|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
CALLing All Learners: An Explanatory Integrative Research Study of EFL Learner-Learner Corrective Feedback Patterns Within On-Line Synchronous Environments by Annmarie Gorenc Zoran A dissertation submitted in partial fulfilment of the requirement for the degree of Doctor of Philosophy Department of Secondary Education College of Education and Department of World Language Education College of Arts and Sciences University of South Florida Co-Major Professor: Antony K. Erben, Ph.D. Co-Major Professor: Carine Feyten, Ph.D. Anthony J. Onwuegbuzie, Ph.D. Ann Cranston-Gingras, Ph.D. Date of Approval: October 30th, 2006 Keywords: chat, peer feedback, error types special education nee ds, clarification request, elicitation, explicit correction, me talinguistic feedback, recast, repetition, education in Slovenia, middle school, high school Copyright 2006, Annmarie Gorenc Zoran
DEDICATION To delo posve am svojemu partnerju Marjanu, najboljemu prijatelju in moji sorodni dui ter svojima sinovoma Marku in Alexu. Brez vae neskon ne ljubezni, pomo i, podpore in nesebicnosti mi ne bi uspelo. Iz srca hvala tudi Francu in Silvi Zoran, ki sta vedno in iz srca nesebi no pomagala svojim otrokom. Ravno tako posve am to delo svojima stare ma, Marjanci in Petru Gorenc, ki vse svoje ivljenj e rtvujeta in delata za bol jo prihodnost svojih otrok. Neskon no sem vama hvalena. Nobena be seda ne more izraziti moje hvalenosti in iskrene ljubezni, zato preprosto HVALA! And to all the unheard children!
ACKNOWLEDGMENTS This dissertation would not have been possible without the assistance and support of many amazing and talented individuals. I def initely would not be here today, if it were not for Dr s. Feytens, Flaitzs and Zhus confidence in accepting me into the SLA/IT Ph.D. program. Thank you very much for believing in me and providing me with such an extraordinar y experience. Thanks to my committee member Dr. Ann Cranston-Gingras for offering her support, guidance, and compassion. Thank you for always being there and encouraging me to continue and push myself. Also, I would like to thank Dr. Tony Onwuegbuzie for introducing me to the rigorous nature of mixed method analysis. Thank you for taking the time to talk with me for endless hours on the phone, for the numerous constructive revisions, and for sho wing me alternative ways of analyzing problems. My utmost gratit ude goes out to my co-major professors Dr. Erben and Dr. Feyten. Thank you, Dr. Feyten for always being there for me and for being such an amazing professional. You truly insp ire me in more ways than you know! To Dr. Erben, thank you for helping me make sense of all the data and for guiding me towards the finish line. Your tireless pursuit of knowledge and being a voice for the unheard has inspired me to become a better person and educator. My sincerest gratitude goes to the SLA/IT Ph.D. students, cohort 2001, and graduates -for their support, dedicati on, and ambition. I want to sincerely
thank each and everyone of you. Each of you have motivated me to be a better educator and researcher. To my amazi ng friends Iona, Aline, and Chen: You have been with me through all the bad and good times, as colleagues and friends. It is difficult to de scribe in words the appreciation I have in you as friends. I will forever treasure our past, our present, and our future. I would like to express my gratitude to the participants in my study, to the dedicated teachers and princi pals who took the time to accommodate my needs. Thank you, Janja for all your help! My appreciation also goes out to the National Education Institute of Slovenia for assisting me with the research protocol and in securing relevant research sites. Last, but definitely not least to all my family in Sloveniatoo many to name, but all are in my t houghts and my heart! I would like to thank my family; to my mother-, father-, and sister-in-law, thank you for your support, love, endless baby-sitting and housekeeping duties. To my brother, his family, and my parents, thank you for showing me the world, exposing me to other languages, and helping me to be the person that I am t oday. I will forever treasure your kindness, your help, unconditional love, and determinat ion. My love will always be with you. Mami, thank you for this opportunity. To my husband, friend, and partner. No amount of words can express the amount of gratitude for y our endless support, consistently encouraging me to continue and sacrificing so much! To my inspirations Mark and Alex. You have given me the experience of unconditional love for which I will eternally treasure in my heart.
i Table of Contents LIST OF TABLES.................................................................................................vi LIST OF FI GURES .............................................................................................viii LIST OF A CRONYMS..........................................................................................ix ABSTRACT..........................................................................................................x CHAPTER 1 IN TRODUCT ION.............................................................................1 Statement of the Probl em..........................................................................1 Theoretical Fr amewor k..............................................................................5 Purpose of the St udy.................................................................................8 Quantitative Resear ch Ques tions ............................................................13 Hypothes es..............................................................................................13 Qualitative Resear ch Ques tions ..............................................................14 Educational Si gnifican ce..........................................................................15 Definition of Terms...................................................................................16 Limitati ons................................................................................................23 Delimitati ons............................................................................................24 Organization of the Re maining Ch apters.................................................25 CHAPTER 2 REVIEW OF LITERA TURE...........................................................26 Overvi ew..................................................................................................26 Theoretical Overvi ew...............................................................................26
ii Negotiation of Meani ng............................................................................29 Feedback .................................................................................................47 Feedback Studies Within Teac her-Learner Inte ractions...............53 Feedback Studies with Chil dren Intera ctions ................................64 Summary of F eedback Studi es.....................................................70 Computer-MediatedCommunica tion.......................................................71 Overvi ew.......................................................................................71 Criteria and Attributes for CALL Int egratio n..................................74 SLA and CMC Re search...............................................................79 Is Synchronous Discourse Writing or Speaki ng?..........................80 Discourse, Affective Factor s, and Language Pr oduction...............81 Summary of Language Development in Online Discussion Environm ents................................................................................88 Corrective Feedba ck and CM C.....................................................89 CALL, CMC and Corrective Feedback Su mmary........................100 Foreign Languages and Special N eeds................................................. 101 Historical Overview of the Cont ext.........................................................108 Education in Slovenia ............................................................................110 Foreign languages and Technol ogies in Sl ovenia .......................114 Chapter Su mmary..................................................................................121 CHAPTER 3 METH ODOLOGY ........................................................................123 Overview ................................................................................................123 Participants for Quant itative St udy......................................................... 125
iii Participants for Qua litative St udy........................................................... 132 Participants for Interview.............................................................133 Ethical Consi derations ...........................................................................134 Instrument s............................................................................................135 Questionnair e..............................................................................135 Qualitative Task Instrum ent........................................................ 136 Tool for Colle ction.......................................................................137 Qualitative Intervie w Instrum ent..................................................138 Pragmatist Pr ocedure ............................................................................139 Research De sign...................................................................................141 Data Collection Procedures ...................................................................145 Unit of A nalysis ...........................................................................153 Data Analysis Procedur es...................................................................... 158 Quantitative Analysi s Procedur es..........................................................166 Qualitative Analysis Procedures of Conversation Analysis....................174 Qualitative Analysis Procedures of Interview Protocol ...........................184 Summary...............................................................................................185 CHAPTER 4 DATA ANALYS IS AND RESU LTS..............................................187 Overview ................................................................................................187 Questionnaire Re sults.................................................................187 The database ..............................................................................190 Data Anal ysis.............................................................................. 192 Results ........................................................................................219
iv Results of Null Hypothesis 1.......................................................222 Results of Null Hypothesis 2.......................................................224 Results of Null Hypothesis 3.......................................................230 Results of Qualitat ive Analys is.................................................... 235 Chapter Su mmary..................................................................................264 CHAPTER 5 SUMMARY, DISCUSSION, RECOMMENDATIONS, AND IMPLICATION S...........................................................................265 Overview ................................................................................................265 Discussio n.............................................................................................265 Discussion of Resear ch Questi on 1............................................ 266 Discussion of Resear ch Questi on 2............................................ 268 Discussion of Resear ch Questi on 3............................................ 271 Discussion of Qualit ative Resu lts................................................273 Future Res earch.................................................................................... 280 Implications and Re commendati ons...................................................... 282 Limitati ons..............................................................................................285 Conclusi on.............................................................................................287 Referenc es....................................................................................................... 289 Appendix A The Structure of the Education System in Slovenia............328 Appendix B Foreign Language in G eneral Secondary Schools .............330 Appendix C Task................................................................................... 332 Appendix D IRB Approval for Pilot Study............................................... 334 Appendix E IRB Approval for Present Study.......................................... 336
v Appendix F IRB Course Completi on......................................................338 Appendix G IRB Core C ourse Comple tion............................................. 341 Appendix H Permission to Conduct Res earch.......................................343 Appendix I Student Backgr ound Questionna ire .................................... 345 Appendix J Codebook ............................................................................351 Appendix K Instructions to Participants in Slovene ................................356 Appendix L Instructions to Pa rticipants in English .................................358 Appendix M Corrective Feedback Codi ng..............................................360 About the Aut hor.....................................................................................E nd Page
vi LIST OF TABLES Table 1 : Negotiation of Fo rm Leading Toward s Repair.....................................58 Table 2 : Overview of Pa rticipants by Grade....................................................126 Table 3 : Overview of Dyad Members by Gr ade and Gender ...........................150 Table 4 : Research Questions wi th Data Analysi s Procedure..........................167 Table 5 : Total Errors by Type Across Grade Levels........................................201 Table 6 : Frequency and Percentage of Corrective Feedback Types by Grade Level..................................................................................................213 Table 7 : Percentage of Corrective F eedback and Learner Turns with Error by Grade Level .......................................................................................215 Table 8 : Frequencies of Corrective Feedback and Learner Error for All Grades ..........................................................................................................217 Table 9 : Mean and Standard Deviation for In cidence Variables as a Function of Grade Level .......................................................................................221 Table 10 : Corrective Feedback and NonFeedback Incidences by Grade Level .........................................................................................................223 Table 11 : Observed Frequencies and Percentages of Corrective Feedback Incidences by Ty pe and Grade ........................................................226 Table 12 : Mean and Standard Deviation for Corrective Feedback Incidences as a Function of Gr ade Level ................................................................229
vii Table 13 : Contingency Table of Obse rved Frequencies of Corrective Feedback and Learner Error Incidenc es for Gr ade 7.......................................231 Table 14 : Contingency Table of Obse rved Frequencies of Corrective Feedback and Learner Error Incidenc es for Gr ade 8.......................................232 Table 15 : Contingency Table of Obse rved Frequencies of Corrective Feedback and Learner Error Incidenc es for Gr ade 10..................................... 233 Table 16 : Contingency Table of Obse rved Frequencies of Corrective Feedback and Learner Error Incidenc es for Gr ade 11..................................... 234 Table 17 : Interv iew Them es............................................................................247
viii LIST OF FIGURES Figure 1 Varonis and Gass (1985) model of nonunder standing........................38 Figure 2. Varonis and Gass (1985) model of nonunderstanding including comprehension checks and interlocutor s stepping out of conversations. ............................................................................................................40 Figure 3 Lyster and Rantas (1997) error treatment sequenc e..........................56 Figure 4 Gap in the literatur e.............................................................................90 Figure 5 Selection of participant s....................................................................128 Figure 6. Research de sign............................................................................... 143 Figure 7 Procedures of t he study. ....................................................................146 Figure 8 Screen-shot c hat screen. ...................................................................148 Figure 9 Present error treat ment sequenc e.....................................................154 Figure 10 Coding proc ess...............................................................................156 Figure 11 Research implement ation proc ess..................................................165 Figure 12 Content of IRF routines ...................................................................181 Figure 13 Personal computer usage across grade le vels................................189 Figure 14 Corrective feedback frequencies per error type across grade levels. ..........................................................................................................218
ix LIST OF ACRONYMS CA. Conversation Analysis L1. First language L2. Second Language CACD. Computer Assisted Classroom Discussions CALL. Computer Assisted Language Learning CMC. Computer-Mediated Communication EFL. English as a Foreign Language ESL. English as a Second Language FL. Foreign Language LAN. Local Area Network NS. Native Speaker NNS. Non-native Speaker IRF. Initiation/response/follow-up SEN. Special Education Needs SLA. Second Language Acquisition TELL. Technology Enhanced Language Learning TL. Target Language
x CALLING ALL LEARNERS: AN EXPLAN ATORY INTEGRATIVE RESEARCH STUDY OF EFL LEARNER-LEARNER CORRECTIVE FEEDBACK PATTERNS WITHIN ON-LINE SYNCHRONOUS ENVIRONMENTS Annmarie Gorenc Zoran ABSTRACT This mixed methods research study centers on learner-learner interactions; thus, contributing to the on-going investigation within negotiation and interaction, computer-mediated-co mmunication and its role in second language learning. The specific aim was to investigate corrective feedback types, incidences, and the relationship betw een error and feedback type among peers within online synchronous env ironments in EFL classes in Slovenia, Europe. Interactional characteristics of co rrective feedback with learners having a documented special need (SN) also were explored using qualitative analyses. The study encompassed 208 students that were randomly placed into 104 dyads within intact classes of Grades 7, 8, 10, and 11. There were 32 dyads in Grade 7, 16 dyads in Grade 8, 24 dyads in Grade 10, and 32 dyads in Grade 11. Three participants had a documented specia l need. Quantitative analysis did not reveal statistical significant difference in the incidence of corrective feedback and grade level, the relationship among t he type of corrective feedback and grade level, or the relationship between learner error and type of corrective feedback
xi across grade levels. Corrective feedback types were similar to those studied in traditional classroom research (i.e., explic it corrections, recasts, negotiation of form). However, descriptive statis tics and qualitative analyses revealed conversational techniques that are s pecific to text-based online discourses providing insight into interactional c haracteristics among interactants within a discourse environment that differs both from speech an d written texts. Consequently, an additional corrective feedback type emerged from the data, coded as feedback request The most frequent corrective feedback type provided was explicit corrections. Frequency dat a revealed that corrective feedback tended to decrease as the grade level increa sed. Data with SN learners indicated distinctive discourse techniques. Overall, low incidences of correct ive feedback and error types might have been affected by the learners development al levels, social readiness, and/or psychological readiness (Oliver, 1998), as well as the learners individual conversational styles and socio-cultural fa ctors. Consequently, further research is warranted in examining thes e factors. In addition, longitudinal studies are warranted in examining whether online negotiat ed work lead towards L2 acquisition. Finally, the role of phantom corrective moves when coding qualitative online text data also need to be examined further.
1 CHAPTER 1: INTRODUCTION Statement of the Problem The question is not whether technology should be used, but how best to integrate technology on the basis of t heory and our current understandings of second language processes (Oxford, River a-Castillo, Feyten, & Nutta, 1997) for the benefit of all learners. Stemming from new advances in technology, the widespread use of the world-wide web, along with its new possibilities of including authentic information and incor porating new media of communication, has influenced the pedagogy and research of foreign l anguage classrooms. Thus, furthering foreign language methodol ogy into incorporating technologies for communicative based teachings. Within language learning classrooms new online communication media, among other factors, influence: (a) the nat ure of the discourse; (b) the affective influences, interactive competences, lingui stic output, and cognitive processes on language learning; and (c) the pedagogy of foreign language education (Beauvois, 1994; Beauvois & Eledge, 1996; Blake, 2000; Castaeda, 2005; Chapelle, 2001; Chun, 1994; Cubillos, 1998; Erben, 1999; Iwasaki, 2000; Negretti, 1999; Pelletieri, 2000; Warsc hauer, 1996, 1997; Warschauer & Healy, 1998). More importantly, communication methods such as online synchronous text based tools or chat rooms are on the rise within work place communications (Garcia & Jacobs, 1999; ONeill & Martin, 2003), being used as an informal
2 communication tools across methods am ong various generations for instant communication (Alvestrand, 2002), within fore ign language classrooms as a tool for foreign language instruction (Bradley & Lomicka, 2000; Bush, 1997; Cubillos, 1998; Egbert & Hanson-Smith, 1999; Ke lm, 1992, Kern, 1995; Morris, 2005; Warschauer, 1996; Wilson, 2000), and within foreign language teacher education (Egbert, Paulus, & Nakamichi, 2002); and has been investigated within second language acquisition (Beauvois, 1992, 1994, 1997; Blake, 2000; Castaeda, 2005; Chapelle, 1998, 2001; Chun, 1994; Chun & Plass, 1996; Erben, 1999; Gonzalez-Bueno, 1998; Gonzalez-Bueno & Perez, 2000; Gonzalez-Edfelt, 1990; Iwasaki, 2000; Iwasaki & Oliver, 2003; N egretti, 1999; Pellettieri, 2000; Sotillo, 2000; Sullivan & Pratt, 1996; Warschauer 1997; Warschauer & Healy, 1998). For learners with special education needs (SEN), who are also being mainstreamed into regular classrooms, the role and use of technology can be even more crucial, depending on the seve rity of their disability. Technology usage within classroom settings may be used as an assistive device (e.g., augmentative communication) or used as an educational tool. In either case, technology integration may assist learner s with special needs to become more active learners within mainstream classrooms. Noting the importance of technology usage within the United States for learner s with special needs, Hasselbring and Glass (2000) highlight: For example, use of computer technology for word processing, co mmunication, research, and multimedia projects can help the three million students with specific learning and emotional disorders keep up with their nondisabl ed peers (p. 102). Even though the
3 number of students reflect the population of learner s with special education needs in the United States, the applicabi lity of technology for learners with special needs also is related to other se ttings, more specifically, in Sloveniathe context of this study. Even though online-communication tools: (a) are readily available on the world-wide web, (b) influence the language learner in the classroom, and (c) are imperative in meeting the needs of l earners success in todays ever-growing technological world; the sole usage of technology without purpose is just a means in itself. To investigate the pr ocesses of second language learning using technology, one should base it on existent theory and research. Many areas of research have influenced the methodology of foreign language teaching as well as the research agenda within second l anguage acquisition. Theoretical and research advances in first language ac quisition, anthropology, sociology, linguistics, cognitive sciences, psyc hology, philosophy, and second language acquisition, have provided models, theorie s, and principles that represent our current understanding and know ledge of the underlying processes and factors influencing second language learning and te aching (Ellis, 1994; Johnson, 2004). As such, the field has progressed from the general notion t hat learners learn languages through imitation, stimulus/response, cognitive abilities and processes, interaction and feedback with other individuals, and/or as an active participant within their social environments. More specifically, researchers and linguis ts from both the fields of first and second language acquisition have investi gated the type of language input that
4 language learners receive and its influenc e on the quality and processes of language learning. Researchers have indi cated that feedback given on ill-formed utterances (i.e., negative feedback) withi n teacher-learner and learner-learner interactions leads learners to notice t heir gaps in knowledge and, in turn, to revise and construct their interlanguage (i .e., the stage through which a learner passes within language acquisition) into more target-like utterances (Gass & Varonis, 1994; Iwasaki & Oliver, 2003; Long, 1981, 1983, 1988, 1991, 1996; Lyster, 1998a; Lyster & Ranta, 1997; Morris, 2002; Oliver, 1995, 1998, 2000; Panova & Lyster, 2003). Negative feedba ck then can be used to hypothesize, notice, and/or confirm tar get language utterance, as such providing the learner the opportunity not only to notice their errors, but also the opportunity to reconstruct in a more correct manner, ther eby facilitating their language learning. Researchers within negative feedback have shown that learners are provided with feedback by native speakers (N S) in their roles as teachers (Gass & Varonis, 1994; Lyster, 1998a; Lyster & Ranta, 1997; Panova & Lyster, 2003), as NS interlocutors (Iwasaki & Oliv er, 2003; Oliver, 1995) or as non-native speakers (NNS) (Gass & Varonis, 1994; Mo rris, 2002; Oliver, 1995) as either NNS young adults (i.e., university student s) or as NNS children (Oliver, 1998, 2000). Also, feedback has been shown to be incorporated within subsequent conversational turns (Gass & Varonis, 1985; Oliver, 1995) and t hat negotiation of meaning or corrective feedback facilitates l earners to push their output into a modified, more target-like utterance.
5 As such, researchers within the inte ractionist field have argued that learners who receive negative feedback to t heir ill-targeted utterances have their language development facilitated and, as such benefit from these interactions. If research finds that within classroom interactions, negative feedback does indeed promote second language development, then it is worthy to investigate the interactional characteristics that lear ners have with teachers and other learners. Furthermore, the widespread use of online communication tools have been integrated with mainstream learners with or without spec ial educational needs, as well as in foreign language classes, where its efficacy and usage have been researched with second language acquisiti on. However, no research was found that examines corrective feedback as pr oficiency changes or research that includes mainstreamed learners with or without special educational needs within an online synchronous environment. Theoretical Framework The theoretical foundation for this investigation is based on the interlanguage theory and interactionist theory (Long, 1996; McLaughlin, 1987; Pienemann & Johnston, 1987) situat ed within studies of negative feedback (Lapkin, Hart, & Swain, 1991; Long & Robi nson, 1998; Lyster & Ranta, 1997; Mackey & Oliver, 2002; Schachter, 1991; Schmidt, 1993; Tomasello & Herron, 1988; White, 1991). Various cognitive theories within second language acquisition have examined the internal /mental processes of second language learning (L2); more specifically, on the L2 input learners receive and the cognitive processes that are entailed for coherent L2 linguistic output. One of the better-
6 known cognitive theories is the concept of interlanguage, coined by Selinker in 1972. In general, interlanguage represents certain stages that learners must pass through to achieve target-language competence (Lars en-Freeman & Long, 1991). Interlanguage is neither the first language (L1) or the target language (TL), but it is its own language. Within the interlanguage process, learners hypothesize about the rules of the L2. This is called hypothesis-testing, in other words, a learner forms her/his own hypothes is of the linguistic rules of the TL, and then based on linguistic input received, t he learner may accept or reject the linguistic hypothesis (McLaughlin, 1987). Lingui stic structures are accepted by the learner when the hypothesis has been confirmed or rejected if negative evidence (i.e., implicit or explicit correction) had been received (Ellis, 1994; McLaughlin, 1987). Following the progression of interlanguag e theory, there is evidence that learners progress through specific stages of acquisition. Early research based on research from Krashen (1977, 1981, 1985) and Dulay and Burt (1973, 1974, 1975) reveal that second language learner s have a natural order of acquisition, regardless of the learners L1. Morpheme acquisition studies (e.g., Dulay & Burt, 1975; Krashen, 1977, 1981) show that learner s first progress from the linguistic structure of progressive (i.e., continuous) ing, plural forms, the copula to be through the irregular past and progressive au xiliary towards the stage of article usage, regular past, third person singular s and possessive s endings (Krashen & Terrell, 1983). Furthermore, Pienemann (1984, 1989) and Pica (1983) have found that classroom instru ction does not seem to modify the
7 developmental sequences of acquisition orders. Based on this natural order of acquisition of linguistic structures hypothesis, the argument is that comprehensible input is necessary for the target language to be developed. However, researchers found that adjusted input of the target language is insufficient in itself (Swain, 1985, 1995). Stemming from knowledge on interlanguage development, the interacti onist theory has invoke[d] both innate and environmental factors to explain language learning (Larson-Freeman & Long, 1991, p. 266). Even though Pieneman n (1987, 1989) found that classroom instruction does not alter stages of progr ession, he also found that the pace and ultimate progression to the target languag e is influenced by formal instruction. Formal instruction is beneficial when the le arners interlanguage is prepared for a new linguistic structure that are mo rphosyntactively and cognitively more complex than previous struct ures learned. More specifically, when learners are prepared to accept more complex struct ures (i.e., the learnability hypothesis) then teaching (i.e., teachability hypothesis) is said to be a noticable variable (Pienemann, 1984). Thereby, teachability is dependent on the learnability stage of the language learner. Furthermore Pienemann and Johnston (1987) found that learners acquisition of grammatica l structures is explained by memory processing rather than grammatical comple xity. As learners progress through the developmental stages, they become more proficient; whereby more complex structures are integrated within their interlanguage. It is hypothesized in this study that as learners acqui re new linguistic structures at the same time as
8 proficiency (i.e., grade level) increases t hat additional feedback will be provided on more complex linguistic structures. Interactionist theorists also contend t hat structures can be acquired if they are noticed (Alanen, 1992; Lightbown & Spada, 1990; Long, 1991, 1996; Tomasello & Herron, 1989). More specifica lly, learners notice their gap in current target language knowledge by negative ev idence in context, whereby it is hypothesized in this study that correctiv e feedback types might differ based on learners awareness of their peers erroneous utterances .The focus of learners negotiating among each other while obtai ning negative feedback may assist with the achievement and pace of target languag e development within interlanguage. As mentioned earlier, the interactionist field, acknowledges both internal and external factors and furthermore indicates that negotiation promotes interlanguage development and that learner s are most likely to negotiate if opportunities are provided (Long, 1996). Mo re specifically, there is some evidence that there is a connection among conversation, negotiation, and interlanguage development (Long, 1996) As such, negative feedback and negotiations among interlocutors can be fact ors wherein learners notice their TL gaps (i.e., ill-formed structures) and com pare these TL-utterances with their own interlanguage processes (Tomasello & Herron, 1988). Purpose of the Study This dissertation stems from research findings in the fields of foreign language education, comput er-mediated-communication, and special education, as is more specifically extrapolat ed in the next chapter. As online
9 communications become a regular part of the language classroom, the field of second language acquisition and teaching is compelled to investigate how foreign language learners use online communication technologies and its usability as a teaching tool. Researc hers have discussed the ability to see learning in progress within online syn chronous environments (Beauvois, 1992; Kelm, 1992; Kern, 1995). In addition, some indicati on exists that within synchronous discussions learners report less anxiety, greater peer-to-peer participation, increased l anguage production and aware ness of their L2 errors, and utilization of a variety of discourse forms and structures (Beauvois, 1992; Chun, 1994; Gonzalez-Bueno, 1998; Gonz alez-Edflet, 1990; Johnston & Milne, 1995; Morris, 2005; Pellettieri, 2000; Sotillo, 2000). More specifically, synchronous discussions appear to be a fac ilitative tool for learners who are atrisk to fail (Beauvois, 1992). Most of the studies within compute r-mediated communication (CMC), as is reviewed in Chapter 2, have examined the interactions and benefits of computermediated communication within language lear ning. However, relatively few studies directly examine correct ive feedback within online synchronous environments and none to the researcher s knowledge has examined learnerlearner corrective feedback across grade le vels within an onlin e environment. Situated within the work of negotiati on and interaction in SLA, research has focused on both comprehensible input and output in terms of the occurrence and forms that lead to acquisition (Oliv er, 1995). As such, understandings and
10 research findings within the area of in teraction and negotiation of meaning within SLA highlight the following: Comprehensible input is necessary as an innate process triggering an internal process (Krashen, 1985; Schwartz, 1993). Comprehensible input makes target language forms more salient and, therefore, learners are more aware of them (Gass & Varonis, 1994; Long, 1996; Pica, 1994). Two features of interaction can lead to modified output: form-focused negotiation and negative (corrective) feedba ck (Long, 1996; Oliver, 1995; Swain & Lapkin, 1995, 1998; Tomasello & He rron, 1988; Varonis & Gass, 1985). Recent studies also have examined peer corrective feedback with adults in traditional face-to-face foreign languag e classrooms (e.g., Morris, 2002) and in English-as-a-Second-Language (ESL) cla ssrooms (e.g., Mackey, Oliver & Leeman, 2003). However, there has been very little research (Oliver, 2000) on feedback with children in traditional ESL classrooms (Mackey et al., 2003; Oliver, 1995, 2000) and foreign language classroom s. Most adolescent participants have been studied within immersion settings (e.g., Chaudron, 1977, 1986, 1988; Hamayan & Tucker, 1980; Lyster, 1998a, 1998b; Lyster & Ranta, 1997; Morris, 2005). However, Oliver (2000) stresses the importance of further research examining implicit negative feedback as it corresponds to the age of the learner. The need for further research within negotiation of meaning; more specifically, with corrective feedback, stem from the in teractionist framework, where such negotiations are facilitative and essential to second language
11 development (Long, 1996). Moreover, research within corrective feedback and the inclusion of special needs children are scant. Initial research is needed, where special need students are include d as participants with second language acquisition studies, especially with the onset of mainstreaming special need students. Furthermore, because technology is increasingly being integrated within foreign language lear ning, further research is merited on the usage of corrective feedback within a technological environment. Also, there is scant research with special needs students with respect to corrective feedback. Therefore, there is a n eed to explore the nature, frequency, and relationship of corrective feedback of EFL adolescents in online synchronous environments who may and may not have special learning needs Finally, there is no research found by the researcher that investigates whether corrective feedback differs based on proficiency--more specifically, the gr ade level of the foreign language (FL) learner. Therefore, the specific ai m of the present research was to: (a) investigate incidences of corrective feedback am ong EFL adolescent learners within an online synchronous environment, (b) ex amine the type of feedback, (c) investigate the relationship between error and feedback type, and (d) explore the interactional conversation characteristi cs of interlocutors in dyads when one or more of the learners hav e a documented special need. The present study was based on the underpinnings of the Intera ction Hypothesis ( Long, 1996) within interactionist theory. In addition, this in vestigation is build on Lyster and Rantas (1997) work on corrective feedback char acteristics and types with immersion
12 teachers and whether there are similar c haracteristics if the participant type differs (i.e., if learner-learner dyads al so provide similar types and amount of feedback as do teachers). The synchronou s mode or real time, as opposed to asynchronous or delayed-time, was chosen based on Olivers (1998) research showing that based on the nature of whol e class interactions, students had fewer occasions to respond to feedback when it was provided to them. Oliver (1998) further noted that because of the teacher s control over language production in the class, students also had fewer opportunities to risk-take. Childrens ability to risk-take, is a possibl e explanation for the larger in cidence of corrective feedback provided in learner-learner dyads (Mo rris, 2005). As such, the synchronous environment was chosen to provide oppor tunities for students to take risks without a teachers presence, and provide students with opportunities to respond to their peers feedback. It is important to note that within error correction and negative feedback research studies, the following terms t hat are similar in concept are used differently depending on the fi eld of study. When studying error correction from a linguistic perspective, the term negative evidence is used; within discourse analysis, the term repair is most common; psychologists use negative feedback ; the term focus on form is predominantly found withi n classroom second language acquisition research; and corrective feedback is the phrase used by second language teachers. As such, corrective feedback, instead of the aforementioned terms, was used throughout the present study.
13 Quantitative Research Questions The following research questions were addressed in the quantitative phase of the study: Research Question 1. What is the difference in the incidence of corrective feedback in online-synchronous envir onments provided by adolescent EFL learners to other dyad members as a function of grade level? Research Question 2. What is the relationship between type of corrective feedback in online-synchronous environments provided by EFL learners to other dyad members and grade level? Research Question 3. What is the relationship bet ween the type of learner errors and type of corrective feedback in online-synchronous environments provided by EFL learners to other dyad members and grade level? Hypotheses The following null hypotheses and nondirectional research hypotheses were tested: Null Hypothesis 1. There is no difference in t he incidence of corrective feedback in synchronous online environm ents provided by adolescent EFL learners to other dyad members as a function of grade level. Research Hypothesis 1. There is a difference in the incidence of corrective feedback in synchronous onlin e environments provided by adolescent EFL learners to other dyad member s as a function of grade level.
14 Null Hypothesis 2. There is no relationship between the type of corrective feedback in online synchronous envir onments provided by adolescent EFL learners to other dyad me mbers and grade level. Research Hypothesis 2. There is a relationship between the type of corrective feedback in online synchronous environments provided by adolescent EFL learners to other dy ad members and grade level. Null Hypothesis 3. There is no relationship betw een learner error and type of corrective feedback in online syn chronous environments provided by adolescent EFL learners to ot her dyad members and grade level. Research Hypothesis 3. There is a relationship between learner error and type of corrective feedback in online synchronous environments provided by adolescent EFL learners to ot her dyad members and grade levels. Qualitative Research Questions The following research question was addr essed in the qualitative phase of the study: Research Question 4. What interactional conversation characteristics by dyad members are present in onli ne-synchronous environments when one or more of the interlocutors are learners with special needs? Qualitative analysis was used for Research Question 4. The general question framed to guide the qualitative analysis was on the interactional characteristics of conversation when one or more of the interlocutors are learners with special needs. Interactional characte ristics of conversation are defined as the type of corrective feedback, error ty pes, responses to previous turns or
15 previous requests, questions, prompts, in vitations, and so forth. Qualitative analysis was used in order to make fe w assumptions about the nature of the participants and population (Fr aenkel & Wallen, 2003). As the research questions show, findings on learner uptake are not presented or analyzed. The impor tance of learner uptake is an important variable and will be reported in follo w-up studies. However, data to follow up on learner uptake were collected simultaneously. Educational Si gnificance It was hoped that the present rese arch would provide additional information on the nature of correcti ve feedback within learner-learner interactions in their second langua ge (L2) development within online synchronous environments. Additionally, it was hoped that the findings from this study would provide a better underst anding on the linguistic environments of various aged learners, the nature and impact of corrective feedback of mainstream learners wit h special learning needs, and ways to enhance student learning and differentiate instructi on through opportunities for feedback through online tasks. Lastly, it was hoped that this investigation would contribute to the research in second language acquisitio n in terms of examining corrective feedback with EFL participants within an educational and geographical setting that has not yet been included in t he literature on corrective feedback. In addition, there was also a met hodological significance within second language acquisition. As shown in the literat ure review, most of the studies under review were quantitative or descriptive in nature. However, there is a lack of
16 mixed methods studies in the area of corrective feedback and computermediated-communication. Markee (1994) argues that both quantitative and qualitative studies provide more balance and in-depth information to the study. As such, a sequential mixed design was used as the guiding framework for data collection and analysis of qualitative dat a (Tashakkori & Teddlie, 2003). This design was chosen because it reflects t he sequential nature of the quantitative and qualitative research questions and offe rs the opportunities for a broader understanding of the participant s, which is an important factor in pragmatism (Tashakkori & Teddlie, 2003). Definition of Terms Adjacency pairs. An adjacency pair is a unit of analysis within prototypical examples for conversation analysis. Adjac ency pairs are sequences of questions and answers as described by Sacks and Schegloff (1973). Adjacency pairs within this study were used to study the function of the language. Asynchronous. Asynchronous communication is a type of communication that occurs with a time delay (Chapel le, 2001; Warschauer, 1999). Interaction among participants is not in real time and allows interlocutors to respond with a delay. Examples include emails, bul letin boards, and discussion boards. Clarification request. This is one of the corrective feedback types or a negotiation move that provides evidence that the utterance was nontarget-like and that a reformulation is required (Lyster & Ranta, 1997). Exam ples include: I dont understand, What?, What did you mean? For the purpose of this study,
17 clarification requests in response to errors defined within the codebook (Appendix J) rather than c ontent form were examined. Computer-assisted language learning. Computer Assisted Language Learning (CALL) is an area of inquiry within Second Language Acquisition. It examines computer facilitation of language learning based on theories and principles from SLA and other fields (Chapelle, 1998). Computer-assisted language learning includes technology such as software, CDs, DVDs, Internet, chat rooms, word processing pr ograms, and web site building. Computer-mediated communication. A field of inquiry in computerassisted language learning (CALL) and Technology Enhanced Language Learning (TELL) is computer-mediated co mmunication (CMC), which examines computer usage with human interact ion (Blake, 2000; Warschauer, 1997). Computer-mediated communication includes asynchronous (e.g., bulletin boards, discussion boards, email) and synchronous (e .g., chat, video conferencing, audio conferencing) interactions. Conversation analysis. Conversation Analysis (CA) is a method used to examine conversational st ructure and the practices us ed among interlocutors for achieving comprehensible communication (Heritage & Atkinson, 1984; Markee, 2000). Within-CA sequences of adjacency pairs and initiation/response/follow-up structures were determined. Corrective feedback. Corrective feedback is a term used to indicate error correction studies by second language t eachers. More specifically, for the purposes of this study, the term corrective feedback is defined as feedback
18 moves that are provided by learner-learner interactions or corrective feedback to the dyad members errors. In this study specific corrective feedback categories include: explicit correction, recasts, elicitation, meta linguistic feedback, clarification requests, repetit ion, multiple, and emergent. Elicitation. Elicitation is a form of correct ive feedback that brings out the correct form from the inte rlocutor who created a nontar get-like utterance (Lyster & Ranta, 1997). Elicitation can take the form of leaving a blank for the interlocutor to complete, using questions, or aski ng to reformulate the nontarget-like utterance. Examples include: This is an .., How do you say in English? Can you please repeat what you just said? Error. For the purpose of this study, an e rror is defined as a non-target (illformed) utterance that is unacceptabl e in the target language. This study considered the following errors: gra mmatical, lexical, orthographical, typographical and spelling, and unsolicit ed use of the first language (L1). Error treatment sequence. The error treatment sequence in this study is the initial learners (P1) ill-formed utte rance, with corrective feedback provided by the interlocutor (P2) and t he initial learners (P1) re sponse to the feedback. The error treatment sequence was us ed as the unit of analysis. Grammatical error. A grammatical error is a type of error that violates the grammar of the target language. Interactional feedback. Interactional feedback are negotiated interactions among interlocuters. Intera ctional feedback is referred by researchers (e.g., Mackey, 2000) as recasts and negotiation moves.
19 IRF sequence. The initiation/response/fo llow-up (IRF) sequence (Mehan, 1985; Ohta, 1993, 1994, 2001; Sinclair & Coulthard, 1975) was used as an additional unit of analysis withi n conversation analysis. The initiation turn can be a question or a statement and includes an error, the response is an immediate turn to the initiation and considered as f eedback to the error in the initiation turn, and the follow-up is praise from the teacher and/ or repair of the error in the initiation turn based on the feedback in the response turn. L1. In the field of Second Language Acqui sition, L1 is the first language of a second language (non-native) speaker. L2. In the field of Second Language Acquisition, L2 is the second language of the non-native speaker. Lexical Error. Lexical error is a type of erro r that uses the incorrect word (vocabulary unit) in the utterance (Cas taeda, 2005; Morris, 2005). These lexical errors include inappropriate or inaccurate uses of structural derivations (i.e., nouns, verbs, adverbs, adjectives). Metalinguistic feedback. A metalinguistic feedback is an implicit response from the interlocutor that the utterance was nontarget-like in some form (Lyster & Ranta, 1997). Metalinguistic feedback can be seen in the form of meta-analysis of the error. Examples of metalinguistic feedback can be: Is that singular? and Can you find your error? Negative evidence. Negative evidence is a term used in the field of linguistics to indicate studies on e rror correction (Bohannon, MacWhinney, &
20 Snow, 1990; Krashen, 1985). Other term s include negative feedback, repair, corrective feedback, and focus on form. Negative feedback. Negative feedback is a term used by psychologists to indicate studies on error correction and feedback (Schachter, 1991). Other terms include negative evidence, repair, co rrective feedback, and focus on form. Negotiation moves. Negotiation moves is a term used for feedback types such as confirmation checks, clarificat ion requests, and repetition (Mackey, 1999; Mackey, Gass, & McDonough, 2000). Negotiation of form. Negotiation of form in cla ssroom instruction is focused on grammatical points rather than on th e meaning of content (Long, 1983, 1985, 1991). Orthographic errors. Orthographic errors represent omissions of letters unique to the English language. These include q, w, x, y. In addition errors may include additions of letters unique to the Slovenian alphabet, such as , and . Orthographic errors were combined within the typographical and spelling error category because it was difficult to plac e these errors into their own separate categories. Recast. Recast is a reformulation of all or part an ill-formed utterance, excluding the error (Long, Inagaki, & Ortega, 1998; Lyster & Ranta, 1997; Mackey & Philp, 1998). Recasts also have been referred to as paraphrase (Spada & Frhlich, 1995), repetition with change, and repetiti on with change and emphasis (Chaudron, 1977).
21 Repair. In the field of negotiation of meaning, repair refers to nonunderstanding that occurs and ends with a re solution of some sort or correction (Kasper, 1985) following some type of feedback. Repetition. Repetition is a type of corre ctive feedback where the peer repeats the nontarget-like u tterance created by the l earner (Lyster & Ranta, 1997). The repetition of the ill-formed uttera nce is in isolation usually with or without intonation. In a CMC environm ent, this can be denoted with a question mark, exclamation point, an emoticon, and so forth. Examples include, a children? and this horses? Special educational needs. Special educational needs (SEN) or learners with special needs are those students w ho need extra or different types of assistance due to emotional or behaviora l disturbances, physical impairments, chemical imbalances, and/or difficu lty understanding and developing higherthinking skills (Ministry of Education and Spor ts, 2000). It is more difficult for such students to learn or access appropriate education. The follo wing documented special needs were considered for inclus ion in this study: Trainable Mentally Handicapped, Speech Impair ed, Language Impaired, Deaf or Hard of Hearing, Visually Impaired, Emoti onally Impaired, Specific Learning Disabled, Profoundly Mentally Handicapped, Dual Sensory Impaired, Autistic, Severely Emotionally Handicapped, Traumatic Brain Injury, De velopmentally Delayed, and Educable Mentally Handicapped (Indivi duals with Disabilities Education Act [IDEA], 1997; Ministry of Education and Sports, 2000).
22 Synchronous environment. A synchronous environment is a real-time communication mode, wherein interlocutor s can meet anywhere and at the same time. In traditional senses, a telephone conversation can be considered real time; in a technology environment, chat and conferencing are considered real time. The present study utilized t he chat portion of the synchronous environment. Target language. The target language is the language that the person is learning, and does not include the pers ons first language. The first language of the participants in this study was Sloven e, and the target language was English. Turn. For the purpose of this study, a tu rn in the synchronous environment is considered when a message is composed and sent into the chat room either by clicking the send button or by pressing enter on the keyboard. Typographical and spelling error. A typographical error is a type of error that results in misspelled words because of keyboarding inexperience, rushing, not paying attention. A spelling erro r is one made when forming words with letters and the letters are not put in the correct order. Due to the ambiguous nature of typing and spelling errors, both of these forms were included under one category. Unsolicited use of L1. Unsolicited use of L1 is t he learners intentional or unintentionally usage of their native langu age (L1). Use of L1 was considered as a factor in this study to investigate responses by the dyad member to the learners use of L1 (e.g., causing both dyad members to shift to L1, both members redirecting to L2, or ignor e the L1 and continue with the topic).
23 Limitations Both external and internal validity limited the findings of this study. Onwuegbuzies (2003) framework for possi ble external and internal validity threats to a study was used as a guide in this study. Possible threats to external validity included the following: (a) ecolog ical validity was a threat because the participants were limited to learners of English as a foreign language from a specific geographic area in Europe; (b) population validity was a threat because the sample sizes from the combined schools were relatively small; (c) temporal validity threatened external validity because of the limited time of data collection; and (d) reactive arrangements, the effect of participants reactions by being aware that they were partici pating in the study, could hav e influenced the validity of the findings. Several threats to internal validity of the findings were considered: (a) the amount of data might hav e generated responses that did not yield data saturation; (b) intact classes with learners that have a differential or too similar of a range of proficiency was anothe r threat to validity; (c) researcher bias also was a threat that because certain cat egories might have been constructed or collapsed based on personal beliefs of the researcher (i.e., illusory correlation); (d) time constraints was a threat bec ause there was only one collection time used for analysis; however, more participant s were chosen from various schools to somewhat alleviate this limitation; and (e) instrumentation was a threat pertaining to the reliability and validity of the coded data. To alleviate somewhat external and internal validity threats of the quantitative data, inter-rater and intra-
24 rater checks were performed, as well as peer debriefings and the completion of a questionnaire prior to data collection (Lin coln & Guba, 1985; Miles & Huberman, 1994). Finally, research validity in qualitative research was considered in terms of (a) descriptive validity, (b) interpretive validity, and (c) theoretical validity. To obtain descriptive validity, researcher tr iangulation was used. The researcher of the current study used both questionnaires as well as follow-up interviews with 5% of the participants, which inclu ded extreme points within the data set and special need learners. Also, field notes during dat a collection and data analysis were used throughout the process. Interpretive validity was achieved by accurately supplementing student accounts with a selection of direct quotes obtained through interviews. Fina lly, theoretical validity was obtained by including two other peers to review the data, interp retation, and conclusions of the study. Delimitations The delimitation of this mixed met hod study imposed by the researcher included the choice of which grade levels to study. For the purposes of this study Grade 7 was chosen initially because students already had approximately two years of EFL experience, thereby having some foreign language experience, at a beginner or upper-beginner le vel of English. Following grades were mainly chosen by students, and teachers av ailability and quantity of students. Therefore, Grade 7, 8, 10, and 11 were the final choices.
25 Organization of the Remaining Chapters The remainder of this dissertation in cludes the review of literature in Chapter 2 on interaction, feedback, co mputer-mediated communication, foreign language learning in Sloven ia, and foreign language learning with mainstream and special needs students. The dissertat ion then continues with Chapter 3 where the research design, procedure, instruments, data collection, and data analysis are described. The results are pr esented in Chapter 4. Finally, Chapter 5 provides a summary of the fi ndings, discussion, recommendations, and implications.
26 CHAPTER 2: REVIEW OF LITERATURE Overview Because this study examined the ty pes and distribution of corrective feedback between learner-learner interactions the literature review reflects the interdisciplinary nature of this study and combines it with a theoretical framework that guided this investigation. Thus, the first section describes research and main findings of studies on feedback and co mputer-mediated communication within second language acquisition. The second section discusses foreign language learning in Slovenia, as well, as t he description of special learning needs and inclusion within mainstream classrooms. A summary concludes this chapter. Theoretical Overview Research into the role of feedbac k and negotiation of meaning in SLA goes back more than 20 years, begi nning with Krashens (1982, 1985) arguments that natur al approaches can lead to mast ery of the target language. His works have resulted in many debates furthering the current knowledge of language acquisition. He cont ends that the subc onscious processes, the natural approach, along with comprehensible input, are factors that lead to acquisition. Krashen proposed the following five hy potheses on the phenomena of second language: (a) the acquisition/learning hypot hesis, (b) the monitor hypothesis, (c) the natural order hypothesis, (d) the affe ctive filter hypothesis, and (e) the input hypothesis (Krashen, 1985). In the ac quisition/learning hypothesis, Krashen
27 distinguishes between language as being ac quired (i.e., similar to first language acquisition) versus learned (i.e., classroom instruction). Kras hen argues that the conscious processes of language practice cannot cross over to the unconscious or the acquired language system. Speaker s utilize the learned or conscious process to focus on form (i.e ., grammatical structures), thereby monitoring their output. Learners who focus on meaning rather than on form develop their acquired (versus learned) linguistic system (Krashen, 1976, 1982, 1985), which is posited within the monitor hypothesis. T he natural order hypothesis states that there is a natural order of acquiring linguis tic structures that are not altered even with formal instructions. Furthermore, Krashen claimed that affective factors (e.g., anxiety, motivation, stress) were posited to influence second language acquisition. The affective filter hypothesis causes a filter to be raised (i.e., a mental block) when the affective fact ors are negative (e.g., higher anxiety), whereby linguistic input may not be comp rehensible to the lear ner. Or the filter may be lowered, which may be positive towards comprehensible input. Comprehensible input is the central clai m within the input hypothesis, wherein input that is received needs to be underst ood in order to be acquired. Krashen (1983) illustrates progress with the i + 1 structure, where l earners receive input that is one stage beyond their current level of second language development (i.e. interlanguage), which in returns pushes linguistic improvement. Krashens five hypotheses stem from Chomskys (1965) innatist view that language acquisition is a subconscious pr ocess and that language acquisition is based on an internal language device. Krashens hypotheses have been
28 universally accepted within the teac hing field; however, researchers have criticized his failure to explain a hy pothetical device, kn own as the language acquisition device (LAD) that allows people to acquire language innately (Chomsky, 1965) for second language l earners. Some researchers have contended also that these hypothes es are non-testable (Gregg, 1984; Pienemann & Johnston, 1987) and express concern about Krashens use of only anecdotal or introspective methods to obtain data (McLaughlin, 1978, 1987). In addition, Krashen has been criticized for hi s sole emphasis on comprehensible input (Long, 1991; Swain, 1985), wher eby comprehensible input, within Krashens framework, is t he language that is understand able to the learner by producing language that is less complex or simplified. However, Swain (1985) stated that not only is comprehensible input an important factor, but comprehensible out put, or the language produced by the L2 learner, should not be overlooked as a fa ctor in second language learning. In comprehensible output, learners notice a gap in their L2 production and remodify to produce target language i nput. Learners achieve comprehensible output by modifying and approximating their production eventually to produce successful target-like output (Swain, 1985) It has been further argued that when learners modify their output, the interl anguage utterances for greater message comprehensibility are rest ructured and affect the L2 learners knowledge base (Swain & Lapkin, 1995). Gass and Seli nker (1994) further highlighted the distinction between comprehensib le input and comprehended input. Comprehensible input is controlled by the person providing input and
29 comprehended input is controlled by the lear ner, wherein the learner is or is not undertaking all the work to understand t he intended message. In their model of second language acquisition, Gass and Seli nker include comprehended input to encompass the various levels of comprehension that exist, including both comprehension of st ructure and meaning. Based on the then current understanding of second language acquisition, the communicative approach to teaching had steadily received more widespread acceptance in the foreign language teaching field as a viable way to facilitate foreign language learning. This was bes t actualized through Canadian French immersion programs. In these immersi on programs, children learned to speak French fluently; however, it was found (Har ley & Swain, 1984; Swain, 1985) that the immersion learners accuracy in syntax and morphology was poor. An argument given was that the learners di d not have sufficient opportunities to speak nor to negotiate meaning. Various researchers have attempted to explain these phenomena. The following section reviews current understandings as well as reviewing literature on negotiation of meaning, and cont inuing with the role of feedback in second language acquisition. Negotiation of Meaning From the current research on comprehensible input, output, and interlanguage development, Pica, Holl iday, Lewis, and Morgenthaler (1989) argued that negotiation in terms of negotiation of input also is a mediating factor in language acquisition. Stevick (1976, 1980) also contended that to facilitate acquisition, there needs to be active involvement. Long (1996) furthered this and
30 updated his original Interactionist Hypothes is. Negotiation of meaning, according to the updated Interactionist Hypothesis is the negotiation of meaning between the learner and usually a more profici ent speaker of the language (Long, 1996). This type of negotiation is an important el ement in language acquisition in that learners, because of the overflow of in formation, focus on meaning rather than on form (Long, 1996). Lyster and Ranta (1997) also proposed, based on van Liers (1988) distinction of conversati on and didactic functions, that negotiation in L2 classrooms has two functions. The first function, classroom function, involves the negotiation of meaning, which has been an important component of immersion classrooms. The second functi on, didactic function, involves the negotiation of form, which includes not only comprehensibility of a message, but also the encouragement of self-re pair and feedback. Negotiation can be influenced by several examples such as the type of task, characteristics of participants, structure of participants (E llis, 1994), and context. Several research studies have examined negotiation from these perspectives, as will be shown below. Savignon (1972) examined the contex t of communicative classrooms (e.g., informal instruction) and the role of focusing on form or the grammatical structures (e.g., formal instruction) within college French language classes. In this study, students who received form-f ocused instruction were compared to students who received form focused plus an additional hour of communicative tasks. The results revealed that the st udents receiving the additional hour of communicative tasks outperformed the group with no additional communicative
31 tasks, but there were no differences in the linguistic measure between the two groups. The results also were similar in Montgomery and Eisens teins (1985) research on form-focused instruction in combination with a more natural communicative interaction. The resu lts of their study showed that the communicative instruction group showed hi gher gains on linguistic measures (accent, comprehension, grammar, and vo cabulary) than did the grammar-based English as a Second Language (ESL) group. A far less researched area is within communicative contexts where the emphas is is on grammar. However, Beretta and Davies (1985) did examine this area in ESL schools in India. The results showed that learners in communica tive courses performed better on communicative tests and outperfo rmed grammar-based programs on contextualized grammar and dictation te sts. Participants in grammar-based programs performed better t han did those in communicative courses on discrete point grammar tests. Additionally, S pada (1987) investigated time spent on grammar instruction in communicati ve adult ESL programs and found that learners who received more explicit gr ammar instruction (i.e., focus on form) received similar results or even perfo rmed better on grammatical measures than did those learners who received less expl icit grammar instruction. Students in both groups received similar results on the communicative measures. The above noted research examined the role of focus on form in communicative settings and its effect on second language acquisition. Tomasello and Herron (1988, 1989) further examined when attention to form is most
32 functional, with college students in a Frenc h foreign language course focusing more on language form than on communicati on. The participants in the study were divided into two groups: in the fi rst group, learners were corrected after making an error, whereas in the other gr oup students were alerted beforehand of certain rules, exceptions, possible places for error, and so fo rth. They found that the former group performed better as meas ured by immediate and delayed posttests, thereby favoring turns that in cluded repetition with change and repetition with change and emphasis, also known as re casts. However, caution should be noted when generalizing such findings to communicative classrooms because the instructional context of this stud y was more focused on language than on communication. Researchers also have ex pressed caution regarding the validity of Tomasello and Herrons (1989) findings, namely due to internal validity (Beck & Eubank, 1991) and external validity (Long, Inagaki, & Ortega, 1998) concerns. The role of form-focused instruction in primarily communicative contexts was explored by Lightbown and Spada (1990). The data for their study were from 1,000 students in 40 intensive ESL cla sses and approximately 200 students in ESL programs. Their database included four intact classes in Grades 5 and 6, which amounted to 100 second language le arners. Based on their initial observations, there were different li nguistic results depending on the type of instruction received. The authors then further explored this issue by asking, Are there other differences in learner language outcomes that may be related to differences in instruction? (p. 435). First, the researchers used the modified version of the Communicative Orientat ion of Language Teaching (COLT; Spada
33 & Frhlich, 1995) scheme--Part A to colle ct data for both the macro-level and micro-level analysis. At the macro level, real life coding was taking place to describe activity type, studentversus teacher-centered material, macro skills, and whether the focus was on meaning or form and, if on form, if vocabulary, pronunciation, grammar, or discourse was targeted. The micro-level analysis used audiotapes and transcripts of the audiotapes to classify teachers behaviors as being either instructional or reacti ve. Instructional behavior was defined as teachers presenting a certain point and allo wing students to practice it, whereas reactive behavior was conceptualized as being a reaction to a students error. The results of Lightbown and Spadas (1990) study showed that all four classes were communicative; however, the instruct ional time on focus on form differed as well as did the instructional behaviors of the teachers. Direct grammar lessons were almost never taught; however, gr ammar lessons were given more as a reaction to learners errors. Based on these initial findings the authors hypothesized that the learner language in each class might show signs of the influence of specific items on which an i ndividual teacher had chosen to focus (p. 437). To verify the hypothesis a picture card game was created where a learner described a picture until the interv iewer could guess which one was being described. The task was audio taped and tran scriptions were made for the data to be interpreted. Using an analysis of variance (ANOVA), differences among the classes were found in grammatical accuracy of the plural verb and progressive ing (e.g., books and sitting). With regard to adjective placement in noun phrases, two of the four classes studied (Cla ss 2 and Class 4) were statistically
34 significantly different using Tukey s multiple comparison procedure. The possessive determiners were ascertained by the accuracy of his/her usage and the number of students who used both h is and her correctly. Class 2 had the least accurate results in both situations The authors suggested that these results were due to their development levels, which might have been somewhat different from those of the other classes. Lightbown and Spada (1990) caution t hat the data for this study were taken after the fact and that the data could not be generalized. However, they suggested that based on the fact that the participant s had similar backgrounds and exposure to ESL, the differences found might be related to the type of instruction provided, as shown by the fact that Class 1 outperformed all other groups on all the grammatical items in terms of knowledge and accuracy and had a teacher who focused on fo rm most frequently. This was in contrast to Class 4 where the teacher did not focus on gra mmar at all during the observations and which had the lowest grammatical accura cy. The authors did c onfirm, certain teachers seemed to have a particular set of structural features on which they placed more emphasis and for which they had greater expectations for correct use (p. 443). In addition, the results in this study provide further evidence that form-focused instruction within communicati ve contexts are more beneficial in terms of higher levels of linguistic knowledge and performance than just purely communicative classrooms. Further studies have examined negotiati on of meaning and conversational interactions among various interlocutor s. For example, Varonis and Gass (1985)
35 examined conversational interactions between native (NS) and non-native speakers (NNS), where the major purpos e was to see how conversations between non-native speakers differ from those between native speakers on the one hand and between native speakers and non-native speakers on the other hand (p. 71). Varonis and Gass (1985) contextualiz ed their study by briefly describing research already conducted between NS and NNS and then by describing conversational discourse between NNS based on data gathered for their study. The authors assumed that linguistic activi ty between NNS are different than that between other types of discourse espec ially with respect to negotiation of meaning. They based this assumption on a NS-NS discourse study conducted by Schegloff, Jefferson, and Sacks (1977), w ho found that other-correction (as opposed to self-correction) can be embarrassing and does not provide interlocutors with status of equality while participating in the discourse. Varonis and Gass argued that when interlocutor s have a shared competence (as with NNS-NNS discourse), it would give t he interlocutors more opportunity for negotiation of meaning. Varonis and Gass s uggested that simplified input (i.e., simplified vocabulary and grammar) is not as beneficial as the input based on negotiation of meaning. This suggestion wa s documented in Scarcella and Higas (1981) study that compared NS-NNS ch ildren with NS-NNS adolescents, where simplified input was greater with children participants; however, it was found that adolescents worked harder at keepi ng conversations flowing.
36 Based on Scarcella and Higas (1981) findings, Varonis and Gass (1985) examined the role of negotia tion of meaning among vari ous participants: NS/NS, NS/NNS, and NNS/NNS. The database included 22 dyads, of which 14 dyads were between NNS, 4 dyads were between NS and NNS, and the remaining 4 dyads were between NS. None of the par ticipants had previously met, and the 14 NNS-NNS dyads were matched for gender The participants were from the University of Michigan, where the English as a Second Language (ESL) NNSNNS dyad members attended the Eng lish language program. The NS-NNS consisted of conversation partners, and the NS-NS were university students. Each dyad was audio-recorded to speak freely in English. No other instructions were given. The first five minutes of each conversation was used for analysis. Based on previous research on discour se progression in conversations with interlocutors who have similar backgr ounds, Varonis and Gass proposed that when interlocutors are not on equal f ooting (p. 73), nonunderstandings occur. Nonunderstandings within their study were defined as those exchanges in which there is some overt indication that understanding between par ticipants has not been complete (p. 73). In order to build a model of negot iation of meaning, they suggested that nonunderstanding routines have one of two functions: (a) negotiation of nonunderstanding and/or (b) continuation of conversation. Misunderstandings that have gone unrecognized by one of the interlocutors were excluded from the database, whereas nonunderstandings were included. A proposed model was illustrated by the authors for nonunderstanding. The first part of the model consist ed of a trigger (an indication that a nonunderstanding
37 occurred from the hearer of the utter ance). The second part of the model, the resolution, consisted of: (a) an indicato r, which is a suggestion to the speaker that a nonunderstanding has occurred on th e part of the hearer, wherein the normal flow of conversation is inte rruptedalso known as negative input whereby an indication t hat the utterance is in some way inappropriate (Schachter, 1984); (b) a response, which is the recognition on the part of the speaker that a nonunderstanding has occurred; and (c) reaction to response, which is an optional turn that ma y occur to the nonunderstanding before continuing with the previous conversation pat h. This model is displayed in Figure 1.
38 Figure 1 Varonis and Gass (1985) model of nonunderstanding. Key: T = Trigger. A trigger is an indication that a nonunderstanding o ccurred from the hearer of the utterance. I = Indicator. An indicator is a suggestion to the speaker that a nonunderstanding has occurred on the part of the hearer. T he normal flow of conversation has been disturbed. It is also termed as negative input by Schachter (1984). R = Response. A response is the recognition on the part of the speaker that a nonunderstanding has occurred. RR = Reaction to Response. A reaction to response is an optional turn that may occur to the nonunderstanding before conti nuing with the previous conversation path. Trigger Resolution T I -> R -> RR
39 Varonis and Gass (1985) expanded the model in Figure 1 to include comprehension checks (CC) with interl ocutors optional stepping out of conversations as denoted by the arrows in Figure 2. Comprehension checks may occur before or after any turn in t he model, following a trigger. Comprehension check utterance or utterances may be ex pressed by the speaker or the hearer.
40 Figure 2. Varonis and Gass (1985) model of nonunderstanding including comprehension checks and interlocutor s stepping out of conversations. T = Trigger. A trigger is an indication that a nonunderstanding o ccurred from the hearer of the utterance. CC = Comprehension Checks. Comprehension checks o ccur before or after any turn in the model, beginning after a trigger. I = Indicator. An indicator is a suggestion to the speaker that a nonunderstanding has occurred on the part of the hearer. The normal flow of conversation has been disturbed. It is also termed as negative input by Schachter (1984). R = Response. A response is the recognition on the part of the speaker that a nonunderstanding has occurred. RR = Reaction to Response. A reaction to response is an optional turn that may occur to the nonunderstanding before conti nuing with the previous conversation path. T -> (CC)-> I -> (CC) -> R -> (CC) -> RR-(CC)
41 Using this model to analyze the dat a, the authors confirmed their assumption that the highest incidence of negotiation rout ines were found in those instances where the interlocutors did not share the same language or proficiency level. The lowest incidence of nonunderst anding routines occurred in exactly those dyads that shared a language and pr oficiency level. The results were analyzed with t -tests (comparing the means between and withi n) the two groups. Based on their findings, the authors sugges t that the NNS-NNS interaction is an important factor for NNS when acquiri ng a language, because it provides a common ground to practice skills and provides the availability of comprehensible input through negotiation that facilitat es SLA. Gass and Varonis (1991) conducted a follow-up study on the i ssue of nonunderstanding and towards a model of negotiation. They concluded that when there is incomplete understanding then repair (or correction; Ka sper, 1985) occurs and is shown in the form of negotiation of meaning (Ellis, 1994), which can be seen through, for example, confirmation checks, clarificati on requests, self, and/or other repair. In other words, negotiation of meaning is the interaction and effort between interlocutors to achieve mutual underst anding using various strategies (Ellis, 1994; Long, 1996). However, how does modified interaction differ with teacher-directed lessons and students working with in groups? Doughty and Pica (1986) conducted a follow-up investi gation from an initial study (i.e., Pica & Doughty, 1985), in which the researchers hypot hesized that there would be more conversational modification by student s in groups versus teacher-fronted
42 lessons. Modified interaction in both studies was defined as interaction which is altered in some way (either linguistically or conversationally) to facilitate comprehension of the intended message m eaning (p. 306). The hypothesis was not confirmed in the initial study. T he authors suggested that there were two main reasons for lack of conversational modifications: the type of task and the role of group members. In the initial study, an optional one-way information gap task was used, where participation among al l learners was not required. Also, the role of group members might have had an effect on the results. Possibly, because certain members may have been more proficient and more dominant, thus not allowing or provid ing opportunity for other gr oup members to participate. In addition, the role of proficien cy might have had an additional effect. In particular, high-proficient interlocutor s understood all utterances such that no modification was needed; whereas low prof icient interlocutors did not respond due to nonunderstanding, or in some cases, unwillingness. Ther efore, a follow-up study was conducted by Doughty and Pica (1986) to examine both the type of task (required vs. optional information exchange) and participation pattern (teacher vs. group vs. dyads). As such, the aim of this study was fourfold: 1. Compare teacher-directed and gr oup interactional pattern with both optional and requir ed information tasks; 2. Compare modified interaction ac ross teacher-directed versus group modified interaction where the task is held constant; 3. Examine the role of repetition; and 4. Assess the total amount of interaction within the tasks.
43 The purposes of the study were based on the hypothesis that: (a) information exchange activities would gen erate more modified interaction than from those activities where excha nges are an optional task, and (b) more interaction would take place in dyad pai rings rather than in group situations, which should result in more opportunities for modification than in teacher-directed lessons. The latter purpose was based on t he authors assumption that teachers would be less likely to seek clarificati on or confirmation, and more proficient students would not check comprehension, whereas less profic ient students might feel reluctant or embarrassed (Doughty & Pica, 1986, p. 309) with clarification or confirmations in teacher-fronted lessons. Consequently, the researcher hypothesized that within group settings, t he amount of modification would be higher than with teacher-fronted lessons with fewer chances for embarrassment and the highest amount of interaction wit hin dyads, wherein only two participants interact at one time. The participants chosen were six interm ediate adult ESL classes (three for the current study and three fr om the previous study used as archival data). The teachers chose at random to place st udents both in dyad and group situations. The data for their follow-up study were co llected in the same manner as in the previous study. The tasks were pilot-test ed and showed that they were not too difficult for the students. The two-way information gap activity used in all three settings was a felt board garden activity, where eac h participant received only pieces of information; however, when the information is put together, it revealed the complete activity.
44 To control for practice effect, the t eacher first provided a demonstration lesson with frequent comprehension checks. For th e teacher-directed lesson the teacher began the lesson, stopped after 15 minutes for questions and answers, and then continued. In all three interactional pa tterns (teacher-directed, group, and dyad) the activity was in progress at least 20 minutes before a 10-minute sample was taken. Modified interaction was the uni t of analysis and included clarification requests, confirmation checks, and comp rehension checks. Repetition was considered as taking place when communication broke down or when both interactants actively continued or created further topics. The results for effects on task and parti cipation pattern on the modification of interaction showed that required in formation exchange produced statistically significantly more interaction as anal yzed via a two-way analysis of variance (ANOVA). As such, researchers who ex amine any type of participation pattern should take into account tasks as a vari able when examining participation pattern and negotiation. Results also found a statistically si gnificant interaction between task and participation pattern; however, participati on pattern alone did show a main effect. A one-way ANOVA did reveal a statisti cally significant main effect for participation pattern as an independent variable, wherein modification of interaction was higher in the group vers us the teacher-fronted lessons. It is interesting to note that here was no difference between group and dyad participation patterns. A possible ex planation outlined by Doughty and Pica
45 (1986) might be the interactional experienc e, as has been argued also by Pica and Long (1986), between NS-NNS conversations. Statistically significant results were found on the task type, where required information exchange resulted in more m odified interaction, and statistically significant results were found between ta sk and participation pattern (Doughty & Pica, 1986). The researchers further investigated the role of repetition, which was tested by eliminating all instances of repetition in the database in order to determine effect on tasks and participation pattern. Similar results emerged as with those instances where repetition was included. This is not to say that repetition is not an import ant component of modified interaction. Indeed, Pica, Doughty, and Young (1985) found quite t he opposite. These researchers have attempted to define repetition, and have found that repetition might be the most critical component of inte ractional modification. Doughty and Pica (1986) also exami ned the total amount of interaction. This was tested using the sum of a ll T-units and fragments based on Hunts (1970) description. The results showed that the amount of speech increased when the task was required, as opposed to being an optional task, that the teacher-fronted interaction on required tasks generated more interaction, and that the group generated the least amount of interaction on optional tasks. Based on these results, Doughty and Pica ( 1986) concluded that when students are engaged in required information tasks, t he students will speak more and that modified interactional will increase when students work in groups. These results are supported by other findings (e.g ., Newton, 1995), where two-way tasks
46 resulted in higher frequencies of negotiation of meaning. Alongside the findings of the initial and follow-up study, Doughty and Pica (1986) ar gued that both group work and pair work provides students with opportunities for target language production and modified interacti on, but that the sole use of group work is not suggested. L2 learners produce many ungramma tical utterances that tend to be corrected by the teacher who is the sole input of correct utterances in the classroom. Thus, the teachers role, task type, and interactional patterns all are factors that affect modified interaction and amount of input. In summarizing the above review of liter ature, it can be said that the type of input, conversational interactions with both opportunities for input and output, and negotiation facilitate second language development to a various degree. It is not just the above interactions that in crease possibilities for successful target language attainment, but al so the negotiation between interlocutors provide successful contribution to a conversation. The type of task also influences the frequency of interaction. Doughty and Pica (1986) showed that required information, through two-way tasks, pr oduced more interaction. Even though Gass and Varonis (1985) did not find any difference in the two-way tasks as measured by indicators of negotiation, arguments made by Long (1989) suggest that there is enough evidence to show the usefulness of negotiated work, as well as more productivity with two-way tasks. Negotiation of meaning can be a factor where learners are able to notice their gaps. Long et al. (1998) point out that negotiation of meaning elicits negativ e feedback, including recasts. Such feedback draws learners attention to mismatches between input and output (p.
47 358). The following section focuses on the role of feedback and studies that have been conducted to determine its precise influence. Feedback Within the area of feedback, there hav e been various definitions and terms used depending on the field of study. Schac hter (1991) outlines the differences among feedback terms in the literature. Negative feedback tends to be used within the domain of psychology or conc ept learning, negative data or negative evidence within the field of linguisti cs or language acquisition, and corrective feedback is a term used in the pedagogic al field of second language teaching and learning. Lyster and Ranta (1997) al so note that corrective feedback is a term used by second language teachers, w hereas focus on form is used within classroom SLA research. For the purpos es of the present study, the term corrective feedback was used for the fo llowing three reasons: (a) corrective feedback is situated within the pedagogical realm, whereas the other terms belong within other related fields; (b) to examine types of corrective feedback and whether they are similar to those pr ovided by second language teachers; and (c) to determine if corrective feedback tec hniques differ based on participation type (i.e., within learner-learner dyads and the role of grade level of the learner dyads). Other terms were used, whenever necessary to reflect certain domains and fields of feedback. The notion of corrective feedback, as it is known in the field of second language teaching/learning, has its roots in the field of first language acquisition, which also has been integrated within the field of second language acquisition.
48 Earlier research focused on the signifi cance, existence, utilization, and perception of corrective feedback in in structional and nonpeda gogical settings (e.g., Lyster, 1998a, 1998b; Mackey & O liver, 2002; Oliver, 1995), and recent studies have explored corrective feedback within different pedagogical contexts (e.g., Iwasaki & Oliver, 2003; Morris, 2005; Panova & Lyster 2003). When discussing feedback, research findings on learner errors also should be provided. Hendricks on (1978) described the hist orical perspective of learner errors, and the then current research on learner errors in the classroom. Guiding his review of classroom resear ch on error correction, he outlined the following questions: 1. Should learner errors be corrected? 2. When should learner errors be corrected? 3. Which learner errors should be corrected? 4. How should learner errors be corrected? 5. By whom should learne r errors be corrected? Based on his review of the research, he summarized that learner errors (both oral and written) shoul d be corrected; however, there is no consensus from the literature on when to correct those e rrors, especially erro rs that seriously impair communication, stigmatize l earners, or are frequently produced (Hendrickson, 1978). Furthermore, direct co rrective techniques have been shown to be least beneficial (Hendrickson, 1978) and that peer correction might be more helpful to students as an effective inst ructional strategy than might teacher correction of learner errors. Indeed, earlie r research suggested that learners tend
49 to correct each others errors once the corrector already has overcome certain lexical and grammatical problems (Hendr ickson, 1978). In regards to the questions outlined by Hendrickson (1978), Lyster and Ranta (1997) argue that research findings on such fundamental questions are still inconclusive. Research studies on whether lear ner errors should be corrected have been examined within experim ental studies of classroom-instructed SLA. The when which and how have been examined within observational studies, and the who has been studied in the area of negotia tion of meaning. However, gaps still exist in the research on learner errors with more rigorous analyses that need to be carried out. From the linguistic perspective, l earners have two types of linguistic information available to them; these are known as positive evidence and negative evidence (Long, 1996; Long & Robinson, 1998). Positive evidence is defined as: (a) providing the correct form of input to the language learner ; or (b) learners exposure to utterances that tends to be well formed (Long & Robinson). Conversely, negative evidence helps the le arner to notice the gaps in their own learning by giving the learner informa tion of target l anguage and non-target language samples (Long & Robinson). Positi ve evidence might be authentic or simplified/elaborated, dependi ng on the learners profic iency level. Negative evidence can be pre-emptive (e.g., based on t he learners error, rules are given) and represent reactive negative feedback (whi ch can be explicit with overt error correction) or implicit (Long et al., 1998; Long & Robinson, 1998). Negative evidence, as opposed to positive evidence, has been challenged by first
50 language acquisition researchers (e.g., Beck & Eubank, 1991; Pinker, 1989). Working within an innatist paradigm first language acquisition researchers believe that the quality and quantity of negat ive evidence is too inconsistent for language learning to occur (Grimshaw & Pinker, 1989; Pinker, 1989) and that language is acquired through Universa l Grammar (UG; Chomsky, 1975), whereas negative evidence has little impact on UG and does not alter the interlanguage system of the learner. The most cited research studies on Canadian French immersion students (Lapkin, Hart, & Swain, 1991) have shown t hat linguistic errors are very much evident in immersion learners speech, even though learners achieve fluency in their L2. Schmidt (1993) also argued that not icing errors is an additional factor in acquisition, and White (1989, 1991) contended that with positive evidence alone, certain structures would not be acquired. In first language (L1) acquisition (e.g., Pinker, 1989) and L2 (LarsonFreeman & Long, 1991; Schachter, 1991; Swain, 1985) studies, negative feedback has been a point of contention. Diffe rent theorists have viewed the role of negative feedback as inc onsequential, such as the innatists (Grimshaw & Pinker, 1989; Pinker, 1989), or relevant (Tomasello & Herron, 1988; White, 1991). Long et al. (1998) separate negativ e feedback into explicit feedback and implicit negative feedback, and define the difference between the two forms of negative feedback as the following: e xplicit feedback .with the speakers attention overtly directed at problematic code features. With implicit negative
51 feedback, on the other hand, the message, not t he code, remains the interlocutors primary att entional focus (p. 358). Long et al. (1998) argue that the ro le of negative feedback is not only concerned with ultimate attainment, but al so with the rate of attainment (Ellis, 1994; Long, 1983, 1988), which also s upports Pienemann, Johnston, and Brindleys (1988) contention that instru ction does not have an effect on certain developmental sequences, but may have an e ffect on the variational features of the target language. Instruction does not cause learners to skip developmental stages. However, instruction does increas e the chance on the rate and ultimate attainment (i.e., quality) of the target language dev elopmentconsistent with Longs (1996) updated version of the In teraction Hypothesis, which states, Negative feedback obtained in negotiation work or elsewhere may be facilitative of SL development, at least for voc abulary, morphology, and language-specific syntax, and is essential for learning cert ain specifiable L1L2 contrasts (Long, 1996, p. 414). Focus on form within a meaningful context has been argued as being an important factor in language learning (G ass & Varonis, 1994; Long, 1996; Spada & Lightbown, 1993). Besides selective attent ion to form with nego tiation, negative (corrective) feedback also leads to modi fied output (Long, 1996; Lyster & Ranta, 1997). Negative feedback gives an opportunity for learners to compare target-like utterances with their own interlanguage utte rances (Tomasello & Herron, 1988), whereby the type of feedback can be either explicit or implicit. An example of explicit corrective feedback can be:
52 A: He go home B: No, you should say he goes home. Here, the response to the ill-target utter ance included an explicit correction. In contrast, implicit negative feedback to t he above ill-target utterance can be seen as: A: He go home B: John goes home everyday. This form of corrective feedback woul d be considered a recast, because the ill-formed original utterance is incor porated into the corrective feedback with the target form supplemented. Researchers in the area of recasts and negotiation moves (i.e., confirmation checks, clarification r equests, and repetition) have examined the effects of L2 learners participating in interaction (Ellis & He, 1999; Gass & Varonis, 1994; Mackey, 1999); benefits of interactional feedback (DeKeyser, 1998; Long et al., 1998; Lyster & Rant a, 1997; Mackey & Philp, 1998; Swain, 1985, 1995; Swain & Lapkin, 1998); the individual types of feedback in interactional conversations (Lyster, 1998a, 1998b; Lyster & Ranta, 1997); and in which way participation supports linguis tic development (Long, 1996; Pica, 1994). Benefits of interactional feedback hav e shown more target-like output by the L2 learners (Lyster & Ranta, 1997; Oliver, 1995)--leading towards modification of output (DeKeyser, 1998; Swain, 1985, 1995) and promoting L2 development (Pica, 1992). There have been mi xed findings regarding the specific utility of certain feedback types, namely, recasts. Long et al. (1998) and Mackey
53 and Philp (1998) have found advantages with those learners who have been exposed to recasts. However, Lyster and Ranta (1997) have found that recasts represented the least effective feedback type to lead to learner repair. Feedback Studies Within Teacher-Learner Interactions Initial speculation on the potent ial of teacher feedback and the instructional process had been first menti oned by Alwright (1975) He argued that error treatment was imprecise, incons istent, and ambiguous (p. 574). Fanselow (1977) examined corrective techniques of teaching in adult ESL classrooms and found that corrective techniques were confusing for learners. Roberts (1995), who examined Japanese learners ability to identify teacher f eedback, found that almost one-half of the recasts were not identified by the learners. Further, Doughty (1994) examined corrective feedbac k with adult learners of French and found that the learners responded to one-third of the recast moves. Based on this finding, Doughty concluded t hat learners tended to notice teachers feedback, even though one-third could be consider ed a low number to generalize noticing feedback. Chaudron (1977) examined the re lationships among type of error, feedback, and learner-repair and devel oped a comprehensive model of corrective discourse from his database on immersion students. He found that the most common type of feedback was teac hers reformulation of learners utterances with the inclus ion of emphasis, reduction, expansion, and repetition. Slimani (1992) studied y oung ESL learners notice of forms and self-repair and found that students did not not ice error correction in those instances when the teacher reformulated learner utterances implicitly; consequently no further
54 involvement from t he students occurred. Not notici ng error correction could be attributed to the developmental and proficiency level of individual ESL students. Of particular relevance, and a basis of the present study, is Lyster and Ranta's (1997) study on corrective feedback and learner uptake in four immersion primary classrooms. Here the authors argued the need for further research among different variables in a variety of teaching contexts. The purpose of their study was to develop an analytic model of error treatment sequences and apply such a model to the primary cl assrooms. The purpose of developing a model and applying it was to determi ne the frequency, distribution, and responses of corrective feedback. The complete database consisted of six French immersion primary school classrooms in the Montral ar ea. However, data used for this study were of four classes in one grade level: three Grade 4 classes and a split Grade 4 and 5. The data included 27 lessons from French l anguage arts and subject matter courses and totaled 18.3 hours or 1,100 minutes. All teachers were experienced, with more than five years of experience, and were selected based on their willingness to participate in the study. The lessons were audio taped and the Communicative Orientation to Language Teaching (COLT) coding scheme (Spada & Frhlich, 1995) was adapted for the immersion classr oom. The COLT was used to capture teacher-student interactions. The aut hors combined the COLT coding scheme with Doughty's analysis of fine-tuning f eedback to develop a model of error treatment sequence. The error treatm ent sequence model (Figure 3) was developed based on data from the study and was used as the main unit of
55 analysis. The sequence consists of lear ner error that can lead to teacher feedback or topic continuati on. There are two options after teacher feedback, either topic continuation with a teacher or student or learner uptake. Learner uptake being the students immediate respon se to the teachers feedback, which indicates students attention towards thei r erroneous utterance (Lyster & Ranta, 1997). If there is uptake, then the utte rance is repaired or they can still need repair. If the utterance still needs repai r, then additional corrective feedback can be provided. If no feedback is given, then t here is topic continuat ion. If there is repair, then either topic c ontinuation or some reinforcement is given by the teacher, after which there is topic continuation.
56 Figure 3 Lyster and Rantas (1997) error treatment sequence. Figure 3. Lyster and Rantas (1997) Error Tr eatment Sequence, used as the unit of analysis for coding of error and correct ive feedback types, as well, as learner uptake. Note. From Corrective Feedback and Learner Uptake: Negotiation of Form in Communicative Classrooms by R. Lyster and L. Ranta (1997), Studies in Second Language Acquisition 20, 37-66. Copyright by Cambridge University Press.
57 Coding for error consisted of student tu rns that contained an error or not, excluding hesitations, false starters, and those without prominence. Errors were classified as phonological, lexical, gr ammatical, gender-based, L1, and, where more than one error occurred at the sa me time, as multiple. Only language learner errors were included, whereas errors in content were not. Feedback coding consisted of six differ ent types of categories: explicit correction, recasts, clarif ication requests, metalinguisti c feedback, elicitation, and repetition. Explicit correction refers to feedback that was expl icitly corrected and indicating that it was incorrect. Recasts in volved feedback that was not explicit in nature but included different degrees of implicitness. Recasts also have been referred to as paraphrases in the COLT scheme (Spada & Frhlich, 1995), repetition with change, and repetiti on with change and emphasis (Chaudron, 1977). Translations also were included as recasts, namely because they served the same function and were infrequent in nature. Clarification requests were defined according to Spada and Frhlichs definition that provide students an indication that their utterances were ill-formed and that follow-up as either repetition or reformulation is required. Clarification requests also can be due to inaccurate content; however, only clarif ication requests due to student errors were included. Clarification request may include the repetition of the error or include phrases (see Table 1 for examples of each feedback type).
58 Table 1 Negotiation of Form Leading Towards Repair Corrective Feedback Type Definition Example Clarification Request May include the repetition of the error or include specific phrases What did you mean in X? Pardon me Metalinguistic feedback Refers to non-explicit comments on the non-target like utterance of the learner Can you find your error? Do we say that in English? No. No, not X Elicitation Contains three techniques used by teachers. First, contains a strategic pause either including the error or not. Second, teachers can use questions and, finally, the teacher can ask students to reformulate the utterance The dog can runs? The dog can . How do we say X in English Repetition Refers to repetition of the students non-target utterance with or without intonation of the error A children?
59 Metalinguistic feedback, on the other hand, refers to non-explicit comments on the nontar get-like utterance of the l earner, whereas elicitation contains three techniques used by teac hers. First, elicitation can contain a strategic pause either including the e rror or not. Second, teachers can use questions and, finally, the teacher can a sk students to reformulate the utterance. The final feedback type, repetition, refers to repetition of the st udents non-target utterance with or without in tonation of the error. Uptake, the final variable in the er ror treatment sequence, was defined, as a students utterance that immediately follows the teachers feedback and that constitutes a reaction in some way to the teachers intention to draw attention to some aspect of the students initial utteranc e (Lyster & Ranta, 1997, p. 49). As is seen in Figure 4, after the teacher provides feedback, there can be topic continuation by the teacher and/or student or it can l ead to learner uptake. If learner uptake occurs, it can result in needs-repair or in repair. Repair can be seen as repetition, incorporation, selfor peer repair, and is defined as the correct reformulation of an error as uttered in a single turn and not to the sequence of turns resulting in the correct re formulation; nor does it refer to selfinitiated repair (p. 49). For the purposes of Lyster and Rantas (1997) study, only those repair types were analyzed that occurred after prompting and did not include those that were self-correct ed. The needs-repair category consisted of the following six types of utterances: acknowledgement, same error, different error, off target, hesitat ion, and partial error.
60 The final category in corrective discour se is reinforcement. If there is repair, then either topic c ontinuation or reinforcement by the teacher is seen. Reinforcement refers to the teacher in some form, reinforcing the repair with acknowledgment, words of praise, and repet ition. After reinforcement, there is topic continuation. The results of Lyster and Rantas ( 1997) study showed that out of the six different feedback types by teachers, re casts were the most frequent, followed, respectively, by elicitation, clarific ation request, metalin guistic feedback, and explicit correction, with the least frequent being repetition. Ho wever, when Lyster and Ranta (1997) examined uptake as repair and needs repair, recasts have been shown as the least likely to lead to uptake, with explicit correction as the next least likely feedback type, as meas ured by frequency tabulations. The most likely type of feedback to lead to uptake is elicitation. The other types of feedback types leading to uptake were clarificat ion requests, metalinguistic feedback, and repetition (Lyster & Ranta, 1997). The authors further broke down the data by separating peer and selfrepair from repetition and incorporation. The purpose for further analysis was to examine the relationship between f eedback type leading to repair and the allowance for negotiation of form. The results showed that elicitation is responsible for almost one-half of the repairs, whereas recasts and explicit correction did not lead to repair. Based on the results of the study, the authors concluded that the four feedback types t hat allow for negotiation of form (i.e.,
61 elicitation, metalinguistic feedback, clar ification requests, and repetition) are the feedback types that lead to student-generated repair. Lyster and Ranta (1997) further contended that t he level of learners proficiency is a key indicator of the succe ss of negotiation of form as well as the different types of feedback used. However, if certain feedback types lead to more student-generated repair, we need to step back and ask whether learners even notice the feedback received or perceive it as such (Mackey et al., 2000). Mackey et al. (2000) researched the area of noticing fe edback and learner perception of interactional feedback. Interactional f eedback in their study was defined as negotiation moves, which are confirmation checks, clarification requests, and repetition. Re sults from their study have shown that learners seldom perceived feedback of morphosyntact ic errors (i.e., grammatical accuracy of structures) as such, but perceived it as various other types of feedback on error types. Furthermore, a post-hoc analysis on the type of feedback and error type tentatively found that recast was the most fr equent type of feedback for morphosyntactic errors. However, with feedback on phonology (speech sounds) and lexical (word/vocabulary) errors learners perceived them with more accuracy. Feedback types used with phono logy and lexical errors were negotiation and combination types. The author s pointed out several reasons that morphosyntax was not perceived correctly. First, this might be due to the communicative nature of the in teraction. They highlight Picas (1994) claim that negotiated interaction may be more benef icial for lexical errors, but less beneficial for morphosyntaxic errors. Sec ond, it might be that morphosyntax is
62 used with recast, and learners might not perceiv e it as such. An additional reason might be the cognitive overload of learners; it might not be feasible for learners to perceive all feedback types correctly. Finally, a limitation of this study, as pointed out by the authors, was the limited number of parti cipants, as well as the language barrier limited English proficiency of the participants in providing correct feedback to the researchers on t he stimulated recall procedures. An additional study with adult ESL st udents that also speculated on the perception of recasts is Panova and Lyst ers (2003) research. These researchers examined the relationship between f eedback types and learners responses. They used Lyster and Rantas (1997) m odel of error treatment and corrective discourse on teacher/student interaction. The database for their study included 25 beginning/early intermediate-level adu lt students, where 20 of the students shared French as a common L1. The teache r was a French/English bilingual with 13 years of experience, who was chosen based on her experience and willingness. The instructional approac h of the classroom represented communicative language teaching with mini mal linguistic forms, which was based on Spada and Frhlichs (1995) Comm unicative Orientation to Language Teaching (COLT) coding scheme. The COLT coding scheme revealed that students engaged in oral exchanges 90% of the time. Instrum ents for the study included the COLT scheme, observation of 18 hours of cla ssroom time during Weeks 6-9 (in a 9-week course), res earcher field notes, and audio recordings from the classroom. The dat abase for the current study contained 10 hours of all student utterances. Coding was adapted from Lyster and Rantas (1997) study
63 based on the error treatment sequence: learner error, teacher feedback, and learner uptake (with repair or needs-repair). Analysis included counting all errors, coding errors as phonological, grammatical, or lexical. From the analysis, the teacher utilized seven types of feedback: those delineated by Lyster and Ranta (1997), plus translation. Lyster and Rant a included the few translations from their database within recast; however, because of the high number of L1 utterances in the current study, these were coded as a separate category. Uptake and repair also were coded as per Lyster and Rantas definitions. The database included 1,716 student turns and 1,641 teacher turns. Almost one-half of the student turns that contained errors (857 turns) rece ived corrective feedback. Recasting and translation, respectively, were the two most frequent types of feedback (77% total). Learner uptake was evident in 47% (192 out of 412) of feedback moves where learner repair was coded 16%, and onl y 8% were repaired after teacher feedback. The highest rate of uptake and repair occurred with clarification requests, elicitation, and r epetition, whereas the lo west rate occurred with recasts, translation, an d explicit correction. The findings in Panova and Lysters ( 2003) study are similar to those in nonexperimental studies (e.g., wherein re casts tend to be the most frequently used type of feedback). Panova and Lyster consider the students proficiency level as a factor in the types of fe edback provided by the teacher. The authors also related their findings to Lin and Hedgcocks (1996), Mackey and Philps (1998), and Nettens (1991) conclusions that less proficient learners may not notice recasts, whereas more advanc ed learners regard recasts as negative
64 evidence. Explicit correction was minima l in Panova and Lysters study, whereas in Lyster and Rantas (1997) study, this type of feedback led to repair. The reason also might be attributed to proficiency level and the number of participants, which is an area in need of further research (Ellis, 1994). Another explanation might be the unique situation in which learning occurs in an ESL setting where class participants share a common L1, which may not be typical in an ESL classroom, where ESL learners are of different linguistic backgrounds. Participants were neither truly in an ESL se tting, nor were they in a strict foreign language setting, but fostered a different setting in itself Lyster and Ranta (1997) have argued that in order to understand better the nature of corrective f eedback, other variables and instructional contexts should be taken into account. Participant type (i.e., adult and children) and context type (i.e., immersion and ESL setti ng), as well as learners perception and noticing of feedback, were a few of the variables that were examined in the preceding literature review. The following review of research focuses on the age of the participants and the ro le of children either thr ough their roles with other children, with native speaker s, or non-native speakers. Feedback Studies with Children Interactions Building on research within implic it negative feedback and the role of negotiation and recast, Oliver (1995) ex amined the patterns of interaction between native-speaker and non-native spea ker (NS-NNS) dyads. The basis of her study stems from research in fi rst language acquisition, where it has been shown that children do use negative feedba ck (Brown & Hanlon, 1970 as cited in
65 Oliver, 1995). However, the question is if native speakers modify their interactions and provide feedback wit h their NNS peers, what type of modifications do they utilize? The partici pants in this study were 96 child dyads from four primary school s between the ages of 8 and 13 years. Eight NS-NNS dyads were formed based on age, gender, a nd proficiency level. The non-native speakers came from different linguistic backgrounds. Their proficiency levels were assessed by the researcher and teacher using the Australian Second Language Proficiency Rating scale from De partment of Immigration and Ethnic Affairs (Oliver, 1995). The native speakers were from the mainstream classrooms and were chosen based on their ability, status, and interactions with other second language learners. The pairs were audioand video-recorded twice (with one-week difference) while working on a one-way and two-way activity. The first 100 utterances for each pair and each task were used from the transcript for analysis, where all of the speech was included. T he coding categories were based on the interactive nature of conversations and were determined as non-native speaker initial turn, native speaker response, and non-native speaker reaction. A second rater also coded one-quarter of the sample and a high inter-rater reliability rate was calculated. Nine interactional pattern s were determined from the data. Each interaction was assigned into one of the categories. Wi thin the NNS initial turn category, the initial turn was classifi ed as incorrect, incomplete, and complete. The NS response category examined t he preceding turn and determined if negative feedback was provided, in the form of recast or negotiation, or if their
66 was topic continuation. The final categor y, NNS reaction, examined if their feedback was incorporated or if there was topic contin uation. The results were presented via frequencies, percentages, standard deviations, and the mean. These findings showed that within ch ildren dyads, when working on tasks, children interacted in multiple ways. The amount of negative feedback was very high, wherein 61% of errors were prov ided with feedback. In addition, 37% of NNS error turns did receive reactive implicit negative feedback. The author argued that this shows the existence t hat negative feedback is not rare or nonexistent as other researchers (e.g., Grimshaw & Pinker, 1989) have contended. The results also showed that the type of feedback given was related to the error of the non-native speaker. In instances of single errors, recasts occurred more often and with multiple errors when the responses were negotiated. The results also showed that non-native speakers incorporated the feedback when they had the opportunity to do so, and provided evidence that feedback is used in interlanguage production. Findings from Olivers (1995) study are important to the purpose of the current study. There is some exist ence of negative feedback within childrens interactions. More importantly, the proce sses of interaction may facilitate second language acquisition. Alongside the ro le of negative feedback in second language acquisition is the role of interloc utor types, namely the age and type of the interlocutor within task-based interact ion. Mackey et al. (2003) examined this area with adults and children. Their dat abase included 96 participants wherein one-half were adults and the other half we re children between the ages of 8 and
67 12 years. Within the age groups, the participants were randomly assigned and gender matched to native speaker (NS) and non-native speaker (NNS) dyads. This assignment yielded 12 native speakernon-native dyads and 12 non-native speakernon-native speaker dyads. Both children and adult non-native speakers came from a variety of L1 backgrounds and their proficiency level was assessed as being lower-intermediate. The pr oficiency level was based on the developmental sequence of morphosyntact ic forms by Pienemann and Johnston (1987). The adults in the NNS-NNS dyads were from an intensive English language program at a university in the Un ited States and the adult participants in the NS-NNS were in a similar progr am in Australia. The adult and children native speakers were from Australia, with the child native speakers being from the mainstream schools and the adults being at the same university as the nonnative speakers. Each dyad completed a one-way task, which required a drawing of a scene in the park, and the other participant had to describe it to her/his partner and then recreate it. The two-way task wa s a picture of a kitchen, where both dyad members collaborated to place the it ems in the correct place. Analysis of the transcripts included the first 100 utte rances for each task for a total of 9,600 utterances. An utterance was defined a ccording to Crookes and Rulons (1985) definition, consisting of one intonation contour, bounded by pauses, with a single semantic unit. Categories coded from the data were defi ned as initial learner utterances, interlocutor responses to nontargetlike learner utterances, and learner
68 responses to feedback. Initial learner u tterances were defined as targetand nontarget-like utterances. Only the nontar get-like utterances were used in the analyses. Next, all nontarget-like utterances were classified according to whether or not negative feedback (defined as recasts, confirmation checks, and clarification requests) were provided. If the topic continued without any negative feedback, then it was classified under no feedback. Along the same category of interlocutor response to nontarget-like utte rances, instances of opportunities for modified output was examined. If negative feedback was provided and opportunity was given for the learner to modify their output, then the utterances were coded as opportunity for modified output; however, if the learner did not have an opportunity to modify their output then it was coded as no opportunity for modified output. Under the category of learner response to feedback, the original ungrammatical utterances t hat were coded as feedback with opportunity for modified output were re-examined to see if they had been corrected. The results of the overall data se t were reported using the means, standard deviations, and ranges of the age and type of interlocutor dyads, along with the interactional structure. The following results are based on NNS-NNS and NS-NNS dyads. A chi-square analysis of the frequency of responses to nontarget-like utterances with negative f eedback revealed that the adult dyads provided statistically significantly mo re negative feedback than did the children dyads. Opportunities for learners to pr oduce modified output were examined and showed that across all the dyad types oppor tunities were provided as calculated by the frequency tabulation. A chi-square analysis also revealed that in the adult
69 NNS dyads, more opportunities for modifi ed output were offered than in the feedback provided by NS, and both child dyad s produced statistically significantly more modified output than did the adult dyads. The next set of results was based on adult versus child dyads. There were no statistically significant differences between adults and children in the amount of feedback, nor in the opportunities to use feedback in NS-NNS dyads. Results did show a statistically significant re sult with response to feedback in NNS-NNS dyads, where children produced statistically significantly more output than did adults. The overall results showed no stat istically significant differences between NNS-NNS and NS-NNS dyads other than the native speakers providing more feedback than the non-native speakers. These findings differ from other studies, wherein NNS interacted more with other NNSs (e.g., Varonis & Gass, 1985). Mackey et al. (2003) suggested that t he way they operationalized their data collection steps, in that grammaticality of the original utterance was taken into account, might have influenced the result s of the study. Adult NNSs provided less feedback than did the NSs, and within the child dyads, there was statistically significantly more modified output wit hin the non-native speaker dyads than within the NS-NNS dyad. The author s suggested that non-native speaking children seem to utilize more of the f eedback when it is from another non-native speaker. As such, both types of dyad (N S-NNS and NNS-NNS) are statistically significant, as is the age of participant ty pe (i.e., age was the significant factor among the NNS-NNS dyads, but not am ong the NS-NNS dyad). A possible
70 explanation of statistical significance fo r age is that children are great risk-takers and that they have fewer inhibitions in correcting others. Summary of Feedback Studies Thus, in summary, current research in feedback generally shows: 1. There is a positive relati onship between interaction and development (Mackey, 1999). 2. Interaction can serve as an a ttention-getter to learners in the area in which they need to improve (Gass, 1997). 3. In SLA the role of noticing is contentious (Schachter, Rounds, Wright, & Smith, 1998; Truscott, 1998), 4. Findings have shown some evidence that noticing plays an important role in language acqui sition (Mackey et al., 2000). 5. The age, status, and proficiency le vels of the interlocutors play a role in the amount and type of interactional feedback (Mackey et al., 2003; Oliver, 1995). 6. The type of task used also is an additional factor affecting the amount of interaction that ta kes place (Doughty & Pica, 1986; Long, 1989). 7. Proficiency levels of learners might have an effect on instructional behaviors and the ty pe of feedback provided by the teachers (Lyster & Ranta, 1997).
71 8. Different types of feedback can have an effect on the opportunity for modified output and the use of the feedback in SL development (Lyster & Ranta, 1997; Panova & Lyster, 2003). 9. Interactions with negative feedback occur within children dyads (Mackey et al., 2003; Oliver, 1995). 10. Level of proficiency and appr opriate uses of feedback can be based on the learners readiness as well as her/his attention towards feedback types (Lyster & Ranta, 1997; Mackey et al., 2000). The results noted here have not a ll been conclusively accepted and clearly more research is needed in this ar ea. Moreover, as argued by Lyster and Ranta (1997), additional research is needed using different variables in various instructional contexts. Research ha s been conducted within immersion and ESL settings. However, there is a gap in the literature in correct ive feedback within K12 foreign language settingsmore specif ically, in the learner-learner interactions in these types of instructional settings. Computer-Mediated-Communication Overview Theoretically, computer integration depends on the ro le of the computer (Levy, 1997). This can involve logical and physical considerations (Levy, 1997). If the computers role is that of a tuto r, then the logical problem centers around what work should be completed at the computer and which ones should be completed in the classroom. There is a cl ear distinction of computer-related work
72 and non-computer related item s. The tutors role is to evaluate, whereas the computer as a tool does not. The tutors role also is a temporary substitute for the teacher. It gives the lear ner an opportunity to undertake not only drill and practice, but also interactive and individualized activities. The physical (Levy, 1997) consideration in the computer as a tutor role would be computers in one space (in one r oom/space) and a classroom without a computer. The advantages of this type of work are mostly fo r the teacher, where he/she has more time to fo cus on oral work while st udents are working at their computers, and tasks are easily divided am ong proficiency levels. In the tutor role, the framework of the ta sks is given by the teacher. If the computer serves as a tool, then the logical and physical considerations are difficult to extrapolate because of t he supportive nature of the tool. The computer as a tool offers full int egration and collaborat ion among students, computer(s), and teacher. As such, the com puter is not the central focus of the activity but functions as a s upport to the teacher and learner. Much of the research on second l anguage acquisition also has begun to influence the area of computer-assist ed language learning (CALL) and other technological environments (Chapelle, 2001). Doughty (1987) provided possible theoretical orientations to CALL, and Chun (1994) was one of the first to examine foreign language learning and CALL using di scourse analysis within SLA. Other types of initial research conducted included (a) Nagata (1993), who compared learners who received feedback to thos e who only received minimal feedback; and (b) DeKeyser (1995) within computer-a ssisted SLA research, who examined
73 deductive versus implicit inductive lear ning. Another approach involved using the computer as a data-gatherer, as seen in Bland, Noblitt, Armington, and Gay (1990) and Hulstijin (1993), wherein computer s were used to collect data to make inferences about interlanguage and proce ssing strategies within classroombased learning. Currently, much of the CALL research has focused on the effects of using technology and how language learners intera ct with certain technologies to support language development (Chapelle, 2001; Egbert & Hanson-Smith, 1999; Lee, 1997; Warschauer, 1996, 1997). Ev en though there is some common ground on what learning should look lik e in technology-enhanced environments, research on the effectiveness of technol ogy has shown mixed results, from statistically significant gains to nonsta tistically significant gains (Milken Exchange, 1999). Similar results are rele vant within the field of computerassisted language learning research (Chapelle, 1998). The field has progressed in such a manner that we should not be concerned whether technology should be used or whether it is effective, but how technology should be used. Pusack and Otto (1997) argue that there are few areas of SLA t heory and research that do not impact the development and use of mu ltiple forms of media in foreign language teaching. However, they conti nue that seldom do theorists and methodologists reflect on the changes that multiple media have made to the way instruction is delivered. They further argue that the value of instruction, role of grammar instruction, and error correcti on, as well as the impact of accurate
74 speech development are all issues withi n SLA theory and are applicable within multiple forms of media in foreign language instruction. In addition to using technology with various approaches and skills, technology also is suitable for various learner populations. Otto and Pusack (1996) concluded that computers and te chnologies promot e student-centered instructional content. The use of technologi es builds on critical thinking skills and is appropriate for individual students levels and needs. Criteria and Attributes for CALL Integration Researchers in the field of instru ctional technology (Reigluth, 1999; Wilson, 2000) and second language acquisi tion contend that when affordances and benefits are interconnected within the whole philosophy and the whole curriculum, then certain gains will be evident. In the SLA, ESL, and learning theory literature, research repeatedly points to eight conditions that when present in the language learning environment, in some form and in some amount, seem to support optimal classroom learning. Egbert and H anson-Smith (1999) suggest that in an ideal environment eight principl es of optimal learning also should be used in computer-assisted l anguage learning. These principles are: (a) learners having the opportunity for interaction and negotiation of f eedback; (b) learners are provided with appropriate time and feedback; (c) learners autonomy is supported; (d) learners possess ideal leve ls of stress/anxiety; (e) learners interact in the target language; (f) lear ners are guided through a mindful process; (g) learners work with authentic tasks; and (h) learners have opportunities for varied and target language output. As can be seen from the eight conditions for
75 language learning, certain fa ctors (e.g., opportunities for feedback, learners autonomy, ideal levels of stress) are required to support effective and successful language learning experiences (Egbert & Hanson-Smith, 1999; Krashen, 1982; Long & Crookes, 1987; Pica, 1996). Research further shows that if CALL is appropriately integrated into the curricu lum, the language learning experience can accomplish the following: 1. support experi ential learning; 2. give students practice in a variety of modes; 3. provide effective feedback to learners; 4. enable pair and group work; 5. promote explorator y and global learning; 6. enhance student achievement; 7. provide access to authentic materials; 8. facilitate greater interaction; 9. individualize instruction; 10. provide multiple sources of information; and 11. motivate learners; (Egbert & Hanson-Smith, 1999; Egbert et al., 2002; Warshauer & Healy, 1998). However, for benefits to be evident, Pusack and Otto (1997) conceptualized the following attributes of multiple forms of media (multimedia) and technology integration: 1. The combination of multiple media; 2. control; and
76 3. interactivity. Usage of multimedia (e.g., text, motion video, photo images, sound, graphics) controlled via computer, complements Pu sack and Ottos (1997) philosophy of language learning through its potent ial to enhance students learning experiences. Using multiple forms of media also is known as TechnologyEnhanced-Language-Learning (TELL), a term us ed to incorporate not only CALL, but also all other usages of technology within language learning (Bush, 1997). While interacting with multiple forms of media (hereafter mentioned as multimedia), students can become more mo tivated to engage with more complex issues than with simple drill and skill. Students engage more by interacting with interactive programs and authentic materi al (Erben, 1999). Multimedia support contextualized learning to prepare students to apply what they have learned in an appropriate context (Reeves, 1992). Howeve r, the use of authentic material might lead to great frustrat ion and little benefit if no additional support is provided (Pusack & Otto, 1997). Tasks need to be supported in accordance with students levels of proficiency (Chapelle, 2001; Omaggio-Hadley, 2001) and developmental levels, and build on experiences and k nowledge that the students already possess. In other words, teachers need to build on students schema (Reeves, 1992). Technology-enhanced-lan guage-learning, if appropria tely chosen, can be a suitable platform for using authentic material, and build on language learning bridging students control over the program with other systems. A further factor when utilizing tec hnology within foreign/second language classrooms is the evaluation of tasks, curricula, and activities of computer-
77 assisted language learning (Chapelle, 2001). They should be assessed based on the language learners potentia l, learners fit, meaning fo cus, authenticity, impact, and practicality. This is also suppor ted by task-based research with learnerlearner dyads within online synchronous environments, as found by Pellettieri (2000). Therefore, when in tegrating technology into the foreign language classroom, the attributes of technology need to be evaluated as well as the task. When developing tasks (or curricular activities) the following questions should be asked (Chapelle, 2001): Learning Potential : Do task conditions present sufficient opportunity for beneficial focus on form? Learner Fit: Is the difficulty level of the ta rgeted linguistic forms appropriate? Meaning focus: Is the attention of learners directed primarily toward the meaning of language? Authenticity: Will learners be able to see the connection between the CALL task and tasks outside the world? Impact: Will learners learn more about the target language and about strategies for language learning through the use of the task? Practicality: Are there sufficient resource s for the task to proceed? Technology enhanced language learning gives control to teacher and learner over the pace of materials. Howe ver, Pusack and Otto (1997) caution that students may not choose the appropriate strat egies for effective learning if it is structured as an independent task. This is especially more true with low-ability students and students with insufficient backg round experience, or if complex
78 tasks are presented without support. In other words, teachers need to specify clearly defined tasks while interacting with materials (Omaggio-Hadley, 2001; Pusack & Otto, 1997). Another characteristic of technology is in its interactivity. When utilizing technology there are many factors to cons ider: navigation, user interface design, lesson architecture, task formats, and student inputs. Gay and Mazur (1989) provide the following recommendations in foreign language contexts: I. Begin with an epistemological analysis of the knowledge; a. analyze the competencies, b. analyze underlying methodological theories, and c. conceptualize the structure, and then II. Build a framework for interactions and activities that reflect this analysis. Thus, such a framework leads to t he optimal design of foreign language learning building on learners proficiency levels of both technology and the target language, as well as underpinning tec hnology with theoretical understandings. As such, for the purposes of the current study, the current discussion on utilizing technology is essential because a task needs to be evaluated for appropriateness and fit (Chapelle, 2001), appr opriately structured (Pusack & Otto, 1997), and use technology that is methodologically and theoretically based (Gay & Mazur, 1989), serving as a suppor t to learners and/or teachers (Levy, 1997) that centers on l earners competencies,
79 SLA and CMC Research Research in SLA and CALL has focused on the effectiveness of technology and the learning outcomes and th e interactions between the learner and the mode (Chun & Plass, 1996; Egbert et al., 2002; Lee, 1997; Warshauer, 1996, 1997). Liu, Moor e, Graham, and Lee (2002) and LeLoup and Ponterio (2003) recently conducted an overview of research that has been undertaken in second language acquisition and te chnology. LeLoup and Ponterio (2003) examined the research from an interacti onist and sociocultural perspective and argued that the research is troublesom e because of the varied data collection methods, population differences, lack of re search in the K-12 environment where it is most needed, no control of negativ e effects of the computer, and scant empirical research using ei ther quantitative or qualitativ e techniques. Liu et al. (2002), in their review of 246 articles fr om 1990-2000, also argued that there are a lack of research studies that are theor etically grounded, and they also called for more research within a K-12 school setting. CALL research also includes a spec ific type of communication entitled Computer-Mediated-Communication (CMC), which provides learners with an opportunity to interact with peers, inst ructors, native s peakers, and non-native speakers using synchronous or asynchronous interactions. Synchronous interactions occur in real time, with in teractants participating at the same time (Beauvois, 1992; Chun, 1994; Kelm, 1992). Examples of synchronous interactions include chat, video conferenc ing, audio conferencing, and telephone conversations. Asynchronous interacti ons, on the other hand, occur with a time
80 delay in which interactants do not have to exchange messages at the same time (Beauvois, 1992; Chun, 1994; Kelm, 1992). Examples of asynchronous include email, postal mail, discussion boards listservs, pda, or cell phone textmessages. Research in CMC indicates that w hen learners self-reported anxiety is lower (Beauvois, 1992; Kelm, 1992), there is greater student pa rticipation (Chun, 1994) and peer-to-peer interaction (Kern, 1 995). Research also provides some evidence that there is gr eater cultural awareness with students using CALL and that there is a greater par ticipation with online discour se than with regular faceto-face classroom interaction (Cub illos, 1998; Warschauer, 1997). Further, Gonglewski (1999) and Salaberry (1996) found that students who use online communication in their L2 are more awar e of their errors. Warschauer (1996) also reported that students who participate in online discussions in their L2 have more coherent and cohesive discourse. Is Synchronous Discourse Writing or Speaking? Synchronous discourse provides the opportunity for quick feedback, and learners can participate in one-to-one conv ersations, one-to-many conversations, or many-to-many communication events. Synchronous discourse provides the opportunity for learners to plan and shape their language before sending it for viewing to their interlocutor and, as such is different from the traditional oral classroom, where discourse happens more quickly, with great er likelihood of interruption and increased levels of anxiety (Beauvois, 1992; Kelm, 1992; Warschauer, 1997). Kern (1995) argues that during the synchronous local area
81 network (LAN) discussions, the student s operated largely wit hin a framework that resembles that of oral communicati on, even though the medium is written (p. 460). Tannen (1988 as cited in Kern, 1995) also states that just because the discourse is written does not mean that it should be consider ed a written genre. Thus, it has been argued that synchr onous discussions ar e on a continuum between oral and written discourse or s peak-writing (Erben, 1999, p. 239), with unique characteristics in a distinctive c ontext, with a unique language. Also, skills gained through speak-writing can be facilitat ive towards further education. In addition, skills gained through the medium of synchronous chat also will be facilitative to language learners in their future studies and provide experience in fine-tuning their skills in electroni c communications (Chapelle, 2001). Discourse, Affective Fact ors, and Language Production Research within synchronous chat has shown that learners use a wide range of discourse structures (Chun, 1994, Kelm, 1992; Kern, 1995; Sotillo, 2000). The quantity of production in synchron ous chats are great er than in oral discussions, and synchronous chats have an impact on the quality of learner utterances (Chun, 1994; Kern, 19 95; Sotillo, 2000; Warschauer, 1996). Researchers also report on the various discourse features of non-native speakers while participating in discussion through a synchronous program on local-area-networks (LAN). More specif ically, Chun (1994) conducted one of the first studies examining discourse rout ines within synchronous environments. She investigated the efficacy of class discussions on a computer network in increasing interactive competence, as well as a way for learners to manage
82 various discourse routines in different contexts. The data were collected with first-year German students. The firs t semester included 14 students and the second semester involved 9 students8 student s of the original first-semester students and a new student. The software program used for synchronous discussions was InterChange, which has been used in other studies (Beauvois, 1992, 1994, 1997; Beauvois & Eledge, 1996; Bump, 1990; Kelm, 1992; Kern 1995; Sullivan & Pratt, 1996; Warschauer, 1996). Students were given oral instructions with the written questions available on the discussion program. The questions asked were open-ended in nature concerning weekend activities, travel experiences, parental complaints about students, and so forth. The students hel d discussions among themselves, but were free to reply to any given cla ss member. After the 14 sessions were completed, the transcripts were prin ted and analyzed for frequency and length of turns held by each student, syntactic and gr ammatical complexity, and discourse structures. All turns were classified as questions and answers, statements and imperatives, and discourse management. Under each category, the frequency data showed that (a) students replied and questioned teachers, as well as other students; (b) the learners took initiatives in answering other students; (c) students used requests for clarification when questions were not understood; and (d) learners provided feedback to other students by agreeing, apologizing, requesting clarification, and pr oviding appropriate social expressions. The results showed that the quality of language pr oduction varied, with some learners producing simple sentences and others producing more complex sentences.
83 Also, participation was focused more on peer interaction rather than on teachers input. In an observation of online discour ses, Kelm (1992) investigated how synchronous discussions were used as comm unicative tools rather than tools for reading and writing. He found cert ain benefits, which included increased participation among group members, r eduction of anxiety, and individualized identification of errors. Similar findings also were found with Beauvois (1992) observational study of synchronous di scussions with university students participating in a foreign language c ourse and Chuns (1994) research on interactive competence. Some negative findings outlined by Kelm (1992) include offensive comments shared by learners in their syn chronous discussions (also known as flaming), in which comments were blun t, direct, and honest. Without the teachers presence, there was more usage of the first language by learners, and also more time constraints with completion of the ac tivity. Each of these limitations is interesting in itself. The honesty and di rectness of students while engaging in non-native discourse can reflect the in terlanguage competencies of pragmatic skills. Students are still in the process of learning and this type of medium can assist the instructors in providing mo re tailored feedback and also gaining further skills to obtain the necessary skills to communicate. The teachers roles in the three studies mentioned were that of non-dominance. The instructors did not heavily contribute to the discussions, but were present online during the discussion task. Kelm (1992) notes t hat students tended to overlook the
84 structures of the target l anguage. However, in order to prevent this, the instructor printed out a record of their discourse s and highlighted crucia l areas for them optionally to correctafter which, Kelm (1992) reports that the learners were more aware of the target language. Interest ingly, in all three studies, where the teacher was present, the nature of the activities represented open-ended questions that were geared towards discussion. Pellettieri (2000) argues that the role and object ive of the task is an important factor to consider with re spect to the successfulness of online negotiation. More specifically, because of the open-ended nature of the tasks, it is believed that the teachers role is more critical during open-ended discussion type questions. However, tasks that are more form-focused and/or required tasks have limited amount of outcomes. Of most interest was Kelms (1992) observation on the ab ility of viewing the learners interlanguage processes as they were occurring during discourse participation. Kelm observe s, that interlanguage [computer assisted classroom discussions] CACDs can aid in incr eased second language development (p. 449). As such, the observable interlangua ge processes include the increased capacity to read for main ideas, usage of a range of verb fo rms and grammatical structures that otherwis e might have not been used, and increased quantity of language. Kelm also noted t hat students did not frequently correct each other, which reflects the communicative nat ure of computer-assisted synchronous discussions. However, the quantity of corrections shared between the learners could have been influenced by the presence of the instructor in the discussions
85 and the fact that the instructor prin ted out the discussions and highlighted the learner errors for them to correct. This influence was not specifically mentioned in Kelms observation study, but could have inadvertently influenc ed the amount of error correction among students. Conver sly, learners giving feedback and requesting for clarification, as well as negotiation for meaning were found to be evident in Chuns (1994) study. Similarly, Beauvois (1992) expl ored synchronous discussions between university students in an intermediate Po rtuguese class taught by Kelm. Based on the results, Beauvois (1992) also ex plored synchronous discussions with one high school student attending a French fore ign language class. As a basis of evaluating CALL, she used Underwoods (1984 as cited in Beauvois, 1992) criteria for evaluating CALL, which more precisely evaluates the communicative nature of CALL, aiming at: acquisition rather than learning; grammar being implicit and integrated withi n the lesson; facilitating students to generate original messages; not being a judge or evaluator of what the student does; not telling students t hat they are wrong; not being overly rewarding with vari ous external symbology (lights, bells, whistles); not being cute; using only the L2; being flexible;
86 being exploratory; being facilitative and feeling natural; being unique and not performing activiti es that can be undertaken with a textbook; and having fun. Using these criteria, she examin ed transcripts from the Portuguese university class and found certain advantage s in that there was little use of the L1, students were self-encouraged to problem solve and to ask each other questions. Also, in accordan ce with the above listed criteria, students were generating their own u tterances without judgments or accusations. Because the discussi on was on a synchronous program, grammar was integrated into the less on (as per Underwoods criteria) by the instructor. However, after some time, students were talking using the synchronous program, but were using the target language inconsistently. To focus their attention on the gra mmatical structures and without placing judgment onto the learners, the instructor highlig hted a printed copy with grammatical errors and distributed it to the learners to review and, optionally, to correct. This helped the learners to focus on accuracy, also leaving evidence of the discussion, as it was available to review, and having all students participate almost simultaneously, which would not be able to happen with oral classroom discussions. Based on the several advantages f ound from the Portuguese classroom,
87 Beauvois (1992) used synchronous di scussions with a pupil who was having serious difficulties in French. Even though the pupil did not pass the course at the end of the data collection se mester, the author noted th e following benefits that did occur with the pupil: attitudinal change, more talk with other students than with the teacher, and greater language production. The au thor suggested that such a medium could be appropriate for students who do not seem to flourish. The process itself might have had an in fluence, where the pupil was centered only on one activity and/or the reading ( listening) and writing (speaking) were being self-paced in accordance with the l earners ability and proficiency level. Finally, this study was one of the initial r eports on the possible facilitative role of synchronous discussions towards negotiation of meaning, and their superfluous benefits for at-risk learners. The amount of target language produc ed was examined by Kern (1995) with Level 2 French students at a univers ity. He compared language production with oral class discussions versus online class discussions and found that learners produced a range of various cl ause types and verb forms. Advantages of synchronous discussions noted by Kern also were similar to those noted by Kelm (1992) and Beauvois (1992), where l earners had a greater opportunity to talk and produced greater language produc tion through complex structures and morphosyntactic features, with reduc ed anxiety and increased motivation. However, linguistic accuracy was not as evident and suggested that the electronic medium did not facilitate fo rmal accuracy. Kern (1995), however, strongly pointed out the disadvantages by st ating, On the other hand, the use of
88 InterChange introduces changes that may be unsettling. Teacher control is compromised. The fast pace of the discu ssion can tax learners reading ability. Grammatical accuracy suffers and consequent ly learners read defective French (p. 470). However, until further research can show otherwise, these statements may be too early and without basis. The defective language might occur because of the increased language producti on, where the ill-formed utterances are interlanguage processes in play (Kelm, 1992) or, as Pellettieri (2000) argues, the role of the task and the negotiation of meaning that occur also influence the amount, quality, and type of negot iation between learners. Summary of Language Development in Online Discussion Environments Results have shown an array of findings from no advantages in lexical and grammatical accuracy (Gonzalez-Bueno & Perez, 2000) to no significant differences in oral discussions (Sulliv an & Pratt, 1996). Overall benefits show positive learner attitudes and motivation (Beauvois, 1994; Bradley & Lomicka, 2000; Lee, 1997), increased student participati on including students who tend to be marginalized (Bump, 1990), increased le arner collaboration (Gonzalez-Edfelt, 1990), increased language production (B eauvois, 1992; Gonzalez-Beuno & Perez, 2000, Johnston & Milne, 1995; Ke lm, 1992; Kern, 1995) with a variety of discourse functions in synchronous mode (Chun, 1994; Sotillo, 2000), and syntactically more complex language out put in asynchronous mode (Sotillo, 2000).
89 Corrective Feedback and CMC Most of the studies noted have examined the interactions and benefits of CMC. However, relatively few investi gations directly have examined corrective feedback within online synchronous envir onments. The few researchers who have investigated corrective feedback within synchronous environments have examined it from NS-NNS (Castaeda, 2005; Iwaskai & Oliver, 2003), NNS-NNS (Pellettieri, 2000), and between child-child in teractions (Morris, 2005), each of which is significant for the purposes of this study. However, the current dissertation differs from previous research studies in that this study was: (a) situated with English as Foreign Languag e students, (b) conduc ted with learnerlearner adolescent foreign language l earners, and (c) analyzed using mixed methods methodology (see Figure 4).
90 Figure 4 Gap in the literature. Key : Language: SFL=Spanish-as-a-Foreign-Language JFL=Japanese-as-a-Fo reign-Language EFL=English-as-a-Foreign-Language Participation Type: L-L=learner-learner NS-L=native speaker-learne r T-L=teacher-learners NNS=Non-native Speaker Synchronous Tool: RTA=Remote Technical Assistance IRC=Internet Re lay Chat BB=Blackboard MSN=MSN Messenger
91 Castaeda (2005) conducted one of the most recent investigations on corrective feedback within both synchr onous and asynchronous environments. Her investigation was on corrective feedback types provided by four instructors of Spanish as foreign language instructor s to students at a large southeastern university. Interestingly, the results re vealed that instructors provided a greater amount of corrective feedback within t he asynchronous mode (i.e., bulletin board) than within the synch ronous mode (i.e., chat). Approximately 15% of errors received corrective feedback. In fa ct, instructors tended most frequently to use explicit correction in the bulleting boar ds and recasts in the chat room, where one instructor did not attempt to prov ide any corrective feedback to her/his students. Similarly, Iwasaki and Oliver (20 03) examined whether negative feedback even exists within online communication, mo re specifically within NS/NNS dyads of Japanese as a foreign language. Thei r research examined the provision and use of negative feedback, that is, recast s and negotiation of meaning within chat environments. The study stems from res earch on negative feedback in face-toface verbal interactions and current under standings of Internet applications within language learning. The aut hors argue that a paucity of research has been undertaken examining second/foreign la nguages with Internet applications more specifically, the linguistic benefits of such usage. As su ch, they examined whether negative feedback exists with native speaker and non-native speaker dyads on the Internet.
92 The participants were gender-matc hed NNS with native speakers of Japanese. The NNS were 12 university students studying Japanese at an Australian university. This was an intact class; however, the par ticipants were at two different proficiency levels. The prof iciency level was based on length of time studying a foreign language. The NNS par ticipants also had previous experience with Japanese word processors and reported that they were confident in using a Japanese word processor. The native s peakers were young adults in Japan, and had no previous chat experiences. The data were collected on three o ccasions one week apart. Before data collection began, a handout with instructions in their L1 was distributed asking the participants to use a Japanese script while chatting, not to use English or a dictionary and not to ask classmates or the researcher any questions during collection. They were also not to prepare any drafts while waiting for a response from their dyad members. A dditionally, they were also asked to practice with the Internet application called Internet Rela y Chat (IRC). IRC was chosen because it allows direct communication with native spea kers. Also, it rese mbles face-to-face communication in that continuous mess ages are flowing back and forth and the time is not sufficient to allow learners to review their messages. The Secret Chat portion of IRC was chosen to allow only two people to exchange messages without any external intrusions to the conversation. Upon data collection, the participants were asked to talk freely in all three sessions. Thus, the database included a total of 2,441 minutes ex changed between 12 dyad members across three separate intervals.
93 Data were categorized based on turns and were classified under the following procedure: (a) NNS initial turn s, (b) NS response to non-target language, and (c) NNS reactions to turns. First, coding was determined if the NNS initial turns consisted of a target or non-target language utterance. If it included a non-target language (NTL) utte rance with at least one form, it was coded as NTL. If the NNS provided a ta rget language utteranc e, or if the NNS corrected themselves within the same turn or in the subsequent turn, then it was not coded as NTL. Next, the non-target language forms were determined for type of error, which ranged from ungrammatical use of verbs, adjectives, copulas, and participles, misuse of tense and/or word order, mismatch of subjects and predicate, and typographical errors. Typographical errors were based on previous research on error classification, which includes typographica l errors, wrong conversions of Chinese characters (Chinese characters are us ed in the Japanese languageKanji), and errors in loan words and place names in foreign countries. Following classification of error type, all non-target language forms were examined for native speakers responses to the NTL fo rm. Two options were evident from the data set: either the NTL was ignored by the NS, or negative feedback was provided as a recast or negotiation of m eaning. A recast was defined as the NS modifying the ill-target u tterance without changing the original meaning of the NNS turn. Negotiation of meaning included cl arification requests or confirmation checks without the use of recasts. Fina lly, all turns that were provided with feedback then were classified for NNS r eaction to the feedback, which was
94 classified as either (a) ignoring the negative feedback, (b) no opportunities were given for response, or (c) response to negative feedback. If a response to negative feedback was given, it was then examined to det ermine if the response included incorporation of t he recast or modifying the ill-utterance towards more target language forms. The results of the study showed that the percentage of negative feedback and the NNS use of negative feedback provid ed were lower in frequency than for other studies of face-to-face interactions Also, the findings showed that negative feedback was mostly a response to typographi cal, grammatical, lexical, and other errors, respectively. Most feedback was ignored with typographical errors. The percentage of negative feedback frequen cy and provision of negative feedback according to error type ranged from 10 to 19.35. Frequencies on use of negative feedback and error type were between 4 an d 8, or 11.63% to 66.67%. The results did show use of recasts and negotia tion of meaning in subsequent turns. This might have been due to the relative ly low frequency levels and the number of dyads. The authors also argued that the low negative f eedback rate might be due to students perceptions of the errors (i.e., typographical e rrors are not that serious, whereas grammatical feedback more likely to be used and incorporated) and type of media used (email and chat vs. face-to-face). Another possible explanation might be t he role of the task, where it was structured as open-ended discussions. As previous research within oral interactions has shown, the type of task and the number of outcomes have an
95 effect on the amount of negotiation and t he type of production (Brock, Crookes, Day, & Long, 1986; Long, 1996; Pica et al., 1989). The role of tasks within online c hat environments has been quite maticiously examined by Pellettieri (2000). She examined, in contrast to Iwasaki and Oliver (2003), learner-learner explicit and implicit corrective feedback in synchronous environments and the development of grammatical competence with university students of Spanish-as-a-fo reign language. She suggested that the role of task can affect the amount of negotiation, qualitat ive and quantitative output, and learner modification when tasks ar e not conversationally oriented, but goal oriented. The role of tasks has been examined within transitional face-toface classroom research, where the ty pe of task affects t he type of production (Brock et al., 1986; Long, 1996; Pica et al., 1989). Accordingly, Pellettieri examined negotiation in terms of the role of tasks and it s effect on grammatical development within online environm ents among 20 undergraduate students learning Spanish-as-a-foreign language. More specifically, she examined if negotiation of meaning occurs in task-bas ed chatting, if negotiations facilitate mutual comprehensions, if the modifi ed output produced by learners are both meaning and form focused, and if negotiated interaction provide opportunities for corrective feedback and incorporation of such feedback. Five communicative tasks were creat ed ranging from open conversations to more closed tasks, where two tasks had an additional subtask. Before data collection commenced, practice sessions we re provided for learners to become more acquainted with the task. Also, befor e actual sessions began, tasks were
96 explained and instructions were given to use only the target language during task involvement. The participants were paired into seven mixed dyads and three same-sex dyads, and were visually separated dur ing the data collection sessions. The program used was ytalk (a UNIX bas ed program; Yenne, 1990) and the NCSA Telnet (National Center for Supercom puting Applications, 2000) was used to capture the transcripts. The data analysis was descriptive (based on frequencies and percentages), and the data analysis was based on Gass and Varoniss (1985) model of negotiation: triggers, signals, responses, and reaction to the responses. Based on the data, triggers we re classified as lexical and semantic (i.e., vocabulary and its correct meaning), morphosyntactic (i.e., grammatical accuracy), and content trigger s (i.e., entire content is not appropriate). Nontargetlike utterances within negotiation trigger s were calculated and the responses were categorized according to whether (a ) a modification o ccurred; (b) type of modification was lexical, morphosyntact ic, or semantic; and (c) the modification was target-like. Incorporations were analyzed as to whether corrective feedback was identified. Corrective feedback was cl assified as being either explicit or implicit. All types of corrective feedback were counted and determined for linguistic type and whether the utterances were target-like. The results of the studies revealed t hat in all five tasks learners negotiated for meaning in the task-based interacti ons and that learners both provided and reciprocated corrective feedback. The five different tasks produced different types of negotiation. The two tasks that included a more focused activity
97 produced more morphosyntactic negotiati ons than did the other three tasks. Interestingly, out of all the five tas ks, the second task, which had one possible outcome, generated the largest amounts of negotiation. The author suggested that this reflects research findings w herein one possible outcome generates the largest amount of negotiation (Pica, K anagy & Falodun, 1993) and that the level of task difficulty, which was somewhat higher than the lear ners proficiency levels, affected the amount of negotiation. Other research findings have shown that decision-making tasks and jigsaw puzzles (Blake, 2000; Morris, 2005) created more negotiation; however, it shoul d be noted that Smith (2003) did not find a statistically significant effe ct due to communication and task type. The issue of negotiation and its facilitation towards successful communication among one another showed that learners worked laboriously towards mutual understanding. This was determined by the analysis of transcripts as well as task completion. All of the tasks were successfully completed, except for the second task. The accuracy rate for those dyads that completed the task was more than 60%. The one dyad that did not complete the task had only an accuracy rate of 50%, and the author suggested that their lack of negotiation was surely detrimental to their performance (Pellettieri, 2000; p. 77). Again, the level of task difficu lty was another factor regarding task completion and accuracy rate. Determining whether negotiations to modified output were produced that were both form focused and meaning focuse d revealed that in response to negotiations and corrective feedback, lear ners produced linguistic modifications
98 (i.e., lexical, syntactic, and se mantic). Interestingly, 8 out of the 15 instances of errors were modified by the learners to wards the target form, and there was only one instance where the modification wa s away from the target language. Similarly, when examining provisions of corrective feedback and incorporation of target language forms, the quality of feedbac k was quite high, wherein only 6 of the 31 instances produced non-target fo rms and only 2 then were incorporated into subsequent turns. The author noted that none of the imp licit non-target feedback was incorporated into learners subsequent turns, suggesting that this might provide some evidence of the benef its of recasts withi n corrective feedback in NNS discourse, as argued by Long (1996). Also, incorporation of target-like forms has been discussed by Gass and Varoni s (1985), who state that learners know which utterances are correct and in correct. Pellettieri (2000) suggests that learners who can distinguish such utter ances have a high level of metalinguistic awareness. Also, learners within chat env ironments have an added benefit in that the talk is visual, provides learners with more time to process both explicit and implicit feedback, and discriminates both ta rget and non-target fo rms (Pellettieri, 2000). Pellettieri contended that her results contradict Kerns (1995) contention that the quality of production in el ectronic environments is questionable, inasmuch as the language produced is interlanguage, which is no more flawed than the traditional face-to-face oral interactions (Kelm, 1992). Additional studies on corrective feedback incorporation among learnerlearner dyads contextualized within inte raction, corrective feedback, CMC, and primary learners is provided by Morris (2005) research on fi fth-grade Spanish
99 immersion students. The study was conducted with three sections of a fifth-grade computer laboratory class containing a total of 46 par ticipants. The participants were randomly paired and completed a ji gsaw puzzle with their partners using the Blackboard 5.0 chat tool (Blackboard Inc., n.d.). The task also instructed students to draft an essay after completing the jigsaw puzzle. Data analysis consisted of coding for learner errors, learner corrective feedback, and response to errors and learne rs repair. The errors were coded as syntactic errors, lexical errors, and uns olicited uses of L1. Frequencies were used to analyze the data. The author reported 135 errors, with 76 following corrective feedback. The majority of the corrective feedback moves were, respectively, lexical errors (58%) or syn tactic errors (40%), with only 2% of unsolicited uses of L1. Of the correctiv e feedback moves, the majority was in a form of negotiation, with only 5% in recast s and none were evident within explicit correction. The highest rate of repair was for lexical errors (86%). Morris (2005) suggests that the rate of implicit feedback and the rate of repair is higher due to the fact that children are greater risk takers, as has already been highlighted in the present literature review (e.g., Mackey et al., 2003). The high rate of implicit feedback and repair also supports other st udies documenting that negotiation is one of the most common forms of feedback (Lyster & R anta, 1997; Mackey et al., 2003; Oliver, 1995, 2000, 2002). Morris however cautions that the results in his study might be due to other external factor s such as learners learning styles and strategies. However, he argues that mo re work and more rigorous experimental designs should be develope d to study further this area of interest.
100 CALL, CMC and Corrective Feedback Summary Successful technology integration into a classroom requires it to be situated within a sound theoretical fr amework, integrat ing methodological theories and examining t he precise role of the technology. All these are precursors that have been shown to pr ovide an optimal environment. However, tasks and activities also need to be evaluated based on their fit, potential, and level. Research findings have shown that when considering criteria for evaluation, synchronous discussion is a fac ilitative tool for l earners who are atrisk to fail either because of their proficiency levels or because of developmental readiness. If appropriately designed CALL ac tivities can assist the learner to visualize the talk process and have a more flexible and open environment that does not judge, evaluate, or tell them that they are wrong, but a llows them to ask questions, discuss, and seek assistance from other peers or inst ructors. Morris (2005) has utilized the synchronous t ool with immersion children and found encouraging results, where corrective feedback was provided and subsequently learners repaired their erro rs. Other benefits also have been noted, with learners reporting less anxiety and greater peer-topeer participation, noticing their L2 errors, and using a variety of discourse forms and structures. However, there is a paucity of resear ch on technology integration in the K12 foreign language program with at-risk second language learners. More research is needed to determine better pedagogical tasks and implications of using various tools and participation patte rns with second language classrooms. Based on the researchers current review of literature, all le arners included to
101 participate within research studies met t he minimum proficiency level; however, determining if learners have any documented special needs were not requirements for exclusion or inclusion. It is, therefore, impor tant when designing research studies to predetermine any special education needs of participants, which also may have an effect on the interaction pattern between dyad members. Foreign Languages and Special Needs It is a common belief in the field of education that for students with disabilities who are experiencing difficulty learning to read and write in their first language, literacy instruction should be in their L1 (Baca & Cervantes, 2004). This common notion is namely because t he disability interferes with native language (Baca & Cervantes, 2004). Resear ch shows that students, even with mild to severe disability levels, benefit fr om native language instruction in their L1 while immersed in an L2 environment (B ruck & Herbert, 1982; Cloud, 2002; de Valenzuela & Niccolai, 2004; Greenlee, 1981; Gutierrez-Clellen, 1999; Rondal, 2000). Even with these initial findings, qualit ative and quantitative research in early foreign language learni ng is not vast, especially wit h respect to the area of foreign language learning for special needs students, in which few articles have been published. Rosenbusch (199 8) states currently, very little information specific to the field is available to fo reign language teachers of young students to help them in this inaesthe (p. 59). In addition, specific teaching methods for foreign language students wit h special needs also are lacking, and those that exist are limited (DiFino & Lombardi no, 2004; Sparks & Ganschow, 1991).
102 However, awareness is increasing, reports that are more descriptive are being collected, and initial qu estions are being raised. Kretschmer and Kretschmer (1998) contended that foreign language teac hers need to know how the disability influences the language lear ning process. These authors classified disabilities with regard to foreign l anguage learning into four broad categories (this classification considers only one prim ary disability and not more). These categories are (a) hearing and visual im pairment, (b) severe motor control disabilities, (c) disturbances in neur ological and biochemical development, and (d) severe socio-emotional problems. Students who are cl assified as hearing and visually impaired usually have sufficient c ognitive abilities for learning languages, but lack communicative and language abilitie s because of the lack of exposure to the aural/visual environment and sensory disabilities. Severe motor control disabled children also have sufficient c ognitive abilities but are physically and communicatively impaired in expr essing the language. Children with disturbances in neurological and bioc hemical development usually are cognitively/neurologically impaired to various degrees and cannot acquire various aspects of the language such as the syn tactic, pragmatic, and lexical forms of words. The last category, children with severe socio-emotional problems have, obstacles to their language learning mainly with the semantic forms of language. Kretschmer and Kretschmers (1998) cl assification includes important factors in that not all special needs lear ners have similar abilities, and that their disabilities may range from suffi cient to less-sufficient cogni tive abilities. As such, special need students may over come obstacles by adapting educational material
103 to their strengths and not their limitati ons. For example, the ability to learn another language is possible when individu alized solutions are developed and obstacles are overcome with support fr om the immediate social environment; however, these obstacles are even more di fficult to overcome when they are due to severe language disorders, developm ental delays, and severe barriers to learning (Kretschmer & Kretschmer, 1998) However, descriptive studies have indicated that special needs children of various degrees and types are capable of learning other languages. For example, Candelaria-Greene (1996) reported on children in Kenya diagnosed with mental retardation (MR) and their ability to acquire fluency in three or more languages. She had found that because the social discourse environment required in dividuals to communicate in various languages, depending on with whom they we re communicating, children with MR also became fluent in the languages ar ound them. This might hint at language learning that is not solely dependent on cognitive ability. Gouin (1998), Holobow (1998), and Genesee (1987) reported on immersion programs that included special needs with learning-di sabled children. Gouin stated that accommodati ons need to be determined based on individualized needs. These needs includ e adapting activities, alternative assessments, pair/group work, and individu al attention. Holobows (1998) and Genesees (1987) reports also have show n that there are some benefits of language-disabled children in immersive env ironments: (a) they have been able to learn an additional foreign language sl owly and gradually (Bruck, 1982), or (b) they have achieved below average result s similar to their monolingual learning
104 disabled peers, but had the added benef it of a second language (Andrade, Kretschmer, & Kretschmer, 1989). Wings (1996) also reported on children with special needs within various foreign language settings and provided an exce llent example of a school district that values and encourages foreign langua ge education. The author describes a Foreign Language in Elementary School (FL ES) program in Putnam City School, Oklahoma City, which offers forei gn language programs to 18 elementary schools from Grade K -12. Inclusion in these schools represents students with learning disabilities, physically impair ed, and English language learners. Some of the characteristics of a school system adapting to a more diverse population have been opportunities for pr ofessional development, providing opportunities for teachers, special education, and forei gn language educators to consult with one another. Important aspects in teaching early foreign language learners with special needs are individualiz ation, inclusion, addressing students abilities on an individual basis, instruction, and progr am types (Genesee, 1987; Gouin, 1998; Holobow, 1998; Torres, 1996; Wing, 1996). Overall, from the review noted above, it can be surmised that an individualized approach has been utilized. In addition, strong parental support also has been weaved into important fact ors of success. Yet, empirical data are limited in the area of early foreign l anguage learning/teaching of special needs (Wing, 1996). From current understandings of forei gn language research with learning disabilities, early findings show that all educators and learners should believe
105 that foreign languages can be attai ned (Mabbott, 1994a); however, the degree of attainment will differ across a continuum Furthermore, second language learning should begin and develop after the first language has been sufficiently acquired (Andrade et al., 1989); however, when exactl y first language had been attained is not yet clearly defined. There is some evidence that early foreign language learning can be a predictor of success in foreign language learning for learning disabled learners (Bruck, 1982) and that immersion settings have shown to be conducive to language learning for both nonlearning disabled and lear ning disabled learners (Mabbott, 1994b). Within immersion settings learning-disabled learners have acquired the necessary tools to utilize the foreign language; however, difficulties within their specific areas of disability still remain (Mabbott, 1994b). Furthermore, research also has shown that forei gn language instruction should involve appropriate identification and pedagogical instruction that includes all modalities of visual, aural, oral, and inaesthetic learning (Ganschow & Myer, 1988; DiFino & Lombardino, 2004) and using material in classrooms that steadily progresses from familiar topics and contexts to unfamiliar topics and contexts (Andrade et al., 1989). Sparks and Ganschow (1991, 1993) and Sparks (1995) have devoted much of their research toward high school and university at-risk students and students with learning disabilities. For ex ample, Sparks, Ganschow, Pohlman, Skinner, and Artzers (1992) study of high school learning disabled students (mean age of 14 years) showed that by using direct instruction with the
106 Multisensory Structured Language (MSL) ap proach in both Spanish and English, students significantly improved in their native language phonology and vocabulary skills. The MSL approach involves using explicit and direct instruction of a foreign languagephonology, mor phology, and grammar, linking visual, aural/oral, and kinaesthetic modes t ogether (Moats & Farre ll, 2005; Sparks, 1995). An additional method, following a bottom-up approach to foreign language uses a dynamic method that combines various learning styles beginning with sounds and progressing towards wri tten discourse (Sparks, Ganschow, Kenneweg, & Miller, 1991). This approach al so is known as the Orton-Gillingham (Sparks et al., 1991) approach and was in vestigated in high school students. These students showed benefits and incr eased improvements in phonology development (Sparks et al., 1991). The imp lication of the above noted research findings for existing foreign language pr ograms. However, more research and information is needed involving various methods into early foreign language learning with young learners. Research findings within bilingu al special education (see Baca & Cervantes, 2004 for an overview) have not been included in the review of literature for the present study, even though disability types may be similar; however, learners needs are intrinsi cally different. Within foreign language settings, learners levels of academic success does not hinge on their ability to learn the language because a ll high-stake exams are in the learners L1; however, within bilingual spec ial education, le arners have to learn the second language to succeed academically, because all classes are held in the learners
107 L2. If learners do not succeed then placement into special education classes are warranted. Therefore, fo reign language studies have been reviewed, whereas learners with special needs, who are al so English language learners have been excluded from the review. Furthermore, due to the scant amount of research in the field of foreign language and specia l needs, empirically based research studies need to investigate the areas of inclusive environments and foreign language learning/teaching (Rosenbusch, 1998) ; the effects of various program types and disability (Holobow, 1998); t he relationship between the types of disability and foreign language learning; and additional research within primary schools providing immersion, dual language, or other foreign language programs. Further questions need to be asked on the ro le of instructional contexts, using technology as a tool to facilitate l earning and as a platform for expressing different learning styles and modalities of learning, as well as additional information on the nature of the inte ractions between students with special learning needs and those students who do not have special needs. Most importantly, when examining r egular classrooms it would be remiss not to include children with special needs in the study. With the inclusion and focus on individual learners within an in tegrated mainstream classroom, these factors can provide further informat ion on (a) the dynamics of classroom interactions, (b) the process of learning in progress, and (c) alternative ways to facilitate the language learning experience of learners with various needs. This is much more prevalent with the onset of mainstreaming an in creasing number of children into regular classr ooms across all grade levels.
108 As such, the present study attempts to build on current knowledge, as well bridges three areas of interest and cu rrent needs: second language acquisition and teaching, computer-mediated communi cation, and special needs education. Following is a description of the contex t in which the research study was conducted. The historical influences and linguistic background of the Republic of Slovenia, the country of residence for the participants in the current study, follows. Historical Overview of the Context A historical overview and its linguisti c and societal influences are briefly reviewed below. The summary and time below are based on the works of Prunk (1996), Eurydice (2001/2002), and Granda (n.d.). The present Republic of Slovenia has undergone a relatively turbul ent history politically and socially. Separate regions of Slovenia have been und er various rules dating back to the Celtic and Roman Empire in the fourth and third centuries B.C. Due to invasions, rebellions, and shifts in political goals, the Slovenes were ruled by various kingdoms. The Slovenes, in the 7th century A.D., were under King Samos tribal confederation, now known as the Czech Republic; in the 8th century under Frankish rule; then in the 10th century it was included in the medieval German Holy Roman Empire; and from the 14th century until the beginning of the 20th century Slovenia has been under the ru le of the Habsburgs (Eurydice, 2001/2002). During the 16th century of Turkish invasion s and the Napoleonic war, the first Slovene books were published along with the first Slovene grammar book in
109 1584. Under Emperor Joseph II (1765-1790) compulsory and primary education began and so did national interest in Sl ovenia among its people. Towards the end of the 19th century, Slovenia became part of the Austro-Hungarian monarchy, and during the First World War, more specifically in October 1918, it was part of the independent state of Slovene s, Croats, and Serbs. However, this was short lived. Due to pressures fr om Serbs to unify into one state and occupation of territories by the Italians, the independent states were united in December of 1918 into the Kingdom of Se rbs, Croats, and Slovenes. In 1929, it was renamed into the Kingdom of Yugoslavi a. This too was short lived. During the time of World War II, the Kingdom was disintegrat ed and divided by Hungary, Italy, and Austria (Granda, n.d.; Prunk, 1996). At the end of the Second World War, Slovenia joined five other republics and two autonomous regions and formed t he Federal Republic of Yugoslavia, which was later renamed as the Socialist Federal Republic of Yugoslavia. In 1980, after the death of Josip Broz Ti to, more demands were made by the Slovene people for independence. In 1991, the Slovenes adopted a new constitution and became an independent state. The Republic of Slovenia is now an independent republic wit h a parliamentary democra cy (Eurydice, 2001/2002; Granda, n.d.). The official language of the republic as well as the language of instruction is Slovenian. In ethnic minority areas, namely the Italian and Hungarian minorities, the o fficial languages also are Italian and Hungarian. In May 2004, Slovenia joined the European Union as a full member (Eurydice, 2001/2002; Granda, n.d.).
110 Education in Slovenia Many changes were made after the di ssolution of Yugoslavia to the political, economic, and social areas. O ne important change, and of interest in this review, is the educational system. A reform in the education system began in 1992 through research initiatives and discussion with experts in the field. The results of these initiatives were brought together in the Bela knjiga o vzgoji in izobrazevanju v Republiki Sloveniji (Krek, 1995), with an English version published one year later entitled, White Paper on Education in the Republic of Slovenia (Krek, 1996). It provides a basis of organization for pre-university and pre-school education. The aim of the White Paper was to restructure the educational system and base it on human ri ghts and law. The main objectives of the educational system is to include preschool children into appropriate programs; link the existing pre-school classes (a lso known as Kin dergarten) with the eight-year elementary school, and change it into a compulsory nine-year elementary school. The reason outlined is to provide successful completion of school for all pupils; encourage pupils to inae in general, technical, and vocational secondary schools; provide equal opportunities for both genders; provide opportunities for adult education; make possible transferr ing between programs; and provide opportunities for children with special needs (Eurydice, 2001/2002).
111 Eurydice, the information network on education in Europe, provides a detailed outline of Slovenias educational sy stem. In the report, Eurydice outlines the current framework gover ning education in Slovenia. The legislative laws governing education are: the Co nstitution, which gives a right to free education and provisions for minorities and Slovenes abroad; The White Paper on Education, which is the basis of Sloven ias international st anding in education; The Organization and Financing of Educat ion Act; the Elementary School Act; the Gimnazijski Act, Vocational Educat ional and Training Act, and the Adult Education Act. The basis throughout t he education system is the European Dimensions in Formal Education, which aims for an educational orientation, environmental protection, and healthy wa y of life. The European dimensions encompass the curriculum, role of the teac her, in-service teacher training, faculty and personnel, information and communication skills in foreign languages, international mobility, scholarship, yout h actions, and international exchanges of volunteers (Eurydice, 2001/2002). According to the Education Systems in Slovenia (Lakota & Gajgar, 2003), curricular reform followed in 1996 to 1999 consisting of 500 experts in the National Curriculum Council. Changes to the existing curricula were in the syllabi, goals and objectives, and timetables for the pre-school, elementary, and secondary schools, as well as in the curriculum for the linguistically and ethnically mixed areas. Currently, Slovenia is wo rking with the European Union in joint activities and participating with the Youth, Leonardo da Vinci, and Socrates
112 programs to achieve international com parable curricula and towards increasing knowledge in the European Unio n (Lakota & Gajgar, 2003). The education system (for a visual representation see Appendix A) consists of pre-school education, basic education, upper secondary education, post secondary vocational education, and higher education. Specialized educational programs within the educational program s include music and dance education, adult education, special needs education, and programs for linguistically and ethnic minority areas (Lakota & Gajgar, 2003). Preschool education, which includes pre-school program s at public or privat e institution, or at home, is optional and is subsidized if certain financial requirements are met (Eurydice, 2001/2002). Children attending pre-school program s are between the ages of one and six years. The approved cu rriculum is entitled, the Curriculum for Pre-school Institutions, and refers to six areas of activities: art, language, movement, mathematics, nature, and society ( Lakota & Gajgar, 2003). Basic education in Slovenia is free and has a required curriculum (Lakota & Gajgar, 2003). Basic, compulsory educat ion, has gradually expanded since the 1999/2000 academic year from an eight-year to a nine-year program and has completed the process of transformation to a nine-year program in the 2003/2004 school year (Eurydice, 2005). At the age of six years, all children are required to enter first grade, unless exceptions have been made by the committee for classification of learners, where it is determined that the child is not yet developmentally ready for entrance into the first grade (Eurydice, 2005).The nine-year elementary school consists of three cycles. The first cycle is from
113 Grades 1 through 3, the second cycle is fr om Grades 4 through 6, and the third cycle is from Grades 7 through 9. Students complete their basic education by Grade 9. After Cycles 2 and 3, external assessments are given to provide feedback on achievement to the parents, teachers, school, and pupils (Eurydice, 2005). The final compulsory external assessment in Grade 9 must be successfully completed at least in two out of three courses in order to continue their education in high school. A Year 10 of elementary school also is available for students who fail or wish to retake t he external assessment in the final cycle (Lakota & Gajgar, 2003). Secondary education consists of secondary vocational and technical education and general secondary education (i.e., gimnazija). The latter is divided into short-term programs (one-and-a-half to two-and-a-half years), secondary vocational programs (three years), or the technical education programs (four years). The secondary vocational and te chnical education program prepares students for entering the job market (Lakot a & Gajgar, 2003). Upon completion of the secondary vocational and technical pr ograms, students are able to continue their education in a higher education or post-secondary vocational institute, but are required to complete successfully t he external examination called matura. Students who enroll after elementary school into a short-term vocational program are not able to continue their educati on at a post-secondary or a higher education institute. The general secondary education program ( gimnazija) is divided into two groups: general and professional programs. Both programs last for four years
114 and end with an external examination called the matura. Upon successful completion of the external examination, students are able to enroll at the postsecondary vocational educational institut es or at higher education institutes (academic universities and prof essional-oriented studies). In the year 2000, a new law was pass ed for the education of children with special needs. It is an important legi slation, because it gives students with special needs the opportunity to attend school with thei r mainstream peers and learn in inclusive environments (Lakota & Gajgar, 2003). In addition, curriculum accommodations and modifications have been developed to assist students in achieving the standards set out for them. Modifications to the curriculum also have been made for the linguistically and ethnically mixed minoritie s. The area of Prekmurje in Slovenia has both Slovene and Hungarian as the languages of instruction (Eurydice, 2005). In Slovenian Istria, the language of instructi on is either Slovene or Italian. Where Slovene is the language of instruction, Italian must be learned as the second language. If Italian is the language of instruction t hen Slovene must be added as the second language (Eurydice, 2005). In addition to learning both languages, pupils also learn the hi story, culture, and her itage of both countries. Foreign languages and Technologies in Slovenia Special areas in education that have a priority in the nations education program are in health education, civic education, computer literacy, and the teaching and learning of foreign languag es (Lakota & Gajgar, 2003). The latter two are of particular relevance to the pur poses of this study. As a result of
115 prioritizing computer liter acy as a nationwide significance, Slovene schools were modernized with information and comm unication technologies through the Computer Literacy Project enacted by t he School Tolar Act (Lakota & Gajgar, 2003). The objectives of the project were to train students to use technology, thereby providing more quality educat ion, implementing more appropriate organizational structures in schoo ls, equipping schools with appropriate hardware, software, and facilitating re search conducted by students and faculty with new technologies in education (Lakot a & Gajgar, 2003). Slovene schools are now a part of the European School Netw ork UN-School net, which provides students with free access to the Internet. The Academic Research and Education Network of Slovenia (A RNES) provide support fo r students and teachers with Internet technologies. Another priority set out by the Sl ovene educational system was the critical learning of foreign lan guages, as set forth in the White Papers : The knowledge and skill to communicate the capacity to understand and express oneself (in the broadest sense of the word) in the Slovene as well as foreign languages is of utmost im portance. Developmental trends of education systems in the world show that, in addition to a thorough teaching of the Slovene language inseparably connecte d with its literature, it is necessary to begin teaching a first foreign language as soon as possible and soon afterwards (often already during the compulsory schooling) also a second and a third one. This is extremely important for
116 us, since we belong to a group of sm aller European countries. (Krek, 1995, English translation 1996, p. 5) Besides restructuring the educati onal system, foreign language education has gone through various changes as well. The eight-year elementary school system required 375 hours per school y ear of foreign language education in Grades 5 through 8 (Eurydice, 2001; Grosm an et al., 1999). With the changes in the nine-year elementary education system, the number of required hours has increased to 656 hours per school year for one foreign language (Ministry of Education, Science and Sports, 1998) pl us an additional 210 hours for a second foreign language (Eurydice, 2001 ; Grosman et al., 1998). The existing foreign language curricul um for the eight-year and nine-year elementary school was revised and modifi ed by the committee for the English language under the auspices of the Nati onal Curriculum Council (Eurydice, 2001; Grosman et al., 1999). According to the of fice of the Ministry of Education, Science and Sports (2004), the eight-year elementary school will be completely phased into a nine-year elementary school by the 2008-2009 school year. Hence, both curricula (for the eightand nine-year elementary school) developed for the English language are valid. Fore ign language education for the eight-year elementary school is required from Gr ades 5 through 8; however under the new nine-year elementary school, all pupils between Grades 4 and 9 will be required to take one foreign language and may add an additional foreign language from Grades 7 through 9 (Eurydice, 2001; Grosman et al., 1998).
117 The goal of the English foreign language curriculum, for both eightand nine-year systems, is for the learners to be able to use English in various contexts. Knowledge about the language per meates the curriculum (Eurydice, 2001; Grosman et al., 1999). In other word s, English is studi ed around themes and topics while using all macro skills and focusing on formal properties of the language whenever appropriate and necessary (Eurydice, 2001). In the eightyear curriculum, grammatical items ar e to be explained through lexical understandings, especially in the earlier gr ades, and not to teach explicitly grammatical functions as belonging under a s pecific category (e.g., I ran, past tense, verb run) (Grosman et al., 1999). Conversely, in the nine-year curriculum, the teaching of grammar should be implic it and have a facilitating role in the learning of languages, where students wil l learn the grammatical structure through its form and function (G rosman et al., 1998). The focus of both curricula is on the proficiency and development of the learner, based on their needs, interests, and learning styl es, as well as in learni ng English through exposure, input, interaction, output, and feedback (Grosman et al., 1999). The L1 (i.e., Slovene) can be used at earlier stages when certain structures might be above the learners proficiency level; it can also be used to save teaching time and use L1 to clarify when needed and to undertake a quick check of L2 understanding (Eurydice, 2001; Grosman et al., 1999). In the beginning grades, verbal communi cations are placed in the forefront with reading and writing bei ng gradually introduced and in accordance with the learners proficiency level. Reading and writing gradually increases to an even
118 level with speaking and listening in the upper grades of elementary school (Eurydice, 2001). This is not to say t hat all macro skills are not being developed from the beginning. Speaking, listening, reading, and writing skills are all integrated through various differentiated ac tivities. The curriculum is based on a communicative approach of learning fore ign language, while still emphasizing the need to learn the properties of the language in order to be able to communicate successfully in writing and orally (Euryd ice, 2001). Thus, not only are the verbal and nonverbal communicative goals of t he foreign language curriculum outlined, so are the grammatical, sociocultural, and cognitive and affective aspects. The focus within each aspect is on learners, specif ically on their levels of proficiency, while providing enough support to gain prof iciency. Because of the dual focus on communicative learning while focusing on t he form of the languag e, activities are typically based on (a) interactivity among peers, groups and teachers, (b) taskbased activities, (c) usage of songs and chants, (d) integration of various intelligences (e.g., multiple intel ligences), (e) Total Physical Response (inaesthetic activities), (f) project wo rk, (g) usage of audio and visual realia, (h) independent research, and (i) integration of technology. Throughout the learning process the teachers role is that of a facilitator and not the sole keeper of knowledge (Eurydice, 2001; Grosman et al., 1998; Gr osman et al., 1999). The dual function of the curriculum also is seen in assessment procedures. Both traditional and alternative assessments ar e highlighted. Sugge stions from the curriculum for ongoing assessments are: teachers observing learners in various contexts;
119 students submitting written work either as a formal test or as a written product; students completing portfolios that show their development in the target language; students carrying out self-evaluations; and teachers and students evaluat ing homework activities (Grosman et al., 1999). In addition, English should be used acro ss subjects within the school. The curriculum also delineates collaboration among English language teachers and subject matter teachers. The main purpos e of the curriculum is to bring the language across various contexts, for the foreign language to have purpose for the learner, and for the learner to develop linguistic awareness of their first and other languages and to devel op their own identities. Similarly, changes are being gradually implemented at the secondary level, due to the restructuring of basi c education (see Appendix B). English-as-a foreign-language is one of the subject ma tter classes that is required in general education ( gimnazija ). However, depending on th e foreign language taken in elementary school, English can be the first foreign language or the second foreign language beginning in the general secondary school. If English were the first language then the learner would have co mpleted a total of eight years of English upon graduation from the general secondary school However, if English is not the learners first foreign lan guage in elementary school, then English can be chosen as the second foreign language. If English were chosen as the second foreign language, then the learner would have spent a total of four years studying
120 English. Obligatory final exte rnal examinations (i.e., matura ) in English-as-aforeign-language are identical for all st udents, those with four years and those with eight years of English. Under the new nine-year elementary school and the restructuring of general education, t he total amount of time-spent learning English-as-a-first-foreignlanguage is 10 years. If English was chosen as an optional second foreign language in elem entary school, then the total number of years spent learning English would be sev en. If English was chosen as a second language at the onset of gener al education, then the tota l would be four years (Eurydice, 2001; Grosman et al., 1998). The focus on abilities (linguistic, so ciolinguistic, discourse, strategic, sociocultural, and independent learning) within the general secondary school curriculum is identical to that in the elementary school curriculum; however, the content is more rigorous. The goals of learning English-as-a -foreign-language through the general secondary education program are fo r students to be able to use English to assist with their studies and be able to read foreign professional literature for their studies in higher educ ation, to be able to communicate with individuals either professionally or per sonally, and to pass the final external examination of English as a requir ed or chosen subject (Eurydice, 2001; Grosman et al., 1998). In addition, the focus on language teac hing is similar to that in the elementary school, where the learne r-centered approach with cooperative learning through various activities is encouraged. Learners also are required to
121 read various literature, as well as be able to be competent in English both productively and receptively in all four macro skills. Chapter Summary Research shows that children do participate in negotiation and provide feedback whether in the role of the native speaker (Oliver, 2000) or in the role of non-native speaker (Mackey et al., 2003) Furthermore, re search within computer-mediated-communication has show n that learners communicate more online and are able to recognize more easily their errors in online environments than in traditional face-to-face cla ssrooms. Additionally, communicative classrooms that are also focused on t he form of the language have been shown to facilitate better second language develop ment than do classrooms that are just communicative in nature (Savignon, 1972) However, detailed comparisons of language learners interactions in forei gn language classrooms is a vital area to explore more in-depth, especially t he type and amount of corrective feedback learners provide among each other in online synchronous environments. In addition, by examining classrooms that are both communicative and focus on form (e.g., explicit gra mmar instruction), additional insights can be generated on the facilitative role of corrective feedba ck within such instructional contexts. Exploring the interactive env ironments of foreign langu age learning in Slovenia, additional evidence can be prov ided regarding the role of interaction from various linguistic groups as well as different inst ructional contexts. By studying learnerlearner interactions, more information c an be obtained in terms of: (a) how to conduct pair work within instructional progr ams that integrate technology, (b) the
122 facilitative role of a peer in the negotiati on process, and (c) inclusion of learners with diverse needs to examine the dynam ics of pair work. From a linguistic perspective, research on lear ner-learner interactions co ntributes to the current on-going research within negotia tion and interaction and its role in the process of second language learning.
123 CHAPTER 3: METHODOLOGY Overview The overarching purpose of the pres ent study was to examine corrective feedback within an online synchronous environment that occurs within adolescent leaner-learner dyads in foreign language classrooms. Equally important, this study was designed to in clude a few learner-learner dyads that have a documented special need. Correctiv e feedback was examined by using a commercially available (Ligon, T annenbaum, & Richardson Rodgers, 1991) twoway task (see Appendix C) within an onli ne synchronous environment. Similar two-way tasks have been discussed in research studies with: (a) corrective feedback in oral classroom discussions (Mackey et al., 2003), (b) feedback and task-based interaction (Mackey et al., 2003), and (c) chat environments (Morris, 2005; Pellettieri, 2000). Because the contex t of the study and research questions guided the research design underpi nned by the pragmatist philosophy (Tashakkori & Teddlie, 1998), qualitative data collection (i.e., text data) with quantitative and qualitative analysis tec hniques was utilized. Therefore, integrative research (i.e., mixed methods ) was employed to answer the research questions. The specific aim of the pres ent research was to: (a) investigate incidences of corrective feedback am ong EFL adolescent learners within an online synchronous environment, (b) ex amine the type of feedback, (c)
124 investigate the relationship between error and feedback type, and (d) explore the interactional conversation characteristics of interlocutors in dyads when one or more of the learners have a documented special need. The first three purposes were addressed via quantitat ive analysis of qualitative data using both inferential and descriptive statistics. The final purpose was address ed via qualitative conversation analysis. The database c onsisted of data from 208 participants, which were collected from: (a) a twoway information gap activity within a synchronous chat room, (b) a questionnair e, and (c) semi-structured interviews with 10 participants. The transcripts from the two-way information gap activity within the chat environment were used for quantitative and qualitative data analyses. The purpose of the intervie w and questionnaire was to add breadth and scope to the study. Namely, the questionnaire was utilized to acquire participants personal background info rmation, language experiences, and computer experiences. Partic ipants for the interview we re collected from extreme cases, as well as, participants with special needs. The aim was to obtain additional insight into the learners perceptions, attitudes, usefulness, and perceived effectiveness of communicating in a foreign language using an online synchronous tool. The researcher also k ept a journal to enter any observations, thoughts, and comments from pa rticipants or teachers to triangulate the collected data. The researcher revi ewed the data analysis, interp retation, and final report with participants instructors for fi nal feedback and comments. All personal information (i.e., first na mes, surnames, place of residence, name of school, telephone numbers, and pers onal addresses) were kept confidential. Names
125 were changed into identific ation numbers and were known only to the researcher of the present study. Analyses of data occurred within a mi xed methods framework, following the stages of data reducti on, data display, data transformation, and data integration (Onwuegbuzie & Teddlie, 2003). This chapter reviews a description of the participants, research design, and data analyses procedures. This study concludes with the results in Chapter 4 and the summary, discussion, recommendations and imp lication in Chapter 5. Participants for Quantitative Study Participants for this study were student s from Grades 7, 8, 10, and 11. At the time of the study, participants we re attending English-Foreign-Language (EFL) classes in mainstream public school s in Slovenia, Europe. The ages of the participants ranged from 11 to 19 years. Members of the study comprised learners from approximately tw o to three sections of Gr ade 7, 8, 10, and 11 from various schools in Slovenia (see Table 2 for demographics on participants).
126 Table 2 Overview of Participants by Grade Grade Level Number of Schools Number of Sections Students with Special Needs Age (mean) Female n % Male n % Total Participants Grade 7 2 4 3 12.36 30 46.88 34 53.13 64 Grade 8 3 4 0 14.38 25 78.13 7 21.88 32 Grade 10 2 6 0 16.92 35 72.92 13 27.08 48 Grade 11 2 5 0 17.97 44 68.8 20 31.3 64 Total 5a 19 3 15.41 134 64.42 74 35.58 208 a Total of 5 different schools. Both seventhand eighthgrade parti cipants were from the same school, except for an additional section/school in Grade 8. Grade 10 and 11 participants were from the same two schools.
127 In Slovenia, there are 450 prim ary schools and 160 secondary schools (The National Education Institute of the Republic of Slovenia, 2004). Because mixed methods research also require s mixed methods sampling to increase internal validity/trustworthiness as well as generalizability/tr ansferability (Kemper, Stringfield, & Teddlie, 2003) in the pr esent study, the researcher selected participants using a multilevel approach, that is by contacting all schools, using homogenous case sampling followed by simple random sampling (see Figure 5 on sampling).
128 Figure 5 Selection of participants. Figure 5. A visual representation on mixed-method sampling techniques .
129 First, all schools were contacted th rough the National E ducation Institute of the Republic of Slovenia and the EFL association for teachers entitled the International Association of Teaching En glish as a Foreign Language Slovenia. The National Education In stitute provides training, consultation, resource material, research informa tion, placement assistance, parental information, teacher materials, and other school-rela ted assistance for various types of schools. School types range from day care, kindergarten, elementary school, secondary education, university studies, vocational education, special needs education, adult learning, and e-learni ng for private and public schools in Slovenia (The National Educ ation Institute of the R epublic of Slovenia, 2004). The same request also was made by t he researcher to the International Association of Teaching English as a Foreign Language Slovenia to provide contact information of all English Fore ign Language teachers to the researcher. Upon the school principals and teacher s agreement to participate in the study, a homogenous case sampling strat egy was used. All schools had to meet the following criteria to be placed in t he pool of applicable participants: (a) have an EFL program from Grade 5 onwards in the elementar y schools; or be a high school wherein Grade 10 and Grade 11 lear ners are enrolled in a general secondary school (i.e., gimnazija ); (b) have a computer l aboratory or a classroom with a minimum of one computer per st udent participating in the study or be willing to divide the class so that one lear ner is using a computer at a time; (c) possess Internet connection on all com puters; (d) be willing to download the MSN Messenger program on t he computer or to use the web version; and (e)
130 have teachers and students who are willing to participate. Out of the 10 schools that volunteered to participat e, 5 schools met the criter ia above. A total of 238 students had agreed to participate in the st udy. However, transcripts were eliminated or deleted due to incomplete data, sole use of L1, technical glitches, electrical outages, students not correctly saving their chat sessions, or, as in one instance, lost data on a disk due to the floppy disk malfunction. Other participants transcripts were elimi nated from the data analysis due to the following reasons: (a) odd number of student s (i.e., not having a partner), (b) whole transcript being off task, (c) noncompletion during the practice data sessions, and (d) absenteeism between the practice and actual sessions. One dyad was eliminated from the data analysis for using profanity in all turns. Consequently, out of 2 38 students enrolled to parti cipate in the study, 208 students completed both the practice se ssion and actual data collection period, met the guidelines for inclusion crit eria, and, completed the background questionnaire. The number of participants pe r school and per grade is shown in Table 2. According to Stevens (2002) Power Sample Size Table, a sample size of 256 was needed to detect a moderate effect size (i.e., d = 0.75) with an acceptable statistical power of .8 at t he .05 level of sign ificance. However, because of the low number of schools and t eachers willing to participate in the study or not meeting the inclusion criter ia, only 238 participants were available. Furthermore, due to the above noted reasons another 34 students were excluded. Data collection in a subsequent school year was considered; however, because the students would be the same participants in the following school
131 year, this would have violated inde pendence among the grades. For example, students in Grade 7 would be the same st udents in Grade 8 the following year and students in Grade 10 would be the same students in Grade 11 the following year. The fact that the sample size obtained was smaller t han that suggested by the a priori power analysis is consi dered a limitation of this study. Thus, the sampling frame consis ted of 208 participants attending a mainstream public school selected in Sp ring 2005. Because the Slovene school system gradually is implementing a ni ne-year elementary school system, some students were in either Grade 8 of an eight -year elementary school or Grade 9 of a nine-year elementary schoolin both situat ions the pupils were in their final grade of basic education. For the pr esent study, Gr ade 8 students were combined with the Grade 9 students in the dat a set. In essence, they had spent a similar amount of time st udying English as a foreign language and were of the same age group. The participants were from intact classes and the researcher randomly assigned the participants into dyads as they entered the class. Of the 208 participants, 104 dyads were formed and of these matched pairs, 64.42% were female. Students mean age in Grade 7 wa s 12.36, in Grade 8 was 14.38, in Grade 10 was 16.92, and in Grade 11 wa s 17.97. All students were of a Caucasian background; however their native language did slightly differ. Almost 94% of the students nativ e language was Slovene, 3% Serbian, 2.5% SerboCroatian, and 0.5% of t he students reported both Croatian and German as their native language. However, the students re spective teachers reported that none
132 of the students whose L1 was not Slovene were receiving any type of special instruction for the Slovene langua ge and had been schooled in the Slovene language since first grade. The length of foreign language study also varied among grades. Length of English-as-a-f oreign-language study encompassed extra-curricular English classes thr ough private language schools and private lessons, as well as through formal instruction through the public schools. Grade 7 students reported an average length of 3.96, Grade 8 of 6.38, Grade 10 of 6.90, and Grade 11 of 8.36 years of EFL study. Participants for Qualitative Study Participants for the qualitative study were learners with special needs. Although, data from learner s with special needs were included in the quantitative analysis, the data were extracted for further qualitative analysis, more specifically conversation analysis. The purpose for a follow-up qualitative analysis was to review interactional characteristics of conversation among learners with special needs in terms of their corrective feedback moves, error types, and responses to given prompts by their fellow dyad mem ber. Participants with special needs were determined by the teachers official r eport of any documented special needs, that is, by an issuance of an individualized pl an or an official report by the school. In addition, identification of students with special needs was determined by the schools willingness to provide the informa tion to the researcher or the parents and learners willingness to disclose such information. If students, parents, or teachers did not disclose any special ne eds, then the student s were identified as students with non-specia l needs and were not included in the follow-up
133 qualitative data analysis. Ou t of the 208 participants, three students were documented with a special need and had an individualized education plan. The special needs consisted of a neurological disorder and ep ilepsy, attention deficit hyperactivity disorder (ADHD), and a lear ning disability. Of the three learners with special needs, two were males and one was a female in the seventh grade. The first language for all three learners was Slovene and all had had experience with using a computer and participating in chat rooms. One ma le reported using computers for six years, the other male for four years, and the female for three years. All reported having had previ ous experience with chat and being comfortable using the com puter and participating in ch at rooms. The length of English-as-a-foreign-language study was r eported equally for all three students, that is, three years. Participants for Interview Stratified purposeful sampling (Patton, 2002) had been used to choose participants for the oral interview. Stra tified purposeful sampling is defined as illustrate[ing] characteristi cs of particular subgroups of interest; facilitate[ing] comparisons (Patton, 2002, p. 244). Interview participants were chosen based on the number of turns, quantity of errors while completing the task relation to the rest of the class, and teachers report of work in class. More specifically, participants were chosen based on the extr emities on each end of the continuum (i.e., high learners and low learners). In addition, learners with special needs were automatically included in the interview pool. As such, a total of 18 participants from the 208 participants we re chosen for the final stage of data
134 collection, that is, to participate in a semi-structured interview with the researcher. However, out of the 18 chosen only 10 parti cipants were included in the interview analysis. A total of five par ticipants declined to participate in the interview. They did not provide a r eason. In addition, data collected from an additional two participants were not audible and one additional participant responded with I dont know on all ques tions and did not wish to comment. Consequently, 10 students or 5 dyads we re interviewed based on the following structure: (a) one high-high learner dyad (students who were above average in Englishon the high end of the continuum), (b) one low-low learner dyad (students that are below average in Engl ishthe low end of the continuum), (c) one special need-special need learner dyad, (d) one high-special need learner dyad, and (e) one low-high learner dyad. The interviews were conducted at different times, depending on participants availability, but no more than two days after data completion. Ethical Considerations Prior to conducting this investigati on, a proposal was presented to the University of South Floridas Institutional Review Board (IRB) for approval of the pilot study (Appendix D) and the current investigat ion (see Appendix E). The researcher also completed the re quired continuing and core education requirements to conduct research (A ppendix F & G). Data for the actual investigation were collected afte r all approvals were obtained. Permission to enter the schools was se cured from the National Education Institute of the Republic of Slovenia (A ppendix H). Any information received from
135 the data or through the data co llection processes that revealed the identity of the participants were changed and altered to protect their anonymity. All hard copy information pertaining to the disability of the participants in the study were kept in the researchers locked file cabine t and all electronic data were passwordprotected on the researchers personal co mputer. All data colle cted electronically also were saved to a disk and locked in the researchers file cabinet. All names from the questionnaire were changed to identification numbers and any identifying information in the data set wa s changed. Only the researcher of the present study had access to personal info rmation. Inter-raters had access to the data for data coding; however, all i dentifying information were changed beforehand. Ethical issues such as the characteri stics of the participants were taken into consideration. The informed cons ent form that had been created by the National Education Institute of the Repub lic of Slovenia and the researcher was distributed to the students and their parent s (i.e., if underage) one to two weeks before data collection commenced. The participants were provided with the opportunity to withdraw at any stage from the study for any reason and without any penalty or consequence. Instruments Questionnaire A questionnaire was distributed to the students in Grades 7, 8, 10, and 11 during the practice sessions. They were instructed to read the questionnaire and return it to the researcher the same day. The purpose of the questionnaire was to
136 determine demographic information of st udents: age, gender, native language, onset of learning English, motivation fo r learning English, previous use of computers, any known special needs, and whether the respondent was retained or skipped grade levels (see Appendix I). Th e questionnaire was modified from ORelly (1999) and consisted of 22 items, sub-divided into seven sections. There were four general headings in the questionnaire: Demographics, Background, Foreign Language, and Technology. The Demographics section contained items that extracted information on gender, age, grade level, and school type. The Background section solicited informati on on native language, special needs, and whether participants r epeated grade levels. The Foreign Language category elicited information on native language, fo reign languages bei ng learned, length of time studying English-as -a-foreign-language, levels of motivation for studying, and amount of exposure to the English language outside of t heir classrooms and countries. Technology, the final por tion of the questionnaire, requested background information on the participants computer usage, reasons for using computers, level of comf ort, and previous experienc es with discussion boards and chat programs. All item s either provided an option to check off yes/no answers, complete fill-in-the-blank it ems, write open-ended responses, or to respond to multiple-choice items. Qualitative Task Instrument Based on the literature review and curr ent research findings, a similar twoway task (see Appendix C) within dyads (Mackey et al., 2003; Oliver, 2000) was used. The two-way information gap task was used within an online synchronous
137 environment using the chat tool MSN Messenger (Microsoft Corporation, 2005a). The two-way task used in the current study was similar in type to those used in other feedback studies conducted by Macke y (1999), Oliver ( 1995), and Silver (2000). The task also complements Chape lles (2001) criteria on tasks (i.e., learning potential, learner fit, me aning focus, authenticity, impact, and practicality). The two-way task included 10 different pi ctures that, as a whole, depicted a story. Each pair of students received five different pictures from the set of 10. With their dyad member, the students were to place the pictures in the correct order according to the time sequence of events depicted on the pictures. As such, each member within a dyad was missi ng information that the other member of the dyad had. Thus, they were to communicate with one another to describe their pictures for the purpose of determining the sequence of events. Tool for Collection MSN Messenger was used as the text-bas ed discussion (chat) tool for the two-way task to be implemented. MS N Messenger is available as a downloadable program (Microsoft Corpor ation, 2005a) or as an online web version (Microsoft Corporation, 2005b). MSN Messenger was chosen because of its practicality (i.e., it is available to all worldwide users without cost), ability to download or use the web version, and it s usability on most operating systems and platforms (Microsoft Corporation, 2005a 2005b). It also allows the users of the program to see when their online chat partner is typing, by seeing a message at the bottom of thei r screen that says, user name is typing; therefore, for the
138 most part, it can mirror conversations that take place in face-to-face discussions. Other programs, such as InterChange, I RC, ESL Webchat, and ytalk, which have been used in previous research studies (Beauvois, 1992, 1997; Iwasaki & Oliver, 2003; Kelm, 1992; Kern 1995; Negretti, 1999; Pellettieri, 2000) also were considered; however, because of the cost, risk of invasions of outside speakers, constraints on downloading UNIX based programs on school computers, and easiblity of use (see for example, Baron, 2003; Orthmann, 2000), MSN Messenger were chosen as being best compat ible with the design, accessibility, and participants in the study. Qualitative Interview Instrument Interviews were conducted with 10 par ticipants who were chosen based on stratified purposeful sampling (Patton, 2002). The semi-structured interview was designed to provide participants the opportunity to add information and ideas, while allowing the researcher to facilitate the interview based on the participants responses. The pre-defined quest ions asked participants on their: (a) impressions, barriers, and advantages of co mpleting the activity within the online synchronous environment, (b) reaction to t heir partner in terms of language level, attitude, and knowledge, and (c) perceptions on the usefulness of completing an activity online. The interviews were approximately 10 minutes per each interview participant. Interviews were audiotaped and verbatim transcripts were created. The researcher then translated the intervie w transcripts into English for further inter-rater analysis. A colleague, who is also an educator in the Slovenian public schools, reviewed the original Slovene and translated transcripts for accuracy.
139 The colleague received both the original transcription in the Slovene language and the translated version from the resear cher. She verbally completed a reverse translation (i.e., in front of the researcher of this study verbally read the English translation, provided the Slovene equivalency and reviewed the original manuscript for accuracy). The review entailed a 100% consistency between the interview transcript and the translation. The researcher then reviewed the translations after one-week, which also entailed a 100% consistency score. Interview prompts and identified themes are further explored in Chapter 4. Pragmatist Procedure A sequential mixed methods study was used, in which both quantitative and qualitative approaches were utilized in the research process. In this design both the quantitative and qualit ative phases had an equal status. That is, they represented a QUAN QUAL sequential balanced design (Morse, 2003). This design typically is used when the quant itative and qualitative methods are conducted sequentially. The first phase is a quantitative sample followed by another qualitative sample. T he qualitative data are used to provide explanation of the quantitative results (Morse, 2003). More specifically, because both quantitative and qualitative models were int egrated to complement the research, thereby supports the pragmatist worldv iew of mixed met hods (Maxcy, 2003; Tashakkori & Teddlie, 2003). For pragmatis ts, the research question drives the method used. In addition, the strengths of both quantitative and qualitative methods are being utilized within mixed met hods as well as giving the researcher
140 the opportunity to use various ways in answering the questions at hand (Johnson & Onwuegbuzie, 2004). The present study includes participant s who were adolescent learners with or without special needs. Because of the participant characteristics and the research purposes, the pragmatist view of mixed methods is most suitable. Additionally, the pragmatist philosophy also was relevant for the studys research design in that it allowed in tegration of other theoretical or conc eptual frameworks. Such an allowance gave the researcher an opportunity to discover and explore findings as they emerged. Within the sequential mixed methods design, data were collected and analyzed separately; however, the results of both types of data were compared by the researcher at the inference st age (Erzberger & Kelle, 2003; Miller, 2003; Onwuegbuzie & Teddlie, 2003; Tashakko ri & Teddlie, 1998, 2003). Greene, Caracelli, and Graham (1989) lists five purposes of mixed methods studies: triangulation (seeking convergence of results), complementarity (seeking clarification of phenomena of results of one method with results from another), initiation (discovering paradoxes), devel opment (one method informs the other), and expansion (adding breadth and scope to a study). The presents study purpose of mixed methods was to develop an initial framework by examining: (a) participants with special needs qualitativ ely and the overall pattern of online corrective feedback quantitatively, and (b) extreme cases with follow-up interviews that this would add to the current body of knowledge of SLA, online communication, as well as provide po ssible new knowledge of second language
141 learners with or without special needs. T hus, the researcher hoped to integrate the findings by incorporating the strengt hs of both approaches. More specifically, the purpose of a mixed methods research design was complementarity (Greene et al., 1989). Finally, there is a paucity of research in second language acquisition and computer-mediated communi cation incorporating mixed method or mixed model methodologies. Thus, it was hoped that this study would add to the existing body of literatur e in the area of SLA. Research Design This research design utilized a mixed-methods or integrative research framework (Johnson & Onwuegbuzie, 2004) that includes mixed methods sampling strategies (Kemper et al ., 2003) situated within a pragmatist philosophy. For the present study, mixed methods was defined using Johnson and Onwuegbuzies (2004) definiti on, that mixed methods is the class of research where the researcher mixes or combi nes quantitative and qual itative research techniques, methods, approaches, concepts or language in a single study (p. 17). Because the present study int egrated both quantitativ e and qualitative practices, the pragmatist approach was considered most appropriate. Another factor in determining the appropriatene ss of designs within the pragmatist approach stemmed from the nat ure of the research questions. As in any study, the research questions are to be consi dered the most fundamental. Therefore, depending on the type of research questi ons guiding the study, appropriate
142 designs should be selected to complement them (Johnson & Onwuegbuzie, 2004). The overview of the research desi gn processes is depicted in Figure 6. The pragmatist philosophy allows the re searcher to examine the data from both a logico-deductive and a heuritistic-in ductive approach. Thus, the research questions, situated within the theoretical framework of an Interactionist perspective (Long, 1996), guided the quantitat ive portion of t he study. At the onset of data coding, the codebook (Append ix J) was used to code the data numerically. The data were coded bas ed on errors and corrective feedback patterns from previous research studi es (Castaeda, 2005; Doughty, 1994; Gass & Varonis, 1985; Iwasaki & Oliver, 2003; Lightbown & Spada, 1990; Long, 1991, 1996; Long et al., 1998; Lyster & Rant a, 1997; Mackey, 1999; Mackey & Oliver, 2002; Mackey et al., 2003; Morris, 2002, 2005; Oliver, 1995, 1998, 2000, 2002; Pellettieri, 2000; Pica et al., 1985; Schac hter, 1991; Sotillo, 2000). After coding, all dyads that included st udents with special needs were further analyzed using conversation analysis. Conversation analys is (CA), the qualit ative stage of the current study, also complemented the pragmatist philosophy, thereby allowing the researcher to approach the data without a priori assumptions or questions (Pomerantz & Fehr, 1997). Therefore, the researcher stepped outside the Interactionist theoretical framework ( Long, 1996) and focused on the data itself. A general question was posed to guide t he researcher; however, as per the assumptions of CA, it allowed the re searcher to examin e the data without predetermined theories and have the questions arise out of the data (Psathas, 1995).
143 Figure 6. Research design. Figure 6. A visual representation of the re search design for the current study incorporating both quantit ative and qualitative techniques and methods. Pragmatist Philosophy Heuristic-inductive QUAL Logico-deductive QUAN
144 Or, as stated by Pomerantz and Fehr (1997), it rejects the use of investigator-stipulated theoretical and c onceptual definitions of research questions (p. 66). Mixed-method designs can vary depending on data collection implementation, priority of research methodology, stage of data integration, and theoretical perspective (Creswell, Plano Clark, Guttmann, & Hanson, 2003). Based on the nature of the research questions, design, type of data, and guidelines for data collection implantation, this study used a balanced sequential mixed-method design (Creswell et al ., 2003; Tashakkori & Teddlie, 1998). A sequential mixed method design is defined by the collection and analysis of quantitative data followed by the collect ion and analysis of qualitative data. Priority is typically given to the q uantitative data, and the two methods are integrated during the interpretation phase of the study (Creswell et al. 2003, p. 223). Creswell et al. note that the s equential explanatory designs may be used to characterize individuals along certain tr aits of interest related to the research question. (p. 227). However, Creswell et al. caution that t he main weakness of this design is the length of time involved in data collection to complete the two separate phases (p. 227). However, Creswell et al. further note that by giving equal priority to both the quantitativ e and qualitative study may be more appropriate. Furthermore, to alleviate limit ations within the qualitative study both data coding and data interpretation, interand intra-rater reliability was used to assess the consistency of the coding.
145 Finally, mixed methods had been chosen to add to the current field of second language acquisition by combinin g both methodologies, thereby adding to the development of t heories. As Markee (1994) argues, the hermeneutic scientific traditions should not be deemed to be less serious, empirical, rigorous or even less informing, but that the qualitat ive and quantitative studies are in reality complementary ways of creating new knowledge (p. 91). As such, by integrating both methods, t he researcher hoped that the integrative nature would not only add to the field of SLA, but woul d also promote further research using integrative methods. Data Collection Procedures After approvals to conduct the study were obtained from the schools, the researcher contacted the schools to discuss the research study, technical requirements, number of par ticipants, and conduct a site visit. At this time, the researcher and teacher discussed requirem ents for participating in the study, the researcher requested the t eacher to distribute the informed consents to be signed by participants and parents, and possi ble dates for data collection were scheduled. The researcher requested two dates. The first date involved completing the questionnaire and practi ce session. The second date was scheduled for the actual data collecti on session. The informed consent and permission form were taken home for parents and participants to review, complete, and sign, which were subsequently returned to the researcher. After returning the consent/permission forms and completing the questionnaire, the students partook in the mandatory practi ce session. The second date set aside
146 Figure 7 Procedures of the study. I. IRB Approvals VII. Interpretation & Final Report VI. Data Analysis V. Stratified Purposeful Interviews with Selected Participants IV. Practice Session & Data Collection (Questionnaire and Task) III. Selection of Participants & Consent Forms II. Pilot Study Procedures
147 was for the actual session that took pl ace no more than two weeks after the practice session (see Figure 7 for data collection procedures). Before data collection would begin, the researcher created userids and passwords for the students to sign-on into MSN Messenger. The userids were unique to each participant and consis ted of alphanumeric symbols. The password was generic. In addition, befor e the practice and data collection sessions began, the researcher had alr eady entered the appropriate userids and passwords onto the computer termi nals. The purpose of entering the identification numbers was th reefold: (a) to ascertai n if registration of the identification numbers were su ccessfully completed, (b) to verify the validity of the passwords and userids, and (c) to match dyads online using predetermined identification numbers. Based on the expe riences of the pilot study that was conducted a year prior to the current st udy, these procedures allowed for more time to be allocated towards the task and for dyads to be already paired up via identification numbers. Students were r andomly assigned their identification numbers at the onset of collection and based on thos e identification numbers dyads were created (i.e., the student who received an identific ation number of 1a was automatically paired wit h the student that receiv ed an identification number of 1b and so forth). The transcripts of the data received fr om MSN Messenger included all entries by the learners. All student names or other identifiable information were deleted by the researc her and replaced by the aforementioned identification numbers. A sample of a c hat screen is available in Figure 8.
148 Figure 8 Screen-shot chat screen. Figure 8. Screen shot of MSN Messengers chat function. The typing area is where messages are created and sent into the center area, which is common to both members in the dyad. Chat area recipient recipient Typing Area
149 Upon entering the computer laboratory, the students were given identification numbers at random and instructed by the res earcher where to sit, in order to prevent dyad members from sitting too close to one another. The dyads were mainly matched by gender; however, due to (a) odd number of male/female pairs in class sections, (b) eliminating cert ain dyad members and restructuring dyads due to technology problems (e.g., com puter freezing, Inte rnet not working properly), or (c) less frequently, unwillingnes s to work with certain individuals, some dyad members were grouped into mixed gender dyads. Consequently, of the 104 total dyads, there were 56 female-female dyads, 23 female-male dyads, and 25 male-male dyads (for more information see Table 3).
150 Table 3 Overview of Dyad Members by Grade and Gender Gender Dyads / Grade Level Female Female Female Male Male Male Total Dyads n % n % N % n % Grade 7 9 16.07 12 52.17 11 44.00 32 30.77 Grade 8 11 19.64 3 13.04 2 08.00 16 15.38 Grade 10 17 30.36 2 08.70 5 20.00 24 23.08 Grade 11 19 33.93 6 26.09 7 28.00 32 30.77 Total 56 53.85 23 22.12 25 24.03 104
151 For both the practice and actual dat a sessions, the students were given written and verbal instructions in Slov ene and English (see Appendix K and L), similar to instructions given in previous research studies (Iwasaki & Oliver, 2003; Pellettieri, 2000), and asked if they had any additional questions before proceeding. The instructions consisted of us ing only English during the activity, to be as accurate as possible, not asking t heir peers or teachers oral questions, not preparing drafts of answers on a piece of paper or on any other platform, and focusing on the task. After all questions had been answered, two similar two-way communicative tasks (see Appendix C) we re distributed to the students. One version was distributed during the prac tice session and the second version during the actual data session. Both the researcher and the teacher we re present for both the practice and actual sessions, except on two occasions when one teacher returned towards the end of the period and another had to leave to attend to another class. From the pilot study analysis that was conducted earlier in the school year, the teachers presence was seen as beneficial, inasmuch as students were more focused on the task instead of being distra cted by the researchers presence. In addition, the pilot study showed that approximately 25-30 mi nutes of the 45 minutes allotted were for actual task work The actual data session did reflect the experiences of the pilo t data, notwithstanding extern al variations, such as technology breakdowns and student relu ctance in completing the task. Five minutes before the end of class, the students were asked to finish and to move away from the keyboard. T hey were instructed to wait for the
152 researcher to approach their computers to save the data on a floppy disk and hard drive. The chat archives were sav ed in a Word document (.doc) format in order to preserve the emoticons (i.e., te xt format did not pres erve emoticons). The final piece of data collection included informal interviews with 10 participants after the two-way task had been completed. The interview included: (a) three learners with special needs, (b ) one dyad member who chatted with one of the learners with special needs, and (c) three low learners and three high learners (i.e., extreme cases). These dy ads were chosen based on number of turns, words, error leve l, corrective feedback moves, and class standing. The purpose of the semi-structured interview wa s to solicit additional data to include in the discussion of the participants per ceptions of the effectiveness and usefulness of the conferencing tool in En glish language learning. Data collected from the interview were transcribed into Microsoft Word. A colleague, who also was a teacher of an elementary school, was asked to review the interview and the transcriptions for accuracy. The resear cher then translated the interview from Slovene into the English language. After all data collection had been comple ted, the data from the two-way task were imported into Microsoft Excel for coding. The Excel workbook included formulas automatically to differentiate and sum the number of corrective feedback types, error types, and repair. Fu rthermore, each participants turns were tabulated in Microsoft Excel for num ber of turns, word s, error level, corrective feedback moves and class standi ng in order to determine extreme
153 cases for the interview. The interviews were transcribed into Microsoft Word, where themes were extracted by both the researcher and inter-raters. Next, all completed questionnaires were inputted into Survey Gold (Golden Hills Software, Inc., 2005/2006), a downloadable Internet software program that specializes in survey co llection and analyses. Detailed information on the coding processes, units of analysis and the analyses of data collected via online chats are presented below under the Data Analys is Procedures section. Unit of Analysis The unit of analysis from the chat transcripts represents a modified version of Lyster and Rantas (1997) e rror treatment sequence (Figure 9). Only the first two units were examined: peer error and learner corrective feedback. Peer response (i.e., uptake) will be furt her examined in a subsequent study.
154 Figure 9 Present error treatment sequence.
155 Analyses were conducted based on turn s (see Figure 10). For the purpose of the study, a turn is defined as when a message is composed and sent into the chat room. First, initial peer errors cont aining at least one form were tallied and calculated. Next, the ty pe of error was determined based on pre-existent categories (Lyster & Ranta, 1997; Morri s, 2005; Oliver, 1995) and one category was left open for any emergent categories that might not fall under the six predetermined categories. Peer responses to the non-target language form were rated as ignored or provided with correct ive feedback. If prov ided, all types of corrective feedback were identified and cl assified according to the corrective feedback codebook (Appendix J) then talli ed using the coding sheet (Appendix M), and then evaluated. Total instances of corrective feedback were tallied and evaluated for quality (target-like vs.nontar get like). Finally, after feedback was provided, the peers response to the f eedback was examined as (a) ignored, (b) no opportunity given to respond, or (c) response to peers feedback. If the feedback was ignored or no opportunity wa s provided, then the response was coded as topic continuation as per the unit of analysis model. If feedback response was acknowledged then the peers feedback was classified, as either incorporation (repair), needs repai r, or an emergent category.
156 Figure 10 Coding process.
157 For ease of coding for both the ra ter and inter-raters, the Corrective Feedback Coding form (Appendix M) was ent ered into Microsoft Excel. Additionally, formulas were included automat ically to sum totals of each column and tally different types of errors and corrective feedback types within each worksheet for each grade level separately. Finally, a separate worksheet was created to calculate the sum of all tota ls (i.e, error and corrective feedback types) across all grade levels. The results of these frequency counts were used for descriptive accounts and to assist in the interpretations of the results. For further inferential statistics (chi -squares, Fishers exact tests, and Multiple analysis of variance [MANOVA] with discriminant analysi s), all instances of corrective feedback moves or error moves within one dyad were collapsed into a count of one incidence. This was nec essary for the sole purpose of not violating the assumption of independence (Glass & Hopkins, 1996; Onwuegbuzie & Daniel, 2003; Stevens, 2002). If frequen cy counts of each corrective feedback or error move within one dyad had been used for the analysis, the independence assumption would have been violated bec ause one type of corrective feedback or error type provided by a member dy ad might influence the corrective feedback and/or error types provided by the peer dy ad. Some studies in this area, that involve the use of inferential statisti cs (e.g., Blake, 2000; Mackey et al., 2003; Morris, 2005), are flawed by the fact that the independence viol ation is violated by using an incorrect unit of analysis. Therefore, the frequencies were collapsed either to zero or one instance of co rrective feedback within each dyad, or for error counts a zero or one instance was calculated within each dyad. These
158 frequencies within each grade level then were used for inferential statistics and analysis. Data Analysis Procedures Data collected for analysis consisted of transcripts created from the online tasks, which were originally saved in a Word document format and then imported into Microsoft Excel for coding. Each question was analyzed separately. As the research objectives reflec t, findings on learner uptak e will not be presented. Learner uptake is an important variable t hat will be reported in follow-up studies; however, data on learner uptake were coded simultaneously. Onwuegbuzie and Teddlies (2003) framework for analyzing data in mixed methods studies was used as a guide for t he analysis of data in the current study (see Figure 11). Onwuegbuzie and Teddlie s data analysis framework includes data reduction (Stage 1), which involves re viewing, organizing, and reducing data that were obtained from the data collection phase. Next is data display (Stage 2), which involves visual representing the data, via tables, graphs, diagrams, lists, and so forth. Then, data transformation (S tage 3) might follow, which includes quantitizing (i.e., converting text data in to numerical forms) or qualitizing (i.e. converting numerical data in to qualitative codes) the data. The subsequent three stages (i.e., data correlation, conso lidation, and comparis on) occur depending on the types of data collected. When bot h quantitative and qualit ative data are collected for each research participant, then data correlation (Stage 4) occurs. However, if a new set of variables or c onsolidation of variables from two data types are the focus of the study, then data consolidation (Stage 5) occurs.
159 Depending on the research focus, data corre lation or data consolidation need not occur. Another option might be data compar ison (Stage 6). Data comparison is used when the intent is to compare different data sources (Onwuegbuzie & Teddlie, 2003). The last stage in the dat a analysis framework is data integration (Stage 7). Data integration may be the last stage following dat a correlation, data consolidation, or data co mparison or it might follow directly after data transformation. Within this last stage of data analysis data integration occurs when all data are integrated into a c oherent whole or two separate sets of coherent holes (quantitative or qualitat ive) (Onwuegbuzie & Teddlie, 2003, p. 377). The present study used the follo wing mixed method stages of data analysis of the above-described fram ework (see Figure 11 for a visual representation): Stage 1: Data Reduction Stage 2: Data Display Stage 3: Data Transformation Stage 6: Data Comparison Stage 7: Data Integration The data transformation (Stage 3), data display (Stage 2) and data reduction (Stage 1) were reversed, respecti vely, while data correlation (Stage 4) and data consolidation (Stage 5) were not used for this study, namely because these stages did not fit within the framew ork of the current study. The purpose of the current study was co mplementarity (Greene et al., 1989). As such, data
160 correlation (Stage 4) was eliminated by the researcher, because this stages main focus was to triangulate (Greene et al., 1989). Additionally, the qualitative data in this study did not include all of the participants of the quantitative data, which is recommended for data correla tion (Onwuegbuzie & Teddlie, 2003). Furthermore, data consolidation (Stage 5) entails consolidating data to create new variables (Onwuegbuzie & Teddlie). In lieu of Stages 4 and 5, the data comparison stage (Stage 6) was incl uded, where data are compared for triangulation, complementar ity, or initiation purpos es (Greene et al., 1989). Johnson and Onwuegbuzie (2004) and Onwuegbuzie and Teddlie (2003) contend that even though the stages are s equential they are not linear and the analyst can skip or chose only the most applicable stages. The first step of the mixed methods data analysis process entailed data transformation (Stage 3). This involved orga nizing the data collected as a result of synchronous chat into Microsoft Excel using the codebook format (Appendix J), and transcribing the interviews. Quant itizing data was based on the coding process (Figure 10). The data from syn chronous chat went through an initial review of all turns, while organizing and copying data into Microsoft Excel. Those dyads that chatted solely in the L1 (i.e ., Slovene) or did not participate in the practice session were eliminated from t he study. Consequently, five dyads in total were eliminated. In addition, during this stage, the synchronous chat data were coded using the codebook and code form (Appendix M). Data transformation included quantitizing synchr onous data. Statistical procedures and analyses were used for the fi rst three questions.
161 Within the same stage, that is the data transformation stage, conversation analysis (Markee, 2000) was used to analyze the final research question. Conversation analysis (see Qualitative Analys is section of this chapter for more information) is defined as: a form of analysis of conversation data (ACD) that accounts for sequential structure of talk-in-interaction in terms of interlocutors real-time orientations to the prefer ential practices that underlie, for participants and consequently also for analysts, the conv ersational behaviors of turn-taking and repair in different speech exc hange systems. (Markee, 2000, p. 25) The next stage in the data analysis framework included data display (Stage 2). The quantitized data were displa yed in the form of tables for each separate question; whereas t he qualitative data were presented in rubric form. Quantitative analysis was based on codi ng for corrective feedback and error types. Qualitative analysis included conver sation analysis of the three special need learners and their dyad members, as well as, interviews with seven additional participants. The interv iew data were analyzed using Miles and Hubermans (1994) matrix building. Data from the interview did not address any specific research questions. Its purpose was to add to the data interpretation phase of the data analysis. Within this phase, data were submi tted three times to two additional interraters. The first time was at the initial stage of coding using the unit of analysis of the chat transcripts, as well as, IRF patterns and adjacency pair classification for conversation analysis, and theme identification for interview
162 themes. After this stage, coded data fo r quantitative and qualitative analysis were redefined based on discussions among the interraters. The fi nal analysis was completed when all themes were refined and a final evaluation of the codes was reviewed. Both inter-raters were the researcher s colleagues, had experience coding with linguistic data, and were familiar with the error-sequence patterns. One of the inter-raters had previously coded data using a modified version of the codebook in this study. The other inte r-rater was an instructor of English linguistics at a large southeastern univers ity. Each inter-rater coded 13% of the quantitized data. Initial relia bility for each inter-rater was calculated at 90.88% and 95.88%, respectively. The researcher and inter-raters discussed the discrepancies. As a result, the init ial codebook was modified and after subsequent coding of 14% of the data for quantitative analysis, a 99.64% and 99.85% interrater reliability was achiev ed by each interrater, respectively. Intercoder reliability was calculated using Miles and Hubermans (1994) formula, where intercoder reliability was calculat ed as the number of agreements divided by the total number of agreem ents plus disagreements or: number of agreements reliability = total number of agreements + disagreements (Miles & Huberman, 1994, p. 64) Additionally, the data went through three intra-rater checks of the researchers coding. First, immediatel y after initial coding was completed; second, after initial inter-rater feedback was submitted back to the research; and
163 the final intra-rater check was complet ed before the data were subjected to further statistical analysis, or approximatel y three months after initial data coding. Intra-rater reliability also was calc ulated using Miles and Hubermans (1994) intercoder reliability. Reliability scores were 90.13%, 98.19%, and 99.58%, respectively. All final discrepancies of intrarater and interrater scores were reviewed with the interraters. Inter-reliability level for the IRF sequences and adjacency pair was calculated at 98.4% and 97.8%. After additi onal discussions with the interraters a final 100% inter-rater reliability was ac hieved. In addition, the researcher calculated an intra-rater reliability appr oximately two weeks after IRF sequences and adjacency pairs were determined. A relia bility score of 100% was achieved. For the interview data, the whole transcript was reviewed by both interraters. An initial 92.4% and a 94.0% reliability score was calculated. After another round of discussions among the res earcher and interraters, a 95.3% and 96.5% reliability level was achieved. After reviewing discrepancies the interreliability level was calculated at 100%. Finally, the researcher calculated an intra-reliability score for her coding, appr oximately two weeks after inter-reliability was calculated. An intra-reliabili ty score of 98.7% was achieved. Next, data reduction (Stage 3) co mmenced. Data reduction included inputting the questionnaire into Survey Gold (Golden Hills Software, Inc., 2005/2006). In addition, quantitized data were input and descriptive statistics were calculated using SAS (SAS Institute Inc., 2004).
164 Following, was the data comparison st age (Stage 6), where the qualitative data, the quantitized data, and interview data were compared for consistencies, outliers, and emerging themes. Next, was the integration stage (Stage 7) wherein both data types were integrated as two separate wholes (Onwuegbuzie & Teddlie, 2003). Within this stage, the l egitimation process also began. Once the researcher believed that the data were legitimate, in other words, that there were no other possible explanations, then bot h the quantitative and qualitative data were interpreted and the final report wa s written. To ensure the process of legitimation, the two in terraters who coded during the data display stage reviewed both data interpretation and the conclusions. Results also were submitted to the participants teachers fo r final review (see Figure 11 for an overview of the research process).
165 Figure 11 Research implementation process. Two-wa y task w/ text chat III. Data Interpretation QUAN + QUAL IV. Final Report / Discussion Final checks with participants and interraters Data Integration of Q UAN Data Integration of Q UAL Statistical Results IRF & A d j acenc y c D ata R educt i o n b. Data Display I nter/Intraraterchecks d. Data Comparison Coding/memosinterview I. Data Collection Quest i o nn a ir e II. Data Analysis a. Data Transformation Quantitizing Conversation Analysis IRF & Adjacency Pairs Purposeful Interviews Coding
166 Quantitative Analysis Procedures After the data had been quantitized and cod ed by both the researcher and the inter-raters, the data were analyzed using SAS (SAS Institute Inc., 2004) software [version 9.1.3]. SAS was used for descriptive statistics, measures of central tendency, standard deviation, chi-squa res, and four Fishers Exact Tests. SPSS version 11.0.1 (SPSS for Windows, 2001) was used for the multivariate analysis of variance (MANOVA) and subs equent discriminant analysis. In addition, the data were exam ined for deviation from no rmality by examining the skewness and kurtosis coefficients. Afte r this assumption check, statistical analyses were used to address the research questions. For ease of reading, the applicable statistical method is described under each null hypothesis, as well as being available in Table 4, which descri bes the coding process and statistical procedures in relation to the research question.
167 Table 4 Research Questions with Data Analysis Procedure Research Question Research & Null Hypothesis Analysis Type of Analysis 1. What is the difference in the incidence of corrective feedback in online synchronous environments provided by adolescent EFL learners to other dyad members as a function of grade level? Null Hypothesis 1. There is no difference in the incidence of corrective feedback in synchronous online environments provided by adolescent EFL learners to other dyad members as a function of grade level. Research Hypothesis 1. There is a difference in the incidence of corrective feedback in synchronous online environments provided by adolescent EFL learners to other dyad members as a function of grade level. Step 2b of coding process (Total Tally of Grade) 4 x 2 Chi-Square (12) Effect size measured by Cramers V
168 Table 4 Research Questions with Data Analysis Procedure (continued) Research Question Research & Null Hypothesis Analysis Type of Analysis 2. What are the differences in the nature of corrective feedback in online synchronous environments provided by EFL learners to other dyad members as a function of grade level? Null Hypothesis 2. There is no difference in the relationship among the type of corrective feedback in online synchronous environments provided by adolescent EFL learners to other dyad members and grade level. Research Hypothesis 2. There is a difference in the relationship among the type of corrective feedback in online synchronous environments provided by adolescent EFL learners to other dyad members and grade levels. Step 2a of coding process (Types of Corrective Feedback P2) 4 x 4 Chi-Square Effect size measured by Cramers V and MANOVA discriminant analysis effect size as measured by 2
169 Table 4 Research Questions with Data Analysis Procedure (continued) Research Question Research & Null Hypothesis Analysis Type of Analysis (12) What is the relationship between the type of learner errors and type of corrective feedback in online-synchronous environments provided by EFL learners to other dyad members? Null Hypothesis 3. There is no relationship between learner error and type of corrective feedback in online synchronous environments provided by adolescent EFL learners to other dyad members and grade level. Research Hypothesis 3. There is a relationship between learner error and type of corrective feedback in online synchronous environments provided by adolescent EFL learners to other dyad members and grade levels. Step 2d of coding process (Error with Corrective Feedback Type) 4 Fishers Exact Tests Cramers V
170 Table 4 Research Questions with Data Analysis Procedure (continued) Research Question Research & Null Hypothesis Analysis Type of Analysis 4. What interactional conversation characteristics by dyad members are present in online-synchronous environments when one or more of the interlocutors are learners with special needs? Characteristics in terms of corrective feedback, errors, and responses to previous turns or previous requests, questions, prompts, invitations, and so forth. 1 & 2 of coding process Conversation Analysis (using IRF & adjacency pair coding)
171 Null hypothesis 1. There is no difference in the incidence of corrective feedback in synchronous online environm ents among adolescent EFL learners working in dyads across grade levels. To te st this hypothesis, first all turns were counted for total number of learner turns and turns containing errors. Then, within each dyad, the number of incidences of corrective feedback was reduced. If there was an occurrence of corrective feedback within one dyad then a frequency of 1 was entered; or if there were not any occurrenc es, then a frequency of 0 incidences of corrective feedback move s was entered for each dyad. The sums of incidences of corrective feedback fo r each grade were used for statistical analysis. A chi-square analysis was used to co mpare the amount of corrective feedback across the grade levels. The i ndependent variable was grade level and the dependent variable was amount of corr ective feedback received when errors occurred. This yielded a 4 x 2 chi-squar e contingency table. Huck (2004) notes that a chi square analysis is appropriate for comparing categorical data. Assumptions that were reviewed and considered before conducting the chi-square were that the data r epresented independent observations and mutually exclusive row and column variabl es that include a ll observations (Glass & Hopkins, 1996). The researcher care fully examined each category before a chi-square was applied. The frequencies of each cell size as well as the sample size of learners providing corrective feedback were determined after the coding. The observed and the expected frequencies were computed and the effect size was measured using Cramers V The observed frequency was compared with
172 the expected frequency. If the observed 2 was larger than the expected frequency then the null hypothesis was reject ed at the .05 level. If rejected it could be concluded that t here is some association between the two variables. Null hypothesis 2. There is no relationship between the type of corrective feedback in online synchronous environment s provided by EFL learners to other dyad members and grade level. After the total number of corrective feedback incidences was tallied and converted into a percentage score, the total number of learner turns with error receiving corre ctive feedback was coded. Again, these were collapsed within dyads to either zero or one incidence of corrective feedback and error. The types of corrective feedback were coded and sorted under the following categories: explicit correction, recasts, elicitation, metalinguistic feedback, clarification request, repetition, and emergent. Lyster and Ranta (1997) found t hat four types of corrective feedback (i.e., elicitation, metalinguistic feedback, clarification request, and repetition) lead to learner repair. These four types were considered as one category, namely negotiation of form Similarly, Castaeda (2005), in her study of online corrective feedback moves by instructors of Spanish-as-a-fo reign-language, also collapsed the four types of corrective feedback leading to repair; however, she categorized them under the category of opportunity to negotiate. Lyster (2004) and Lyster & Mori (2006) also collapsed these four feedback types into prompts Conversely, explicit correction and recasts were found not to lead to students repair, or led to a low rate and, therefore, were left as separate categories. Emergent is a category that was left open for any types of corrective feedback that might have
173 arisen during the coding process that were different from t he a priori defined categories. To test Null Hypothesis 2, the c oded and tabulated data on the frequency of corrective feedback was used to conduc t a 4 x 4 chi-square analysis via SAS and a MANOVA via SPSS to evaluate the overall level ( n = 104 = total number of dyads) and the corrective feedback level ( n = 62 = total number of each corrective feedback incidences), respecti vely. The dependent variables were the four categories (i.e., explicit corre ction, recast, negotiation of form, and emergent) and the in dependent variable was the grade level (i.e., Grade 7, Grade 8, Grade 10, and Grade 11). For the chi-square, a 4 x 4 contingen cy table was used. A chi-square had been chosen to determine the relations hip between grade level as the independent variable and corrective feedbac k type as the dependent variable. Assumptions that were accounted in t he previous null hypothesis (independence and frequency counts in each cell) also were reviewed. All assumptions were met and a 4 x 4 chi-square analysis was conducted. A MANOVA using SPSS version 11.0.1 was used to determine whether the four grade levels di ffer across the four dependent variables of corrective feedback. The independent variable was the grade level (7, 8, 10, and 11) and the dependent variables were the different types of corrective feedback (i.e., explicit correction, recast, negotiation of form, and emergent). A MANOVA rather than an analysis of variance (ANOVA) was c hosen to increase statistical power. The alpha level for the statistical test was set at .05. When the results were
174 statistically significant ( p < .05), the effect size was measured via 2, and the data were computed and interpreted to as sess the practical significance of the results (McLean & Ernest, 1998). Any statisti cal significance re sulting from the MANOVA led to a discriminant analysis to identify which corrective feedback types discriminated the four groups (Cody & Smith, 1997). Null hypothesis 3. There is no relationship be tween learner error and type of corrective feedback in online syn chronous environment s among adolescent EFL learners working in dyads across grade level. To test this hypothesis, four Fishers exact tests were computed, one for each grade leve l. The independent variable was type of error (i .e., grammatical, lexical, typographical/spelling, and usage of L1) and the dependent variable was corrective feedback type. The Bonferonni adjustment was used to control for Type I error. Specifically, each Fishers exact test was conducted at the .0125 (i.e., .05/4) level of significance. In addition, quality of peers feedb ack was examined and categorized according to whether the corrective f eedback was target-like or nontarget-like. This was examined holistically and added to the overall findings of the study. Additionally, the total number of utter ances per grade level was measured to indicate the quantity of chat. Qualitative Analysis Procedur es of Conversation Analysis Roger and Bull (1989) defi ne conversation analysis as examin[ing] the procedures used in the production of ordi nary conversation (p. 3). Conversation analysis (CA) is used to understand structures of conversational action and members practices for conversing (Hopper, Koch, & Mandelbaum, 1986). CA
175 also resists final categorization and codi ng to preserve detail that would be lost through such processes (Hopper et al., 1986). Further, CA is situated within ethnomethodology (Roger & Bull, 1989) and combines both hermeneuticdialectic and logico-analytic perspecti ves (Heritage, 1987; Markee, 2000; Mehan, 1978). Within SLA, Firth and Wagner (1997) argued that the field of second language should be expanded in that SL A theory needs a more emic-focused research within talk-in-interaction. Gass (1998), Kasper (1997), and Long (1997) contended that conversation analysis fo cuses on language use (i.e., social interactions) and not on acquisition (i .e., cognitive processes). However, investigations within conversational pr actices (i.e., turn-taking, repair, sequencing) are processes that are both social and cognitive (i.e., socially distributed cognition). Markee (2000) argued that because SLA studies examine such processes CA would be a viable as a, methodological arsenal of the sequential and other resources that speakers use to modify each others talk and thereby to comprehend and learn new languages (p. 32) and would play directly into the research program out lined by Long (1985) on the role played by comprehensible input in SLA (Markee, p. 32). CA does not develop arguments on a priori theory (formal, constructivist, nomothetic); develop argument s based on quantitative data of frequency, or lead to generalizations (Heritage & Atkinson, 1984; Markee, 2000; Negretti, 1999; Sacks, Schegloff, & Jefferson, 1974; Tarone, 1994). Within CA, a turn is defined
176 where one speakers turn is beginning as the others turn end (Sacks et al., 1974). Turns are constructed in relations hip to previous and subsequent turns. CA was chosen because it: (a) allows for analysis of turns rather than utterances (Sacks et al., 1974), (b) allows the data to change, adapt, or modify the questions (Heritage & Atkinson, 1984; Markee, 2000), and (c) as Tarone (1994) argues, CA Show[s] what successful input looks like for a single learner in a very particular context. What it cannot show is that successful input always looks this way for all learners in all contexts (p. 327). The heuristic-inductive approach of CA allows the researcher to integrate the pragmatist philosophy of centering the focus of t he study onto the research purpose. Certain evidence shows that due to the interactional context of chat, that text based chat (vs. audio chat) is a unique communication tool that differs from both oral and written media (Negretti, 1999). Negretti (1999) argues that because of the unique structures that learners produce and the unique cont ext of the discourse, conversation analysis is the most useful and fruitfu l because such a hypothesis-generating method is a good way to begin the study of new interaction/acquisition situations (p. 76). For the current study, conversation analysis provided an opportunity for the researcher to explore in-depth t hose dyads where learners with special needs were included. The researchers objectives of the last question was to explore key types and relations among co rrective feedback, learners response, and type of corrective feedback. Next, CA provided the researcher the opportunity to review the data collect ed from individual pa rticipants without
177 generalizing, in which the data only could be generalized to the participant itself (Heritage & Atkinson, 1984; Markee, 2000; Negretti, 1999; Sacks et al., 1974; Tarone, 1994). CA also provided exploration of the nature of at-risk second language learners interacti on and feedback negotiations, by allowing the data to produce the questions. Finally, CA provi ded the opportunity for new discoveries to emerge. A guiding question highlight s the researchers inte rest at the onset, but also allows the researcher to change, adapt, or modify the ques tion. In addition, as per the limited number of partici pants, the results did not lead to generalizations, but rather to discovery of L2 acquisition (Negr etti, 1999) and any findings were limited to the participants t hemselves. The final question framed for qualitative analysis within this study wa s: What interactional conversation characteristics by dyad members ar e present in online-synchronous environments when one or more of the in terlocutors are learners with special needs? Interactional conversation charac teristics were extrapolated through initiation/response/follow-up sequenc es and adjacency pairs to determine corrective feedback moves, error types, and response to prompts. Markees (2000) articulation and assump tions of CA were used as a tool for analysis. The four assumptions underlying CA are: (a) conversation has structure; (b ) conversation is its own autonomous contextthat is, the meaning of a par ticular utterance is shaped by what immediately precedes it and also by w hat immediately follows it; (c) there is no a priori justification for belie ving that any detail of conversation,
178 however minute, is disorderly, accident al, or irrelevant; and (d) the study of conversation requires naturally occurring data. (Markee, 2000, p. 98) It was expected that the conv ersation in the chat would have structure (i.e., a. conversation has structure) in that turns would be initiated and responded with additional turns. The conversations we re preserved by the researcher and included all turns preceding and followin g the data examined, where none of the utterances was disregarded, (i.e., b. c onversation is its own autonomous context and c. no a priori justification). Finally conversation also is considered as naturally occurring (i.e., d. conversati on requires naturally occurring data) talk and is situated within real time (Negretti, 1999). Procedures for CA, as outlined by Markee (2000), first examine the prototypical examples (p. 99). Prototypic al examples involve examination of the data set as a whole and analysis based on qua litative research criteria. It is not meant for the data to be quantified; however, quantitat ive analyses may be used for follow-up research or for present ing regularities in numerical form. Prototypical examples are sequences of questions and answers or adjacency pair as described by Sacks and Schegloff (1973). However, as Negretti (1999) noted, adjacent pairs in online chats are more sequential and do not adhere to the time pattern of adjacent pairs (p 81), where turns in face-to-face conversations are serially located or adjacent to one another. Negretti (1999) found that most responses to initial tu rns were delayed or instantaneously mixed with other turns. As such, the flow of the conversati on is atypical in that a response may not appear immediately afte r the question posted. In this study,
179 participants with special needs were extract ed out of the initial data set and their turns were coded based on the error treatment sequence (see Figure 9), adjacency pairs, and initiation-response-fo llow up sequences (see Figure 12). Both adjacency pairs and initiation/response/follow-up (IRF) sequences were used as prototypical examples. Adjacency pairs (Sacks et al., 1974; Schegloff & Sacks, 1973) were used to extrac t insight into the function of the language. Any type of question, invitati on, request with an applicable response were considered adjacency pairs. Within adjacency pairs, turns were examined for sequential ordering and completion of turn s. For example, whether a question was followed-up by a response, a reques t or invitation was replied with an acceptance or denial. The initiation/response/follow-up (IRF) sequence (Mehan, 1985; Ohta, 1993, 1994, 2001; Sinclair & Coulthard, 1975) was used as a guide for the delayed turns in the chat room and to determine whether the turns went beyond a traditional adjacency pair in te rms of their complexity and relation to the type of error and corrective feedback types. IRFs we re used also to determine structure of conversation (Markee, 2000). The initiation turn can be a question or a statement and/or incl udes an error, the response is an immediate turn to the initiation and/or considered as feedback to th e error in the initia tion turn, and the follow-up is praise from the teac her and/or repair of the erro r in the initiation turn based on the feedback in the response turn. Figure 12 depicts possible content for IRF routines. However, because the third tu rn need not be evaluative in nature, the term follow-up (Sin clair & Coulthard, 1975) is used here rather than
180 the term evaluation (Mehan, 1985), which also was used in Ohtas (2001) study. The content of the follow-up turn vari es, depending on content of the response turn. Following drill or mechanical practice a follow-up turn is likely to contain a response to comprehending the previous turn.
181 Figure 12 Content of IRF routines. Figure 12. Content of IRF Routines modified from Ohtas (2001) study. Italic font indicates how IRF routines were considered within the current study. Arrows depict possible flow of IRF routines, but typical conversational routines do not necessarily follow this direction. Initiation Turn Question Error Response Turn Answer Corrective Feedback Follow-Up Turn Indication of comprehension (minimal or extended) Uptake Acknowledgment Initiation Turn Drill prompt Content Error Response Turn Response Content Feedback Follow-Up Turn Evaluation / Aligned assessment / confirmation question Uptake Acknowledgment
182 Within the IRF sequence, Initiation was coded when a statement or prompt was given to solicit a response from the peer whenever an error in the L2 occurred (see Appendix J for types of errors ). After this stage, the turns were examined for any responses or feedback gi ven on that particular error. If feedback and/or a response was given, t hen that particular turn was coded as response Next, the data were examined if the learner, who committed the error and was given feedback on that error, also incorporated the feedback and corrected her/his initial error. If correct ion was attempted, then that turn was coded as follow-up These coding steps are similar to the error treatment sequence, which is used as the unit of analysis in quantitative analysis. It was expected that the conversati on in the chat room would have instances of structure in terms of IRF sequences and adjacency pairs, wherein turns would be initiated and responded to by further turns; however, the researcher also believes that the sequences and/or pai rs would be less frequent and interrelated. The IRF sequences and adjacency pairs from the chat environment were expected to be dispersed in a visibly vertical sequence with overlapping turns being evident and presented. The researc her believed that when special need dyad participants were included, the IRF sequence, as well as adjacency pairs, would be more unstructured and incomp lete, wherein initiation within IRF sequences and questions (within adjacency pai rs) would lead towards infrequent responses and follow-up turns. As such, by using adjacency pairs and IRF sequence, the structure and nature of corrective feedback were examined.
183 Adjacency pairs and IRF sequences were identified using the whole chat transcript and not being limited sole ly to the immediate turns. Finally, after prototypical exampl es had been identified, data were examined to identify and/or corroborate claims and structures, as well as go through artificial falsification (Markee, 2000, p. 99). This entails the data being examined in identifying prototypical examples, corroborating data, and using outside data to strengthen further the results. This final step also corresponds to Seliger and Shohamys (1989) criteria for validity control of qualitative data and used in Negrettis (1999) study The criteria are: (a) data retrievability, (b) data confirmability by supporting assertions with examples from the collected data, and (c) data representativeness. As far as accessibility of data or data retrievability is concerned, data were easily accessible. When data collection was complete, the discussions were saved and printed. The results chapter of this study prov ides various examples of data confirmability. However, data repr esentativesness was more complex to determine. Data saturation or repr esentativesness might have been reached; however, it is speculative whether three participants with special needs identified in Grade 7 accurately represent the data. It is, however, addi tional information that should motivate further research. Be cause the focus of the research was on learner-learner dyads, the researcher did not participate or observe normal behavior in the actual chat environments during data collection. However, to reduce this limitation, rich examples we re provided to show representatives of the data by using various sections within the same grade level.
184 The original conversations of lear ners with special needs were preserved by the researcher. All turns were incl uded in the qualitativ e data analysis, and none of the utterances or turns was disregarded. Following Negrettis (1999) study, the present study also could be considered to be situated in a natural setting, where the medium used was part of a learners exposure to language learning in their regular classrooms. Fina lly, it was the researchers intent to focus on the data as they presented themselves and to generate any findings relevant to the participants using the data. In addition, two other raters, who were colleagues and familiar with coding classroom data, also were trained to code the data, as well as were provided the opportunity to negotiate with the researcher, whenever inconsistencies occurred. Qualitative Analysis Procedur es of Interview Protocol The interviews lasted approximately 10 minutes per participant and were audiotaped, and verbatim transcripts were created. The researcher then translated the interview transcripts into English for follow-up inter-rater analysis. The data were sorted, organized, and comp ared to establish themes. First, the researcher read through the complete transcript and developed initial themes. Second, each individual interview was re coded according to the original scheme. Third, the interview translated transcripts and preliminary themes were submitted to the inter-raters. The two inter-raters and researcher collaborated on the themes where an initial 92.4% and 94% in ter-rater reliability score was obtained for each inter-rater, respectively. The pre-determined interview schemes were compared with one another and fine-tuning of the interview themes occurred.
185 Fourth, another round of inter-rater c oding occurred to determine the newly established themes. Inter-rater reliabi lities rates of 95.3% and 96.5% were calculated for both inter-raters, respecti vely. Fifth, interview themes were developed as umbrella terms to capture students suggestions and reactions to their experiences. Finally, the researc her recoded the data to assess for final agreements. An intra-rater reliability of 100% was calculated, which established reliability of the coding. Summary An explanatory sequential mixed method research design was used to guide the data analysis procures. This study was guided by the following objectives on learner-learner feedback wit hin online synchronous environments: (a) to investigate the difference in incidences of corrective feedback between peers in online synchronous environments and, if so, to examine (b) the type of feedback and (c) the relationship between the error and feedback. To answer these questions the researcher decided on quantitizing the qua litative data of synchronous text-based chat. The data we re subjected to both descriptive and inferential statistical analysis. The fi nal objective was to (d) explore the interactional conversation characteri stics among learners with a documented special need. To respond to the latter question, conversation analysis, using adjacency pairs and IRF sequence, was used to analyze the distribution and types of occurrences. The results of this study represented a quantitative and qualitative description of correctiv e feedback within computer-mediated
186 communication, among peer dyads, where one of the interlocutors may or may not be documented with a special need.
187 CHAPTER 4: DATA ANALYSIS AND RESULTS Overview This study was designed to determi ne corrective feedback patterns among pairs working in a synchronous online environment. Furthermore, a particular interest also was to include learne r-learner dyads that have a documented special need. Corrective feedback was ex amined by using a two-way task within an online synchronous environment, which has been shown to result in corrective feedback within oral classroom discussi ons (Mackey et al., 2003) and in chat environments (Morris, 2005; Pellettieri, 2000). The research questions guided the design of the study and, thus, both quantitative and qualitat ive methodologies were employed to answer the three quantitative research questions and one qualitative question, thereby yi elding a mixed method design. Questionnaire Results Background information was collect ed from the participants using a questionnaire (see Appendix I). A total of 208 participants participated in the study and completed the questionnaire. The participants were from intact classes and were randomly assigned into dyads within the classes. One hundred and four dyads were formed ( n = 208), and of these matched pairs, 64.42% were female (see Table 2 and 3 for an overview on the participants).
188 Length of English-as-a-foreignlanguage study encompassed extracurricular English classes through private language schools, private lessons, as well as formal instruction through public schools. Grade 7 students reported an average length of 3.96 year s, Grade 8 of 6.38 years, Grade 10 of 6.90 years, and Grade 11 of 8.36 years of EFL study. All of the 208 participants stated that they had expe rience with computers. More specifically, when the participants we re asked for what purposes they use a computer, they replied t hat they most often used the computer for wordprocessing activities (21.48%), follow ed by games (20.07%), browsing the internet (18.97%),and for emails (17. 01%), and less frequently on electronic bulletin boards (5.45%) and using the comp uter for programming (4.03%). With regard to chat usage, 129 students (62. 02%) stated that they use chat for personal communication, whereas only 43 students (20.67%) had used it as part of their coursework. Figure 13 provides a visual representation of computer usage across all grade levels.
189 Figure 13 Personal computer usage across grade levels. Computer Usage Across Grades in Percentageschat word games internet email bulletin program 0 10 20 30 40 50 60 70 80 90
190 Across grade levels, their English class grades, as determined by their instructors for the current year, also varied. An equivalent of an A grade was reported by 34 students (16.50%), a B grade by 52 students (25.24%), a C grade by 65 students (31.55% ), and a D grade by 55 students (26.70%). None of the students were failing their English class at the time of data collection. Again, data collection took part to wards the end of the school year. Finally, most of the students ( n = 173 or 83.17%) also were studying an additional foreign language besides English as part of the class curriculum. This reflects the Slovene curriculum as outlined in Chapter 2, where students in Grades 7, 8, and 9 may choose an additi onal foreign language as an elective; however, students in general high school hav e two foreign languages as part of their mandatory curriculum. The database. In addition to the data collected from semi-structured interview data, the qualitative data (i.e ., transcripts) collected through the chat room served as the databas e for quantitative and qualitativ e analysis. Initially, the transcripts were sorted, organized, and re viewed. The data first were sorted by grade and organized into turns. For the purposes of this study, a turn was defined as one message being typed and sent to another member. In MSN Instant Messenger or Web Messenger, one participant typed a message in the text box and when s/he was ready for t he partner to read their message, the participant sent the message by clicking on send or hitting the enter key on the keyboard. This one message sent to thei r partner constituted a turn. The data also were reviewed for any turns that would not be applicable to the study, which
191 would inappropriately incr ease the number of turns, and as such improperly inflate the amount of data co llected. More specifically the researcher deleted those turns where introductions in the L1 were used. Introductions in the first language were eliminated, as they were not part of the instruct ions; they served only a purpose of students finding out, who they were paired with and, most importantly, they were not conducted in the L2 and, theref ore, were not an objective of this study. For similar purposes dyads were deleted if their sole chat was in L1. Finally, chat transcripts were deleted if dyad members participated in the actual data collection peri od, but did not complete the practice run. The latter were deleted, because these participants were not exposed to the same treatment as other participants and, t herefore, would not correctly reflect participants understanding of the task nor fi nal results. In addition, there was one case of lost data on a disk due to a floppy disk malfunction. Other students were eliminated from the data analysis due to not having a partner in class (odd number of students), not s howing up between the practice and actual sessions, or not being in a general high school but rather being students on a technical track. One dyad was elimi nated from the data analysis for using profanity in all turns. However, actual turns were includ ed and only identifying information were altered, such as names, addresses, telephone numbers, and email addresses. Emoticons and punctuations also were preserved. Such text-based symbols were a substitute for facial expre ssions and emotions and therefore were
192 included in the turns. In addition, off-topic turns in the L2 were preserved, as they did elicit corrective feedback. As such, after refining the data, th e database serving for analysis included a total of 4,590 turns among 104 dyads in Grades 7, 8, 10, and 11.The turns were not equally distributed among learners or grade. A total of 922 turns were provided in Grade 7, Grade 8 had 600 turns, in Grade 10 there were 1,163 turns, and 1,905 turns in Grade 11. These tu rns then were coded for corrective feedback and error types within Microsoft Excel, using the codebook (Appendix J), which is based on a modified version of Lyster and Rantas (1997) error treatment sequence. Data Analysis The unit of analysis for this study was based on the modified error treatment sequence of Lyster and Rant a (1997). The error treatment sequence constituted an initial turn containing an error, following the learners possible response to the error, and a possible peers reaction or response to the correction (see Figure 9). The actual codi ng reflected that out of the seven a priori categories of error types, five of the categories were existent. The seven pre-determined error categories were: (a ) grammatical, (b) lexical, (c) orthographical, (d) typographica l/spelling, (e) usage of L1, (f) multiple, and (g) emergent. Following are def initions and examples of each error code. A grammatical error. A grammatical error co nstituted a participant producing a grammatical construction that violated the grammar conventions of
193 the English language. In addition, inappropria te word order or usage of articles and syntactical errors also were coded as grammatical errors. Example 1 (Grade 10) Line 735 Student A then he tell her to made bed tense and article error = grammatical Line 736 Student B At 3 oclock he went sleeping What is in your third picture? Tense error = grammatical Even though there were multiple errors of grammar within Line 735 in Example 1, this was counted as one grammatical erro r in the coding of individual turns containing errors. A lexical error. A lexical error constitutes the usage of an inappropriate word or missing lexical item in an utter ance (i.e., missing lexical items such as prepositions, nouns, adjectives). Howeve r, whenever article errors were committed those were coded as grammatica l. More specifically, articles are functional not lexical free morphemes and their usage is related to rule application in an utterance. Examples of lexical errors include inaccurate, imprecise, or inappropriate choices of lexical items and non-target derivations of nouns, verbs, adverbs, and adjectives. Ex ample 2 and 3 show examples of lexical errors. Example 2 (Grade 8)
194 Line 30 Student A At tuesday she fell asleep at 3 Inappropriate use of preposition Example 3 (Grade 7) Line 128 Student A The girl stood up at 6:30 Inappropriate choice of lexical item Orthographical errors. Orthographical conventions consisted of errors including omissions of accent marks and letters unique to the English alphabet (i.e., q, w, x, y) or tr ansfer of letters unique to t he Slovene language (i.e., , and ). For example, an orthographic erro r would be evident, if the learner spelled the lexical item cherry as erry, reflecting the Slovene orthographic convention for the ch shound. Typographical / spelling errors. Typographical/spelling errors created while inputting text via a keyboard. Such an erro r is made despite the user knowing the spelling of the word. This usually result s from the persons inexperience using a keyboard, from rushing, quick typing, not paying attention, or carelessness (see Examples 4 and 5). A spelling error is one made when forming words with letters, and the letters are not put in the acceptable order. However, in this study, it was almost impossible to determine whether the learner made a typographical error or spelling error, or if it was an erro r of orthographical conventions. Therefore, orthographical and typographica l/spelling were combined into one category because it was difficult to determine if t he omission of a certain letter unique to the English alphabet as opposed to the Slovene alphabet was due to an
195 omission due to the speed of typing, spelling error, or in essence, if it was a true orthographic error. Example 4 (Grade 11) Line 87 Student B do you know what happend in my story Error typographical / spelling of happend Example 5 Grade 10 Line 323 Student A I have nuber 3 too Error typographical / spelling error of nuber Unsolicited use of L1. Usage of L1 consisted of utterances, where the participants used the native l anguage or the Slov ene language as in Example 6. One of the specific ations in the instructions (see Appendix K and L) included usage of only English. Therefore, usage of L1 was considered as an error. This category also was created to examine weather and how peers react with any form of corrective feedback to the unsolicited use of the L1. Example 6 Grade 7 Line 427 Student B po tevilkah, tako kot jaz. medva mava vsak svojo zgodbo pol jo morava pa skupi sestavt [with numbers, like I did. Usage of L1 (L1 here is colloquial)
196 we both have our stories and then we have to put them in order] Multiple errors Multiple errors were coded when more then one type of error occurred in a student turn (for ex ample, lexical and grammatical) and, as such, these were coded as multiple errors. Example 7 provides a sample of multiple errors. Example 7 Grade 10 Line 730 Student A On mine one man eat an ice-cream mnjam...On a visit came his friend. He talk him something....OK...That is when the time is 9.00.Than they go watc TV Multiple errors (including grammatical, lexical, typographical / spelling) An emergent category. An emergent category was created to allow for any error types that were not foreseen. Not surprisingly, there were no instances of any emergent error categories. However, there were deviations that were not classified as errors and were not included in the frequencies of errors but might serve us eful in additional studies on corrective feedback. One such category is usage of L3 which are utterance(s) that contain neither the L1 (S lovene) or L2 (English), but is the third
197 language being studied by the participants (i.e., German). As such, an error is neither lexical (wrong voc abulary unit) or a typographi cal/spelling in the L2. Therefore, 11 turns in Gr ades 10 and 11 were coded as usage of L3 These turns were not coded as emergent because it was not certain if these errors were due to interlanguage development, transfer, or intentional use. If usage of L3 within the transcripts reflected inter-language dev elopment or transfe r, then it would have been justified to enter this type of error into the emergent category (A. Erben, Ph.D., personal communica tion, February 27, 2006). There were also turns that included content feedback, using an L1 term for clarification purposes. These instances were not errors that would promote corrective feedback, but rather generate content/question feedback, as illustrated in the following example: Example 8 Grade 7 Line 766 Student B How do you say POJDI SPAT Usage of L1 for the purpose of content feedback. Pojdi spat = go to bed] In Example 8, this turn was coded as content feedback with L1 ; however, they were not included in the error counts Within this example, the student intentionally used the L1 fo r the purpose of receiving a question on the English translation of go to bed Therefore, this was not a st ructural error, but rather a content question.
198 A final category was created called orthographicons which included emoticons, exaggerations, and abbreviations. These instances were coded under a subcategory of orthographic conventi ons, and, again, were not counted as errors. Punctuation and/or capitalizati on were not coded as an error and were wholly ignored by all participants. This is probably due to the type of interaction, which is neither a written nor a traditional face-to-fa ce format, similar to a combined verbal plus email interaction. In addition, it was interesting to note that almost every turn included either capita lization or punctuation errors and in none of the instances did the punctuation or capitalization receive any type of corrective feedback. Therefore, these we re coded as separate categories, but were not included in the frequency counts of errors. In certain instances, punctuation, as well as, emoticons, were used to display facial expressions and emotions and as such enhanced the text -based conversational interaction among the dyads. Examples 9 through 12 show ty pical examples of capitalization, abbreviation, and punctuation errors. Example 9 Grade 7 Line 443 Student A whats your first picture Apostrophe and period are not included, as well as not capitalizing the beginning of a sentence
199 Example 10 Grade 10 Line 324 Student A sooo lets start with 4 Punctuation and capitalization are not included, as well, as exaggeration of the utterance so Example 11 Grade 10 Line 466 Student B fine tnx u Capitalization missing at the beginning of the sentence; abbreviation of thanks and you Example 12 Grade 11 Line 481 Student A lol abbreviation of laugh out loud lol After all coding of errors had been completed, a total of 1,957 grammatical, lexical, orthographical /typographical/spelling, usage of L1, and multiple errors were found across all grade levels. When examining Table 5, which represents total error by type across all grades, the least frequent errors created are under t he categories of usage of L1 and lexical errors. The most frequent errors were grammatical, mu ltiple and orthographical/typo-spelling
200 errors. Overall, Grade 8 produced the least amount of errors and Grade 11 produced the greatest amount of errors. The total number of errors might have had a relationship with the total number of turns, because Grade 8 had the least number of turns (i.e., 600) and Grade 11 had the greatest amount of turns (i.e. 1,905). However, as these results are based on individual turns and the independence assumption is violated; thus statistical analysis was not justifiable.
201 Table 5 Total Errors by Type Across Grade Levels Grade 7** Grade 8** Grade 10** Grade 11** Total*** n (%) n (%) n (%) n (%) n (%) Grammatical 109(25)66(27)139 (29)234(29)548(28) Lexical 42(10)39(16)81 (17)94(12)256(13) Ortho / Typo / Spell* 99(23)66(27)103 (21)191(24)459(23) Usage of L1 48(11)5(2)25 (5)60(8)138(7) Multiple 132(31)72(29)136 (28)216(27)556(28) Emergent 0(0)0(0)0 (0)0(0)0(0) Total*** 430(22)248(13)484 (25)795(40)1957(100) Key: Ortho / typo / spell = orthographic, typhographical and spelling Collapsed orthographic and typographical/spelling into one category. ** Percentage calculated as cell frequency divided by column total *** Percentage calculated as column or row total divided by the grand total
202 The next step in the coding and analysis process was determining corrective feedback moves from the erro rs committed. From the review of literature, the researcher determined six a priori types of corrective feedback, of which all six types were found withi n the data. Corrective feedback types identified were: (a) explicit correction, (b) recast, (c) clarif ication request, (d) metalinguistic feedback, (e) elicitat ion, and (f) emergen t. Following are a description of each corrective feedback type with examples. Explicit correction. Explicit correction is an unambiguous and clear provision of the correct form, where a learner explicitly corrects their dyad members error(s). Example 13 shows an ex plicit correction in Line 91 based on a grammatical error in Line 90. Here, the learner gave an explicit corrective feedback move. The dyad member, who committed the error, noticed the feedback and responded with her own modifica tion of the grammatical error she committed in Line 90. Example 13 Grade 7 Line 90 Student A FIRSTMONDAY P.M. AT 7 O'CLOCK SHE COOK BREAKFAST Grammatical error Line 91 Student B SHE COOKS with explicit Line 92 Student A *COOKED correction Recasts. Recasts are a learners reformulation of all or part of her/his dyad members utterance excluding the erro r. Example 14 shows an example of a dyad member committing a lexical error in Line 332 and instantaneously in the
203 following turn clarifying the typographical/ spelling error by using L1. The learner then recasted the error Line 334 by reit erating the dyad members sentence without the error. Example 14 Grade 7 Line 332 Student A 5 She has stomacheak Line 333 Student A I writ e (da ga boli trebuh) good Line 334 Student B 5 She has stomachache, and her mother takes her to the hospital. Recast of lexical error Clarification request. Clarification request indica tes to the dyad member either that the learner does not understand t he utterance or that the utterance is ill-formed in some way or that a repetiti on or a reformulation is required on the part of the dyad member as in Line 104 in Example 15.
204 Example 15 Grade 8 Line 102 Student A 2. They are ate something and one girl tol something at 9. o clock Line 103 Student A ok? Clarification request of Line 104 Student B what does TOL mean typographical / Line 105 Student A Told spelling error Line 106 Student A Sory Metalinguistic feedback. Metalinguistic feedback involves comments that indicate that there is an error somewher e. These comments can be in the form of grammatical metalanguage or can point to t he nature of the erro r. In Example 16, Student A did not correctly form the past tense in Line 270. The peer then used metalinguistic feedback in Line 272, by commenting and exagger ating with a no utterance. After no response was gi ven, Student B then followed up with an explicit correction in Line 273 by capi talizing or emphasizing the error.
205 Example 16 Grade 7 Line 270 Student A her eggs did not good Tense error Line 271 Student A Later she was hungry With meta Line 272 Student B nonononononononon! linguistic feedback Line 273 Student B Her eggs WASN'T good and explicit Line 274 Student A yES I BELIVE YOU correction Elicitation feedback. Elicitation is w hen the learner directly elicits the correct form from her/his dyad member. These elicitations may be in various forms. The learner can allow the dyad me mber to fill in the blank, can use questions to elicit the correct form, or can ask the dyad member to reformulate the utterance. In Example 17, a dyad me mber committed a lexical error in Line 1149. The learner did not under stand the utterance and elic ited in Line 1152 the correct lexical item. However, either the dyad member did not provide an answer because of topic continuation or the dy ad member did not perceive turn 1152 as a request for correcting the error. Example 17 Grade 11 Line 1149 Student A i have a mote Lexical error Line 1150 Student A Sory Line 1151 Student B at 8:00 or at 6:30 when she woke up Line 1152 Student B I have a ? Elicitation Feedback
206 Repetition. Repetition is another type of corrective feedback move, when a learner repeats the dyad mem bers erroneous utterance in isolation. In Example 18, in Line 642, the dyad member committed a multiple error consisting of a grammatical and typographical/spelling erro r. The learner in Line 643 used a repetition move and isolated the typographica l/spelling error to provide feedback that the utterance was ill-fo rmed. However, the dyad me mber did not provide any acknowledgments on either receiv ing the feedback or correcting the typographical/spelling error. Example 18 Grade 7 Line 642 Student A 3.it's tuesday a.m. at 3.00 she slepeng. Grammatical and Lexical error (multiple with) Line 643 Student B slepeng? Repetition feedback Emergent feedback Emergent feedback or request for feedback was an a priori category created for the purpose of additional feedback types that might emerge from the data that were not accounted for in previous studies. An additional corrective feedback type emerged, named request for feedback. Instead of the learner providing a form of corrective feedback to the dyad member, the learner themselves r equested feedback be given. This was considered as an additional corrective feedback type because it provided an opportunity to negotiate with persons dyad me mber on a linguistic structure that had not been yet fully articulated or acquire d by the learner soliciting the proper structure. Thus, r equest for feedback is defined as the dyad member herself
207 implicitly (see Example 14, Line 333) or explicitly (see Example 19 below) requesting feedback based on their own err oneous error. In this corrective feedback move, the learner acknowledges t heir error and solicits a correction to their ill-formed or ill-structured utteranc e. In Example 19, Student A did not complete their turn as s/he stumbl ed on an unknown lexica l item. In the immediate turn, Student B requested feedba ck on the lexical item in the L1 and immediately received the feedback, whic h was then incorporated in Line 316.
208 Example 19 Grade 11 Line 313 Student B my first: at 6.30 she woke up, her mother baked a Did not complete sentence as lexical item was not known Line 314 Student B Kako se napie zajtrk? [how do you write breakfast?] request for feedback on unknown lexical item Line 315 Student A breakfast Line 316 Student B At 6.30 girl woke up and her mother had already baked breakflast for her. Learner incorporates feedback, but with a typographical / spelling error Line 317 Student A It is 7 o'clock and she made a breakfast Example 20 below shows another instance of a request for feedback. Here, Student A uses the L1 as he is uns ure of a vocabulary item. Initially, Student B thought it was funny and us ed an onomatopoeic interjection. However Student A continues and in Li ne 961 explicitly requests for the English translation of the unknown voc abulary unit. Two lines further, Student A requests an answer to his request and notices that Student B already provided implicit feedback to the erroneous term in Line 963. Even though
209 feedback was provided, it was not an accura te translation or vocabulary unit because the correct translation would be frying rather than baking the steak in a pan. Example 20 Grade 11 Line 959 Student A im pe i the steak [Im frying the steak] Line 960 Student B ha ha Line 961 Student A pe i in English [fry in English] request for feedback Line 962 Student B in the morning at six thirty i woke up Line 963 Student B at seven o clock mother bake me a steak Implicit feedback recast Line 964 Student A answer me please Line 965 Student A aja bake are you sure Line 966 Student B at eight o clock i ate my steak Example 21 and 22 below reveal additional examples on request for feedback. Again, these requests are for unknown lexical items. In Example 21, Student A does not remember the lexical item for couch or sofa so she requests it in L1 from her dyad mem ber. The dyad member does provide the appropriate answer in Line 510, which is immediately incorporated into the learners turn in Line 511.
210 Example 21 Grade 8 Line 508 Student A in my last picture girl was sitting on the i dont now how is english you said kav Request for feedback on lexical item kav [couch / sofa] Line 509 Student B at 7.00 girl eats brekfast Line 510 Student B sofa is kav Explicit feedback provided Line 511 Student A so in my last picture girl was sitting on the sofa and she is looking very bad, i think she is sick Incorporation of feedback However, in Example 22 the learner reques ts the translation for plate from his dyad member. The dyad member replie s with two possible options, of which the learner who requested the feedback explicitly chooses one of the two vocabulary units suggested. However, he does not incorporate the feedback requested in any of the turns follo wing acknowledgment of the feedback provided.
211 Example 22 Grade 7 Line 572 Student B kaku se re e kronik Request for feedback on lexical item kronik [how do you say kronik] Line 573 Student A ful razumm Line 574 Student A plate One possible translation offered Line 575 Student A or soucher Another lexical option Line 576 Student B *plate Student chooses first option Another type of error correction that was found in the data, but was not included as a corrective feedback type, were instances of self-correction Selfcorrection is when students correct their errors within the same or immediate turn. It is coded separately because it does not belong within the scope of corrective feedback by another learner, did not promote interaction, but resulted in correction within themselves--similar to one verbally correcting oneself outloud. Self-identified errors occurred ac ross all grades and were distributed as follows: (a) 12 (1.3%) in Grade 7,(b) 9 (1 .5%) in Grade 8, (c) 17 (1.5%) in Grade 10, and (d) 40 (2%) in Grade 11. Howeve r, these might have been instances where self-correction gave the other dyad member the opportunity not to commit
212 an ill-formed utterance or structure and mi ght have been facilitative to the dyad members conversational chat. However, there was no way to determine whether this had happened in this study. Interview or other reflective tools would have been facilitative to determine if th is is a plausible premise. Out of the 4,590 total chat turns, 88 represented va rious corrective feedback moves. Table 6 shows the frequency and percentage of corrective feedback types across the grade levels. More specifically, explicit correction (42%) was the most frequent type of co rrective feedback move, followed by recasts (23%) and the emergent cat egory request for feedback (20%) when examining the frequencies across the grade levels. The least frequently used corrective feedback type was in the category of opportunity for negotiation (Castaeda, 2005) or negotiation of form which Lyster and Ranta (1997) coined as an umbrella term for elicitation, meta linguistic feedback, clarification request, and repetition. Together negotiati on of form accounted only for n = 13 or 14% of total corrective feedback moves.
213 Table 6 Frequency and Percentage of Corrective Feedback Types by Grade Level Grade 7 Grade 8 Grade 10 Grade 11 Total n (%) n (%) n (%) n (%) n (%) Explicit correction 17(45) 6(43) 4 (25) 10(50) 37 (42) Recasts 5(13) 5(36) 5 (31) 5(25) 20 (23) Elicitation 0(0) 0(0) 0 (0) 1(5) 1 (1) Metalinguistic feedback 2(5) 0(0) 0 (0) 0(0) 2 (2) Clarification request 5(13) 1(7) 1 (6) 0(0) 7 (8) Repetition 1(3) 0(0) 2 (13) 0(0) 3 (3) Emergent 8(21) 2(14) 4 (25) 4(20) 18 (20) Total 38 (43) 14 (16) 16 (18) 20 (23) 88 (100) Note. Cell percentages were calculated as the sum of each cell divided by column totals. Row and column percentages were calculated as row total divided by grand total or column total divided by grand total, respectively.
214 When examining percentage of correct ive feedback and learner turns with error by grade level (see Table 7), the percentage of learner turns with errors were approximately equal across all grade levels (between 41%-47%); wherein percentage of turns with e rrors receiving feedback dec reased as the grade level increased. When examining both error and corrective feedback turns across grade levels, only 4% ( n = 88) of student turns with erro rs received corrective feedback. Out of the 88 learner turns wit h error receiving corrective feedback, Grade 7 had the second least amount of errors and total turns compared to the other grades, but the highest amount of corrective f eedback. On the other hand, Grade 11 had the highest amount of turns and errors, but the least amount of corrective feedback moves.
215 Table 7 Percentage of Corrective Feedback and Learner Turns with Error by Grade Level I. Total Number of Learner Turns II. Total Number of Learner Turns Containing Errors III. Percentage of Learner Turns with Error a IV. Total number of Learner Turns with Error Receiving Corrective Feedback V. Percentage of Student Turns with Error Receiving Corrective feedback b n n %% % Grade 7 922 43047 38 9 Grade 8 600 2484114 6 Grade 10 1163 4844216 3 Grade 11 1905 7954220 3 Total 4590 19574388 4 a calculated as total number of learner turns with error divided by total turns, b calculated as corrective feedback divided by learner error
216 When examining the sums of frequencie s for all grades (see Table 8), the most frequent corrective feedback move was explicit correction for orthographical/typographical/spelling e rrors (21%) and emergent request for feedback on lexical errors (21%). As visually represented in Figure 14, the proportion of error types receiving correct ive feedback reflects the rate at which various error types by turn occurred in the database. Usage of L1 received the least amount of corrective feedback with only 2% allocated towards explicit correction whereas lexical errors received the highest proportion (42%) of overall co rrective feedback moves. The least amount of corrective feedback moves were in negotiation of form and the least amount of errors that receiv ed feedback were in usage of L1.
217 Table 8 Frequencies of Corrective Feedback and Learner Error for All Grades Learner Error Grammatical Lexical Ortho / Typo / Spelling L1 Multiple Total n % n % n % n % n % n % Explicit Correction 7 (8) 9 (10) 18 (21) 2 (2) 1 (1) 37 (42) Recast 4 (5) 6 (7) 5 (6) 0 (0) 5 (6) 20 (23) Negotiation of Form 5 (6) 4 (5) 4 (5) 0 (0) 0 (0) 13 (15) Emergent 0 (0) 18 (21) 0 (0) 0 (0) 0 (0) 18 (21) Total 16 (18) 37 (42) 27 (31) 2 (2) 6 (7) 88 (100) Note. Cell percentages are calculated as cell total divided by grand total.
218 Figure 14 Corrective feedback frequencies per error type across grade levels. 0 5 10 15 20 25 30 35 40 ExplicitRecastNOFEmergent Corrective Feedback Types Multiple L1 Ortho/Typo/Spell Lexical Grammatical
219 Results Following are the results of the res earch question within this study (see Table 4). For the three null hypotheses t hat were tested, all occurrences of corrective feedback or errors within one dyad were collapsed to either zero or one incidences. For example, if one dyad ha d three different types of corrective feedback (e.g., explicit correction, reca st, and emergent) this was coded as one incidence of corrective feedback for this particular dyad. Similarly, the same procedure was used to code error types. The collapsing of incidences of both corrective feedback and error types was conducted not to violate the independence assumption. Because one mem ber within the dyad might influence the type of corrective feedback and/or erro r, this established a possible influence among the dependent variables. By collaps ing turns into incidences, the independence assumption was not violated and statistical analyses could be undertaken, of course, taking into consideration other assumptions. Descriptive statistics were computed to assess the normality assumption. To be normally distributed variables, ske wness and/or kurtosis coefficients (divided by their standard e rrors) should be within the 3 range (Onwuegbuzie & Daniel, 2002). The skewness and kurtosis c oefficients were reviewed for each of the variables, that is, for each type of corrective feedback incidences (explicit correction, recast, negotiation, and emergent ), corrective feedback as a whole, and individual error types (grammatical lexical, orthographical/typo/spelling, usage of L1, and multiple). D eparture from normality was i ndicative for two of the four corrective feedback types and three out of the five error types. More
220 specifically, for the followin g corrective feedback types: (a) negotiation (skewness coefficient = 2.98; kurtosis coefficient = 7.04) and (b) emergent (skewness coefficient = 2.44; kurtosis coefficient = 4.05). For error types, the following were not within the limits of normality: (c) grammatical (skewness coefficient = -4.29; kurtosis coefficient = 16.70), (d) orthographical / typographical / spelling (skewness coefficient = -3.85; kurt osis coefficient = 13.07), and (e) multiple (skewness coefficient = -4.29; kurtosis coefficient = 16.70). Because the overall kurtosis coefficients were greater than 3 they suggested a leptokurtic distribution. Due to the fact that corrective fee dback types negotiation and emergent, as well as the error types grammatical, orthographi cal/typo/spelling, and multiple did not fall within the domain of normality, additional cauti on should be exercised in interpreting any inferential analysis in volving the aforementioned variables. Table 9 presents descriptive statistics of the variables as a function of grade level.
221 Table 9 Mean and Standard Deviation for Incidence Va riables as a Function of Grade Level Grade 7 Grade 8 Grade 10 Grade 11 Corrective Feedback Types n M SD n M SD n M SD n M SD Explicit 32 .34 .48 16 .25 .45 24 .13 .34 32 .19 .40 Recast 32 .16 .37 16 .19 .40 24 .17 .38 32 .16 .37 Negotiation 32 .16 .37 16 .06 .25 24 .08 .28 32 .03 .18 Emergent 32 .16 .37 16 .13 .34 24 .08 .28 32 .09 .30 Total feedback 32 .47 .51 16 .31 .48 24 .29 .46 32 .34 .48 Error Types Grammar 32 .91 .30 16 .94 .25 24 1.0 .00 32 .97 .18 Lexical 32 .56 .50 16 .94 .25 24 .83 .38 32 .94 .25 Ortho/Typo/Spell 32 .91 .30 16 .94 .25 24 .96 .20 32 .97 .18 Usage of L1 32 .47 .51 16 .19 .40 24 .33 .48 32 .59 .50 Multiple 32 .97 .18 16 .81 .40 24 1.0 .00 32 .97 .18
222 Results of Null Hypothesis 1 Null hypothesis 1. There is no difference in the incidence of corrective feedback in online synchronous envir onments provided by adolescent EFL learners to other dyad members as a f unction of grade level. Table 10 depicts corrective feedback and non-feedback incidenc es of error turns across all grade levels. Non-feedback incidences were calculated based on error incidences, where no corrective feedback was provi ded. Similarly, corrective feedback incidences were calculated by taking a ll error turns that provided corrective feedback. Incidences were defined as co llapsing all subtyp e corrective feedback levels into one category. Specifically, in cidences had a value of zero, where there was no corrective feedback provided to the error or a value of one, where there was one or more corrective feedback types provided. As such, the total sample was 104, which appropriately corresponds to the number of dyads.
223 Table 10 Corrective Feedback and Non-Feedback Incidences by Grade Level Total Learner Error Incidences Corrective Feedback Non-Feedback Total Grade 7 Frequency of incidences n = 15 n = 17 n = 32 % within grade 47% 53% 100% % within total feedback 40% 26% 31% Grade 8 Frequency of incidences n = 5 n = 11 n = 16 % within grade 31% 69% 100% % within total feedback 13% 17% 23% Grade 10 Frequency of incidences n = 7 n = 17 n = 24 % within grade 29% 71% 100% % within total feedback 18% 26% 23% Grade 11 Frequency of incidences n = 11 n = 21 n = 32 % within grade 34% 66% 100% % within total feedback 29% 32% 31% Total Frequency of incidences n = 38 n = 66 n = 104 % within grade 37% 64% 100% % within total feedback 100% 100% 100%
224 From the observed frequencies and perce ntages in Table 10, there were more non-feedback incidences within each grade level than feedback incidences. Across all grade levels, 64% were non-feedback incidences and 36% were corrective feedback incidences. Grade 7 had the highest amount (40%) and Grade 8 had the lowest amount (13%) of total corrective feedback incidences across all grade levels. However, the low amount of total corrective feedback in Grade 8 might be due to fewer participants. Before proceeding with the chi-square analysis, assumptions were reviewed for randomness, independenc e, and frequency of expected observations. The data were collect ed from a random sample, the frequency within each cell was collapsed to incidenc e level to prevent the violation of independence assumption, and the degrees of freedom and expectancy counts were reviewed. None of the expected ce lls contained a count of less than five. Because all assumptions were met, t he researcher proceeded with the chisquare. Using a contingency table, a 4 x 2 chi-square was conducted via SAS to test this null hypothesis. The analysis revealed no statistically significant difference in the incidence of correcti ve feedback as a function of grade level, 2(3, N = 104) = 2.30, p > .05. Cramers V the effect size measure indicated a relatively moderate relationship, with V = .51. This effect size suggests that the small sample size prevented a statistically significant relationship from emerging. Results of Null Hypothesis 2 Null hypothesis 2. There is no relationship between the type of corrective feedback in online synchronous envir onments provided by adolescent EFL
225 learners to other dyad mem bers and grade levels. To test this hypothesis, the tabulated data from Table 11 was used to conduct a 4 x 4 chi-square analysis. The sample size for the chi-square was 62. The sample size is based on the incidence level of each correct feedback subtype: (a) explicit correction, (b) recast, (c) negotiation of form, and (d) emergent request for feedback, respectively. Incidences within each dy ad were collapsed for each subtype. In addition, MANOVA, a discriminant analysis, and an effect size as measured by 2 were computed and interpreted to assess the statistical and practical significance of the results, respectively. A MANOVA was conducted to assess results at the dyad level ( n = 104); therefore, inci dences were calculated as the overall corrective feedback type incidences, whereby each subtype was collapsed to either zero or one incidence. A chi-square analysis was used to det ermine whether the four grade levels differed across the four dependent va riables of corrective feedback. The independent variable was grade level (7, 8, 10, and 11) and the dependent variables were the corrective feedback types (explicit correction, recast, negotiation of form, and/or emergent). Negotiation of form resulted from collapsing clarification requests, elic itation, metalinguistic feedback, and repetition. Lyster and Rant a (1997) collapsed these four corrective feedback types, as they are ones that implicitly ask for feedback on ill-formed utterances. Within CMC, Castaeda (2005) also collapsed clarification request s, elicitation, metalinguistsic feedback, and repetition; however, she named the collapsed category as opportunity for negotiation (Castaeda, 2005).
226 Table 11 Observed Frequencies and Percentages of Corrective Feedback Incidences by Type and Grade Explicit Correction Recasts Negotiation Emergent Total n (%) n (%) n (%) n (%) n (%) Grade 7 11(18)5(8)5(8)5(8)26(42) Grade 8 4(7)3(5)1(2)2(3)10(16) Grade 10 3(5) 4(7) 2(3) 2(3) 11(18) Grade 11 6(10) 5(8) 1(2) 3(5)15(24) Total 24(39)17(28) 9(14)12(19)62(100)
227 Before proceeding with the chi-square analysis, assumptions were reviewed for randomness, independenc e, and frequency of expected observations as in null hypothesis one. Because all the assumptions were met, a chi-square was computed using Table 12 s observed frequency counts on types of corrective feedback incidences by grade level (see Table 11). However, caution should be exercised in interpre ting the findings because 75% of the expected counts were less than five. The results revealed no statistically significant relationship between the type of corrective feedback provided by adol escent EFL learners to other dyad members and grade level, 2(9, N = 62) = 2.9323, p > .05. Cramers V was used to measure the effect size, which reflects a low relationship, with V = .13. An additional test, a MANOVA, was computed to assess the relationship between type of corrective feedback and grade level. T he independence, equality of variance-covariance, linear ity, and normality assumptions were reviewed before proceeding with the MANOVA. SPSS for Windows (2001) was used for the statistical procedure. Boxs M test was reviewed to determine homogeneity of the variance-covariance ma trix involving the corrective feedback types. Boxs M statistic was 60.02, suggesting heterogeneity of the covariance matrices ( F [30, 15046] = 1.84, p = .003). Although Boxs M is very sensitive to departures from normality; discriminant analysis and MANOVA are robust to this violation. As such, caution is not ed when interpreting these results as heterogeneity appeared to be present.
228 Next, a MANOVA was conduct ed at the dyad level. In order to detect a moderate effect size with four variables a total of 64 participants (or 32 dyads) were needed per group or per each level of the independent measure (Stevens, 2002). However, because some data were eliminated from the analysis in Chapter 3, a total of only 26 dyads (i.e., 104 dyads / 4 groups) for each grade level were used for analysis. Analysis was completed using SPSS version 11 (SPSS for Windows, 2001). The hypothesized effect was used generating Wilks Lambda to evaluate the statistical signifi cance of the variables. Wilks Lambda was used to measure the difference in means among the gr oups, where the greater the value of lambda the smaller the differences. The relationship between corrective feedback type to other dyad members and grade level fell short of statistica l significance (F[ 12, 257] = .59, p > .05 ; Wilks Lambda = 0.93. Further, the effect size, as measured by 2, associated with grade differences wa s .04. Discriminant analysis was not undertaken because none of the variables were statisti cally significant. These results failed to support Hypothesis 2. Thus, it appears t hat there is no relationship between corrective feedback and dyads as a functi on of grade level. Table 12 presents descriptive statistics pertaining to these four variables as a function of grade.
229 Table 12 Mean and Standard Deviation for Corrective Feedback Incidences as a Function of Grade Level Explicit Recast Negotiation Emergent Grade levels n M SD n M SD n M SD n M SD Grade 7 32 .34 .48 32 .16 .37 32 .16 .37 32 .16 .37 Grade 8 16 .25 .45 16 .19 .40 16 .06 .25 16 .13 .34 Grade 10 24 .13 .34 24 .17 .38 24 .08 .28 24 .08 .28 Grade 11 32 .19 .40 32 .16 .37 32 .03 .18 32 .09 .30 Total 104 .23 .42 104 .16 .37 104 .09 .28 104 .12 .32
230 Results of Null Hypothesis 3 Null hypothesis 3. There is no rela tionship between learner error and type of corrective feedback in online syn chronous environment s among adolescent EFL learners working in dyads across grade levels. To determine statistical significance among learner error and type of corrective feedback, four Fishers exact tests were computed. Fishers exact tests were chosen because the observed frequencies were quite low and it was anticipated that the expected frequencies would be five or less (see T able 13 through 16). A chi-square would not have been an appropriate statistical procedure because chi-square assumes expected frequencies of five or more per cell. The independent variable was the error type and the dependent variabl e was the feedback type.
231 Table 13 Contingency Table of Observed Frequencies of Correct ive Feedback and Learner Error Incidences for Grade 7 Learner Error Grammatical Lexical Orthographic / Typographical / Spelling L1 Multiple Total n (%) n (%) n (%) n (%) n (%) n (%) Explicit Correction 3(10) 3(10) 5(17) 1(4) 1(4) 13 (45) Recast 1(4) 0(0) 2(7) 0(0) 2(7) 5(17) Negotiation of Form 2(7) 2(7) 2(7) 0(0) 0(0) 6(21) Emergent 0(0) 5(17) 0(0) 0(0) 0(0) 5(17) Total 6(21) 10(35) 9(31) 1(4) 3(10) 29(100) p=0.11, Fishers exact test
232 Table 14 Contingency Table of Observed Frequencies of Correct ive Feedback and Learner Error Incidences for Grade 8 Learner Error Grammatical Lexical Orthographic / Typographical / Spelling L1 Multiple Total n (%) n (%) n (%) n (%) n (%) n (%) Explicit Correction 1(7) 1(7) 3(21) 1(7) 0(0) 6(43) Recast 2(14) 1(7) 1(7) 0(0) 1(7) 5(36) Negotiation of Form 0(0) 0(0) 1(7) 0(0) 0(0) 1(7) Emergent 0(0) 2(14) 0(0) 0(0) 0(0) 2(14) Total 3(21) 4(29) 5(36) 1(7) 1(7) 14(100) p=0.67, Fishers exact test
233 Table 15 Contingency Table of Observed Frequencies of Correct ive Feedback and Learner Error Incidences for Grade 10 Learner Error Grammatical Lexical Orthographic / Typographical / Spelling L1 Multiple Total n (%) n (%) n (%) n (%) n (%) n (%) Explicit Correction 1(8) 2(17) 1(8) 0(0) 0(0) 4(33) Recast 0(0) 1(8) 2(17) 0(0) 1(8) 4(33) Negotiation of Form 1(8) 1(8) 1(8) 0(0) 0(0) 3(25) Emergent 0(0) 1(8) 0(0) 0(0) 0(0) 1(8) Total 2(17) 5(42) 4(33) 0(0) 1(8) 12(100) p>.99, Fishers exact test
234 Table 16 Contingency Table of Observed Frequencies of Correct ive Feedback and Learner Error Incidences for Grade 11 Learner Error Grammatical Lexical Orthographic / Typographical / Spelling L1 Multiple Total n (%) n (%) n (%) n (%) n (%) n (%) Explicit Correction 1(6) 1(6) 5(31) 0(0) 0(0) 7(44) Recast 1(6) 3(19) 0(0) 1(6) 0(0) 5(31) Negotiation of Form 0(0) 1(6) 0(0) 0(0) 0(0) 1(6) Emergent 0(0) 3(19) 0(0) 0(0) 0(0) 3(19) Total 2(13) 8(50) 5(31) 1(6) 0(0) 16(100) p=.04 > p.013 (Bonferonni adjustment), Fishers exact test
235 Care should be used in interpreting these results, because several cells contained zero values. The Bonferroni adjustment was used because multiple comparisons were conducted. After the Bonferroni adjustment the alpha level was calculated as .0125 (i.e., .05 divided by 4 comparisons). The Fishers exact test did not yield a statistical significant relationship between error and corrective feedback type in Grade 7 ( p = .11), Grade 8 ( p = .67), Grade 10 ( p > .99), or Grade 11 ( p = .04). Furthermore, the effect size, measured using Cramers V, were: (a) Grade 7, V = .46; (b) Grade 8, V = .52; (c) Grade 10, V = .39, and (d) Grade 11, V = .53. Consequently, it can be inferred that from the Fishers Tests, there is no relationship between erro r type and type of corrective feedback across grade levels. Results of Qualitative Analysis A qualitative approach was integrated within the study to examine how participants with special needs interact within online synchronous discussions either with or without spec ial needs. Three learners with special needs were identified out of t he 208 participants, who were r andomly selected. The data from the three learners with special needs we re included in the quantitative analysis and were extrapolated for subsequent qualit ative analysis, more specifically conversation analysis. The guiding question for qualitative analysis was: What interactional conversation characte ristics by dyad members are present in online-synchronous environments when one or more of the interlocutors are learners with special needs?
236 Characteristics of interactional conv ersations were defined as corrective feedback moves, error types, and responses to prompts. The data were analyzed using conversation analysis (CA) as initiation/response/follow-up sequences (Mehan, 1985; Ohta, 1993, 1994, 2001; Si nclair & Coulthard, 1975) and adjacency pairs (Sacks et al., 1974; Schegloff & Sacks, 1973). As stated above, the sample provi ded three learners with special needs who were grouped within two dyads in Grade 7. The special needs participants comprised one student who had a neurologic al disorder and epilepsy, one who had attention deficit hyperactivity di sorder (ADHD), and one who had a learning disability. All were classified as havi ng a mild form of disability on their individualized education plans. None of the participants with special needs required any special accommodations or modifications for t he activity, tool, computer equipment, special writing equipment, or length of time as noted in the individualized education plans and by teachers assessment. Of the three learners with special needs, two were males and one was a female. The first dyad analyzed was a special need-special need (SN1 SN2) female / male dyad. The total chat resulted in 18 to tal turns, compared to an average of 29 turns per dyad within the Grade 7 data set. The second dyad analyzed was a special need-high learner (SN3 HL) dy ad. The learner with special needs was a male and the high learner was a female. T heir chat resulted in 17 total turns, wherein the learner with s pecial needs provided 7 total turns and the high learner 10 total turns.
237 Of the 35 total turns created by the two dyads, a total of 30 initiation turns and 5 response turns were delineated. Within the IRF sequences, frequencies and tallies of corrective feedback and erro r types and their relationships were counted to determine any regularities. In addition, eight adjacency pairs were identified. Narrative explanations under representative data for each dyad are provided below. Special need special need dyad. This was a female-male dyad with a documented diagnosis of epilepsy and neurological disorders and Attention Deficit Hyperactivity Disorder (ADHD), respec tively. In the fi rst three lines, the participants used emoticons. Learner SN1 used a positive emoticon, whereas SN2 used negative emoticons representing an angry look in Lines 289 and 290. As it is the beginning of the conversa tion, SN1 began on a positive tone, whereas SN2 either had immediate negative emotions to the task, to th eir conversation partner, or as a reaction to t he situation (see Extract 1). Extract 1 Line 288: SN1: My name is sara ho are you Line 289: SN2: my names is martin Line 290: SN2: at 8 o'clock have a breakfast Even though SN2 had a negative disposit ion, as is evident by the emoticons in Lines 289 -290 and Line 303 (s ee Extract 2), the learner did provide an additional emoticon either to reinforce a turn or to mimic fa cial expression as within traditional face-to-face conversati ons. In Line 299, SN1 committed multiple errors that included grammatical and typographical/spelling errors. SN2
238 acknowledged the content in Line 300 and further attempted to correct the utterance. More specifically, SN2 attemp ted to reinforce corrective feedback that was being provided to SN1 by using an em oticon of a half-moon to visualize and enhance the meaning of sleep (see Extract 2): Extract 2 Line 299 SN1: and shy go to slip INITIATION Line 300 SN2: my too INITIATION RESPONSECONTENT Line 301 SN2: she go to sleep INITIATION-RESPONSE Line 302 SN1: at suesday morning got ap at 7.00 o clock INITIATION Line 303 SN2: By INITIATION While examining the whole data set for this dyad, 3 responses to 17 initations were identified within the IRF sequences, where both responses were targeted towards spelling and or lexical e rrors. In Extract 2, all turns included errors and therefore were initiations for peers to provide feedback. Line 300 included a content response, whereas Line 301 provided a response to the typographical/spelling error in Line 299 committed by learner SN1. In Lines 291 and 292 (see Extract 3), both were coded as initiation turns, as both learners committed errors in t heir turns. SN1 created a multiple error including both grammatical and typographi cal/spelling, whereas SN2 committed only a typographical/spelling error with the utterance slipp. Line 292 also was coded as a response in the IRF sequence because it provided feedback to the grammatical error, but not to the ty pographical/spelling error that SN2 also
239 committed. However, SN2 corrected this typographical/spelling error without any further feedback or prompts la ter in the chat in Line 301 (see Extract 2). A followup turn was not coded in this instanc e because SN1 did not incorporate the feedback in subsequent turns except in Line 299 (see Extract 2), where SN1 included the verb in the present form and somewhat changed the initial spelling of the utterance sleep, but did not use the feedback provided by SN2 in Line 292 (see Extract 3). Extract 3 Line 291: SN1: Shy slipping INITIATION Line 292: SN2: she went to slipp INITIATION-RESPONSE Concerning adjacency pairs, five pairs were identified, where a response was given to a prompt or to an error created by the peer l earner. In Lines 288 and 289 (see Extract 4), the question asked by SN1, did follow a content response, but not a corrective feedback res ponse in Line 289. Similarly, as in the IRF sequence mentioned above SN2 provided a reply to SN1s error; however did not add much to the c ontent of the conversation in terms of conversation. Following are turns where learners are work ing individually on their tasks and it appears as if they were not 'listening' to one another (i.e., reading previous posts by their dyad member) in terms of e rrors, providing content feedback, or furthering the conversation. Learner SN2 in Line 300 reveals that he accepted the content and explicitly acknowledged listening (i.e., reading posts by dyad member) to their interl ocutor. In addition, implicit feedback may have been evident in Line 301 as a response to the initiation error in Line 299.
240 Extract 4 Line 288 SN1: My name is sara ho are you Question 1 (Error) Line 289 SN2: my names is martin Answer 1 (Error) Line 290 SN2: at 8 o'clock have a breakfast Error1 Line 291 SN1: Shy slipping Error 2 Line 292 SN2: she went to slipp Error 3 Reply Error 2 Line 293 SN1: shy go to titch Error 4 Line 294 SN2: at 6.30 woke up Error 5 Line 295 SN2: at 7 o'clock haved a braekfust Error 6 Line 296 SN1: at 9.00o clock come for frend Error 7 Line 297 SN1: shy go to batrom Error 8 Line 298 SN2: she went to comeroency room Error 9 phantom reply to error 8 (line 297) Phantom repair of 'go' to 'went' Line 299 SN1: and shy go to slip Error 10 Line 300 SN2: my too Error 11-Confirming to line 299 Line 301 SN2: she go to sleep Reply to error line 297/299 Line 302 SN1: at suesday morning got ap at 7.00 o clock Error 12
241 Special need high learner dyad. This male female dyad comprised a learner with special needs (SN3) with a documented learning disability. The high learner (HL) was a learner with a hi gh class grade and was recommended by the teacher as a strong language learner. For th ese reasons, the researcher paired this specific high learner with the lear ner with special needs. Of the total turns, SN3 produced six turns in which all but one turn comprised errors. The HL member constructed a total of 10 turns or approximately 59% of the total turns, where 7 of those turns included an e rror or initiation for a response.
242 Extract 5 Line 906 SN3: Hello Line 907 HL: Hello Line 908 HL: At 9:00 2 woman are eating their breakfast INITIATION Line 909 SN3: Monday is the miy INITIATION Line 910 SN3: Vookap is 3.00 clack INITIATION Line 911 HL: Sorry, at 8: 00 2 woman are eating their breakfast. INITIATION Line 912 SN3: A 7.00 clock is the miy INITIATION Line 913 HL: At 10: 30 the old woman is talling young woman that she has to going sleep. INITIATION Line 914 HL: What meens miy? INITIATION-RESPONSE Line 915 HL: At 6: 30 the young woman is very tierd. INITIATION Line 916 SN3: At 9.00 is drink INITIATION Line 917 HL: It' s 7 o' clock and the young woman is eating breakfast. Line 918 SN3: At vokap is INITIATION Line 919 HL: They're going to the EMERGENCY ROOM. Line 920 HL: Do you meen woke up? INITIATION-RESPONSE Line 921 SN3: Tuesday go tuslip INITIATION Line 922 HL: write hours! INITIATION As such, the SN3-HL dyad produced a total of 13 initiations with 2 responses in the IRF sequence (see Extrac t 5). Both responses were corrective feedback from the HL member toward le xical utterances in Lines 914 and 920.
243 Even though the HL member also produced errors in the responses, the errors created by the HL member did not result from any incorrect feedback, but an inappropriate spelling of the lexical it em mean, which in both instances was spelled as meen. This error may imply t hat the HL member created a spelling error versus a typographical mistake. When examining the adjacency pairs, the dyad members created a total of two pairs (see Extract 6). In Line 910, the SN3 learner errors consisted of lexical, grammatical, and typographical/spelling. Th e lexical error vookap was reiterated again by the same learner in Line 918, which evoked a response by the HL in Line 920 to clarify the error. In this in stance, the error evoked a response in creating an adjacency pair, which is evident in subsequent turns. Similarly, the second adjacency pair comprised a lexical item. SN3 posted that within a certain timeframe there was a miy, which ev oked a response by the HL member requesting clarification on miy This adjacency pair was not fully completed because SN3 did not provide further posts on the meaning of miy and was left without a completion of this adjacency pair.
244 Extract 6 Line 906 SN3: Hello Line 907 HL: Hello Line 908 HL: At 9:00 2 woman are eating their breakfast Error 1 Line 909 SN3: Monday is the mi Error 2 Line 910 SN3: Vookap is 3.00 clack Error 3 Line 911 HL: Sorry, at 8: 00 2 woman are eating their breakfast. Error 4 reposting line 908 with content correction Line 912 SN3: A 7.00 clock is the miy Error 5 Line 913 HL: At 10: 30 the old woman is talling young woman that she has to going sleep. Error 6 Line 914 HL: What meens miy? Error 7 clarification for error 5 Line 915 HL: At 6: 30 the young woman is very tierd. Error 8 Line 916 SN3: At 9.00 is drink Error 9 Line 917 HL: It' s 7 o' clock and the young woman is eating breakfast. Line 918 SN3: At vokap is Error 10 Line 919 HL: They're going to the EMERGENCY ROOM. Line 920 HL: Do you meen woke up? Error 11 clarifying error 3 & 10 Line 921 SN3: Tuesday go tuslip Error 12 Line 922 HL: write hours! Error 13 When examining the turns from bot h IRF sequences and adjacency pairs, the errors of the SN3 learner were much more complex and non-understandable than those of the HL. Content errors were self-corrected by the HL as in Lines 908 and 911. In addition, in these turns it appears that the SN3 learner is not commenting on the content or errors, but solely focusing on task completion. Interview themes. To understand better the lear ners experiences in the engagement with the task and to inform the quantitative and conversation
245 analysis results, semi-structured inte rviews were conducted with the three special needs students described above as well as with seven other extreme cases. The interviews were conducted in the Slovene language, after which, the researcher translated the interview. All identifying information was changed to conceal the identity of the participants. A colleague, who was also an educator in a Slovene elementary school, reviewed the interview transcripts for accuracy. It is important to stress beforehand t hat the interview themes may not address all issues and that the experiences by the par ticipants may or may not have been similar to those of the other participants in the study. It does not elucidate how many of the participants had a particular experience, but only how one or more participants chose to talk about various topics in the interview. Following is a description of the protocol and themes derived fr om the interview. General prompts were prepared to guide the interview; however, additional questions were elicited dependi ng on the progression of the interview. As previous research in this domai n had not been conducted, the researcher chose the interview prompts based on the researchers observation during data collection, as well as general questions t hat would solicit additional information from the interview parti cipants. Interview prom pts were prepared based on informal conversations with learners fr om the pilot study in 2004. As such, prompts prepared for the intervie w consisted of the following: a. General impression of the activity and mode; b. Advantages and Disadvantages of task, language, partner type; c. Usage of L2;
246 d. Interactions; and e. Disability type, barri ers, and advantages and/or difficulties of task (specifically asked for learners with special needs). Interviews were analyzed from part icipants deriving from various dyad groupings: two members in Grade 11, which were both considered to be high learner high learner dyads; one low lear ner one high lear ner from Grade 10, one low learner-low learner dyad in Gr ade 8; two learners with special needs from Grade 7, and one high learner one s pecial need learner also from Grade 7. Participants were chosen based on t he number of total tu rns, number of words, error level, corrective feedba ck moves, teacher recommendation, and current standing in class. The intervie w was transcribed and translated by the researcher. A total of 544 lines resulted fr om the interview. A line was counted as a complete turn before another interloc utor (interviewer or interviewee) interrupted or continued the conversation. Interview themes were determined by the inter-raters and researcher, which c onsisted of the fo llowing five main themes: (a) manner, (b) influence of mode, (c) feedback, (d) dyad types, and (e) language. Under each main theme, sub-th emes were determined, as were the number of suggestions or the frequency of sub-themes t hat were mentioned (see Table 17 for an overview of Interview Themes).
247 Table 17 Interview Themes Theme Sub-themes Freq.of sugg. by learner type Num. of suggestions 1. Manner a. Enjoyment/dislike b. Impetus to finish task c. Focus on task a. 3hl, 2ll, 5sn b. 2hl c. 2hl 14 (13%) 2. Influence of Mode d. Focus on errors e. Writing f. Oral g. No difference h. Using other forms i. Typing/writing j. Time k. Perception of interlocutor participation l. Anonymity d. 2hl, e. 3hl, 4ll, 9sn f. 2ll, 1hl, 9sn g. 1hl h. 1hl, 1ll i. 1ll, 1sn j. 1ll, 1sn k. 2hl, 1ll, 2sn l. 1sn 43 (39%) 3. Feedback m. Perception n. No feedback/feedback o. Role of teacher feedback m. 2hl, 1ll, 2sn n. 4hl, 3ll, 3sn o. 2hl, 1ll, 3sn 21 (19%) 4. Dyad types p. Types of pairs p. 8hl, 2ll, 7sn 17 (15%) 5. Language q. Novelty of mode r. Language structure and comprehension s. Attitude q. 1hl, 1ll r. 1hl, 1sn s. 3ll, 8sn 15 (14%) Total 110
248 Under the first main theme of Manner three sub-themes were identified: (a) enjoyment, (b) impetuous to complete the task, and (c) focus on the task. The theme Manner was created because it encompasses learner comments on behaviour, attitudes, perceptions, and reacti ons. All participants agreed that they enjoyed working on the task within the chat room, which was commented by high learners (hl), low learners (ll), and learner s with special needs (sn). Their remarks stated that they enjoy ed it very much, they would definitely partake and work on a similar activity again, and that it is was, literally, refreshing. It was surprising to see that one of the specia l need learners expressed en joyment after their chat reflected emoticons that reflected negat ive emotions (see Extract 1). When asked why they were angry, they noted t heir negative emotions with regard to the mode and writing demands of the tasks, both of which are additional sub-themes described below. Another sub-theme defined was on the need or Impetus to complete the task More specifically, learners were focused on completing the task and provided that as a rationale for not prov iding any feedback. As one high learner states, H1: Because we were trying to finish the task quickly, so I didnt want to ask him. In a sense, it was their drive to co mplete the task within the time period given, rather then being focused on t he task at hand or even providing any feedback in terms of focusing on grammatical accuracy. The final sub-theme identified within the umbrella theme of Manner was Focus on Task When asked the reason for not pr oviding feedback, the learner justified by saying that their focus was on completing the task and that the
249 correctness of the language, while working on it, was not that vital. This is reiterated by the following high learner comments, H2: Well, because we were trying to get the pictures in order and as lo ng as the idea is there. I am not used to writing English out correctly in a chat room. I think that as long as they know what you are talking about, it is fine. However, it was interesting to not e that both sub-t hemes mentioned, namely, the need to complete the task quickly and being focused on the task were provided by only the high learners. A possible reason might be that the high learners in comparison to other partic ipants in the interview had the largest number of utterances, the l east number of errors in relation to the number of turns, and the highest number of correctiv e feedback moves, whereby 30% of the errors they received were given some type of corrective feedback by the high learners. Even though both hi gh learners provided justification and rationales for not providing feedback, one of them provided all the corrective feedback moves allocated within that dyad, where th ree of those corrective feedback moves entailed explicit corrections and one as a recast of grammatical and typographical/spelling errors. An additional theme provided by all ty pes of learners, which also produced the greatest number of prompts or s uggestions, was within the category of Influence of Mode This theme was created to encompass perceptions and attitudes towards the efficacy of the chat tool as a co mmunication tool. As such, nine sub-themes were created. The firs t subtheme was their focus on errors. Despite the literatures cont ention that learners are more aware of their errors
250 within synchronous communication (Gongl ewski, 1999; Salaberry, 1996), the interviews reflected only two comments w here learners were mo re aware of their errors. One learner when asked if they revi ewed their texts or e rrors replied, H4: Yes. That helped me. If I for got then I went back to look. An intriguing comment made by the second learner did not refer to noticing errors as a tool to facilitate their correction, but more of an extrinsi c factor, where others would be aware of their own limitations of languag e competency. She states, H2: Well, you can see your errors and that is bad. But the good part is that only one other person sees your mistake. Interestingly, these comments were made by two high learners. It was expected in this study that the ability to review already-produced utterances would be facilitative to the special need learners. However, the special need learners and low learners had a negative fo cus on errors, as reflected by their comments, and they did not find the tran script to be useful in their language experience. One low-level learner comm ented that he did not even think about using the transcript as a tool to review already-stated utterances or use the transcript for their comm unication interaction. However, the special need learners and low learners reported on being focused on grammar and spelling, which swayed their discussion away from the oral task (see sub-theme on writing below ). Even though t he students claimed not to use the tool to review language produ ction, they did focus on the here and now or their immediate language producti on. It could be hypothesized that based on their comments that the lear ners were not explicitly aware that this type of tool
251 might be facilitative to their learning--more specifically, that they could use the chat tool to reflect on their language production, errors, and interlanguage development. All learners--and most emphatically the special need learners--expressed their preference for oral/aural communication versus Writing which is the second sub-theme identified. This sub-theme produced the highest number of suggestions among all the sub-th emes under the main theme of influence of mode A barrier within chat, suggested 16 times, was the writing aspect. When asked if it was the typing or the writing that was a ba rrier, learners claimed that typing was not a problem, but the fact that not only did they have to concentrate on the grammatical and lexical structures, but also on the spelling of utterances (see Interview Extract 1). Interview Extract 1 Line 335: Question What was hard about it? Line 336: SN3 Writing the words Line 337: Question What was difficult about writing? Line 338: SN3 How to write the words .that was difficult. Line 339: Question What did you find easy? Line 340: SN3 Typing The writing turned out to be stressfu l, especially for the special need learners because they kept on highlighting writing as a barrier to their language production in the task (see In terview Extract 2 and 3).
252 Interview Extract 2 Line 211 Question What was difficult about using the chat room? Line 212: SN1 Writing. Line 213: Question Do you think it would have helped you to have more time or for your partner to help you out? Line 214: SN1 No. I dont like writing When the special need learners were asked if they had any problems completing the task or any problems with concentrati on, they vigorously commented that the only issue that they had was with writt en communication. A learner with special needs stated, SN3: No. The only problem I had was with the writing. More specifically, a special need learner in In terview Extract 3 below stated that oral communication was quicker to comple te than was written communication:
253 Interview Extract 3 Line 246 SN2 Writing. Writing everything in English was difficult Line 247 Question Was it the typing or writing? Line 248 SN2 It was the writing. Line 249 Question Did the typing give you any problems? Line 250 SN2 No, no problems. Line 251 Question If you had to do the same thing that you did on the computer, but in the classroom, which would you prefer? Line 252 SN2 Definitely in the classroom Line 253 Question Why? Line 254 SN2 Because you dont have to write. In class you just say it and its over. The low learners also were asked to mention any barriers and what they had found to be difficult. Similarly, they responded that the writing was the greatest obstacle. A representative comm ent made by the interview participants is reflected in the following thoug hts mentioned by a low learner, I think it would have been better to do it out loud. You can see the person and also talk with them instead of writing the answers. Even the high learners commented on the writing portion of the task. One learner stated, H4: I didnt like to write. I have no problems typing I type fast, but I dont like to write in English and when asked the reason, she stated, H4: Because it is difficult to write words. Based on these comments, it could be hypothesized that wr itten conversation is cognitively more demanding than is oral conversation, despite typing skills or language
254 proficiency. These cognitive demands were commented on by the following learner, H3: I would rather do it out loud. You dont have to think about writing. You just say it and its finished. As such, almost alll learners agreed that Oral responses were much easier than writing. The interview par ticipants recommended that it would have been better to have audio chats versus text-based chats, which is another subtheme under influence of mode Only one high learner noted that there is no difference in using text chat or oral chat ( i.e., no difference subtheme ). Another benefit of using the chat tool noted was the option of using the emoticon functions (i.e., using other forms subtheme ), which as previously noted, either was used to reinforce the content or was used as an attempt to replicate a facial expression typical of face-to-face oral conversations. Other intertwining themes included that Typing (or the physical act of producing conversation) was not an issue, but rather that the cognitive dem ands of written accuracy were an issue. Only one high learner noted that there wa s no difference between oral versus written modes. However, anot her stated that text conv ersation was a limitation, where they had to wait for a response because the conversation did not follow a typical face-to-face conversational pattern, where questions are usually immediately followed by some sort of re sponse. As one high learner stated, H3: The most annoying thing for me when we chatted was not being able to connect with one another, that is be on the same page. Despite high learners concern with t he pace of conversation, low learners and special need learners expressed that this form of conversation provided
255 them with the opportunity to think of t heir responses before they had to reply. This was identified as interview sub-theme Time. Providing adequate time is similar to Warschauers (1995, 1996) cont ention that chat provides learners the opportunity to engage epistemologically wit h their own learning process. An example is where a special need learner responds to a question on what he had found to be easy, SN2: Everything else except writing. That I had time to think before I would write. The actual pace of text-bas ed chat and interaction patterns differ from oral conversations, where responses tend to be immediate without much tolerance for contemplation. Interestingly, the interview responses reflect the researchers notes from data coding that participants we re not listening each other ( perception of interlocutor participation ). For example, a high learne r noted that initially they waited for their peer to respond, but after a while decided to work on their task with minimal reading (or listening) on what their peer posted. The learners with special needs commented that they were focused on their task without waiting for feedback or responses from their part ner. When interview participants were asked the reason for not reading what their partner had posted, learners commented that their focus was on comp leting their portion of the task as completely and as quickly as possible. Another reason prov ided was that the language levels of the postings were non-understandable, eit her the language was severely deficient or highly adv anced, where the peer was not able to decipher what was being posted.
256 The final sub-theme under influence of mode is anonymity Anonymity was defined as the learner conv ersing without outside influenc e. In the extract below, the learner did not wish for anyone besides a fellow peer to be involved in the conversation. When asked the reason, t he learner stated that it heightened her/his anxiety (see Interview Extract 4). Interview Extract 4 Line 295 Question Would you do this exercise again? Line 296 SN2 Yes. But, I would do it only for correspondence and with no one else being able to read it. Line 297 Question Who is no one else? Line 298 SN2 You or the teacher. Line 299 Question Why wouldnt you want anyone else (the teacher or me)? Line 300 SN2 Because I feel like you are watching for my mistakes and that makes me sort of nervous. The third theme identified was Feedback This theme was developed when instances of feedback were mentioned in the interview, either as the role of teacher feedback, perception of feedback, or reasons for providing feedback. Both high learners and a special need l earner provided feedback; however, one high learner stated that content comprehension rather than grammatical accuracy was important for a message to be correctly understood, reflected in
257 the following quote, H2: Again, I think that chatting is really fast. In the Slovenian language we shorten or abbreviate word s and everyone knows what it means. But, not in English and writi ng English in the right way. As long as everyone understands then there is no need to correct, right? But, as mentioned by this learner, familiarity with the language in diffe rent contexts also is a factor in conversing successfully and having enough co mfort with the context and peer to provide feedback. On the other hand, a s pecial need learner did not wish to provide feedback because he was unsure of his self-perceived level of knowledge in the English language was appropria te to provide feedback (s ee Interview Extract 5). Interview Extract 5 Line 263 Question Did you receive any feedback any correction on your errors? Line 264 SN2 No. Line 265 Question Why not? Line 266 SN2 Because I didnt know how to. Line 267 Question What didnt you know? Line 268 SN2 English! This language learner commented severa l times throughout the interview that even though he enjoyed the task, he was quite unsure of his knowledge of English, found writing to be quite an arduous task, and noted that whenever working in pairs or groups, others tended to complete the task without his assistance.
258 Within the theme Feedback another sub-theme mentioned in the interview was Perception of Feedback As one high learner pointed out, he attempted to provide feedback; however, the peer did not accept nor reject the feedback, thereby creating a state of uncertainty by the learner providing feedback. When the learner was asked if he gave feedba ck to his partner, he stated, H1: Yes, I did. I tried to fix any errors, but I wasn t sure if he understood what I was telling him. Similarly, the learner with special nee ds commented that she also provided feedback; however, she did not receiv e any feedback on her work nor on the feedback she provided to her peer. This was also associated with the adjacency pairs and IRF sequences determined in t he qualitative analysis, where prompts did not receive much responses and initia tion of errors did not produce much corrective attention by the peer. Interv iew participants also mentioned that no feedback was provided either because ther e were too many errors, the learner perceived that their dyad members knows more, or that it is the teachers role to make any suggestions on errors (i.e., no feedback/feedback subtheme ). The final sub-theme on the Role of Teacher Feedback revealed that in all instances, learners did not miss the teac hers feedback. In Interview Extract 6 below, the high learner ex plained that the intera ction between the peer and himself/herself provided sufficient assi stance, such that any teachers feedback would not have been deemed useful.
259 Interview Extract 6 Line 57 Question Do you think it would have been better for you if the teacher was also in the chat room? Line 58 H1 No. Why? Line 59 Question Maybe to correct your English. To help you out if you needed it. Line 60 H1 No. I dont see a benefit of having a teacher in the chat room. My partner and I did perfectly fine. Other reasons given by the interv iew participants for not missing any teachers feedback were not having any major problems with the language (as reported by high learners) or a dislike of their teacher (as noted by low learners). Additional comments from low learners included that t he teachers presence or feedback would not have m ade a difference. High lear ners stated that their comfort level would have altered if t he teacher would have been included. One student stated, H4: I dont know. I just feel better without a teacher. You can make jokes without teachers. It could be said that with the teachers presence the conversation would be more formal and more on-task. Similar discussions have been posited by Pellettieri (2000) and Kelm (1992). However, another high learner stated that, H3: No, I wouldnt feel comfortable. I would have liked to have someone who had the same level of English as I do. It would have been better, suggesting that the conversation would have progressed with a more equal dyad member rather t han teachers presence.
260 Consequently, t he fourth theme Dyad was identified based on interview participants comments on the type of pairs member or qualities that dyad members should possess. High learners agr eed that working with a partner with similar knowledge in E nglish was beneficial. One high learner commented on her/his reactions if she/he had been pair ed with a low level learner, H1: Nervous. I would have to work with getti ng information from them and work on the English, also. I think it would be very frustrating. Another high learner also reiterated a similar sentiment conc erning unequal partner s, stating: H2: Yes. I worked really well with my par tner. I think we helped each other out to get the pictures in some order. I dont know if it was correct, but we tried. If I had a par tner that didnt know Englis h, I think that I would have lost my mind. I dont like to waste ti me I like to get the work finished. The high learners who were paired with learners with low-proficient students commented that equal knowledge of E nglish would have been advantageous. A high learner who was par tnered with a lower-level student stated, H3: I think it would have been better if it would have been more equal. Equal. For example, having pairs with the sa me marks. That would make it more equal . The high learner further commented that the conv ersation did not progress as they had anticipated mainly due to the English langu age barrier. Likewise, the high learner partnered with a learner with special needs stated that she was focused more on the writing aspect and did not view it as a speech interaction. The high learner was frustrated by not being able to pr ogress at a pace that was quicker and not being able to build on the task being presented.
261 Similarly, but for different reasons, the low learners commented that they would have preferred someone with equal or greater knowledge of the language. As one student stated, L1: Yes, I think I would like to talk to another person, someone who knows more English so that they could help me. Here, a higherlevel student would have been favored ma inly due to the assistance that the high-level student can provide the low leve l student. In the instance where a low learner was paired with a hi gh learner, the lower-level student stated that she would not have changed her partner (s ee Interview Extract 7 below) and expressed enjoyment when working with her partner. Interview Extract 7 Line 477 Question How was your partner? Line 478 L3 Perfect. It was great. I wouldnt have changed my partner. Line 479 Question Do you think you and your partner were equal in your language knowledge? Line 480 L3 Sometimes. My partner knows a lot of English and we had a good time. Similarly, learners with special needs had the same sentiments as did low learners in that they would have pr eferred a peer with equal or greater knowledge of the language and that self -choosing partners would have been preferred. In Interview Extract 8, the student wit h special needs first stated that a partner with greater or equal knowledge of the l anguage would have been
262 preferred. However, when explicitly ask ed later in the interview if he/she would have wanted a partner wit h a higher working know ledge of the language, the learner commented that they would have pr eferred a peer with equal rather than greater knowledge of the language to control for t he pace of conversation. Interview Extract 8 Line 289 Question Would you have changed anything? Line 290 SN2 Maybe someone else. Line 291 Question Why? Line 292 SN2 I dont know Line 293 Question What do you think would have been better? A friend, someone with a different English level maybe more, less, or the same? Line 294 SN2 Yes, someone with the same or better English. .(later in the conversation) Line 315 Question How about if you are paired in the classroom, which partners do you like to work with? Line 316 SN2 Those that are about the same. Line 317 Question Not more? Line 318 SN2 No. Because they go to fast. Based on the frequent mention of l anguage level within dyad type, the final theme identified was Language Apart from language le vel mentioned in the previous sub-theme, lear ners commented on the following topics, which were
263 identified as sub-themes: Novelt y of mode, language structure and comprehension, and attitude. A possible effect and one that influences other themes (namely, the theme influence of the mode ) is the Novelty of Mode as well as Language Structure and Comprehension Even though, learners had had experience with using chat in the L1, the task itself was novel to them in terms of using it in the L2 and using the tool as a task for a class assignmen t. In addition, because the task was in English, special need students comm ented that the level of language comprehension was a barrier in using the tool more effectively. However, as recounted earlier by a high learner using the English language in a chat room does not necessarily include correct langua ge structure as long as the message is correctly understood by the listener. Most frequent comments under the theme Language were the learners Attitudes toward the language. Similarly, as with Language structure and comprehension attitude towards the language in fluenced the previous identified sub-themes (i.e., Writing/typing, Dyads, Feedback, Task enjoyment/dislike ). When learners were asked about the E nglish language and any difficulties with the whole task, they comment ed that the task itself was enjoyable, but that they disliked English in general, disliked writing in English, or did not feel that they were proficient enough to wr ite in English. Interesti ngly, high learners did not mention difficulties with the English l anguage, only low learners and learners with special needs did so. These attitudes might have influenced the frequency and type of corrective feedback.
264 Chapter Summary The descriptive data revealed that co rrective feedback is provided through synchronous communication; however, there is no statistical significance in terms of the frequencies or types of correc tive feedback across grade levels, or a relationship between learner error types and dyad members type of corrective feedback move provided. Interestingl y, the amount of corrective feedback diminished as proficiency (i.e., grade level) increased. However, an additional MANOVA test was conducted to determine if there was a relationship between error types and grade level. The results did indicate statistical significance. The results and discussion will be expanded in a follow-up study, where learner repairs also will be included to determine any practical significance with regard to error, type of corrective feedback, and learner repairs. As expected, conversation analysis of learners with special needs did reflect communication typical of online environ ments in that they were not serially located, but were dispersed throughout the conversation. In addition, learners commented that the most ar duous task in the activity was the writing aspect, either because of the fast-paced nature of the activity or due to cognitive complexity. Regardless of the reason, they all agr eed that they would have preferred oral communication. The implicat ions of these results are discussed in the following chapter, as well as the pedagogical recommendations and implications for future research.
265 CHAPTER 5: SUMMARY, DISCUSSION, RECOMMENDATIONS, AND IMPLICATIONS Overview The purpose of this dissertation wa s to examine adolescent learnerlearner corrective feedback patterns within a text-based synchronous environment. This final chapter present s the discussion and summary of the results presented in Chapter 4. Addition al recommendations for future research as well as pedagogical implic ations will be provided. Discussion The role of corrective feedback, namely negotiation, has been argued to be facilitative of second language acquisition. From an interactionist perspective for acquisition to take place, there mu st be active involvement (Stevick, 1976, 1980), where conversational interactions contain opportunities for input and output, facilitating second language devel opment to a various degree (Long, 1996). Varonis and Gass (1985) also contend that for learning to take place, learners must stumble upon non-understandings (p. 73). More specifically, the provisions of feedback give the learner an opportunity to co mpare target-like utterances and nontarget-like utterances with their own interlanguage utterances (Tomasello & Herron, 1988). In addi tion, synchronous online conversations assist the learner to visualize the talk process and provide an environment that
266 allows them to ask questions, discuss, in teract, and seek assistance from peers or instructors. As such, this study has attempted to address the overarching question on corrective feedback moves and types of corrective feedback within online synchronous environments among peer-to-peer interact ions, as well as any relationships between the type of errors and their respective corrective feedback moves. Additionally, initial research on t he characteristics of interaction between dyads, where three members are learner s with a documented special need, also was explored. The a priori categories used for coding were based on previous research on corrective feedback of ill-formed utterances (i.e., Lyster & Ranta, 1997; Lyster, 2004; Morris, 2005) with an em ergent category made available for any new discoveries that emerge. Discussion of Research Question 1 The previous chapter addressed both the quantitat ive and qualitative results in detail. These results do conf irm that some learners produce simple sentences and other learners produce more complex structures (Chun, 1994). Turning to research Question 1 on incidenc es of corrective feedback, the results of this study are similar to Iwasaki and Oliv ers (2003) findings in that there was a lower amount of corrective feedback as compared to previous face-to-face feedback research on non-native speak ers (Iwasaki, 2000). There were approximately 37% corrective feedback in cidences (see Table 10). However, when reviewing the data turn by turn, onl y 4% of the participants in this study received corrective feedback (see Table 7). This is a relatively low percentage
267 compared with other online studies, w here 25.6% of negative feedback was provided by non-native speakers (Iwasaki & Oliver, 2003) and 56% with childrenchildren online dyads within a foreign language situation (Morris, 2005). In addition, corrective feedback amounts were relatively lower compared with other face-to-face feedback studies, where feedback was more than 40% (Iwasaki, 2000) and even as high as 61% (Oliver 1995) and 62% (Lyster & Ranta, 1997). It is, however, important to note that in a study of native speaker instructors of foreign languages and their usage of co rrective feedback with their students in synchronous and asynchronous modes, in structors also failed to provide feedback; more specifically, they provided less feedback than anticipated (Castaeda, 2005). Day, Chenoweth, Chun and Luppescu (1984), in a study of face-to-face classrooms, also reported t hat out of 1,595 errors, only 119 or 7.3% received corrective feedback. Furthermo re, adolescents participated in the current study, whereas in previous studies the target participants were university students (Blake, 2000; Castaeda, 2005; Iw asaki & Oliver, 2003; Pellettieri, 2000) and fifth-grade immersion childr en (Morris, 2005). A large amount of corrective feedback might have been evident in Morris (2005) study in that only 135 error turns were accounted for in co mparison with 1,957 error turns in the current study. More importantly, the partici pants in Morris study also were intact immersion students from one grade, which dras tically differs to the participants in this study, who were from traditional foreign language programs, where they were exposed to different teaching me thodologies and pedagogical techniques.
268 Discussion of Research Question 2 The data from this study did not reveal statistical significance with regard to corrective feedback incidences across grade levelthe focus of research Question 2; however, corrective feedback wa s provided, albeit inconsistently in all grades. This supports descriptive research undertaken with university students (Blake, 2000; Pellettieri, 2000) and immersion middle school students (Morris, 2005) that dyads do provide intera ctional feedback to one another. It is also important to note that researchers applying inferential statistics to dyad or group members, while counting specific turns of errors and corrective feedback should take into consideration violat ions of the independence assumption. Therefore, any studies docum enting results statistical significant results within dyad pairings should be scrutinized for assu mption violations (e.g., Blake, 2000; Mackey et al., 2003; Morris, 2005). Another important result was the ident ification of an emergent corrective feedback type, more specifically request for feedback. A possible rationale for the emergent category is because of the medium of the co nversation. Because it was difficult for students to provide faci al expressions or hand gestures as in face-to-face communication, students opted to ask for feedback, once they had stumbled on an incorrect linguistic form Because this also represents a nonunderstanding (Gass & Varonis, 1985, 1994), whereby the student is focusing on what is not known, further research should be ex plored in terms of learner repair, whenever feedback is r equested after the learner has committed an error. In addition, the use of request for feedback also might reveal the
269 learners interlanguage processes at work. More specifically, from the interview results, the participants considered text -based chat as being more complex due to the written aspect of conversation. As such, within online chat the learners might have noticed their gap in the target language, thereby requesting feedback. It is possible that the onlin e chat provided the means of negative evidence in that by visualizing the talk, the learners were more aware of the ill-formed utterances and triggered their attention towards a more appropriate linguisti c structure. It might also indicate that these language learners are now p sycholinguistically prepared to accept instruction on thos e linguistic forms (Pienemann, 1984). There were also incidences of self-i dentification of error, whereby the student immediately self-corrects her/his error without any requests. These might have been due to typographical/spelling mist akes caused by the fast paced nature of the conversation. However, the current analysis was centered on otherinitiated feedback repairs and not on self-repairs. Therefor e, self-identification of errors were not examined in detailed. Surprisingly, the data in this study revealed no statistically significant differences in the incidence of correc tive feedback to other dyad members as a function of grade level. The researc her hypothesized that there would be a difference in the type of correctiv e feedback moves as proficiency and interlanguage development increased (Pie nemann, 1987, 1989). In addition, it was hypothesized in this study that as learners notice erroneous utterances (Alanen, 1992; Lightbown & Spada, 1990; Long, 1991, 1996; Tomasello & Herron, 1989) and negotiate these ill-formed st ructures that based on proficiency
270 (i.e., grade level) interlanguage processes would affect the nat ure of corrective feedback. However, this study revealed no such differences. This would suggest that notwithstanding profici ency levels in the fore ign language the nature of feedback provided did not differ. However, other studies similar in nature to the current research did not compare across grade levels. Such studies examined learners within a similar grade level or pr oficiency level (Blake, 2000; Morris, 2005; Pellettieri, 2000), studied native-speak er interactions with second language learners (Castaeda, 2005; Iwasaki & O liver, 2003), examined learners within face-to-face immersion classrooms intera cting with participants of a similar age level (Lyster & Ranta, 1997) or investigated dyad types that included adults and children native and non-nativ e speakers (Oliver, 1995). Nonetheless, even though the type of corrective feedback moves were not statistically significantly different ac ross grade levels, additional questions do arise and more in-depth research is warranted on the qualit y of corrective feedback moves concerning second langua ge learners stages of interlanguage development. In addition, w hen examining the relations hip between error types and corrective feedback moves, the results revealed no statistically significant relationships within Grade 7, Grade 8, Gr ade 10, or Grade 11. Theoretically, the results may be in line with Pienemann and Johnstons (1987) assertion that acquisition is explained by memory processing ra ther than grammatical complexity. The fast-paced interacti ons might have been a barrier towards noticing of errors and providing correct ive feedback. In additi on, these findings may reaffirm contentions by Schegloff, Jefferson, and Sacks (1977) that other-
271 correction (as opposed to self-corre ction) may be embarrassing to the interlocutors and does not provide member s of the conversati on with equal status while participating in the discourse. Not r eaching statistical significance may be also in line with Kerns (1995) assertion that grammatical accuracy suffers as a result of synchronous discu ssions being its own discourse It may also be due to increased language production that occurred in the text-based chat, whereby the stages of interlanguage are more evident (Pe llettieri, 2000) and, as a result, more errors are obvious. Thus, longitudinal st udies should be conducted on the longterm benefits of corrective feedback and repairs. Discussion of Research Question 3 Both the frequency counts and incidences confirm that children did employ a variety of corrective feedback stra tegies with age-matched peers (Oliver, 1995); however, statistical significance wa s not achieved in this study on the relationship between error and feedback type, the focus of research Question 3. When examining data on the incidences of error and corrective feedback types, there were not any errors that specifically elicit a certain type of feedback. However, when reviewing the analysis of frequency counts, the usage of L1 received the least amount of corrective feedback, with only 2% allocated to explicit correction whereas lexical errors received the hi ghest proportion (42%) of overall corrective feedback moves. The least amount of corrective feedback moves were in negotiation of form and the least amount of errors that received feedback were in usage of L1. This diffe rs with Morris (2005) study, where all usage of L1 received corrective feedback. Recasts accounted for only 23%. This
272 figure is much lower then in other studies In particular, Lyster and Ranta (1997) found that the most frequent of all co rrective feedback moves were recasts, accounting for almost 77% of all corrective feedback moves. Furthermore, learners were more likel y to use explicit correction (42%) than recasts, negotiation of form, or em ergent request for feedback. Lyster and Ranta (1997) posit that a key indicator to the success of negotiation and types of feedback in relation to error type is the learners proficiency level. However, within the current study, Grade 7 as opposed to Grade 11, had the most amount of corrective feedback and was the most diverse. Upon further examination, explicit correction was the most frequent in Grade 7. Possible explanations might include students eagerness to find errors within dyad members conversations or, as Oliver (1995) points out, that children are greater ri sk-takers. Therefore, the larger amount of corrective feedback to errors might be due to the learners attempt to use the language more, but also to challenge their dyads by providing both implicit and explicit corrective feedback. Overall corrective feedback patterns in relation to erro r type might be attributed to the particular language of communicating in English as a forei gn language with Slovene L1 participants; however arguments can also be made that l earner errors within this study reflect the interlanguage processes, which are more evident within synchronous text communications. Furthermore, the discour se patterns may be influenced by the developmental levels, social readiness, and/or psychological differences of participants.
273 Discussion of Qualitative Results This study is exploratory in terms of characteristics of learners with special needs (SN), the focus of Research Questi on 4. CA was used to define initially such interactional characteristics. IRF moves and adjacency pairs were the method used within CA (Markee, 2000) to examine SN learners. As IRF moves reveal the structure of the language and adjacency pairs reveal the function of the language, the preliminar y results elicit further questions concerning SN learners engagement within conversations The preliminary analysis revealed that SN learners engaged quite cautious ly in the conversations. The few instances in which learners with s pecial needs were engaged were limited to invitations. It is only when they were expl icitly asked or requested to reply that they responded. Moreover, the overall lang uage was simple in terms of grammar and lexical choice. As relatively simple complexity of grammatical and lexical items was noticed in the participants tu rns, a follow-up Flesch-Kincaid Grade Level and Flesch-Kincaid Readibility Ease within Microsoft Word was calculated. According to Microsofts Offi ce 2003 Word Help (Microsoft Office, 2003), the Flesch-Kincaid Grade Level calculates t he U.S. grade level, where the derived score corresponds to the grade level. For ex ample, a score of 7.2 is equivalent to writing level of the 7th grade. Furthermore, the Fle sch-Kincaid Readability Ease score was calculated to determine the readabi lity ease of the turns. The higher the readability ease scores the greater the readability ease. A score of 90-100 would be readable to upper elementary sc hools, a score of 60-70 to upper middle school students, and results with 0-30 wit hin the college graduate range. Even
274 though the scores are strict ly quantitative in that they measure length and number of words, syllables, sentences and grammatical structure, they do provide a general idea of the readability of turns. However, it is important to note that these scales are no rmed on native English speak ers and do not represent measurement for non-native speakers (Schuyler, 1982). The readability ease scores were however, used to create a general idea of any differences within the participants grade level and, as such, are not generalizable. All turns with SN learners were calculated and analyzed. The dyad members, where both learner s with special needs were paired together, resulted in a zero score for the grade level. The learner with special needs who was paired with a high learner resulted in a joint 88.1 readability ease score and a 2.8 grade level score. With all three learner s the simplicity of the language was confirmed with these scores. Such results correspond to Kretschmer and Kretschmers (1998) contention that l earners who are cognitively and/or neurologically impaired may not be able to acquire the syntactic, pragmatic, and lexical forms of words, which migh t have been noticeable even more within a text-based synchronous environment. On the other hand, the learners through their semi-structured interviews with the researcher, believe that their language ability, writing barrier, lack of oral intera ction, and quick pace of the task all were prohibitive of interacting more active ly in the conversation. In addition, Pienemann and Johnston (1987) also posit that diffi culty of target language development might not be because of gram matical difficulty, but because of difficulty with short-term memory. The re searcher followed up with the instructor
275 of the three special need learners asking whether the students had difficulty in retaining short-term informa tion The instructor did conf irm that all three of the learners with special needs had difficult ies and needed additional assistance with EFL. Overall, students had a positive atti tude while working on the task and using the medium, as was express ed frequently throughout the interview transcripts by high learners, low l earners, and learners with special needs. However, all contended that they missed the oral communi cation with their peers. In addition, they all were unsure of how the conversational chat was understood by their partner, as well as how the peer perceived any feedback that was given. In addition, there was an indicati on that dyad members were not listening to each other, as revealed by comments that they wanted to complete the task as quickly as possible, which resulted in not being a ttentive to all of their partners posts. Furthermore, they commented that the l anguage proficiency of their partner was also a factor in that it was not understandable either because the language was extremely poor or too advanced to be comp rehensible. This suggests that the decreased number in turns, simplicit y of the language, and low frequency of corrective feedback might have been due to the higher level of comprehensible input received (Krashen, 1985) or inco rrect input, in terms of stages of interlanguage development. If we are to view this from a different theoretical lens, more specifically, from the perspec tive of Vygotsys (1934/1987) theory of learning and development, we could specul ate that the partici pants were not able to reach intersubjectivity. This does not confirm Wells (1999) contention that the
276 non-attainment of intersubjectivity pr omotes a type of dialogic engagement leading to regulation. Furthermore, the learners perceived abi lity of the language also was an influential factor on the production of language. As recounted by one learner with special needs, the reason for not providing feedback is his perceived lack of knowledge. Also, the act of writing versus speaking seemed to be another intertwining barrier among all learners, whereas some perceive the physical act of writing to differ drastica lly from oral conversation, some perceive the task of writing to be more cognitive, and others considering speaking to be much easier than writing. Also, based on the qualitative a nalysis of the interview data, it could by hypothesized that the special need learne rs focused more on the specific task of writing than on the actual task of comp leting the task. It is possible that they did not provide corrective feedback due to t he effect of the discourse type (Kelm, 1992). By examining the whole data set, much of the conversation contained incomplete utterances, colloquialisms, and simplified syntactic structure. Student perception of which errors to correct might have been influenced by the textbased chat. Moreover, students reported that they use chat mainly for informal conversations in the L1 and rarely in t he L2 or in the cla ssroom. Therefore, usage of L1 might not have been perceiv ed as a grave error because these errors provided low incidence of correct ive feedback, whereas grammatical or multiple errors might have been perceiv ed as representing more serious errors. To preclude some of these barriers, future research should examine oral versus text-based online synchronous conversati ons to examine corrective feedback
277 increase in relation to language profici ency and effect of long-term learner uptake. It is important to note that the inte rviewees were engaged in conversation and, therefore, not all issues might have been touched upon in each interview. Not all topics or themes might have em erged as a result of task completion during the interviews. As such, a small number of students may discuss an issue, which may or may not emerge as being fo r a larger number or percentage of students (i.e., data saturation might not have been achieved). Another important factor in codi ng and determining corrective feedback type is the notion of phantom adjacency pairs. A phantom adjacency pair is a response to a posting that might be per ceived as replying to the previous response posted, but could al so be intended to reply to a different post (Garcia & Jacobs, 1999), or it might have not have been a response, but more of a concurrence to the linguistic structure on which the learner is currently working. There could have been instances of phant om corrective feedback moves, within corrective feedback moves identified in th is study, where the implicit feedback provided could be intended as corrective feedback, but it could also provide a response to a previous post or post in the making. For exam ple, in Extract 1 (see Adjacency Pair Extract 1), Learner A misspells the word stomachache, which is then corrected by Learner B. This feedback could be considered as implicit corrective feedback or as a reca st, but it could also be considered as a phantom corrective feedback move in that Learner B did not respond to the error
278 or might not have even have read Learner As post, but used the correct spelling (or considered as post in the making). Adjacency Pair Extract 1 Grade 7 Line 333 A 5 She has stomacheak Line 334 A I write (da ga boli trebuh) good Line 335 B 5 She has stomachache and her mother takes her to the hospital. In addition, Adjacency Pair Extract 2 mi ght reveal a repair of the lexical error stake in Line 639; however, the student might have repaired the utterance because she might have caused a typographical or spelling error, or might have just made a mistake. On the other hand, the error in Line 633 might have been recasted in Line 636 and repaired in Line 639.
279 Adjacency Pair Extract 2 Grade 10 Line 633 A tuesday 3.00 mum wanted to take a rest, so she took a nap while the stake was cooking in the pan Line 634 B tell me the numbers please Line 635 B I THINK THE LAST NUMBER IS 5 do you agree? Line 636 B yap, I was just thinking about it she left the steak in the pan for a day Line 637 A 10 and then 5 Line 638 B i don/t know you have this picture Line 639 A ok i agree steak Conversely, in Extract 3 Learner C mi sspells either man or mum and Learner B corrects the error with explicit correction by providing the correct spelling for mum. However, Learner D wa s not sure if she perceived the error correctly and in the third turn asks for further clarification on the correct lexical item. Here, it is evident that corrective feedback was provided. Adjacency Pair Extract 3 Line 317 C on picture 4 the mam is making the girl breakfast Line 318 D Mam is spelled mum Line 319 D Or did you mean man? Implicit types of feedback, more specif ically recasts, are more difficult to determine and code using text-based synch ronous conversations. Consequently, existing and future research should consider the importance of coding and
280 possible avenues of explanation. Even though inter-raters were included to prevent inaccurate interpretation of data, the existence of phantom adjacency pairs might have been over-looked and theref ore might affect conclusions made in the present research. Additionally, but on the flip side of the coin, the manner in which dyads within groups or whol e class events are being coded and then analyzed using inferential statistics might cause a serious violation of the independence assumption, thereby causi ng existing statistical findings from existent literature possibly to change. Future Research Future research also should take into account the teacher s instructional style. If a teachers pedagogical approach is in a traditional sense as provider of knowledge (Berry, 1981) wherein typical co mmunication in the class is providing traditional questions and students just pr oviding answers (Tharp & Gallimore, 1988), this might have influenced the type of communication t hat learners are accustomed to in the foreign language. Ther efore, it would be useful to examine communication between teacher and students and then dyad members communicative interaction through a dial ogic tradition, as proposed by Johnson (2004). The dialogic tradition takes into c onsideration the dynamic roles of social contexts, individuality, intentionality, and the soci ocultural, historical, and institutional backgrounds of the individual involved in cognitive growth based on Vygotskys (1934/1987) sociocultural t heory and Bakhtins dialogized heteroglossia (Bakhtin, 1981)--more spec ifically, within an activity theory perspective (i.e. Engestr m, 1987; Lantolf & Pavlenko, 2001; Leontev, 1981;
281 Wells, 1999, 2002), which studies dialogic inquiry (Wells, 1999) among learners. In addition, further research should examine the pedagogy strategies used by different teachers. Lyster and Mori (2006) argue that instructional activities and feedback should differ based on the goals of the foreign language classroom. Therefore, further researc her should examine learner-lea rner interactions within classrooms that are predominantly form-f ocused with communicative activities and meaning focused classrooms with form -focused instructional activities. Possible reasons for not achieving a higher amount of corrective feedback also might have been due to the dyad types Varonis and Gass (1985) reported that within their study, the highest am ount of negotiation occurred among those dyads that differed in both language and proficiency compared to those dyads that were more similar or included a nat ive speaker. In addition, the results showed that there were instances w hen dyad members di d not allow other members to participate. Therefore, future research should examine the role of the dyad member, socio-cultural factor learners strategies, communication styles, proficiency, developmental, and social levels on corrective feedback moves. Also, the role of a native speaker as a learner dyad should be included because it is hypothesized by the researc her that the native speaker does not need to concentrate on grammatical structur es and higher cognitive functions in the act of writing and spelling. Another factor is the task. Even though the literature suggests jigsaw puzzles as an appropriate task for negotiati on of meaning (Pellettieri, 2000), it might not always be conducive. Gass and Varonis (1985) did not find any
282 difference in the two-way task, but Long (1989) contends that there is more productivity with two-way tasks. Based on participant feedback, many were focused on merely completing the task than on grammatical correctness or appropriately conveying the message to their dyad partner. Corrective feedback might have been higher if ther e was a teacher or native speaker involved, as was the case in Kelms (1992) observational st udy. As such, future areas of research should include the type of task to be used fo r online discourse, as well as in the area of interlanguage pragmatics. Additional research is warranted in te rms of oral versus text-based chat. The transcripts of the te xt-based data revealed that discourse within text-based chat lies between verbal and email exchan ges (Iwasaki & Oliver, 2003) or is known as speak-writing (Erben, 1999). It is not evident whether this type of discourse provides sufficient ground fo r corrective feedback and its facilitation towards language acquisition. In addition, due to the fast-paced tempo of textbased chat, further research is recommended on learners with special educational needs, more specifically with regard to proficiency of dyad type, as well as differentiating general comm unication or basic interpersonal communication skills (BICS) with writing fo r communicative purposes or cognitive academic language proficiency (CALP). Implications and Recommendations Researchers within the interactionist field have argued that learners who receive negative feedback to their ill-ta rgeted utterances have their language development facilitated and, as such, benefit from these interactions. If research
283 finds that within classroom interact ions, negative feedback does indeed promote second language development, then it is wort hy to investigate the interactional characteristics that learners have wit h teachers and other learners. This dissertation hopes to add to this line of research within the field of second language acquisition and negative f eedback. Researchers who have investigated corrective feedback wit hin synchronous environments have examined it from NS-NNS (Iwaskai & O liver, 2003), NNS-NNS (P ellettieri, 2000) and between child-child interactions (Morr is, 2005), each of which is significant for the purposes of this study (see Figur e 1). In addition, research studies have been carried out within French immersion settings (Chaudron, 1977, 1986; Lyster, 1998a; Lyster & Ranta, 1997) or with teac hers and university students (Blake, 2000; Castaneda, 2005; Iwasaki & Oliver, 2003; Pellettieri, 2000). This study contributes to the gap in research of computer-mediated-communication studies and corrective feedback moves with adolescent learners of English-as-aforeign-language with or without spec ial educational needs. Furthermore, additional insights can be generated on the fa cilitative role of corrective feedback within instructional contexts. The resu lts showed no statistically significant findings with respect to the type of corre ctive feedback or the relationship of error to corrective feedback. However, the st udy did not touch upon learners noticing or repairing their utterances. It does pr ovide initial information and exploratory research on the negotiation process am ong peers with or without special needs and the inclusion of similar tasks and discour se methods within FL classrooms.
284 If synchronous communication, more spec ifically, text-based chat, is used for grammatical tasks situated within a context-embedded activity, teachers should be cautious about the amount of attention students place on linguistic form and structure. Both the qualitativ e and quantitative findings revealed that students tend to focus on the meaning and completion of task, rather than on structural issues, such as grammatical, lexical, or s pelling errors (i.e., focus on form). However, because of the sample size, it is not clear the extent to which the present findings can be generalized. Neve rtheless, perhaps the most significant pedagogical implication to be dr awn, based on the results of the current study, is that instructors of foreign languages s hould be cautious when pairing learners to undertake grammatical tasks. More research needs to be undertaken in understanding how technology improves t he quality of language learning and its facilitative role of noticing gaps in knowledge, attention towards linguistic inaccuracy, and future implications of this new discourse (i.e., speak-writing). In addition, long-term research is warranted in examining whether corrective feedback types lead to L2 acquisition in t he long term (i.e., a l ongitudinal study). Furthermore, it is recommended that teachers share their corrective feedback types with their language students. Specifically, that teachers educate students on the types of implicit f eedback. Students, especially elementary students might not be aware t hat implicit types of feedback exist. It might be conducive for teachers to use Kelms (1992) suggestion to print out transcripts of their text-based conversations so that bot h the learner and inst ructor may review. In such a manner, students then do not over look their errors and are provided
285 with an opportunity to be aware of tar get language utterances (Beauvois, 1992; Kelm, 1992). In addition, archives of su ch transcripts also might shed light on learners language development in the long term. As su ch, longitudinal studies including such techniques with correctiv e feedback and learner repair may reveal the progression of language learning in process This study is explanatory and the results do not permit any definite conclusions on the usage of corrective feedback in the process of acquiring a language. As such, the replications of th is research study are needed, taking into account statistical assumptions needed to undertake inferential statistical analysis, which often has not been the case, especially within pair-work research. Limitations Both external and internal validity thr eats limit the finding s of this study. Onwuegbuzies (2003) framework for possi ble external and internal validity threats to a study were used as a guide in this study. Possible threats to external validity were: Ecological validity, which might have had a possible threat because the participants were limited to learners of English-as-a-forei gn-language from a specific geographic area in Europe; Population validity, becaus e the sample size from the combined schools may not have been large enough to justify generalizations beyond the sample; Temporal validity, because of the limited time of data collection; and Reactive arrangement, as a result of participants reactions to being aware that they are participating in the study.
286 Further, there were several threats to inter nal validity of the findings,including the following: Data Saturation Point: The fact that only one collecti on time was used, due to budgetary and time constraints, may have yielded data that did not reach data saturation point; Differential selection of participants, wherein the composition of the dyads might have affected the findings; Researcher bias also was a threat t hat limited the results, in which certain categories might have been construct ed or collapsed based on personal beliefs of the researcher (i.e., illusory correlation); and Finally, instrumentation threat was a threat pertaining to the reliability and validity of the coded data, although the high inter-rater reliability obtained suggested that this threat was minimal. Finally, the validity of t he qualitative findings was considered in terms of (a) descriptive validity, (b) interpretive validity, and (c) theoretical validity. To obtain descriptive validity, researcher tr iangulation was used. The researcher of the current study used both questionnaires wit h all participants, as well as, follow up interviews with 5% of the partici pants, which included the sampling of participants with extreme scores, includi ng special need learners. Additionally, personal notes written in a journal dur ing data collection were maintained and analyzed throughout the research process. Descriptive va lidity was maximized by presenting student accounts wit h direct quotes stemming from the interview data.
287 Finally, interpretive and theoretical validi ty were addressed by including two other peers to review the data, interpreta tion, and conclusions that emerged. Conclusion This study examined the gap in research within interactionist studies in terms of corrective feedback with adolesc ent learners of Engl ish-as-a-foreignlanguage using computer-mediated communication, more specifically, synchronous online communication. T he study also included learners with special needs as to provide initia l research with special populations. Corrective feedback types that were found in previous research also were found in this study. More specifically, l earners did provide ex plicit corrections, recasts, negotiation of form, and an emergent category. The study did not reveal any statistically significant results; however, other import ant issues emerged more specifically, the following findings : (a) an emergent category entitled request for feedback emerged; (b) the not ion of phantom adjacency pairs within coding was discussed; (c) the importanc e of appropriate st atistical analysis procedures within research with dyad members were highlighted; and (d) learners with special needs partake in conversational interactions and have a limited focus on developing further, that is, more in-depth conversation. The amount of corrective feedback in rela tion to error types was less then expected, suggesting that proficiency levels, language background, task type, text-based discourse mode, social, p sychological, and cognitive development might all be factors influencin g the results of the study. In addition, the impact of foreign language methodology and pedagogy style, as well as types of
288 communication students are accustomed to in the foreign language may all have influences on interactions among interloc utors. Most import antly, the type of discourse, which is neither strictly an oral or written genre, also might not have been contributing towards sustainability and usability of corrective feedback. The results of the study show that further research, more specifically, longitudinal integrative research is needed to build on the present study. Longitudinal studies are warranted in examining whether corrective feedback types lead to L2 acquisition over time. If research does indeed reveal, that learners pr ogress in their language developed based on their active participatio n and negotiation, then it is important that with research, we strive to not only understand, provide, and assist, but, most importantly, to involve language l earners in their development. It is only through further inquiries using various theor etical insights that greater knowledge of the specific needs of learners be atta ined and the path of language acquisition be understood. Through these means appr opriate tools and support will be mediated towards involving all students to interact with other cultural and linguistic invidiuals, regardle ss of their individual needs. "Tell me and I will forget; show me and I may remember; involve me and I will understand." Aristotle
289 LIST OF REFERENCES Alanen, R. (1992). Input enhancement and rule presentation in second language acquisition. Masters in ESL Thesis, University of Hawaii at Manoa, Honolulu. Alvestrand, H. ( 2002, September 11). Instant messaging and presence on the Internet Internet Society (ISOC). Retr ieved September 14th, 2006, from http://www.isoc.org/briefings/009/. Alwright, R. (1975). Problems in the study of the language teachers treatment of learner error. In M. K. Bu rt & H. C. Dulay (Eds.), On TESOL 75 (pp. 96109). Washington, DC: TESOL. Andrade, C., Kretschmer, R., & Kretschme r, L. (1989). Two languages for all children: Expanding to low achievers and the handicapped. In K. E. Muller (Ed.), Language in elementary schools (pp. 177-203). International Education Series. New York: The Am erican Forum for Global Education. Baca, L., & Cervantes, H. (2004). The Bilingual Special Education Interface (4th ed.) Upper Saddle River, NJ: Pearson Education. Bakhtin, M. M. (1981). The dialogic imagination. Austin: University of Texas Press.
290 Baron, N. S. (2003). Langua ges and the Internet. In A. Farghali (Ed.), The Standford handbook fo r language engineers (pp. 59-127). Stanford, CA: CSLI. Beauvois, M. H. (1992). Computer-assist ed classroom discussion in the foreign language classroom: Conversation in slow motion. Foreign Language Annals, 25 455-464. Beauvois, M. H. (1994). E-talk: Attit udes and motivation in computer assisted classroom discussion. Computers and the Humanities, 28, 177-190. Beauvois, M. H. (1997). High-Tech, High-T ouch: From discussion to composition in the networked classroom. Computer-Assisted Language Learning 10 (1), 57-70. Beauvois, M. H., & Eledge, J. (1996). Personality types and megabytes: Student attitudes toward computer medi ated communication (CMC) in the language classroom. CALICO Journal, 13 (2&3), 27-46. Beck, M. L., & Eubank, L. (1991). Acquisi tion theory and experi mental design: A critique of Tomasello and Herron. Studies in Second Language Acquisition, 13 73-76. Beretta, A., & Davies, A. (1985). Ev aluation of the Bangalore project. English Language Teaching Journal, 39 121-127. Berry, M. (1981). Systemic linguistics and discourse analysis: A multi-layered approch to exchange structure. In M. Coulthard & M. Montgomery (Eds.), Studies in discourse analysis (pp. 120-145). London: Routledge and Kegan Paul.
291 Blackboard Inc. (n.d.). Blackboard Inc. Retrieved October 30, 2006 from http://www.blackboard.com Blake, R. (2000). Computer mediated communication: A window on L2 Spanish interlanguage. Language Learning & Technology, 4 (1), 120-136. Bland, S. K., Noblitt, J. S., Armington, S., & Gay, G. (1990). The nave lexical hypothesis: Evidence from comput er-assisted language learning. Modern Language Journal, 74 440-450. Bohannon, J. N., MacWhinney, B., & S now, C.E. (1990). Negative evidence revisited: Beyond learnability or w ho has to prove what to whom? Developmental Psychology, 26, 221-226. Bradley, T., & Lomicka, L. (2000). A case study of learner interaction in technology-enhanced language learning environment. Journal of Educational Comput ing Research, 22 347-368. Brock, C., Crookes, G., Day, R., & Long, M. (1986). The differential effects of corrective feedback in native speaker conversation. In R. Day (Ed.), Talking to learn: Conversati on in second language acquisition (pp. 229236). Rowley, MA: Newbury House. Bruck, M. (1982). Language impaired chil drens performance in an additive bilingual education program. Applied Psycholinguistics, 3 45-60. Bruck, M., & Herbert, M. (1982). Correla tes of learning disabled students peer interaction. Learning Disability Quarterly, 5, 353-362. Bump, J. (1990). Radical changes in class discussion using networked computers. Computers and the Humanities, 24 49-65.
292 Bush, M. (1997). Introduction. In M. Bush (Ed.), Technology-enhanced language learning (pp. xi-xviii). Lincolnwood : National Textbook Company. Candelaria-Greene, J. (1996). A paradigm fo r bilingual special education in the USA: Lessons from Kenya. The Bilingual Research Journal, 20 545-564. Castaeda, M. (2005). Corrective feedback in online asynchronous and synchronous environments in Spani sh as a Foreign Language (SFL) classes. Unpublished Dissertation, University of South Florida, Tampa. Chapelle, C. A. (1998). Mult imedia CALL: Lessons to be learned from research on instructed SLA. Language Learning & Technology, 2 (1), 22-34. Chapelle, C. A. (2001). Computer applications in second language acquisition: Foundations for teaching, testing, and research Cambridge: Cambridge University Press. Chaudron, C. (1977). A descripti ve model of discourse in the corrective treatment of learners errors. Language Learning, 27, 29-46. Chaudron, C. (1986). Teachers priorities in correcting l earners errors in French immersion classes. In R. Day (Ed.), Talking to learn (pp. 64-84). Rowley, MA: Newbury House. Chaudron, C. (1988). Second language classrooms. New York: Cambridge University Press. Chomsky, N. (1965). Aspects of the theory of syntax MIT Press, Cambridge. Chomsky, N. (1975). Reflections on Language: The Whidden lectures New York, NY: Pantheon Books.
293 Chun, D. M. (1994). Using computer netwo rking to facilitate the acquisition of interactive competence. System, 22 (1), 17-31. Chun, D. M., & Plass, J. L. (1996). Effects of multimedia annotations on vocabulary acquisition. Modern Language Journal, 80, 183-198. Cloud, N. (2002). Culturally responsive inst ructional planning. In A. Artiles & A. Ortiz (Eds.), English language learners with spec ial needs: Ident ification, assessment, and instruction (pp. 107-132). McHenry, IL: Center for Applied Linguistics, Delta Systems Co. Cody, R. P., & Smit h, J. K. (1997). Applied statistics and the SAS programming language (4th ed.), Upper Saddle River NJ: Prentice Hall, Inc. Creswell, J. W., Plano Clark V. L., Guttmann, M. L. & Hanson, E. E. (2003). Advanced mixed methods research desig n. In A. Tashakkori & C. Teddlie (Eds.), Handbook of mixed methods in social and behavioral research (pp. 209-240). Thousand Oaks, CA: Sage. Crookes, G., & Rulon, K. (1985). Incorporation of corrective feedback in native speaker/non-native speaker conversations. Honolulu, HI: University of Hawaii at Manoa, Center for Second Language Classroom Research. Cubillos, J. H. (1998). Tec hnology: A step forward in the teaching of foreign languages. In J. Harper, M. Lively, & M. Williams (Eds.), The coming of age of the profession: Issues and em erging ideas for the teaching of foreign languages (pp. 37-52). Boston: Heinle & Heinle.
294 de Valenzuela, J. S., & Niccolai, S. L. (2004). Language developm ent in culturally and linguistically diverse students with special education needs. In L. Baca & H. Cervantes (Eds.), The bilingual special education interface (4th ed.) (pp. 125-161). Upper Saddle River, NJ: Merrill. Day, R., Chenoweth, A., Chun, A., & Luppescu, S. (1984). Corrective feedback in native nonnative discourse. Language Learning, 34 19-45. DeKeyser, R. M. (1995). Learning second language grammar rules: An experiment with a miniature linguistic system. Studies in Second Language Acquisition, 17 379-410. DeKeyser, R. M. (1998). Beyond focu s on form: Cognitive perspectives on learning and practicing second la nguage. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 4263). Cambridge: Cambridge University Press. DiFino, S. M., & Lombardi no, L. J. (2004). Language lear ning disabilities: The ultimate foreign language challenge. Foreign Language Annals, 37 390400). Doughty, C. (1987). Relating secondlanguage acquisition theory to CALL research and application. In W. F. Smith (Ed.), Modern media in foreign language education: Theory and implementation (pp. 133-167). Lincolnwood, IL: National Textbook Company.
295 Doughty, C. (1994). Fine-t uning of feedback by competent speakers to language learners. In J. Alatis (Ed.) Georgetown University Round Table on Languages and Linguistics 1993: Strat egic interaction and language acquisition (pp. 96-108). Washington, DC: Ge orgetown University Press. Doughty, C., & Pica, T. (1986). "Informati on gap" tasks: Do they facilitate second language acquisition? TESOL Quarterly, 20 305-325. Dulay, H., & Burt, M. (1973). S hould we teach children syntax? Language Learning, 23, 245-258. Dulay, H. & Burt, M. (1974). Errors and strategies in child second language acquisition. TESOL Quarterly, 8 129-136. Dulay, H. & Burt, M. (1975). Creative construction in second language learning and teaching. In M. Burt & H. Dulay (Eds.), New directions in second language learning, teaching, and bilingual education. Washington, D.C.: TESOL, 21-32. Egbert, J., & Hanson-Smith (Eds.) (1999). CALL Environments: Research, practice, and critical issues. Alexandria, VA: Teachers of English to Speakers of Other Languages. Egbert, J., Paulus, T. M., & Nakamich i, Y. (September, 2002). The impact of CALL instruction on classroom computer use: A foundation for rethinking technology in teacher education. Language Learning & Te chnology, 6 108-126. Ellis, R. (1994). The study of second language acquisition Oxford University Press.
296 Ellis, R., & He, X. (1999). The roles of m odified input and output in the incidental acquisition of word meanings. Studies in Second Language Learning, 21, 285-301. Engestrm, Y. (1987). Learning by expanding: An activity-theoretical approach to developmental research Helsinki: Orienta-Konsultit. Erben, T. (1999). Constructing learning in a virtual immersion bath: LOTE teacher education through aud iographics. In R. Debski & M. Levy (Eds.) World CALL: Global perspectives on computer-assisted language learning. The Netherlands: Swets & Zeitlinger Publishers. Erzberger, C., & Kelle, U. (2003). Making inferences in mixed methods: The rules of integration. In A. Tashakkori & C. Teddlie. Handbook of mixed methods in social and behavioral research (pp. 423-455). Thousand Oaks, CA: Sage. Eurydice. (2001). Foreign language teaching in schools in Europe. Brussels, Belgium: European Commission. Retr ieved September 14, 2006, from http://oraprod.eurydice.org/resso urces/eurydice/pdf/025EN/008_chap5_02 5EN.pdf. Eurydice. (2001/2002). The education system in Slovenia. Brussels, Belgium: European Commission. Retrieved September 14, 2006, from http://22.214.171.124/Eurybase/Applicat ion/frameset.asp?country=SI&lan guage=EN.
297 Eurydice. (2005). National summary sheets on educ ation systems in Europe and ongoing reforms. Brussels, Belgium: European Commission. Retrieved September 14, 2006, from http://www.eurydice.org/ressources/ eurydice/pdf/047DN/047_SI_EN.pdf. Fanselow, J. F. (1977). The treatment of error in oral work. Foreign Language Annals, 10, 583-593. Firth, A., & Wagner, J. (1997). On discourse, comm unication, and (some) fundamental concepts in SLA research. Modern Language Journal, 81 285-300. Fraenkel, J. R., & Wa llen, N. E. (2003). How to design and evaluate research in education (5th ed.). New York, NY: McGraw Hill. Ganschow, L., & Myer, B. (1988). Profiles of frus tration: Second language learners with specific learning di sabilities. In J. Lalande (Ed.), Shaping the future of foreign language education (pp. 32-53). Lincolnwood, IL: National Textbook. Garcia, A. C., & Jacobs, J. B. (1999). The eyes of the beholder: Understanding the turn taking system in quas i-synchronous computer-mediated communication. Research on Language and Social Interaction, 32 337367. Gass, S. (1997). Input, interaction and the second language learner. Mahway, NJ: Lawrence Erlbaum.
298 Gass, S. (1998). Apples and oranges: Or why apples are not oranges and dont need to be. A response to Firth and Wagner. The Modern Language Journal, 82 83-90. Gass, S., & Selinker, L. (1994). Second language acquisition: An introductory course Hillsdale, NJ: Lawrenc e Erlbaum Associates. Gass, S., & Varonis, E. (1985). Ta sk variation and non-native/non-native negotiation of meaning. In S. Gass & C. Madden (Eds.), Input in Second Language Acquisition (pp. 149-161). Rowly, MA: Newbury House. Gass, S., & Varonis, E. (1991). Mi scommunication in non-native speaker discourse. In N. Coupland, H. G iles, & J. M. Wiemann (Eds.), Miscommunication and problematic talk (pp. 121-145). Newbury Park, CA: Sage. Gass, S., & Varonis, E. (1994). In put, interaction, and second language production. Studies in Second Language Acquisition, 16 283-302. Gay, G., & Mazur, J. (1989). Concept ualizing a hypermedia design for language learning. Journal of Research on Computing in Education, 22 119-126. Genesee, F. (1987). Learning through two languages: St udies of immersion and bilingual education. Rowley, MA: Newbury. Glass, G. V., & Hopkins, K. D. (1996). Statistical methods in education and psychology (3rd ed.). Boston: Allyn and Bacon. Golden Hills Software, Inc. (2005/2006). Survey gold (version 7.0) [software program]. Retrieved onlin e July 24, 2006, from http://surveygold.com/index.htm.
299 Gonglewski, M. R. (1999). Linking the Internet to the national standards for foreign language learning. Foreign Language Annals, 32 348-362. Gonzalez-Bueno, M. (1998). The effect s of electronic mail on Spanish L2 discourse. Language Learning and Technology, 1 (2), 55-70. Retrieved September 14, 2006, from http://llt.m su.edu/vol1num2/article3/default.html Gonzalez-Bueno, M., & Perez, L. C. ( 2000). Electronic mail in foreign language writing: A study of grammatica l and lexical accuracy, and quantity of language. Foreign Language Annals, 33 189-198. Gonzalez-Edfelt, N. (1990). Oral intera ction and collaboration at the computer: Learning English as a second language with the help of your peers. Computers in the Schools, 7 (1-2), 53-90. Gouin, D. (1998). Report on current prac tice: Immersion program in Montgomery County, Maryland. In M. Met (Ed .), Critical issues in early second language learning: Building for our childrens future (pp. 62-64). Glenview, IL: Addison-Wesley. Granda, S. (n.d.). A brief history of Slovenia. Retrieved September 14, 2006, from http://www2.arnes.si/~krsrd1/co nference/Speeches/Granda_a_brief_histor y_of_slovenia.htm Greenlee, M. (1981). Specif ying the needs of a bilingual developmentally disabled population: Issues and case studies. NABE: The Journal for the Association for Bilingual Education, 6 (1), 5-76.
300 Greene, J. C., Caracelli, V. J., & Graham, W. F. ( 1989). Toward a conceptual framework for mixed-me thod evaluation designs. Educational Evaluation and Policy Analysis 11 255-274. Gregg, K. (1984). Krashens mo nitor and Occams razor. Applied Linguistics, 5, 79-100. Grimshaw, J., & Pinker, S. (1989). Positive and negat ive evidence in language acquisition. Behavioral and Brain Sciences, 12, 341-342. Grosman, M., Kogoj, B., Gost ia, B., Kukovec, M., Poto ar, M., aubah D., & Tomai N. (1998). Osnovna ola u ni na rt angle ina: 4.razred9.razreda. [Elementary school cu rriculum for the English Language: Grades 4-9]. Ljubljana, Slovenia: Mini stry of Education, Science and Sports. Grosman, M., Kogoj, B., aubah-Kovi D., Gostia, B., Kukovec, M., Poto ar, M., & Tomai N. (1999). Posodobitev sedaj veljavnega u nega na rta za Angleki jezik: Priloga k sedaj veljavenmu u nemu na rtu. [Curriculum revision for the English language: A supplement to the current valid curriculum]. Ljubljana, Slovenia: Mini stry of Education, Science and Sports. Gutierrez-Clellen, V. F. (1999). Language choice in in tervention with bilingual children. American Journal of Speech-Language Pathology, 8, 291-302.
301 Harley, B., & Swain, M. (1984). The interlanguage of immersion students and its implication for second language teaching. In A. Davies, C. Criper, & A. P. R. Howatt (Eds.), Interlanguage (pp. 291-311). Edinburgh, Scotland: Edinburgh University Press. Hamayan, E., & Tucker, R. (1980). Language input in the bilingual classroom and its relationship to second language achievement. TESOL Quarterly, 14, 453-468. Hasselbring, T. S., & Glaser, C. H. (2000) Use of computer technology to help students with special needs. Future of Children, 10 102-122. Hendrickson, J. M. (1978). Error correct ion in foreign language teaching: Recent theory, research, and practice. Modern Language Journal, 62, 387-398. Heritage, J. (1987). Ethnomethodology. In A. Giddens & J. H. Turner (Eds.), Social theory today (pp. 224-272). Cambridge: Polity Press. Heritage, J. C., & Atkinson, J. M. (1984). Intro duction. In J. M. Atkinson and J. C. Heritage (Eds.), Structures of social action: Studies in conversation analysis (pp. 1-15). Cambridge: Cambridge University Press. Holobow, N. (1998). A report on research: The suitability of immersion for all majority-language childr en. In M. Met, (Ed .), Critical issues in early second language learning: Building fo r our childrens future (pp. 68-73). Glenview, IL: Addison-Wesley.
302 Hopper, R., Koch, S., & Mandelbaum, J. (1986). Conversation analysis methods. In D. G. Ellis & W. A. Donohue (Eds.), Contemporary issues in language and discourse processes (pp. 169-186). Hillsdal e, N.J.: Lawrence Erlbaum. Huck, S. W. (2004). Reading statistics and research (4th Ed.). Boston: Pearson Education. Hulstijin, J. (1993). When do foreign l anguage learners look up the meaning of unfamiliar words? The influence of task and learner variables. Modern Language Journal, 77 139-147. Hunt, K. W. (1970). Syntactic matu rity in school children and adults. Monographs of the Society For Research in Child Development, 35 (1, Serial No. 134). Chicago: University of Chicago Press. Individuals with Disabilit ies Education Act. (1997). Individuals with Disabilities Education Act [IDEA]. Retrieved Sept ember 14, 2006, from http://www.ed.gov/offices/OSERS/Policy/IDEA/regs.html. Iwasaki, J. (2000). The existence of re casts in Japanese as a foreign language (JFL). Proceedings of International Asso ciation of Applied Linguistics, Tokyo 1999 [CD]. Tokyo: AILA. Iwasaki, J., & Oliver, R. (2003). Chat -line interaction and negative feedback. Australian Review of A pplied Linguistics, 17 60-73. Johnson, M. (2004). A philosophy of second language acquisition New York: Yale University Press.
303 Johnson, R. B., & Onwuegbuzie, A. J. (2004). Mixed methods research: A research paradigm whose time has come. Educational Researcher, 33 (7), 14-26. Johnston, J., & Milne, L. (1995). Scaffolding second language communicative discourse with teacher-controlled multimedia. Foreign Language Annals, 28 315-329. Kasper, G. (1985). Repair in foreign language teaching. Studies in Second Language Acquisition, 7 200-215. Kasper, G. (1997). A stands for acquisi tion: A response to Firth and Wagner. The Modern Language Journal, 81 307-312. Kelm, O. R. (1992). The use of sync hronous computer networks in second language instruction: A preliminary report. Foreign Language Annals 25, 441-545. Kemper, E., Stringfield. S., & Teddlie C. (2003). Mixed methods sampling strategies in social science research. In A. Tashakkori & C. Teddlie (Eds.) Handbook of mixed methods in social and behavioral research (pp. 273296). Thousand Oaks, CA: Sage. Kern, R. (1995). Restructuring classroom interaction with networked computers: Effects on quantity and characterist ics of language production. Modern Language Journal, 79, 457-476. Krashen, S. (1976). Formal and informal linguistic environm ents in language acquisition and language learning. TESOL Quarterly, 10 157-168.
304 Krashen, S. (1977). The monitor model of adult second language performance. In M. Burt, H. Dulay, & M. Finocchiaro (Eds.), Viewpoints on English as a Second Language. New York: Regents, 152-161. Krashen, S. (1981). Second language acquisition and second language learning. Oxford: Pergamon. Krashen, S. (1982). Principles and practice in second language acquisition London: Pergamon Press. Krashen, S. (1983). Newmarks ignor ance hypothesis and current second language acquisition theory. In S. Gass & L. Selinker (Eds.) Language transfer in language learning (pp. 135-153). Rowley, MA: Newbury House. Krashen, S. (1985). The input hypothesis: Issues and implications London: Longman. Krek, J. (Ed.) (1995). Bela knjiga o vzgoji in izobrazevanju v Republiki Sloveniji. Ljubljana: Ministry of Educ ation, Science and Sport. Krek, J. (Ed.) (1996). The White Paper on Education of the Republic of Slovenia. Ljubljana: Ministry of Education and Sport. Kretschmer, L. W. Jr., & Kretschmer, R. R. (1998). Wh at special challenges do learners with disabilities bring to t he foreign language classroom? In M. Met, (Ed .), Critical issues in early se cond language learning: Building for our childrens future (pp. 65-68). Glenview, IL: Addison-Wesley. Lakota, A. B., & Gaj gar, M. (Eds.) (2003). Education system in Slovenia Ljubljana: Ministry of Educ ation, Science and Sport.
305 Lantolf, J. P., & Pavlenko, A. (2 001). (S)econd (L )anguage (A)ctivity: Understanding learners as peopl e. In M. Breen (Ed.), Learner contributions to language learning: New directions in research (pp. 141 158). London: Pearson. Lapkin, S., Hart, D., & Swain, M. ( 1991). Early and middle french immersion programs: French language outcomes. The Canadian Modern Language Review, 48 (1), 11-44. Larson-Freeman, D., & Long, M. H. (1991) Introduction to second language acquisition research London, UK: Longmann. Lee, L. (1997). Using Internet tool s as an enhancement of C2 teaching and learning. Foreign Language Annals, 30 410-427. LeLoup, J. W., & Ponterio, R. (20 03). Second language acquisition and technology: A review of the res earch. EDO-FL-03-11, Retrieved September 15, 2006, from http://www.cal.org/resour ces/digest/0311leloup.html Leont'ev, A. N. (1981). The pr oblem of activity in psyc hology. In J. V. Wertsch (Ed.), The concept of activity in Soviet Psychology (pp. 37-71). Armonk, NY: Sharpe. Levy, M. (1997). Computer-assisted language learning: Context and conceptualization. Oxford: Oxford University Press. Lightbown, P. M., & Spada, N. (1990). Focus-on-form and corrective feedback in communicative language teaching. Studies in Second Language Acquisition, 12, 429-448.
306 Ligon, F., Tannenbaum, E., & Richardson Rodgers, C. (1991). More picture stories: Language and problem-posing activities White Plains, NY: Longman. Lin, Y. H., & Hedgcock, J. (1996). N egative feedback incorporation among high proficiency and low-proficiency Chi nese speaking learners of Spanish. Language Learning, 46 567-611. Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry. Beverly Hills, CA: Sage. Liu, M., Moore, Z., Graham L., & Lee, S. (2002). A look at the research on computer-based technology use in second language learning: A review of the literature from 1990-2000. Journal of Research on Technology in Education, 34 250-273. Long, M. H. (1981). Input, interacti on and SLA. In H. Winitz (Ed.) Native language and foreign language acquisition (pp. 259-278). New York: New York Academy of Sciences. Long, M. H. (1983). Native speaker/ non-native speaker conversation and the negotiation of comprehensible input. Applied Linguistics, 4 126-141. Long, M. H. (1985). Input and second langu age acquisition theory. In S. M. Gass & C. G. Madden (Eds.), Input in second language acquisition (pp. 377393). Rowley, MA: Newbury House Publishers. Long, M. H. (1988). Instructed interlangua ge development. In L. Beebe (Ed), Issues in Second Language Acquisition (pp. 115-141). Cambridge, MA: Newbury House.
307 Long, M. H. (1989). Task, group, and task-group interactions. University of Hawaii Working Papers in ESL, 8 1-26. Long, M. H. (1991). Focus on form: A design feature in language teaching methodology. In K. De Bot, R. B. Ginsberg, & C. Kramsch (Eds.), Foreign language research in cross-cultural perspective (Vol. 2, pp. 39-52). Philadelphia: John Benjam ins Publishing Company. Long, M. H. (1996). The role of the lin guistic environment in second language acquisition. In W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of second language acquisition (pp. 413-468). New York: Academic Press. Long, M. H. (1997). Construct validity in SLA research: A response to Firth and Wagner. The Modern Language Journal, 81 318-323. Long, M. H., & Crookes, G. (1987). Inte rvention points in second language classrooms. In B. Das (Ed.), Patterns of interaction in classrooms in Southeast Asia (pp. 177-203). Singapore: R egional English Language Centre. Long, M. H., Inagaki, S., & Ortega, L. (1998). The role of implicit negative feedback in SLA: Models and re casts in Japanese and Spanish. Modern Language Journal, 82 357-371. Long, M. H., & Robinson, P. (1998). Focus on form: Theory, research, and practice. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 15-41). New York: Cambridge University Press.
308 Lyster, R. (1998a). Negotiation of form, recasts, and explic it correction in relation to error types and learner repair in immersion classrooms. Language Learning, 48 183-218. Lyster, R. (1998b). Recasts, repetition, and ambiguity in L2 classroom discourse. Studies in Second Language Acquisition, 20 51-81. Lyster, R. (2004). Differential effects of prompts and recasts in form-focused instruction. Studies in Second Language Acquisition, 26 399-432. Lyster, R. & Mori, H. (2006). Intera ctional feedback and instructional counterbalance. Studies in Second Language Acquisition, 28 269-300. Lyster, R., & Ranta, L. (1997). Co rrective feedback and learner uptake: Negotiation of form in communicative classrooms. Studies in Second Language Acquisition, 20, 37-66. Mabbott, A. S. (1994a). Students labeled learning disabled and the foreign language requirement: Background and sugges tions for teachers. In C. A. Klee (Ed.), Faces in a crowd: The individual learner in multisection courses (pp. 325-353). Heinle & Heinle. Mabbott, A. S. (1994b). An exploration of reading comprehension, oral reading errors, and written errors between English and a second language by subjects who are labeled learning disabled. Foreign Language Annals, 27 293-324. Mackey, A. (1999). Input, interacti on and second language development: An empirical study of question formation in ESL. Studies in Second Language Acquisition, 21 557-587.
309 Mackey, A. (March, 2000). Interactional feedback on L2 morpho-syntax: Learners perceptions and developmental outcomes. Paper presented at the American Association of Applied Linguistics conference, Vancouver, British Columbia, Canada. Mackey, A., Gass, S., & McDonough, K. (2000). How do learners perceive interactional feedback? Studies in Second Language Acquisition, 22 471497. Mackey, A., & Oliver, R. (2002). Inte ractional feedback and childrens L2 development. System 30 460-477. Mackey, A., Oliver, R., & Leeman, J. (2003). Interactional input and the incorporation of feedback: An explor ation of NS-NNS and NNS-NNS adult and child dyads. Language Learning, 53 (1), 35-66. Mackey, A., & Philp, J. (1998). Conversa tional interaction and second language development: Recasts, responses, and red herrings? Modern Language Journal, 82 339-356. Markee, N. P. (1994). Toward an ethnomet hodological respecification of secondlanguage acquisition studies. In E. E. Ta rone, S. M. Gass & A. D. Cohen (Eds) Research methodology in second language acquisition (pp. 89-116). Hillsdale, NJ: Lawrence Erlbaum Associates. Markee, N. (2000). Conversation analysis. Malwah, NJ: Erlbaum Associates.
310 Maxcy, S. J. (2003). Pragmatic threads in mixed methods research in the social sciences: The search for multiple modes of inquiry and the end of the philosophy of formalism. In A. Tashakkori & C. Teddlie (Eds.), Handbook of mixed methods in soci al and behavioral research (pp.51-89). Thousand Oaks, CA: Sage. McLaughlin, B. (1978). T he monitor model: Some methodological considerations. Language Learning 28 309-332. McLaughlin, B. (1987). Theories of second language learning. London: Edward Arnold. McLean, J. E., & Ernest, J. M. (1998). The role of statisti cal significance testing in educational research. Research in the Schools, 5 (2), 15-22. Mehan, H. (1978). Structur ing school structure. Harvard Educational Review, 48, 32-64. Mehan, H. (1985) The structure of classroom discour se. In T. A. Van Dijk (Ed.) Handbook of discourse analysis (Vol. 3, pp. 119131). London: Academic Press. Microsoft Corporation. (2005a). MSN Messenger. Retrieved July 17, 2006, from http://messenger.msn.com/Xp/Default.aspx. Microsoft Corporation. (2005b). MSN Web Messenger Retrieved July 17, 2006, from http://webmessenger.m sn.com/?mkt=en-us. Microsoft Office. (2003). Microsoft Office Word Help. Microsoft Corporation. Miles, M. B., & Huber man, A. M. (1994). Qualitative data analysis: An expanded sourcebook Thousand Oaks, CA: Sage.
311 Milken Exchange. (1999). A question of effectiveness: Does education technology work? Retrieved September 18, 2006, from http://www.mff.org/edtech/discussi on.taf?_function=detail&Content_uid1= 123. Miller, S. (2003). Impact of mixed methods and design on in ference quality. In A. Tashakkori & C. Teddlie. Handbook of mixed methods in social and behavioral research (pp. 423-455). Thousand Oaks, CA: Sage. Ministry of Education, Science and S ports. (1998). Predmetnik devetletne ole [The nine-year elementary school ti metable]. Slovenia: Ministry of Education, Science and Sports. Ministry of Education and Sports. (2000). Placement of Children with Special Needs Act. Official Gazette of t he Republic of Slovenia No. 54/2000. Ministry of Education, Science and S ports. (2004). Uvajanje programa 9-letne ole [Implementing the nine year school]. Slovenia: Ministry of Education, Science and Sports. Moats, L. C., & Farrell, M. (2005). Definition, history, and research on multisensory instruction. In J. Birsh (ed.), Multisensory structured language teaching (2nd ed.), Baltimore: Paul Brookes. Montgomery, C., & Eisenstein, M. (1985). Re al reality revisited: An experimental course in ESL. TESOL Quarterly 19 317-333. Morris, F. A. (2002). Negotia tion moves and recasts in relation to error types and learner repair in the fore ign language classroom. Foreign Language Annals, 35 395-404.
312 Morris, F. A. (2005). Child-to-child interaction and corrective feedback in a computer mediated L2 class. Language Learning & Technology, 9 (1), 2945. Morse, J. M. (2003). Principles of mi xed methods and multimethod research design. In A. Tashakkori & C. Teddlie (Eds.), Handbook of mixed methods in social and behavioral research (pp. 209-240). Thousand Oaks, CA: Sage. Nagata, N. (1993). Intelligent com puter feedback for second language instruction. Modern Language Journal, 77 330-339. National Center for Supercomp uting Applications. (2000). NCSA Telnet [computer program]. UrbanaChampaign, IL: University of Illinois UrbanaChampaign. Negretti, R. (1999). Web-based activities and SLA: A conversation analysis research approach. Language, Learning & Technology, 3 (1), 75-87. Retrieved September 17, 2006, at http://llt.msu .edu/vol3num1/negretti. Netten, J. (1991). Towards a more language oriented second language classroom. In L. Malave & G. Duquette (Eds.), Language, culture and cognition (pp. 284-304). Clevedon, England: Multilingual Matters. Newton, J. (1995) Task-based interaction and incidental vocabulary learning: A case study. Second Language Research, 11 159-177.
313 Ohta, A. S. (1993). Activity, affect and stance: Sentential particles in the discourse of the Japanese as a foreign language classroom. Doctoral dissertation, University of Califor nia, Los Angeles, 1993. (University Microfilms Internat ional No. 9411371). Ohta, A. S. (1994). Socializi ng the expression of affect: An overview of affective particle use in the Japanese as a foreign language classroom. Issues in Applied Linguistics, 5 303325. Ohta, A. S. (2001). Second language acquisition processes in the classroom: Learning Japanese Mahwah, N.J.: Lawrence Erlbaum Associates. Oliver, R. (1995). Negative feedback in child NS-NNS conversation. Studies in Second Language Acquisition, 17 459-481. Oliver, R. (1998). Negotiation of meaning in child interactions. Modern Language Journal, 83 372-356. Oliver, R. (2000). Age differences in negotiation and feedback in classroom and pair work. Language Learning, 50 (1), 119-151. Oliver, R. (2002). The patterns of negotia tion for meaning in ch ild interactions. Modern Language Journal, 86 (1), 97-111. Omaggio-Hadley, A. (2001). Teaching language in context (3rd ed.). Boston: Heinle & Heinle.
314 ONeill, J., & Martin, D. (November, 2003). Text chat in action. Paper presented at the International ACM SIGGROU P conference on supporting group work at Sanibel Island, FL. New York, NY: ACM Press. Retrieved September 16, 2006, from http://www.dirc.org.uk/publications /inproceedings/papers/93.pdf#search= %22text%20chat%20in%20action%22. Onwuegbuzie, A. J. (2003). Expanding the framework of internal and external validity in quantitative research. Research in the Schools, 10 (1), 71-89. Onwuegbuzie, A. J., & Daniel L. G. (2002). Uses and misu ses of the correlation coefficient. Research in the Schools 9 (1), 73-90. Onwuegbuzie, A. J., & Dani el L. G. (2003, 19 February) Typology of analytical and interpretational errors in quant itative and qualitat ive educational research. Current Issues in Education [On-line], 6 (2). Available: http://cie.ed.asu.edu/volume6/number2/. Onwuegbuzie, A. J., & Teddlie, C. ( 2003). A framework for analyzing data in mixed methods research. In A. Ta shakkori & C. Teddlie (Eds.) Handbook of mixed methods in soci al and behavioral research (pp. 351-383). Thousand Oaks, CA: Sage. ORelly, L. (1999). The effect of focused versus unfocused communication tasks on the development of li nguistic competence during negotiated interaction (Doctoral dissertation, Universi ty of South Florida, 1999). Dissertation Absracts International, 60, 726.
315 Orthmann, C. (2000, Decem ber). Analysing the communi cation in chat rooms: Problems of data collection. Forum Qualitative Sozialforschung/Forum: Qualitative Social Research 1 (3). Retrieved September 18, 2006, from http://qualitative-research.net/fqs/fqs-eng.htm Otto, S. K., & Pusack, J. P. (1996). Technological choices to meet the challenges In B. H. Wing (Ed.) Foreign language for all: Challenges and choices (pp. 141-186). Lincolnwood, IL: Nort heast Conference Reports. Oxford, R., Rivera-Castillo, Y. Feyten, C., & Nutta, J. (1997). Computers and more: creative uses of technology fo r learning a second or foreign language. In V. Darleguy, A. Ding, & M. Svensson (Eds.), Les Nouvelles Technologies Educatives dans l'appr entissage des langues vivantes: rflexion thorique et applications pratiques (pp. 90-110). [Educational Technology in language learning:Theoret ical considerations and practical applications]. Lyon, France: Centre de Ressources en Langues, INSA Lyon. Panova, I., & Lyster, R. (2003). Patterns of corrective feedback and uptake in an adult ESL classroom. TESOL Quarterly, 36 572-595. Patton, M. Q. (2002). Qualitative evaluation and research methods (3rd ed.). Newbury Park, CA: Sage. Pellettieri, J. (2000). Negot iation in cyberspace: The role of chatting in the development of grammatical competenc e. In M. Warschauer & R. Kern (Eds), Network-based language teac hing: Concepts and practice (pp. 5986). Cambridge, UK: Cambridge University Press.
316 Pica, T. (1983). Adult acquisition of Eng lish as a second language under different conditions of exposure. Language Learning, 33 465-497. Pica, T. (1992). The textual outcomes of native speaker-non-native speaker negotiation: What do they reveal about second language learning? In C. Kramsch & S. McConnell-Ginnet (Eds.), Text in context: Cross disciplinary perspectives on language study (pp. 198-237). Lexington, MA: Heath. Pica, T. (1994). Research on negotiation: What does it reveal about secondlanguage learning conditions, processes, and outcomes? Language Learning, 44 493-527. Pica, T. (1996). Second language lear ning through interaction: Multiple perspectives. Working Papers in Educational Linguistics, 12 (1), 122. Pica, T., & Doughty, C. (1985). The role of group work in classroom second language acquisition. Studies in Second Language Acquisition, 7 233248. Pica, T., Doughty, C., & Young, R. (1985, July). Does the modification of interaction lead to acquisition? Paper presented at Summer TESOL, Georgetown University, Washington, D.C. Pica, T., Holliday, L., Lewis, N., & Morgenthaler, L. (1989). Comprehensible output as an outcome of lingui stic demands on the learner. Studies in Second Language Acquisition, 11, 63-90.
317 Pica, T., Kanagy, R., & Falodun, J. ( 1993). Choosing and using communication tasks for second language instruction. In G. Crookes & S. Gass (Eds.), Tasks and language learning: Int egrating theory and practice (pp. 9-34). Clevedon, UK: Multilingual Matters. Pica, T., & Long, M. H. (1986). The lingu istic and conversational performance of experienced and inexperienced teacher s. In R. R. Day, (Ed.), Talking to learn : Conversation in second language acquisition (pp. 85-98). Rowley, MA: Newbury House. Pienemann, M. (1984). P sychological constraint s on the teachability of languages. Studies in second language acquisition, 6, 186-214. Pienemann, M. (1987). Determi ning the influence of instruction on L2 speech processing. Australian Review of A pplied Linguistics, 10 83-113. Pienemann, M. (1989). Is language teachable? Psycholinguistic experiments and hypotheses. Applied Linguistics, 10 52-79. Pienemann, M., & Johnston, M. (1987). Fa ctors influencing the development of language proficiency. In D. Nunan, (Ed.), Applying second language acquisition research (pp. 45-142). Adelaide: Na tional Curriculum Resource Centre. Pienemann, M., Johnston, M. & Brindley, G. (1988). Constructing an acquisitionbased language assessment. Studies in Second Language Acquisition, 10 217-243.
318 Pinker, S. (1989). Resolving a learnability paradox in the acquisition of the verb lexicon (pp. 13-62), In M. L. Rice & R. L. Schiefelbusch (Eds.), The teachability of language. Baltimore, MD: Paul H. Brookes. Pomerantz, A., & Fehr, B. J. (1997). Conversation analysis: An approach to the study of social action and sense-making practices. In T. Van Dijk (Ed.), Discourse as a social interaction (pp. 1-37). Thousan d Oaks, CA: Sage. Prunk, J. (1996). A brief history of Slovenia Ljubljana; Zaloba Grad. Psathas, G. (1995). Conversation analysis: The study of talk-in-interaction. Thousand Oaks, CA: Sage. Pusack, J. P., & Otto, S.K. (1997). Taking control of mult imedia. In M. D. Bush & R. M. Terry (Eds.) (pp. 1-46), Technology-enhanced language learning ACTFL Foreign Language Series Linc olnwood, IL: National Textbook Company. Reeves, T. C. (1992). Research foundatio ns for interactive multimedia. In Promaco Conventions (Ed.), Proceedings of the Inte rnational Interactive Multimedia Symposium 177-190. Retrieved S eptember 16, 2006, at http://www.ascilite.org.au/aset-a rchives/confs/iims/1992/reeves.html Reigluth, C. M. (Ed.) (1999). Instructional-design theories and models (Vol. 2). Mahwah, NJ: Lawrence Erlbaum Associates. Roberts, M. (1995). Awareness and the effica cy of error correction. In R. Schmidt (Ed.). Attention and awareness in foreign language learning (pp. 162-182). Honolulu: University of Hawa ii, Second Language Teaching and Curriculum Center.
319 Roger, D., & Bull, P. (1989). Introducti on. In D. Roger & P. Bull (Eds.), Conversation: An interdisciplinary perspective (pp. 21-47). Clevedon: Multilingual Matters. Rondal, J. A. (2000). Bilingualism in mental retardat ion: Some prospective views. Saggi: Child Developm ent and Disabilities, 26 (1), 57-64. Rosenbusch, M. H. (1998). Is foreign la nguage education for all learners? In M. Met, (Ed .), Critical issues in early se cond language learning: Building for our childrens future (pp. 57-59). Glenview, IL: Addison-Wesley. Sacks, H., & Schegloff, E. (1973). Opening up closings. Semiotica, 8, 289-327. Sacks, H., Schegloff, E. A., & Jefferson, G. (1974). A simplest systematics for the organization of turn-taking for conversation. Language, 50, 696-735. Salaberry, M. R. (1996). A theoretical foundation for the development of pedagogical tasks in computer mediated communication. Calico Journal, 14 (1), 5-34. SAS Institute Inc. (2004). SAS [version 9.1.3] Language Reference: Concepts. Cary, NC: SAS Institute Inc. Savignon, S. (1972). Communicative competence: An experiment in foreign language teaching Philadelphia: Center for Curriculum Development. Scarcella, R., & Higa, C. (1981). Inpu t, negotiation and age differences in second language acquisition. Language Learning, 31 409-438. Schachter, J. (1984). Negative in put. In W. Rutherford, (Ed), Second language acquisition and language universals. Amsterdam: John Benjamins.
320 Schachter, J. (1991). Corrective f eedback in historical perspective. Second Language Research, 7 89-102. Schachter, J., Rounds, P. L., Wri ght, S., & Smith, T. (1998). Comparing conditions for learning syntactic patterns: Attention and awareness Unpublished manuscript, University of Oregon, Eugene. Schegloff, E. A., Jefferson, G., & Sacks, H. (1977). The preference for selfcorrection in the organization of repair in conversation. Language, 53 361-382. Schegloff, E. A., & Sacks, H. (1973). Opening up closings. Semiotica, 7 289327. Schmidt, R. W. (1993). Interaction, acculturation, and the acquisition of communicative competence. In N. Wolfson & J. Manes (Eds.). Sociolinguistics and sec ond language acquisition (pp. 137-174). Rowley, MA: Newbury House. Schuyler, M. (1982). A readability fo rmula for use on microcomputers. Journal of Reading, 26 560-591. Schwartz, B. (1993). On explicit and negative data affecti ng competence and linguistic behavior. Studies in Second Language Acquisition, 15 147-163. Seliger, H. W., & Shohamy, E. (1989). Second language research methods. Oxford: Oxford University Press. Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics, 10, 209-231.
321 Silver, R. E. (2000). Input, output, and negotiation: Conditions for second language development. In B. Swierzbin, F. Morris, M. E. Anderson, C. A. Klee, & E. Tarone (Eds.), Social and cognitive factors in second language acquisition: Selected proceedings of the 1999 Second Language Research Forum (pp. 345-371). Somerville, MA: Cascadilla Press. Sinclair, J. M., & Coul thard, R. M. (1975). Towards an analysis of discourse: The English used by teachers and pupils London: Oxford University Press. Slimani, A. (1992). Evaluation of classroom intera ction. In C. Anderson & A. Beretta (Eds.) Evaluating second language education (pp. 197-221). Cambridge: Cambridge University Press. Smith, B. (2003). The use of communication strategies in computer-mediated communication. System, 31 29-53. Sotillo, S. (2000). Discourse functions and syntactic complexity in synchronous and asynchronous communication. Language Learning and Technology, 4 (1), 82-119. Retrieved S eptember 17, 2006, from http://llt.msu.edu/vol4 num1/sotillo/default.html Spada, N. (1987). Relationships between in structional differences and learning outcomes: A process-product study of communicative language teaching. Applied Linguistics, 8, 137-161. Spada, N., & Frhlich, M. (1995). COLT. Communicative Orientation of Language Teaching observation scheme: Coding conventions and applications Sydney, Australia: National Centre for English Language Teaching and Research.
322 Spada, N., & Lightbown, P. M. (1993). Instruction and the development of questions in L2 classrooms. Studies in Second Language Acquisition, 15 205-224. Sparks, R. (1995). Examini ng the linguistic coding differences hypothesis to explain individual differences in foreign language learning. Annals of Dyslexia, 45, 187-214. Sparks, R. L., & Ganschow, L. (1991, Spring). Foreign language learning differences: Affective of native language aptitude differences? Modern Language Journal, 75 (1), 3-16. Sparks, R. L., & Ganschow L. (1993). The impact of native language learning problems on foreign language learning: Case study illustrations of the linguistic coding deficit hypothesis. Modern Language Journal, 77 58-74. Sparks, R. L., Ganschow, L., Kenneweg, S ., & Miller, K. (1991) Use of an OrtonGillingham approach to teach a foreign language to dyslexic/learningdisabled students: Explicit teachi ng of phonology in a second language. Annals of Dyslexia, 41 97-118. Sparks, R., Ganschow, L., Pohlman, J. Skinner, S., & Artz er, M. (1992). The effects of Multisensory Struct ured Language instruction on native language and foreign languag e aptitude skills of at-risk high school foreign language learners. Annals of Dyslexia, 42, 25-53. SPSS for Windows. (2001). Statistical Package for the Social Sciences (Rel. 11.0.1) [Computer Software]. Chicago, IL: SPSS Inc.
323 Stevens, J. (2002). Applied multivariate statis tics for the social sciences (4th ed.). Mahwah, NJ: Lawrence Erlbaum Associates. Stevick, E. (1976). Memory, meaning and method. Rowley, MA: Newbury House. Stevick, E. (1980). Teaching languages: A way and ways. Rowley, MA: Newbury House. Sullivan, N., & Pratt, E. (1996). A co mparative study of two ESL writing environment: A computer-assisted cl assroom and a traditional oral classroom. System, 24 491-501. Swain, M. (1985). Communicative com petence: Some roles on comprehensible input and comprehensible output in it s development. In S. Gass & C. Madden (Eds.), Input in second language acquisition Rowley, MA: Newbury House. Swain, M. (1995). Three functions of output in second languag e learning. In G. Cook & B. Seidlhofer (Eds.), Principle and practice in applied linguistics: Studies in honour of H.G. Widdowson. Oxford: Oxford University Press. Swain, M., & Lapkin, S. ( 1995). Problems in output and the cognitive processes they generate: A step toward s second language learning. Applied Linguistics, 16 371-391. Swain, M., & Lapkin, S. (1998). Intera ction and second language learning: Two adolescent French immersions students working together. Modern Language Journal, 82 320-337.
324 Tarone, E. (1994). A summary: Resear ch approaches in studying secondlanguage acquisition or If the Shoe Fits In E. E. Tarone, S. M. Gass, A. D. Cohen (Eds.) Research methodology in second-language acquisition (pp. 323-336). Hillsdale, NJ: Lawrence Erlbaum Associates. Tashakkori, A., & Teddlie, C. (1998). Mixed methodology: Co mbining qualitative and quantitative approaches. Applied Social Research Methods Series (Vol. 46). Thousand Oaks, CA: Sage. Tashakkori, A., & Teddlie, C. (2003). The past and future of mixed methods research: From data triangulation to mixed model designs. In A. Tashakkori & C. Teddlie (Eds.), Handbook of mixed methods in social and behavioral research (pp. 671-701) Thousand Oaks, CA: Sage. Tharp, R., & Gallim ore, R. (1988). Rousing minds to life. New York: Cambridge University Press. The National Education Institute of the Republic of Slovenia. (2004). Its all about education. Retrieved September 16, 2006, from http://www.zrss.si/default.asp?link=ang Tomasello, M., & Herron, C. (198 8). Down the garden path: Inducing and correcting overgeneralization errors in the foreign language classroom. Applied Psycholinguistics, 9 237-246 Tomasello, M., & Herron, C. (1989) Feedback for language transfer errors: The garden path technique. Studies in Second Language Acquisition, 11, 385395.
325 Torres, I. (1996). Report on current practi ce: K-8 program in Ferndale, Michigan. In M. Met, (Ed.), Critical issues in early sec ond language learni ng: Building for our childrens future (pp. 60-61). Glenview, IL: Addison-Wesley. Truscott, J. (1998). Noticing in second l anguage acquisition: A critical review. Second Language Research, 14 103-135. van Lier, L. (1988). The classroom and the language learner New York: Longman. Varonis, E., & Gass, S. (1985). Non-nativ e/non-native conversations: A model for negotiation of meaning. Applied Linguistics, 6 (1), 71-90. Vygotsky, L. S. (1934/1987). Thinking and speech. In R. W. Rieber & A. S. Carton (Eds.), The collected works of L.S. Vygotsky, Volume 1: Problems of general psychology (pp. 39-285). New York: Plenum. Warschauer, M. (1995). E-Mail for English teaching. Washington, DC: Teachers of English to Speaker s of Other Languages. Warschauer, M. (1996). Comparing face-t o-face and electronic discussion in the second language classroom. CALICO Journal, 13(2&3), 7-26. Warschauer, M. (1997). Computer-mediat ed collaborative lear ning: Theory and practice. Modern Language Journal, 81 470-481. Warschauer, M., & Healy, D. (1998). Computers and language learning: An overview Language Teaching, 31 57-71. Retrieved September 17, 2006, from http://acadweb.snhu.edu/riabov_l yra/Lyra_Web/W_2_MW_%20Computer s%20and%20Language%20Learning%20An %20Overview.htm.
326 Warschauer, M. (1999). Electronic literacies: languag e, culture and power in online education Mahwah, NJ: Lawrence Erlbaum Associates. Wells, G (1999). Dialogic Inquiry: Toward a socioc ultural practice and theory of education Cambridge, UK: Cambridge University Press. Wells, G. (2002). The role of dialogue in activity theory. Mind, culture, and activity, 9 (1), 43-66. White, L. (1989). Universal grammar and second language acquisition (Vol. 1). Philadelphia: John Benjam in's Publishing Company. White, L. (1991). Adverb placement in second language acquisition: Some effects of positive and negativ e evidence in the classroom. Second Language Research, 7 133-161. Wilson, D. R. (July/August, 2000). Building bridges to inclusive foreign language education through appropriately applied technologies. Presentation at FLEAT IV Fourth Conference on Foreign Language Education and Technology in Kobe, Japan Retrieved September 16, 2006, from http://www.specialeducati onalneeds.com/mfl/fleat4/ Wing, B. H. (Ed.) (1996). Foreign languages for all: Challenges and choices. Lincolnwood, IL: National Textbook Company. Yenne, B. (1990). Ytalk (version 1) [Computer so ftware]. Retrieved online September 16, 2006, from http://www. iagora.com/~espel/ytalk/ytalk.html
327 Appendix A
328 The Structure of the Educat ion System in Slovenia. Reprinted from Eurydice (2001/2002).
329 Appendix B
330 Foreign Language Options in General Secondary Schools Translated and reprinted from Grosman et al. (1998 ) English today English after restructuring Module A First foreign language: continuous learning from elementary school. (4 years elementary school + 4 years general secondary education). Total: 8 years First foreign language: continuous learning 6 years elementary school + 4 years general secondary school). Total: 10 years. Module B Second foreign language: just beginning. (4 years general education school). Total : 4 years Second foreign language: just beginning (4 years general education school) Total : 4 years Module C Second foreign language: continuous learning (in elementary school as a required elective for 3 years + 4 years general secondary school). Total: 7 years
331 Appendix C
332 Task. From Ligon, F., Tannenbaum, E., & Richardson Rodgers, C. (1991). More picture stories: Language and problem-posing acti vities. White Plains, NY: Longman.
333 Appendix D
334 IRB Approval for Pilot Study
335 Appendix E
336 IRB Approval for Present Study
337 Appendix F
338 IRB Course Completion
339 Appendix F IRB Course Completion (continued)
340 Appendix G
341 IRB Core Course Completion
342 Appendix H
343 Permission to Conduct Research
344 Appendix I
345 Student Background Questionnaire Thank you for completing this questionnaire Demographics: 1. Ime (Name) : _____________________ 2. Spol (Gender): ____moki (male) ____enska (Female) 3. Starost (Age): ___________ 4. ola (School): _________________ 5. S kricem ozna i vrsto ole, ki jo obiskuje (Check appropriate box) Osemletka (Eight Year Elementary School) Devetletka (Nine Year Elementary School) Gimnazija (General High School) 6. Ozna i kateri razred obiskuje 5. razred (class 5) 6. razred (class 6) 7. razred (class 7) 8. razred (class 8) 9. razred (class 9) 10. razred (class 10) 11. razred (class 11) 12. razred (class 12) Background 7. Ali si kdaj ponavljal razred? (Did you ever repeat a grade?) Da (Yes) _____ (Ne) No______ 7.a e si odgovoril z da, kaken je vzrok, da si ponavljal? (If yes, please note reason for repeating grade?) ______________________________________________________________________________________ ______________________________________________________________________________________
346 Appendix I Student Background Ques tionnaire (continued) 8. Ali ima morda kakne posebne teave (u ne ali razvojne), ki te ovirajo pri u enju? (Do you have any special circumst ances (medical) that makes it more difficult to learn? _____________________________________________________________________________ _____________________________________________________________________________ ____________________________________________________________________ 9. Materin Jezik (Native Language) ________________________ ________________ Foreign Language 10. Ali govori oz. se u i kaken drugi tuji jezik? (Do you speak or study other language/s?) Da (Yes) _____ (Ne) No______ 10a. e si odgovoril/a z da, prosim da ocenis/opie svoje znanje. Na primer: Lahko berem v italijan ini. Lahko berem in piem v nem ini. Lahko govorim kitajsko, vendar ne teko e. Teko e govorim franco ino. (If yes, specify which language/s and how would you grade your ability in each language. For example: I can read in Ita lian; I can read and write in German; I can speak, but not fluently in Chinese; I can speak fluently in French; etc.) __________________ __________________ __________________________ __________________ __________________ __________________________ __________________ __________________ _______________ ___________
347 Appendix I Student Background Ques tionnaire (continued) 11. Kako dolgo se e u is angle ino? (How long have you been studying English?) ________________________ __________________ ____________________ 12. Zakaj se u i angle ino? (Why are you studying English?) ________________________ _____________________ _________________ 13, Ali si kdaj obiskal/a angleko govore o dravo? (Have you visited a English speaking country?) Da (Yes) ____ Ne (No) ____ 13a. Ce si odgovoril/a z da, navedi kat ero dravo si obiskoval/a? Kdaj? Za koliko asa? (If yes, which country? When? For how long?) __________________ __________________ __________________________ __________________ __________________ __________________________ __________________ __________________ _______________ ___________ 14. Ali ima stike z angleko govore imi ljudmi izven razreda? (Do you have any contact with native speakers of English outside the classroom?) Da (Yes) ____ Ne (No)____ 14a. Ce si odgovoril z da, kolikokrat? (If yes, how frequently?) Pogosto (Often) _____ Ob asno (Occasionally) ____ Redko (Rarely) __
348 Appendix I Student Background Ques tionnaire (continued) Technology 15. Ali uporablja ra unalnik? (Do you use a computer) Da (Yes) __ Ne (No)_ 16. e si odgovoril/a z da, koliko asa e uporablja ra unalnik? (How long have you been using comput ers?) ______ (leta/years) 17. Za kakne namene uporablja ra unalnik? Ozna i vse primerne odgovore (What do you use computers for? Check as many as applicable): _____ Elektronska Pota (E-mail) _____ Pisanje (Word-processing -Microsoft Word, WordPerfect, etc.) _____ Igre (Games) _____ Deskanje po spletu (Browsing the Inte rnet -Internet Explorer, Netscape, etc.) _____ Programiranje (Programming) _____ Elektronski chat pogovori (Online Chat -IRC, Yahoo, MSN Inst ant messenger, etc.) _____ Forumi (Electronic Bulletin/Discussion Boards) _____ Ostalo, prosim navedi. (Others, please specify): __________________________ 18. Ali se dobro po uti pri uporabi ra unalnika? (How comfortable are you working with computers?) _____ Zelo dobro (Very comfortable) _____ Dokaj dobro (Somewhat comfortable ) _____ Nezadovoljen (Uncomfortable) _____ Zelo nezadovoljen (Very uncomfortable) 19. Ali kdaj uporablja forume pri pouku? (Do you use electronic bulletin/discussion boards in your classes ?) Da (Yes) _____ Ne (No) _____
349 Appendix I Student Background Ques tionnaire (continued) 19a. e da, kolikokrat na teden (If yes, how frequently? ________ ( krat na teden /times per week) 20. Ali uporablja forume (oz. discussion boards) za osebno uporabo? (Do you use electronic bulletin/discussion boards for personal use ?) Da (Yes) _____ Ne (No) _____ 20a. e da, kolikokrat (If yes, how frequently?) ________ ( krat na teden /times per week) 21. Ali uporablja spletne pogovorne c hat programe, kot so IRC, Messenger, Yahoo, itd. pri pouku? (Do you use chat programs (AOL, Yahoo, MSN Instant messenger, etc.) in your classes ?) Da (Yes) _____ Ne (No) _____ 21a. e da, kolikokrat? If yes, how frequently? ________ (krat na teden/times per week) 22a. Ali uporablja spletne pogovorne chat programe kot so IRC, Messenger, Yahoo za osebno uporabo? (Do you use chat programs (AOL, Yahoo, MSN Instant messenger, etc.) for personal use ?) Da (Yes) _____ Ne (No) _____ 22a. e da, kolikokrat? (If yes, how frequently?) ________ (krat na teden/times per week) Najlepsa hvala! This questionnaire was adapted from ORelly (1999), p. 157
350 Appendix J
351 Codebook Corrective Feedback Coding Scheme Interaction Analysis Codebook (a dapted from Castaeda, 2005) Unit of Data Collection : The unit of analysis for this research study is the error treatment sequence. The error treatment se quence refers to the learners initial turn containing an error (P1), the dyad mem bers response (P2) to the error, and the learners reaction or response to the co rrection (P1). If a l earner is identified with a special need the notat ion in the codebook is indicated with an s at the end of the abbreviation. For example: P1S or P2S. Error: An error is defined as an ill-formed language utterance, an unacceptable utterance in the target language. The vari ous types of errors below served as the a priori categories in the present study. New varieties of errors were not found and therefore a new emergent theme or category was not warranted. E-01 Grammatical: Grammatical errors produce a grammatical construction that violates the grammar of the target language. Inappropriate word order or usage of articles and syntactical errors also are coded as a grammatical error. E-02 Lexical: Lexical errors are the use of the wrong word or missing lexical item in an utterance (i.e. missing lexical items su ch as prepositions, nouns, adjectives; however, not including articles as ar ticles are functional not lexical free morphemes and their usage is related to rule application in an utterance. Inaccurate, imprecise, or inappropriate choices of lexical items and non-target derivations of nouns, verbs, adverbs, and adjectives constitute examples of lexical errors. E-03 Orthographic Conventions: These types of errors include omissions of accent marks and letters unique to the Engl ish alphabet. These include : q, w, x, y. In addition, errors ma y include additions of lette rs unique to the Slovenian alphabet. These include , . E-03a: Orthographicons: These also include emoticons, exaggerations, and abbreviations. These instanc es are coded under orthographic conventions, but were not counted as errors. Punctuation and/or capitalization were not coded as an erro r and were ignored; namely due to to the type of interaction, which is nei ther a written nor an oral format, the Appendix J Peer (initial turn with error(P1) Learner response (P2) Reaction/Response (P1)
352 Appendix J Codebook (continued) frequency of capitalization and punctuati on errors in almost every turn, and in none of the instances did the punctuation or capitalization receive any type of corrective feedback. E-04 Typographical and Spelling: A typographical error is one made while inputting text via a keyboard, the error is made despite the user knowing the spelling of the word. This usually result s from the persons inexperience using a keyboard, from rushing, fr om not paying attention, or carelessness. A spelling error is one made when forming words with letters and the letters are not put in the acceptable order. In this study, it is impossible to know whether the learner made a typographical error or spelling error and therefor e these were put in the same category. It should also be noted that omission or addition of specific orthographic letters (under Orthographic Conventions) were combined with the typographical and spelling category, as it was difficult to determi ne if an omission or addition of orthographic convention we re not really typographical or spelling errors. E-05 Unsolicited use of L1: Use of the native language (L1) is not an error per se, but it is interesting to examine at whic h points students turn to L1 and their peers reaction to the unsolicited use of the L1. Usage of L1 was counted in the error turns. E-06 Multiple: When more than one type of error occurs in a student turn (for example, lexical and grammatical) these were coded as multiple. E-07 Emergent: An emergent error category was not found. X-L1: Content feedback with L1 : When a turn includes a content question that includes an L1 term for clarification (e.g. how do you say POJDI SPAT), this was not coded as an error in relation to the corrective feedback, but to the content/question feedback. The L1 used was fo r puposes of content clarification. Therefore, only the s pecific feedback to the conten t/question were coded, if there as an error. X-L3: Usage of L3: when a turn contains utter ances with the usage of a third language, which is neither English or Slov ene. Within this category, utterance with L3 or the third language being studied by the partici pants were included, but not included in the overall data set. These were counted as lexical error, but were separately coded to view instances of L3 usage.
353 Appendix J Codebook (continued) Corrective Feedback: Corrective feedback is defined as a response to a learner error made by the dyad member that pr ovides the learner with information about what is acceptable and unacceptable in the target language. Using Lyster and Rantas (1997) findings of the various types of corre ctive feedback, the following a priori categories for corrective feedba ck were used in the present study. A different variety of corrective feedback was found, namely due to the nature of discourse taking place in a synchronous environment as well as the interaction among peers. This constituted th e emergent theme or category. If a learner is special needs the notati on indicate an s at the end of the abbreviation. For example: P1S or P2S. CF-O1 Explicit correction: This is the explicit (direct) provision of the correct form. CF-02 Recasts: The learner dyad members (P2) re formulation of all or part of a learners (P1) utterance excl uding the error is a recast. CF-03 Negotiation of form: Negotiation of form was used following Lyster and Rantas (1997) definition of negotiation of form. Elicitation, metalinguistic, clarification request, and repet ition are types of correc tive feedback that were compressed into the single category of negotiation of form. These feedback types can elicit or lead the learner to r epair. In contrast, as Lyster and Ranta (1997) found, recasts and expl icit correction lead to lo w rates of student repair because they already provide the lear ner with the correct form or forms. Elicitation, metalinguistic, clarification request, and repetition types of corrective feedback can, on the other hand lead to student generated repair and can be considered negotiati on of form. CF-04 Clarification requests: These indicate to the lear ner (P1) either that the utterance is not understood by the dyad mem ber (P2) or that the utterance is illformed in some way or that a repetition or a reformulation is required on the part of the learner (P1). CF-05 Metalinguistic feedback: Metalinguistic feedback are comments that indicate that there is an error somewher e. These comments can be in the form of grammatical metalanguage or can point to the nature of the error. Peer (initial turn with error(P1) Learner response (P2) Reaction/Response ( P1 )
354 Appendix J Codebook (continued) CF-06 Elicitation: Elicitation is when, the dyad me mber (P2) directly elicits the correct form from the learner (P1). These elicitations can come in various forms. The dyad member (P2) allows the learner to fill in the blank, may use questions to elicit the correct form, or can ask the learner (P1) to refo rmulate the utterance CF-07 Repetition: Whenever, a dyad member (P2) repeats the learners (P1) erroneous utterance in isolation th is is defined as a repetition. CF-08 Emergent-Feedback Request: Feedback request is when, a student requests feedback from their peer by usi ng either the L1 or L2. For example: mum took the girl to the emergency room because she had a stomackacke (how do you spell this?) X-SC: Self-correction: Self-correction is when students self-identify their error within the same or within their immediat e turn after the e rror. It is coded separately, because it doesnt belong withi n the scope of corrective feedback as other initiated, but within themselves.
355 Appendix K
356 Instructions to Participants in Slovene Navodila Pred vama je 10 slik nekega dogodka. Ti imas polovic slik tega dogodka in tvoj partner ima drugo polovico. Tvoja naloga je, da postavi slike v pravilni vrstni red. Zapomni si, da ti imas polovic o zgodbe in tvoj partner im a drugo polovico. Sodeluj s svojim partnerjem, tako da ugotovita pravilni vrstni red zgodbe in nato skupaj sestavita pravilni vrstni red o dogajanju na slikah. Torej, s partnerjem preko konferen nega orodja MSN: 1. Opiita slike 2. Slike postavita v pravilni vrstni red glede na dogajanju na slikah (1, 2, 3, 4, 5, 6, 7, 8, 9, in 10) 3. Prilepita (copy/paste) celoten pogovor v WORD (ali textpad) 4. Dvignita roko, ko kon ata. Zapomni si: 1. Uporabljaj samo angleki jezik. 2. Ne sprauj soseda ali profesorja. 3. Ne uporabljaj slovarja. 4. Bodi im bolj natan en slovni no, pri rkovanju in pri izbiri besedi a 5. Vpraaj partnerja e kaj ne ve 6. Obvesti raziskovalko z dvigom roke, ko s partnerjem kon ata. Raziskovala bo prila k vama in bo shranila kon ano nalogo na disketo. Najlepa hvala za sodelovanje! Annmarie G. Zoran University of South Florida
357 Appendix L
358 Instructions to Participants in English Instructions: You and your partner have 10 pictures of an event. Your partn er has half of the pictures and you have the other half. Your task is to place the pictures in the correct order. Remember! You have only half of the whole story. Your partner has the other half. So, using the MSN conferencing tool you and your partner will: 1. Accurately, describe what the pictures are about; 2. And place them in the correct sequence of events (1, 2, 3, 4, 5, 6, 7, 8, 9, and 10); 3. Copy/Paste your finished transcr ipt into WORD (or textpad). 4. Raise your hand when you are finished. Remember: 1. Use only English. 2. Do not ask questions to your neighbor or teacher. 3. Do not use a dictionary. 4. If you are unsure, ask your partner 5. Be as precise as possible in both grammar, spelling, and vocabulary Ask your partner, if you are unsure about anything. 6. When you are finished, let the researc her know by a raise of hands. The researcher will come to your station and save your finished activity on a disk. Thank you for your participation! Annmarie G. Zoran University of South Florida
359 Appendix M
360 Corrective Feedback Coding Interaction Analysis Coding Form for Synchronous Interaction (modified fr om Castaeda, 2005) Dyad member 1 (P1): ________________________ ___________________ Dyad member 2 (P2): ________________________ __________________ Date of Interacti on: ________________ ___________________ Coder: __________________ ___________________________ Column 1 Column2 Column 3 Column 4 Column 5 Column 6 Column 7 (follow up study) Column 8 (follow up study) Notes for Special Needs Interlocutors Turn # Error Yes/No Error Type Corrective Feedback Yes/No Corrective Feedback Type Learner Response Yes/No Learner Response Type Turn Note: If a learner with special needs the notation will indicate an s at the end of the abbreviation. For example: P1S or P2S Peer (initial turn with error(P1) Learner response (P2) Reaction/Response (P1)
361 About the Author Annmarie Gorenc Zoran received her Bachelors Degree from the University of Illinois at Chicago. She taught EFL to elementary school students and to adults in Slovenia, while attending the Pedagogy and Andragogy program at the University of Ljubljana. Her inte rest in education and research led her to enroll in the Second Language Acquisition a nd Instructional Technology (SLA/IT) Ph.D. program at the Univer sity of South Florida. While in the PhD program, Ms. Zor an was quite active She has taught ESL classes, university level face-to-fa ce and distance learning, worked as the SLA/IT program assistant, CALL consul tant, curriculum builder for online courses, served as Secretary and Pres ident of the SLAQ student organization, has served on several committees, has presented papers and published/copublished several papers and book chapters. Her research interests are in computer-mediated-communicati on, teacher education, whilst furthering research and awareness within special education and SL learning. Ms. Zoran is currently residing in Slovenia.
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001914913
007 cr mnu|||uuuuu
008 071030s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001772
Gorenc Zoran, Annmarie.
CALLing all learners :
b an explanatory integrative research study of EFL learner-learner corrective feedback patterns within on-line synchronous environments
h [electronic resource] /
by Annmarie Gorenc Zoran.
[Tampa, Fla] :
University of South Florida,
ABSTRACT: This mixed methods research study centers on learner-learner interactions; thus, contributing to the on-going investigation within negotiation and interaction, computer-mediated-communication and its role in second language learning. The specific aim was to investigate corrective feedback types, incidences, and the relationship between error and feedback type among peers within online synchronous environments in EFL classes in Slovenia, Europe. Interactional characteristics of corrective feedback with learners having a documented special need (SN) also were explored using qualitative analyses. The study encompassed 208 students that were randomly placed into 104 dyads within intact classes of Grades 7, 8, 10, and 11. There were 32 dyads in Grade 7, 16 dyads in Grade 8, 24 dyads in Grade 10, and 32 dyads in Grade 11. Three participants had a documented special need. Quantitative analysis did not reveal statistical significant difference in the incidence of corrective feedb ack and grade level, the relationship among the type of corrective feedback and grade level, or the relationship between learner error and type of corrective feedback across grade levels. Corrective feedback types were similar to those studied in traditional classroom research (i.e., explicit corrections, recasts, negotiation of form). However, descriptive statistics and qualitative analyses revealed conversational techniques that are specific to text-based online discourses providing insight into interactional characteristics among interactants within a discourse environment that differs both from speech and written texts. Consequently, an additional corrective feedback type emerged from the data, coded as feedback request. The most frequent corrective feedback type provided was explicit corrections. Frequency data revealed that corrective feedback tended to decrease as the grade level increased. Data with SN learners indicated distinctive discourse techniques.Overall, low incidences^ of corrective feedback and error types might have been affected by the learner's developmental levels, social readiness, and/or psychological readiness (Oliver, 1998), as well as the learner's individual conversational styles and socio-cultural factors. Consequently, further research is warranted in examining these factors. In addition, longitudinal studies are warranted in examining whether online negotiated work lead towards L2 acquisition. Finally, the role of phantom corrective moves when coding qualitative online text data also need to be examined further.
Dissertation (Ph.D.)--University of South Florida, 2006.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 361 pages.
Adviser: Antony K. Erben, Ph.D.
Special education needs.
Education in Slovenia.
x Secondary Education
t USF Electronic Theses and Dissertations.