linguistic competence to adjust trouble indicator forms to the demands of the interaction. 7.3.2 Individual Portraits of the Nonnative Speakers In this section, the results are presented and discussed for the individual informants to show the development of NNS trouble indicators over time. In addition, influences of such personal factors as conversational style and gender can be determined. 7.3 Nonnative Speaker Trouble Indicators 125 Fatitna When a problem in understanding occurred Fatima almost exclusively reacted nonverbally (83qo) (see Table 7.7). Fatima indicated trouble nonverbally in all but one case in the institutional conversations. This one case was also an implicit (global) indicator. Only in the informal conversations did she use regularly specific 8z implicit indicators as well. Brceder (1993) referred to other data from the ESF project, the self-confrontations, to explain her trouble indicating behavior. In these self-confrontations, the earlier taped video-recordings were played back for the NNSs and they were asked in their source languages about problems in understanding which occurred and their strategies in dealing with them. Fatima used a `wait-and-see strategy' in the first cycle (see Brceder, 1993: 84). She often did not react at all or reacted with a gratuitous concurrence (jalyes) at transition relevant places. In the self-confrontations, she stated that she had understood almost nothing in the first institutional conversation. There may also be a gender influence, mentioned before in chapter 6, which may have made her hesitant to actively indicate trouble to a male authority. Cycle 1 2 3 Total qo Nonverbal 16 35 39 90 82.6 Global 8t implicit 0 1 0 I 0.9 Specific 8c implicit 1 10 3 14 12.8 Global8t explicit 0 0 0 0 0.0 Specific 8c explicit 1 1 ] 3 2.8 Other 0 1 0 1 0.9 Total 18 48 43 109 100.0 Table 7.7: Fatima's trouble indicator rypes per cycle Gender may also have played a role in her frequent use of specific 8c implicit indicators in cycle 2. She requested words by using an unfinished declarative (see also Table 7.8). Fatima used this strategy solely with Janet in the informal conversations. She did not use global 8c explicit indicators, but a couple of times she used specific 8z explicit indicators realized by minimal linguistic means. In Example 15, she used an ungrammatical short Wh-question like Bart also did. (15) Informal conversation with Fatima, Cycle 3 ts Janet: zijn jullie al naar ti 126 Fatima: de woningdienst geweest? (..J wat dienst? have you been to the housing service? (...) what service? - ~what do you mean by `service'?~ Chapter 7 Indicating Trouble in Understanding 1 2 3 Total Nonverbal 16 36 39 91 83.5 Minimal feedback question 0 1 2 3 2.8 Wh-question 1 l 1 3 2.8 Declarative 0 8 1 9 8.3 Other 1 2 0 3 2.8 To[al 1 ~S .1~ -t ~ 109 Cycle ~ qo 100.0 Table 7.8: Fatima's trouble indicator fonns per cycle Furthermore as can be seen in Table 7.8, there did not seem to be any linguistic change over time for Fatima. She even increasingly used more nonverbal indicators. The declaratives in cycle 2 are word requests she used with Janet. T'his is a remarkable outcome because although Fatima was a slow learner she did acquire some Dutch during the data collection period (Broeder, 1991). However, that apparently did not influence her choice of trouble indicator forms much. Mohamed As can be seen from Table 7.9, Mohamed appeared to have two basic strategies for dealing with problems in understanding. In particular in the first cycles, he reacted nonverbally. In the second and third cycles, he most frequently used global 8c implicit indicators. Therefore, the majority of his trouble indicators were minimal feedback questions which require little linguistic competence (see Table 7.10). Cycle 1 2 3 Total Nonverbal 22 10 14 46 33.6 Global 8c implicit 14 25 26 65 47.4 Specific 8c implicit 1 9 7 17 12.4 Global8c explicit 0 1 4 5 3.6 Specific 8z explicit 1 1 2 4 2.9 Other 0 0 0 0 0.0 Total 38 46 53 137 qo ]00.0 Table 7.9: Mohamed's trouble indicator types per cycle 7.3 Nonnative Speaker Trouble Indicators 127 Mohamed also used specific 8c implicít indicators in cycles 2 and 3(12.4qo). In cycle 3 there appeared to be some growth towards more diversity in types and forms. He started to use more global 8z explicit indicators, which were most frequently declaratives, and specific 8c explicit indicators, which were mainly Whquestions (see Table 7.10). qo 1 2 3 Total Nonverbal 22 10 14 46 33.6 Minimal feedback question 14 28 28 70 51.1 Elliptic question 1 0 0 1 0.7 Yeslno question 0 0 1 1 0.7 Wh-question 1 3 6 10 7.3 Declarative 0 5 4 9 5.8 Other 0 0 0 0 0.0 Total 38 46 53 137 Cycle 100.0 Table 7.10: Mohamed's trouble indicator forms per cycle It can be concluded that Mohamed generally stuck to minimal verbal indicators, which may have been effective enough for his goals. He was the least co-operative NNS informant, especially towards the end of the project. Broeder (1993: 73) showed that Mohamed's indicators in the later part of the data collection are more response preparing strategies than real indicators of trouble. After a question he immediately reacted with an specific 8r implicit indicator. Often with little to no clarification from the NS, Mohamed would react with an appropriate response in the next turn, as can be seen in Example 16. (ltí) Institutional conversation with Mohamed, Cycle 3 where is that? ts Bart: waar is dat? mhm? ti Mohamed: hm? Bart: waar is dat? where is that? tc here in tilburg mr Mohamed: hier in tilburg Mahmut As can seen in Table 7.11, Mahmut did not limit himself to one or two trouble indica[ing strategies, but used quite a few. He used specific 8z implicit indicators in almost two-thirds of the cases. Quite frequently he used the keyword strategy, repeating a problematic word from the NS's trouble source to elicit a repetition or explanation. He used several utterance forms for this strategy, such as declaratives and elliptic questions (see also Table 7.12). In Example 17, Mahmut suggested that he might get extra money from the social welfare institution if he were to get a more expensive apartment. Paul denied that, referring to subsidy regulations. Mahmut did not understand Paul's utterance 128 Chapter 7 Indicating Trouble in Understanding Cycle Total 3 qo 1 2 Nonverbal 4 4 3 11 14.5 Global 8L implicit 2 5 5 12 15.8 Specific 8c implicit 18 19 11 48 63.2 Global 8t explicit 0 0 0 0 0.0 Specific 8t explicit 0 1 2 3 3.9 Other 0 2 0 2 2.6 Total 24 31 21 76 100.0 Table 7.11: Mahmut's trouble indicator types per cycle and repeated the word subsidie without rising intonation to receive clarification. Such keyword repetition is a very ambiguous strategy because the interlocutor often does not know whether the repeated word has been understood or not. (17) Institutional conversation with Mahmut, Cycle 1 ts Paul: ti tc ri Ic Mahmut: Paul: Mahmut: Paul: cc co tc Mahmut: Paul: dcen ze niet a--F they won't do it ~ niet helpen? nee die kunnen niet helpen t mag niet [t] not help? no they cannot help t not allowed [t] want die liggen boven de subsidiegrens subsidie ja dat kan niet kan niet kan geen subsidie op gegeven worden because these are above the subsidy limit subsidy yes that cannot be done cannot cannot be given a subsidy Apparently in the above example, Paul thought Mahmut had understood subsidie but not the rest and thus he reformulated the rest. After Mahmut's second keyword repetition, he rephrased the total trouble source. Mahmut checked this by rephrasing part of the clarification and the trouble appeared to be resolved. Mahmut also used nonverbal and global 8c implicit indicators each about 15qo of the time. He started to use specific 8t explicit indicators in cycles 2 and 3, but very infrequently. Only global 8z explicit indicators were not used at all, so he appeared to be co-operative and face protecting. In general, there is little development over time. As can be seen in Table 7.12, in Mahmut's forms there is minimal development fewer elliptic questions, more minimal feedback questions (wat?) and more Whquestions. It might be that Mahmut early on discovered an effective range of trouble indicators and stuck to this pattern, even though they may not all have been nativelike. However, his minimal feedback questions do develop in their form towards more standardlike usage. In the first cycle Mahmut used a Tilburg dialect form wasda? (wat is dat - what is that?). In the second and third cycles he shifted to the standard wat? (what?). 7..3 Nonnative Speaker Trouble Indicators 129 Cycle qo 1 2 3 Total Nonverbal 4 4 3 11 14.5 Minimal feedback question 2 7 8 17 22.4 Elliptic question 8 6 2 16 21.1 Declarative question 1 0 0 1 1.3 Wh-question 0 1 3 4 5.3 Declarative 9 11 5 25 32.9 Other 0 2 0 2 2.6 Total 24 31 21 76 100.0 Table 7.12: Mahmut's trouble indicator forms per cycle Ergun Ergun used a variety of trouble indicators. Whereas Mahmut used specific indicators, Ergun used global 8z implicit indicators frequently, in particular in cycles 1 and 2 (Table 7.13). Cycle ~lo 1 2 3 Total Nonverbal 14 4 2 20 17.7 Global8t implicit 18 14 8 40 35.4 Specific 8r implicit 5 13 11 29 25.7 Global 8z explicit 4 6 7 17 15.0 Specific á explicit 2 3 1 6 5.3 Other 0 1 0 1 0.9 Total 43 41 29 113 100.0 Table 7.13: Ergun's trouble indicator types per cycle In the first and second cycles of the institutional conversations Ergiin used the Tilburg dialect form welk? (which) as an equivalent to hè? (huh?) and wat? (what?) as can be seen in Example 18. (18) Institutional conversation with Ergun, Cycle I ts Paul: ik zal eens kijken I'll have a look of je ingeschreven staat whether you are registered ben je al eerder hier geweest? have you been here before? ti Ergun: ~ welk? f which? tc Paul: ben je al eerder hier geweest? have you been here before? tr Ergiin: ja yes 130 Clwpter 7 Indicating Trouble in Understanding In the third cycle, he stopped using this form and used hoezo? (why?), which is a more powerful and more appropriate indicator, even though he was not aware of the difference in meaning with hè? (huh?) and wat? (what?) (Broeder, 1993: 81). Both Ergiin and Mahmut apparently acquired dialect forms first but later shifted to more standard indicators. Ergun also used specific 8c implicit indicators such as the keyword strategy quite frequently (25.7qo). These are often elliptic questions. The fact that Mahmut and Ergun both used keyword repetition regularly might be related to their source language background. In Turkish, repetitions of (parts of) the interlocutors utterance play an important role (Brceder, 1991: 187). They apparently used repetitions to indicate trouble as well. olo 1 2 3 Total Nonverbal 14 4 2 20 17.7 Minimal feedback question 19 14 9 42 37.2 Elliptic question 4 11 1 16 14.2 Yeslno question 0 0 2 2 1.8 Or-choice question 0 0 1 1 0.9 Wh-question 1 6 7 14 12.4 Declarative 5 6 7 18 15.9 Other 0 0 0 0 0.0 Total 43 41 29 113 Cycle 100.0 Table 7.14: Ergiin's trouble indicator fonns per cycle Ergiin sometimes indicated trouble nonverbally (17.7qo). In addition, he was the only NNS informant who regularly indicated trouble explicitly ( 20.3qo) He used global 8t explicit indicators quite frequently, as can be seen in Example 19. (19) Informal conversation with Ergun, Cycle 3 ts Janet: en al die vragen die je mcet invullen? ti ErRiin: ik weet ik niet cc Janet: weetje niet? ti Er,~iin: nee ik weet ik niet and all those questions you have to fill out? I don't I know you don't know no I dnn't I know Ergiin's indicators are idiomatic in two ways. Firstly, a NS would not use the verb `know' but `understand', saying ik begrijp het niet (I don't understand). Secondly, he used the reduced inverted spoken form weet ik niet instead of dat weet ik niet (that I don't know) and put an extra subject in front of it: ik weet ik niet (I don't I know). He kept using this frozen expression in all three cycles, even though he also used correct variations from the first cycle on: dat weet ik niet (that I don't know), 7.3 Nonnative Speaker Trouble Indicators 131 weet ik niet precies (I don't know exactly). He also used a positive version weet ik wel (that I know). This may be a compensation for the face-threatening explicit indicators (Broeder, 1993: 79). Ergun showed some development over time, using fewer nonverbal and global 8z implicit indicators, and more specific 8c implicit indicators. The explicit indicators show no development over time. They are typical for Ergun because the other NNSs hazdly used them. He apparendy prefer,-ed to go on record regularly. As far as the forms are concerned Ergun most frequently used minimal feedback questions in all three cycles, but there appeazs to be some development towards more diversity (see Table 7.14). The number of Wh-questions increased in cycles 2 and 3, while the minimal feedback questions decreased after cycle 2. This may be related to his linguistic development. Ergun's elliptic questions showed an increase in cycle 2 but they almost disappeazed in cycle 3. In the third cycle, he also used a couple of yeslno and or-choice questions which may indicate some linguistic development in trouble indicating forms. In conclusion, there are differences between the NNSs in trouble indicating behavior. Fatima most frequently used nonverbal trouble indicators in the institutional conversations. This may show a strong need for face protection. Only in the informal conversations did she use some other types (word requests). Her linguistic forms did not develop. Mohamed, in addition to nonverbal indicators, used global and implicit indicators as well. These generally had the form of minimal feedback questions. He showed some development over time to other more linguistically demanding types and forms. Mahmut mainly used specific 8c implicit indicators (keyword repetition) but he also sometimes used nonverbal and global 8z implicit indicators. Explicit indicators he hazdly used at all, so he appeazed to be co-operative and face protecting. In general, there is little development over time in types and forms. He did drop a dialect form for a more standard trouble indicator. Ergun used global 8cimplicit indicators most frequently. He also used specific 8c implicit indicators, which were keyword repetitions. In addition, he was the only NNS informant who regulazly indicated trouble explicitly, showing the least concern for face. He showed some development over time towards more specific indicators and more diversity in fonm. He also exchanged a dialect form for a standazd trouble indicator. 7.4 Comparing NS and NNS Trouble Indicators In this section, NS and NNS indicators, which were discussed separately in sections 7.2 and 7.3, will be compared to each other. As can be seen from Table 7.15 and as had already become clear in Chapters 5 and 6, the NNSs used four times as many trouble indicators as the NSs. 132 Chapter 7 lndicating Trouble in Understanding NS Trouble indicator types NNS ~ `~r ~ ~o Nonverbal indicator 12 10.9 167 38.4 Global 8c implicit indicator 18 16.4 I 18 27.1 Specific 8c implicit indicator 51 46.4 108 24.8 Global 8c explicit indicator 7 6.4 22 5.1 Specific 8c explicit indicator 19 17.3 I6 3.7 Other functions 3 2.7 4 0.9 110 100.0 435 Total 100.0 Table 7. l5: A comparison of NS and NNS trouble indicator h~pes However, when the NSs did use trouble indicators, they preferred specific indicators over global ones and implicit over explicit indicators. The majority of cases were specific 8z implicit indicators, but global 8t implicit and specific 8c explicit indicators were used regularly as well. All other means were also used by the NSs, which may show that they adjusted their strategies depending on the NNS's communicative competence and the type of trouble source. T'he NNSs, on the other hand, most frequently used nonverbal means to indicate trouble. They had an even stronger preference than the NSs for implicit over explicit indicators. In contrast to the NS, they preferred global over specific indicators, particularly in the initial stages of language development. They used many global 8r. implicit indicators, which require little communicative competence. T'he NNSs also used specific 8c implicit indicators but less so than the NSs. More communicative competence is necessary for the production of these indicators, but it still can be done with minimal means as can be seen in Example 20. Mahmut just mentioned, at this particular point, that he had had a talk with a colleague about this colleague's plans to take some days off from work. Therefore, he was confused when Janet asked him about his days off. He apparently did not expect a question about himself. (20) Informal conversation with Mahmut, Cycle 3 ts tUSI cc tc Janet: Mahmul: Mahmut Janet ja wat ga je dcen? wie? hij? nee jij wat ga jij dcen yes what are you going to do? wlui? he? no you what are you going to do in die vrije dagen?in those days off? Problems with reduced deictic pronouns like in this example appear to arise quite frequently with all NNSs. They occur particularly often when the reduced form of the 2nd person subject pronoun in Dutch (je for jij) is used. The NNSs then used specific óz implicit indicators such as `who?' and confirmation checks like `me?'. 7.4 Comparing NS and NNS Trouble Indicators 133 Explicit indicators were used by the NNSs in 9qo of the cases, whereas the NSs used them in 25qo of the cases. The NNSs may thus protect their own positive face more to compensate for lack of linguistic competence. NS Trouble indicator forms NNS ~ 01o q qo Nonverbal 12 ]0.9 168 38.4 Minimal feedback question 25 22.7 132 30.3 Fill-in-the-blank question 7 6.4 0 0.0 Elliptic question 9 8.2 33 7.6 Yeslno question 0 0.0 3 0.7 Or-choice question 3 2.7 1 0.2 Wh-question 42 38.2 31 7.1 Declarative ques[ion 0 0.0 1 0.2 Declarative 7 6.4 61 14.1 Imperative 2 1.8 0 0.0 Other Forms 3 2.7 5 1.2 110 100.0 435 Total ]00.0 Table 7.16: A comparison of NS and NNS trouble indicator forms As can be seen in Table 7.16, both NSs and NNSs used minimal feedback questions regularly to indicate trouble, showing a general preference for efficiency and minimizing effort. The NNSs, in addition, frequently indicated trouble with nonverbal means, while the NSs used these indicators only l0010 of the time. However, the fact that the latter used nonverbal indicators too does indicate that it is in certain cases an appropriate way of indicating trouble. Wh-questions made up only 7010 of the NNS trouble indicators, while the NSs used this kind of question most frequently (38010 of the time). Here the difference in communicative competence can be seen. To formulate such a Wh-question requires knowledge of specific question words, which refer to different types of information, such as place, time, or manner. For the NNSs however, these question words caused many problems. 1'hey often did not understand which type of information was requested and could not use them themselves. Also the NS's role, particularly in the institutional conversations, where most Wh-questions were used, may influence the distribution. In both interaction types the NSs more or less controlled the topics and thus took more initiative. This may have led to the use of more Wh-questions. Fill-inthe-blank questions were only used by the NSs. These seem to be a typical NS 134 Chapter 7 Indicating Trouble in Understanding strategy of helping a speaker with limited competence to clarify a problem in understanding. The NNSs used a remarkable amount of declaratives, while the NSs used them rarely. These declaratives were used for statements of non-understanding and for word requests. Both appear to be typical NNS behaviors, because they request help from the NS in understanding input and producing output. 7.5 Conclusions In this chapter, the types and forms of NS and NNS trouble indicators were described and compared to each other. In addition, the influence of NNS communicative competence and interaction type on trouble indicator types and forms were investigated. The classification for trouble indicator forms made it possible to show differences between individuals and between the NSs and NNSs as a group on three dimensions: nonverbal-verbal, implicit-explicit, global-specific rather than on one (implicit-explicit) as Bremer et aL (1993) did. As faz as the types of trouble indicators aze concerned, it can be concluded that the NSs generally tended to be specific and generally implicit (off record) in indicating trouble. Within a sequence, first implicit indicators are used and when this dces not have the desired effect explicit indicators are used later in the sequence. However, they also used global, explicit and nonverbal trouble indicators. This may show their ability to adapt the trouble indicators to the NNS communicative competence and to the trouble source. In contrast, the NNSs most frequently indicated trouble in nonverbal and global ways which may be caused by their lower linguistic competence. However, the NNSs also increasingly used specific 8c implicit indicators which may reflect their language acquisition. Explicit indicators are not used much. Generally they are used, as with the NSs, later in the sequence if implicit indicators did not lead to the desired clarification. Both NSs and NNSs used nonverbal indicators and minimal feedback questions, which may be the least effortful and least interruptive way to indicate trouble. T'he separate classification of trouble indicator forms made it possible to describe differences between NSs and NNSs as a group, and between the NNSs over time better than was possible with Bremer et al.'s classiiication (1993). The NSs showed more variety of linguistic form than the NNSs but there was less difference than expected. The NSs used more Wh-questions for their specific 8z implicit indicators, which require a considerable amount of linguistic competence. Only the NSs used fill-in-the-blank questions, which provided the NNSs with a model which had to be completed. This seem to be a typical helping strategy. The NNSs on the other hand, most frequently used minimal feedback questions and only infrequently other question types. They did use quite a few declaratives. These were used to explicitly indicate trouble and to request words. 7.5 Conclusions 135 As far as the individual NNSs are concerned, there is a growth in diversity for the three male informants but not for Fatima. She most frequently kept using nonverbal cues for indicating trouble. There may be a gender influence as hypothesized in Chapter 6. Fatima might have felt uncomfortable actively indicating trouble. However, she also had the lowest linguistic competence of the four NNSs. There are some systematic differences between the slow learners Fatima and Mahmut on the one hand and the fast learners Mohamed and Ergun on the other. Although Mahmut from the first cycle on employed a broader range of trouble indicator types and forms than Fatima, both showed little development over time in trouble indicator types and forms. Mohamed and Ergun showed some development, particularly in the forms. Thus communicative competence is an explaining variable for the differences in NS and NNS trouble indicator types and forms. There are also differences in personal style. Ergun is the only NNS informant who regularly went on record and used global 8r explicít indicators. Mohamed, in contrast, had a preference for global 8z implicit indicators, which may show his somewhat more passive attitude. Word requests were a type of specific trouble indicator which appeared to be a typical NNS strategy for requesting assistance from the NS for obtaining certain vocabulary. The Turkish informants differed somewhat from the Moroccan informants. They appeared to favor keyword repetition. They also both first acquired dialect trouble indicator forms and later replaced them with standard forms. There was also some influence of interaction type. In the institutional conversations, the NSs used more global means to indicate trouble than in the informal conversations. This difference is caused by one NS in particular, Bart, the housing official. Whether this originates from his professional role or personal style is not clear. In general, the three NSs analyzed individually showed personal preferences in trouble indicating strategies. In general, the NSs used a greater variety of forms in the informal conversations. In particular, more 611-in-the-blank questions were used. In the institutional conversations, the NSs used Wh-questions more frequently than in the informal conversations (SOqo). Furthermore, in about 30qo of the cases minimal feedback questions were used in these conversations. The NNSs most frequently used nonverbal trouble indicators in the institutional conversations. Fatima, in particular, used this type only. This may show a strong need for face protection. In the informal conversations, the NNSs generally used global 8z implicit and specific 8t. implicit means. They used word requests only in this context. This shows their motivation to make their output more comprehensible. In addition, it may indicate that they could use word requests because they had less need to protect their and the NS's face as in an asymmetrical context (see also Woken 8z Swales, 1989). The fact that they also used specific á explicit indicators a little more agrees with this hypothesis. No major differences in trouble indicator forms were found for the two interaction types. This may indicate that the NNSs were not able to adjust their forms to the demands of the interaction. I36 Chapter 7 Indicating Trouble in Understanding Chapter 8 Clarifying Trouble in Understanding 8.0 Introduction After a discussion of the different types and forms of trouble indicators in the previous chapter, we now focus on trouble clarifications. Trouble clazifications are moves with which generally the speaker who created trouble tries to repair it. A trouble clarification manipulates the content and form of a trouble source. Trouble clarifications regularly form an adjacency pair with the trouble indicators discussed in Chapter 7. The following reseazch questions will be answered for the part which concerns the trouble clarifications: 4a. How can NS and NNS trouble clarifications be described in terms of their types and forms? 4b. What is the influence of the NNS communicative competence and interaction type on the variety and quantity of trouble clarifying behavior of the NS and NNS? In section 8.1 follows a classification of trouble clarifying types (8.1.1). For the forms, only a few additional remarks are made in section 8.1.2 because the basic classification has been described in Chapter 7. In section 8.1.3 some expectations aze formulated on the outcomes concerning the use of NS and NNS trouble clarification types and forms in the different conditions. In section 8.2, the results for the NS group will be presented. The influence of interaction type will be discussed in section 8.2.1. Portraits of the individuals are given in section 8.2.2. In section 8.3, the results for the NNSs as a group will be presented and discussed. The influence of interaction type on the NNS trouble clarifications is described (section 8.3.1). In section 8.3.2 follows a discussion per NNS. Longitudinal analyses are discussed only when there was any development. In section 8.4, a comparison will be made between the NSs and NNSs as a group. The chapter ends with conclusions in section 8.5. 8.0 Introduction 137 8.1 A Classification of T~ouble Clarification T~pes and Forms In this section, a taxonomy of trouble clarification types is presented which can be applied to NS and NNS trouble clarifications and confirmation checks (8. I.1). This taxonomy emerged through the coding process and was partly based on the literature on NS adjustments. For the trouble clarification forms, the classification presented in Chapter 7 was used. Therefore only some additional remarks are made in section 8.1.2. Some expectations about the NS and NNS trouble clarifications are formulated in section 8.1.3. 8.1.1 Trouble Clarification Types One of goals of this study was to develop classifications for negotiation moves which could be applied to both NS and NNS behavior. For the development of trouble clarification types, literature on NS adjustments (foreigner talk and comprehensible input) as well as on NNS adjustments (comprehensible output) was studied. The literature pertaining to trouble clarification types focusses mainly on NS adjustments which simplify the input and make it comprehensible. LarsenFreeman 8z Long (1991: 125-126) and Hatch (1983: 183) give an overview of such adjustments, which may occur either before or after trouble. There is a great deal of líterature on NS adjustments to the NNS (see Chapter 2) but less on NNS adjustments, that is, comprehensible output (Pica et al., 1989). The early work on NS adjustments focussed on ungrammatical input. Ferguson (1975: 4) described three processes of adjustments in foreigner talk which lead to ungrammatical input: 1. omission (deletion of articles, copulas etc.); 2. expansion (addition of lexical material with or without repetition of the original material, such as addition of `you' to imperative utterance); 3. replacementlrearrangement (substitution of e.g. personal pronouns by object pronouns such as in `me no like'). Long (1983b) typifies such ungrammatical adjustments as linguistic adjustments. Linguistic adjustments are replacements of lexical items or phrases by equivalents (synonyms). These adjustments may occur in the areas phonology, morphology, lexicon, and syntax. They do not always lead to ungrammatical input. In fact, much more often linguistic adjustments lead to grammatical input (see also Chapter 2). Long (1983b) claimed that another kind of adjustment is much more powerful for the creation of comprehensible input. This kind of adjustment he called conversational adjustments. Conversational adjustments concern the content or the structure of the interaction. Content adjustments may be a more here-and-now orientation of topics. Structural adjustments may be more repetition, elaborations, and more question-answer strings. These lead to an increase of the number of turns, that is to side-sequencing. In fact, any negotiation of ineaning move leads to such conversational adjustments, whether or not they also manipulate linguistic ma- 138 Chapter 8 Clarifying Trouble in Understanding terial. As stated in Chapter 2, Ellis (1985) apparently had the same opinion when he, based on the work of Ferguson, Long and others, distinguishes three accumulative types of foreigner discourse: 1. with conversational adjustments 2. with conversational and grammatical linguistic adjustments with conversational and grammatical and ungrammatical linguistic adjust3. ments. The processes reported by Ferguson (1975) for ungrammatical NS input were also found in NNS morphological adjustments by Pica et al. (1989: 189). When they coded NNS adjustments they divided them into addi-tion, substitution, and deletion of inflectional morphemes or functors. The fact that these three processes appear to be equal to the foreigner talk processes supports Ellis's view (1985). He claimed that the "formal characteristics of [NNS] interlanguage, pidgins and foreigner talk aze very similar (..) and may therefore be the result of a single underlying process" (Ellis, 1985: 138). In other words, the same kinds of inechanisms may be operating on NS and NNS codes. The processes may be operating a priori or after trouble has been indicated. In this study, the three main processes of adjustment, therefore, were applied to trouble clarifications of both the NS and NNS which occur after trouble is indicated. Thus the adjustment processes are related to an eazlier part of the interaction, that is generally the trouble source. A priori adjustments are not analyzed. The three processes of NS and NNS adjustments are defined as: 1. deletion (when parts of the original trouble source aze left out in the trouble clarification) 2. addition (when linguistic material is added to the trouble source) 3. substitution (when the trouble source's linguistic material is reworded). Along with the distinction in three processes of adjustments, Long's distinction (1983b) between conversational and linguistic adjustments was considered for classification. However, it turned out to be difficult to apply this distinction in the way Long defined it. First, content adjustments, such as here-and-now orientation which he called conversational are very different from structural adjustments. They did not occur much in this study because the topics demanded use of past and future tenses. Such content adjustments serve to keep the conversation simple, which is not always possible. Structural adjustments such as repetition, on the other hand, which occurred very frequently, probably serve to create redundancy and a second opportunity for processing. Another problem with the distinction was that a category such as elaboration (expansion) is called a conversational adjustment by Long (1983b) because it leads to structural adjustments (more turns). However, it also adds new linguistic material to the interaction, but because the material expands on the trouble source. Therefore, it also appears to create linguistic adjustments as well. Apparently, Long's category of conversational adjustments includes linguistic adjustments in the form of additions. It seems he only called an adjustment linguistic when substitution took place. Thus different adjustment types, of content and structure, of repetition and addition, are taken together in the category conversational adjustments. The distinction between linguistic and 8.1 A Classification of Trouble Clarification Types and Forms 139 conversational is not an easy one to draw. In negotiation of ineaning sequences, almost every clarification move leads to conversational adjustments. In addition, they may include some linguistic adjustments. Therefore, it is probably better to speak of degree of linguistic adjustment in trouble clarifications. Some categories are more conversational, others more linguistic. This distinc-tion is important in terms of comprehension and production. A trouble clarification which includes both repetition and replacement of one element of the trouble source requires less linguistic processing or production than when a trouble source is fully paraphrased. Another relevant distinction which emerged from the coding is the level of operation. Some trouble clarifications operate locally. They repeat or substitute one word or phrase out of the original trouble source which is perceived as the main problem or add one word or phrase to the original trouble source. Thus they stimulate the interlocutor to focus on this element. Other clarifications operate globally. They repeat or rephrase the complete trouble source or expand on it. Eventually a classification of 11 categories was established through the open coding of the data (see Figure 8.1). The use of all of these categories leads to structural adjustments, that is side-sequences. Thus all are conversational adjustments. However, only a few categories are pure conversational adjustments (full and partial repetition). Many categories appeared to combine conversational with linguistic adjustments. A few are classified as strong linguistic adjustments (e.g. paraphrase, word supply). In addition, a twelfth category (Other) was added for checks which could not be coded as any of the 11 categories because they were partly unintelligible or because they were expressed in another language. The classification described above is applicable to trouble clarifications as well as confitYrtation checks. Although confirmation checks indirectly signal a problem in understanding they also manipulate content and form. Therefore, they also serve to clarify the problem. It will, however, become clear from the next chapter that quantitative differences appear in the distribution of [ypes between the two moves. 1. Full repetition A trouble clarification is coded as full repetition when an utterance is fully repeated with exactly the same words. Then none of the three adjustment processes is occurring (see Example 1). However, other changes in intonation, prosody and speed may occur. (1) Institutional conversation with Ergun, Cycle 1 tc Paul: Ergun: Paut.~ wat voor huis wil je? hè? wat voor huis wit je? what kind of house do you want? huh? what kind of iu~use do you want? 2. Partial repetition The next type is used for utterances of which only part is repeated without any additions or changes. This is called partial repetition, which is a conversational adjustment (see Example 2). 140 Chapter 8 Clarifj~ing Trouble in Understanding Degree of linguistic Trouble clarification adjustment type Level of operation NO LINGUISTIC l. Full repetition global ADJUSTMENTS 2. Partial repetition local 3. Additive repetition 4. Elaboration local global 5. Substitutive repetition 6. Complex repe[ition local local 7. Paraphrase global local WEAK L[NGUISTIC ADJUSTMENTS STRONG LINGUISTIC ADJUSTMENTS 8. Hypothesis forming 9. Correction 10. Word supply local local locaUglobal 11. Translation Figure 8.1: Trouble clarification types (2) Institutional conversation with Mohamed, Cycle 3 ts ti tc Bart: Mohamed: Bart: ik maak maar een grapje hm? ik maak een grupje I' m only making a joke mhm? I'm mnking a joke 3. Additive repetition An utterance may also be repeated with exua material added. This is called additive repetition (see Example 3). (3) Institutional conversation with Mohamed, Cycle 1 ts Paul: waar wil je` [wonen]? ti tc Mohamed: Paul: `[als uh] hm? waar wil je ergens wonen? where do you want` [to live?] `[if uh] mhm? where ubout do you want to live? 4. Elaboration The trouble clarification may contain only new elements continuing on the same topic or explicating underlying presuppositions. T'his move has been called expan- 8.1 A C[assiftcation of Trouble Clariftcation Types and Forms 141 sion or elaboration. Vazonis á Gass (1985b: 47) define this type of clarification as information which was not included in the original response but which is added after a trouble indicator to expand upon and clarify a point made in the original response. They do not make clear whether elaborations which are combined with repetitions are included. In the current study this is not the case (see Example 4). (4) Institutional conversation with Fatima, Cycle I tc PauL kun je daar niet zolang blijven? can't you stay there for the moment? gc Fatima: ja yes tc Pauk totdat er een huis is unti! there is a house Elaboration also includes what has been called `decomposition' by Larsen-Freeman 8c Long (1991) and `faktorisieren' by Hinnenkamp (1989). These terms are used when an utterance or idea unit is broken down into smaller, more manageable parts because it is perceived to be difficult to process as a whole (see Example 5). Often decomposition clarifies underlying presuppositions. (5) Informal conversation with Ergun, Cycle 2 tc Janet: krijg je ook reiskosten? ti Ergun: ~reiskostetu? - ~imitates~ co ]anet ja ti Ergun: wat reiskos[en? tc Janet: ja die vorige keer in hetl in de studio gc ErgOn: ja tc Janet: alsjij vraagt naar werk in de broodfabriek [dan] gc Ergun: [ja] tc Junet: vraagjij krijg ik ook geld voor uh het reizen tussenl met de trein of inet de bus? do you get Iravel expenses? aravelexpenses~? - ~imitates~ yes wat Iravel expenses? yes the last [ime in thel in the studio yes iJyou ask about work in the bread factory [then] [yes] do you ask do 1 get money for uh the travelling betweenl by train or by bus? 5. Substitutive repetition When substitution or a word of phrase occurs in a repeated trouble source, this is called substitutive repetition. For instance, a lexical item from a trouble source as the formal term used by the housing official 'woning' (residence) is replaced by a more common synonym such as `huis' (house). However, substitutive repetition could also be a NS repetition of a NNS ungrammatical utterance in which the NS provides a grammatical model. This is a form of modelling which may function as an off-record implicit correction (Day et al., 1984). 6. Complex repetition The three adjustment processes deletion, addition and subsiitution do not always appeaz sepazately but also in combination. When for example a combination of 142 Chapter 8 Clarifying Trouble in Understanding deletion and addition occurred, it was called complex repetition. Trouble clarifications with one of the three adjustments and a rearrangement of word order were also coded as complex repetition. 7. Paraphrase A revision of the linguistic material may also happen on utterance level substituting for the trouble source rather than repeating it. Then it is called paraphrase. Often these pazaphrases concern idiomatic expressions (see Example 6). (6) Institutional conversation with Fatima, Cycle 3 Bart: ts is hij afgekeurd? ti tc Fatima: Bart t mag hij niet meer werken? is he rejected? (as medically unfit) t is he not aUowed to work anymore? The term semantic repetition is also used by Long (1983a) and Pica et al. (1989) includes paraphrase as well as substitutive repetition. Here a distinction is made, because in the first case there is no repetition of linguistic elements, whereas in the second there is. In addition, substitutive repetition operates on a local level and pazaphrase on global level. Pazaphrase replaces the full trouble source. Pazaphrases are strong linguistic adjustments. 8. Hypothesis forming Furthermore, the trouble clarification may merely formulate a hypothesis on the other's intended meaning (hypothesis forming). It is an adjustment or guess of perceived but not explicitly or clearly formulated content of the interlocutor rather than of the speaker's own utterance. Thus it could be seen as a form of otherpazaphrase. This last category is typical for confirmation checks but dces incidentally occur as trouble clarification (see Example 7). (7) Institutional conversation with Fatima, Cycle 3 ts Bart: was jij eerder hier were you here before dan man of niet? husband or not? ti Fatima: t t Bart: wie was ;eerder in nederland? who was ;first in the netherlands? tc tc was jij ;eerder hier we.re you ;first here in nederland in the netherlands tc of man eerder hier? or husband here first? ti Fatima: t f tc Bart: eerst mun hier in nederland? first husband here in the netherlands? mr Fatima: ja mijn man yes my husband A more extensive discussion of the type hypothesis forming can be found in Chapter 9 (section 9.1.1). 8.1 A Classification of Trouble Clarification Types and Forms 143 9. Word supply Some trouble clarifications aze bound as a second part of an adjacency pair to a specific trouble indicator. A word request is often followed by ihe requested item. The trouble clarifications which supply these requested items have been called word supply (see Example 8). These usually occur as single elements without repetition. (8) Informal cooversation with Fatima, Cycle 2 tslti Fatima: co ti tc t~tc tc cc )anet: Fatima: Janet: Fatima: Janet: Fatima: tc Janet: met uh ~dcea~ - ~m-medicine~ ~ash smitce~ with uh ~dcea~ - ~m-medicine~ ~ash smitce~ ja waz zegt die t? raólet ~dcea~ [alles] [medicijnenJ alles medicijnen mediclmedic(x) o Irnedici - daughs~ medicijnen yes what dces that say t? tablet - ~m-how do you call it?~ tablet of - an-how do you call it?~ tablet or ~dcea~ [everything] [medicine] everyrhing [medicineJ mediclmedic(x) ~~~medici - daughs~ medicine 10. Correction In a few cases the speaker rephrases part of the interlocutor's utterance specifically, indicating that it is a correction by the use of declaratory intonation. This is called on-record cotrection by Day et al. (1984: 23). Correction can be signalled by adding a rejection (no). Here this linguistic adjustment will be simply called correction. They differ from substitutive repetition, which can also model a correct form, in the explicitness with which the correction is signalled. Corrections may concern the content or the form of an utterance. Thus they have alongside the goal of clarifying understanding an additional goal of `teaching' the other about the world or about language use. In Example 9, Fatima tried to explain a Moroccan headache remedy of putting pieces of lemon on the head and covering them with a shawl. Janet hypothesized wrongly that the lemon was put into water. This hypothesis Fatima corrected. (9) Informal conversation with Fatima, Cycle 2 Fatima: alleen jij hoofdpijn [s cc cm tc tc tc l~ met uh t ~citron~ dcen Janet: Fatima: )anet Fa[ima: - ~-French pronunciation~ ja [cilcitrcen] [jij weet]? en (water]? [citrcen] met nee citroen met sjaal ~hien - ~points~ only you headache with uh t lemon do - r-French pronunciation~ yes lellemon [you know?] and [water?] [lemon] with no lemon with shawl ~here~ - ~points~ Chapter 8 Clarifying Trouble in Understanding 11. Translation In some cases the speaker tried to clarify the trouble by giving a translation into another language, such as the source languages of the NNS informants, Turkish and Moroccan Arabic, or French ( see Example 17). Translations are linguistic adjustments because they substitute an utterance, phrase or word from one language with an equivalent in another language. They thus can operate locally or globally. 12. Other The `other' category contains reactions and responses which could not be classified as any of the above. As can be seen from the tables, the category was included in the quantitative results but this category was only infrequently used. Most trouble clarifications could be classified with the above mentioned categories. This category will not be further discussed. 8.1.2 Trouble Clarification Forms In Chapter 7 an overview of the linguistic forms was given based on the division of declarative, interrogative and imperative utterances which is also applicable for the trouble clarifications ( see section 7.1.2). There are a few interrogative types which were not used as trouble indicators but do appear as trouble clarifications. These are negative questions. Some other questions such as indirect, rhetorical, fill-in-the blank, and minimal questions appeared so rarely that they were taken together in one category ( other questions). There is no nonverbal category. Ungrammatical utterances were not systematically coded as such. Almost all NNS utterances were ungrammatical and only a few of the NSs'. Some examples of the ungrammatical NS trouble clarifications will be discussed in section 8.2.3. In this study, the adjustments coded for the trouble clarifications and confirmation checks for practical reasons do not include phonology because the transcripts were the basis for analyses and they were more or less based on standard orthography. 8.1.3 Expectations Several expectations can be formulated about the differences in NS and NNS trouble clarification types and forms. The degree and type of NS and NNS adjustments may depend on several factors such as the NNS's proficiency, interaction type, and roles (asymmetry). According to Bingham Wesche (1994: 233), NS adjustments provide NNSs with well-formed, simpler, more regular, more explicit and more redundant language. Long (1983b) claims that NSs favor conversational adjustments over linguistic adjustments. In addition, Ellis (1985), by hypothesizing an accumulation of (ungrammatical) conversational (and linguistic) adjustments, also presumes conversational adjustments to be most frequent. This is certainly true for negotia- 8.1 A Classification ojTrouble Clarification Types and Forms 145 tion moves, which by definition lead to conversational adjustments. The question is what is the function of linguistic and conversational adjustments. Several principles appear to be at work to achieve these adjustments. T'hrough negotiation of meaning, well-formed and more regular forms can be achieved by adding to original elliptic utterances. Simplification may also be reached through substitution of complex forms by easier equivalents. Bingham Wesche (1994: 228) hypothesizes that paraphrase may provide NNSs with richer lexical and structural data. More explicitness can be reached through elaboration of underlying propositions, detail, examples, etc. These are linguistic adjustments. Simpler utterances may be achieved through deletion, which may focus the interlocutor on the important part of the message. Redundancy can be achieved through repetition. T'hese are conversational adjustments. Long (1983c) claims that from an educational perspective NS linguistic adjustments might be counterproductive. What happens is that the NS takes out of their language those elements that cause problems (and even replaces them with ungrammatical phrases) and thus takes away the opportunity of the NNS to learn that particulaz form. This may be the case for certain types of substitution, but it is not necessazily the case for all types. It is expected that trouble clarifications are accumulative in clarification sequences. They are always conversational adjustments but they may include some form of linguistic adjustment as well. The NS is expected to use a full range of conversational and linguistic adjustment types and forms applying all three adjustment processes, deletion, addition and substitution. The NNS differs from the NS in that he has fewer linguistic resources for clarification. He might use more types and forms of clarification which require limited linguistic competence. He may prefer clarifications which require little to no linguistic adjustment. In other words, the NNS may prefer partial repetitions over additive repetitions and even more over substitutive repetitions. He may prefer elaboration over paraphrase. However, for the NNSs there may be a growth in quantity and variety of, in particular, linguistic adjustments over time so that a full repertoire of trouble clarifications develops. As far as grammaticality is concerned, the NNSs' adjustments may often be ungrammatical due to their limited communicative competence. NS trouble clarifications, on the other hand, may normally be grammatical, but, in some cases ungrammatical linguistic adjustments may result from extreme attempts to simplify the input. It is expected that ínteraction type may influence the use of ungrammatical NS adjustments. In the institutional conversations, in which a more formal and technical register is used by the NS, and which is goal-oriented, more ungrammatical trouble clarifications aze expected to occur. Furthermore, it can be expected that while there is a diversity in forms for the NSs, this diversity might be less for NNSs in the initial stages but may grow over time. Interaction type may also influence types and forms of trouble clarifications. There may be a wider range of trouble clarifications in the informal conversations, where there are no constraints on negotiation of ineaning in terms of asymmetry. There may be more terminology and formal register which needs to be pazaphrased in the institutional conversations. As far as the forms are concerned, it was not clear beforehand what the influence of interaction type might be. 146 Chapter 8 Clarifying Trouble in Understanding 8.2 Native Speaker ~-ouble ClariFcations In this section, the types and forms of the NSs will be discussed. In section 8.2.1 the two interaction types will be compared. In section 8.2.2 the characteristics of the individual NSs will be discussed. The results for the NS trouble clarifications are presented in Table 8.1. The clarification types are ordered starting with pure conversational adjustments (full partial repetition), followed by weak linguistic adjustments (additive - complex repetition), and then by strong adjustments (paraphrase - translation), ending with the other category. The NSs use a broad range of clarification types and forms: SOqo are repetitions which were mainly questions, most frequently complex, but also partial, additive, substitutive and full repetition. About a third are elaborations. The great majority of these were in the form of declaratives. In addition, IOqo are paraphrases, which were generally questions. ó Linguistic form .y o Trouble clarification t yp e 6~ ~ 3 Q~ ...~ U ~ ~ o O Ó y c O ~o '~; ~ 7 O ,..r N ~ aJ ~ 7 ~ 7 7 ~ ~ ~ y ~ ~ ~y ~ ~~ ~ zo :~ ~ ~ ~ ~ ~ ~ ~ I I 13 Full repetition 11 8 3 4 Partial repetition I1 11 1 8 Additive repeti[ion l0 1 l0 Elaboration 10 1 IS 4 9 16 Substitutive repetition 7 1 10 1 2 3 Complex repetition 19 S 15 5 13 7 Paraphrase t] 3 Is z 10 1 3 3 18 Hypothesis forming C ~ 3 1 42 5.9 59 8.3 227 31.8 2S 49 6.9 3 57 12a 17.4 7 3 2t 72 lo.t a t a 3a a.s i 11 12 1.7 zi z2 3.1 7 I.o 2 0.2 S 1 t 163 1 7 Other clc e; 1 z9 Translation TOtal ,~ o F 9.0 Correction Word supply ~ ~ ba 9 z8 o z 79 12 90 19 74 41 IS 374 10 11.2 1.7 12.6 2.7 10.4 5.7 2.1 52.3 4.4 714 I00.0 Table 8. I: Types and linguistic forms of NS trouble clarifications 8.2 Native Speaker Trouble Clarifications 147 The NS trouble clarifications were most often adjustments of the interactional structure. All types of repetitions result in such adjustments. Typical for clarifications is that almost all these repetitions were self-repetitions rather than other-repetitions, in contrast to trouble indicators, which were most frequently other-repetitions (e.g. of keywords Chapter 7). Full repetition was used in only Sqo of the cases. In these cases minor adjustments of intonation, prosody and speed may have occurred. Apparently the NSs estimated that the NNSs needed more additional help than such minor adjustments, by either focussing (through deletion), regularization (addition) or by simplification (substitution) (Bingham Wesche, 1994). In addition, the NSs tried to clarify by being more explicit through the use of elaborations. As discussed before, through elaboration a trouble source can be decomposed into several small parts to make it more processible for the NNS. Through the clarifications a restructuring of the interaction takes place, because the NNS has to confirm reception of every piece of information with a gratuitous concurrence ja (yes). This can be seen in Example 10. (10) Institutional conversation with Ergiin, Cycle 1 ts PauL nou dan ga ik eens kijken well, I'll have a look of dat ik daar uh ruimte whether [ can find uh voor je kan vinden accommodation for you woonruimte living accommodation ti Ergiin: wat? what? rc Paul: ik ga kijken 1'll have a look go Ergiín: ja yeah rc Pau1: of daar een woning leeg is whether there is a vacant residence go Ergiin: ja yeah tc Paul: of leeg laimt or whether there is going ro be one vacant These vertical structures resemble the vertical constructions Scollon (1980) found in mother-child interaction or the so-called scaffolded interaction that Hawkins (1987) found in an elementary school ESL class with Low English Proficient children (see also Hatch, 1983b: 165-173). As can be seen in the above example, elaborations add new linguistic material but also may contain some adjustment of original linguistic material (woonruimte (living space) versus een woning (a residence). Also questions were used, in particular yeslno and declarative ones for elaborations (see 8.2. I). As stated before, the NS did not frequently use substitutions which revise linguistic material of the trouble source. The biggest category here are the paraphrases. It may be that it was the fotmal register in particular in the institutional conversations that was reformulated. Whether this is the case will be discussed in section 8.2.1. Furthermore, hypothesis forming that proposes models for the other's intended meaning was only used about Sqo of the time. Clarifications typically manipulate content and form of one's own original utterance rather than of the other's. This strategy will be discussed more extensively in the next chapter on confirmation checks because it is used much more frequently there. Three specific types of trouble clarifications were used in far fewer cases: word supply, correction and translation. Translatíon was used the least. Apparently the 148 Chapter 8 Clarifying Trouble in Understanding NS could not or would not resort to ihe use of other languages to clarify a problem. Word supply are reactions to NNS requests for a specific word, which appeared mainly in the informal conversations (see Chapter 7). In Example 11, Janet supplied the word for niece and explained the difference with nephew. (11) Informal conversation with Fatima, Cycle 2 ts Fatima: uh mijn t ook ~neeb - ~p-nlb ti docht uhldochter van mijn tante tc Janet: nichr cc Fatima: nicht tc Janet een neef is een jongen Fatima: uh nicht is een meisje tc Janer nicht meisje ac Fatima: uh my t also nephew daught uh~daughter of my aunt niece niece a nephew is a boy uh niece is a girl niece girl Corrections appeazed even less frequently. Sometimes, the NS corrected him or herself. In so far as corrections were used by the NS to correct the NNS, they pertained the content of the NNS's output more often than to the form. Such lack of form correction was also found in Gaskill (1980). An explanation for the lack of form corrections may be `face' related. Form corrections put into focus the difference between the communicative competence of the speakers, which threatens the NNS's positive face. Day et al. (1984) found that `on-record' corrections were used quite regularly by NSs who talked to NNS friends, in pazticular with lowlintermediate proficiency. In the current study, the interlocutors were strangers or only acquainted with each other through their participation in the research project. T'herefore, the imposition represented by a correction may be fairly great. It may be that even fewer cotrections occur in the institutional conversations because of the asymmetry in roles (see section 8.2.1). Apparently, familiarity is a more heavily weighing factor than level of communicative competence. The NNS informants of this study probably had a lower competence than the ESL college students from Day et al.'s study, but their low level of competence did not lead to many corrections. Over the three cycles, the number of corrections is more or less constant. The NSs used quite a few complex repetitions. They could be partial repetitions with substitutions or additions. In Example 12, Janet and Fatima pondered the fact that it is not habitual in the Netherlands to have guests for more than one or two days, the way it is in Morocco. (12) Informal conversatíon with Fatima, Cycle 1 ts ]anet: ja ik weet het niet uh f ik denk dat het komt dat de huizen zo klein zijn t ti tc Fatirna: Janet: ti tc Fatima: Jttnet: denk je niet? t waarom denk jij dat nederlanders uh dat niet dcen? tt ti~ ucunrn is dat? 8.2 Native Speaker Trouble Clarifications yes I don't know uh t I think it is that the houses are so small t don't you Ihink? t why do you think that the Dutch uh don'[ do that? ~ why is thar? 149 If we now look at the forms of the NS trouble clarifications, we can see that more than SO~o of the trouble clazifications were in the form of a declarative. About 20qo were elliptic declaratives. However, several question types also appeaz regularly, in particular yeslno questions, Wh-questions and elliptic questions. These questions may be repetitions or pazaphrases of trouble sources which were also questions. Long (1981) stated that NSs appear to have a preference for questions in foreigner talk discourse. In this manner, the NS initiates new content with new linguistic material and the NNS can react to it with a minimal answer (yes~no question) or by using material from the NS question (or-choice question). To answer NS Wh-questions may still be difficult because they request a specific type of information with relatively similar words without providing an answer model. A NS could simplify a vouble source by changing the question type (Hatch 1983b); for instance a Wh-question could be reformulated as an or-choice question, providing the NNS with an answer model. A yes~no or declarative question would make the linguistic task even easier because they require only confirmation (a nod or `yes') instead of the production of complex output. Long, for instance, found significantly more NS or-choice questions in foreigner talk discourse, which may be evidence of such adjustments (Long, 1983b: 180181). Some sequences with such adjustments could be identified, such as Example 13, in which Janet reformulated a Wh-question as a yeslno question. (13) Informal conversation with ErgGn, Cycle 1 ts Janet: tí tc ti co~tc Ergiin: lanet: Ergiin: Janet: Ergun: Janet: Ergiin: Janet ti mc ti tc hij woontl hij gaat ook in uh [fJ he? waar gaat hij wonen? waar? jajouw broer? weet ik niet ja weet je wel o f- daughs~ woont hij hierin tilburg? he livesf he is also going to [t] huh? where is he going to live? where? yes your brother? 1 don't know yes you know o t- daughs~ does he live here in tilburg? In Chapter 6, example 4 contains a sequence in which the NS rephrases his original Wh-question wat voor huis zoek je? (what kind of house aze you looking for?) in three ways from or-choice back to Wh- and then to elliptic and then to yeslno question. Another question form which did not appeaz very often was the negative question. This type will be discussed in Chapter 9 because they were more frequent in confirmation checks. 8.2.1 The Influence of Interaction Type In this section a compazison will be given of the relative frequency of NS clazification types and forms used ín the informal conversations and in the institutional conversations. It should be kept in mind that the NSs participating in the in- 150 Chapter 8 Clarifying Trouble in Understanding formal conversations differed from the ones in the institutional conversations. In the informal conversations, except for one conversation with Peter, the scores represent Janet's clarifying behavior. The scores in the institutional conversations reflect both Bart, who pazticipated in two-thirds of the conversations, and Paul, who participated in one-third. As for the distribution of types, we can see in Table 8.2 that, in general, the percentages do not differ much. There appears to be an influence of interaction type. The use of repetitions is more or less equal except that complex repetition occurred more often in the instítutional conversations; in the informal conversations, paraphrase and hypothesis forming are used more. NS trouble clarification types Informal conversations Insti[u[ional conversa[ions ~ ol0 4k olo Full repetition 22 7.0 20 5.0 Partial repetition 27 8.6 37 9.3 Additive repetition 22 7.0 37 9.3 Elaboration 124 39.5 103 25.8 Substitutive repetition 20 6.4 29 17.3 Complex repetition 44 14.0 80 20.0 Paraphrase 12 3.8 60 15.0 Hypothesis forming 6 1.9 28 7.0 Correction ]0 3.2 2 0.5 Word supply 20 6.4 2 0.5 Translation 7 2.2 0 0.0 Other 0 0.0 2 0.5 Total 314 100.0 400 100.0 Table 8.2: Distribution of NS clarification rypes per interaction type In the informal conversations, elaborations and word supply aze more frequently used. Correction and translation occur in both types infrequently but clearly more in the informal conversations. Elaborations were relatively often declarative questions in the institutional conversations. Bart used information he received earlier in the conversation to explain to Mahmut the system of urgency credits with which Mahmut was not acquainted. Credits could be received, for example, for social problems such as Mahmut had with the noise level of the discotheque next door (see Example 14). Bart used something Mahmut had already told him to explain the urgency credit system. 8.2 Native Speaker Trouble Clarifications 151 (14) Institutional conversation with Mahmut, Cycle tc Bart: als bijvoorbeeld problemen zijn gc Mahmut ja Batt: tc in de buurt gc Mahmut: ja tc Bart: he problemen met buren Mahmur. gc ja Barr: tc je praat over t een disco Mahmut: ja gc tc Bart: he t naastje deur t veel herrie gc tc gc tc Mahmut: Bart: Mahmut Bart: gc tc gc Mahmut-. Bart: Mahmut: cm Bart: tc gc tc gc tc gc tc gc tc Mahmut: Bart: Mahmut: Bart: Mahmut: Bart: Mahmut: Bart: ac Mahmut: ja kindje weinig slapen ja vrouw weinig slapen t dan kunnen we dad er zijn formulienjes ja mcet je invullen ja 3 if for example problems are yes in the neighborhood yes huh problems with neighbors yes you talk about t a discotheyue yes huh t next door t a lot of noise yes baby little sleep yes wife little sleep t then we carJ there are forms yes you have to fill out yes ja yes die mcet je hier bij mij weer terug brengen ja met dat formuliertje ja kan ik naar een commissie tce ja en kan ik daar over praten ja en vragen of dat we je eerder mogen helpen ja waar is die ~formub halen? - r-formulien those you have to bring back to me yes with that form yes I can go to a committee yes and 1 can talk about it yes and ask whether we can help you earlier yes where is that ~formub get? - r-form~ The difference in outcomes between the two interaction types can be related to the differences in NS~NNS role and the goal-orientedness of the conversation. In the institutional conversation, the housing ofiicial had the task of filling out a form with the information provided by the NNS. This information transfer may have been impeded by specific terminology and indirect language used by the housing official. This may explain why there is more paraphrase. One might expect that paraphrase would lead to simplification. However in Example 15, Bart, the housing official first replaced one idiomatic expression to refer to trouble between father and son with another expression just as opaque. Both expressions are more or less euphemis[ic. When this did not help, he used an ungrammatical phrase (no verb and no possessive pronoun) with the more common word `ruzie' (quarrel), replacing the idiomatic phrase with one that is also much less face protective. The original idiomatic expressions are not quite as direct as the ungrammatical paraphrase, which may indicate that the housing official considered the topic to be sensitive. He probably used the less face protective phrase because he had to acquire this information on such private matters to determine the need for independent housing. 152 Chapter 8 Clarifying Trauble in Understanding (IS) Institutional conversation with Ergiin, Cycle 3 Bart: ja je kunt met elkaar niet yeah, you can't get along ts meer overweg? with each other? Ergun: wat? what? ti tc Barr. je kunt niet goed opschieten you can't get on well met je vader? with yourfather? ti Ergun: t ik begrijp niet wat t I don't understand what bedcelt means rc Barr: nou ruzie met vader? we!!, yuarre! with father? mr Ergun: ja yeah Bingham Wesche (1994: 228) hypothesizes that paraphrase may províde NNSs with richer lexical and structural data. This is certainly the case in the first trouble clazification. However, for less low proficient speakers such rich data are not necessazily effective in making input comprehensible. Probably Ergun did not analyze the first (trouble source) nor the second expression even after he understood the gist by the third utterance. The second clazification, which is less rich and even ungrammatical, is definitely effective from a communicative point of view. The asymmetry in the institutional conversations, in addition, may lead to more euphemistic, indirect, and formal register, which may be used to protect the NNS positive face when sensitive topics aze discussed. In the informal conversations, there is less role asymmetry. This may explain why correction occurs more in this context. However, even there, most corrections were self-correction directed towazds content. In Example 16, Janet used an on-record correction of Mahmut's choice of words. She did this in cycle 3 when she and Mahmut had gotten to know each other (see Day et al. 1984). Interestingly, she did not precede it with a`no' but with a`yes'. This may soften the correction. (1~ Informal conversation with Mahmut, Cycle 3 ts~ti Mahmut ja ~passab die met [t] - ~brand name~ lanet: [~~] - ~whisdes~ Mahmut: cc ti cc coltc tc tc cc co aclcc co Janet-. Mahmut: Janet Mahmut: caravan he dieldie dingen ja t passat die dingen diesel bestel? wat? bestel? nee bijna bestel ja Janet: t3t ja caravan he caravan ja dat is un bestel f Mahmut: Janet: Mahmut-. caravan is un bestel t zo'n auto als ik heb jaja zelfde jaja bestelauto bestelauto 8.2 Native Speaker Trottble Clarifications yes ~passat~ those with [t] - ~brand name~ [O] - whisdes~ cazavan huh thoselthose things yes f passat those things diesel pickup? what? pickup? no almost pickup yes t3t yes caravan huh caravan yes rhat is a pickup t caravan is a pickup t a car like 1 have yeah yeah same yeah yeah pickup wck píckup uvck 153 The results differ from those of Crookes 8t Rulon (1985: 15), who found that `corrective feedback' was given less in free informal conversations than in task-related conversations. Day et al. (1984) found that correction occurred equally ofren in informal conversation as in task-related game activities. The institutional conversations were task-oriented, but still little correction occurred, probably because of the status differences. Also, word supply occurred more often in the informal conversations. This might also have resulted from the more symmetric and less face threatening situation. Woken 8c Swales (1989) found that when the NNSs had expert status, the NSs never offered verbal help, such as offering vocabulary explanations, and the NNSs never asked for it. NNS expert status may thus inhibit negotiation of ineaning because it would threaten the NNS's status. In the institutional conversations in which the NSs had higher status, there was little word supply either. Appazently status asymmetry in either direction prevents NSs and NNSs from directly negotiating vocabulazy items. As to the relative frequency of forms, there is a difference in the distribution of declazatives versus interrogatives (see Table 8.3). Declaratives appeaz most frequently in the informal conversations. As we have seen earlier these were most frequently elaborations. In the institutional conversations, more questions were used (5óqo versus 34qo in the informal conversations). This may be explained by the format of the housing ofiice conversation, which is an interview. The housing official asked many questions dealing with many kinds of information and thus were not always understood by the NNS. However, problems could not be left as easily because of the conversational goal of information exchange. This may lead to rephrasings of questions. Yeslno, elliptic and Wh-questions were used regularly as could been seen in earlier examples. The declaratives used in the institutional conversations were less frequently elliptic than in the informal conversations where the distribution is about equal. 8.2.2 Individual Portraits of the Native Speakers In the next section the NS's trouble clarifications will be discussed per informant to show where they differ from the average picture. The distribution of types and forms is not presented in tables but will be given in percentages in the text as faz as they are relevant. Janet In the informal conversations, Janet most frequently used elaborations (approximately 40qo). The average percentage over the four NNSs decreased more than SOqo over time. These were mainly declaratives, indicating that she was explaining and telling things-rather than asking questions-when problems occurred. She in fact used decomposition regularly (Larsen-Freeman 8z Long, 1991). Twenty percent of the time she used adjustments with revisions of linguistic material, which is relatively frequent. As stated before, she in pazticular used word 154 Chapter 8 Clarifying Trouble in Understanding NS trouble clarification forms Informal conversations Institutional conversations ~ elc fl clo Wh-question 21 6.7 58 14.5 Or-choice question 3 0.9 9 2.3 Yeslno question 37 11.8 53 13.3 Negative question 8 2.5 1I 2.8 Elliptic question 20 6.4 54 13.5 Declarative question 12 3.8 29 7.3 Other questions 3 1.0 12 3.0 Full declarative 105 33.4 ]O8 27.0 Elliptic declarative 96 30.6 65 16.3 Other 9 2.9 I 0.3 Total 314 100.0 400 100.0 Table 8.3: Distrihntion of N.S elarification forms per interaction rype supply, correction and translation which have an educational function in addition to a communicative one. Two reasons can be given for this. Firstly, the type of conversation allows space for these kinds of negotiations. There was no time pressure caused by a script or obligatory information exchange. Secondly, Janet worked with speakers of Dutch as a second language in an educational context and thus she may have felt inclined to help and educate her interlocutors. She first used more partial repetitions and later more additive repetitions, which are probably harder to process. Although she was instructed to interact in Dutch, Janet was the only NS who used translations. She had some competence in Moroccan Arabic and particularly in the first cycle with Fatima and Mohamed, she sometimes resorted to translations, as can be seen in Example 17. (17) Informal conversation with Mohamed, Cycle 1 ts Janet ik wil graag pra[en over t I would like to talk about t verschillen tussen marokko differenccs between morocco en nederland and the netherlands ti Mohamed: t t tc Janet: ~el fary~ ~elfarq~ - r-m-the difference~ - ~-m-the difference~ cm Janet: ja? yes? co Mohamed: ja ~el farq~ yes ~el farq~ - ~-m-the difference~ - r-m-the difference~ tc Janet: ~fary~ tussen nwmkko ~farq~ between morocco en nederland [tj and the netherlands [t] - ~-m-difference~ - ~-m-difference~ 8.2 Native Speaker Trouble Clarifications 155 Paul In the first cycle of the institutional conversations, Paul used a variety of clarification types. He used elaborations in about 25qo of the cases, which is less often than Janet. He used repetitions in about SOqo of the cases too. This is a relatively high score. These repetitions create redundancy, which may support understanding and give a second opportunity for processing. Repetition is the least creative and thus the easiest way of clarifying. Paul, in particulaz, used full and partial repetitions frequently. This may indicate that Paul was relatively more at loss when a problem occurred than the other NSs. However, he also used paraphrase quite frequently. Paul did not appear to be very sensitive in adjusting his input. He even used uncommon proverbs, as in [he following example. As can be seen in Example 18, his paraphrase did not resolve the trouble either. Apparently he had trouble adjusting to a level which Mohamed could comprehend and thus gave up and closed the conversation. (18) Institutional conversation with Mohamed, Cycle 1 Mohamed: ik uh ft mcet uhl I uh tt must uhl ik mcet een huis hebben I must have a house Paul: ja yeah ts we kunnen geen uhl you can't uhl ti tc Mohamed: Paul: tc ti es Mohamed: PauL we kunnen geen ijzer met handen breken he you can't break iron with your hands huh ~we can't do the impossible~ wat? her is moeilijk he we hebben niet zoveel huizen er is woningnood hè [tf] [tt] (xx) het is half zes t uhtneet volgende maand terugkomen what? it is difficult huh we don't have many houses there is a housing shortage huh [tt] [~] (xx) it is five thirty t uhtnot next month come back He did not use word supply or translations. Paul's clarification were in SSqo of the cases questions. Bart Like Paul, Bart used elaborations about 25ofo of the time. Thus, both housing officials used elaborations less than Janet. Bart used forms of repetitions 45010 of the time (mainly complex ones). Like Paul, he used questions more than declaratives. This may have to do with the type of interaction, which does have an influence. In particular, like Paul, he used paraphrase regularly and only a couple of times correction and word supply. These paraphrases were much more frequently ungrammatical in form than those of the other NSs. This was also reflected in his more frequent use of incomplete elliptic questions and declaratives (43qo versus Paul 20qo and Janet 35qo). In the next section, some aspects of NS ungrammatical adjustments are discussed more extensively. From a language acquisition focus, it is important to determine the conditions which lead to ungrammatical input which does not support acquisition of standardlike target language. 156 Chapter 8 1 Clarifying Trouble in Understanding 8.2.3 Ungrammatical Foreigner Talk One may wonder why Bart used ungrammatical adjustments regularly when Paul en Janet did not. Conditions which may predict the use of ungrammatical foreigner talk are formulated by Long (1983b: 179): 1. The NNS has a very low proficiency in the L2 2. the NS perceives himself as being of higher social status 3. the NS has prior FT experience with low proficient NNSs 4. the conversation occurs spontaneously. Long claims that conditions I, 2, and 4 seem to be necessary conditions and that no single one is sufficient. The first two conditions concerning NNS communicative competence and status are applicable for both native speakers in the institutional conversations. As to the third one, only a guess can be made, since there are no data available of interactions with Paul and Bart in their former work contacts with NNSs. From the relaxed way Bart conducted the conversations and the knowledge that he indeed had regular contact with NNSs similar to those in the study, it may be inferred that Bart was in the habit of using ungrammatical foreigner talk. One of the participating researchers mentioned that after an assistant on the project actually confronted Bart with this foreigner talk use, he became more self-conscious and diminished his ungrammatical use (Brceder, personal communication). The fourth condition of spontaneity in conversation is harder to apply to the conversations studied. Long (1983b) exemplifies what he means by spontaneous by referring to task-oriented communication of the factory floor and contrasts it with arranged encounters with strangers in the research laboratory. The institutional conversations studied are in between these two extremes. However, conversation even if it is task-oriented or an arranged role play in a research lab, is still spontaneous as far as the language used is concerned. In this interpretation, the fourth condition of spontaneity would also be applicable, in particular with Bart, who is playing his own professional role and thus slipping in his routine interaction behavior. Therefore, it may not be surprising that he used ungrammatical foreigner talk. Snow et al. (1981) and Jakovidou (1993) hypothesize topic choice may influence the occurrence of ungrammatical foreigner talk. Snow et al. (1981) state that abstract complex topics lead to more ungrammatical adjustments. This may particularly be true when grammatical adjustments did not lead to comprehension. From the data, it appears that ungrammatical foreigner talk was often used by Bart after a grammatical utterance did not work, as in Example 152. Thus ungrammatical foreigner talk is sequentially placed after grammatical talk. In Example 19, Bart tried to find out what Fatima's current living conditions were. He first asked how many people lived there. Fatima answered three. Apparently he did not trust this answer because after another sequence he again resumed this topic and did finally get another answer. The initial questions yeslno question whether Fatima lives there on her own is grammatical and the continuation en hoeveel familie woont daar?(and how much family lives there?) is not correct. It should be en hoeveel familieleden wonen daar?(and how many members of your family live there?). It could, how- 8.2 Native Speaker Trouble Clarifications 157 ever, be acceptable in everyday speech. The questions in between are all ungrammatical. (19) Iastitutional conver~sation with Fatima, Cyde 2 ts ti tc Bart: Fatima: woon je daar alleen? t do you live there alone? t eart: zelj tc tr Fatima: ac~tc tr Bart Fatima: geen man geen [kinderen]? [nee] ik uh heb t lik trouwt ik heb een kinder ac Bart: tr cc co ac Fatima: Bart: Fatima: Bart: ts ti Fati ma: tc Bart: mr Fatima: cc colmr Bart: Fatima: selj ja waar is man met kind? met ookl[met] no man no [children]? [no] I uh have t II marries 1 have one children yes where is man with child? with alsol[with] uh familie t woon ook bij familie wonen? ja jaja uh family t live also live with family? yes uhuh hoeveel mensen samenwonen in da[ huis? zes how many people together live in that house? six six people? yes woman man four children [ook] [also] en hceveel familie woont daar? and how much family lives there? f t zes mensen? ja vrouw man vier kinder Paul and Janet used ungrammatical clarifications very rarely. They may have been more influenced by the experimental setting. Paul, the social worker, did not resort to ungrammatical talk but used repetitions most frequently. Impressionistically, Bart appeared much more communicatively effective than Paul. For Bart the ungrammatical adjustments apparently have a communicative function. As can be seen from Examples 15 and 19 the simplifications appear to increase understanding. As Ellis (1985) concluded, goal-oriented (e.g. institutional) communication may lead to ungrammatical adjustments because the focus is not on language form but on goal achievement. For Janet, condition one concerning the NNS communicative competence is applicable. Condition two concerning status is more questionable. It cannot be doubted that in the project she as a highly educated native researcher has a higher status than the NNS informants. However, in the conversational context itself, her status is of no importance for the subject matter discussed (differences between Turkey~Ivlorocco and the Netherlands and other personal topics). The NNS informants had as much or even more expertise on these matters. From this it could be concluded that the second condition also needs to be specified, in the sense that the status difference should be relevant for the topics discussed in conversation (Zuengler 8t Bent, 1991). Hatch (1983a) hypothesized that the goals of foreigner talk may be to promote communication, to create an affective bond, or it may function as an implicit teaching mode. However, Hatch, Shapira 8t Gough (1978) reported ungrammatical use also between a native-speaker and a nonnative-speaker who were friends, a situa[ion in which status did not play a role at all. The fact that Janet also used elliptic utterances quite regularly may point to simplification. Therefore, 158 Chapter 8 Clarifying Trouble in Understanding this condition may not be necessary. Another influence on lack of ungrammatical foreigner talk may be related to condition three. Her experience with NNSs in her professional role as a Dutch-as-a second-language consultant has provided her with more of a teacherlike register than foreigner talk register, which she resorted to if necessary. She appeared quite adept at adjusting to the NNS's level of comprehension. The fourth condition is also more doubtful for her. As a project researcher, she may have been very much influenced by the experimental condition, which may have inhibited the use of foreigner talk. In conclusion, NS adjustments provide NNSs with well-formed, simpler, more regular, more explicit and more redundant language (Bingham Wesche, 1994: 233) through the application of three adjustment processes, deletion, addition and substitution. The processes of deletion and addition aze used more frequently than substitution. Repetition was used in combination with these processes, which led to more redundancy. Therefore, the types of adjustment used are more conversational than linguistic. Addition led to more well-formed and more regular forms. More explicitness was reached through frequent use of elaboration. Simpler utterances were mainly achieved through deletion and less through substitution. There is considerable variation among the NSs. This outcome is similaz to what was found by Long (1983b: 184), Bingham Wesche (1994) and Jakovidou (1993). Janet used more implicit language-teaching devices, such as word supply and correction. Paul used most simple full and partial repetition, which may show his discomfort in adjusting his language behavior. Bazt, in particulaz used ungrammatical adjustments. NS ungrammatical adjustments may be influenced by several factors, such as leazner characteristics, experience in dealing with NNSs, status differences, type of conversation, topic choice and sequential placement. Bart's `foreigner talk' may not be conducive to language learning, but it was effective in communication. 8.3 Nonnative Speaker T~ouble Clarifications In this section, the types and forms of the NNSs will be discussed. In section 8.3.1 the two interaction types will be compazed. In section 8.3.2 the characteristics of the individual NNSs will be discussed. The scores for the trouble clazification types and forms were also calculated per cycle for every NNS. However, there generally appeared to be little development in trouble clarifying behavior. Therefore, the longitudinal results are discussed per NNS informant only where applicable. The tables of the longitudinal results are presented in Appendix 3(Tables 5-12). In Table 8.4, we can see that the NNSs used repetitions, in particular partial, complex and additive ones frequently (SOqo). Similar to the outcomes for the NSs, these NNS repetitions almost exclusively turned out to be self-repetitions. In addition, elaborations were used 40010 of the time. 8.3 Nonnative Speaker Trouble Clarifications 159 Linguistic fortn Trouble clarification type ~ ; ó rr ~' 3 Full repetition 1 Partial repetition a Additive repetition i á' u ~ L ~ ~ o ~o o ? p. ~ o' ~ - C } t -~ ~ c ~ ~ i - ~ ~ y ó ó ~ io 1 iz 3.6 2 as z s6 ~7.0 t 33 36 io.9 izb 38.3 Elaboration ~2a Substitutive repetition 13 13 4.0 45 51 15.5 a 5 I.5 z o.6 20 6J o o.o 6 ~.8 2 0.6 COmplex repetition 3 1 I 1 Paraphrase i Hypothesis forming z z COrrec[ion 19 1 Word supply Translation 6 Other i i TOtal 10 I 2 6 I 297 t2 ~0 3.0 0.3 0.6 1.8 0.3 90.3 3.6 329 100.0 Table 8.4: Types and linguistic forms of NNS trouble clarifications Pazaphrases and translations each occurred only in 1.Sqo of the cases. Corrections occurred remarkably often. There may be individual preferences (see 8.3.2). Word supply did not occur, indicating that is a typical NS strategy. As far as the forms are concerned, the NNSs used declaratives in 90qo of the cases. SO~Jo were elliptic declaratives indicating that many trouble clarifications were sentence fragments (see Table 8.6). The high occurrence of partial repetitions proves this too. The NNS's comprehensible output in oral interaction does not consist of grammatical, full, finite sentences. Only 6qo of the clarifications were questions. This may show that the NNSs either were not initiating topics by asking questions or that only their declaratives caused problems and had to be clarified. Given the reactive role the I~1NSs generally played in the conversations, the first explanation seems most plausible. Wh-questions were used most regularly. This may be explained by individual preferences, as can be seen in section 8.3.2. 160 Chapter 8 Clarifying Trouble in Understanding 8.3.1 The Influence of Interaction Type After the general presentation of the NNS results, the distribution of clarification types over the two interaction types will now be discussed. The scores of the informal and institutional conversations present the same speaker only a month apart in recording. As can be seen in Table 8.5, there are few differences in distribution. This seems to indicate that interaction type has little influence on NNS clarification behavior. This was not the case with NNS trouble indicating behavior. For the trouble indicators there was some difference in distribution for the NNSs, probably due to face protection, which occuned more in the institutional conversations. Apparently trouble clarification behavior is not influenced by face concerns. One small difference found was that translations appeared only in the informal conversations. For the Moroccan informants it is understandable that they attempted clarification in Moroccan Arabic because Janet had a minimal competence in this language. They also used some French to clarify problems. However, from the transcripts it appeared that the Turks Ergun and Mahmut also attempted clariiication through the use of Turkish in the informal conversations, even though Janet did not understand this language. This phenomenon will be NNS trouble clarification Informal conversations Institutional conversations lk qo fk qo 7 4.7 5 2.8 Partial repetition 27 18.1 29 16.1 Additive repetition 17 11.4 19 10.6 Elaboration 57 38.3 69 38.3 3 2.0 10 5.6 22 14.8 29 16.1 Paraphrase 2 1.3 3 1.7 Hypothesis forming 0 0.0 2 1.1 Correction 8 5.4 12 6.7 Word supply 0 0.0 0 0.0 Translation 6 4.0 0 0.0 Other 0 0.0 2 1.1 Total 149 typeS Full repetition Substitutive repetition Complex repetition 8.3 Nonnative Speaker Trouble Clarifications 100.0 180 100.0 161 discussed further in section 8.3.2 (Ergun). In the institutional conversations, neither NS had any competence in Moroccan Arabic or Turkish. Only French could have been of use for the Moroccan informants in clarification. However, apparently none of the NNSs attempted to use other languages to clarify in the institutional conversations. The distributions of the forms of the NNS trouble clarifications also differed only marginally (see Table 8.6). In both interaction types declaratives were used overwhelmingly often. The distribution of full and elliptic declaratives is more or less the same. Only elliptic questions appeared a few times more in the institutional conversations. The `other' category is relatively large for the infortnal conversations. This is due to utterances in other languages such as the translations which were categorized under this category. These utterances in other languages were not coded for form. NS trouble clarification forms Informal conversations Institutional conversations ~ ~lo lk ~lo Wh-question 4 2.7 6 3.3 Or-choice question 1 0.7 0 0.0 Yeslno question 1 0.7 1 0.6 Elliptic question 1 0.7 5 2.8 Declara[ive question 0 0.0 1 0.6 Full declarative 54 36.2 72 40.0 Elliptic declarative 76 51.0 95 52.8 Other 12 8.1 0 0.0 Total 149 100.0 180 ] 00.0 Table 8.6: Distribution of NNS trouble clarificationforms per interaction rype 8.3.2 Individual Portraits of the Nonnative Speakers In the next section, the results will be discussed for every NNS separately. Their results are presented in raw scores and percentages in Tables 8.7 and 8.8. Fatima Over the three cycles, Fatima used all the types of trouble clarifications except word supply, which was used only by the NSs, and hypothesis forming (see Table 8.7). She used elaborations most frequently but also used complex, partial and 162 Chapter 8 Clarifying Trouble in Understanding Fatima NNS trouble Mohamed Mahmut tk qo tk olo ~ olo fk Full repeti[ion 3 4.0 2 6.1 4 2.5 3 4.7 Partial repe[ition IO 13.3 I1 33.3 25 15.9 10 15.6 Additive repetition 8 10.7 3 9.1 l8 I I.5 7 10.9 Elaboration 26 34.7 4 12.1 63 40.1 33 51.6 Subs[itu[ive repetition 3 4.0 1 3.0 8 5.1 1 1.6 Complex repetition 14 18.4 5 15.2 25 15.9 7 10.9 Paraphrase 3 4.0 0 0.0 2 1.3 0 0.0 Hypothesis forming 0 0.0 0 0.0 2 1.3 0 0.0 Correction 5 6.7 5 15.2 8 5.1 2 3.1 Translation 3 4.0 1 3.0 1 0.6 1 1.6 Other 0 0.0 1 3.0 1 0.6 0 0.0 Total 75 100.0 100.0 64 ] 00.0 clarification types 33 oIo Ergun 100.0 157 Table 8.7: NNS trouble clarification rypes per informant additive repetitions regularly. As for the longitudinal development, the number of elaborations decreased somewhat over time in favor of complex repetitions (see Appendix 3, Table 5). This may indicate a small increase in vaziety and complexity of the trouble clarifications, which might be related to her growing communicative competence. In Example 20, Fatima used complex repetition when she tried to make clear to the housing official that if the housing corporation did not help them, she might put an ad in the newspaper. This attempt caused trouble because her utterance did not contain a subject. Bart, therefore, requested specification of the subject. However, Fatima did not react by specifying the subject, but instead clarified by substituting krant (newspaper) with nieuws (news) and by deletion of the rest of the utterance. Still this complex repetition is minimal in form. (20) Institutional conversation with Fatima, Cycle 3 ts Fatima: misschien uh straks maybe uh later schrijf in die krant write in the newspaper ti Bart: wíe? who? tc Fatima: nieuws news tilcc Bart: wie? man? who? husband? coltc Fatima: nee die mijn uh huis no that my uh house misschien die mensen uh [f] maybe those people uh [t] gc Bart: [ja] [yes] Fatima: n11I exchange 8.3 Nonnative Speaker Trouble Clarifications 163 cc co Bart: Fatima: ruilen? ja exchange? yes The forms of Fatima's trouble clarifications did not show much variety nor any development (see Appendix 3, Table 6). She used mainly declaratives from which almost half were elliptic declaratives. The other category contains only utterances in French and Moroccan Arabic, which were either translations or repetitions (of the trouble source), which Fatima apparently resorted to when she could not produce any comprehensible output in Dutch. These translations may reflect her limited linguistic competence in Dutch. Fatima NNS trouble Mohamed Mahmut Ergun ~ ~o ~k olo il elo tl olo Wh-question 0 0.0 6 18.2 3 1.9 I I.6 Or-choice ques[ion 0 0.0 1 3.0 0 0.0 0 0.0 Yeslno question 1 l.3 1 3.0 0 0.0 0 0.0 Elliptic question 0 0.0 1 3.0 3 1.9 2 3.1 Declarative question I 1.3 0 0.0 0 0.0 0 0.0 Full declarative 25 33.3 8 24.2 65 41.4 28 43.8 Elliptic declarative 43 57.3 15 45.5 82 52.2 30 46.9 Other 5 6.7 1 3.0 3 1.9 3 4.7 Total 75 100.0 33 100.0 157 100.0 64 100.0 clarification forms Table 8.8: Forms of NNS trouble clarifications per informant Mohamed Mohamed used trouble clarifications relatively infrequently (33 over 6 conversations compazed to 157 for Mahmut). He used most trouble clarifications in the third cycle (see Appendix 3, Table 7). Mohamed, in contrast to the general picture, did not use elaboration regularly, but instead opted most frequently for partial repetitions. He also used some correction in later cycles. Mohamed was the least cooperative informant, in particular at the end of the data collection. The low number of trouble clarifications and the most frequent use of the least demanding strategy of partial repetition may be evidence of this attitude. In Example 21, Mohamed wanted to compare the municipality of the Moroccan city Casablanca to the Dutch town Tilburg. This resulted in a lengthy clarification sequence. The main problem is caused by the fact that the concept of municipality differs in the Netherlands and Morocco. The town of Tilburg belongs to one larger municipality (county), whereas the city of Casablanca consists of five municipalities. The problem started at the point where that Mohamed answered Peter's question how big Casablanca is 164 Chapter 8 C[arifying Trouble in Understanding in square meters with a counterquestion which Peter could not relate to his former question. He, thus, did not understand Mohamed's pronunciation of the word gemeente (municipality) because the last `e' (schwa) lacked. This made the word look like a past participle of the verb menen-gemeend (intend-intended). This confused the NS. Mohamed initially clarified through the use of three partial repetitions of the perceived problem word in the trouble source gemeente (municipality). Only when that clearly did not resolve the problem did he use other means. (21) Informal conversation with Mohamed, Cycle 3 Mohamed: hier in uh ~ tilburg? hceveel gemeent zijn hier in t? Peter: hceveel? Mohamed: gemeent Peter: t hceveel mensen? Mohamed: Peter: Mohamed: Peter: Mohamed: Peter: Mohamed: Peter: Mohamed: Peter: Mohamed: Peter Mohamed: Peter Mohanxd: Peter: cc coltc ti tc ti tc Mohamed: Peter: Mohamed: Peter: Mohamed: mr cc co Peter: Mohamed: Peter: gemeent t gemeenr ja in tilburg? (tJ [ja] hoeveel gemeent? ik begrijp het niet gemeent weet je wat is gemeent? een gerneente? ja t weet je wat is? ja gemeente`[ja] `(ja] hoeveel gemeent hier in tilburg? ~ ik begrijp uhlvierkantl vierkante meter bedcel jij? nee gemeente t hoeveel gemeent hier? t hoeveel een gemeent twee gemeent in tilburg? eentje eentje maar ja here in uh t tilburg? how many municipality are here in t? how many? rnunicipality f how many people? municipality t municipality yes in tilburg? (t J [ja] how many municipaliry? 1 don't understand municipality do you know what is municipality? a municipality? yes t do you know what is? yes municipality`[yes] `[yes] how many municipaliry here in tilburg? tt I understand uhlsquare square meter you mean? no municipaliry t haw many municipaliry here? t how many one municipaliry two municipality in tilburg? one only one yes Peter was totally at loss as to what Mohamed intended.lfierefore, he regularly did not know how to respond and made wild guesses. Mohamed varied his clazifications with minimal changes. After the partial repetitions of the perceived problem word gemeent, he repeated ottter parts of his original question. Then he checked Peter's understanding of the word gemeente. Consequently, he repeated his question with hypotheses for possible answers. He appeared quite resourceful in the clarification of trouble. 8..3 Nonnative Speaker Trouble Clarifteations 165 About the fotms of Mohamed's clarification nothing more can be said except that he mainly used declaratives, most of which contained ellipsis (see also Appendix 3, Table 8). In the third cycle, he also used quite a few Wh-questions which may reflect his growing competence in Dutch (see Example 21). Mahmut Mahmut clazified a great deal and most frequently used elaborations to do so (40qo). In addition, Mahmut used partial, additive, substitutive and complex repetitions in more than SOqo of the cases. He used corrections more than any other speaker. These trouble clarifications probably followed confirmation checks of the NS, in which a wrong hypothesis was given of Mahmut's intended meaning. This hypothesis was then corrected. Apparently face issues played less of a role when the correction came from the lower status NNS. All corrections concerned the content rather than the form of the NS's utterance and were accompanied by a rejection (nolnot), as can be seen in Example 22. (22) Institutional conversation with Mahmut, Cycle 2 ts Mahmut en dan vijftienl and then fifteenl vijftien gulden bij fifteen guildets with cc Bart: vijftíen? fifteen? co Mahmut: ja vijftien gulden yes fifteen guilders cc Bart: nu tweehonderdvijfenvijftig? now two hundred fifty-five? coltc Mahmut: nee tweehonden~ijfenzestig no two hundred sixty-ftve Numbers (concerning amount of rent etc.) caused quite a few problems. Another correction came after Bart checked Mahmut's own misiake when he said he annually sent tweehottderdvijftig (two hundred fifty) guilders and meant tweeëtteenhalfduizend (two-and-a-half thousand) to his mother in Turkey. There is no evidence of development towazds more variety of trouble clarification types. The forms show a homogeneous picture too. Mainly declaratives were used in all three cycles (with or without ellipsis) (see Appendix 3, Table 9 and 10). Ergun Of the four NNS informants, Ergun turned out to use the greatest number of elaborations (SOqo); however, he also used partial and additive repetitions regularly. Again for Ergun there appears to be little longitudinal development (see Appendix 3, Table 11). Like Mahmut, he used a translation once, although Janet did not understand Turkish. He more often used Turkish words if he did not know the Dutch ones for what he wanted to express. In Example 23, he talked to Janet about weddings. His brother was going to marry and Janet wanted to know whether they were going to make a song or an act about the bride and groom as is common at Dutch weddings. Ergun did not understand what Janet was referring to. He thought she was talking about the wedding dress. He expressed that he did not know the Dutch word. When Janet provided him with the word jurk (dress) he checked it by using the Turkish word. However, this she did not understand. 166 Chapter 8 Clarifying Trottble in Understanding (23) Infortnal conversation with Ergiin, Cycle 1 ts Ergiin: ~gelinliklgelinlib - ~-t-wedding dress~ ti Janet he? tc Ergun: gelinlik ti Janet gel? tc Ergiin: gel - in - lik cc Janer. gelinlik co Ergun: ja tgelinliklgelinlik~ - ~-t-wedding dress~ huh? gelinlik gel? gel - in - lik gelinlik yes Maybe Ergun and Mahmut sometimes used translation of their own initiative because Janet at other times requested translations. She may have hoped that some recognizable borrowings from Arabic or French, of which there are many in Turkish, would clarify the trouble. In the forms, there is also little development (see also Appendix 3, Table 12). Ergun mainly used both full and elliptic declaratives. In conclusion, the NNSs used several repetitions most frequently (SOqo). The frequent use of repetitions by the NNSs shows that they preferred the easiest means of clarification, in particular, partial and additive repetitions. Elaborations were used frequently too, in particular by Ergun and Mahmut. This may indicate that addition in the form of continuations and expansions of a topic is also quite easy and can be done by minimal means. Trouble clarification types which required substitution of linguistic form were used very infrequently. From the trouble clarification types which contained linguistic adjustments, corrections were used the most, in particular by Mohamed and Mahmut. These corrections concerned false NS hypotheses of their own intended meaning. By far most trouble clarifications were declaratives, and often elliptic ones. Mohamed was the only one who used Wh-questions a few times. There was little longitudinal development for all four NNSs. This is a remarkable outcome. Apparently the NNS found the minimal means they used effective enough. If this is the case, negotiation of ineaning does not appear to lead to much NS comprehensible output, even if the NNS is given the opportunity to create it. Interaction type had little influence too. Only translations were found more often in the informal conversations. This may indicate that the NNSs felt somewhat freer in this context to try any means for clarification and were more restrained by face considerations in the institutional conversations. However, this difference may very well be related to the NS Janet's personal style than to the interaction type. 8.3 Nonnative Speaker Trouble Clarifcations 167 8.4 Comparing NS and NNS Trouble Clarifications In this section, NS trouble clarifications will be compared to those of the NNSs. The results will be tied back to the expectations about the types and forms of their clarifications in the beginning of this chapter. In general, the analyses showed that trouble clarifications generally concern adjustments of the speaker's own former utterance. In other words, they usually are (other-initiated) self-repairs (Schegloff, Jefferson 8c Sacks, 1977). It was expected that the NSs would use a greater range of trouble clarification types and forms. The NNSs were expected to show a smaller range, preferring simple adjustments such as paztial repetitions, particulazly, in the eazlier stages of the data collection. They were expected to develop in the direction of the NS. Trouble clarification typeS NS NNS li olo ~ olo Full repetition 42 5.9 12 3.6 Partial repetition 64 9.0 56 17.0 Additive repetition 59 8.3 36 10.9 Elaboration 227 31.8 126 38.3 Substitutive repetition 49 6.9 13 4.0 Complex repetition 124 17.3 51 15.5 Paraphrase 72 10.1 5 1.5 Hypothesis forming 34 4.8 2 0.6 Corcection 12 1.7 20 6.1 Word supply 22 3.1 0 0.0 Translation 7 1.0 6 1.8 Other 2 0.2 2 0.6 Total 714 100.0 329 100.0 Table 8.9: A comparison ofNS and NNS trouble clarification rypes As is shown in Table 8.9, the NSs used twice as many trouble clarifications as the NNSs, with a wide range of trouble clarification types. They used repetitions most frequently (SOqo). Repetitions create redundancy and give a second opportunity to process the input to achieve understanding. Full repetition was used by the NSs in only Solo of the cases. This may indicate that the NS estimated the problem in un- 168 Chapter 8 Clarifying Trouble in Understanding derstanding as not being an incidental lack of attentiveness. They perceived it to be a problem concerning the input itself or the processing of it by the NNS. Complex and substitutive repetition were used relatively often. Adjusted repetitions may also show that the NS diagnosed a particular part to be problematic and thus tried to highlight this part. Elaborations were used about 30qo of the time. They provide more explicit context to support understanding. Both repetition and elaboration aze classified by Long (1983b) as conversational adjustments. However, they represent two different processes. Whereas with repetition portions of the original trouble source are reused, with elaboration new but content-related material is added. Both deletion and addition occurred more frequently than substitution. Linguistic adjustments, such as paraphrase, fully substitute one utterance for another. Paraphrases were used about lOqo of the time by the NSs. Chaudron (1983) found that a conversational adjustment such as simple repetition promoted recall better than a linguistic adjustment such as the use of synonyms in lecturettes for ESL students, which proves the effects of increasing redundancy. Pica (1987) also found that repetition improved NNS comprehension. In addition, Chaudron á Parker (1987) report on research which showed that elaboration was effective in improving NNS comprehension. The NSs appeared to have a sense that conversational adjustments, such as repetition andlor addition, may be more effective than paraphrase. Paraphrase may be less effective in creating comprehension. Kelch (1985) found the effect of linguistic adjustments on comprehension to be inconsistent. Long (1983c) claims that from an educational perspective, NS linguistic adjustments might be counterproductive if linguistic material is replaced by simpler or even ungrammatical equivalents. It was found that this kind of replacement indeed occurred sometimes, although not always. Overly simplified ungrammatical NS adjustments sometimes occurred when grammatical adjustments were not comprehended. Like the NSs, the NNSs used repetitions SOqo of the time but they favored partial repetition. This agrees with findings by Pica (1994: 516) that NNSs segment portions of their initial utterances, which is a form of decomposition (Larsen-Freeman 8t Long, 1991). Strong linguistic adjustments such as paraphrase were rarely used by the NNSs. For the NNSs paraphrase might be linguistically the most demanding strategy. However, the NNSs used one type of strong linguistic adjustment, that is corrections more than the NSs. These were mainly reactions to false hypotheses by the NS of the NNS's intended meaning. In contrast, word supply was only used by the NSs, in particular by Janet. This appeared to be a typical NS strategy as a reaction to a NNS word request. In total, the NNSs used somewhat fewer global level trouble clazifications such as full repetition and paraphrase, than the NSs. This may be caused by their limited communicative competence. It is easier to repeat part of an utterance than the whole. The range of forms of the NSs and NNSs differed considerably. Whereas the NSs used declazatives and questions in a 50-50 distribution, the NNSs opt almost solely for declaratives (see Table 8.10). Partly this difference can be explained by the fact 8.4 Comparing NS and NNS Trouóle Clariftcations 169 NS Trouble clarification NNS ~ olo ik qo Wh-question 79 11.1 10 3.0 Or-choice question 12 1.7 1 0.3 Yeslno question 90 12.6 2 0.6 Negative question 19 2.7 0 0.0 Elliptic question 74 10.4 6 1.8 Declara[ive question 41 5.7 1 0.3 Other questions 15 2.1 0 0.0 Full declarative 213 29.8 126 38.3 Elliptic declarative 161 22.5 171 52.0 Other 10 1.4 12 3.6 Total 714 I OO.U ~~~9 l0U (1 forms ~ Table 8.10.~ A comparison of NS and NNS trouble clarifzcation forms that the NS as part of the research team took initiating roles in the interaction, which leads to question behavior. In particular, the authoritative role of the NS in the institutional conversations led to rephrasing of questions with which new topics were initiated. Yeslno, elliptic and Wh-questions were used for such paraphrase. Such frequent use of questions may also reflec[ a foreigner talk characteristic of more question-answer strings (Larsen-Freeman 8z Long, 1991). Remarkable is that or-choice questions were not much used, whereas Long (1981) reported a significantly higher proportion of those in foreigner talk discourse. The NNSs, on the other hand, almost only used declaratives and a few elliptic questions. Contrary to the NSs, most of their declaratives were also elliptic, indicating that minimal linguistic means were employed to achieve the communicative goal of clarification. Imperatives were not used by any of the NSs or NNSs. 8.5 Conclusions In this chapter, the types and forms of NS and NNS trouble clarifications were described with the same classifications. This made it possible to compare NSs and NNSs in the way they clarify trouble. Individual portraits were also presented and, for the NNSs, some data on longitudinal development. 170 Chapter 8 Clarifying Troub(e in Understanding The NSs provided the NNSs with well-formed, simpler, more regular, more explicit and more redundant language (Bingham Wesche, 1994: 233) through the application of three adjustment processes: deletion, addition and substitution. Deletion leads to easier processing and highlighting of the trouble source. Addition gives extra material to support processing of the trouble source. Substitution may simplify trouble sources if one item is replaced by another easier one. However, substitution did not always lead to simplification. In fact, deletion and addition are used more than substitution by both NSs and NNSs. This may indicate that substitution, which requires considerable linguistic adjustments, is more difficult to produce and process. This agrees with findings that repetition and elaboration (addition) support comprehension more consistently than linguistic adjustments such as substitution of lexical items (Chaudron 8c Parker, 1987). The distinction between NS conversational and linguistic adjustments from Long (1983b) is empirically a difficult one to apply to NS as well as NNS adjustments. In clariiication sequences, all clarifying moves lead to structural adjustments which Long calls conversational. However some moves also modify the linguistic input more or less strongly. The processes of addition and substitution both lead to linguistic adjustments. Long (1983b) classified adjustments which add to a trouble source, such as elaboration, as a conversational adjustment. However, in this study elaboration is seen as a form of addition because it adds new linguistic material to the original trouble source. Thus it is a linguistic as well as a conversational adjustment. Yet, addition may be less demanding than substitution. Substitution requires the replacement of one lexical item by a synonym or a paraphrase. This is more linguistically demanding. Long appeared to only see substitution as a linguistic adjustment. Based on the data of the current study, the division between conversational and linguistic adjustments needs revision. Long's (1983b) dichotomy of linguistic and conversational adjustment does not do justice to the three adjustment processes operating on NS as well as NNS clarifications, which may all lead to different degrees of conversational and linguistic adjustments. Another distinction which was found to be useful is the distinction between trouble clarifications which operate locally on part of an utterance of the trouble source, such as partial repetition, and others which operate globally on a full utterance, such as paraphrase. The NSs used global trouble clazifications somewhat more that the NNSs. This may have been related to level of communicative competence. It is linguistically more demanding to repeat or paraphrase a full utterance than a part. The NS clarifications were as often questions as they were declaratives. Questions occurred more frequently in the institutional conversations, in particular. The original trouble source in this context may have been a question too, which may be due to the interview format. There is considerable variation among the NSs. Janet used more ]anguageteaching devices, such as word supply and correction. Paul relatively frequently used simple full and partial repetition, which may show his discomfort in adjusting his language behavior in a more resourceful way. Bart used the strongest adjust- 8.5 Conclusions 171 ments, which were even ungrammatical sometimes. These adjustments appear to have made understanding easier. NS ungrammatical adjustments may be influenced by several factors, such as his experience in dealing with NNSs, his expert status, the institutional context, and sequential placement. The NNSs used repetitions as frequently as the NSs, but preferred partial, complex and additive ones. Also for the NNSs these repetitions almost exclusively turned out to be self-repetitions. Repetition is linguistically the least demanding method of clarification because portions of the initial utterance can be used instead of producing new material (Pica, 1994: 516). It is a weak form of adjustment. In addition, the NNS used elaborations more frequently than the NSs did. This confirms the expectation that the NNSs would use the linguïstically simpler means of clarification. Addition apparently was used with as much ease as repetitions. Trouble clarifications that required stronger linguistic adjustment, such as those based on substitution, were hardly used at all, except for corrections, which were used more frequently by the NNSs than by the NSs. Translations were used rarely by the NNSs. It is remarkable that there is little to no development in NNS clarification types and forms. With the NNS trouble indicators, there was some development. Apparently trouble clarification can be effectively done with minimal means. However, such minimal NNS clarifications may not create `comprehensible output' which is more appropriate and grammatical in terms of Swain (1985). In addition, the NNSs used almost only declaratives, in particular, elliptic sentence fragments, to clarify trouble. The NNS's comprehensible output in oral interaction does not consist of grammatical, full, finite sentences. As the NNSs also rarely used linguistic adjustments, NNS adjustments do not appear to result in much `comprehensible' output. With the NNSs there was also some individual variation. Although most NNSs used elaboration frequently, Mohamed did not use it much. Instead he most frequently used partial repetition. He was also the only NNS who used Wh-questions as well as declaratives. Corrections were used somewhat more frequently by Mohamed and Mahmut. These corrections concerned false NS hypotheses of their own intended meaning. 172 Chapter 8 Clarifi~ing Trouble in Understanding Chapter 9 Requesting Confirmation 9.0 Introduction In the preceding two chapters, a description was given of NS and NNS trouble indicator and clarification types and forms. In this chapter, the NS and NNS checking behavior will be described. Confirmation checks were defined as expressions "immediately following an utterance by the interlocutor which are designed to elicit confirmation that the utterance has been correctly heard or understood by the speaker" (Long, 1983a: 137). The following questions will be answered for the part which concerns confirmation checks: 4a. How can NS and NNS confirmatíon checks be described in terms of their types and forms? 4b. Wha[ is the influence of the NNS communicative competence and interaction type on the variety and quantity of checkíng behavior of the NS and NNS? In this chapter, first the classification presented for trouble clarification types in Chapter 8 will briefly be discussed for confirmation check types (section 9.1.1). In addition, some remarks are made on the application of the form classification presented in Chapter 7 to confirmation checks (section 9.1.2). In section 9.1.3 some expectations are formulated on the outcomes concerning the use of NS and NNS confirmation check types and forms in the different conditions. The NS's confirmation checks will be discussed for combinations of types and forms used (section 9.2). The NS types and forms are compared over the two interaction types (section 9.2.1). Then individual portraits of the NSs follow (section 9.2.2). Subsequently, the same aspects will be discussed for the NNSs' confirmation checks, but for this group some specific attention will also be given to their longitudinal development where relevant (section 9.3). Then a comparison of NS and NNS confirmation check types and forms follows (section 9.4). T'he chapter ends with conclusions in section 9.5. 9.0 Introductíon 173 9.1 A Classification of Confirmation Check T~pes and Forms For the description of confirmation check types, the same taxonomy is used which is described for the trouble clarifications in Chapter 8 since both moves manipulate the content of the trouble source. This classification of types will briefly be discussed for confirmation checks to emphasize some specific features for this move. Similarly for the forms, only specific features are discussed which are expected to be characteristic for confirmation checks. 9.1.1 Confirmation Check Types The three main processes of NS and NNS adjustments were: 1. deletion, 2. addition and 3. substitution. These three processes can operate on local or on global level. They can occur in combination with repetition and in combination with each other. They adjust the conversational structure and sometimes also the linguistic input. This led to a classification of 11 categories of trouble clarification and confirmation check types (see Figure 8.1 in Chapter 8). As stated before, the classification is applicable to trouble clarifications as well as confirmation checks. However, confirmation checks usually manipulate the content of the other speaker's utterance, and thus are other-initiated other-repairs (other-corrections) in terms of Schegloff, Jefferson 8z Sacks (1977), whereas trouble clarifications concern the speaker's own original utterance (other-initiated self-repairs). Confirmation checks are used when the listener is not certain he understands the other su~ciently, and wants to double-check. If he has understood enough, he may be able to manipulate the other's content by repeating it in some way or another. If he understood less, he may formulate some hypothesis concerning the intended meaning. Thus confirmation checks can be rephrased as one of two questions: "Did you say X?" or "Did you mean X?" (Holliday, 1988). T'he confirmation checks which reflect the first question may copy the other speaker's utterance with or without minimal linguistic adjustments. They are characterized by repetition, with rising intonation of all or part of the speaker's preceding utterance. For the categories partial and full repetition there is no linguistic adjustment at all. Minimal adjustments may occur when the speaker is reasonably sure he has understood the other. Thus he may use a declarative format but with some linguistic adjustment, such as additive, substitutive and complex repetition. Additive categories may show some confidence. In adding to the other's original utterance, elaborating on it, or supplying a requested word, the speaker must be somewhat confident he understood what the other intended to say. It may be clear that the repetitions which function as checks 174 Chapter 9 Requesting Confirmation are different from keyword repetitions, which were classified as trouble indicators. With keyword repetition the interlocutor echoes the sounds of a word of the other speaker's utterance. Generally it was clear from the pronunciation, intonation and the reaction to this repetition from the other speaker whether the repetition was a contïtmation check or a trouble indicator. Substitution, for example, may occur when the speaker who caused the trouble did not use standard formulations, such as NNSs often do. The NS may model the correct form. Confirmation checks which reflect the second question `Did you mean X?', check intended (implied) meaning or underlying presuppositions. These checks can be classified as several types, such as hypothesis forming, paraphrase, correction, or translation (see also Pica, 1988: 53). Hypothesis forming may be used when there is considerable uncertainty about the other's intentions. Such checks propose a model for the other's intended meaning, manipulating not only the other's content but also the linguistic form. Paraphrase, cotrection and translation may be used when the speaker is better able to guess intended meaning but still creates a different linguistic form for this meaning. In Example 1, Mahmut started to explain that the living room in his current house was very narrow. He used the word dun (thin) instead of the correct form snial (narrow). Bart first checked Mahmut's intended meaning by repeating this word. This is an example of the `Did you say X?' question (Holliday, 1988). When Mahmut confirmed the word, Bart hypothesized that dun (thin)referred to the thickness of the walls instead of to the width of the room. He used hypothesis forming with a one word elliptic question and received an erroneous confitmation. (1) Institutional conversation with Mahmut, Cycle 2 Mahmut: grolbeetje teeveekamer hè [f] ~bi~la bit TV room huh [t] -~big~ [yes] Bart: [ja] hele dunne hè very thin huh ts Mahmut: thin? cc Bart: dunne? yes thin notrnot t coltc Mahmut ja dunne nietlniet t what thin? ti Bart: wat dun? tc Mahmut: zo lange of dunne hè this long or thin huh cc co Bart: Mahmut muren ? ja walls? yes Correction and translation may show greater confidence and be realized as declaratives. The greater the deviation of the original utterance, the more uncertainty there is about the equivalence with the other's intended meaning. 9.1.2 Confirmation Check Forms The form division between declaratives, intetrogatives, and imperatives introduced in Chapter 7, was also used for the confirmation checks ( see section 7.1.2). In particular, questions which manipulate content and which can be reacted to with a confirmation or rejection may be used as confirmation checks, namely: yeslno, elliptic, negative and declarative questions. 9.1 A Classifecation of Confirmation Check Types and Forms 175 9.1.3 Expectations Several expectations can be formulated on the variety of confirmation check types and forms used by the NSs and NNSs in the two interaction types. Confirmation checks compared to trouble clarifications have an additional function. They clarify the problem and at the same time direct the attention of the NNS to a trouble source. Therefore, it is expected that quantitative differences may occur in the distribution of types and forms compared to the results in Chapter 8 for the trouble clarifications. Varonis 8z Gass (1982: 131-132) claimed it may be typical NS behavior to check assumptions based on NNS speech and to request confirmation of this assumption. Such assumptions are in this study coded as hypothesis forming. Hypothesis forming may require more linguistic competence than repetition because strong linguistic adjustment may be needed. Thus it is to be expected that NSs will use hypothesis forming more than NNSs. The NS may use it to support the NNS by correctly rephrasing intended meaning at a level beyond the NNS current productive linguistic competence. The NNS's task is then to confirm or reject the hypothesis. Based on NS usage in NS-NS interaction (Beun, 1990), it is expected that the NSs will use declarative questions for such hypotheses. However, it might be that NS speech differs in interaction with NNSs and that incomplete question types are used more frequently. The NNSs are expected to use repetition more frequently, in particular partial repetition, for which the least linguistic competence is required. Hypothesis forming, which requires strong linguistic adjustment and creative production, may not be used as much. There may also be some sequential order of use of confirmation checks. NNS communicative competence may influence NS and NNS confirmation check types and forms. Due to limited NNS communicative competence it is expected that the NNSs will show less variety and complexity in types of confirmation checks than the NSs, but that this variety will increase when their communicative competence increases. NSs on the other hand, are expected to adapt their types and forms depending on factors such as the context and level of communicative competence of the NNS, thus using a wider range of confirmation check types and forms. Interaction type may also influence NS and NNS confirmation check types and forms. In the institutional conversations, the NSs and NNSs may both use hypothesis forming most frequently, because specific information had to be exchanged. They could not switch topics as easily as in the informal conversations. In the informal conversations, the speakers may be less restricted and thus use a greater variety of ineans. The NS Janet may opt for means which support the NNS in expressing intended content. She may thus use hypothesis forming, word supply, correction and translation more regularly than the NSs in the institutional conversations. If she used declaratives and declarative questions frequently, it would show her confidence in understanding the NNS's intended meaning. 176 Chapter 9 Requesting Confirmation 9.2 Native Speaker Confirmation Checks As was shown in Chapter 5, native speakers used many more confirmation checks than trouble indicators when they had trouble understandíng the NNSs. This accounts for the high total score of almost 500 NS checks to approximately 150 of the NNSs. The NSs used many confirmation checks of several types. The results for the NS confirmation checks are presented in Table 9.1. The confirmation check types are ordered starting with pure conversational adjustments (full and partial repetition), followed by weak linguistic adjustments (additive repetition, elaboration, substitutive and complex repetition), then by strong linguistic adjustments (paraphrase, hypothesis forming, word supply, correction and translation), and ending with the other category. Hypothesis forming occurred 35qo of the time (see Table 9.1). Most of these hypotheses have the form of elliptic or minimal questions, indicating the number of linguistic units were kept low. Yeslno and negative questions also occurred quite regularly with hypothesis forming. These forms, like elliptic, minimal, and declarative questions, only require an affirmative or negative answer (yes or no). Complex repetition was also used regulazly (22010) in the form of declaratives. The use of declaratives may show that the NSs were quite certain of the correctness of their repetition. Alternatively, elliptic, yeslno and declarative questions were used. All other repetition categories also appeared (7- lOqo) and regularly had a declarative form. Full repetitions were used more as confirmation checks than as trouble clarifications. Like substitutive repetition they frequently had the form of declaratives and elliptic questions. Partial and additive repetition frequently also had the form of elliptic questions. In total, however, the NSs used declaratives mainly for repetitions. Word supply was only used in a few cases. Elaboration and paraphrase occurred rarely. Apparently those are typical clarification categories. Translation and correction were as marginal as confirmation checks as they were as trouble clarifications. The forms of the NS confirmation checks were less diverse when compared to the trouble clarifications. Only question types which require a yes or no answer were used. Elliptic questions (fragments) were by far the biggest category. Declara[ive questions and declaratives also occurred quite often. Imperatives did not occur at all, which is logical, because confirmation checks do not demand but suggest something. The frequent occurrence of hypothesis forming alongside repetition, may show that on many occasions the NSs were not certain about the NNSs' intended meaníng. Quite regularly confirmation checks of the hypothesis forming type were negative questions. Such negative questions sometimes, instead of contributing, to the resolution of a problem, created problems because of the ambiguity of the NNS answer following it. This, for example, appeared to be the case with Fatima. In Example 2, Bart used a negative question to check Fatima's negative statement that she did not have a house. Fatima intended to express agreement but used an affirm- 9.2 Native Speaker Confirmation Checks 177 Linguistic form C O y 7 Q' ó ó " y e ~ 7 h ~ y 7 Q' ~ ~ ~ Q ~ 9 7 W ? Confirmation check tYPe ~ ~, y '~, on z -a w ~ ~ ~ é~ ~ ~~ t~ r Ó N R F ~ Full repetition 1 3 15 2 6 ts z a7 9.s Par[ial repetition 1 2 23 1 3 17 3 50 10.1 Additive repetition 6 4 14 3 1[ 38 7.7 2 4 2 8 1.6 17 37 7.5 Elabora[ion Substitutive repetition I 1 l0 6 2 Complex repetition 14 4 39 24 28 109 22.0 Paraphrase 2 1 2 3 3 11 2.2 Hypothesis forming t5 ta 6t 25 2t t7a 35.I 3 3 0.6 t7 3.a t o.z t t o.2 496 38 Correction Word supply 1t 5 t 1 Translation Other Total 40 29 17I 68 49 132 7 oIo 8.1 5.8 34.5 13.7 9.9 26.6 1.4 100.0 Table 9.1: Types and linguistic fotms of NS confitmation checks ative rather than a correct negative response. Therefore, Bart checked again with a positive declarative question and a Wh-question. T'he latter forced Fatima to manipulate content herself. Only then could Bart make out for sure that she indeed did not have a house for herself. (2) Institutional conversation with Fatima, cycle 2 Fatima: ik geen huis ts Bart: cc je hebt geen huis? co Fatima: ja tc Bart: u woont nu? cc coltc cc Fatima: Bart.~ u hebt nu een huis? nee ik geen huis nou u hebt geen huis? tc Bart: waar woont u nu? tr Fatima: met familie ac cc co Bart: bij familie ja op een kamenje? ja co 178 Fatima: Fatima: ja I no house you don't have a house? yes you live now? you have a house now? no I no house now you don't have a house? yes where do you live? with family with family yes in a small room? yes Chapter 9 Requesting Confitmation Minimal questions were also used quite frequently. These are one-word feedback questions which often check whether an affirmative answer to an earlier check was not gratuitous but indicated real agreement. In Example 3, Bart used such a minimal questíon. At that point, Bart was not play-acting but was díscussing steps Mahmut could take in real life to improve his housing situation. Bart used the minimal feedback question, to check whether Mahmut was really going to fill out and send away the registration form before a certain date. This was important so that Mahmut would become eligible for other housing. Mahmut's answer to Bart's earlier check invullen? (fill out?) was ja kan tirel. This literally can be translated as `yes, it could be done' which apparently did not convince Bart that he really would do it. Mahmut at this point may have been ambivalent because he was not sure whether they were play-acting or not. Later in this conversation, Mahmut checked Bart's real identity and was surprised to find out that Bart actually was the housing official his brother-in-law had referred to some time before. Maybe he realized only then that Bart was real[y trying to help him. (3) Institulional conversation with Mahmut, cycle 2 ts Bart: ja dat moet je invullen yes you have to fill that out cc Mahmut deze? this? mr ik niet weten 1 not know cc Bart: niet weten? [ff] not know? [~] ti Mahmut: (tt] [~] cc Bart: co cc Mahmut: Barl: invullen? ja kan wel ~a? fill out? yes okay co Mahmut: ja yes Ves ~ Declarative questions and declaratives occurred quite often. Beun (1990) found that about SOolo of the declarative questions had question intonation. In this study, the NS used question intonation in about 40qo of all cases. The majority had no question intonation. In Example 4, both declarative question forms can be seen. Mohamed wanted to make clear that he was going to marry next month, but he could not distinguish past from future forms. Therefore, the housing official was confused about whether he was already married or not. Surprised, the NS rephrased Mohamed's utterance, in which Mohamed mistakenly used vorige (last) for volgende (next), with a declarative question. Mohamed did not understand this question. The NS stepped back to a lower degree of certainty and rephrased with a yeslno question. When this question (erroneously) received an affirmative answer, the NS rephrased with a declarative question without question intonation which indicates more certainty again. He received another erroneous affirmative answer from Mohamed and realized that they had not achieved understanding only later when the issue came up again. (4) Institutional conversation with Mohamed, Cycle ts Mohamed: `(ja] nou uh ik wil uh trouwt Paul: t zeg dat` ti Mohamed: `trouwt trouwt tc cc Paul: ga je trouwen? coltr Moharned: ja nou niet 9.2 Native Speaker Confirmation Checks t `(yes] now uh 1 want uh marries t say that` Urtames mames are you going to many? yes now not 179 ti ts Paul: Mohamed: cc ti tc mr cc co Paut: Mohamed: Paul: Mohamed: Paul: Mohamed: maar ik wil tt uh wanneer ga je trouwen? tt misschien uh t vorige maand of uh zoiets vnwr~je bent al getrouwd? hm? ben je nu al getrouwd? ja ja trouwt je bent al getrouwd ja but 1 want tf uh when are you going to many? tt maybe uh t last month or uh something IaAalyou're rrwrried already? mhm? are you married yet? yeah yeah rttarries ynu are already married yes Many confirmation checks are followed by a confirmation, even when it is not at all clear whether the NNS has actually understood or is merely showing attention and willingness to continue the conversation (gratuitous concurrence, Liberman, 1984). Therefore, hypothesis forming carries in itself the risk of false agreement and misinterpretation, as in the above example and in Example 1. This may even be experienced as manipulation in the goal-oriented interaction, as may have been the case in the example in which Mohamed is not able to keep the option of a house with a garden (see Example 4, Chapter 6). Complex repetition was also used often (22qo). Apparently when the NSs thought they understood the NNS, they adjusted the NNS's intended meaning in more than one way (deletion andlor addition andlor substitution, or word order changes). These repetitions may be cases of modeling to standard usage. Although many sentence fragments (elliptic utterances) occutred, hardly any ungrammatical confirmation checks were used. This finding contrasts with the NS trouble clarifications. Apparently, confinnation checks have the function of modeling correct fotTns. Quite regularly declarative forms were used which show the NS's certainty that he understood the NNS. In Example 5, Bart asked Mohamed how many children he would like. He segmented Mohamed's answer as `I hope~ two girl and boy'. Bart checked by using an elliptic declarative to repeat part of Mohamed's answer and he added the plural form, a cardinal number and an article to explicate his interpretation of Mohamed's answer (two girls and a boy). However, Mohamed had apparently segmented his utterance differently as `I hope twol girl and boy'. Therefore, he explicated his intention through the use of the ordinal number `one'. (5) Institutional conversation with Mohamed, Cycle ts Bart: kindjes hceveel? ti Mohamed: hè? tc Bart: hceveel? tr Mohamed: t ik hoop twee meisje en uh t jongen cc Bart: twee mei.cje.r en een jangen coltc Mohamed: nee één meisje één jongen ac Bart: een meisje en een jongen t tja je hebt het met voor te kiezen hè 180 3 children how many? huh? how many? t I hope two girl and uh t boy two Rirls and a boy no one girl one boy one girl and one boy t well you can't pick them out can you Chapter 9 Requesting Confirmation In conclusion, the NSs used several types of repetition and hypothesis forming to check intended meaning. The forms reflected the degree of certainty in these checks. Repetitions reflect a considerable degree of certainty and thus most often had a declarative form. Hypothesis forming reflects less certainty and thus most freyuently was realized by the use of elliptic questions and not, as expected, declarative yuestions. Apparently the NSs kept the linguistic means for checking to a minimum. However, there were almost no ungrammatical checks, so these minimal utterances were still standardlike and thus could function as models from which a NNS could learn. 9.2.1 The Influence of Interaction Type In this section the influence of interaction type on the distribution of confirmation check types and forms will be described. It was expected that the NSs in the institutional conversations would use hypothesis forming more frequently than in the informal conversations. The opposite turned out to be the case. NS confirmation check Informal conversations Institutional conversations ~ oIo lk olo Full repetition ]0 5.7 37 11.6 Partial repetition 22 12.5 28 8.8 Additive repetition 8 4.5 30 9.4 Elaboration 0 0.0 8 2.5 Substiwtive repetition 1I 6.3 26 8.1 Complex repetition 28 15.9 81 25.3 Paraphrase 3 1.7 8 2.5 Hypothesis forming 79 44.9 95 29.7 Correction 3 1.7 0 0.0 Word supply 11 6.3 6 1.9 Translation 1 0.6 0 0.0 Other 0 0.0 1 0.3 Total 176 types 100.0 320 ]00.0 Table 9.2: Distribution of NS confirmation check types per interaction type Hypothesis forming was used more in the informal than in the institutional conversations by the NSs (see Table 9.2). This might indicate that the NSs in these con- 9.2 Native Speaker Confirmation Checks 181 versations guessed more, because they had fewer expectations based on the structure and content of the conversation. The NSs in the institutional conversations had a script to support their understanding. Possible intended meanings may have been more restricted and thus repetition occurred more frequently (ó3qo as against 45qc in the informal conversations). Word supply was used more frequently in the informal conversations. This was also the case with word supply which occurred as trouble clarifications. This may be related to Janet's personal style (see section 9.2.2.). Full and additive repetitions were used more in the institutional conversations. This may indicate that the NSs used confirmation checks to achieve a higher degree of certainty about the information received and to obtain complete answers. Similar to the results of the trouble clarifications, there are not many differences in forms compared over the two interaction types (see Table 9.3). This is remarkable because there are different speakers in the two interaction types. In the informal conversations, elliptic questions and full declaratives occurred somewhat more. NS confirmation check forms Informal conversations Institutional conversa[ions ~ `~~ ~ `~r Yeslno-question 11 6.3 29 9.1 Negative question 9 5.1 20 6.3 Elliptic question 70 39.8 101 31.6 Declarative question 10 5.7 58 18.1 Minimal feedback question 18 10.2 31 9.7 Full declarative 18 10.2 16 5.0 Elliptic declarative 35 19.9 63 19.7 Other questions 5 2.8 2 0.6 Total 176 ]00.0 320 100.0 Table 9.3: Distribution of NS confirmation check forms per interaction type Declarative questions occurred much more often in the institutional conversations as did yeslno questions. Declarative questions may reflect the authoritative role of the housing official. The use of these forms may be explained by the housing official's behavior of checking information already given in order to fill out the form accurately. This type of exact information transfer is less important in the informal conversations. Bart used declarative questions also as a way of confirming common ground before requesting more detailed infotmation. T'hese questions often caused confusion, because the NNSs did not always seem to understand that Bart went over earlier supplied information for this reason. In particular, Fatima often became confused when, after she had given the information requested, another request for confirmation came. 182 Chapter 9 Requesting Confirmation In the part of the conversation preceding Example 6, Fatima had told Bart she lived in an apartment on the sixth floor and wanted to move to a house with a garden so that her child could play outside while she could keep an eye on him. She answered the following question `what do you want then?' by repeating that she wanted a house. Bart then went back to the current living conditions (the apartment) to establish common ground, using a declarative question without question intonation. Miscommunication occurred because Fatima probably understood him to be talking about future wishes. (6) Institutional conversation with Fatima, cycle 3 Bart: t en je wil graag ts - ~writes down~ verhuizen je wil graag ergens anders naartce wat wil je dan? tr Fatima: ~ die huis cc Burt: je woont nu op de jlat co Fatima: nee cc Bazt: nu op de Flat co cc co tc ti tc ti cc ti tc mr ti tr Fatima: Bart: Fatima: Bart: Fatima: Bart: Fatima: Bart: Fatima: Bart: Fatima: Bart: Fatima: cc colmr ac Bazt: Fatima: Batt: vind je het niet leuk nee je wil iet.s ruiders ja die ander wat is dat? t wat wil je dan? tt een denb - ~jokes~ t ~of een caravan~ - ~jokes~ die t `[stad] `[wat wil jeJ? ja t misschien uh f naast die centrum tc tc ti cc ti cc cc colmr cc co cc co Fatima: Bart: Fatima: Bart: Fatima: Bart: Fatima: Bart: Fatima: in het centrum ja beter voor mij ja enlenl en wat mcet dat zijn? een huis of uh t iets anders? [f] [t] huis? t of een andere Flat? een lage flat nee t flat nee helemaal geen flat? nee ja maar een huis ja 9.2 Native Speaker Confirmation Checks t and you would like to - ~writes down~ move you would like to go somewhere else what then do you want? tt that house you live in an uparrment nnw no now in an apartment you don't like that no you want something else yes that other what is that? t what do you want? tt a ~tent~ - ~jokes~ t ~or a trailen - ~jokes~ that t `[city] `[what do you want]? yes t maybe uh t neaz that center in the center yes better for me yes andlandl and what should that be? a house or uh t something else? [t] house? t or ano[her apartment? a lower apaztment ( building) no t apar[ment no no apartment at all? no yes but a house yes 183 In conclusion, there were a few differences in NS confirmation check types and forms between the interaction types. The NSs used hypothesis forming most frequently in the informal conversations. In the institutional conversations, more (complex) repetition was used. Form differences were even smallec Only declarative questions were clearly used more in the institutional conversations. These questions may reflect the authoritative role of the housing o~cial. 9.2.2 Individual Portraits of the Native Speakers In this section the NS's confirmation checks will be discussed per individual to show where they differ from the average picture. The distribution of types and forms (over the cycles) will be given in raw scores or percentages in the text where informative. Jmiet For Janet 45qo of her confirmation checks were hypothesis forming, which is lOqo above the average score. This shows Janet put effort into making sense of the NNSs' utterances and tried to model correct forms for their intended meanings. Another type which she used regularly were complex repetitions, which increased somewhat over time. Example 7 shows that complex repetitions may also be a kind of modeling. Ergun's utterance is an incorrect reaction in Dutch to the question when he got his driver's license. Janet modeled the correct form. (7) Informal conversation with Ergun, Cycle 3 ts Ergiin: bijna anderhalfjaar cc Janet anderhuljjuur ~;eleden? co Ergiin: -~sighs~ ac Janet: jaja almost a year-and-a-half u year-and-u-halJugn -~sighs~ yeah yeah Partial repetition decreased sharply after cycle 1. It appears that Janet shifted from this linguistic strategy which is easiest for the NNS to process to more informative and demanding ones. The other types of repetition she did not use much at all. Word supply is a favorite type, which she, like with the trouble clarifications, specifically used in the second cycle (9 times as against 1 in the other cycles). This strategy may be especially supportive, because the NNS noticed the trouble and was motivated enough to request the item from the NS (Pica, 1994). In the second cycle, the NNSs had reached a level at which they still missed many words but were capable of requesting them. They did this, in particular in interaction with Janet, which may show that they felt comfortable doing this. Her `teacherlike' attitude may have played a role here (see also Chapter 8, Example 10). Later there may have been fewer lexical problems and this may have led to fewer requests for words. There was a slight increase in the number of yeslno, negative and elliptic questions Janet used. The latter type was used in 40qo of the cases, which showed that this NS also thought it appropriate to use minimal means. There was a decrease in the use of minimal questions and declaratives. The total number of declaratives 184 Chapter 9 Requesting Confirmation was still high (30010), indicating the relative degree of certainty Janet felt about her understanding of the NNSs' utterances. Paul Paul used hypothesis forming in only 15010 of the confirmation checks. This is a remarkably low score compared to the average NS score of 35qo. His major strategy was complex repetition, adjusting the NNS utterance in more than one way. He also used additive, substitutive, and full repetition each more than lOoJo of the time. With substitutive repetitions he regularly modeled the correct form. In Example 8, Paul replaced the preposition, changing Mohamed's meaning and also the article de with the cotrect neuter fotm het. (S) lnstitutional conversation with Mohamed, Cycle mr Mohamed: bij de centtvm cc Puul: in het centrum co Mohamed: ja 1 near the center in the center yes The fact that Paul used repetition more and hypothesis forming much less than Janet shows that he less tried to guess the NNS's intended meaning but just repeated parts of the NNS's original utterance. Tfiis may indicate a lack of confidence or motivation to try to rephrase the NNS's intended meaning. Another difference with Janet is that Paul used declarative questions most regularly (33qo), followed by elliptic questions (23010), and yeslno questions (18qo). He used these question types for complex repetitions in particular. It was expected that the NS would use declarative questions most frequently for their confirmation checks. The question remains whether Paul's frequent use of them indicates that he did not adjust to NNS speech but spoke as he would have with other NSs. Bart Hypothesis forming was used by Bart in 35010 of the cases, which is average. He also used complex repetition often (21oIo), as in Example 9, in which he repeated a problem word with modified pronunciation while deleting the rest of Fatima's utterance. On the question what kind of work Fatima did in a motel, she answered the following: (9) Institutional conversation with Fatima, Cycle 3 ts Fatima: cc tc cc tc ac Bart Fatima: Bart: Fatima: Bart: voor alles buffet schoonmaak die ~bEsti.b voor alles plastiek? ~bEsti.k~ bestek vork uh lepel uh mes ja ja je staat in de keuken? for everything buffet clean that ~bEsti.b for everything plastic? ~bEsti.k~ silverware fork uh spoon uh knife yeah yeah you stand in the kitchen? Bart's use of hypothesis fotming decreased in the third cycle in favor of complex 9.2 Native Speaker Confirmation Checks 185 repetition, which increased. This may indicate that the NNSs' utterances were closer to native-like output and thus could be repeated to be checked, instead of Bart having to hypothesize about intended meaning. Bart used hypothesis forming more in the second cycle than in the third, which supports the expectation that hypothesis forming decreases as the NNS's linguistic competence increases. In addition, other forms of repetition were used regularly by Bart, such as full, partial and substitutive repetition. Bart used elliptic questions most regularly (35qo) for hypothesis forming (see the above Example 9), but also for several forms of repetition. He also used declaratives in more than 25qo of the cases. These were most frequently used as complex and full repetitions. There was an increase of the use of declaratives in the third cycle compared to the second. Declarative and minimal questions were also used in more than l0010 of the time. In conclusion, the NSs used hypothesis forming frequently, in particular in the informal conversations. They also used all forms of repetition. Contrary to expectations, elliptic questions were used most frequently rather than declarative questions. Declarative questions were used frequently by both housing officials, but mainly by Paul, which may show that he adjusted the least to the NNSs. Declaratives were used frequently too, in particular by Janet and Bart. This may be related to both their greater comfort in communication with NNSs and to a degree of certainty they felt in interpreting the NNSs' utterances. There were considerable individual differences between the confirmation check types used. Janet favored hypothesis forming and word supply, which may show her concern for NNS language development. Paul, on the other hand, favored partial repetition, the least demanding and supportive strategy. 9.3 Nonnative Speaker Confirmation Checks In this section, the NNS confirmation checks will be discussed. The nonnative speakers used confirmation checks much less frequently than the native speakers. The frequencies per type and form are often so marginal that few conclusions can be drawn. This is particularly the case for the comparison over time. Like with the trouble clarifications, there is little to no development in the NNS confirmation checks. Therefore, the longitudinal scores are presented in Appendix 3. The compiled NNS individual scores will be presented in two overview tables, 9.7 and 9.8, in section 9.3.2. In this section, individual development and variation will be discussed. In section 9.3.1, the influence of interaction type will be described. The NNSs used hypothesis forming in about 40qo of the cases, most frequently in the form of either elliptic or minimal questions (see Table 9.4). In addition, partial repetition was used in 25qo of the cases. Those were either elliptic questions or declaratives. Furthermore, complex repetition was used in 15qo of the cases, 186 Chapter 9 Requesting Confirmation Linguistic form Confirmation check type I ~ ~ .. ; o 7 ~ ~ q. 3 ó I ~ c a ~ -. ~ ~ ~ ~ -~, I , ~ ~ i z I r.u Full repetition Partial repeti[ion ~ ~ a ~ L .o ~ E i ~ ~ C~ L p [-~ o s 8 1 tz 7.s 16 39 25.5 3 6 3.9 i i o.7 z 7 a.6 2 2a is.7 z 2 t.3 i 60 39.2 i i o.7 i o.7 3 1 2I Additive repetition 1 3 Elaboration Substitu[ive repetition Complex repetition ~ 5 ~ 2 19 Paraphrase Hypothesis forming i z 30 2a 2 Correction Word supply Translation ~ Other I i Total 2 ] 4 84 27 32 3 QIo 1.3 0.7 2.6 54.9 17.6 20.9 2.0 153 100.0 Table 9.4: Types and linguistic forms of NNS confirmation checks which were almost all elliptic questions. The NNSs may have had trouble repeating complete NS utterances which probably were beyond their productive linguistic competence. Full repetition did, however, occur in some cases, probably only from elliptic NS utterances (see Table 9.6). Other types of repetition occurred much less often. Correction and word supply were not used at all by the NNSs, whereas they did occur as trouble clarifications. Neither were any declarative questions used. This is remarkable because declarative questions were expected to be typically used for checking the other's intentions. This non-occurrence may reflect the asymmetry of the roles or lack of communicative competence. In the first case, the NNSs may not be willing or able to step into the shoes of the NSs. In the second case, the absence of declarative questions may be just one manifestation of a total lack of full sentence types. The latter generally appeared to be the case. The NNSs had a strong preference for elliptic questions which did not require full sentence patterns. Imperative forms were not used at all, which is understandable, because checks do not demand a certain action but just a confirmation fmm the other speaker. 9.3 Nonnative Speaker Confirmution Cheeks 187 9.3.1 The Influence of Interaction Type If we look at the distribution of confirmation check types over the two interaction types, we see that hypothesis forming was used more by the NNSs in the institutional than in the informal conversations (see Table 9.5). This may show that the NNSs had more trouble guessing the housing off'icial's intended meaning. A great number of topics had to be discussed in considerable detail. In particular, in the earlier conversations the NNSs appeared to only catch a keyword with which they could establish the topic (e.g. apartment, work), but they were guessing about the type of information they were asked to give about that topic (when, where, who). For this reason, complex repetition may also have been used more in these conversations than in the institutional conversations. In the informal conversations, the NNSs used more partial repetition, where it was used as often as hypothesis forming. This would be the linguistically easiest type of checking. In addition, the NNSs used more full repetition which may indicate that in some cases, Janet's input was reproducible for the NNSs. Furthermore, the NNSs used more additive repetition which requires expansion of the NS's content. NNS confirmation check types Informal conversations Ins[itu[ional conversations ~ 01o Ik olo Full repetition 10 13.7 2 2.5 Partial repetition 23 31.5 16 20.0 Additive repetition 5 6.8 1 1.3 Elaboration I 1.4 0 0.0 Substitutive repetition 2 2.7 5 6.3 Complex repetition 6 8.2 18 22.5 Paraphrase 0 0.0 2 2.5 Hypothesis forming 24 32.9 36 45.0 Correction 0 0.0 0 0.0 Word supply 0 0.0 0 0.0 Transla[ion 1 1.4 0 0.0 Other 1 1.4 0 0.0 Total 73 100.0 80 100.0 Table 9.5: Distribntion ofNNS confirmation check rypes per interaction type 188 Chapter 9 Reguesting Confirmatiai As far as the distribution of the forms is concerned, elliptic questions appear considerably more often in the institutional conversations, whereas (elliptic) declaratives are used more in the informal conversations (see Table 9.6). Again, the use of questions may be caused by the interview format of the institutional conversations. The NNSs frequently checked the housing official's questions with counterquestions, as can be seen in Example 10. (10) Institutional conversation with Mohamed, Cycle 2 ts Bart: ja en hebben jullie altijd yes and have you always in oisterwijk gewoond? lived in oisterwijk? cc Mohamed: toen ik hier? wlten 1 here? co Bart: ja yes mr Mohamed: ja yes ac Bart: ja yes NNS confirmation check Informal conversations Institutional conversations ~ ~e ~ ~lo Wh-question 0 0.0 2 2.5 Yeslno question 1 1.4 0 0.0 Negative question 1 1.4 3 3.8 forms Elliptic question 33 45.2 51 63.8 Minimal feedback question 12 16.4 IS 18.6 Full declarative 3 4.1 0 0.0 Elliptic declarative 20 27.4 9 11.3 Other questions 3 4.1 0 0.0 Total 73 100.0 80 100.0 Table 9.6: Distribution of NNS confirntation check forms per interaction type Only in the institutional conversations were Wh-questions used (see Ergun, section 9.3.2). Minimal questions occurred in both settings. The other category which only occurred in the informal conversations concerned mainly checks in another language which could not be categorized. In conclusion, there are some differences in distribution of NNS confirmation check types over the two interaction types. Partial and full repetition were used more frequently in the informal conversations, whereas hypothesis forming and complex repetition were used more in the institutional conversations. There were a few form differences. Elliptic questions were used more in the institutional conver- 9..i Nonnative Speaker Confirmation Cheeks 189 sations, probably caused by the interview format, whereas elliptic declaratives were used more in the informal comersations. 9.3.2. Individual Portraits of the Nonnative Speakers In this section, overview tables of confirmation check types (Table 9.7) and forms (Table 9.8) of the individual NNSs are presented. Tables with longitudinal scores per NNS can be found in Appendix 3. The results will be discussed per NNS informant and longitudinal information will be given when informative. Fatima NNS confirmation check types Mohamed Mahmut Ergun ~ ~o ~ ~~ ~ ~~ ~ ~o Full repetition 2 7.4 2 6.9 4 8.9 4 7.7 Partial repetition 4 14.8 9 31.0 12 26.7 14 26 9 Additive repetition 0 0.0 2 6.5 1 2.2 3 S R Elaboration 0 0.0 0 0.0 1 2.2 0 Q(i Substitutive repetition I 3.7 1 3.4 3 6.7 2 3.h Complex repetition 5 18.5 5 17.2 3 6.7 11 21.2 Paraphrase 1 3.7 1 3.4 0 0.0 0 0.0 Hypothesis forming 14 51.9 9 31.0 20 44.4 17 32.7 Correction 0 0.0 0 0.0 0 0.0 0 0.0 Translation 0 0.0 0 0.0 0 0.0 1 1.9 Other 0 0.0 0 0.0 1 2.2 0 0.0 Total 27 100.0 29 100.0 45 100 52 100.0 Table 9.7: NNS confirmation check types per informant Fatima Fatima used few confirmation checks. In half of the cases, she used hypothesis forming, most frequently in the form of elliptic questions (see Tables 9.7 and 9.8). In the third cycle in particular, the checks almost all had the form of elliptic questions, which require a minimal linguistic competence. In Example 11, Fatima misunderstood Bart's closing question whether she had any further questions. She thought he asked whether she wanted to continue talking. (11) Institutional conversation with Fatima, Cycle 3 ts 190 Bart: wil je mij nog iets vragen? do you want to ask me anything else? Chapter 9 Requesting Confirmation cc Futima: t t ~ ~ nog uh niet klaar? - daughs~ tt o not uh finished yet? - daughs~ ti Bart: wat? cc Fatima: nog uh praat? co Bart: ~a what? sti[l uh ralk? yes 1 don't know whether you wantto ask me anything else ik weet niet of jij rnij nog iets wilt vragen tc She furthermore used partial, complex and full repetition. Linguistic adjustments such as substitutive repetition were rarely used. Fatima NNS confirmation Mohamed Mahmut Ergiin ~ qo ~ olo tl olo fk qo Wh-question 0 0.0 0 0.0 0 0.0 2 3.8 Or-choice question 0 0.0 0 0.0 0 0.0 0 0.0 Yeslno question 0 0.0 0 0.0 l 2.2 0 0.0 Negative question 1 3.7 1 3.4 1 2.2 1 I.9 Ellip[ic question 16 59.3 22 75.9 16 35.6 30 57.7 Minimal feedback q. 5 18.5 2 6.9 12 26.7 8 15 -l Declarative question 0 0.0 0 0.0 0 0.0 0 OU Full declarative 0 0.0 I 3.4 1 2.2 1 4.7 Elliptic dec(arative 4 14.8 3 10.3 14 31.1 8 15.4 Other 1 3.7 0 0.0 0 0.0 2 3.8 Total 27 100.0 29 100.0 45 I00.0 52 100 (i ~ -~ check forms Table 9.8: Forms of NNS confirntation checks per informant Mohamed Mohamed used confirmation checks mainly in the second cycle. Hypothesis forming and partial repetition were each used in one third of the cases. Of all his checks 75qo had the form of an elliptic question, which requires minimal linguistic competence and effort. In the third informal conversation, Peter substituted for Janet because the relationship between her and Mohamed had deteriorated to such a point that normal conversation was not possible. In Example 12, Peter tried to find out whether everyone in Morocco had an address with a street name and number. He referred to the situation in Turkey where that is not the case in some small villages. Mohamed reacted with partial repetition in elliptic question form. (12) Intormal conversation with Mohamed, Cycle 3 ts Peter iedereen kent mekaaz everyone knows each other 9.3 Nonnative Speaker Conf rmation Checks 191 altijd t dus niemand t dus zeg een herder in het atlasgebergte die heeft ook straatnaam uh voor zijn hutje hè cc co mr Mohumed: Peter: Mohamed: atlasgebergte? ja maar daarldaar die ken ik niet zo gced ben ik daar nooit geweest always t so nobody t so [ake a shepherd in the atlas mountains he has a street name too uh for his hut huh atlas mountains yes but therelthere those I don't know well have 1 never been there He used complex repetition a few times. The rest of the categories only occutred marginally (e.g., as full and additive repetition). Declaratives and minimal questions were used in a few cases. Mahmut Mahmut also used confirmation checks mainly in the second cycle. Of his checks 45qo were cases of hypothesis forming and about 25qo of the cases were partial repetitions. He used elliptic questions and declaratives each in about one third of the cases. The frequent use of declaratives may be related to Mahmut's relaxed conversational attitude and may express confidence in the correctness of his checks. In addition, he used minimal questions in a quarter of the cases. These questions often checked the referent of the NS's question. In the part of the conversation preceding Example l3, Mahmut told Bart he had trouble with the neighborhood children calling him `crazy' or `bad Turk'. He referred to one of them with the word kindje which is literally a diminutive form of kind (child) but which translates into baby or toddler. Bart confirmed that the children's behavior is not acceptable and then requested the age of the child involved. Bart was confused because Mahmut used the Dutch word for baby for an older child. (13) Institutional conver.cation wilh Mahmut, Cycle 2 ts Bart: hè nee het is niet gced huh no it is not good how old is he? hce oud is die? huh? ti Mahmut: hè? tc Bart: hce oud is die? how old is he? cc Mahmut: ik? me? coltc Bart: nee het `[kindje] no the `[litde child] mr Mahmut: `[oo] kindje of tien twaalf `[oh] little child or ten twelve Furthermore, Mahmut used all other types of repetition, but very infrequently. Ergiin Ergun most frequently used hypothesis forming for confirmation checks (33qo). However, he also used partial and complex repetition quite regularly ( 27oIo and 21 qo). As far as the forms are concerned, he used elliptic questions by far the most frequendy (58qo). He was also the only one who twice used Wh-questions. One was a partial repetition of a NS trouble source. Only the Wh-question in Example ] 92 Chapter 9 Requesting Confirmation 14 was an original rephrase created by Ergun himself, which was effective though nowhere near standardlike. However, neither was Bart's second phrase of the trouble source turn. (14) Institutional conversation with Ergun, Cycle 3 ts Bart: en wat heb je (xx) and what do you have (xx) Iwat ziek zijn? Iwhat be ill? cc co nv Ergiin: Bart: Ergun: wat is dat? what's that? wuarvoar die ziek van mijn? ja ik heb operatie gehad nou (x) knie what fiir rhat ill q~ rnine? yes 1 have had surgery now (x) knee Ergiin also used minimal questions and elliptic declaratives each in 15oIo of the caties. In conclusion, the NNSs most frequently used hypothesis forming, followed by partial and complex repetition. They used elliptic questions for more than SOqo of their confirmation checks. Furthermore, they mainly used declaratives and minimal questions. In the informal conversations, they used full and partial repetition most frequently, and hypothesis forming and complex repetition in the institutional conversations. The latter two types require more linguistic adjustment and may thus stimulate the creation of comprehensible output. However, also in the institutional conversations, mainly minimal forms were used. The NNSs did not differ greatly in their use of confirmation check types and forms. Fatima used hypothesis forming more frequently [han the three male NNSs. Mohamed mainly used partial repetition. Ergun was the only one who used a variety of confirmation checks types. Three types he used quite frequently ( hypothesis fortning, partial and complex repetition). Mahmut used elliptic declaratives more frequently than the other NNSs, which may indicate greater confidence in the correctness of his checks. 9.4 Comparing NS and NNS Confirmation Checks In this section, the distribution of types and forms of the NS confirmation checks will be compared to those of the NNSs. The results will be connected to the expectations about confirmation check types and forms in section 9.1.3. A problem with the comparison is that the NNSs used relatively few confirmation checks (153 as against 496 for the NSs). In general, the analyses showed that confirmation checks generally manipulate the content of the other speaker's trouble source. In other words, they usually are (other-initiated) other-repairs. It was expected that hypothesis forming would be typical NS checking behavior (Varonis 8c Gass, 1982). The NNSs were expected to use hypothesis forming less, 9.4 Comparing NS and NNS Confirmation Cltecks 193 because it requires manipulation of the NS's content and linguistic adjustment of form. They are in a sense a substitution of one speaker's words with words of the other speaker. It was found that hypothesis forming is the type of confirmation check most frequently used by both NSs and NNSs (35qo and 39qo, see Table 9.9). It was expected that the NNSs would use repetition most frequently, in particular, partial repetition, for which little communicative competence is required. The NNSs used partial repetition frequently (2óqo), but less than hypothesis forming. Such partial repetitions also adjust the structure of the interaction. They delete part of the trouble source and repeat the other part. They also used full repetition in 8qo of the cases, which are also pure conversational adjustments. These may have been repetitions of short NS utterances. Such repetitions may create extra time for the planning of an appropriate response. In 15010 of the cases, complex repetition was also used, in which some linguistic adjustment may occur. However, these adjustments generally appear to be minimal. Apparently, in some cases the NNSs judged it necessary to use more than one adjustment strategy to check the NS's intended meaning. The other categories were used irregularly or not at all. The NNSs were thus limited in the types of confirmation checks used. This limitation may originate from a lack of communicative competence, from status influences which made the NNS hesitant to speak for the NS, or it might also partly be a coincidental finding ínfluenced by the limited number of NNS occurrences. The NSs used all types of repetition. They used complex repetition most frequently (22qo), so they apparently judged it necessary to adjust the NNS's input in more than one way. However, they also used partial, additive, substitutive repetition and full repetition each in 5 to l00~0 of the cases. In addition, they used correction and word supply which were not used at all by the NNSs. These are strong linguistic adjustments. Other adjustments typical for trouble clarifications appeared only rarely as confirmation checks. These aze elaboration and paraphrase. Paraphrase can be seen as the alternative of hypothesis forming in manipulating the content of one's own utterance instead of the other's. Thus it is logical that it is rarely used. Extensive elaboration on the interlocutor's trouble source may be difficult to achieve. It would go beyond best guesswork. Altogether, the NSs, as expected, showed more variety in types of confirma[ion checks, using both conversational and linguistic adjustments. The NSs probably adjusted their checks depending on their estimate of the problem, the NNSs' communicative competence and the context of the interaction. They may have first used simple repetition to check what has been said and if that did not solve the problem they may have rephrased with complex repetition and hypothesis forming (see Example 1). It can be concluded that for the confirmation checks linguistic adjustments were used frequently. Long (1983a) developed the distinction for linguistic and conversational adjustments to describe NS adjustments to input, that is, trouble clarifications. The reseazch is consistent on the more positive effect of conversational rather than linguistic adjustments (Pica et al.. 1987; Chaudron 8t Parker, 1987; Larsen-Freeman 8c Long, 1991). However, linguistic adjustments in trouble clarifications may differ from those of confirmation checks because trouble clarifications manipulate the NS's own trouble source. If processing time is needed, one could 194 Chapter 9 Reguesting Confirmation easily see that such new linguistic input could frustrate ra[her than support comprehension. However, NS linguistic adjustments used in confirmation checks concern the NNS's utterance. These may be much easier to understand because they rephrase the NNS's own intended meaning. Thus, the effect of linguistic adjustments on comprehension in confirmation checks may be different from the eftèct on trouble clarifications. Compared to those of the trouble clarifications, confirmation check forms were quite restricted for the NSs as well as the NNSs (see Table 9.10). The NSs used quite a few forms which are limited in the number of linguistic elements, and which require a yes or no answer (about 60qo are elliptic, declarative and minimal questions). Or-choice and Wh-questions were not used by the NSs because they request more open answers. NS Confirmation check NNS lk ~Io ~l alo Full repetition 47 9.5 12 7.8 Partial repetition 50 10.1 39 25.5 Additive repetition 38 7.7 6 3.9 Elaboration 8 1.6 1 0.7 Substitutive repetition 37 7.5 7 4.6 Complex repetition 109 22.0 24 15.7 11 2.2 2 1.3 35.1 60 39.2 types Paraphrase Hypothesis forming 174 Correction 3 0.6 0 0.0 Word supply 17 3.4 0 0.0 Translation 1 0.2 1 0.7 Other I 0.2 1 0.7 Total 496 ] 00.0 153 100.0 Table 9.9: A comparison of NS and NNS conJïnnation check types The NNSs used elliptic questions in more than half of the cases. These require minimal linguistic competence. The NSs also used these questions in 30010 of the cases. Apparently the elliptic question type is most characteristic for a confirmation check and not, as expected, the declarative question. Declarative questions were not used at all by the NNSs, and only in about 14qo of the cases by the NSs. It might be more difficult for NNSs to produce and comprehend this type because of the full linguistic form required. In addition, the NNSs might have preferred to 9.4 Comparing NS and NNS Confirnuition Checks 195 NS Confirmation check forms NNS ~ `~r ~ qo 0 0.0 2 1.3 Yes~no question 40 8.1 1 0.7 Negative question 29 5.8 4 2.6 Elliptic question 171 34.5 84 54.9 Declarative question 68 13.7 0 0.0 Minimal feedback question 49 9.9 27 17.6 Full declarative 34 6.8 3 2.0 Elliptic declarative 98 19.8 29 19.0 Other 7 1.4 3 2.0 Total 496 Wh-question ]00.0 153 100.0 Table 9.10: A comparison ojNS and NNS confirmation check forms use other types instead of a form which takes the perspective of the other because of their assumed lower status in the conversations. Both NSs and NNSs apparently prefer sentence fragments or 'satellite units' (Bygate, 1988). The use of such satellite units is typical for speech but may be even more prevalent because of the low proficiency level of the NNSs all through the data collection period. The rest of the categories the NNSs used were either minimal questions (18qo) or elliptic declaratives (19010), which both also require only minimal linguistic competence. The latter category was used as often by the NSs. The NSs used slightly more negative questions than the NNSs. Those NS confirmation checks which were negative questions may have caused new problems because of the ambiguity of the yes~no answers received. The NNSs sometimes answered positively when a negative answer would have been correct, and thus created confusion. The NSs also used more yes~no questions and some full declarative forms. The yeslno and declarative questions were present in the institutional conversations, in particular, and may thus reflect the authoritative role of the housing official. This context and the interview format may have led to more formal language, that is complete sentences, and to the role division in which the housing official stepped into the NNS's shoes. If we look at the opportunities for language acquisition provided by confirmation checks, Pica et al. (1989: 77-78) conclude that confirmation checks restrict the NNS output following it. Thus the NS would be better off using trouble indicators to stimulate the NNS to produce comprehensible output. This is not what they generally did. They most frequently used confirmation checks, to which the NNSs replied with yeslno answers, sometimes combined with partial repetition. One may 196 Chapter 9 Requesting Confirmation wonder why this is the case. The NNSs did not use confirmation checks very frequently and if they did use them they were very restricted in type and form. Thus they do not seem to provide much opportunity for producing comprehensible output. Pica et al. (1989) state that we cannot conclude from these kinds of outcomes that confirmation checks are not supportive of language acquisition. They do present NNSs with models for their own intended meanings, that is with comprehensible input. However, the question is whether these models are noticed, comprehended, reproduced and more permanently stored in memory. It is not clear whether the NNSs would focus on the form of proposed NS models. Even if they were to notice the form or even request a word, acquisition does not automatically occur. Hatch (1983b: 179) describes an anecdote in which she requested the word for file cabinet from a NS of Arabic as she was readying herself to leave. By the time she left the room she realized she had already forgotten it again. Particulary, with adult learners, linguistic forms may slip out as quickly as they came in. Alternatively, sometimes the models confirmation checks provide may lead to more permanent acquisition of new lexicon and structure, particularly when they are repeated more than once by both speakers. 9.5 Conclusions In this chapter, the types and forms of the NS and NNS confirmation checks were discussed. Confirmation checks were used much more frequently by the NSs than by the NNSs. In contrast to trouble clarifications, confirmation checks manipulate the content and form of the other's utterance and thus contain some degree of uncertainty. Therefore, many hypothesis forming types were used by the NSs. They were realized with different, often minimal, forms such as elliptic, minimal feedback questions but also by declarative or yes~no questions and declaratives. Contrary to expectations the NNSs used hypothesis forming most frequently as well but they only used minimal forms to hypothesize. Hypothesis forming is a strong linguistic adjustment because they proposed another model for the interlocutor's intended meaning. However, all kinds of repetition also occurred frequently for the NSs as well as the NNSs, when the speaker was more certain about the other's intended meaning and only minimal rephrasing was necessary. These result in conversational adjustments. Substitutive and complex repetition and elaboration are both conversational and weak linguistic adjustments. Strong linguistic adjustments such as paraphrase, translation and correction were used quite rarely. These types are apparently typical for trouble clarifications only, whereas repetitions and word supply were used for both clarifying and checking. Thus, even though the same taxonomy can be used for analyzing trouble clarifications and confirmation checks, there are clearly different preferences for both moves. Confirmation checks are more frequendy than trouble clarifications linguistic adjustments. However, NS 9.5 Conclusions 197 linguistic adjustments used in confirmation checks concern the NNS's utterance. These may be much easier to understand for the NNS than linguistic adjustments in trouble clarifications because they rephrase the NNS's own intended meaning rather than the NS's meaning. Thus, the effect of linguistic adjustments on comprehension in confirmation checks may differ from the effect on trouble clarifications. There were few differences between the distribution of NS confirmation check types over the two interaction types. More hypothesis forming occurred in the informal conversations, showing the NSs had fewer clues for interpreting the NNS's utterances. In the institutional conversations, more complex repetition was used by the NSs. There were individual differences between the NSs in the distribution of confirmation check types. Janet preferred hypothesis forming and word supply. Thus she appeared to support the NNSs in finding the correct Dutch wording for their intended meaning. Paul, on the other hand, used partial repetition most frequently, which may reflect his trouble in making sense of the NNS's utterance. He stayed on the safe side by repeating wha[ he did understand. Declarative questions were used frequently by Paul which may show he used a form for checking which is more regularly used in NS-NS interaction, and thus he may have adjusted the least to the NNSs. Bart used hypothesis forming more in the second cycle than in the third which supports the expectation that hypothesis forming decreases as the NNS's linguistic competence increases because the NNS becomes easier to understand. The NNSs, in the informal conversations, used full and partial repetition most frequently, whereas they used hypothesis forming and complex repetition more frequently in the institutional conversations. The NNSs did not differ much in their distribution of coniirmation check types. Nor did they differ much in their forms. In the institutional conversations, they used more elliptic questions and in the informal conversations more elliptic declaratives. This may show greater NNS confidence in the latter type of conversation or it may simply reflect the declarative character of normal conversations and the more interrogative character of institutional conversations due to goals and roles. As far as the forms of confirmation checks are concerned, as expected, generally questions which require a confirmation or rejection were used most frequently. Typically no full sentence types are used for checking. Elliptíc questions and elliptic declaratives were used quite frequently. It was expected that declarative questions would be the most typical question type used for checking. This was, however, not the case. Probably, full sentences were still too demanding for the NNSs to produce and to comprehend. The NSs only used declarative questions in some cases and then in the institutional conversations in particular. Apparently this type of question is role and context dependent. Por the rest, the NSs used mainly elliptic questions. Declaratives were used quite frequently too, in particular, by Janet and Bart. This may show their relative confidence in interpreting the NNS's intentions. The NNSs used elliptic questions in more than SOqo of their checks and some elliptic declaratives and minimal questions. No development could be shown over the points of ineasurement. It may be that such development did not appear because of the relatively low NNS frequencies. Alternatively, requesting confirmation can be l98 Chapter 9 Reguesting Confirrnution done with minimal means, [hus there may be no incentive to create comprehensible output in fuller linguistic forms. By coding NS and NNS with the same coding system, quantitative and qualitative conclusions can be drawn on the asymmetry in behavior. Quantitative differences were considerable but qualitative differences were also found. Only the NS used word supply and declarative questions. Altogether, the NSs as expected, showed more variety in confirmation checks types. They used adjustments which were sometimes purely conversational, but more often, weakly or strongly linguistic. The NSs probably adjusted their checks depending on their estimate of the problem, the NNS's communicative competence and the context of the interaction. The NNS confirmation checks were limited in types and forms. This may originate from a lack of communicative competence or from status intluences which made the NNS hesitant to speak for the NS. 9.5 Conclusions 199 Chapter 10 Conclusion 10.0 Introduction In this final chapter, first a summary and discussion of the results will be given (section 10.1). This summary will be followed by a general discussion of the theoretical implications of the findings for intercultural communication and second language acquisition theory (section 10.2). In section 10.3 some implications of the findings for the practice of intercultural communication and language teaching will be discussed. The chapter ends with a discussion of the scope and limitations of the study and presents some suggestions for further research (section 10.4). 10.1 Summary and Discussion of the Results The current study aimed to investigate what intercultural communicative competence entails for a native as well as a nonnative speaker in dealing with trouble in communication, that is, in negotiating meaning. One goal was to investigate the influence of NS and NNS communicative competence on NNS and NS negotiation of ineaning behavior, and on the amount of the negotiation of ineaning. Furthermore, the study aimed at investigating the influence of different types of interaction on NS and NNS negotiation of ineaning behavior and on the amount of the negotiation of ineaning. In order to investigate these issues, a longitudinal multiple case study was carried out to study clarification sequences in which meaning is negotiated in conversations between NSs and NNSs of Dutch. The study examined (i) the organization of clarification sequences, (ii) the amount of the conversations spent on negotiation, (iii) the distribution of main negotiation moves, trouble indicators, trouble clarifications and confirmation checks over NS and NNS, (iv) strategy types and forms of the main moves used by NSs and NNSs. The interaction data used for this study were collected in a European Science Foundation (ESF) project on spontaneous second language acquisition by adult immigrants (Perdue, 1993a1b; Feldweg, 1991). Two Moroccan NNSs (Fatima and I0.0 Introduction 201 Mohamed) and two Turkish NNSs (Mahmut and Ergun) participated in the study. They had little or no formal instruction in Dutch. Two types of conversations were recorded about a month apart from each other. The first type consisted of informal conversations with a project researcher (Janet), which generally concerned the topic of cultural differences between the Netherlands and Turkey or Morocco. The second type were institutional conversations. These were housing office registration interviews with a housing official (Bart) or a social worker (Paul). In these conversations, the NS and NNS participants followed a script and held the institutional roles of institutional representative and client respectively. All conversations were recorded in a university studio about one to one-and-a-half years after the NNSs' arrival in the Netherlands, and the conversations were twice repeated after a nine-month period. In this section, the main findings will be discussed per research question and related to former research to draw some conclusions. The first part of the first research question was: la. Which moves do clarification sequences in NS-NNS interaction consist of? Twelve different kinds of negotiation moves were described which are part of NSNNS clarification sequences (see Chapter 4). This is a much larger number than the four moves in a model for negotiation of ineaning described by Varonis 8r. Gass (1985a) for NNS-NNS interaction. A clarification sequence consists of main line moves concerning the topic in progress at the moment trouble arises, and side line moves which are inserted to clarify this trouble. Such a sequence generally starts with a trouble source. This is a main line move. Other main line moves are main line and thematic responses, which generally end the clarification sequence and take up the main line of the conversation again. In addition, escapes were considered to be main line moves because they break off the negotiation and bring the conversation back to the main line, by switching to a different topic. The side line in which the negotiation takes place consists of several moves. Two side line moves manipulate the content of the trouble source: trouble clarification and confirmation check. Trouble clarifications manipulate the content and form of the speaker's own trouble source in order to clarify it. Confirmation checks manipulate the content and form of the other speaker's by proposing a model for intended meaning and thus implicitly indicate trouble as well. Another group of side line moves can be realized through feedback signals. The most important kind of feedback move is the trouble indicator. Trouble indicators are feedback elicitors because they request clarification and thus play an important role in initiating negotiation of ineaning (Allwood, 1993). Other feedback moves are confirmations, gratuitous concurrence, and acknowledgements which function as feedback givers (mainly in the form of jalyes). Confirmations indicate agreement with the former speaker. Gratuitous concurrence moves just indicate attention is being paid and with such moves continuation of the other speaker is explicitly requested. Acknowledgements indicate to the interlocutor that a problem is sufficiently resolved 202 Chapter 10 Conclusion and that a move back to the main line of the conversation can be made. The negotiation moves are organized into adjacency pairs: trouble indicator-trouble clarification, confirmation check-confirmation, and trouble source with postponed second part pair-main line or thematic reaction often bracketing the inserted side line moves. Gratuitous concurrence often forms an adjacency pair with trouble clarifications, when one clarification after the trouble indicator is not enough to close the negotiation. Acknowledgements are third parts often closing the side line. Finally, there are two side line moves which explicitly address the problematic character of the interaction. These are comprehension checks, which check the other's understanding, and meta comments, which evaluate conversational progress. Adults have, in contrast to children, metalinguistic and communicative awareness and thus they have ability to comment on the progress of the conversation (Slobin, 1993: 249). It is not entirely clear how generalizable the moves found are to other types of data, in particular NS-NS interaction. However, some of the categories are similar to categories described in other studies on NS-NNS interaction in Swedish and English with different source language speakers (see Kalin, 1995; Shehadeh, 1991; Long, 1983; Pica 8z Doughty, 1988) and in studies on NS-NS interaction (Schegloff, Jefferson 8c Sacks, 1977). The main negotiation move categories, that is trouble indicators, trouble clarifications, and confirmation checks, appear to be generalizable over interaction with different source and target languages. However, many of the NS-NNS studies did not focus on negotiation behavior in their interactional context of a side-sequence but concentrated on one or two of the main moves. Neither did these studies systematically distinguish content moves from feedback moves. This is an omission because feedback signals play an important role in interaction with speakers with limited proficiency, allowing them to participate with a minimum of linguistic means. The second part of the first research question concerned the organization of clarification sequences in NS-NNS interaction. lb. How are negotiation moves organized in clarification sequences? As a result of differences in NS and NNS communicative competence, it was expected that clarification sequences would be organized differently depending on whether the NS or the NNS caused the trouble. Therefore, clarification sequences were divided into two types, the ones starting with a NS trouble source and the ones with a NNS trouble source (See Figure 10.1). The NS was expected to alleviate the NNS of the task of clarification, which requires manipulation of content and form. The analyses show that NS sequences occur more frequently than NNS sequences (Chapter 4). Thus more NS than NNS trouble sources were indicated and negotiated. Not only do the number of NS and NNS sequences differ, but so does the organization of these seyuences. The NS sequences most frequently contain a NNS trouble indicator which is a feedback elicitor as a second turn, generally followed by a NS trouble clarification, which manipulates the content and form of the NS 10.1 Summary and Discussion of the Results 203 irouble source confirmation trouble source ~~ confirrnation check ,,,~,,,:.: trouble clarification acknowledgement `~ ~~ trouble indicator mainline reaction Figure 10.1: Model of a NNS sequence (left) and mode[ of a NS sequence (right) trouble source. The NNS sequences, on the other hand, most frequently contain a NS confirmation check as a second turn which proposes a model for meaning intended in the NNS's trouble source. Such checks are usually followed by an affirmative feedback signal of the NNS. This indicates that in both NS and NNS sequences, the NS manipulates content and form, and the NNS uses feedback signals most frequently. The NS not only requests clarification but indeed tries to relieve the NNS from the task of clarification by proposing a model for repair. Thus there appears to be asymmetry in NS and NNS negotiation of ineaning behavior. The next research question concerned the amount of interaction which is dedicated to negotiation of ineaning. This amount may influence communicative effectiveness as well as opportunities for language acquisition. Pica (1994) and Pica et al. (1989) hypothesized that if there is a great deal of negotiation of ineaning, many opportunities for language acquisition occur. Negotiation of ineaning will lead to more comprehensible input by the NS and to more opportunities to create comprehensible output by the NNS (Swain, 1985). However, communicative effectiveness may at the same time be hindered. If there is a great deal of negotiation, the interlocutors may not be able to achieve their goals. The first part of the question was: 2a. What is the amount of interaction spent on negotiating meaning? The percentage of turns (utterances of a speaker which end another speaker's prior turn) which were part of a clarification sequence was calculated. The relative share of turns spent on negotiation of ineaning was variable but high in most conversations (from around 20oIo up to 70oIo; see Chapter 5). Other studies such as Gumperz (1990), Mazeland (1992), and Glahn (1985) found lower shares, even though most included self-initiated self-repairs, which were excluded in this study. The high levels of negotiation of ineaning in the current study may, on the one hand, create opportunities for language learning. On the other hand, such high levels may also threaten communicative goals and create face risks. Two factors which may have affected the amount of negotiation of ineaning were investigated. The second part of the research question concerned these factors: 204 Cltapter 10 Concluslon 2b. How do NNS communicative competence and interaction type influence the amount of interaction spent on negotiating meaning? It was expected that the amount of negotiation would decrease as the NNS's communicative competence increased, at least after a certain threshold level was reached. For most NNS informants, there only appeared to be a decrease over time in the institutional conversations. Apparently the factors NNS communicative competence and interaction type interact with each other in influencing the amount of negotiation of ineaning. In the institutional conversations, the amount of negotiation of ineaning probably decreased because the NNSs acquired the script, terminology and concepts of a housing office registration interview. In the informal conversations, the level of negotiation of ineaning over the three points of ineasurement was high, but more or less stable around SOqo. No relationship between level of NNS communicative competence and amount of negotiation was found. This may indicate that in less goal-directed conversations more continuous negotiation takes place. The relatively high level of negotiation in the informal conversations may be caused by several factors. At all three points of measurement, the NNSs were speakers with relatively low proficiency. New topics may have created new problems, especially because a script that could support best guessing was lacking. This may have kept the proportion of negotiation of meaning more or less constant. The level of negotiation in NS-NNS interaction would probably remain higher than in NS-NS interaction, until the NNS reaches near-native competence and shares more cultural background with the NS. In the current study, even at the third point of ineasurement NNS communicative competence was still quite low compared to other studies because the NNSs received little or no instruction in Dutch. In addition, the semi-authentic conversations may have differed from spontaneously occurring ones in the motivation of the speakers (in particular the NSs) to keep the conversation going in spite of trouble. In naturally occurring conversations, NSs may more quickly decide not to negotiate but abandon the conversation instead. The third question concerned asymmetry in NS and NNS roles in negotiating meaning (see Chapter 6). The outcome of research question lb pointed to unequal NS and NNS participation. Perdue (1993: 254) stated that asymmetry in NS-NNS interaction leads to a negative learning environment. However, the literature did not show unequivocally that NSs always dominate conversations with NNSs. Thus this issue was studied in more detail for the main negotiation moves, that is, trouble indicators, trouble clarifications and confirmation checks. Due to limited communicative competence and interactional role, the NNSs were expected to display different preferences in the types of moves they used. The NS was expected to clarify trouble and propose models for the NNS's intended meaning more often, because the NS had the linguistic means to manipulate content. The NNS was expected to use feedback signals to indicate trouble most frequently, for which less communicative competence is required. This leads to asymmetric participation in negotiating meaning and the NS doing most of the linguistic work. Such unequal 10.1 Summan~ and Discussion of the Results 205 participation may limit NNS opportunities for language acquisition, in particular if NNSs do not clarify trouble frequently and thus do not create comprehensible output. In addition, it may affect the outcomes of the negotiation process. The first part of the third research question was: 3a. Is there asymmetry in NS and NNS participation in negotiating meaning? This question was answered positively. There appeared to be a great deal of asymmetry between NSs and NNSs. The results show a complementary role distribu[ion in negotiation of ineaning between the NSs and NNSs. The NNSs indicated trouble most often, whereas the NSs clarified trouble most often. Most sustained asymmetry was found for the confirmation checks. When the NSs had trouble understanding the NNSs, they used confirmation checks more often than trouble indicators. With such checks a speaker can manipulate the content and sometimes also the outcomes of the interaction. Thus the use of confirmation checks may be indicative of NS dominance. Two factors which may explain the observed asymmetry in negotiation of ineaning were NNS communicative competence and interaction type. The second part of the third research question was: 36. How do NNS communicative competence and interaction type influence the distribution of negotiation moves over the NS and NNS? The role distribution can best be explained by the different levels of communicative competence. There is some development over time towards more symmetry for NS and NNS trouble indicating and clarifying behavior, in particular, in the informal conversations. However, even at the last point of ineasurement, there was still considerable asymmetry. Interaction type also had some influence on the asymmetry in NS and NNS interactional behavior, particularly on trouble clarifications and confirmation checks. In the institutional conversations there appeared to be somewhat more sustained asymmetry over the three points of ineasurement than in the informal conversations; however, not as much as expected. A factor which may have diffused the results is expert status, which consists of authoritative role and topic knowledge (Zuengler 8z Bent, 1991; Woken 8c Swales, 1989). In the institutional conversations, the native speaker housing official Bart had both an authoritative role and topic knowledge based on his training and experience on the job. Yet in both interaction types, the NNSs had topic knowledge and thus were informal experts as well, when the conversation concerned their own life experiences. Zuengler 8z Bent (1991) found that when the NS and NNS have equal knowledge there is no clear pattern of asymmetry. Rather, on some measures, the NS and NNS score equally (topic moves, interruptions), whereas on others one or other party scores highec They state that the NS talks more and the NNS uses more fillers and backchannels. In other words, the NS is seen as the active speaker and the NNS as the active listenec In the current study, this role distribution manifests itself in that the NS manipulated content more frequently and the NNS used trouble indicators and 206 Chapter IO Conclusion other feedback signals more frequently. It is necessary to detine more cleazly what expert status consists of, and how authoritative role and topic knowledge influence role distribution. Topic knowledge obtained through formal study and training should be distinguished from informal expert knowledge based on, e.g., cultural background. In addition, the fact that the institutional conversations were semi-authentic play-acting scenes may be a cause for asymmetry not materializing as much as expected. Nothing was really at stake, after all. The sustained asymmetry in the informal conversations could be explained by the NSs having higher social and institutional status. The three main negotation moves, i.e. trouble indicators, trouble clarifications, and confirmation checks, were further analyzed as to strategy type and linguistic form. Differences were determined between the NSs and NNSs in quantity and variety of types and forms. In addition, these analyses could show NNS development in negotiating meaning as their communicative competence increased. The results were compared for both interaction types to determine the influence of a script and institutional roles. Thus, the fourth research question was: 4a. How can NS and NNS trouble indicators, trouble clarifications and confirmation checks be described in terms of their types and forms? 4b. What is the influence of NNS communicative competence and interaction type on the variety and quantity of trouble indicating, clarifying and checking behavior of the NS and NNS'? The results of this research question will be discussed separately for the three moves, trouble indicators, trouble clazifications, and confirmation checks, consecutively. The types and forms of the NS and NNS trouble indicators were described and compared in Chapter 7. The trouble indicator types could be classified on several dimensions. First, trouble indicators could be verbal or nonverbal. Nonverbal indicators were significant silences with or without nonverbal signals, followed by trouble clarifications. Verbal indicators could be further positioned on two dimensions according to the explicitness with which a trouble source was characterized and according to the specificity of the characterization of the trouble source. This resulted in four combinations. For example, a global 8z implicit indicator would be `what?'. A specific 8c explicit indicator would be `I don't understand X'. As far as the types of trouble indicators are concerned, it can be concluded that the NSs generally tended to be specific and implicit (off-record) in indicating trouble; however, they also used global, explicit and nonverbal trouble indicators. This shows their ability to adapt the trouble indicators to the NNS and to the trouble source. In contrast, the NNSs most frequently indicated trouble in nonverbal and global ways, which requires little linguistic competence. The NNSs increasingly used more specific 8c implicit indicators, which reflects some development in their linguistic competence. 10.1 Stunmarv and Discussion of the Results 207 As far as trouble indicator forms are concerned, there was one major difference between NSs and NNSs. The NSs used many more complex Wh-questions for their specific 8z implicit indicators. Differences in communicative competence may explain this outcome. Wh-questions require a considerable amount of linguistic competence. There were a couple of qualitative differences between NSs and NNSs. Word requests were typically used by the NNSs and not by the NSs. Gaps in NNS vocabulary probably led to such requests. In contrast, only the NSs used fill-in-the-blank questions, which provided the NNSs with a model to be completed. Thus the use of such questions may be a typical NS helping strategy. The NNSs used mainly minimal feedback questions and only infrequently other question types. As far as the individual NNSs are concerned, there is a growth in diversity of trouble indicator types and forms for the three male informants but not for the Moroccan female Fatima. She generally kept using nonverbal cues for indicating trouble. There are some systematic differences between the relatively slow learners Fatima and Mahmut on the one hand, and the relatively fast learners Mohamed and Ergun on the other. Although Mahmut from the first cycle on employed a greater diversity of trouble indicator types and forms than Fatima, both showed the least development in trouble indicator types and forms. Mohamed and Ergun, on the other hand, developed in the range of types and forms they used. Thus communicative competence appeared to be an explaining variable for the differences in NNS trouble indicator types and forms. There also appeared to be some influence of source language. The Turks used keyword repetition more frequently than the Moroccans. The use of repetition is a typical conversational characteristic of Turkish speakers (Allwood, 1993). There was also some influence of interaction type. In the institutional conversations, the NSs used more global means of indicating trouble than in the informal conversations. This outcome is caused by one NS in particular, namely Bart, the housing official. Whether his use of global means originates from his professional role or from personal style is not clear. In general, the NSs used a greater variety of forms in the informal conversations. In particular, more fill-in-the-blank questions were used. In the institutional conversations, the NSs used Wh-questions more frequently than in the informal conversations. Furthermore, in about 30010 of the cases, the NNSs used minimal feedback questions in these conversations. The NNSs also most frequently used nonverbal trouble indicators in the institutional conversations. This may show a strong need for face-protection. In the informal conversations, the NNSs most frequently used global8c implicit and specific 8c implicit trouble indicators. The NNSs used word requests only in this context. This shows their motivation to communicate and acquire words needed to express intended meaning in this context. In addition, the use of more explicit and specific indicators may indicate that they had less need to save their own or NS's face, as in an asymmetrical context (see also Woken át Swales, 1989). No great differences in NNS trouble indicator forms were found for the two interaction types. This indicates that the NNSs did not adjust their forms to the demands of the interaction. 208 Chapter 10 Conclusion The types and forms of NS and NNS trottble clarifications were described in Chapter 8. These mainly concern a trouble source of the same speaker. Thus trouble clarifications manipulate the content and form of a speaker's own utterance rather than of the other. Clarification can be realized in three ways: by deleting from, adding to, ancUor substituting (part ot) the trouble source. These three processes can occur with or without repetition of (part ot) the trouble source. They may operate on a local level, manipulating part of the trouble source, or they may operate on a global ]evel, repeating, rephrasing or expanding the whole trouble source. Clarification sequences always lead to adjustment of the interactional structure and thus all trouble clarifications are conversational adjustments (Long, 1983b). The dimensions mentioned above led to a tripartite classification of trouble clarification types: those with pure conversational adjustments (full and partial repetition), those with weak linguistic adjustments which requïre some linguistic manipulating (e.g. elaboration, additive repetition), and those with strong linguistic adjustments which requïre major linguistic manipulation of the trouble source (e.g. paraphrase, hypothesis forming, translation). This last type of clarification is particularly difficult for NNSs with limited linguistic competence to comprehend and to produce. Repetition and elaboration were the trouble clarification types most frequently used by both NSs and NNSs. Intuitively, NSs and NNSs appeared to use those strategies that were found to be effective in increasing comprehension (Pica et al., 1987; Chaudron 8z Parker, 1987). Strong linguistic adjustments, the effects of which on comprehension were found to be less consistent, were used infrequently. Paraphrase, whích requires strong linguistic manipulation, was used more by the NSs than by the NNSs, but still not as much as an adjustment type such as repetition or elaboration. The NSs rarely used corrections, translations and word supply but when they did, they were mainly in the informal conversations. These conversations generally appeared to be more conducive for negotiation with a focus on languagelearning. Similar to the NSs, the NNSs used several kinds of repetitions frequently, but they preferred partial and complex ones. In addition, the NNSs used elaborations, which are characterized as weak linguistic adjustments, even more frequently than the NSs. The adjustment process of addition was apparently used by the NNSs with as much ease as repetítions. Strong linguistic adjustments were hardly used at all, except for corrections. These were used more frequently by the NNSs than by the NSs and mainly concerned content of the speaker's own utterances rather than form. The outcomes confirm the expectation that NNSs would mostly prefer pure conversational adjustments. However, weak linguistic adjustments were also used. The NS clarifications generally were in the form of questions as often as they were declaratives. However, in the institutional conversations, questions occurred more. The interview format may have led to many NS questions. Such questions were repetitions and paraphrases of original trouble sources, which were questions too. The housing official Bart used many ungrammatical clariiications. Long (1983b: 179) hypothesized that ungrammatical adjustments occur when (i) the NNS has a very low proficiency in the L2, (ii) the NS perceives himself as being of higher so- I0.1 Sirmman' and Discussion of the Results 209 cial status, (iii) the NS has prior foreigner talk experience, and when (iv) the conversation occurs spontaneously. Snow et al. (1981) added that abstract complex topics lead to more ungrammatical adjustments. All these conditions are more or less fulfilled for the housing official Bart. Yet Paul, the social worker for whom all these conditions were also applicable, did not resort to ungrammatical talk. Therefore, the choice for ungrammatical talk also shows a personal preference. Habit may explain the differences. Bart regularly worked with NNS minority members in his job at the housing office. Thus he may have developed his foreigner talk register to deal with the demands of the situation. Bart's `foreigner talk' may not be conducive for language leazning, but it appeared to be effective for the communication because he often used it when grammatical means had failed. He generally appeared more communicatively effective than Paul, who appeazed to have more trouble adjusting optimally to the NNS's communicative competence. The NNSs used almost only declazatives, in particular sentence fragments, to clazify trouble. Thus the NNS's comprehensible output in oral interaction did not consist of grammatical, fu11, finite sentences. There was little or no development over time in the NNS clarification types and forms, while there was some development for the trouble indicators. This is a remarkable outcome. Apparently the NNSs found the minimal means they used effective enough. If this is the case, negotiation of ineaning does not lead to much NNS comprehensible output even if the NNS is given the oppor[unity to create such output. Finally, a qualitative difference between the NSs and the NNSs was that only the NSs used word supply, in response to NNS word requests. The types and forms of the NS and NNS confcrmation checks were discussed in Chapter 9. Confirmation check types were classified with the same categories as trouble clarifications. The difference between a trouble clarification and a confirmation check is that the latter manipulates the content of the other's utterance rather than of one's own utterance. Confirmation checks, thus, often contain a high degree of uncertainty. Hypothesis forming types which propose models for the other's intended meaning were frequently used by the NSs as well as by the NNSs. These are strong linguistic adjustments. However, all kinds of (adjusted) repetition occurred frequently also, when the speaker was more certain about the other's intended meaning and only minimal rephrasing was necessary. In that case more pure conversational adjustments (full and partial repetition) were used. Confirmation checks were used much more infrequently by the NNSs than by the NSs. The NNSs most frequently used hypothesis forming, followed by partial and complex repetition. In the informal conversations, they used full and partial repetition most frequently and in the institutional conversations hypothesis forming and complex repetition. Due to the frequent use of hypothesis forming, confirmation checks were more frequently strong linguistic adjustments than trouble clarifications. Such adjustments were hypothesized to be more difficult to comprehend for the NNS. However, NS linguistic adjustments used in confirmation checks concern the NNS's utterance. Such adjustments may be much easier to understand for the NNS than 210 Chapter 10 Conclusion linguistic adjustments in trouble clarifications because the NNS's own intended meaning rather than the NS's meaning is rephrased. Thus, the effect of linguistic adjustments on comprehension in confirmation checks may be different from their effect on trouble clarifications. As far as the forms of confirmation checks are concerned, as expected, questions which only require a simple feedback signal for an answer were used most frequently. It was expected that declarative questions would be the most typical question type used for checking. This was, however, not the case. In general in oral interaction, there is heavy reliance on incomplete structures or satellite units (Bygate, 1988). This may partly explain the relative absence of declarative questions with complete sentence structure. However, full sentence structures may also not have been used because they were still too demanding for the NNS to produce and comprehend. Instead, elliptic questions were used. Generally no full sentence types were used for checking by the NSs, and even less by the NNSs. The NSs only used declarative questions in some cases, in particular in the institutional conversations. The rest of the time, the NSs mainly used elliptic questions. Declaratives were used quite frequently too, in particular by Janet and Bart. This may show their relative confidence in interpreting the NNS's intentions. Qualitative differences between NSs and NNSs were again the sole use of word supply and of declarative questions by the NSs. The NNSs used elliptic questions in more than SOoIo of their checks. No development could be shown over the points of ineasurement. It may be that such development did not show because of the relatively low NNS frequencies. Alternatively, requesting confirmation can be done with minimal means, thus there may be no incentive to create comprehensible output in fuller linguistic forms. There were few differences in the distribution of NS confirrnation check types over the two interaction types. More hypothesis forming occurred in the informal conversations, showing the NSs had fewer clues to interpret the NNS's utterances. In the institutional conversations, more complex repetition was used. There were individual differences between the NSs in the distribution of confirmation check types. In the informal conversations, Janet preferred hypothesis forming and word supply. Thus she appeared to support the NNSs in supplying the correct Dutch wording for their intended meaning. In the institutional conversations Paul, on the other hand, used partial repetition most frequently, which may reflect his difficulty in making sense of the NNS's utterance. He stayed on the safe side by repeating what he did understand. In conclusion, even though the same taxonomy could be applied for trouble clarification and confirmation check types, there are different preferences for the two moves. Clarifications manipulate the content of the speaker's own trouble source, and therefore, they were mainly conversational adjustments, such as repetitions and elaborations. Confirmation checks manipulate the other speaker's trouble source. Thus they often contain strong linguistic adjustments, such as hypothesis forming, which is used when the other speaker's intention is perceived to be widely disparate from the linguistic realiaation. 10.1 Summary and Discussion of the Resu[ts 211 10.2 Theoretical Implications of the Findings In this section, the implications of the results for both intercultural communication and second language acquisition theory will be discussed. 10.2.1 Intercultural Communication Theory To investigate what intercultural communicative competence entails in negotiation of ineaning an interactionist perspective was chosen. Intercultural communicative competence was defined as not only the nonnative speaker's competence to deal with his limited proficiency but also as the native speaker's ability and willingness to accommodate to the NNS (Dirven 8t Putz, 1994). Conversation analysis methodology was used to describe NS and NNS interactional behavior in clarification sequences in detail. Sequential analyses showed how NS and NNS behavior affects the other speaker's contributions, communicative effectiveness, and role distribution. In addition, both NS and NNS negotiation moves were coded with the same taxonomies for move types and forms. The classification of NS and NNS negotiation moves made it possible to show differences between NSs and NNSs. Intercultural communicative competence in negotiation of ineaning can be defined as the ability (i) to use feedback signals to indicate trouble and continue the negotiation (ii) to manipulate content and form, and (iii) to metacommunicate about the problematic character of the interaction. In this study, asymmetry was determined in NS and NNS quantitative and qualitative contributions to negotiation of ineaning. Trouble indicating was done most frequently by the NNSs. It can be done with minimal feedback signals and thus little linguistic competence is needed. This is the easiest way for the NNS to exercise some control over the outcome of the conversation. However, he has to balance such intervention with demands for face-protection and prevention of communication breakdown. Indicating trouble stimulates the NS to tune his input to a comprehensible level. For trouble clarifications and confirmation checks some manipulation of content and sometimes form is necessary and thus these moves require more linguistic competence. Consequently, the NNS clarifies less and, in particular uses fewer confirmation checks. The latter require stepping into the other's shoes, which is more difficult for a NNS, and it may be more risky too when there is a status difference. The NS on the other hand, also takes a risk by using confirmation checks, in particular, those in the form of hypotheses, which present assumptions about the NNS's intended meaning. The NNS often agrees too easily with proposed models of intended meaning. This can lead to miscommunication which may remain unrecognized but which may affect the outcome of the conversation negatively. If the NS uses many confirmation checks, the NNS loses control over the content and outcome of the interaction. If instead the NS uses trouble indicators, the conversation may proceed more slowly. The NS may be perceived as uncooperative because he has means at hand to relieve the NNS from clarification. However, in 212 Chapter ]0 Conclusion some cases, indicating trouble and leaving the clarification to the NNS may lead to more genuine and effective information transfer. NS-NNS interaction appears to be asymmetric through several factors. In interethnic communication, minority members usually have lower ethno-linguistic status. In institutional conversations, they have lower status because the institutional representive has expert status. If the minority member is a nonnative speaker, he also has less control over negotiation of ineaning because he has less ability to manipulate content. In addition, when the minority member is a woman, this may also lead to lower status. The effects of accumulative lower status were most visible for the female Fatima, whose results deviated from those of the three male informants. Little influence of source language background could be detected, whereas other studies in the framework of the ESF project found considerable source language influence (Slobin, 1993). NSs and NNSs appeared to use the same strategy types and forms of negotiation moves, although the NNSs lacked some variety and prefer linguistically simple means. Thus negotiation of ineaning competence may be a more universal problem-solving ability than a language or culture specific one. Whether this hypothesis is correct should be determined by research on NNSs from other source languages, learning different target languages, and holding different levels of language proficiency. 10.2.2 Second Language Acquisition Theory There are two aspects of second language acquisition theory relating to negotiation of ineaning which deserve some special attention. The first aspect concerns Pica's claim (1994) that negotiation of ineaning may stimulate second language acquisition. The second aspect is the distinction between linguistic and conversational adjustments posited by Long (1983b) for NS input. The main goal of interaction is understanding. However, interaction can for NNSs also have a positive effect on language acquisition. Pica (1994) claims that negotiation of ineaning, in particular, might stimulate language acquisition. Yet different negotiation moves may influence opportunities for language acquisition in different ways. Negotiation of ineaning in problematic communication may have positive effects for comprehensible input and comprehensible output and in particular, it may sensitize the NNS to noticing target language forms and functions. A distinction should be made between moves which may manipulate content to create NS comprehensible input or NNS comprehensible output, and those moves which can be realized with minimal feedback signals. The latter may be more effective for communication than for language acquisition. The NSs used trouble clarifications and confirmation checks regularly which most likely increased the amount of comprehensible input, which may have stimulated language acquisition. The NNSs most frequently used trouble indicators which stimulated the NSs the adjust their mput to a comprehensible leveL However, they used trouble clarifications less frequently than the NSs. Thus the NNSs had some 10.2 Theoreticul Lnpllcations of the Findings 213 opportunity to create comprehensible output, which requires morphosyntactic analysis and which may be important for language acquisition as well (Pica 8c Doughty, 1988). However, when the NNSs claritied trouble they rarely adjusted linguistically and mainly used sentence fragments. In addition, the NNSs used confirmation checks infrequently. Those confirmation checks they did use, even if content and form were manipulated, were realized with minimal forms. It is questionable how comprehensible NNS output was in terms of Swain's definition, that is, whether it "extends the linguistic repertoire of the learner as he or she attempts to create precisely and appropriately the meaning desired" (Swain, 1985: 252). It is necessary to define more exactly what comprehensible output consists of. It is not clear whether the NNS adjustments following other-initiations lead to comprehensible output. It may be that learner's self-initiated self-repairs are more effective for the creation of comprehensible output, as Shehadeh (1991) found. Such self-initiated self-repairs concern language problems the NNS notices spontaneously. Therefore, such attention may be an important factor in second language acquisition (Pica, 1994). Negotiation of ineaning may also increase NNS attention to content and language form. If the NNSs indicate trouble themselves, this is proof of their attention. NNS word requests display NNS attention to the target language forms. The sequences following such word requests often result in the provision of the requested word and several repetitions, which may support long-term storage of that word (see e.g. the negotiation of ineaning of oorbel - earring Chapter 5, Example 4). In addition, when the NS indicates trouble or uses confirmation checks, the NNS is forced to pay attention to the NS's input and his own output. It appears obvious that negotiation of ineaning may heighten NNS attention to content and probably also to linguistic form. However, the forms brought to attention may not be at the NNS's `i f 1' level (Krashen, 1981, 1982). Therefore, acquisition may not take place. It is worth remarking that the negotiation of ineaning mainly concerned content, lexicon, and pronunciation. Sometimes negotiations of ineaning concerned tense aspects. However, those were often difficult and unsuccessful. Other linguistic aspects such as morphology were not negotiated at all. This may also relate to the type of language Dutch is. In Dutch one can communicate a great deal without using the morphology. It might be that learners of agglutinative target languages such as Finnish, where morphology can barely be circumvented, have to negotiate those aspects of linguistic form (personal communication, Suni 1995). It can be concluded that access to comprehensible input is a necessary but not a sufficient condition for language acquisition, which is present in clarification sequences. Comprehensible output is probably also a necessary but not a sufficient condition. In any case, the analyses showed that the NNSs did not create a great deal of comprehensible output. NNS attention to form may be another necessary condition of which there was some evidence in the clarification sequences. However, the three conditions together, NS comprehensible input, NNS comprehensible output and NNS attention to form may still not be sufficient for language acquisition. Many NNSs in the ESF project managed to communicate with a basic language variety which consists of "first- and second-person pronouns, one or two locative prepositions expressing generalised topological relations. sentence-level 214 Chapter 10 Conclusion negation, a coordinating conjunction, and some temporal adverbs" (Slobin, 1993: 241). One may wonder why a NNS would move beyond this stage if it works in communication. O[her sociopsychological factors, such as motivation and attitudes probably account for this, as research on speech accommodation has shown (Giles et al., 1987). Mohamed, for example, was the most reluctant communicator in this study. However, he was the best learner, who learned Dutch through daily interaction with his Dutch girlfriend. The fact that he had this girlfriend apparently motivated him to move beyond a basic language variety. The other aspect of second language acquisition theory which deserves attention concerns the distinction between conversational and linguistic adjustments. NS and NNS adjustments in trouble clarifications and confirmation checks were classified into three processes, that is deletion, addition and substitution. In addition, Long's distinction between NS linguistic and conversational adjustments was applied to NS as well as to NNS negotiation of ineaning behavior. Long (1983b) hypothesized that the type of NS adjustment may matter for language acquisition. He hypothesized conversational adjustments were more conducive to language learning, because they leave the linguistic input intact but provide more processing time and redundancy. However, the distinction between the two types of adjustments is problematic for the negotiation moves in the present study. T'here appeared to be no strict differentiation in many cases. All negotiation moves lead to adjustment of the interactional structure and are thus conversational. Some moves, such as elaboration, which Long (1983b) classifies as conversational adjustments, also add new linguistic material and thus also result in linguistic adjustments. In addition, a typical conversational adjustment such as repetition is sometimes accompanied by linguistic adjustments such as an addition or a substitution for a word by a synonym. Such combinations are, therefore, in the current study classified as weak ]inguistic adjustments. Other adjustments more strongly manipulate linguistic form, such as paraphrase and hypothesis forming. If Long's claim is correct, these strong linguistic adjustments may generally be less effective in improving comprehension and stimulating acquisition. However, strong linguistic adjustments such as hypothesis forming may have a very different impact on comprehension than on paraphrase. Through the use of hypothesis forming a speaker rephrases the other speaker's intended meaning. Such rephrasing may be much easier for the other speaker to comprehend than a paraphrase, which rephrases the content of the trouble source of the speaker himself. Long's classification of conversational versus ]inguistic adjustments also does not do justice to the three adjustment processes operating on NS as well as NNS clarifications: deletion, addition and substition. Long (1983b) did not make a distinction between deletion and addition. From an acquisitional perspective, it is useful to make such a distinction between deletion and addition because they create comprehensible input in different ways. Deletive repetitions decrease processing demands by reducing the total amount of input and by creating more redundancy, whereas additions increase the amount of input to create additional context for understanding. 10.2 Theoretical Intplications of the Fi~uluigs 215 10.3 Practical Implications of the Findings In this section, some suggestions are given for the implications of the findings for the practice of intercultural communication and language teaching. The results are reviewed for those characteristics of negotiation of ineaning whích may increase communicative effectiveness and opportunities for language acquisition. Given the fact that this is an explorative multiple case study, conclusions drawn are only tentative and should be backed up by further research. When a NNS indicates trouble, he creates an opportunity to receive more comprehensible input, which enhances the chances for communicative success and of which he may make use for his language acquisition. Generally, implicit and global trouble indicators will be sufficiently effective. NNSs are able to use these from the initial stages of language acquisition onward. Implicit indicators are generally preferred over explicit ones, because they are more face-protective. NNSs prefer global indicators over specific indicators, because the exact identification of trouble requires more linguistic competence. Yet in some cases, it is important to specify the trouble through a specific indicator. It helps the interlocutor to address the trouble and tune his input. The NSs used specific indicators more regularly than the NNSs. In second language classrooms, students could be trained to indicate trouble in various ways. However, for NNSs with very limited proficiency, it would probably be impossible to segment a trouble source and pick out a specific element if little processing takes place. When the NS has trouble understanding the NNS, he tends to use confirmation checks rather than trouble indicators. Hence the NS tends to take on the linguistic task of manipulating the content and form, independently of whether it is he himself who causes the trouble or whether it is the NNS. However, if the NS does most of the manipulation of the content, the NNS has little opportunity to do such manipulation himself, and thus to create comprehensible output which may stimulate language acquisition. Therefore asymmetry in negotation of ineaning limits the opportunities for language acquisition in terms of NNS productive opportunities. In addition, asymmetry creates a communïcative risk. When the NS provides a model for the NNS's intended meaning, the NNS generally tends to accept this model, although he may not have fully understood its meaning. He may trus[ the NS in his cooperative attitude to support him in communication. This may or may not be justified. In any case, when the model is not correct, the NNS may not notice this or may not able to correct it. This may influence the NNS's communicative effectiveness negatively. Such negative effects particularly occur in those institutional contexts in which little time is available for the realization of the NNS's communicative goals. When there is time, the NS should make space for the NNS to at least occasionally clarify trouble himself. This would be the most reliable way of checking the NNS's intended meaning. In addition, the NNS may profit from the opportunity to clarify from an acquisitional point of view, because he is pushed to produce comprehensible output which is more correct and appropriate. 216 Chapter 10 Conclusion 10.4 Scope of the Study and Suggestions for Further Research This study concerned negotiation of ineaning in NS-NNS interaction. Both NS and NNS behavior was studied. The NS data were used as baseline data with which the NNS data were compared. It was hypothesized that NS negotiation of ineaning behavior in interaction with NNSs is qualitatively similar to such behavior in interaction with other NSs. There were, however, no comparable NS-NS data available to test this claim. A suggestion for further research is to compare negotiation of meaning in NS-NNS interaction with NS-NS interaction to determine whether NS negotiation of ineaning behavior is indeed similar in both contexts. Such comparison may better show the impact of NNS communicative competence on the interaction. Yet it may be difficult to find enough comparable data, because negotiation of ineaning occurs much less frequently in NS-NS interaction in which speakers share the same linguistic and cultural background. However, in particular the comparison of negotiation of ineaning in institutional conversations, in which both NS and NNS are clients, may be revealing as to the influence of NNS communicative competence on communicative effectiveness in this type of interaction. In this study, the influence of communicative competence remained unclear in various aspects. There were differences between NSs and NNSs, particularly in the frequency of confirmation check use and in the types and the linguistic realizations of trouble indicators and clarifications. However, the longitudinal dimension through which NNSs could be compared with themselves at three differents points in time, did not show as much development as expected. Only NNS [rouble indicators showed some development, but NNS clarifications and confirmation checks did not. There may be several explanations for this outcome. First, negotiation of ineaning may depend on general problem solving strategies which are independent of second language competence, because such strategies have been acquired earlier in the first language. However, there were clear differences between NSs and NNSs in the quantity and quality of negotiation moves, so there is at least some influence of level of communicative competence. Secondly, the interval of 9 months between the points of ineasurement may not have been long enough. Extension of the period between the points of ineasurement may lead to results showing more development over time. In addition, study of NNSs who receive language instruction may show more development over a shorter period of time. Other longitudinal studies should determine how negotiation of ineaning is affected by the NNS's level of second language proficiency. Other variables which influenced the results of this study are factors such as gender, cultural background, and expert status. In particular, the female NNS differed in some aspects from the three male NNSs in negotiation behavior. Pica et al. (1991) and Gass 8t Varonis (1986) found that gender affects the negotiation of ineaning process. Therefore, this variable should be studied more systematically in further research. Little ínfluence of cultural background could be traced on negotiation of 10.4 Scope of the Study~ and Suggestions for Further Research 217 meaning. Again this may originate from the fact that negotiation of ineaning depends on general problem solving competence. However, with bilingual researchers it may be possible to trace source language and cultural influences better. Another factor which diffused the results of the current study somewhat was expert status. Expert status consists of authoritative role and topic knowledge. Other research has shown that asymmetry in NS-NNS interaction may have been more influenced by expert status than by communicative competence (Woken 8c Swales, 1989). In particular, topíc knowledge appeared to have an ímpact on the interaction. Such knowledge may make someone an informal expert, even if he dces not possess an authoritative role. Therefore, more research is needed on how authoritative role and topic knowledge influence expert status. In particular, research on NS-NNS interaction in which the NNS has expert status may clarify the influence of this factor on negotiation of ineaning. In this study, semi-authentic data were used. The analyzed conversations were recorded at the university studio. For the housing office conversations, a housing official and a social worker were asked to participate. In the informal conversations, project researchers conversed with the NNSs. Considering the goals of the ESF project, this was the best choice. For the current study, this choice has advantages and disadvantages. The advantages from an acquisitional point of view are that interaction data from four NNS informants at three different points in time could be compared; in addition, it was not necessary to gain access to an institution, which might have refused cooperation to protect the privacy of its personnel and clients. The disadvantage is that it is not clear in which aspects negotiation of meaning behavior might differ under real time pressure, when there is really something at stake. Efficiency, economy, time pressure and background knowledge are all aspects which influence institutional conversations (Agar, 1985). The first three aspects limit the amount of time available to the participants. These restrictions are present in all forms of business talk (see Stajpers, 1993). In an authentic setting, the NSs might have been less cooperative, thus limiting the amount of negotiation of meaning and the opportunities for NNSs to achieve their communicative goals. Research such as that executed by Clyne (1994) on intercujtural communication at 'work' between NNSs of different linguistic backgrounds shows the possible impact of authenticity on NNS language behavior. Another effect of the semi-authenticity of the conversations is that the presence of a videocamera and researchers who recorded the sessions appeared to influence the conversation once in a while. In the next example, Bart appeared not only to be addressing Ergiin, who probabjy had not the faintest idea what Bart was talking about, but also the researchers who were recording the session. (1) Institutional conversation with Ergiin, Cycle 2 Bart: ja maar diel er is geen baby in het spel t vrouw is niet in verwachtíng `(vrouw] Ergun: `(nietJ wachten? Bart: vrouw krijgt geen baby? 218 yes but thad there is no baby yet wife is not pregnant `f wifej `[not] wait? wife is not having a baby? Chap~er 10 Conclusion Ergiin: Bart: Ergun: Bart: Ergun: Bart: Ergiin: Bart: Ergun: Ban: Ergiin: Bart: o0 of wel? t `[(xx)] `[krijgt] vrouw baby? nu? nee nou niet nu? geen baby hè? nee nee nee dat gaat geen jaren duren hè? ja het gaat geen jaren duren als een vrouw in verwachting is t er staat negen maanden voor ja bij olifanten duurt het negenjaar geloofik t oh or is she? t `[(xx)] ~s wife having baby? now? no not now now? no baby huh? no no no that dcesn't take years huh? yes it dces not take years íf a woman is pregnant t it normally takes nine months yes with elephants it takes nine yeazs 1 believe t dat is wat anders that's something else Therefore, some influence of the experimental setting was found. On the linguistic level, there may have been less influence. In particular, the NNSs may not have been capable of manipulating their language. However, they may have been more or less reticent because nothing was really at stake. Fatima and Mohamed, who were the least comfortable in the sessions, appeared quite reticent. Mahmut on the other hand, appeared to enjoy the conversations and spoke freely. Yet, in his conversations with Bar[, the housing oftïcia[, another problem occurred due to the semi-authenticity of the conversations. At a certain point in one of the conversations, reality and play became confounded. Bart gave some real advice on Mahmut's housing situation. Mahmut became very confused, because he did not realize that Bart's advice was serious. Another limitation of the study relates to the coding procedure. Once the coding scheme was established, the reliability of this scheme for the quantitative analyses could have been improved by a second rater. An interrater reliability test could have established the reliability of the coding scheme and thus its utility in future research by other researchers on other types of data. However, this was not possible within the constraints of the current study. The codings of the negotiation moves were added to the examp[es to give the reader a chance to check the researcher's coding decisions. Finally, the current study was a multiple case study and there appeared to be a great deal of individual variation between the informants. Therefore, it is difficult to generalize from the results. Now that a coding scheme has been established, it could be applied by other researchers to larger numbers of NS and NNS informants. In particular, larger groups of different source language speakers could provide more insight into the influence of source [anguage background on negotiation of ineaning processes. It could thus be determined which aspects of negotiation of meaning behavior are based on universal problem solving strategies and which aspects are languagelculture specific. 10.4 Scope of the Study and Suggestions for Further Research 219 List of References Adelswaerd, V., Aronsson, K., Jánsson, L. 8c Linell, P. (1987). The unequal distribution of interactional space: Dominance and control in courtroom interaction. Text, 7, 313-346. Agar, M. (1985). [nstitutional discourse. Text, 5, 147-168. Ainsworth-Vaughn, N. (1992). Topic transitions in physician patient interviews: Power, gender, and discourse change. Language and Society, 21, 409-426. Allwood, J. (1986). Some perspectives on understanding in spoken interaction. Acta Philosophica Gothoburgensia, 1, 13-59. Allwood, J. (1993). Feedback in second language acquisition. C. Perdue (Ed.) The crosslinguistic study of second languages. ~olume II: The results. Cambridge: Cambridge University Press, pp. 196-235. Aries, E. (1987). Gender and communication. P. Shaver 8t C. Hendrick (Eds.) Sex and gender. Newbury Park: Sage, pp. 149-176. Asante, M. c4r Gudykunst, W. (Eds.) (1989). Handbook of international and intercu[tural Communication. Newbury Park: Sage. Aston, G. (1986). Trouble shooting in interaction with leamers: The more the merrier? Applied Linguistics, 7, 128-143. Banks, S., Ge, G. 8r Baker, J. (1991). Intercultural encounters and miscommunication. N. Coupland, H. Giles 8r J. Wiemann, (Eds.) Miscommunication andproblematic talk. Newbury Park: Sage Publ., pp. ]03-120. Beebe, L. 8r Giles, H. (1984). Speech-accommodation theories: A discussion in terms of second language acquisition. International Journal ojthe Sociology of Language, 46, 5-32. Bejarano, Y. (1987). A cooperative small-group methodology in the language classroom. TESOL Quarterly, 21, 483-501. Berenst, J. (1994). Relationeel taalgebruik. Conversationele strategieën in interpersonele en interculturele communicatie. Doctoral Disserta[ion, Amsterdam: Thesis Publishers. Berger, J., Cohen, B. 8z Zelditch, M. Jr. (1972). Status characteristics and social interac[ion. American Sociological Review, 37, 241-255. Beun, R. (1990). The recognition of declarative questions in information dialogues. ITK Researrh Report No. 13, Tilburg, Netherlands: Institute for Language Technology and Arti ficial Intelligence. Bialystok, E. (1990). Communication strategies: A psychological analysis of second language use. Oxford: Blackwell. Bialystok, E. 8t Frohlich, M. (1980). Oral communication strategies for lexical di~culties. Interlanguage Studies Bulletin, 5, 3-30. 221 Bilmes, J. (1986). Discourse and behavior. New York: Plenum Press. Bingham Wesche, M. (1994). [nput and interaction in second language acquisition. C. Gallaway 8z B. Richards (Eds.) Input and interaction in language acquisition. Cambridge: Cambridge University Press, pp. 219-249. Bremer, K., Broeder, P., Roberts, C., Simonot, M. 8c Vasseur M.-T. (1988). Ways of achieving understanding: Communicating to learn in a second language. Final Repor[ Volume I the ESF project "Second language acquisition by adult immigrants", StrassbourglL.ondon. Bremer, K., Broeder, P., Roberts, C., Simonot, M. 8r Vasseur, M.-T. (1993). Ways of achieving understanding. C. Perdue (Ed.) The crosslinguistic study of second languages: t'olume ll The Results. Cambridge: Cambridge University Press, pp. 153-195. Broeder, P. (199 ]). Talking about people. A multiple case study on adult language acquisition. AmsterdamBerlin: Swets 8c Zeitlinger. Broeder, P. (1993). Leaming to understand in interethnic communication. lssues in Applied Linguistics, 4, 57-89. Brouwer, D. Bc De Haan, D. (Eds.) (1987). Women's language, socialization and self-image. Dordrecht: Foris Publications. Brown, P. 8c Levinson, S. (] 989I1978). Politeness: Some universals in language usage. Cambridge: Cambridge University Press. Brown, G. 8c Yule, G. (1983). Discourse analysis. Cambridge: Cambridge University Press. Bygate, M. (1988). Units of oral expression and language leaming in small group interaction. Applied Linguistics, 9, 59-82. Canale. M. 8c Swain, M. (1980). Theoretical bases of communicative approaches to second language testing. Applied Linguistics, 1, 1-47. Carbaugh, D. (1993). Cultural pragmatics. Reflections on some soviet and american encounters. R. Wiseman 8c J. Koester (Eds.) Intercultural communication competence. Newbury Park: Sage, pp. 168-183. Celce-Murcia, M. 8c Larsen-Freeman, D. (1983). The grammar book. An ESUEFL teacher's course. Rowley, Mass.: Newbury House. Chaudron, C. (1983). Simplification of input: Topic reinstatements and their effec[s on L2 Learners' Recognition and Recall. TESOL Quarterly, 17, 437-458. Chomsky, N. (1965). Aspects of tfee theory of syntax. Cambridge, Mass.: MIT Press. Clark, H. á Schaefer, E. (1989). Contributing to discourse. Cognitive Science, 13, 259-294. Clyne, M. (1977). Mul[ilingualism and pidginiza[ion in Aus[ralian indus[ry. Ethnic Studies, l, 40-55. Clyne, M. (1978). Some remarks on foreigner talk. N. Dittmar, H. Haberland, T. Skutnabb-Kangas 8z U. Teleman (Eds.) Papersfrom the first Scandinavian-German symposium on the language of immigrant workers and theirchildren. Linguistgruppen, Roskilde Universiteitscen[er, pp. 155-169. Clyne, M. (1994). Intercultural communication at work: Cultural values in discourse. Cambridge: Cambridge Uníversity Press. 222 List of References Coupland, N., Giles, H. 8c Wiemann, J. (1991). Miscommunication andproblematic talk. Newbury Park: Sage Publ. Crookes, G. 8z Rulon, K. (1985). Incorporation of corrective feedback in native speakerlnon-native speaker conversation. Technical Report, nr. 3, Centerfor Second Language Classroom Research, Manoa: Social Science Research ]nstitute, University of Hawaii a[ Manoa. Davies, A. (1989). Communicative competence as language use. Applied Linguistics, ]0, 157-170. Davis, K. (1988). Power under the microscope. Dordrecht: Foris Publications. Day, R., Chenoweth, A., Chun, A. 8t Luppescu, S. (1984). Corrective feedback in native-nonnative discourse. Language Learning, 34, 19-45. Deen, J. (1991). Comparing in[eraction in a coopera[ive learning and teacher-centered foreign language classroom. ITLReview ofApplied Linguistics, 93194, 153-181. Deen, J. (1991). [nteractie in interetnische communica[ie, Toegepaste Taalwetenschap in Artikelen, 39, 46-62. Deen, J. (1994). Negotiation of ineaning in institutional and informal conversations with non-native speakers. H. Purschel, E. Bartsch, P. Franklin, U. Schmitz 8c S. Vandermeeren (Eds.) Intercultural communication. Proceedings of the 17th international L.A. U.D. symposium Duisbeerg 23-27 March 1992. Frankfurt am Main: Peter Lang, pp. 159-184. Deen, J. 8c Van Hout, R. (1991). Clarification sequences in native-non-native interaction. R. van Hout 8c E. Huls (Eds.) Artikelen van de eerste sociolinguïstische conferentie, Delft: Eburon, pp. 121-138. Dietrich, R. (1982). Selbstkorrekturen. Fallstudien zum mundlichen Gebrauch des Deutschen als Fremdsprache durch Erwachsenen. Zeitschrift f'tir Literaturwissenschaft und Linguistik, 45, 120-149. Dirven, R. 8z Putz, M. (1994). Intercultural communication. H. Purschel, E. Bartsch, P. Franklin, U. Schmitz 8c S. Vandermeeren (Eds.) Intercultural communication. Proceedings of the 17th internationalL.A. U.D. symposium Duisburg 23-27 March 1992. Frankfurt am Main: Peter Lang, pp. 1-33. Ehrlich, S., Avery, P 8z Yorio, C. (1989). Discourse S[ructure and the Nego[iation of Comprehensible Input. Studies in Second LanguageAcquisition, 11, 397-414. El6s, R. (1985). Understanding second language acquisition. Oxford: Oxford University Press. Ensink, T. (1993). Vormen van miscommunicatie. Tijdschrift voor Taalbeheersing, 2, 93-108. Erickson, E(1976). Gate-keeping encounters: A sociai selection process. P. Sanday (Ed.) Anthropology and the public interest. New York: Academic Press. Faerch, C. Kasper, G. (1982). Phatic, metalingual and metacommunicative functions of discourse: Gambits and repairs. N. Enkvis[ (Ed.) Impromptu speech. Abo: Abo academi, pp. 71-103. Faerch, C. 8z Kasper, G. (1983a). Plans and stra[egies in foreign language communication. C. Faerch 8z G. Kasper (Eds.) Strategies for interlanguage communication. London: Longman, pp. 20-60. 223 Faerch, C. 8c Kasper, G. (1983b). On identifying communication strategies in in[erlanguage production. C. Faerch 8c G. Kasper (Eds.) Strategies for interlanguage communication. London: Longman, pp. 201-237. Faerch, C. 8t Kasper, G. (1985). Repair in learner-native speaker discourse. E. Glahn 8z A. Holmen (Eds.) Learner discourse. Anglica et Americana 22, Copenhagen: University of Copenhagen, pp. 11-23. Feldweg, H. (1991). The European Science Foundation second language databank. Nijmegen, Netherlands: Max Planck Institut fur Psycholingustik. Ferguson, C. (1975). Towards a characterization of English foreigner talk. Anthropological Linguistics. 17, 1-14. Gaskill, W. (1980). Correction in native speaker-nonnative speaker conversation. D. Larsen-Freeman (Ed.) Discourse analysis in second language research. Rowley. Mass: Newbury House, pp. 125-137. Gass, S. 8c Varonis, E. (1985a). Task variation and non-nativeJnon-native negotiation of meaning. S. Gass 8c C. Madden (Eds.) Input and second language acquisition. Rowley, Mass.: Newbury House, pp. 149-161. Gass, S. 8c Varonis, E. (1985b). Variation in native speaker speech modification [o non-native speakers. Studies in Second Language Acquisition 7, 37-57. Gass, S. 8z Varonis, E. (1986). Sex differences in nonnative speaker-nonnative speaker discourse. R. Day (Ed.) Talking to learn. Rowley, Mass.: Newbury House, pp. 327-351. Gass, S. 8c Varonis, E. (1991). Miscommunication in nonnative speaker discourse. N. Coupland, H. Giles 8z J. Wiemann (Eds.) Miscommunication andproblematic talk. Newbury Park: Sage, pp. 121-145. Geerts, G., Haeseryn, W., De Rooy, J. 8c Van den Toom, M. (1984). Algemene Nederlandse spraakkunst. Groningen: Wolters-Noordhoff. Giles, H., Mulac, A., Bradac, J. 8z Johnson, P(1987). Speech accommodation theory: The first decade and beyond. McLauglin, M. (Ed.) Communication yearbook 10. Beverly Hills: Sage, pp. 13-48. Glahn, E. (1985). Turntaking in conversations between leamers and native speakers. E. Glahn 8t A. Holmen (Eds.) Learner discourse. Anglica et Americana 22, Copenhagen: University of Copenhagen, pp. 24-41. Glaser, B. 8t Strauss, A. (1979). The discovery of grounded theory. Strategies for qualitative research. Chicago: Aldine. Gonzalez, Ch. (1989). Translation. M. Asante BZ W. Gudykunst (Eds.) Handbook of international and intercultural communication. Newbury Park: Sage, pp. 484-501. Gudykunst, W. 8c Nishida, T(1989). Theoretical perspectives for s[udying in[ercultural communication. M. Asante 8t W. Gudykunst (Eds.) Handbook of internationaland intercultural communication. Newbury Park: Sage, pp. 17-46. Gudykunst, W., Ting-Toomey, S., Hall, B. 8c Schmidt, K. (1989). Language and intergroup communication. M. Asante 8c W. Gudykunst (Eds.) Handbook of internationaland intercultural comrnunication. Newbury Park: Sage, pp. 145-162. Gumperz, J. (1982). Interethnic communication. J. Gumperz (Ed.) Discourse strategies. Cambridge: Cambridge University Press, pp. 173-186. Gumperz, L(1990). The conversational analysis of interethnic communication. R. Scarcella, E. Andersen 8c S. Krashen (Eds.) Developing communicative competence in a second language. New York: Newbury House, pp. 223-238. 224 List of Referencés Gumperz, J. (1991). Cross talk. Program from BBC Education and Training, Mosaic, BBC Education, London W52PA. Gunthner, S. (1989). Interkulturelle kommunikation und Fremdsprachen unterricht. Injonnation Deutsch als Fremdsprache, 4, 431-447. Hammer, M. (1989). Intercultural Communication Competence. M. Asante 8t W. Gudykunst (Eds.). Handbook of internatianaland intercultural cotrununicntion. Newbury Park: Sage, pp. 247-260. Harman, R. 8c Briggs, N. (1991). SIETAR survey: Perceived con[nbutions of the social sciences to intercultural communication. lnternational Journalof Intercultural Relations, 15, 19-28. Hatch, E. (1978) (Ed.). Second langtuige acquisition: A book of readings. Rowley, Mass: Newbury House. Hatch, E. (1983a). Simplified input and second language acquisition. R. Andersen (Ed.) Pidginization and creolization as lmtguage acquisition. Rowley, Mass.: Newbury House, pp. 64-86. Hatch, E. (1983b). Psycholinguistics. A second language perspective. Rowley, Mass.: Newbury House. Hawkins, B. (1987). Scaffolded classroom interaction in a language minority setting. Los Angeles: Report for the Center for language Education and Research, UCLA. Hinnenkamp, V(1989). Interaktlortale Soziolinguistiek und interkulturelle Kn~nmunikation. Gesprrichsmanagement zwischen Deutschen tutd Turken. Tubingen: Max Niemeyer Verlag. Hofstede, G. (1980). Culture's consequences: International difj`erences in work-related values. Beverly Hills CA: Sage. Holliday, L. (1988). Let them talk: A study of native-non-native interaction. Unpublished manusctipt, University of Pennsylvania, Philadelphia. Holmen, A. (1985). Distribution of roles in leamer-native speaker in[eraction. E. Glahn 8c A. Holmen (Eds.) Learner discourse. Anglica et Americana 22, Copenhagen: University of Copenhagen, pp. 70-89. Hou[koop, J. (1987). Establishing agreement: An analysis ofproposal acceptance sequences. Unpublished doctoral dissertation, University of Amsterdam. Hymes, D. (1971). Competence and performance in linguistic theory. R. Huxley 8z E. Ingram (Eds.) Language acquisition: Models and methods. London: Aeademic Press, PP- 3-28. Hymes, D. (1972). On communicative competence. J. Ptide 8z J. Holmes (Eds.) Sociolinguistics: Selected readings. Harmondsworth: Penguin, pp. 269-293. lakovidou, A. (1993). Funktion und t'ariation im 'ForeignerTalk'. Tiibingen: Gunter Narr Verlag. Jefferson, G. (1972). Side sequences. D. Sudnow (Ed.) Studies in social interaction. New York: Free Press. Kalin, M. (1995). Coping with probletns in understanding. Repairsequences in conversations between native ond non-native speakers. Unpublished doctoral dissertation, University of Jyv~skylíi, Finland. Kasper, G. (] 986) (Ed.). Learning, teaching and communication in the foreign language classroom. Aarhus, Aarhus University press. 225 Kelch, K. (1985). Modified input as an aid to comprehension. Studies in Second LanguageAcquisition, 7, 81-89. Klein, W. (1986). Second language acquisition. Cambridge: Cambridge University Press. Klein, W. 8t Perdue, C. (1988). Utterance structure. Strasbourg: Final Report Volume VI of the ESF project "Second language acquisition by adult immigrants". Knapp, K. 8t Knapp-Pothoff, A. (1987). Instead of an introduction: Conceptua] issues in analyzing intercultural communication. K. Knapp, W. Enninger 8c A. Knapp-Potthoff (Eds.) Analyzing intercultural communication. Berlin: Mou[on de Gruyter, pp. 1-15. Koole, T. 8c Ten Thije J. (1994). The construction of interrulturaldiscourse. Team discussions of educational advisers. Amsterdam, Atlanta GA: Rodopi. Krashen, S. (1981). Second language acquisition and second language learning. Oxford: Pergamon Press. Krashen, S. (1982). Principles andpractice in second language acquisition. Oxford: Pergamon. Larsen-Freeman, D. (1985). Sta[e of the art on input in second language acquisition. S. Gass á C. Madden (Eds.) Input and second language acquisition. Rowley, Mass.: Newbury House, pp. 433-444. Larsen-Freeman, D. 8c Long, M. (1991). An introduction to second language acquisition research. New York: Longman. Leet-Pelligrini, H. (1980). Conversation dominance as a function of gender and expertise. H. Giles, W. Robinson Bc P. Smith (Eds.) Language: Social psychological perspectives, New York: Pergamon, pp. 97-104. Levinson, S. (1983). Pragmatics. Cambridge: Cambridge University Press. Liberman, K. (1984). The hermeneutics of intercultural communication. Anthropological Linguistics, 26, 53-83. Liberman, K. (1990). An ethnomethodological agenda in the study of intercultural communication. D. Carbaugh (Ed.) Cultural communication and intercultural contact. Hillsdale, N.: Lawrence Erlbaum, pp. 185-192. Lindstrám, A. (1994). Identification and recognition in Swedish telephone conversations. Language in Society, 23, 231-252. Linell, P., Gustavsson, L. 8c Juvonen, P. (1988). In[eractional dominance in dyadic communication: A presentation of initiative response analysis. Linguistics, 26, 415-442. Long, M. (1981). Questions in foreigner talk discourse. Language Learning, 31, 135-157. Long, M. (1983a). Native speakerlnon-native speaker conversation and the negotiation of comprehensible input. Applied Linguistics, 4, 126-141. Long, M. (1983b). Linguistic and conversa[ional adjustments to non-native speakers. Studies in Second Language Acquisition 5, 177-93. Long, M. (1983c). Native speakerlnon-native speaker conversation in the classroom. M. Clarke 8c J. Handscombe (Eds.) Washington, DC, TESOL, pp. 207-225. Long, M. (1985). Input and second language acquisition theory. S. Gass 8c C. Madden (Eds.) Input and second languoge acquisition. Rowley, Mass.: Newbury House, pp. 377-393. Long, M. Bt Porter, P. (1985). Groupwork, interlanguage [alk and second language acquisition. TESOL Quarterly, 19, 207-228. 226 List of References Long. M. 8t Sato, C. (1983). Classroom foreigner talk discourse: Forms and functions of teacher's questions. H. Seliger 8c M. Long (Eds.) Classroom oriented research in second language acquisition. Rowley, Mass.: Newbury House. Markova, I. 8r Foppa, K. (1991). Asyntmetries in dialogue. Hempstead, England: Harvester Wheatsheaf; Savage, MD: Bames 8c Noble. Mazeland. H. (1992). Vraag~antwoord sequenties. Unpublished doctoral dissertation, Free University of Amsterdam, Amsterdam: Stichting Neerlandistiek VU. McLaughlin, M. (1984). Conversation: How talk is organized. Beverly Hills: Sage. Mcerman, J. (1988). Talking culture.~ Ethnography and conversation analysis. Philadelphia: University of Pennsylvania Press. Mulac, A. (1989). Men's and women's talk in same-gender and mixed-gender dyads: Power or polemic? Journal of Language and Social Psychology. 8, 247-270. Muller, F. (1991). Mariuccia, Mirella attd many more. The interaetive management of callers in an Italian radio phone-in program. Arbeitspapier nc 19, Konstanz: Fachgruppe Sprachwissenschaft, Universit~t Konstanz. Nofsinger, R. (1991). Everyday conversation. Newbury Park: Sage Publications. Parker, K. Bz Chaudron, G. (1987). The effects of linguistic simplication and elaborative modifica[ions on L2 comprehension. University of Hawaii: Working Papers in ESL, 6, 107-33. Perdue, C. (1984). Understanding, misunderstanding, and breakdown. C. Perdue (Ed.) Second language aequisition by adult immigrant.c. A field manual. Rowley, Mass.: Newbury House, pp. 68-97. Perdue, C. (1993a) (Ed.). Adult language acyuisition: Cross-linguistic perspectives. Volume l: Field methods. Cambridge Universi[y Press: Cambridge. Perdue, C. (1993b) (Ed.). Adult lnnguage acquisition: Cross-linguistic perspectives. Volume 11: The results. Cambridge University Press: Cambridge. Peters, V., Wester, R 8t. Richardson, R. (1989). Kwalitatieve analyse in de praktijk en handleiding bij Kwalitan. Nijmegen: Vakgroep Methoden, Fakulteit Sociale Wetenschappen, Katholieke Universiteit Nijmegen. 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. (1994). Review Article. Research on negotiation: What does it reveal about second-language learning conditions, processes, and outcomes? Language Learning, 44,493-527. Pica, T. 8c Doughty, C. (1985). Input attd interaction in the communicative classroom: Comparison of teacher-fronted and group activities. S. Gass 8c C. Madden (Eds.) Input and second language acquisition. Rowley, Mass.: Newbury House, pp. 1 IS-132. Pica, T. 8c Doughty, C. (1988). Variations in classroom in[eraction as a function of participant pattern and task. J. Fine (Ed.) Second Language Discourse, Norwoord, New Jersey: Ablex, pp. 41-55. Pica, T., Holliday, L., Lewis, N. Bt Morgenthaler, L. (1989). Comprehensible output as an outcome of linguistic demands on the learner. Studies in Second Language Acquisition, 11, 63-90. 227 Poulisse, N. (1989). The use of compensatory strategies by Dutch learners of English. Doctoral dissertation, Katholieke Universiteit Nijmegen, Enschede: Sneldruk. Pouw, A. (1990). Waardeoordelen in gesprekken. Een pragmatische studie van het Frans gesproken door Nederlanders. Unpublished Doctoral Dissertation, Rijksuniversitei[ Groningen. Quirk, R. 8c Greenbaum, S. (1974). A universiry grammar of English. London: Longman. Rie[veld, T. á Van Hout, R. (1993). Statistical techniques in the study of language and language behaviour. Berlin: Mouton de Gruyter. Roelandt, Th., Roijen, J. 8c Veenman, J. (1992). Minderheden in Nederland. Statistisch vademecum 1992. Den Haag: SDU UitgeverijlCBS Publicaties. Ros[, M. (1989). Sprechstrategien in ' freien Konversation". Tubingen: Gunther Narr Verlag. Sacks, H., Schegloff, E. 8t Jefferson, G. (1974). A simplest systematics for the organization of turn-taking in conversation. Language, 50, 696-735. Schegloff, E. (1972). Sequencing in conversational openings. J. Gumperz 8c D. Hymes (Eds.) Directions in sociolinguistics. The ethnography of communication. New York etc.: Holt, Rinehart 8c Winston, pp. 346-380. Schegloff, E. Bc Sacks, H. (1973). Opening up closings. Semiotica, 7, 289-327. Schegloff, E., Jefferson, G. 8c Sacks, H. (1977). The preference for self-correction in the organization of repair in conversation. Language, 53, 361-382. Scollon, R. 8c Scollon, S. (1980). Arhabaskan-English interethnic communication. Center for Cross-cultural Studies, University of Alaska, Fairbank. Scollon, R. 8t Scollon, S. (1995). Intercultural communicatlon. A discourse approach. Oxford: Blackwell. Schwartz, J. (1980). The negotiation for meaning: Repair in conversations between second language leamers of English. D. Larsen-Freeman (Ed.) Discourse anal}~sis in second language acquisition. Rowley Mass.: Newbury House, pp. 138-153. Selting, M. (1987). Verstándigungsproblemen: Eine empirische Analyse am Beispiel der Burger-Uerwaltungs-Kommunikation. Tubingen: Niemeyer. Shehadeh, A. (1991). Comprehension andperfonnance in second language acquisition: A.rtudy of second language learner'sproduction ojmodified comprehensible output. Unpublished doctoral dissertation, School of English, University of Durham, England. Shuy, R. (1983). Three types of interference to an effective exchange of information in the medical interview. S. Fisher 8t A. Todd (Eds.) The social organization of doctor-patient corruneenication. Washington: Center for Applied Linguistics, pp. 189-202. Slobin, D. (1993). Adult language acquisition: A view from child language study. C. Perdue (Ed.) Adult language acquisition: Cross-linguistic perspectives. Yolume 11: The results. Cambridge: Cambridge University Press, pp. 239-252. Snow, C. (1976). T~e language of the mother-child rela[ionship. S. Rogers (Ed.) They don't speak our [anguage. Essays on the language world of children and adolescents. London: Edward Amold, pp. 63-79. Snow, C., Van Eeden, R., Muysken, P. (1981). The interactional origins of foreigner talk. Journalof the Sociology of Language, 28, 81-91. 228 List ofReferences Stalpers, J. (1993). Progress in discourse. The impact ofjoreign language use on business talk. Unpublished doc[oral dissertation, Tilburg University. Stern, H. (1983). Fundamental concepts of language teaching. Oxford: Oxford University Press. Swain, M. (1984). Large-scale communicative language testing. A case study. S. Savignon 8c M. Bems (Eds.) Initiatives in communicative language teaching: A book of readings. Reading, Mass.: Addison Wesley. Swain, M. (1985). Communicative competence: Some roles of comprehensible input and comprehensible output in its development. S. Gass 8r. C. Madden (Eds.) Input and second language acquisition. Rowley, Mass.: Newbury House, pp. 235-256. Tarone, E. (1980). Communication strategies, foreigner talk, and repair in interlanguage. Language Learning, 30, 417-431. Taylor, T. (1986). Do you understand? Criteria of understanding in verbal in[eraction. Language Bc Communication, 6, 171-180. Taylor, D. (1988). Meaning and use of the term 'competence' in linguistics and applied linguistics. Applied Linguistics, 9, 148-168. Thorne, B. 8c Henley, N. (1975). Difference and dominance: An overview of language, gender, and society. B. Thorne 8r N. Henley (Eds.) Language and sex: Difference and dominance. Rowley, MA: Newbury House, pp. 5-42. Van den Branden, K. (1992). Een onzekere zekerheid. De rol van betekenisonderhandeling bij tweedetaalverwerving. GrumrncJlTf, I, 211-224. Van den Branden, K. (1995). `Zijn er nog vragen?' Over betekenisonderhandeling in he[ taalonderwijs. E. Huls 8r J. Klatter-Folmer (Eds.) Artikelen van de tweede sociolinguistische conferentie. Delft: Eburon, pp. 99-109. Van de Craats-Oosterwold, I. (1994). "Hier is woont een van mijn vader vriend woont hier". Een meervoudige casestudie naarde verwerving van woordvolgorde en funetionele categorieën bij volwassen tweede-taalleerders. Unpublished Master Thesis, Tilburg University. Van Gussenhoven, C. (1992). Phonetic Representation Dutch. Journalof the International Phonetics Association, 22, 45-54. Van der Wijst, P. (1991). Beleefdheidsmarkeringen in Franse en Nederlandse verzoeken. R. van Hout 8c E. Huls (Eds.) Artikelen van de eerste sociolingu'lstische conferentie, Delft Eburon, 473-489. Varonis, E. 8t Gass, S. (1982). The comprehensibility of non-native speech. Studies in Second Language Acguisition, 4, 41-52. Varonis, E. 8c Gass, S. (1985a). Non-nativelnon-native conversations: A model for negotiation of ineaning. Applied Linguistics. 6, 71-90. Watzlawick, P., Beavin, J. ót Jackson, D. (1967). Pragmatics of human communication: A study of interactional patterns, pathologies, and paradoxes. New York: Norton á Co. Webster's Encyclopedic Unabridged Dictionmy of the English Language (1989). New York: Portland House. Wiemann, J. (1977). Explication and test of a model of communicative competence. Human Communieation Research, 3. 195-213. White, L. (1987). Against comprehensible input: The input hypothesis and the development of second-]anguage competence. Applied Linguistics, 8, 95-110. 229 Wiseman, R. 8c Kcester, L(1993). Intercultural communication competence. Newbury Park: Sage. Woken, M. 8z Swales, J. (1989). Expertise and authority in native-non-native conversations: The need for a variable account. S. Gass, C. Madden, D. Preston 8z L. Selinker (Eds.) Variation in second language acquisition. volume l.~ Discourse and pragmatics. Clevedon, Avon: Multilingual Matters, pp. 211-227. Young, R. 8c Milanovic, M. (1991). Discourse variation in oral proficiency interviews. Studies in Second Language Acquisition, 14, 403-424. Yule, G. 8c MacDonald. D. (1990). Resolving referential conflicts in L2 interaction: The effect of proficiency and in[eractive role. Language Learning, 40, 539-556. Yule, G. 8t Tarone, E. (1991). The other side of the page: Integrating the study of communication strategies and negotiated input in SLA. R. Philipson, E. Kellerman, L. Selinker, M. Sharwood Smith á M. Swain (Eds.) Foreignlsecond language pedagogy research. Clevedon: Multilingual Ma[[ers, pp. 162-171. Zimmerman, D. 8t West, C. (1975). Sex roles, interruptions and silences in conversa[ion. B. Thorne 8c N. Henley (Eds.) Language andsex: Difference and dominance, Rowley, Mass: Newbury House, pp. 105-129. Zuengler, J. (1987). Effect of perceived 'expertise' in interactions between native and non-native speakers. Language 8c Communicntion, 7, ]23-138. Zuengler, J. (1989). Performance variation in NS-NNS interactions: Ethnolinguistic difference, or discourse domain? S. Gass, C. Madden, D. Preston 8c L. Selinker (Eds.) t'ariation in second language acquisition. l~olume I: Discourse andpragmatics. Clevedon, Avon: Multilingual Matters, pp. 228-244. Zuengler, J. 8r Bent, B. (1991). Relative knowledge of content domain: An influence on native-non-native conversations. Applied Linguistics, 12. 397-415. 230 List of References Appendix 1 NS NNS SLA L2 L1 ESF List of Abbreviations - native speaker - nonnative speaker - second language acquisition - secondlanguage - iirst language - European Science Foundation The negotiation moves ts - trouble source mr - main line reaction tr - thematic reaction es - escape tc - trouble clarification cc - confirmation check ti - trouble indicator co - confirmation gc - gratuitous concunence ac - acknowledgement cm - comprehension check mc - meta comment 231 Appendix 2 Transcription Conventions Almost all the transcription conventions described below were used for all ESF transcripts ( see Perdue, 1993a: 214-225). t t3t I ` ~xxx~ [xxx] [t] (xxx) ? x" . , (...) - Unfilled pauses are indicated in the text by plusses. - 3 seconds pause. - Self-interruption is marked with a slash. - Speaker shift with interruption is marked by a backslash. - Angle brackets are used to mark scope of various phenomena. The nature of the phenomena are explained in the comment lines following between angle brackets. - Overlapping utterances are signalled by enclosing the simultaneous parts of the speakers' utterances in brackets. - with a simultaneous utterance of the other speaker should be interpreted as a hesitation pause. - Inaudible stretches are indicated between parentheses. Attempted reconstructions of what has been said may also be given between parentheses. - Utterances interpreted as questions by the transcriber end with a question mark. - Rising intonation on following syllable. - Syllable lengthening. - Word stress is marked by a semi-colon in front of the stressed word. - Omission Non-target language material is marked with an equal sign followed by a one letter code indicating the non-target language spoken: cm-xxx~ - Moroccan Arabic ct-xxx~ - Turkish ~f-xxx~ - French ~p-xxx~ -'p' stands for non-target language pronunciation. In addition, an italic font is used in the examples to highlight a certain move or utterance. 232 Tables Appendix 3 Chapter 5 Informant E ti Cycle 1 Cycle 2 Cycle 3 ~o ~ .v d ~o ~ ~ o ~ sh ~ ' eu W` 'E [i ~o ~ ~ ~ ~ ~ ao W` ~ -- ~ o ri ~ ~ ~ ~ ~`~, ~. Totalturns 409 212 287 217 406 254 507 310 165 415 424 'h9 clari6cationsequencesturns 172 114 124 155 123 153 139 146 80 206 207 147 qo clarification seq. turns 42 54 43 60 27 47 48 50 49 55 71 30 Table 1: Percentage of clarification sequence turns per cycle in the informal conversations Informant Cycle 1 Cycle 2 Cycle 3 .v ~ ~o ~ v ~ s ~ ti ~ ~ ~ 00 tzl E ~~ ti s O ~ L ~ N ~ ~ 00 W '~ fi ~ O ~ t ~ cd ~ ~L ,-. Total tums 181 363 344 298 374 547 1294 477 547 305 902 S~4 clarification sequences turns 75 134 198 200 243 120 355 316 260 78 223 177 olo clarification seq. turns 41 37 58 22 27 26 25 3-3 67 65 66 48 Table 2: Percentage of clarification seguence turns per cycle in the institutional conversations Chapter 5 233 Chapter 6 Cycle I(Janet) Cycle 2(Janet) ~ Informant É cc~. ~ ó i ~ ~ ~ É ~ W [c`~. ~ NS tr. ind. 7 1 1 2 4 NNS tr. ind. 13 16 9 16 15 Ln NSINNS Cycle 3(lanetlPeteri ~ ó ~ ~ E ~ É ~ Ltl IL ~ ~ ~ ó .~ E ~ ~ - -~ [i~ 3 3 0 2 7 8 4 30 14 13 14 36 12 14 -0.59 -2.40 -1.85 -1.89 -1.24 -2.16 -1.42 -3.30 -1.76 -1.58 -0.39 -1.I7 NS tr. clarif. 17 25 38 32 17 37 20 23 9 38 39 19 NNS tr. clarif. 25 0 4 9 20 5 9 15 4 15 25 18 -0.38 3.93 2.15 1.92 0.77 0.42 0.75 0.91 0.44 0.05 NS conf.check 22 7 5 22 l5 6 17 13 l7 13 21 18 NNS conf. chk. 4 5 7 6 ]0 7 5 4 3 7 10 5 1.61 0.31 -0.31 1.24 0.39 -0.14 1.16 1.10 1.61 0.59 0.72 1.21 Ln NS~NNS Ln NSINNS 1.22 -0.16 Table .J: Logits of NS vs. NNS main negotiation moves in the informal conversations Cycle 1 ( Paul) Cycle 2 ( Bart) ~ Informant ~ ri ~ Cycle 3(Bart) ~ E ~ ~ ~ rit E ri ~ ó ~ ~ ~ ~ w` ~ ri ~ ~ i ~ U r.n NS tr. ind. 2 4 14 1 3 2 10 8 10 2 10 2 NNS tr. ind- 5 22 15 27 33 16 17 28 29 17 9 IS 0.10 -1.82 Ln NSRVNS - 0.79 -1.61 -0.07 -2.91 - 2.26 -1.89 - 0.51 -1.21 -1.03 -1.95 NS tr. clarif. 14 25 21 47 58 22 27 59 43 20 37 27 NNS tr. clarif. 1 6 36 3 3 3 65 12 22 4 18 7 2.27 1.37 - 0.53 2.61 2.82 1.86 -0.86 1.56 0.66 1.52 0.71 1.30 NS conf.check 11 14 38 18 39 I1 59 47 31 4 26 22 Ln NS~ItiS 1 53 3.27 0.-13 I?R I]? I.~,1 1-10 ? 02 (l51 Ln NSINNS NNS conf. chk. 2 2 4 9 ! 7h ' 2 15 , 0.67 1 7 16 15 7 1 3 t3 Table 4: Logits of NS vs. NNS main negotiation moves in the institu[ional conversations 234 Appendix 3 Tables Chapter 8 Cycle 1 2 3 Total Exact repetition 3 0 0 3 4.0 Partial repetition 2 5 3 10 13.3 Additive repetition 2 ] 5 8 ]0.7 Elaboration 13 6 7 26 34.7 Paraphrase 0 1 2 3 4.0 Correction 1 4 0 5 6.7 Translation 2 1 0 3 4.0 Subst.repetition 0 1 2 3 4.0 Complex repetition 3 4 7 14 18.4 Total 26 23 26 75 100.0 ; 90 Table 5: Fatima's trouble clariftcation n~pes per c~~cle Cycle qo 1 2 3 Total Yeslno question 0 1 0 1 1.3 Declarative question 0 0 1 1 1.3 Full declarative 1I 6 8 25 33.3 Elliptic declarative ]3 13 17 43 57.3 Other 2 3 0 5 Total 1R -ix -13 109 ~ 6.7 ]00.0 Table 6: Forms of Fatima's trouble clarification per cycle Chapter 8 235 Cycle qo 1 2 3 Total Exact repetition 0 0 2 2 6.1 Partial repetition 2 3 6 1] 33.3 Additive repetition 1 2 0 3 9.l Elaboration 1 1 2 4 12.1 Correction 0 1 4 5 15.2 Translation 0 0 1 1 3.0 Subst. repetition 0 ] 0 1 3.0 Complex repetition 2 0 3 5 15.2 Other 0 0 1 1 3.0 Total 6 8 19 33 100.0 Table 7: Mohamed's trouble clarification types per cycle oIc 1 2 3 Total Wh-question 0 0 6 6 18.2 Or- question 0 0 1 I 3.0 Yeslno question 1 0 0 I 3.0 Elliptic question 0 0 1 1 3.0 Full declarative 3 3 2 8 24.2 Elliptic declarative 2 5 8 15 45.5 Other 0 0 1 1 Total 6 8 19 33 Cycle 3.0 100.0 Table 8: Forms of Mohamed's trouble clarifications per cycle 236 Appendix 3 Tables Cycle I 2 Exact repetition 0 2 Partial repetition 5 Additive repetition 4 Elaboration qo 3 Total 2 4 2.5 9 25 15.9 8 6 18 11.5 17 31 15 63 40.1 Paraphrase 0 2 0 2 1.3 Hypoth. form. 1 1 0 2 1.3 Correction 2 5 1 8 5.1 Translation 0 0 1 1 0.6 Subst. repetition 2 3 3 8 5.1 Complex repetition 9 ]0 6 25 15.9 Other 0 1 0 I 0.6 Total 40 74 43 157 11 ~ I ~ ! 100.0 Table 9: Mahnuet's trouble clarification types per cvcle Cycle oIo 1 2 3 Total Wh-question 1 I 1 3 1.9 Elliptic question 2 1 0 3 1.9 Full declarative 19 31 15 65 41.4 Elliptic declarative 18 40 24 82 52.2 Other 0 0 3 3 I 1.9 ~ Table 10: Fonns of Mahmut's trouble clarifications per cycle 237 Chapter 9 Cycle 1 2 3 Total Exact repetition 1 0 2 3 4.7 Partial repetition 0 7 3 10 15.6 Additive repetition 3 3 1 7 10.9 Elaboration 5 13 15 33 51.6 Correction 0 0 2 2 3.1 Translation 1 0 0 1 1.6 Subst. repetition 0 1 0 1 1.6 Complex repetition 2 3 2 7 10.9 Total 12 27 25 64 ~Io ] 00.0 Table 11: Ergiin's trouble clarification types percycle Cycle qo 1 2 3 Total Wh-question 0 0 1 1 1.6 Elliptic question 0 2 0 2 3.1 Full declarative 7 8 13 28 43.8 Elliptic declarative 3 16 11 30 46.9 Other 2 1 0 3 4.7 ~ ~~ , ~ , -; . „~~ ~ Table l2: Forms of Ergun's trouble clarifcations per cycle 238 Appendix 3 Tables Cycle 1 ~ 2 3 Total olo Full repetition 0 2 0 2 7.4 Partial repetition 1 3 0 4 14.8 Paraphrase 0 0 1 1 3.7 Hypoth. form. 4 5 5 14 Subst. repetition 0 0 1 1 3.7 Complex repetition 1 1 3 5 18.5 Total 6 11 10 27 100.0 I 51.9 Table 13: Farinu,'s confinnation check ty~es per cycle Cycle qo 1 2 3 Total Negative question 0 0 1 1 Elliptic question 3 5 8 16 59.3 Minimal question 2 2 1 5 18.5 Elliptic declarative 1 3 0 4 14.8 Other 0 1 0 1 3.7 3.7 Table 14: Forms of Falima's confinnarion checks per cyc[e 239 Cycle i ~ 2 3 Total Ful( repetition 2 ~ 0 0 2 6.9 Partial repetition 1 4 4 9 31.0 Additive repetition 0 0 2 2 6.5 Paraphrase 1 0 0 1 3.4 Hypoth. form. 2 6 1 9 31.0 Substit. repetition 0 1 0 1 3.4 Complex repetition 1 3 1 5 17.2 Total 7 14 8 29 ] 00.0 3 Total clo Table I5: Mohamed's confirmation check t~~es per cycle Cycle I 2 Negative question 0 1 0 I Elliptic yuestion 4 1I 7 22 75.9 Minimal question 1 1 0 2 6.9 Full declarative 1 0 0 1 3.4 Elliptic declarative 1 1 1 3 10.3 Total 7 14 8 29 100.0 I ~ ~lo 3.4 Tnble l6: Forms of Mohamed's confirmation checks per cy~cle 24n Appendix 3 Tables Cycle 1 ~ 2 ~ 3 Total qo Full repetition 1 2 1 4 8.9 Partial repetition 6 2 4 12 26.7 Additive repetition 0 0 1 1 2.2 Elaboration 0 0 1 1 2.2 Hypoth. form. 3 13 4 20 44.4 Subst. repetition 0 2 l 3 6.7 Complex repetition 1 2 0 3 6.7 Other 0 0 1 1 2.2 Total 11 21 13 45 ~ 100 Tabfe 17~ Mahmut's confirmation check types per cycle Cycle I 2 3 Total Yeslno question 0 0 1 1 2.2 Negative question 1 0 0 1 22 Elliptic question 5 7 4 16 35.6 Minimal declarative 1 8 3 12 26.7 Full declarative 0 0 l 1 Elliptic declarative 4 6 4 14 Total I1 21 13 45 qo 2.2 31.1 ]00 Table I8: Forms of Mahmut's confirnuztion checks per cycle 241 Cycle Qlo I 2 3 Total Full repetition 2 1 1 4 Partial repetition 4 3 7 14 Additive repetition 1 2 0 3 Hypoth. form. 6 7 4 17 Translation 0 1 0 1 1.9 Substit. repetition I 0 1 2 3.8 Complex repetition 1 5 5 11 21.2 Total 15 19 18 52 ~ 7.7 26.9 5.8 32.7 100 Table !9: Ergun's confirmation check types per c~~cle Cycle qc 1 2 3 Total Wh-question 1 0 1 2 3.8 Negative question 0 1 0 1 1.9 Elliptic question 8 9 13 30 57.7 Minimal question 2 4 2 8 15.4 Full declarative 0 0 1 1 1.9 Elliptic declarative 4 4 0 8 ] 5.4 Other 0 1 1 2 3.8 Total 15 19 18 52 100 Table 20: Fornu of Ergun's confirniation checks per cycle 242 Appendi.r 3 Tables Samenvatting Nederland heeft sinds de Tweede Wereldoorlog nieuwkomers ontvangen uit verschillende delen van de wereld, zoals de ex-koloniën en de landen rond de Middellandse Zee. Autochtone Nederlanders krijgen daardoor in toenemende mate te maken met landgenoten die een andere culturele achtergrond hebben en het Nederlands in mindere mate beheersen. Dit proefschrift behandelt de vraag hoe moedertaalsprekers en tweede-taalsprekers problemen proberen op te lossen als ze elkaar niet begrijpen. Het proces waarin sprekers met elkaar een begripsprobleem proberen op te lossen wordt ook wel betekenisonderhandeling genoemd. Betekenisonderhandeling leidt tot onderbrekingen van de hoofdlijn van de interactie. Men gaat als het ware op een zijspoor om het probleem op te lossen om daarna de rode draad van het gesprek weer op te pakken. Zulke `zijsporen' worden ophelderingssequenties genoemd. In zulke sequenties kunnen de sprekers verschillende gedragingen vertonen. Zij kunnen bijvoorbeeld aangeven dat er een probleem is door middel van een zogenaamde probleemindicator; zij kunnen een uiting die tot een probleem leidde herhalen om zo het probleem op te lossen door middel van een zogenaamde probleemopheldering; of zij kunnen controleren of ze de ander goed begrepen hebben door middel van een zogenaamd bevestigingsverzoek. Deze gedragingen worden onderhandelingszetten genoemd. In deze longitudinale meervoudige casestudie worden de volgende aspecten van ophelderingssequenties onderzocht: (i) de opbouw, (ii) het aandeel van betekenisonderhandeling in de interactie, (iii) de verdeling van de belangrijkste onderhandelingszetten over de eerste-taalspreker (T1-spreker) en tweede-taalspreker (T2-spreker) te weten, de probleemindicatoren, probleemophelderingen en bevestigingsverzoeken, en (iv) de strategietypen en linguïstische vormen van deze drie belangrijkste onderhandelingszetten. De gegevens voor de studie werden verzameld in het kader van het European Science Foundation Project 'Tweede-taalverwerving door volwassen immigranten'. Gesprekken in het Nederlands van twee Marokkanen (Fatima en Mohamed) en twee Turken (Mahmut en Ergun) met Nederlanders lagen ten grondslag aan de uitgevoerde analyses. De T2-sprekers hadden geen of nauwelijks formele instructie gehad in de Nederlandse taal. Het ging om twee typen gesprekken. Het eerste type gesprekken waren informele gesprekken tussen Nederlandse projectonderzoekers en T2-sprekers over culturele verschillen tussen Nederland enerzijds en Marokko en Turkije anderzijds. Het tweede type gesprekken waren institutionele gesprekken tussen een huisvestingsambtenaar en een T2-spreker. Deze gesprekken waren semi-authentiek. Ze volgden een bepaald script waarbij de T2-sprekers gevraagd werden de rol te spelen van een jonge verloofde die wil trouwen en daarom een huis zcekt. De eerste informele en institutionele gesprekken vonden plaats ongeveer één tot anderhalf jaar na de aankomst van de T2sprekers in Nederland. Deze gesprekken werden twee keer herhaald met een tussenperiode van negen maanden, zodat er van ieder type gesprek uiteindelijk drie vergelijkingsmomenten waren. De eerste onderzoeksvraag betrof de opbouw van ophelderingssequenties. Uit de analyse met behulp van een computerprogramma voor kwalitatieve analyse (Kwalitan) bleek dat twaalf soorten onderhandelingszetten kunnen worden onderscheiden in ophelderingssequenties. Deze zetten kunnen verdeeld worden in hoofdlijnzetten die de overgang markeren van de hoofdlijn van het onderwerp van de interactie en zijlijnzetten waarmee het begripsprobleem wordt onderhandeld. Een ophelderingssequentie begripsprableem Fatima: probleemindicator Bart: probleemapheldering Fatima: probleemopheldering Ban: bevestigingsverzoek bevestiging Fatima: afsluiting Bart: ja ook eh mijn zoon nou wil eh loop in die jardin waar loopt ie? in diejardinl buiten die~in die huis in de tuin? ja tuin ja Het bleek dat een ophelderingssequentie meestal opent met een hoofdlijnbeurt die het begripsprobleem bevat. Bijvoorbeeld Fatima antwoordt op de vraag waarom ze wil verhuizen mijn zoon nou wil eh loop in die jardin. Dit antwoord wordt gevolgd door een zijlijnzet van de betrokken huisvestingsambtenaar waarmee deze aangeeft dat er een begripsprobleem is. Dit is een probleemindicator (waar loopt ie?). De T2-spreker reageert op de probleemindicator met een probleem-opheldering (in die jardin, huiten die~in die huis). De T2-spreker manipuleert zo de inhoud en woorden van haar probleembeurt door middel van een herhaling met een toevceging. De T1-spreker is er nog niet zeker van dat hij Fatima goed heeft begrepen en controleert met een bevestigingsverzoek waarbij een model voor de bedoelde inhoud en vorm van de andere spreker geeft (in de tuin?). Met zo'n bevestigingsverzoek manipuleert de T1-spreker dus de inhoud en vormen van de problematische uiting van de T2-spreker. De T2-spreker reageert op een bevestigingsverzoek met een bevestiging of ontkenning (ja tuin). Het probleem lijkt nu opgelost, en de huisvestingsambtenaar sluit de ophelderingssequentie expliciet af met een feedbacksignaal ja. Daarna kan de hoofdlijn van de interactie weer worden opgepakt met bijvoorbeeld een volgende vraag en waar wil je ergens wonen? Soms volgt in plaats van of voor de afsluiting een hoofdlijnreactie die een reactie op de oorspronkelijke probleembeurt betreft. Dit is bijvoorbeeld een verlaat antwoord op een vraag die in eerste instantie niet begrepen werd. Of er volgt 244 Samenvatting een andere niet verwachte inhoudelijke reactie, die hier thematische reactie wordt genoemd. Deze twee zetten brengen het gesprek terug naar de hoofdlijn. Een spreker kan ook terug gaan naar de hoofdlijn door de onderhandeling af te breken en op een ander onderwerp over te gaan (ontsnapping). Er zijn nog een aantal zijlijnzetten die voorkomen als de betekenisonderhandeling niet zo voorspoedig verloopt als in het bovenstaande voorbeeld. Probleemophelderingen worden dan vaak gevolgd door een feedbacksignaal dat ambigue overeenstemming (ja) wordt genoemd, omdat het niet noodzakelijkerwijs instemming aangeeft maar kan betekenen dat verdere opheldering gewenst is. Daarnaast kunnen sprekers ook het problematische karakter van de communicatie expliciet maken door het gebruik van begripscontroles (begrijp je?) en metacommentaar (nog een keer proberen). Samenvattend blijken zijlijnzetten drie verschillende functies in de ophelderingssequenties te hebben: (i) ze geven feedback (ii) ze manipuleren de inhoud en vorm van een begripsprobleem, (iii) of ze verwijzen naar het problematische karakter van de communicatie. T1-sprekers bleken vaker de inhoud van het begripsprobleem te manipuleren wat een behoorlijke communicatieve competentie vereist, terwijl T2-sprekers meer feedback-elementen bleken te gebruiken. Dit leidde tot twee verschillende typen sequenties, aíhankelijk van wie het begripsprobleem veroorzaakte. De T1spreker deed het meeste oplossingswerk ongeacht wie het probleem veroorzaakte. Er lijkt dus sprake te zijn van asymmetrie in de rollen die de T1- en T2sprekers op zich nemen in betekenisonderhandeling. De tweede onderzoeksvraag betrof het aandeel van ophelderingssequenties in de interactie. Op het totaal aantal beurten werd het percentage beurten berekend dat deel uitmaakte van ophelderingssequenties. Dit percentage bleek te variëren van 20 tot 70qc. Het was relatief hoog vergeleken met ander onderzoek. Twee factoren hebben mogelijk geleid tot deze uitkomst: de communicatieve competentie van de T2-spreker en het interactietype. Het bleek dat deze twee factoren elkaar beïnvloedden. Naarmate de communicatieve competentie van de T2spreker toenam was er alleen een afname van betekenisonderhandeling te constateren in de institutionele gesprekken. De T2-sprekers leerden waarschijnlijk het scenario van een huisvestingsgesprek en de gebruikte vaktermen. Daardoor konden problemen sneller worden opgelost. In de informele gesprekken was het aandeel van de betekenisonderhandeling stabiel rond SOolo. Dit kan betekenen dat in dit soort gesprekken nieuwe onderwerpen steeds nieuwe problemen creëerden omdat beide sprekers weinig verwachtingen hadden over wat de ander zou vertellen. Voortbouwend op de gevonden verschillen in structuur van de ophelderingssequenties, betrof de derde onderzceksvraag de mate van asymmetrie in de rollen van de Tl-spreker en T2-spreker in betekenisonderhandeling. De literatuur wees niet op automatische dominantie van TI-sprekers over T2-sprekers. Daarom werd onderzocht wat de verdeling over de T1-sprekers en T2-sprekers was van de drie belangrijkste onderhandelingszetten: probleemindicatoren, probleemophelderingen en bevestigingsverzoeken. 245 Een analyse van het aandeel probleemindicatoren, probleemophelderingen en bevestigingsverzceken van beide typen sprekers bevestigde dat er asymmetrie was. De T2-sprekers indiceerden problemen het meest frequent, terwijl de T1-sprekers ze vaker ophelderden. De T1-sprekers bleken echter vooral veel meer bevestigingsverzoeken te gebruiken. Met zulke bevestigingsverzoeken kan een spreker de inhoud van de interactie sterk beïnvloeden omdat deze als het ware de T2spreker woorden in de mond leggen. Deze verzceken kunnen dus wijzen op dominantie van de T1-spreker. Er werd ook nagegaan of de communicatieve competentie van de T2-spreker of het interactietype verklarend was voor de gevonden asymmetrie. De invloed van beide factoren is niet erg groot. Als de communicatieve competentie van de T2-spreker toeneemt is er wel enige afname van asymmetrie in de rolverdeling voor het aangeven en ophelderen van problemen, maar ook in de laatste gesprekken is er nog asymmetrie. Ook interactietype had minder invloed dan verwacht. In de institutionele gesprekken was wel iets meer asymmetrie wat veroorzaakt kan zijn door de institutionele rol van de huisvestingsambtenaar. De vierde onderzoeksvraag betrof de drie belangrijkste onderhandelingszetten, probleemindicatoren, probleemophelderingen en bevestigingsverzoeken, die vervolgens in meer detail zijn bestudeerd naar hun strategietypen en linguïstische vorm. De probleemindicatoren bleken op drie dimensies tegen elkaar te kunnen worden afgezet. Een probleemindicator kan of verbaal zijn of nonverbaal (bijv. stiltes na een vraag, het fronsen van de wenkbrauwen). Een spreker kan specifiek aangeven welk deel van de probleembeurt niet begrepen wordt of hijlzij kan alleen globaal aangeven dat er problemen zijn. Een spreker kan verder met een verbale indicator expliciet erkennen dat hijlzij iets niet begrijpt of kan dat impliciet houden. Dit levert vier typen verbale probleemindicatoren op: globaal 8z impliciet (hm?), globaal 8c expliciet (ik begrijp het niet), specifiek 8z impliciet (wie?) en specifiek 8r. expliciet (ik niet begrijp ploeg). De Tl-sprekers gebruikten alle typen indicatoren, maar hadden een voorkeur voor specifieke impliciete indicatoren, die duidelijk aangeven waar het probleem zit maar de T2-spreker niet in verlegenheid brengen. Ze gebruikten alleen waar nodig (bijvoorbeeld na een eerdere mislukte impliciete poging) expliciete indicatoren. De T2-sprekers gebruikten nonverbale indicatoren en globale verbale indicatoren in de vorm van minimale feedbackvragen het meest waarvoor weinig tot geen taalvaardigheid vereist is. Er is wel een tcename in specifieke impliciete indicatoren over tijd, wat voort kan komen uit een tcename van hun communicatieve competentie. Qua linguïstische vorm waren er grote verschillen tussen de T1- en T2-sprekers. De Tl-sprekers gebruikten veel vraagwoordvragen die een behoorlijke communicatieve competentie vereisten. Alleen de T1-sprekers gebruikten invulvragen (je woont in de .... ?) waarmee ze de linguïstische taak van antwoorden voor de T2-spreker verlichtten. De T2-sprekers daarentegen gebruikten soms verzoeken om woorden (beetje witte uh f kleren weet ik nier). Hierop reageerde de T1-spreker met het gezochte woord (trouw)jurk. 246 Samenvatting Het is meestal de spreker die een probleem veroorzaakt die ook het probleem probeert op te lossen door het gebruik van probleemophelderingen. Drie processen kunnen bij de manipulatie van de inhoud en vorm van de begripsprobleembeurt een rol spelen. Bijvoorbeeld Fatima begrijpt de volgende vraag niet: is hij afgekeurd?. De T1-spreker kan ten opzichte van het oorspronkelijke begripsprobleem elementen weglaten (deletie - af~ekeurd?), toevoegen (additie - is hij afgekeurd voor zijn werk?) of vervangen (substitutie - mag hij niet meer werken?). Deze processen van deletie, additie en substitutie kunnen worden tcegepast in combinatie met herhaling van (delen van) de begripsprobleembeurt. Bij deletie en volledige herhaling worden alleen al eerder gebruikte woorden herhaald. Dit worden conversationele aanpassingen genoemd. Bij additie en substitutie wordt ook nieuwe woorden toegevoegd. Dit worden linguïstische aanpassingen genoemd. Deze zijn moeilijker te begrijpen en te produceren dan conversationele aanpassingen. Zowel de T1- als de T2-sprekers hebben een sterke voorkeur voor probleemophelderingstypen met weinig linguïstische aanpassingen: Beiden kiezen dus vooral die middelen die gemakkelijk te produceren en te begrijpen zijn. Daarnaast gebruikten vooral de T1-sprekers ook sterke linguïstische aanpassingen, die moeilijker te realiseren zijn door T2-sprekers, zoals parafrase (mag hij niet meer werken?). In de informele gesprekken maakten de TI-sprekers soms ook gebruik van correctie en vertaling. Dit kan erop wijzen dat in deze context meer expliciet aandacht aan taalproblemen werd gegeven. Wat de linguïstische vorm van de probleemophelderingen betreft vallen een aantal zaken op. Als eerste valt op dat één van de T1-sprekers, de huisvestingsambtenaar, veelvuldig gebruik maakt van ongrammaticale aanpassingen ('foreigner talk' man niet werken?). Een verklaring hiervoor kan zijn dat hij in zijn werk met allochtone volwassenen deze manier van communicatie effectief heeft bevonden. Wat de T2-sprekers betreft valt op dat de vormen van probleemophelderingen die worden gebruikt worden vaak beperkt zijn tot woorden of zinsfragmenten en dat er nauwelijks ontwikkeling in is over tijd. De typen bevestigingsverzoeken konden worden beschreven met dezelfde classificatie als die voor de probleemophelderingen werd gebruikt. In tegenstelling tot probleemophelderingen betreffen bevestigingsverzoeken de inhoud van de probleembeurt van de andere spreker. Dit betekent vaak dat de spreker er vaak niet zeker van is of zijn~haar interpretatie juist is. Daarom werd het type hypothesevorming het meest frequent gebruikt door zowel de T1- als T2-sprekers. Dit type bevestigingsverzoek voorziet de andere spreker van een model voor de bedcelde maar niet 'correct' of duidelijk geformuleerde inhouden van de eerste spreker. Bijvoorbeeld de huisvestingsambtenaar begrijpt Mohamed niet als hij op de vraag wanneer hij gaat trouwen antwoordt: misschien uh vorige maand of uh zoiets. Hij vraagt daarom om bevestiging: je bent al getrouwd?. Deze hypothese bevat nieuwe linguïstisch elementen maar deze zijn echter voor de andere spreker toch vrij gemakkelijk te begrijpen, omdat ze in tegenstelling tot probleemophelderingen immers de inhoud van de ander betreffen. Naast hypothesevorming gebruikten de sprekers als ze zekerder waren van de bedoeling van de ander ook 247 vaak herhalingen van (onderdelen van) de probleembeurt. Bevestigingsverzceken hadden vaak de vorm van vragen waarop met 'ja' kon worden gereageerd. De resultaten hebben de volgende implicaties voor de theorie en praktijk van interculturele communicatie en tweede-taalverwervinglonderwijs. In de sociaalwetenschappelijke literatuur wordt interculturele communicatieve competentie hoofdzakelijk gedefinieerd als algemene interpersoonlijke competentie, zoals het vermogen je in de ander te kunnen verplaatsen of om problemen op te kunnen lossen. Linguïsten daarentegen zien communicatieve competentie als iets dat gebonden is aan de taal en culturele achtergrond van de sprekers. De T1- en T2sprekers die bij dit onderzoek waren betrokken gebruikten alle onderhandelingszetten, al is er, zoals verwacht, sprake van een beperktheid bij de T2-sprekers in variëteit en het gebruik van linguïstisch complexe vormen. Dat T1- en T2-sprekers dezelfde onderhandelingszetten gebruiken kan erop wijzen dat betekenisonderhandeling meer een beroep doet op een universeel probleemoplossingsvermogen dan op een taal- en cultuurspecifiek vermogen. Verder onderzoek naar betekenisonderhandeling met sprekers van verschillende eerste en tweede talen zou de juistheid van deze hypothese kunnen bevestigen. Ook zou uit verder onderzoek duidelijk kunnen worden of de verschillen tussen T1- en T2-sprekers in betekenisonderhandelingsgedrag vooral bepaald worden door verschillen in taalvaardigheid of doordat de sprekers oorspronkelijk verschillend gedrag hebben aangeleerd in hun eerste taal en cultuur. Wat tweede-taalverwervingstheorie betreft verdienen twee aspecten de aandacht: de stelling dat betekenisonderhandeling het leren van een tweede taal kan stimuleren en het onderscheid tussen conversationele en linguïstische aanpassingen. Betekenisonderhandeling kan taalverwerving stimuleren omdat probleemophelderingen en bevestigingsverzoeken van Tl-sprekers tot meer begrijpelijk taalaanbod lijken te leiden. Begrijpelijk taalaanbod is een voorwaarde is voor het leren van taal. Minder overtuigend zijn echter de resultaten wat betreft de hoeveelheid begrijpelijke taalproductie van T2-sprekers. Doordat T1-sprekers vaak bevestigingsverzoeken gebruiken in plaats van probleemindicatoren, gebruiken T2-sprekers minder frequent probleemophelderingen. Als T2-sprekers probleemophelderingen gebruiken zijn ze over het algemeen beperkt qua vorm. Het is de vraag of deze ophelderingen leiden tot uitbreiding van het linguïstisch repertoire van T2-sprekers. , Verder bleek het noodzakelijk de begrippen conversationele en linguïstische aanpassingen beter te definiëren. Deze begrippen die eerder werden toegepast op aanpassingen van Tl-sprekers, zijn in het huidige onderzoek ook toegepast op de aanpassingen van T2-sprekers. Eerder werd onder conversationele aanpassingen ook uitbreidingen (additie) gevat. In de huidige studie worden dit linguïstische aanpassingen genoemd omdat ze nieuwe linguïstisch elementen in de interactie brengen. Het lijkt bovendien de voorkeur te hebben om manipulaties van inhoud en vorm te classificeren op basis van de drie eerder besproken processen: deletie, additie en substitutie. Met deze processen in combinatie met herhaling van eerder gebruikte woorden of frasen konden, alle probleemophelderingen en bevestigingsverzoeken worden geclassificeerd. 248 Samenvatting Voor de praktijk van het tweede-taalonderwijs en interculturele communicatietraining kunnen enige voorlopige conclusies worden getrokken. Voor T2-sprekers is het van belang om middelen te hebben om problemen in communicatie aan te geven. Op deze wijze kunnen ze om meer begrijpelijk taalaanbod vragen, wat zowel het succes van de communicatie als de mogelijkheden tot taalverwerving kan verbeteren. Over het algemeen zijn globale 8z impliciete indicatoren (wat zegt u?) voldoende: deze zijn het minst gezichtsbedreigend voor zowel de T2- als T1-spreker. Daarnaast kan het gebruik van specifieke indicatoren (wat is `medische'?) effectief zijn omdat daarmee de interactiepartner aanwijzingen krijgt over wat hij precies moet ophelderen. In het onderwijs zouden T2-sprekers getraind kunnen worden in zulk betekenisonderhandelingsgedrag. Een aandachtspunt voor T1-sprekers die veel met T2-sprekers in contact komen, is vooral het risico van veelvuldig gebruik van bevestigingsverzoeken. Met deze verzoeken verlicht de T1-spreker de ophelderingstaak van de T2-spreker door te verwoorden wat deze kan hebben bedoeld. Hij neemt daarbij echter het risico dat de T2-spreker te gemakkelijk instemt met deze interpretatie. Dat er wellicht geen werkelijk onderling begrip tot stand is gekomen blijft onzichtbaar als de T1-spreker de ander niet vraagt in eigen woorden te formuleren wat hijlzij bedoelde. Als de T1-spreker de T2-spreker er regelmatig tce aanzet om zelf problemen op te helderen, is de kans op echt onderling begrip groter. Dit kan bovendien goed zijn voor de taalverwerving. Het huidige onderzcek had tenslotte een aantal beperkingen die in vervolgonderzoek kunnen worden ondervangen. Allereerst zou het wenselijk zijn processen van betekenisonderhandeling te onderzoeken in vergelijkbare interactie van alleen TI-sprekers. Dit kan leiden tot een beter inzicht in de invloed van beperkte taalvaardigheid op betekenisonderhandelingsgedrag. Ten tweede zou meer longitudinaal onderzoek kunnen uitwijzen of de beperkte uitkomsten op dit vlak vooral worden veroorzaakt door het feit dat de betekenisonderhandelingsgedrag in grote mate onafhankelijk is van taalvaardigheid en cultuur of dat het ontwerp van de huidige studie tot gebrek aan uitkomsten op dit gebied heeft geleid. Variabelen die in vervolgonderzcek onder controle zouden moeten worden gehouden zijn geslacht en expertise betreffende het gespreksonderwerp. Deze werden in het huidige onderzoek niet systematisch onderzocht maar lijken de resultaten wel te hebben beïnvloed. De invlced van culturele factoren zou voorts wellicht beter zichtbaar kunnen worden als tweetalige onderzoekers bij de analyse van resultaten worden betrokken. Ten derde zou dc opname van echte gesprekken (bijv. op een huisvestingsbureau) duidelijk kunnen maken wat de invloed van authenticiteit op betekenisonderhandelingsgedrag is. Ten vierde, wat de classificaties voor de onderhandelingszetten betreft. zou meer onderzcek met grotere grcepen informanten de betrouwbaarheid van de ontwikkelde classificaties kunnen vergroten. 249 Resume ~ Curriculum vitae Jeanine Deen was born on August 13, 1958 in Groningen. In 1976 she received her diploma Atheneum A. She received her M.A. (doctoraal) in Dutch language and literature at the University of Groningen in 1985. She specialized in Dutch as a second language and staned teaching Dutch to foreigners. Her minors were Applied Linguistics and Education in a Multicultural Society. She went to the U.S. to specialize further in second language acquisition and education, and in 1987 she received a second M.A. in Teaching English as a Second Language from the University of California, Los Angeles. After graduation she taught English as a Second Language at a community college and a school for basic education in Los Angeles, and conducted a research study for AFS Intercultural Programs. In 1989 she returned to the I~Tetherlands and accepted a research assistant position (Assistent in Opleiding) at Tilburg University of which this dissertation is the result. Since 1992, she has been an Assistant Professor at the Department of Language 8r. Communication of the University of Groningen. 250 Bibliotheek K. U. Brabant V'~~II ~ ~ 7 000 O ~ 24 ~ 037 O