Signature Page

THE INTERACTION OF TOPIC CHOICE AND TASK-TYPE

IN THE EFL CLASSROOM

A Dissertation

Submitted

to the Temple University Graduate Board

In Partial Fulfillment

of the Requirements for the Degree of

Doctor of Education

John P. Thurman

August, 2008 ©

2008

John Thurman

iii ABSTRACT

This study was an examination of the effect that three levels of topic choice

(no choice, limited choice, and complete choice) would have on students’ Task Interest and Task Self-efficacy (Study 1, 78 participants), and on three aspects of students’ oral output: Accuracy, Complexity, and Fluency (Study 2, 42 participants in 21 pairs). Also examined were the effects that three types of tasks (descriptive, narrative, and decision- making) exerted on these five variables. Data were collected using a questionnaire for

Study 1 and recording the participants’ conversations for Study 2. Data were collected in nine consecutive treatments to examine the main effects of choice and task, and the interaction effects of choice and task, using two-way repeated-measures ANOVAs.

For Study 1, limited choice promoted Task Interest for the descriptive and narrative tasks, and Task Self-efficacy for the narrative and decision-making tasks to a statistically significant degree. In addition, the descriptive task had the highest Task

Interest for the no choice and complete choice treatments and had the highest Task Self- efficacy for the no choice of topic treatment. The findings generally indicated that the participants were more interested in the task when there was choice, and that this led to higher levels of Task Interest and Task Self-efficacy.

The Study 2 results indicated that Complexity was significantly higher when choice was introduced for the descriptive and narrative tasks. Accuracy and Fluency were not influenced to a statistically significant degree by choice, but they were positively influenced nonetheless. Attentional resources may have been freed up when choice was

iv introduced and the participants may have been more willing to take risks, both possibly causing the significantly higher levels of complexity for choice.

Suggestions for further research include a closer examination of the process students use when choosing a topic and examining ways for a more efficient method of introducing choice into the task-based language teaching syllabus.

v ACKNOWLEDGEMENTS

Many people have helped to bring to fruition this study. First and foremost, I would like to thank Dr. David Beglar. His advice, patience, and guidance have brought this about from a nebulous beginning. Without his help, this study would never have happened.

I would also like to thank the members of my committee for the time they took to read and participate in the defense committee. I would like to thank Dr. Donna Tatsuki, from Kobe City University of Foreign Studies, and Dr. John Norris, from the University of Hawai’i. I would also like to thank Dr. Kenneth Schaefer from Temple University for his support. All of your comments are reflected in this final product Last but not least, I would like to thank Dr. Mark Sawyer from Kwansei Gakuin University. My first class at

Temple University was taught by Dr. Sawyer eight years previous. It was only proper that he should be there at the end.

Others helped as well along the way. Dr. J.D. Brown was influential and helpful in seeing this study to its close. Others have given me their knowledge during this journey.

They are Drs. Dwight Atkinson, David Andrich, Christine Casanave, Nick Ellis, Rod

Ellis, Robert Gardner, Gabriele Kasper, Michael Long, Peter MacIntyre, Sandra McKay, and Steven Ross. All are present in this study and it could not have been done without the knowledge they passed on. For helping me to begin this doctoral program, I would like to thank Dr. Jan Eyring of California State University, Fullerton, Dr. Robyn Najar of

vi Flinders University, and Mr. Kinji Ikuta, formerly of Kochi Women’s University. Without them, I would not have been able to get this far.

Many people have helped with their time to bring this study to completion. I would especially like to thank Hirohito Kusamitsu for arranging help with a great part of the transcribing. I would also like to thank Naoko Okumura, Maki Fujiwara, Ikuko

Higashida, Rieko Tokunaga, and Yuko Tsuji for additional help with transcribing. I would also like to thank Mayumi Matoba for help with the after-task survey. For help with reading the thousands of data cards, I would like to thank Naoki Naito, Kenji

Fujiwara, and Mika Arima for their help with this. I would also like to thank Kiho

Tanaka of Doshisha University for help with survey materials. Lastly, I would to thank

Dr. Toshihiko Yamaoka of Hyogo University of Education, all the members of my doctoral cohort, my family and friends, and everyone else who gave their moral support that helped me to finish this study.

Lastly, but certainly not leastly, I would like to thank my wonderful wife for patience and understanding during this journey. Thank you Setsuko.

vii To my wonderful wife Setsuko

I dedicate this project. Without your

understanding and patience

this would never have been done.

Thank you, my wife.

viii TABLE OF CONTENTS

PAGE

ABSTRACT...... iv

ACKNOWLEDGEMENTS...... vi

DEDICATION...... viii

LIST OF TABLES...... xviii

LIST OF FIGURES...... xxiii

CHAPTER

1. INTRODUCTION �� 1

Task-based Language Teaching �� 1

Task-Based Language Teaching in Japan �� 2

Motivation and Language Learning �� 2

The Aims of This Study �� 3

The Significance of This Study �� 4

The Delimitations of This Study �� 6

The Audience for This Study �� 6

The Outline of This Study �� 8

2. LITERATURE REVIEW �� 10

Tasks in Language Teaching �� 10

Definitions of a Task in Task-Based Language Teaching �� 10

ix Task Features �� 14

Open and Closed Tasks �� 14

Convergent and Divergent Tasks �� 15

One-Way and Two-Way Tasks �� 15

Required or Optional Information Exchange Tasks �� 16

Task Topic Influences �� 16

Discourse Mode �� 17

Task Difficulty �� 18

Robinson’s Model of Task Complexity, Task Difficulty, and Task Conditions �� 19

Comprehensible Output and Task Production �� 21

Assessing Output �� 23

Research on Accuracy, Complexity, and Fluency �� 28

Working Definition of a Task �� 31

Choice �� 31

Culture and Choice �� 32

The Effect of Too Much Choice �� 34

Self-Determination Theory �� 36

Amotivation �� 37

Extrinsic Motivation �� 38

Intrinsic Motivation �� 40

Three Components of Intrinsic Motivation �� 41

Autonomy and the Japanese Self �� 42

x Amae �� 44

Self-Determination Theory and Language Learning Motivation �� 45

Self-Determination Theory and Language Learning Motivation in Japan �� 47

Autonomy and Language Learning �� 49

Definitions and Early Research �� 50

Autonomy as a Degree of Capacity �� 50

Littlewood’s Model of Language Learning Autonomy �� 51

Proactive and Reactive Autonomy �� 53

van Lier’s Model of Language Learning Autonomy �� 54

Nakata’s Model of Language Learning Autonomy �� 56

Oxford’s Perspectives on Language Learning Autonomy �� 57

Cultural Influences and Language Learning Autonomy �� 58

The Link between Autonomy and Motivation in Language Learning �� 60

Working Definition of Language Learning Autonomy �� 61

Gaps in the Literature �� 62

Research Questions �� 63

Study 1 �� 63

Study 2 �� 65

3. METHOD �� 68

Participants �� 68

Research Setting �� 72

The Variables in This Study �� 73

Materials �� 74 xi Task Materials �� 74

Task Materials Used for the Treatment Sessions �� 75

After-task Survey �� 78

Procedures �� 80

The Design of this Study �� 84

Data Analyses �� 85

Using the Rasch Model for Interval Scaling of the Variables �� 86

Procedures for Calculating Production Data �� 89

Calculating Accuracy �� 90

Calculating Complexity �� 92

Calculating Fluency �� 94

Why Some Measures Were Not Used in This Study �� 96

4. PRELIMINARY RESULTS OF STUDY 1 �� 98

Missing Cases and Removing Univariate Outliers �� 98

Descriptive Statistics�� 99

Data Transformation ��107

Removing Multivariate Outliers ��109

Factor Analysis of the Data ��109

Review of the Data Using Rasch Analysis �� 115

5. RESULTS OF STUDY 1 �� 124

Research Question 1 ��124

Research Question 2 ��130

Research Question 3 ��135 xii Research Question 4 �� 141

6. RESULTS OF STUDY 2 �� 147

Descriptive Statistics�� 147

Final Data Analysis �� 155

Research Question 1 ��156

Research Question 2 �� 162

Research Question 3 �� 167

Research Question 4 �� 174

Research Question 5 �� 179

Research Question 6 �� 185

7. DISCUSSION �� 190

Study 1 �� 190

Research Question 1 �� 190

Research Question 2 �� 194

Summary of the Discussion for Task Interest �� 195

Research Question 3 �� 197

Research Question 4 ��200

Summary of the Discussion for Task Self-efficacy �� 201

Study 2 ��202

Research Question 1 ��202

Research Question 2 ��204

Research Question 3 ��205

Research Question 4 ��208 xiii Research Question 5 �� 210

Research Question 6 �� 212

General Discussion of Study 2 �� 214

8. CONCLUSION �� 216

Integrating Task-type and Choice �� 216

Limitations of This Study �� 217

Suggestions for Further Research ��220

Implications of This Study ��222

Conclusion �� 225

REFERENCES CITED �� 227

APPENDICES

A. CLOZE TEST �� 251

B. SYLLABUSES OF THE JAPANESE PROFESSORS FOR THE TREATMENT CLASSES ��256

C. DESCRIPTIVE TASK, NO CHOICE OF TOPIC, FIRST ROUND �� 259

D. DESCRIPTIVE TASK, NO CHOICE OF TOPIC, SECOND ROUND ��260

E. DESCRIPTIVE TASK, LIMITED CHOICE OF TOPIC, FIRST ROUND, TASK A �� 261

F. DESCRIPTIVE TASK, LIMITED CHOICE OF TOPIC, FIRST ROUND, TASK B �� 262

G. DESCRIPTIVE TASK, LIMITED CHOICE OF TOPIC, FIRST ROUND, TASK C �� 263

H. DESCRIPTIVE TASK, LIMITED CHOICE OF TOPIC, SECOND ROUND, TASK A ��264

xiv I. DESCRIPTIVE TASK, LIMITED CHOICE OF TOPIC, SECOND ROUND, TASK B �� 265

J. DESCRIPTIVE TASK, LIMITED CHOICE OF TOPIC, SECOND ROUND, TASK C ��266

K. DESCRIPTIVE TASK, COMPLETE CHOICE OF TOPIC �� 267

L. NARRATIVE TASK, NO CHOICE OF TOPIC FIRST ROUND ��268

M. NARRATIVE TASK, NO CHOICE OF TOPIC SECOND ROUND �� 269

N. NARRATIVE TASK, LIMITED CHOICE OF TOPIC, FIRST ROUND, TASK A �� 270

O. NARRATIVE TASK, LIMITED CHOICE OF TOPIC, FIRST ROUND, TASK B �� 271

P. NARRATIVE TASK, LIMITED CHOICE OF TOPIC, FIRST ROUND, TASK C �� 272

Q. NARRATIVE TASK, LIMITED CHOICE OF TOPIC, SECOND ROUND, TASK A �� 273

R. NARRATIVE TASK, LIMITED CHOICE OF TOPIC, SECOND ROUND, TASK B �� 274

S. NARRATIVE TASK, LIMITED CHOICE OF TOPIC, SECOND ROUND, TASK C �� 275

T. NARRATIVE TASK, COMPLETE CHOICE OF TOPIC �� 276

U. TOPICS FOR DECISION-MAKING TASK, NO CHOICE OF TOPIC �� 277

V. TOPICS FOR DECISION-MAKING TASK, LIMITED CHOICE OF TOPIC, FIRST ROUND, TASK A �� 278

W. TOPICS FOR DECISION-MAKING TASK, LIMITED CHOICE OF TOPIC, FIRST ROUND, TASK B �� 279

X. TOPICS FOR DECISION-MAKING TASK, LIMITED CHOICE OF TOPIC, FIRST ROUND, TASK C ��280

Y. TOPICS FOR DECISION-MAKING TASK, LIMITED CHOICE OF TOPIC, SECOND ROUND, TASK A �� 281 xv Z. TOPICS FOR DECISION-MAKING TASK, LIMITED CHOICE OF TOPIC, SECOND ROUND, TASK B ��282

AA. TOPICS FOR DECISION-MAKING TASK, LIMITED CHOICE OF TOPIC, SECOND ROUND, TASK C ��283

BB. DECISION-MAKING TASK, COMPLETE CHOICE OF TOPIC, HOMEWORK ASSIGNMENT ��284

CC. DECISION-MAKING TASK, COMPLETE CHOICE OF TOPIC, CLASS MATERIAL ��285

DD. AFTER-TASK SURVEY ��286

EE. EXAMPLE OF THE CHOICE PAPER FOR LIMITED CHOICE SESSION, DESCRIPTIVE TASK �� 287

FF. EXAMPLE OF THE CHOICE PAPER FOR LIMITED CHOICE SESSION, NARRATIVE TASK ��288

GG. EXAMPLE OF THE CHOICE PAPER FOR LIMITED CHOICE SESSION, DECISION-MAKING TASK ��289

HH. EXAMPLE FROM THE TRANSCRIPT DATA OF MARKING FOR CORRECT VERB FORMS ��290

II. EXAMPLE FROM THE TRANSCRIPT DATA OF MARKING FOR ERROR-FREE CLAUSES �� 292

JJ. EXAMPLE FROM THE TRANSCRIPT DATA OF MARKING FOR TURNS �� 296

KK. EXAMPLE FROM THE TRANSCRIPT DATA OF MARKING FOR UNALTERED REPETITIONS ��300

LL. EXAMPLE OF A TRANSCRIPT FROM THE DESCRIPTIVE TASK, NO CHOICE OF TOPIC ��302

MM. EXAMPLE OF A TRANSCRIPT FROM THE DESCRIPTIVE TASK, LIMITED CHOICE OF TOPIC �� 303

NN. EXAMPLE OF A TRANSCRIPT FROM THE DESCRIPTIVE TASK, COMPLETE CHOICE OF TOPIC ��306

xvi OO. EXAMPLE OF A TRANSCRIPT FROM THE NARRATIVE TASK, NO CHOICE OF TOPIC ��309

PP. EXAMPLE OF A TRANSCRIPT FROM THE NARRATIVE TASK, LIMITED CHOICE OF TOPIC �� 311

QQ. EXAMPLE OF A TRANSCRIPT FROM THE NARRATIVE TASK, COMPLETE CHOICE OF TOPIC �� 314

RR. EXAMPLE OF A TRANSCRIPT FROM THE DECISION-MAKING TASK, NO CHOICE OF TOPIC �� 317

SS. EXAMPLE OF A TRANSCRIPT FROM THE DECISION-MAKING TASK, LIMITED CHOICE OF TOPIC �� 320

TT. EXAMPLE OF A TRANSCRIPT FROM THE DECISION-MAKING TASK, COMPLETE CHOICE OF TOPIC �� 323

UU. PROCEDURES FOR PREPARING DATA FOR FINAL ANALYSIS �� 327

xvii LIST OF TABLES

Table Page

1. Descriptive Statistics for the Cloze Test �� 69

2. After-task Survey Items and Their Sources �� 79

3. Task Sequence for Groups A and B �� 84

4. Descriptive Statistics for the Descriptive Task with No Choice of Topic �� 100

5. Descriptive Statistics for the Descriptive Task with Limited Choice of Topic �� 101

6. Descriptive Statistics for the Descriptive Task with Complete Choice of Topic �� 102

7. Descriptive Statistics for the Narrative Task with No Choice of Topic �� 103

8. Descriptive Statistics for the Narrative Task with Limited Choice of Topic �� 104

9. Descriptive Statistics for the Narrative Task with Complete Choice of Topic �� 105

10. Descriptive Statistics for the Decision-Making Task with No Choice of Topic �� 106

11. Descriptive Statistics for the Decision-Making Task with Limited Choice of Topic �� 107

12. Descriptive Statistics for the Decision-Making Task with Complete Choice of Topic �� 108

13. Multivariate Outliers Removed From the Analysis �� 110

14. The Constitution of the Dependent Variables �� 111

15. Factor Loadings for the Descriptive Task with No Choice of Topic �� 112

xviii 16. Factor Loadings for the Descriptive Task with Limited Choice of Topic �� 113

17. Factor Loadings for the Descriptive Task with Complete Choice of Topic �� 114

18. Factor Loadings for the Narrative Task with No Choice of Topic �� 115

19. Factor Loadings for the Narrative Task with Limited Choice of Topic �� 116

20. Factor Loadings for the Narrative Task with Complete Choice of Topic �� 117

21. Factor Loadings for the Decision-Making Task with No Choice of Topic �� 118

22. Factor Loadings for the Decision-Making Task with Limited Choice of Topic �� 119

23. Factor Loadings for the Decision-Making Task with Complete Choice of Topic �� 120

24. Descriptive Statistics for Task Interest for the No Choice of Topic �� 125

25. Descriptive Statistics for Task Interest for the Limited Choice of Topic �� 126

26. Descriptive Statistics for Task Interest for the Complete Choice of Topic �� 127

27. Repeated-Measures ANOVA Results for Task Interest �� 128

28. Descriptive Statistics for Task Interest for the Descriptive Task �� 131

29. Descriptive Statistics for Task Interest for the Narrative Task �� 132

30. Descriptive Statistics for Task Interest for the Decision-making Task �� 133

xix 31. Descriptive Statistics for Task Self-efficacy for the No Choice of Topic �� 136

32. Descriptive Statistics for Task Self-efficacy for the Limited Choice of Topic �� 137

33. Descriptive Statistics for Task Self-efficacy for the Complete Choice of Topic �� 138

34. Repeated-Measures ANOVA for Task Self-efficacy �� 139

35. Descriptive Statistics for Task Self-efficacy for the Descriptive Task �� 142

36. Descriptive Statistics for Task Self-efficacy for the Narrative Task �� 143

37. Descriptive Statistics for Task Self-efficacy for the Decision-Making Task �� 144

38. Descriptive Statistics for the Descriptive Task with No Choice of Topic �� 148

39. Descriptive Statistics for the Descriptive Task with Limited Choice of Topic �� 149

40. Descriptive Statistics for the Descriptive Task with Complete Choice of Topic �� 150

41. Descriptive Statistics for the Narrative Task with No Choice of Topic �� 150

42. Descriptive Statistics for the Narrative Task with Limited Choice of Topic �� 151

43. Descriptive Statistics for the Narrative Task with Complete Choice of Topic �� 152

44. Descriptive Statistics for the Decision-Making Task with No Choice of Topic �� 153

45. Descriptive Statistics for the Decision-Making Task with Limited Choice of Topic �� 153

xx 46. Descriptive Statistics for the Decision-Making Task with Complete Choice of Topic �� 154

47. Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the No Choice of Topic Treatment �� 157

48. Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the Limited Choice of Topic Treatment �� 158

49. Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the Complete Choice of Topic Treatment �� 159

50. Repeated-Measures ANOVA for Accuracy (Ratio of Error-Free Clauses) �� 160

51. Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the Descriptive Task Treatment �� 163

52. Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the Narrative Task Treatment �� 164

53. Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the Decision-making Task Treatment �� 165

54. Descriptive Statistics for Complexity (Type-token Ratio) for the No Choice of Topic Treatment �� 168

55. Descriptive Statistics for Complexity (Type-token Ratio) for the Limited Choice of Topic Treatment �� 169

56. Descriptive Statistics for Complexity (Type-token Ratio) for the Complete Choice of Topic Treatment �� 170

57. Repeated-Measures ANOVA for Complexity (Type-Token Ratio) �� 171

58. Descriptive Statistics for Complexity (Type-token Ratio) for the Descriptive Task Treatment �� 174

59. Descriptive Statistics for Complexity (Type-token Ratio) for the Narrative Task Treatment �� 175

60. Descriptive Statistics for Complexity (Type-token Ratio) for the Decision-making Task Treatment �� 176

xxi 61. Descriptive Statistics for Fluency (Word Count) for the No Choice of Topic Treatment �� 180

62. Descriptive Statistics for Fluency (Word Count) for the Limited Choice of Topic Treatment �� 181

63. Descriptive Statistics for Fluency (Word Count) for the Complete Choice of Topic Treatment �� 182

64. Repeated-Measures ANOVA for Fluency (Word Count) �� 183

65. Descriptive Statistics for Fluency (Word Count) for the Descriptive Task Treatment �� 185

66. Descriptive Statistics for Fluency (Word Count) for the Narrative Task Treatment �� 186

67. Descriptive Statistics for Fluency (Word Count) for the Decision-making Task Treatment �� 187

xxii LIST OF FIGURES

Figure Page

1. The Self-determination continuum (from Ryan & Deci, 2000, p. 72) �� 38

2. Littlewood’s model for developing autonomy in foreign language learning (from Littlewood, 1996, p. 432) �� 53

3. van Lier’s model of the growth of proficiency (from van Lier, 1996, p. 41) �� 55

4. Early-stage model of motivational development (from Nakata, 2004, p. 160) �� 56

5. Graphical representation of the data collection procedures �� 81

6. Category probabilities from Rasch analysis �� 122

7. The degree of Task Interest for each level of choice �� 129

8. The degree of Task Interest for each task-type �� 134

9. The degree of Task Self-efficacy for each level of choice �� 139

10. The degree of Task Self-efficacy for each task-type �� 145

11. Profile plot of Accuracy (ratio of error-free clauses) by level of choice �� 161

12. Profile plot of Accuracy (ratio of error-free clauses) by task type �� 166

13. Profile plot of Complexity (type-token ratio) by level of choice �� 172

14. Profile plot of Complexity (type-token ratio) by task type �� 177

15. Profile plot of Fluency (word count) by level of choice �� 184

16. Profile plot of Fluency (word count) by task type �� 188

xxiii CHAPTER 1

INTRODUCTION

Task-based Language Teaching

Task-based teaching language teaching (TBLT) is a vibrant area of second language acquisition research and is an approach to language teaching based on the ideas that language learners can learn the language better by interacting with others and by focusing on the message rather than on the form of the language (e.g., Duff, 1993; Ellis,

2003; Long, 1985; Nunan, 2004; Pica, Kanagy, & Falodun, 1993; Skehan, 1998). TBLT has much to its advantage, including a natural conversational environment, a focus on the learner and the cognitive abilities inherent in the learner, and the provision of a high degree of learner autonomy.

Task-based language teaching has been utilized since the early 1980s, evolving from activities used in communicative approaches to language teaching (Skehan, 2003b, p. 1), such as the Notional-Functional Approach (van Ek & Alexander, 1975, 1976; Wilkins,

1976), which was itself a reaction to the rigid syllabi and the contrived dialogues of the behaviorist approaches popularly used until the mid-1970s and still in use today in some educational contexts (Willis, 2004, pp. 4-5).

Richards, Schmidt, Kendricks, and Kim (2002, p. 540) wrote that task-based language teaching is a teaching approach based on the use of communicative and interactive tasks as the central units for the planning and the delivery of instruction.

1 Interactive tasks help create meaningful communication, interaction, negotiation, and authentic language use. Larsen-Freeman (2000), however, saw a larger picture in her definition of task-based language teaching:

A task-based approach aims to provide learners with a natural context for language use. As learners work to complete a task, they have abundant opportunity to interact. Such interaction is thought to facilitate language acquisition as learners have to work to understand each other and to express their own meaning. By so doing, they have to check to see if they have comprehended correctly and, at times, they have to seek clarification. By interacting with others, they get to listen to language which may be beyond their present ability, but which may be assimilated into their knowledge of the target language for use at a later time (Larsen-Freeman, 2000, p. 144).

Task-Based Language Teaching in Japan

Although task-based language teaching is becoming increasingly popular as a teaching approach, it is not widely used in Japan, in part because of the long history of the use of the methodology known as yakudoku, or Grammar-Translation (Gorsuch,

1998). In high schools and junior high schools, there is strong washback from high- stakes university entrance examinations that is revealed in the English curricula at the high school level by the prevalence of rote memorization that is believed to provide the knowledge needed to pass the university entrance examinations, but little else.

Motivation and Language Learning

Learner motivation in foreign language classrooms plays a very influential role in language acquisition. Motivation in the second or foreign language classroom has a

2 long history of research, beginning with a social-psychological perspective in the late

1950s with Gardner and Lambert (1959). Hypotheses of language learning motivation include, but are not limited to, Gardner’s (1985) integrative and instrumental orientations of language learning motivation, Dörnyei’s (2001; Dörnyei & Ottó, 1998) concept of motivation as a process, and Julkunen’s (1989, 2001) task motivation (see also Dörnyei,

2002, 2003; Dörnyei & Kormos, 2000; Kormos & Dörnyei, 2004). The one common point that these hypotheses share is that more motivated learners, in comparison to lesser motivated learners, will in most cases have higher achievement scores, will continue their studies for a longer time, and will seek opportunities outside the classroom to use the foreign language. Motivating learners seems to be a core concern of language teachers and it is for good reason that motivation has been theorized, strategized, and pedagogized extensively.

The Aims of This Study

The first aim of this study is to investigate how increased autonomy may motivate foreign language learners in a task-based classroom. The technique used to increase the students’ motivation is through the relatively unintrusive method of increased choice of task topic.

The second aim of this study is to investigate the changes that occur in the language output of learners when increased choice is introduced. Common in output research are three measures–accuracy, complexity, and fluency–which are used for

3 assessing output in TBLT. With these assessment guides, the possible changes due to choice in the output the learners produce can be gauged.

The Significance of This Study

The results from this study may help teachers improve their students’ motivation and amount of oral language output, both of which can help learners acquire the language that they encounter in the classroom more effectively. Importantly, this improvement can be attained from a relatively minor change in the curriculum. With the simple introduction of increased autonomy into the syllabus through the choice of the task topic, benefits for increased motivation and language learning that extend beyond the classroom can be gained.

The method of increasing autonomy used in this study is simple and unobtrusive.

It is simple in the sense that preparations needed for increasing autonomy take little time, while the benefits are potentially great. It is unobtrusive compared to other ways of introducing autonomy in the curriculum, such as self-access centers (e.g., Benson &

Voller, 1997), which are sometimes difficult to implement without extensive curriculum revisions, and can be culturally insensitive (e.g., Jones, 1995). With an unobtrusive, simple addition of choice of task topics, cultural barriers towards autonomy can be overcome and students’ motivation can be improved.

There are also gaps in the cross-cultural psychological research literature, the

TBLT research literature, and the pedagogical autonomy literature. From the cross-

4 cultural psychological perspective, researchers (e.g., Iyengar & Lepper, 1999), who claimed that autonomy is not motivating for Asians, used children as their participants. However, it is unclear that this would be the case for adults. One aim of this study is to examine differences in student motivation that occur while they are engaging in tasks when choice is introduced in the Asian setting.

In the task-based language teaching literature, research into task implementation that can help promote motivation is non-existent. Although Long in his seminars, Ellis

(2003), and Dörnyei (2002) have claimed that task-based language teaching in itself may be motivating, little empirical data is available to support these claims. Also, another gap in the TBLT research literature is the role of motivation itself in task-based language teaching. A potential benefit of this study is that it will provide information on this role.

Pemberton, Li, Or, and Pierson (1996), for example, concern the operationalization of self-access centers and self-study centers or program-wide intervention. Although some articles in Barfield and Nix (2003) provide advice for introducing autonomy in the classroom, what is lacking is a method of introducing autonomy in the foreign language classroom that does not require a great deal of curriculum revision and is more under the control of the teacher.

5 Benson (2006) has also written of a lack of empiricism in the current discourse on autonomy in the language teaching profession. The theoretical base is well developed according to Benson, but there is a lack of an “understanding of the ways in which autonomy and the potential for autonomy vary” (p. 34) in diverse settings and situations.

An aim of this study will be to provide this needed support for autonomy in the classroom.

The Delimitations of This Study

Because this study took place in a particular environment, some delimitations need to be mentioned. First, this sample is made up of Japanese young adults. It would be difficult to generalize the results to populations in other cultural or linguistic areas.

A second delimitation is that the oral English language proficiency level of the sample is low. It is possible that different results would be found with students at a higher proficiency level.

A third delimitation concerns the tasks and topics used in this study. Because of the lack of research in this area, it is unclear how the use of different tasks and/or topics would influence the results of this study.

The Audience for This Study

Three groups of professionals make up the audience for this study: researchers interested in motivation and task-based language learning, practicing teachers, and materials designers. For the first group, the use of autonomy as a way to increase

6 motivation adds information to the construct of motivation to speak a foreign language.

Many studies of motivation are snapshots of a specific point in time, with discussions of possible reasons for low or high levels of motivation among diverse students. In contrast, this study provides information about increasing the task interest of the same student performing multiple tasks. For researchers interested in task-based language teaching, this study provides information about a new way to implement tasks. Through this implementation, there may be increased interaction during the task with improved output.

Practicing teachers as well will find an interest in this study, as specific ways to increase students’ interest in pair activities is one focus of this study. Foreign language teachers often lament that capturing students’ interest is often difficult. Many teachers teach classes that are required for university completion, and, as such, the students are not as motivated to engage in tasks as, for example, students who attend elective classes.

The techniques elucidated in this study may help teachers increase students’ interest to engage in speaking tasks.

Materials developers and textbook designers are a third audience for this study.

Incorporating choice may help increase students’ willingness to engage in tasks and provide for more interesting materials that students would engage with more attentively.

Including a choice of task topics in a textbook may introduce a new dimension to materials design.

7 The Outline of This Study

Chapter 2 will begin with a description of the definitions of a language learning task followed by sections on task features, the influence of a task topic, the discourse mode of a task, task difficulty, Skehan’s model of accuracy, complexity, and fluency, and Robinson’s model of task complexity, difficulty, and conditions. Next is a discussion of the Self-

Determination Theory (SDT) of motivation and intrinsic motivation, in which choice plays a large role. This is followed by a section on how autonomy is conceptualized in the foreign language teaching profession and cultural differences found in that body of research.

Chapter 3 begins with a description of the participants and the setting where this study took place. Following this is a section describing the variables in this study.

The next section is a description of the task materials and the survey instrument. The procedures used during the data collection sessions follows. Following this, the design of the study and the order of the task implementation are described. The procedures used for calculating the production data end this chapter.

Chapter 4 contains the preliminary results for Study 1. This chapter includes checking for univariate and multivariate outliers, data transformations, descriptive statistics, and factor analyses. Following the factor analyses is a review of Rasch analysis, which was used to examine the data in regards to its fit to the Rasch model and how well the response categories worked with the surveys that were administered. This chapter concludes with a review of how person measures for each treatment were garnered using

Rasch analysis and how the data file was prepared for final data analysis.

8 Chapter 5 is a report of the results of Study 1 using two-way repeated-measures

ANOVAs. First, the results for the Task Interest dependent variable are presented and

Research Questions 1 and 2 are reviewed and answered. Following this, the results for the

Task Self-efficacy dependent variable are presented and Research Questions 3 and 4 are reviewed and answered.

Chapter 6 is a report of the results of Study 2. Six research questions are answered using two-way repeated-measures ANOVAs. The analysis concerns the accuracy, complexity, and fluency of the output that the students produced while engaged in the tasks.

Chapter 7 begins with a review of the findings as related to the research questions and the hypotheses. Following a review of each research question, interpretations of the underlying reasons for these results are presented.

Chapter 8 is a conclusion to this study. This chapter begins with an integration of task-type and choice according to the results. Following this are the limitations of this study, continuing with suggestions for further research. The implications of the findings of this study follow. A concluding section completes this study.

9 CHAPTER 2

LITERATURE REVIEW

Tasks in Language Teaching

The word task has been a part of English since Norman times, coming from the

Old French. In modern English, the word task may at times have a somewhat negative connotation, as in doing something as if it is drudgery (Oxford, 2006; Simpson & Weiner,

1989). In language teaching, task can have diverse meanings, but it usually refers to an activity engaged in by learners. In recent years, however, task has come to have a specific meaning in the second language acquisition research literature: a task is an activity that will promote language learning under conditions of interaction, attention, and negotiation of meaning. Larsen-Freeman (2000, p. 146) noted the difference between tasks in the Communicative Language Teaching method and tasks used in Task-Based

Language Teaching. In Communicative Language Teaching, a task is used to engage learners in practicing a communicative function. On the other hand, in TBLT, tasks are activities that get the learners to focus on the completion of the activity.

Definitions of a Task in Task-Based Language Teaching

A task in the TBLT context has been defined from various viewpoints and frames of reference. In the short twenty years since the first definition, so many definitions have appeared that some writers have classified the definitions. For example, Kumaravadivelu

10 (1993, p. 70) and Lee (2000, p. 31) have claimed that task definitions can be conceptualized on a continuum; from definitions relating the task to real-world contexts, to tasks in the general education context, and then to tasks in the language teaching context.

Ellis (2003, p. 2), in his review of task definitions, listed six components of a task definition: (a) the scope of a task (general or specific contexts), (b) the perspective from which a task is viewed (the task designer’s or the task participant’s), (c) the authenticity of a task, (d) the linguistic skills required to perform a task, (e) the cognitive processes involved in task performance, and (f) the outcome of the task.

Ur (1981) wrote an early definition of a task that is similar to more recent definitions. To Ur (1981, pp. 13-14), a task is an activity that, among other things, requires thought (i.e., engages cognitive processes), has an outcome, entails interaction, and piques the learner’s interest.

Prabhu (1987) emphasized outcome in his definition when he described a task as:

An activity which required learners to arrive at an outcome from given information through some process of thought, which allowed teachers to control and regulate that process (p. 24).

This definition also emphasized the cognitive processes of the student, a theme that would become more prominent as definitions evolved.

A definition proposed by Breen (1987) focused on the cognitive processes involved in completing the task. As Breen wrote:

The notion of ‘task’ is used in a broad sense to refer to any structural language learning endeavor which has a particular objective, appropriate content, a specified working procedure and a range of outcomes for those who undertake the task. ‘Task’ is therefore assumed to refer to a range

11 of work-plans which have the overall purpose of facilitating language learning-from the simple and brief exercise type to more complex and lengthy activities such as group problem-solving and or simulations and decision-making. Within this broad spectrum, a language test can justifiably be seen as a type of task (p. 23).

Candlin (1987) provided a more narrow definition of a language learning task in the context of task-based language teaching. For Candlin, a language-learning task is:

One set of differentiated, sequencable, problem-posing activities involving learners and teachers in some joint selection from a range of varied cognitive and communicative procedures applied to existing and new knowledge in the collective exploration and pursuance of foreseen or emergent goals within a social milieu (p. 10).

Although it seems at first reading that this is also a broad definition, the atmosphere of the language classroom in which task-based language teaching is implemented is also considered in this definition. In addition, Candlin used the terms “cognitive” and

“communicative” to again place the task in the task-based language teaching classroom and to differentiate it from what could be accomplished in classrooms in which audio- lingual or the grammar-translation methodologies are implemented.

Skehan (1998) defined tasks from a cognitive perspective. To Skehan, a task is (a) an activity in which meaning is primary, (b) there is some communication problem to solve, (c) there is some sort of relationship to comparable real-world activities, (d) task completion has some priority, and (e) the assessment of the task is in terms of outcome

(p. 95). Skehan based this definition on what he believes to be three important criteria for language learning: (a) noticing should occur, (b) learners should analyze the linguistic units that they are using, and (c) the learners should synthesize the language so that it will become a part of their knowledge in a way like that of a first language (Skehan, 1998,

12 p. 91). In the words of Skehan, “the learner needs to be prepared to focus on structure and identify patterns . . . [so that] analyses [will be] reintegrated and synthesized into fluent performance” (p. 92). Skehan also stated that examples of classroom activities that are not tasks are transformation exercises, most question and answer activities with the teacher, and activities where the materials are conducive to the generation of grammatical rules.

Takashima (2000) developed a notion of a task activity that he claims is more suited to the classroom environment in Japan. The task activity incorporates more of the structure focus that Loschky and Bley-Vroman (1993) advocated. For Takashima, a task should: (a) be message-focused, (b) allow learners to have a sense of completion,

(c) invite negotiation of meaning, (d) involve a comparison of structures, (e) include an information gap element, and (f) be of interest to the learners (p. 36). Other than comparison of structures, most of the above features are common in the other definitions of a task. Comparison of structures is manipulated through the design of the task so that the students have to choose a particular form (Takashima, 2000, p. 37) and thereby notice that there are differences between their existing knowledge and the new knowledge

(p. 38). For example, Yamada (1999) developed tasks in which learners must choose between one verb form (the past or present verb forms) over another (the past or present progressive verb forms) in order to complete the task correctly.

Although there are many more definitions of a task in the language learning literature (e.g., Bygate, Skehan, & Swain, 2001, pp. 11-12; Courtney, 2001, p. 9; Crookes,

1986, p. 1; Krahnke, 1987, p. 57; Lee, 2000, p. 32; Long, 1985, p. 89; Nunan, 2004, p. 4;

13 Richards et al., 2002, pp. 539-540), the above definitions were selected because they include a recognition of the cognitive aspect of language learning and the importance of the output that results from doing the task.

Task Features

Features of tasks are important, especially when designing and implementing tasks for the classroom (Ellis, 2003, pp. 86-95). Tasks have generally been categorized as:

(a) open or closed, (b) convergent or divergent, (c) one-way or two-way, and (d) involving a required or an optional exchange of information.

Open and Closed Tasks

Whether the task is open or closed concerns the final solution of the task activity

(Ellis, 2003, p. 89; Long, 1989, p. 18). In open tasks, there does not have to be a final agreement or a solution. An example of a open task is an opinion-giving task in which the learners express their opinions on a certain topic. In closed tasks, a final agreement or correct answer is required. An example of a closed task is a jigsaw or information gap type of activity where there is a set and correct answer. Long (1989, p. 16) stated that closed tasks result in more negotiation than open tasks.

14 Convergent and Divergent Tasks

The second feature of a task is whether it is convergent or divergent (Duff, 1986).

For these two features, learners mutually acknowledge and incorporate each other’s output to produce coherent responses when engaging in convergent and divergent tasks (p. 151).

In convergent tasks, learners are required to arrive at a mutually acceptable solution. For example, in a problem-solving task, such as the desert island problem, participants must agree on which items would be most useful if they were stranded on a desert island. In divergent tasks, learners can arrive at different solutions to the same problem. An example is a task in which learners have independent goals, such as a debate about the pros and cons of television. Ellis (2003, p. 90) suggested that these two types of tasks should be considered a subset of the open type of task because they both allow multiple solutions to the same task.

One-Way and Two-Way Tasks

Tasks can also be classified as one-way or two-way tasks. In a one-way task, one person of a pair holds the complete information and the other person holds jumbled or partial information that must be arranged or completed based on information communicated by the partner. An example is a narrative task where one member of the dyad holds the story in proper order and the other member holds the story in a jumbled order. In a two-way task, each member holds only part of a complete whole and negotiation among the learners must occur if they are to make a complete and correct

15 answer (Ellis, 2003, p. 88). An example is a narrative task where each member of the pair holds separate scenes to a story and together they recreate the story in its entirety. Long

(1989, p. 13) speculated that two-way tasks would require more negotiation to complete when compared to one-way tasks.

Required or Optional Information Exchange Tasks

The next characteristic of tasks concerns whether the exchange of information is required or optional. An example of a required information exchange is a task where each partner must exchange a predetermined set of information, such as a spot the difference task, so that each partner has equal knowledge of the problem at the completion of the task. On the other hand, Ellis (2003, p. 86) stated that opinion-giving and decision- making tasks involve the optional exchange of information because, although each partner contributes to the interaction, what each person says while engaging in the task is at his or her discretion.

Task Topic Influences

Ellis (2003) listed topic familiarity as a factor influencing the completion of a task. Gass and Varonis (1984) and Zuengler and Bent (1991) found that topic familiarity could significantly improve interactions in terms of conversational participation and the facilitation of the comprehension of the interlocutor’s speech. In addition, Zuengler and

Bent found that advanced-level non-native speakers of English held their own quite well

16 in interacting with a native speaker of English, whether the topic was known best by the

native speaker, was known best by the non-native speaker, or whether they were talking

about a topic that they supposedly knew equally well. Plough and Gass (1993) studied

discourse between non-native speakers of English and found that task-nonfamiliar pairs

displayed a greater involvement in the task and that those who were familiar with the

task became somewhat disinterested (p. 50).

Discourse Mode

Different discourse modes and the differences between them have been investigated by Skehan and Foster (Foster & Skehan, 1996, 1999; Skehan & Foster,

1997, 1999, 2001, 2005). These researchers have found that, depending on the task type, participants’ oral output can differ in terms of the accuracy, complexity, and fluency of the language that is used during the task.

The first type of task is the static task. This was defined by Brown and Yule

(1983, p. 109) as a task in which learners describe static relationships and by doing so are describing object properties, the location of objects, and the relationship between the objects. A common version of this type of task uses a picture that has to be described by one of the learners to complete missing information.

The second type of task is known as a narrative task. According to Brown and

Yule (1983, p. 109), this type of task involves dynamic relationships and learners must be able to tell a coherent story using language indicating locations, activities, and states,

17 as well as descriptions so that the hearer can understand and re-create the story. What

is crucial in story-telling tasks is that speakers make it clear whom and what they are

referring to at any point in the story (p. 132). A common version of this type of task,

which was used by Foster and Skehan (1996) and Skehan and Foster (1997, 1999), involves

a learner in telling a picture story to another person who has each frame of the story in

jumbled order and who must place the frames in the correct order.

The third type of task is the decision-making task. According to Brown and Yule

(1983, p. 109), this type of task concerns abstract relationships. Foster and Skehan (1996, p.

307) suggested that speakers have to consider new information, evaluate the information,

and defend their opinions when engaged in this type of task. An example abstract task

is one in which the learners are given a problem and try to find a solution to it. This task

differs from the first two task types in that while there is no correct answer, there has to

be a solution to the problem. A second difference is that both learners in the pair have the

same information, as opposed to the description task and the narrative task where the

learners each have different information.

Task Difficulty

Norris, Brown, Hudson, and Yoshioka (1998) followed Skehan (1996) in defining task difficulty as an interaction of cognitive code, cognitive complexity, and communicative demand (p. 50). Based on this definition, Norris, et al. developed a grid of these three variables. For any one or combination of these, the more attributes there

18 are, the higher the task difficulty. Under code, which is the kind of language involved in successful task performance, there are the subcategories of (a) range, which represents the spread of the language needed to complete the task and (b) the number of input sources.

Cognitive complexity concerns the amount and kind of information processing that a learner engages in to successfully perform the task (pp. 79-80). Under this category, there are two subcategories: (a) the organization of the input and the output, which is the extent to which the learner must organize the information to be successful in the task and (b) the availability of the input, which concerns how much work is needed to search for information to complete the task (p. 80).

The third category is communicative demand, which is the type of communication activity required. The two subcategories for communicative demand are

(a) mode, which is whether the task requires writing or speaking and (b) response level, which is whether the learner is required to handle the input on a real-time basis or not.

Robinson’s Model of Task Complexity, Task Difficulty, and Task Conditions

Robinson (2001a) hypothesized that three influences cause more cognitive resources to be needed to complete a task: (a) task complexity, which are aspects of task design and are manipulable; (b) task difficulty, which is made up of factors related to the learner’s motivation and proficiency; and (c) task conditions, which are the interactive demands of the task. While task complexity can explain the differences in a task, such as one task being interactive or requiring different responses, task difficulty can help explain

19 variation between any two learners’ performance of the same task (p. 295). Task conditions include the features of a task, which are factors such as whether the task is open or closed or whether the task is a one-way or a two-way task, and participant variables, such as grouping, partner familiarity, and the gender of the participants (p. 295).

Robinson speculated that task complexity is operationalized in two dimensions.

The first dimension is resource-directing. Here the resources of the learner are directed to a wider range of linguistic use, when compared to tasks that are more resource-depleting, and aspects of the task encourage learners to use their resources to complete the task. A less complex task would be in the here-and-now (there-and-then leads to more complexity), have fewer elements, and would have no reasoning demands. There will be, according to Robinson, a greater amount of negotiation in the case of more complex tasks (p. 308).

On the negative side, Robinson stated that complex tasks are less likely to be successfully completed, take longer than simpler counterparts, are rated by learners as more difficult, have psychological consequences, and are more susceptible to interference from competing tasks (p. 306). On the other hand, Ellis (2003) suggested that more complex tasks will lead to more sustained interaction, more attempts to repair communication, more pushed output, and a greater use of communication strategies (p. 95).

The second dimension is what Robinson (2001a) called resource-depleting, which concerns aspects of the task that make extra demands on resources that cannot be met through the code of the language (p. 295) and therefore attentional and memory resources are depleted and external assistance is needed (p. 308). In this dimension

20 are the presence or absence of (a) planning time, (b) prior knowledge (not just task familiarity or topic familiarity but also familiarity with parts of the task), and (c) whether the task makes a single demand or multiple demands on the learner.

After task complexity, Robinson (2001a) hypothesized about factors that make tasks more difficult. Task difficulty, which can explain the performance differences between two learners, is the learner’s perceptions of the demands of the task. This includes affective variables such as motivation, anxiety, and confidence, and ability variables such as aptitude, proficiency, and intelligence.

The task conditions in this model include participation variables (e.g., one-way vs. two-way task and a convergent vs. divergent task) as well as participant variables (e.g., the participants’ gender, and power/solidarity relations with his partner). Ellis (2003) in his review of Robinson’s model, eliminated this last category and used a category termed task procedure instead, which concerns implementing a task in the classroom and includes such factors as pre-teaching and planning time.

Comprehensible Output and Task Production

Swain (1985, 1995) proposed an output hypothesis as a response to Krashen’s claim that output only makes an indirect contribution to language acquisition (Krashen,

1982, p. 60). Swain based her output hypothesis on findings comparing native speakers of English studying in a French immersion program with native speakers of French and found that although the immersion students had developed good receptive skills in the

21 foreign language, their productive, lexical, and grammatical performances were not equivalent to that of native speakers despite seven years of intensive input in the target language. Swain hypothesized that part of the reason for this was that students could not practice speaking in communicative exchanges that required a precise and appropriate reflection upon meaning (Swain, 1985, p. 251), even though they used the second language almost all of the time in school.

Swain (1985) wrote that output can enhance language acquisition in two important ways: (a) learners can have meaningful opportunities to use the language and therefore get more practice using the language and (b) output can be “pushed” from the learners in situations where the message needs to be conveyed precisely, coherently, and appropriately. Other roles for output included helping learners test hypotheses, move from semantic processing to syntactic processing, and increase attentional focus because of an expected possible future use of language.

A decade later, Swain (1995) expanded the output hypothesis and focused more on the role of noticing and attention, having followed influential articles by Schmidt (1990) and Schmidt and Frota (1986) on noticing. Swain stated that there were three functions of output. The first of these is noticing initiated by output, which provides a trigger for cognitive processes that can generate new linguistic knowledge in learners. A second function of output is that it can provide learners with an opportunity to test hypotheses about the language as they try out new language forms and structures that stretch their interlanguage. A third function of output is that it can enhance metalinguistic knowledge

22 where learners reflect on the viability of their hypotheses about the target language.

Through the use of collaborative tasks that can enhance collective scaffolding, which in itself can lead learners to utilize strategies for useful language learning, there is the potential to bring a metalinguistic function to output (Swain, 2000). The implication of

Swain’s hypothesis for task-based teaching is that output can be stimulated from the use of collaborative tasks.

Skehan (1998, pp. 16-19) also discussed six roles of output relevant to language learning based upon Swain’s output hypothesis: (a) generating more finely tuned input,

(b) making learners more aware of syntax, (c) allowing learners to test hypotheses, (d) developing automaticity, (e) aiding learners in developing discourse skills, and (f) helping learners develop a personal voice, where they find ways of expressing personal meaning and develop a personal manner of speaking. Output is something that students can see and hear as helping them to learn the language.

Assessing Output

For researchers, output has been important for assessing learner skill and improvement as well as for determining the effects of different tasks in different situations on the language produced. An approach to assessing output that has been especially influential in research into task-based language teaching was developed by

Skehan, who examined learners’ output by assessing accuracy, complexity, and fluency in different situations and with different types of tasks.

23 Based on the work of Swain, Skehan (1998, p. 5) speculated that there are three aspects to oral production: accuracy, complexity, and fluency. This proposal has been developed through the work of Skehan and Foster (e.g., Foster & Skehan, 1996, 1999;

Skehan, 1998, 2001; Skehan & Foster, 1997, 1999, 2001, 2005). A lynchpin in Skehan’s conceptualization of spoken language production concerns how well learners attend to one of these aspects over the others under certain conditions. Attentional resources shift, emphasizing one area and de-emphasizing others, in order to better handle the considerable cognitive load required by producing output (p. 73). Skehan (1995, p. 102) categorized these demands for attention, all of which concern the need to keep up with real-time communication, as follows: (a) cognitive demands, which concern the complexity of the message to be conveyed; (b) linguistic demands, which are relevant to the complexity of the language for effective communication in a certain setting; (c) linguistic criteria, which are attempts of the user to strive for greater accuracy; and

(d) the need to keep up with on-going communication, which involves time, pressure, and unpredictability. According to Skehan, these four influences combine to make communication problems more difficult and pose problems for the limited capacity information processor.

Skehan based his proposal that language output is made up of accuracy, complexity, and fluency on the hypothesized existence of rule-based and exemplar-based systems. L2 learners move between these two systems naturally to meet task demands

(Ellis & Barkhuizen, 2005, p. 142; Skehan, 1998, p. 54). The rule-based linguistic system

24 is a “knowledge of abstract rules that can be used to compute an infinite variety of well- formed utterances/sentences . . . [that] . . . allows complex propositions to be expressed clearly, concisely, and in novel and creative ways” (Ellis & Barkhuizen, 2005, p. 142). A disadvantage of the rule-based system is that it is quite costly in terms of processing effort. In contrast, the exemplar-based system is based on a very large and redundantly structured memory system. This system operates on memorized chunks of language, such as “I told you so,” rather than individual items. These chunks allow learners to conserve processing resources and to formulate speech acts when little time is available for planning. According to Ellis (2001), learners build more and more of these chunks as they use the foreign language. If chunks relevant to the task are available, they help learners to complete a task more quickly than if none relevant to the task are available.

Based on this conceptualization of these rule-based and exemplar-based systems,

Skehan (1998, 2003b) developed a method of assessing output during task performance through the dimensions of accuracy, complexity, and fluency. First, an initial distinction between meaning, which is termed fluency, and form reflects the tension between getting the task done (fluency) and focusing on language development (form). Form is further separated into two entities, control, which is the accuracy of the utterances, and restructuring, which is the complexity of the production that arises from learners’ willingness to take risks.

Skehan (2003a, p. 397) also approached assessing output from a different perspective that contrasts not fluency and form, but change and control. Change is seen

25 as complexifying, which is extending new forms and integrating them into the existing

interlanguage. Control is separated into two entities, form and access. Form is accuracy

and the process of the new form becoming part of the language learner’s repertoire.

Access concerns fluency and the proceduralization and lexicalization of the new language.

From whatever outlook, this troika of accuracy, complexity, and fluency has been

extensively used to measure output in recent TBLT research. The definitions and the information about accuracy, complexity, and fluency in the following three paragraphs, unless otherwise noted, are amalgamated from Ellis (2003), Ellis and Barkhuizen (2005),

Skehan (1996, 2003b), and Skehan and Foster (1999).

Accuracy is performance that is native-like through its rule-governed nature and is connected with a learner’s capacity to handle the language capabilities at whatever level of interlanguage complexity the learner has acquired at the time. Accuracy is also related to the learner’s norms in regards to beliefs about the necessity of accuracy.

Accuracy is a relatively conservative communication strategy in the sense that there is a tendency by the learner to avoid a form unless the learner is sure that he or she has a command of the form. Accuracy is desirable because inaccurate language forms can fossilize, stigmatize and demoralize learners, and impair communicative effectiveness.

Task characteristics that enhance greater accuracy are tasks that are structurally-based, have familiar information, and are more interactive. Task design features that promote the enhancement of accuracy are contextual support, open tasks, and a clear inherent

26 structure. Common ways of calculating accuracy are counting the number of errors in the learner’s output and the target-like use of the language.

Increasing complexity indicates change and development in the interlanguage system and is based on the ability of learners to take risks, use more syntactically complex language, and use more language subsystems with the possibility that such language may not be controlled effectively. Complexity is desirable because it enables a greater degree of acceptance by native speakers. Task characteristics that promote greater complexity are tasks with outcomes that require justifications, interactive tasks, and tasks that have relatively complex outcomes. Task design features that enhance complexity are tasks that have no contextual support, have many elements, involve shared information, pose a single demand, are open with divergent goals, and are narrative tasks. Methods of measuring complexity are interactional (e.g., turns), propositional (e.g., idea units), functional (e.g., frequency of a specific language function), grammatical (e.g., subordination), and lexical (e.g., type-token ratio).

Fluency is the ability to use linguistic resources to the best of one’s ability while communication is taking place and to produce speech at a normal rate of speaking.

Fluent discourse is characterized by an optimal mix of highly automatized chunks of language and learner creativity (Lennon, 2000, p. 32). Fluency is effective when there is an automatization of stored chunks of speech that were restructured on previous occasions.

Fluency is desirable because the results of an emerging and developing restructuring of the interlanguage are evident in speech. Poor fluency can lead to more dissatisfaction

27 with the use of the language and therefore fewer opportunities for interaction. A task characteristic that enhances fluency is familiar information. Task design features that enhance fluency are the provision of contextual support, the presence of few elements, a single demand, closed tasks, and a clear, inherent structure. Fluency in task performance is calculated using temporal variables (e.g., amount of speech and pausing) and hesitation phenomena (e.g., false starts, repetitions and reformulations and replacements).

Research on Accuracy, Complexity, and Fluency

Accuracy, complexity, and fluency of task performance can vary when learners engage in different types of tasks. For example, Foster and Skehan (1996) used personal information exchange, narrative, and decision-making tasks with 16 pairs of participants with English as the medium. Foster and Skehan recorded the interactions produced by the participants and calculated the participants’ accuracy, complexity, and fluency for each task. They hypothesized that the personal information exchange task (i.e., tell your partner how to get to your house and then to turn off the gas) would be the easiest task to do, the decision-making task (i.e., decide the sentences for a list of offenders at a trial) the most difficult, and the narrative task (i.e., construct a storyline from a set of loosely related pictures) somewhere in-between. However, this hypothesis was only partially supported by the results. The researchers found that the personal information exchange task generated the highest degree of accuracy but little complexity. The narrative task engendered the greatest amount of complexity, but little accuracy. The decision-making

28 task did not gain the highest scores for any of the three task performance categories but occupied a level somewhere in-between the other two types of tasks. Often, a trade-off between accuracy and complexity was found depending on the difficulty of a task (Skehan & Foster, 2001). Skehan terms this the tradeoff hypothesis, as attention is allocated to different performance areas under different conditions (Skehan, 2007).

However, according to Foster and Skehan (1996), the personal task produced much more fluent discourse than the narrative task and the decision-making task (p. 317).

Relating his concept of task complexity with the three task performance characteristics of accuracy, complexity, and fluency, Robinson (2001a) speculated that simple monologic tasks (i.e., one-way, open-ended tasks focused on fluency) would promote more fluent but less complex and accurate speech, and complex monologic tasks (i.e., one-way tasks focused on accuracy and complexity) would promote less fluent but more accurate and complex speech. For simple interactive tasks (i.e., two-way tasks focused on fluency), Robinson predicted that learners would produce more fluent but less accurate speech. For complex interactive tasks (i.e., two-way tasks focused on accuracy and complexity), Robinson predicted less fluent but more accurate speech.

For the monologic tasks, Robinson (1995) operationalized a simple task as one in which learners tell a story that is placed in the here-and-now, that is, the students looked at the picture story while they were telling it. The complex task was operationalized as the same kind of picture story, but the students had to turn over the paper while they were telling the story. Therefore, the story was told in the there-and-then context. For

29 this study, Robinson recorded the conversations of 12 intermediate-level students from

Japanese, Korean, Indonesian, and Tagalog L1 backgrounds. The results indicated that the participants produced more accurate and complex speech with the there-and-then task than with the here-and-now task. However, they produced more fluent speech with the here-and-now task.

For the interactive tasks, Robinson (2001b), operationalized a simple task as one in which the information (in a map task) was likely to be known by the participants.

The complex task was operationalized as a map task of an area that was likely to be unknown to the participants. For this study, Robinson recorded the conversations of

44 Japanese university undergraduate participants arranged in pairs. Robinson found that the participants produced more complex utterances when describing the map of the unknown area and they were more fluent when describing the map of a known area.

In general, it seems that the less difficult a task is, the more fluent the performance will be. Skehan and Foster’s personal task and Robinson’s here-and- now and familiar information tasks were the least difficult tasks and both engendered more fluent performance from the learners. In addition, more difficult tasks appear to increase the complexity of learners’ utterances. Skehan and Foster’s narrative task and

Robinson’s there-and-then and unfamiliar information tasks all induced more complex oral production from the learners. Lastly, tasks that promoted learner accuracy were less difficult, as in Skehan and Foster’s personal task, or more difficult, as in Robinson’s there-

30 and-then and unfamiliar tasks and Skehan and Foster’s decision-making task. For this reason, accuracy seems to be more dynamic and unpredictable.

Working Definition of a Task

In this study, a task shall be a communication activity conducted in dyads in which each member of the pair holds information, either mutually equal as in a decision- making task or mutually different as in the descriptive and narrative tasks. The goal of the task is for the two people to match their complementary information to complete the requirements of the task.

Choice

The power of choice to motivate has been shown to exist in several studies (e.g.,

Corah & Boffa, 1970; Geer, Davison & Gatchel, 1970; Geer & Maisel, 1972; Glass, Singer &

Freidman, 1969; Langer and Rodin, 1976; Pervin, 1963; Reim, Glass & Singer, 1971). Even the illusion that there is a choice, such as when gambling (Langer, 1975), has been shown to be a powerful motivator. It has also been shown that as long as people believe that they have chosen to do an activity, they will engage in one that is quite possibly an anathema to them. Zimbardo, Weisenberg, Firestone, and Levy (1965), for example, showed that participants could change their attitudes positively towards eating fried grasshoppers when they believed that they had chosen to do so on their own.

31 Choice was also seen as motivationally beneficial when participants could choose the tasks in some way. In an early study, Stotland and Blumenthal (1964), found that when participants could choose the order in which they took short subsets of a test, they had less anxiety than those who could not choose the order of the sub-tests, even though they all took the same test. Zuckerman, Porac, Lathin, Smith, and Deci (1978) studied university students who were given a choice of a puzzle form to complete and students who were not given a choice. The result was that the students who could choose the puzzle form spent more time completing the puzzle, an indicator of higher intrinsic motivation. Zuckerman et al. (1978) stated, “people’s motivation is greater when they have more rather than less control over their environment” (p. 445).

Culture and Choice

It does not take a lot of time for the American sojourner in Japan to realize the differences of the prevalence of choice in both countries. A published anecdotal account of a Japanese visitor being barraged with choices during a visit to an American home can be found in Doi (1973) and his feelings towards them with a final exasperation, “I couldn’t care less. What a lot of trivial choices they were obliging one to make” (p. 12). An example of the differences in the amount of choice in Japan and America can be found in the serving of school lunches in either country. In Japan, the meal served to elementary school students is usually already chosen. While there is variety of the meals served over the week, all students eat the same school lunch decided by the school. A menu from an

32 elementary school in the US, however, shows a difference. For example, the lunch menu between April 14 and April 16, 2004 for elementary school students in Jupiter, Florida reveals the following choices an elementary school student could then make for those days:

Wednesday–choice of one: roast turkey on a hot roll, hamburger on a bun, peanut butter and jelly sandwich; peas and carrots, mashed potatoes with gravy; Thursday–choice of one: soft taco with Spanish rice, pizza, turkey and cheese sandwich; mixed vegetables; Friday–choice of one: fish fillet on a bun, corndog, hamburger on a bun; seasoned corn (“School Menus,” 2004, p. B6).

Perhaps this shows an early inculcation of the personal value of choice with children in the United States.

Some researchers investigating cultural differences in making a choice have used self-determination theory and one component of intrinsic motivation, autonomy, in their studies. Iyengar set the foundation in this line of research and has built upon her initial studies (Iyengar & DeVoe, 2003; Iyengar & Lepper, 1999, 2002; Iyengar, Lepper,

& Ross, 1999). Using the paradigm of Markus and Kitayama’s (1991) hypothesis of independent cultures (mostly in the West where the self is seen as separated from other groups or the family) and interdependent cultures (mostly in East Asia where the self is seen as being a part of the group or family), Iyengar proposed that individuals from interdependent cultures value independent choice less and will choose according to the group norms or be more highly influenced by others, such as a parent, than those from independent cultures. Iyengar (née Sethi, 1997) and Iyengar and Lepper (1999) found that children from an East-Asian culture (Chinese-American) in the San Francisco area were

33 significantly more motivated to engage in an activity when it was chosen by their mothers than children from Anglo-American cultures. In fact, Iyengar and De Voe (2003) stated that individuals from interdependent cultures, which these authors referred to as dutiful choosers, will have little, if any, intrinsic motivation (p. 163), a motivation that is completely internal to the person.

In response to this body of work, Chirkov, Ryan, Kim, and Kaplan (2003), proposed that the notion of autonomy has a relationship to well-being across cultures and that Iyengar and Lepper (1999) failed to take into account the autonomous value that one might have in submitting to another person choosing for them. Chirkov et al.

(2003) surveyed university students in the US, Turkey, Russia, and South Korea and found that “whatever cultural practices one is considering, there appears to be a positive relation between more internalized or autonomous regulation of those practices and well- being, as measured through both hedonic (happiness) and eudaimonic (self-fulfillment) indicators” (p. 106).

The Effect of Too Much Choice

As in her studies of culture and choice, Iyengar has also been influential in researching the effects of too much choice from a psychological perspective. Iyengar and

Lepper (1999, 2000) discovered that people with many items to choose from at a single time experienced greater frustration in the decision-process and found it more difficult

34 to choose, even though they enjoyed having a great amount of choice available, compared with those who had fewer choices.

Iyengar and Lepper (2000, Study 3) compared two groups of students’ decision- making process when choosing Godiva chocolates; those who had a limited choice of six varieties and those who had a choice from amongst 30 varieties of chocolate. Those who could choose from 30 varieties of chocolate spent more time choosing and enjoyed the decision-making process more than those in the group with six choices. In addition, those in the group with a limited variety to choose from felt that the number of choices was about right and found the decision-making process significantly less difficult and less frustrating, were significantly more satisfied with their choice, and were significantly more likely to choose chocolates as a compensation for participating in the study, compared with those in the no-choice or extensive-choice groups, who were more likely to choose cash as compensation. Iyengar and Lepper (2000) concluded that, contrary to the popular belief that having more to choose from is always best, having too much to choose from can be demotivating. In the end, this can lead to choice overload and a paradox that with great freedom in choice, there is more likelihood that people will depend more on institutions to help them make a decision (Iyengar & Lepper, 2000, p. 1004).

Another researcher in this area, Barry Schwartz, is more person-specific, focusing on how different personalities deal with the choices available. Schwartz’s supposition is that when the number of choices available for one object increases, the benefits to a

35 person’s well-being may decrease because of the person’s lack of skill in dealing with the amount of choice (Schwartz, 2004a).

The solution offered by Schwartz, however, is controversial. With more people with worse well-being, it is better to offer less choice, wrote Schwartz (2000, 2004a, 2004b;

Schwartz, Ward, Monterosso, Lyubomirsky, White, & Lehman, 2002), with the advice that people should “learn to love constraints” (Schwartz, 2004a, p. 235). Such a condition is obviously an anathema to many, and is especially criticized by libertarians. In the words of Solomon (2001), “Schwartz’s conclusion that people would be happier with fewer choices amounts to advocating regression in standard of living as well as liberty” (p. 80). Or, as a reader to the letters to the editor column of the New York Times wrote, “What choice do we have as to who should make the choice?” (Walker, 2004, p. 22).

Self-Determination Theory

A theory of motivation that operationalizes choice is Self-Determination Theory

(SDT). This theory of motivation was begun in the research of Maslow in the 1940s and has as its central axis the concept of will, as philosophized by William James. According to

Deci (1987b), self-determination involves the utilization of the will, which is the capacity to decide and to have those decisions be a part of one’s behavior. According to Ryan and

Deci (2000, p. 68), SDT is an organismic metatheory of human motivation centered on the human capacity for inner personality development and behavioral regulation.

36 Figure 1 shows the continuum of Self-Determination Theory from intrinsic motivation on the top through extrinsic motivation in the center to amotivation on the bottom. Important in this continuum is the perceived locus of causality, from the impersonal locus of causality for amotivation to the internal locus for intrinsic motivation.

The important aspect of the figure is the locus of causality. Each new regulation and its constituent components of perceived loci of causality become increasingly internalized as the continuum moves closer to intrinsic motivation. As the continuum moves in the direction of intrinsic motivation, there is an inherent increase in autonomy; movement in the opposite direction indicates an inherent decrease or absence of autonomy.

Amotivation

Amotivation is the lack of motivation. In most cases, this occurs when people

“lack either a sense of efficacy or a sense of control with respect to the desired outcome”

(Deci & Ryan, 2000, p. 237), or when “he/she has not figured out the goals for the behaviors nor the contingencies between their behaviors and outcomes, thus feeling helpless” (Tanaka & Yamaguchi, 2000, p. 256). With amotivation, people go through the motions of what they are doing with no personal attachment or intent. Amotivation makes people feel that they cannot achieve a desired outcome or outcomes because of a lack of contingency or perceived competence, or they feel no value to the activity or its possible outcomes (Ryan & Deci, 2002, p. 17).

37 Motivation Type Regulatory Styles Locus of Causality Intrinsic Motivation Intrinsic Regulation Internal Integrated Regulation Internal Identified Regulation Somewhat Internal Extrinsic Motivation Introjected Regulation Somewhat External External Regulation External Amotivation Non- Regulation Impersonal

Figure 1. The Self-determination continuum (from Ryan & Deci, 2000, p. 72).

Extrinsic Motivation

In the middle of Figure 1 are four points of extrinsic motivation. Extrinsic motivation is referred to as the Organismic Integration Theory (Deci & Ryan, 2000;

Ryan & Deci, 2000). This theory was developed to “detail the different forms of extrinsic motivation and the contextual factors that either promote or hinder internalization and integration of the regulation of these behaviors” (Deci & Ryan, 2000, p. 72). The information about these different levels of extrinsic motivation in the following four paragraphs, unless otherwise noted, is amalgamated from Deci and Ryan (2000), Ryan and Connell (1989), and Ryan and Deci (2000, 2002).

With external regulation, behaviors are carried out to achieve an external reward, comply with a rule, and avoid punishment. For example, many students who dislike

English attend English classes in Japanese universities because they are required to do so.

External regulation is in evidence when a person’s reason for accomplishing a behavior is to satisfy an external demand or a socially constructed contingency. External regulation is the central focus of operant theories of behavior.

38 Introjected regulation occurs when a person acts from esteem-based pressures, such as avoidance of guilt or because of public self-consciousness. These regulations are within the person but are still relatively external to the self, or, in the pithy phraseology of Perls (1973), introjected regulation is akin to swallowing regulations whole without digesting them (pp. 32-33). An example of introjected behavior would be when a person follows a maxim such as “do unto others as they would do unto you,” not because he or she believes it but because society accepts such maxims. In this case, the regulation is internalized but is not accepted as one’s own and is not part of the integrated self.

A more internal locus of regulation on the extrinsic motivation continuum is identified regulation. With this type of regulation, people recognize and accept the underlying value of a behavior. Although this type of regulation reflects a conscious valuing of a behavioral goal or regulation and is more autonomous, it is still extrinsically motivated because a person will willingly engage in that behavior but without personal attachment. However, although the behavior is more internally regulated and the person identifies with the behavior and personally endorses it, some of these endorsements can be relatively compartmentalized or separated from one’s other beliefs and values, in which case they may not reflect the person’s overarching values in a given situation.

The most internally regulated and autonomous form of extrinsic motivation is integrated motivation. In this case, the regulations are fully assimilated to the self and the person identifies with the importance of the behaviors. An example of identified behavior is when a person will learn a language because it is necessary for them to learn it in order

39 to be able to pursue a hobby or an interest (Dörnyei, 1994). This type of regulation cannot typically become intrinsic motivation because there are still remnants of instrumental reasons for acting. Behaviors are still performed in order to attain personally important outcomes rather than for their inherent interest and enjoyment. However, Ryan and Deci

(2002) suggested that this level of regulation should be promoted by enhancing autonomy and relatedness through the use of autonomy-supporting teaching practices.

Intrinsic Motivation

Lastly, at the top of the continuum, is intrinsic motivation, referred to as the

Cognitive Evaluation Theory by Deci and Ryan (1985). A key characteristic of intrinsic motivation is that the locus of causality lies inside the person. This idea is not new to the thought of man. St. Augustine, in his Confessions, wrote, “It is clear enough that free curiosity has a more positive effect on learning than necessity and fear” (Kelly, 1969, p.

323). Adler (1930) introduced a “striving for superiority” construct that is “an intrinsic necessity of life itself. It lies at the root of all solutions to life’s problems, and is manifested in the way in which we meet these problems” (Adler, 1930, p. 398). The construct of intrinsic motivation has been studied intensively in many fields, but especially in education. The paragraph below is an amalgamation of statements about intrinsic motivation from de Charms (1968), Deci (1971, 1972, 1975, 1987a), Deci and Ryan (1985,

2000), Reeve (1997), and Ryan and Deci (2000).

40 Intrinsic motivation is based in the innate, organismic needs for competence and

self-determination and as such is the innate propensity to seek out novelty, to master

optimal challenges, to extend and exercise one’s capacities, and to explore and to learn.

Intrinsic motivation occurs when individuals experience themselves to be the locus

of causality for their own behavior, when they receive no apparent rewards except the

activity itself, or they perform an activity for no apparent reason other than the activity

itself. Intrinsically motivated activities are those that are freely engaged in out of interest

without the need for external evaluation or reward. More recently, researchers have

striven to understand what factors influence intrinsic motivation and the influences that

intrinsic motivation exerts on learning and other activities.

Three Components of Intrinsic Motivation

According to Deci and Ryan (1985), there are three main components to intrinsic motivation: competence, relatedness, and autonomy. The first component, competence

(White, 1959), is the feeling that an activity is optimally challenging. Competence is characterized by a degree of self-determination (Deci & Ryan, 1985, pp. 58-59).

The second component is relatedness. Here the person feels a sense of security and the desire to feel connected to others (Deci & Ryan, 2000). Based on their research results, Deci and Ryan proposed that competence has the second strongest connection to intrinsic motivation and that relatedness has the weakest.

41 By far the strongest influence in Deci and Ryan’s hypothesis of intrinsic motivation is autonomy, which they defined as “the organismic desire to self-organize experience and behavior and to have activity be concordant with one’s integrated sense of self” and is “the experience of integration and freedom, and it is an essential aspect of healthy human functioning” (Deci & Ryan, 2000, p. 231). Under the theory of self- determination, autonomy occurs when individuals “act in accord with their authentic interests or integrated values or desires” (Chirkov et al., 2003, p. 98), but it can also occur when a person is forced, for example, to accept guidance from a parent or to submit to a traffic policeman, where one sees value in following the commands. Just as importantly,

Deci and Ryan (2000) stated what autonomy is not. Autonomy, in their theory, is not

“equated with ideas of internal locus of control, independence, or individualism” (p. 231).

In Deci and Ryan’s conceptualization of autonomy, the most important component is choice. If there is no choice, there is no autonomy, and if there is no autonomy, there is no intrinsic motivation. According to Dworkin (1988), being autonomous, i.e., human, means to be able to choose on one’s own. “What makes a life ours,” wrote Dworkin, “is that it is shaped by our choices” (p. 81).

Autonomy and the Japanese Self

How do Japanese describe themselves in relation to autonomy? The Japanese conception of choice and how it is inculcated through culture may play a large role in the value of choice by Japanese. In early anthropological writings on the Japanese self,

42 Moloney (1953) wrote that in Japan, “there is no such thing as a true individual who feels, thinks, and acts voluntarily in a self-determined manner” (p. 302). In addition, Caudill

(1963) stated that individual autonomy and mastery are not so much valued in Japan as are the development of collateral relations in the sense of sharing. Also Tsuchida (1992) stated that:

For the American, it is not only a right to exercise control over one’s own destiny, but also one’s duty. Death and life are one’s own private concern. The Japanese, by contrast, have lived for centuries in a highly integrated and contextualized society where even life and death have to be seen as a family affair–if not the affair of the community as a whole–as much as the affair of the particular individual. (p. 321)

However, when considering autonomy and the Japanese self, it would be unfair to make blanket statements. According to Wagatsuma (1983), attributing group behavior to a national character constitutes psychological reductionism. Many different individual needs can motivate what seems to be group-oriented behavior. In fact, Kuwayama (1989) claimed that Japanese can be quite self-assertive to the extent that their own interests often precede those of others at the expense of cooperation rather than conforming to a popular image of the Japanese as a cooperative, harmonious, and self-effacing people.

Kuwayama offered four pieces of evidence for autonomy in the Japanese self, (a) the consistent use of words that indicate the self across diverse situations, (b) an invisible, but irremovable, dividing line between individuals and different religions, (c) a clear reflection of individuality in the fact that it is extremely difficult for Japanese to work together on a community basis, and (d) independence of mind that is veiled behind the household ideology.

43 Amae

An important model explaining the Japanese self is Doi’s (1973, 1993) concept of amae. As Tyler (1983) explained, amae is the “key to understanding the psychodynamics of Japanese culture which is relatively tolerant of dependency feelings and relations” and that “whereas Japanese society encourages and institutionalizes dependency feelings, Western cultures often fail to recognize them or they even ignore them in an unquestioning faith in the self-reliance of ‘rugged individualism’” (p. 49). Johnson

(1993) identified amae as a basic motive or drive (p. 200) and defined amae as, “The need to be responded to, taken care of, and cherished; the mutually interactive attitude or behavior whereby one seeks and (ideally) receives the indulgence of another” (p. 374).

From this, autonomy seems to be a weaker entity in the Japanese mind than a feeling of interdependent dependency.

Recent hypotheses of amae, however, suggest that this construct does not necessarily mean the relinquishment of control by a person and that the hypotheses of primary control and secondary control are important for adding a new dimension to amae (Weisz, Rothbaum, & Blackburn, 1984; Yamaguchi, 2004). Weisz et al. wrote that primary control is an attempt to influence existing realities by influencing the realities to fit the self. Secondary control, on the other hand, is an attempt to accommodate to existing realities by influencing the psychological realities of the self. Both Weisz et al. and Yamaguchi speculated that secondary control is predominant in Japanese culture.

Yamaguchi related secondary control to amae by suggesting that “in amae, individuals

44 attempt to control their environment using someone who is more powerful in the situation than themselves” (p. 30). In addition, those using amae successfully can have someone often more powerful under their control (Nakane, 1970; Yamaguchi, 2004, p. 31).

In summary, although some writers have claimed that there is a lack of autonomy in Japanese students, examined more deeply, significant autonomy seems to exist.

Evidence of autonomy can be found in recent developments of Japanese society, such as increased feelings of individualism, as well as a deeper understanding of amae, which might seem on the surface to be a relinquishment of personal autonomy, but is actually a device to increase one’s autonomy and personal presence in the spheres of daily life.

Self-Determination Theory and Language Learning Motivation

As written in Chapter 1, there are several competing hypotheses of language learning motivation, one of which is sdt. In this regard, Kimberly Noels has recently been in the forefront of research using sdt to help explain language-learning motivation.

Ramage (1990) investigated who would and would not continue studying a foreign language among 138 US high school students studying Spanish and French. The author found that the learners who continued were more likely to have intrinsic motivation associated with (a) general interest in the culture, (b) increasing one’s knowledge, and (c) learning the language (p. 210). Extrinsic motivations associated with continuing to learn a foreign language were rooted in a perceived usefulness of the language in college and future jobs. In addition, Ramage found that attitudes about the learning situation were

45 influential and that those learners who first took the class at an earlier grade (in high school) were more likely to continue learning the language.

As for the learners who did not continue, Ramage found that the reason for taking the course in the first place was to fulfill a requirement and the requirement being filled, they decided not to continue. According to Ramage, discontinuing students had an interest in language learning as a means to other goals with weaker traces of some intrinsic interest in learning a foreign language (pp. 211-212).

Noels, Pelletier, Clément, and Vallerand (2000) surveyed 156 Anglophone adults learning French in Canada using the Academic Motivational scale of Vallerand, Pelletier,

Blias, Brière, Senécal, and Vallières (1992) based on SDT. Noels et al. (2000) found that an autonomy-supporting classroom atmosphere can enhance students’ pleasure in learning the foreign language, therefore supporting the usefulness of the self-determined motivation paradigm in the language classroom. In addition, in this study, strong perceptions of freedom of choice and perceived competence were linked with more self-determined forms of motivation and, conversely, weak perceptions of freedom of choice and perceived competence were linked with amotivation. The authors also found that though some learners may not feel involved in the study of a second language, they nevertheless find pleasure in it, as in an extrinsic identified regulation orientation (p. 75).

McIntosh and Noels (2004, p. 15) studied the interrelation of self-determined motivation with language learning strategies and the need for cognition in 126 undergraduate students at a university in Canada and found that self-determination is

46 associated with a number of specific language learning strategies identified in the L2 literature (e.g., Oxford 1990). These authors wrote that there was a significant relationship between the need for cognition and self-determined motivation, thus introducing the speculation that learners who enjoy effortful thinking for its own sake are likely to begin studying a second language for self-determined reasons.

Self-Determination Theory and Language Learning Motivation in Japan

Kamada (1987), an early contributor to the area of sdt and language learning motivation in Japan, wrote about the effects of English language teaching methodology on the more extrinsic forms of motivation found in many Japanese learners. Kamada stated that learners not only took seriously the highly salient extrinsic goal of studying

English to pass the entrance examination (for high school or college), but also the threat that failing imposes. As a result, these students utilized rote memorization as the major learning strategy. Kamada speculated that the most successful English students in Japan are those who are diligent and have high extrinsic motivation; these students pass into the elite universities and large international companies but end up with a mental block against learning English as they struggle to acquire English speaking skills.

Tanaka and Yamaguchi (2000) surveyed 121 undergraduate university students in

Japan about learning English using the Academic Motivational scale of Vallerand et al.

(1992), which is made up of the categories of intrinsic regulation, identified regulation, introjected regulation, external regulation, and amotivation, in order to assess the

47 students’ degree of autonomy. Using structural equational modeling, Tanaka and

Yamaguchi found that intrinsic motivation and identified regulation had paths (.32 and

.44, respectively) to a mastery orientation and that identified regulation and introjected regulation had paths (.23 and .58, respectively) to a grade orientation. These orientations in turn had paths to deep processing (.52 and .21, respectively), which was correlated with academic achievement (.22) (p. 266). These authors found that identified regulation played a large positive role in the students’ motivational architecture. Tanaka and Yamaguchi proposed that those students with high intrinsic motivation and identified regulation scores were “optimally motivated” and that “reducing feelings of being controlled is as important as enhancing the feelings of autonomy in order for students to achieve higher academic performance” (p. 268). This is the same conclusion reached by Noels, Clément,

Pelletier, (1999, 2001) as well as Noels et al. (2000).

An extensive study of motivation was conducted by Nakata (2004), who used self-determination theory to examine the verbal interaction that Japanese learners engage in from a constructivist perspective. Nakata proposed a model of intrinsic motivation where there are two levels of intrinsic motivation; a surface level in which “the students study with enjoyment but may quit studying in the future” (p. 170), and a core level, that is “strong enough to make learner’s habitual learning behaviour or continuous impulse to learn happen” (p. 170). Students with this core motivation will more likely continue learning long after the class is finished. Nakata related the details of five adult learners and their changes in language learning motivation over the duration of a language-

48 learning course. How these learners fit in his model of intrinsic motivation was key to this research. The author reported five main findings concerning intrinsically motivated learners. First, they became more confident about their English skills. Second, they dealt with anxiety-provoking situations in productive ways rather than becoming debilitated by anxiety. Third, they sought out additional learning experiences after the project was completed when compared to their classmates. Fourth, they realized the importance of studying English for personal enhancement and enjoyment. Fifth, and most importantly, they were autonomous and the students became more self-directed in their learning.

However, even though all the learners increased their intrinsic motivation, only two of them entered into the core area of Nakata’s conception of intrinsic motivation and only these two continued to learn English on their own, as Nakata originally hypothesized.

Autonomy and Language Learning

Although autonomy in the language learning classroom has probably been incorporated into many foreign language curricula for a long time, autonomy in language learning was first defined in the 1970s at the Council of Europe’s Modern Language

Project’s Centre de Recherches et d’Applications Pédagogie en Languages, or CRAPEL for short, in Nancy, France (Benson, 2001, p. 8). Since then, much of the work on autonomy has centered in Europe with the current work of Little in Ireland, Ushioda in England, Riley at

CRAPEL, and Lamb in England. Benson in Hong Kong has also been very influential in researching and conceptualizing autonomy in the foreign language teaching curricula.

49 Definitions and Early Research

Holec’s (1981) Autonomy in Foreign Language Learning is the seminal work on autonomy in foreign language learning, and his definition of autonomy in foreign language learning has exerted a lasting influence on autonomy in language teaching research. According to Holec, autonomy in language learning is “the ability to take charge of one’s own learning” (p. 3). By this, Holec meant that the learner had to take responsibility and to be capable of making decisions about learning in regards to: (a) determining learning objectives, (b) defining the contents and progressions, (c) selecting methods and techniques, (d) monitoring the procedure of acquisition, and (e) evaluating what has been acquired (p. 3). For autonomy to be realized in foreign language learners, however, Holec proposed that two conditions must be fulfilled; learners must know how to take charge of their learning and they must have the possibility for exercising this ability. Subsequently, informing teachers about how learners can accomplish these two conditions has been a guiding force for autonomy researchers.

Autonomy as a Degree of Capacity

In the field of language learning, researchers such as Little (1991, 1999) and Aoki

(1998) claimed that autonomy, rather than being an all-or-none concept, is a matter of degree or capacity. According to Little (1991):

Autonomy is a capacity–for detachment, critical reflection, decision- making, and independent action. . . . The learner will develop a particular kind of psychological relation to the process and content of his learning. The capacity for autonomy will be displayed both in the way the learner

50 learns and in the way he or she transfers what has been learned to wider contexts. (p. 4)

With autonomy, according to Little, learning should be fun, barriers between the classroom and the outside world should not arise, and learners should not experience difficulty transferring their autonomous capacity to all other areas of their lives. Little

(1991) also listed five misconceptions about autonomy: autonomy (a) is the same as self- instruction, (b) requires the teacher to relinquish control and all initiative, (c) is done to the learners or is a new methodology, (d) is a single, easily described behavior, and (e) is a steady state achieved by certain learners (pp. 3-4).

Littlewood’s Model of Language Learning Autonomy

Littlewood (1996) defined an autonomous person as “one who has an independent capacity to make and carry out choices which govern his or her actions” (p. 428). This form of autonomy requires two capacities, ability and willingness, each of which can be broken into two subcategories. Ability depends on the skills needed to carry out choices and the knowledge about what choices have to be made. Willingness involves having both the motivation and the confidence to take responsibility for the choices needed for learning (p. 428).

There are, according to Littlewood, three types of autonomy that a learner can manipulate (p. 431). The first is to be an autonomous communicator, which depends on using the language creatively and with appropriate strategies for communicating meaning. The second is autonomy as a learner, which depends upon working

51 independently with appropriate learning strategies in and out of the classroom. The third is autonomy as a person, which depends on the ability to express personal meanings and to create personal learning contexts.

According to Littlewood (1996, pp. 429-430), there are certain levels of choices that learners can progress through in influencing the learning curriculum. Initially, learners make choices in grammar and vocabulary and progress to making choices in relation to short- and (later) long-term goals in the meanings that they want to express and the communication strategies needed to achieve their communicative goals. After that, learners progress to making decisions that used to be made by the teacher, such as selecting materials and tasks. Lastly, learners can make their own decisions about the nature of the syllabus and its progression.

Using his definition of autonomy with the levels of choice and decision-making required with the three types of autonomy, Littlewood has proposed a model for developing autonomy in foreign language learning (Figure 2). Littlewood uses a circle to emphasize his contention that all the concepts are linked and that focusing on one will affect the others. In the center are the crucial components of autonomy, motivation, confidence, knowledge, and skills. The three boxes on the periphery of the circle are the three domains of autonomy, autonomy as a communicator, as a learner, and as a person.

Also on the periphery are six pedagogical concentrations. For example, by promoting independent work in the classroom, a pedagogical concentration, the students’ autonomy

52 as a learner is strengthened. Through this, there is a direct connection to increased

motivation, confidence, knowledge, and skills.

$PNNVOJDBUJPO -FBSOJOH 4USBUFHJFT 4USBUFHJFT

"VUPOPNZ "VUPOPNZ BTB BTB $PNNVOJDBUPS -FBSOFS

.PUJWBUJPO $POĕEFODF ,OPXMFEHF -JOHVJTUJD 4LJMMT *OEFQFOEFOU DSFBUJWJUZ XPSL

&YQSFTTJPOPG $SFBUJPOPGQFSTPOBM QFSTPOBMNFBOJOHT MFBSOJOHDPOUFYUT

"VUPOPNZ BTB 1FSTPO

Figure 2. Littlewood’s model for developing autonomy in foreign language learning (from Littlewood, 1996, p. 432).

Proactive and Reactive Autonomy

Littlewood (1999) also promoted the notion of proactive and reactive autonomy.

Proactive autonomy is thought to be the prevalent type of autonomy in the West, where learners take charge of their own learning, select methods and techniques for learning, determine their learning objectives, and evaluate what they have done. On the other

53 hand, reactive autonomy, according to Littlewood (1999) “is the kind of autonomy which does not create its own direction but, once a direction has been initiated, enables learners to organize their resources autonomously in order to reach their goal” (p. 75).

In this second case there is more group work and a greater use of collaborative and cooperative learning strategies. It was Littlewood’s contention that Asian learners are more reactive in their autonomy orientation and have experienced few learning situations where proactive autonomy could be exercised. However, Asian learners have the same capacity for autonomy as learners in other regions of the world and they can develop high levels of both proactive and reactive autonomy in group-based forms because language teachers can, without inhibiting learner freedom of choice, give the learners experience in exercising proactive autonomy in the public domain.

van Lier’s Model of Language Learning Autonomy

van Lier (1996) proposed a trinary hypothesis of a “growth of proficiency,” with autonomy playing a central role. In this hypothesis there is awareness, which involves focusing one’s consciousness, authenticity, which van Lier sees as an action done freely and is an expression of what one genuinely feels and believes, and, most importantly, autonomy, which to van Lier involves both responsibility and choice. van Lier, following

Holec, believes that autonomous learners must be able to make decisions about what to learn, when to learn, and what it means to be a learner (p. 13).

54 TPDJBM JOUFSBDUJPO SFDFQUJWJUZ FYQPTVSF "8"3&/&44 QFSDFQUJPO BUUFOUJPOGPDVTJOH

BDDFTTQBSUJDJQBUJPO FOHBHFNFOU

"650/0.: DPHOJUJPO BQQSFIFOTJPO DPNQSFIFOTJPO JOUBLF JOWFTUNFOUQSBDUJDF NBTUFSZ VQUBLF NFNPSZ BDDFTT "65)&/5*$*5: DPOUSPM DSFBUJWJUZ DPNNJUNFOU QSPĕDJFODZ FYUFOTJPO

Figure 3. van Lier’s model of the growth of proficiency (from van Lier, 1996, p. 41).

van Lier places a central importance on social interaction, which is represented in the uppermost wheel (see Figure 3). Social interaction is the driving force, the engine “which moves all the elements of the process . . . that gets the learning wheels of awareness, autonomy, and authenticity turning and thus produces learning like a CD player produces sound” (p. 42).

On the far left are receptivity, access/participation, investment/practice, and commitment. These, van Lier claims, are the conditions necessary for language learning.

On the far right are the outcomes of language learning; perception, cognition, mastery

(uptake), and creativity. In the center, within the larger circles, or wheels, is the process of

55 language learning; exposure, engagement, intake, and proficiency (p. 42). The wheels of awareness, autonomy, and authenticity drive the learner’s progress.

-FBSOJOHFYQFSJFODFT $SPTTJOHBVUPOPNZUISFTIPMEi3VCJDPOw JOUIFFEVDBUJPOBMTZTUFN CZMFBSOFSTXJUIJOUSJOTJDNPUJWBUJPO

.PUJWBUJPOBMDPHOJUJWF TFMGSFHVMBUJPO

-PXJOUSJOTJD 4PDJBMJOUFSBDUJWF 9 NPUJWBUJPO QSPDFTT /0/"650/0.06445"(& "650/0.06445"(&

Figure 4. Early-stage model of motivational development (from Nakata, 2004, p.

160).

Nakata’s Model of Language Learning Autonomy

Nakata (2004) proposed the existence of an autonomy threshold that learners pass through (Figure 4). Taking the idea of Dörnyei’s process model of motivation (Dörnyei,

2001; Dörnyei & Ottó, 1998), Nakata suggested that there is an autonomy threshold, in which only those learners with intrinsic motivation can pass through. Based on this idea, Nakata proposed an early-stage model of motivational development that is based on three suppositions that differ from Dörnyei’s process model. The model (a) focuses on the early stages of motivational development, (b) is more concerned with social motivation than personal motivation, and (c) deals with the motivational process in terms of a habitual learning behavior rather than a particular act (Nakata, 2004, p. 161).

56 Up to the autonomous stage, learners act in a motivated manner and there is a nascent responsibility taken by learners for their own learning. In the autonomous stage, which only intrinsically motivated learners can pass into (p. 158), learners become autonomous, which, to Nakata, includes both cognitive self-regulation (with meta-cognition) and motivational self-regulation.

Oxford’s Perspectives on Language Learning Autonomy

Oxford (2003) offered an outline of how language learning autonomy can be realized in the foreign language learning curriculum using four perspectives of autonomy: (a) a technical perspective that focuses on the physical situation, (b) a psychological perspective that focuses on the characteristics of the learners, (c) a sociocultural perspective (which consists of two sub-perspectives) that focuses on mediated learning, and (d) a political-critical perspective that focuses on ideologies, access, and power. In the technical perspective, autonomy is seen as skills for independent learning situations. In this perspective, learning strategies (i.e., Oxford, 1990) are important. In the psychological perspective, which was first proposed by Benson (1997, p.

23), autonomy is seen as a combination of the learner’s attitude, ability, learning strategy, and style characteristics. In the first sociocultural sub-perspective, autonomy is viewed as a construct of self-regulation gained through interaction with a more capable person in a particular setting. In the second sociocultural sub-perspective, the goal is on social interaction and participation in the community rather than autonomy. In the politico-

57 critical perspective, autonomy concerns gaining access to cultural alternatives, power structures, and the development of a voice.

Cultural Influences and Language Learning Autonomy

As popular as language learning autonomy seems to be, some researchers (e.g., Ho

& Crookall, 1995; Jones, 1995) believe that not all cultures value autonomy equally and, in fact, autonomy in the foreign language classroom may cause some learners consternation.

In addition, some practitioners (e.g., Pennycook, 1997) have claimed that promoting autonomy in some situations may be culturally insensitive. Jones (1995), relating his experience setting-up a self-access center in Cambodia, stated that autonomy remains a

Western idea, and that “to make autonomy an undiluted educational objective in a culture where it has no traditional place is to be guilty at least of cultural insensitivity” (p. 229). Ho and Crookall (1995, p. 237) related the idea of autonomy in the classroom in the Chinese cultural environment and the effect that it would have on the learners who want to protect the face of the teacher. These students, the authors said, would not be comfortable with autonomy because engaging in autonomous behavior may risk presenting opinions that differ from the teacher’s. Pierson (1996) wrote that compared to Americans and

Australians, Hong Kong Chinese place less value on individualism, outward success, and individual competence and that Chinese Hong Kong learners may not regard learner autonomy, independence, or self-direction highly due to cultural forces (p. 52).

58 Nakata (2004, pp. 177-178) stated that several aspects of Confucian teaching

can affect the motivation (and feelings of autonomy) of Japanese students in five ways.

First, the expression of one’s beliefs in public is regarded negatively. Second, students

are required to obey the teacher and respecting one’s teacher is a virtue. Third, making

an effort is such a virtue that some Japanese view such changeable factors as effort as

more important than innate ability. Fourth, persistence is valued over personal ability.

Fifth, harmony in interpersonal relations and the ability to cooperate with other people

are highly valued. These together, which have been inculcated into Japanese students

from preschool age (e.g., Tobin, Wu, & Davidson, 1989), can make for a dynamic very

different from the Western student and possibly incompatible with Western theories of

education and motivation, which places the focus on the self. For example, Tobin et al.

claimed that while American preschool teachers place a high value on self-reliance and

self-confidence, Japanese preschool teachers place a high value on what is termed omiyari, empathy or concern for others (p. 190). Neither self-reliance in Japan nor empathy in

America were disregarded, but they seemed to be mirror opposites of each other in the two cultures. However, for preschool teachers in both cultures, cooperation was equally highly valued. In cultures where self-reliance is important, it is possible that individual autonomy, and therefore choice, may be more highly valued. In cultures where concern for others is emphasized, it is possible that a certain degree of individual autonomy may be relinquished and choice may not be as important as getting along with others.

59 The Link between Autonomy and Motivation in Language Learning

Dickinson (1995) attempted to link autonomy and motivation in language learning through two models of educational psychology, intrinsic motivation and attribution theory. The central tenet of attribution theory is the learner’s perception of success or failure (ability, task difficulty, effort, luck) and the influence that this perception has on subsequent performances. In the best case, greater motivation will occur when students view success as the result of personal effort (p. 171). Teachers can promote higher motivation in the classroom by developing students’ realistic goal setting, planning, personal responsibility, feelings of personal causation, and self-confidence. This approach can be combined with teaching that allows students to make choices and is not perceived by the students as controlling.

Ushioda (1996) proposed a connection between autonomy and motivation through the self-determination theory of motivation. What is new in her research is that she claimed that intrinsic motivation can be supported in the collaborative language learning environment. Ushioda (2004) proposed that researchers investigating the connection between autonomy and motivation have focused too much on the self- determined theory of motivation and have neglected the sociocultural paradigm.

Ushioda also suggested that Vygotskian sociocultural theory can illuminate motivational dimensions and allow different interpretations of data to be revealed. One example is that the interaction between autonomy and motivation can be revealed as learners pursue

60 optimal challenges through the zone of proximal development during collaborative learning. According to Ushioda (1996):

Collaborative learning in itself can create the appropriate psychological conditions for intrinsic motivation, since it explicitly puts the learning initiatives and control of the learning process in the hands of the students themselves, by harnessing their sense of peer-group solidarity and shared responsibility, and minimizing their perception of external direction and control from the teacher. (p. 46)

van Lier (1996, p. 103) as well sees a connection between motivation and autonomy through the theories of Vygotsky. According to van Lier, three clusters of constructs, each existing in a dynamic relationship with one another, can capture the essence of motivation: intentionality (related to choice), affect, and effort. By arousing intentionality, which in the Vygotskian scheme is an effect of the volitional side of human nature, positive affect may also be triggered, and from that learners may expend effort.

Working Definition of Language Learning Autonomy

In this study, autonomy is defined as the amount of choice that students have in selecting the topic of the task. The students will have no choice in the topic, a limited choice of three topics on the same task, or unlimited topic choice confined only by the limitations of the task-type. The reason that a level of limited choice is incorporated in the design is to introduce an intermediate level between complete teacher control of topic and complete student control of topic. This level melds choice with teacher guidance and reflects an intermediate degree of choice that many teachers could implement. The intermediate level of choice also allows the researcher to test hypotheses of linearity in

61 a clearer and a more convincing way compared to a binary difference between all or

none. Finally, as noted above, previous research has indicated that limited choice may be

preferable to complete choice.

Gaps in the Literature

A review of the research indicated that there are three gaps in the literature. These are gaps in the cross-cultural psychological research literature, in the TBLT research literature, and in the pedagogical autonomy literature.

First, in the cross-cultural psychological perspective, researchers (e.g., Iyengar &

Lepper, 1999) claiming that autonomy is not motivating for Asians culture used children as their participants; however, it is unclear that this would be the case for adults. An aim of this study is to examine differences in student motivation while engaging in a task when choice is introduced in the Asian setting.

Second, in the task-based language teaching, task implementation that can in some way help promote motivation is non-existent. Although Long in his seminars, Ellis

(2003), and Dörnyei (2002) have claimed that task-based language teaching in itself may be motivating, no empirical data is available to support these claims. A benefit of this study would be to provide information on the role of motivation using TBLT.

In addition, the participants in previous studies (e.g., Foster & Skehan, 1996;

Skehan & Foster, 1997) were of higher English language proficiency, studying in an

English as a second language environment, and is already motivated in the sense

62 that they had some compelling reason to move from their home country to an ESL environment and then enroll in an elective language class. The participants in this study have a low level of oral proficiency in English, they are attending a required course in

English in an English as a foreign language environment, and many of them are not particularly motivated to develop their oral proficiency in English.

From the pedagogical autonomy literature, a simple way of introducing autonomy in the classroom setting needs support. Although literature that is concerned with promoting autonomy in the general language curriculum exists, many of the articles on autonomy in Benson and Voller (1997), Mackenzie and McCafferty (2002), and Pemberton et al. (1996), for example, concern the operationalization of self-access centers and self- study centers or program-wide intervention. Although some articles in Barfield and

Nix (2003) have advice for introducing autonomy in the classroom, what is lacking is a method of introducing autonomy in the foreign language classroom that does not require a great deal of curriculum revision, is more under the control of the teacher, and is more specific towards the task.

Research Questions

Study 1

The primary purpose of Study 1 is to examine the participants’ task interest and task self-efficacy.

63 Research Question 1: To what degree does the level of task interest change across three levels of choice (i.e., none, limited choice, complete choice)?

Hypothesis 1: It is hypothesized that task interest will increase significantly when choice is available. This hypothesis is based on studies comparing the presence and absence of choice when adults are engaged in a task. However, task interest may decrease when complete choice is implemented. (e.g., Iyengar & Lepper, 2000; Schwartz, 2004a, 2004b).

Research Question 2: To what degree does the level of task interest change between the three types of tasks (i.e., descriptive, narrative, and decision-making)?

Hypothesis 2: It is hypothesized that the descriptive task may engender more task interest because it may be a more familiar task (e.g., Zuengler & Bent, 1991) and because it may better match the proficiency level of the students. Regarding the first reason, as required by government standards (Ministry of Education, Culture, Sports, and

Technology, 1998) students have been exposed to vocabulary and schema to complete the descriptive task from junior high school, as early as the first year of their study of English

(e.g., Sano, Yamaoka, Matsumoto, & Sato, 2006). Regarding the second reason, Brown and Yule (1983) claimed that this is a simple task to complete and this task may better match the beginning level of these students.

Research Question 3: To what degree does the level of task self-efficacy change across the three levels of choice?

Hypothesis 3: It is hypothesized that task self-efficacy will increase significantly when more choice is available. This hypothesis is based on studies comparing the

64 presence and absence of choice when adults are engaged in a task and that more control of the environment increases the ability to do a task (e.g., Monty, Rosenberger, &

Perlmuter, 1973; Stotland & Blumenthal, 1964).

There is no data showing how the level of self-efficacy will change in response to different levels of choice, as there is for task interest. There may be a relation, however, because interest and self-efficacy are both affective constructs. A component of self- efficacy is present in Dörnyei’s (1994) model of language learning motivation, and in the motivational component of Trembley and Gardner’s (1995) revision of the Socio-

Educational model of Gardner (1985).

Research Question 4: To what degree does the level of task self-efficacy change between the three types of tasks?

Hypothesis 4: As in Research Question 2, it is hypothesized that the descriptive task may induce greater self-efficacy because it is a more familiar task (Sano et al., 2006) and because it may better match the proficiency level of the students (Brown & Yule, 1983).

Study 2

The primary purpose of Study 2 is to examine the students’ language production from a qualitative perspective. In this study, the conversations that occurred while participants were engaged in the tasks in this study were recorded, transcribed, and coded for occurrences of accuracy, complexity, and fluency.

65 Research Question 1: To what degree does the level of accuracy change across the three levels of choice?

Hypothesis 1: It is hypothesized that accuracy will increase significantly when choice is available. This hypothesis is based on studies comparing the presence and absence of choice when adults are engaged in a task requiring high levels of attention

(e.g., Dember, Galinsky, & Warm, 1992).

Research Question 2: To what degree does the level of accuracy change between the three types of tasks?

Hypothesis 2: It is hypothesized that the participants will produce greater accuracy with the decision-making task, less accuracy with the descriptive task, and the least accuracy with the narrative task (e.g., Foster & Skehan, 1996; Skehan & Foster, 1997).

Research Question 3: To what degree does the level of complexity change across the three levels of choice?

Hypothesis 3: It is hypothesized that complexity will increase significantly when more choice is available. This is hypothesized because it is possible when choice is introduced in the implementation stage of a task, attentional resources may be freed and allocated towards complexity (e.g., Dember et al., 1992).

Research Question 4: To what degree does the level of complexity change between the three types of tasks?

Hypothesis 4: It is hypothesized that learners will produce greater complexity with the narrative task, the least complexity with the descriptive task, and an

66 intermediate position will comprise the decision-making task (e.g., Foster & Skehan,

1996). However, according to Robinson’s (2001a) results concerning task complexity, output for the decision-making task will be the most complex because of the high reasoning demands inherent in that type of task. Less complex output will be produced for the descriptive and the narrative tasks because the tasks are in the Here-and-Now.

These results were similar in Skehan and Foster (1997).

Research Question 5: To what degree does the level of fluency change across the three levels of choice?

Hypothesis 5: It is hypothesized that fluency will increase significantly when more choice is available because increases in task interest caused by the introduction of choice can positively affect fluency. This could be an effect of an increased willingness to communicate (e.g., MacIntyre, Clément, Dörnyei, & Noels, 1998; Yashima, 2002).

Research Question 6: To what degree does the level of fluency change between the three types of tasks?

Hypothesis 6: It is hypothesized that fluency will be higher for the decision- making task. Because this task is a two-way task, both parties contribute equally to the completion of the task. With the required speaking participation of both students, the word count for this task will possibly be higher (e.g., Long, 1989).

67 CHAPTER 3

METHOD

This chapter begins with a description of the participants and the setting where this study took place. First, the students who participated in Study 1 are described, and this is followed by a description of the participants who took part in Study 2. The next section is a description of the task materials and the survey instrument. The procedures used during the data collection sessions follow. Following this, the design of the study and the order of the task implementation are described. The procedures used for calculating the production data end this chapter.

Participants

The participants in this study were mostly first-year Japanese university students in two similar groups, 111 participants in Group A and 81 participants in Group B. Of these, 96 participants attended class consistently enough to be included in Study 1 (47 participants from Group A and 49 from Group B), and 42 participants (all from Group

B) in 21 pairs who together attended consistently enough to be included in Study 2. Some students from Group B were included in Study 1 but not Study 2 because even though they attended all the sessions, their partner did not, causing their removal from Study 2.

The Group A students attended an English Communication I class in the spring semester

(April to July) of 2005 that met Monday afternoons (14:50 to 16:20, 20 students in Study

68 1), Thursday mornings (9:00 to 10:30, 15 students in Study 1), and Friday mornings (9:00

to 10:30, 12 students in Study 1). The students in Group A were not assigned seats for

the duration of the class. Group B students attended an English Communication II class during the fall semester (October to February) of 2006 that met on Monday afternoons

(13:10 to 14:40, 31 students in Study 1; 30 students, 18 female and 12 male students, two mixed-sex pairs, in Study 2) and Thursday mornings (9:00 to 10:30,18 students in Study

1; 12 students, eight female and four male students in Study 2). The students in Group

B were assigned seats for the duration of the class. Students who had failed the classes previously were assigned to different sections but only one of these students, in the

Monday class for Group B, attended class consistently enough to be included in Study 1.

After the participants completed all the treatments, they completed a cloze test

(Appendix A) that was used to estimate the differences in English proficiency between the different classes. The acceptable-answer method (Brown, 1980) was used to score the test. Table 1 shows the descriptive statistics for the test.

Table 1 Descriptive Statistics for the Cloze Test Year Class M SD N 2005 Monday 9.10 4.94 31

Thursday 14.04 7.01 28

Friday 10.88 6.44 33

2006 Monday 11.78 4.69 40

Thursday 11.90 5.58 30

69 Each class met twice a week for 90 minutes and was taught by two teachers, a Japanese professor and a native English-speaking professor. However, there was no coordination between the teachers in regards to syllabus or assessment. The syllabuses for the Japanese professor for these classes are shown in Appendix B.

The main difference between the groups was that the Group A students attended these classes their first semester at the university and the Group B students attended these classes their second semester at the university. Group B attended an English class,

English Communication I, taught by Japanese professors during their first semester.

However, both groups were similar in that they were attending their first English class at the university taught by a native speaker of English.

A second difference between the two groups occurred in the assignment of the students to the different class sections. The Group A students were assigned according to ascending order of student numbers, and the Group B students were assigned to the section according to a placement test score (Monday class, M = 177, SD = 12.7, N = 34,

Thursday class, M = 203, SD = 4.6, N = 37).

The 96 participants (30 (32%) male and 63 (67%) female students) in Study 1 were mostly first-year Japanese university students. The general proficiency level of the students was beginning. They were an average age of 18.8 years of age and had studied

English six years in middle school and high school. Forty-six of the students had studied

English formally outside the classroom, usually in a cram school or conversation school, from one to eleven years. Twenty-seven (29%) had attended classes taught by a native

70 English teacher from a few days to eight years. Three of the students had lived in an

English-speaking country, two for short stays under one month, but one student had lived in the United States for two years and three months from the age of 3 months.

Twenty-two (23%) of the students had visited a foreign country, mostly in Asia and

Oceania, but also North America and Europe. Lastly, 22 (23%) of the students said that they liked studying English in high school, 33 (35%) said that they did not, and 39 (41%) said that they were not sure. Information from three students is missing.

The 42 participants (16 (38%) male and 26 (62%) female students) in Study 2 formed 21 pairs that participated in a consistent manner through all the production data collection sessions. All but two of the pairs were same-sex pairs. They were an average age of 19.2 years of age and had studied English for six years in middle school and high school. Twenty-eight (67%) of the students had studied English formally outside the classroom, usually in a cram or English conversation school from one to eleven years.

Sixteen (38%) had attended classes taught by a foreigner from one to six years. One student had lived in an English-speaking country for a one-month stay. Eleven (26%) of the students had visited a foreign country, mostly in Asia and Oceania, but also in North

America and Europe. Lastly, eight (19%) of the students said that they liked studying

English in high school, 13 (31%) said that they did not, and 21 (50%) said that they were not sure. All students in Study 2 came from Group B. Thirty students (18 female and 12 male students, and two mixed-sex pairs) were in the Monday class and twelve students

71 (eight female and four male students) were in the Thursday class. All the students gave

their written consent to participate in this study.

Research Setting

This study took place at a university that started as a national university, but with

recent governmental moves to increase privatization (Hara, 2005; Ministry of Education,

Culture, Science, and Technology, 2003) is now called a Kokuritsu Daigaku Hojin

(National University Corporation). Because of this change in status, more autonomy was given to the university in personnel and budgetary decisions. In addition, because the university was from its inception a national university, it has more prestige and attracts better students than many private universities. This university also boasts of a high rate of graduates who pass the difficult examinations to become secondary school teachers in

Japan (Shimizu, 2006).

The university is located about one hour inland north of Kobe in Hyogo

Prefecture. The town in which the university is located has a population of about 25,000, and the university is distant from the center of this town. The university is surrounded by rice fields, bamboo forests, and deciduous woods. The closest store is an hour by foot.

This being the case, most of the first year students live in dormitories located on campus.

Both the Group A students and Group B students met in the same classroom.

The desks included a set of earphones with a microphone attached, a cassette deck near the top of the desk, and a TV to show materials via an overhead camera or a computer.

72 During the treatment classes, Powerpoint slides were sometimes shown through the computer and class materials were often shown using the video OHP. The northern wall was a set of large picture windows facing out to a parking lot.

The Variables in This Study

Two independent variables were used in both Study 1 and Study 2. The first is the type of task, which is the activity that the students performed. Three types of tasks were used: descriptive, narrative, and decision-making tasks. The second independent variable is the level of topic choice for a designated task-type. The first level is no choice when the students engage in the task provided by the teacher. The second level of choice allows the students to choose from amongst three task topics pre-selected by the teacher. Finally, the third level of topic choice allows the students complete freedom to choose the topic within the confines of the task type. For the descriptive task, the students chose a place to describe. For the narrative task, the students told a personal story. For the decision- making task, one topic was selected from amongst a large set of student-suggested topics

(29 on a list or their own topic if not on the list) on current environmental problems and students decided upon a solution to the problem.

In Study 1, there are two dependent variables, Task Interest and Task Self-efficacy.

These variables are based on Deci and Ryan’s concept of intrinsic motivation (Deci &

Ryan, 1985, 2000; Ryan & Deci, 2000, 2002), which states that there are three components in intrinsic motivation; autonomy, competence, and relatedness. Autonomy, is a very

73 strong and vibrant component and in this study is believed to influence Task Interest.

Competence as well is a strong component of intrinsic motivation, and in this study is

called Task Self-efficacy. Relatedness is a weak component and, according to Deci and

Ryan (2000; Ryan & Deci, 2000, 2002), is not detected in some studies; thus relatedness is

not included in this study.

There are three dependent variables in Study 2, Accuracy, Complexity, and

Fluency. These variables are based on the research of Skehan (1998, 2001) and Skehan and

Foster (Foster & Skehan, 1996, 1999; Skehan & Foster, 1997, 1999, 2001).

Materials

Task Materials

Different pedagogical materials were developed for the descriptive task, narrative

task, and decision-making task and for building the students’ schema in preparation for

engaging in the tasks. The same tasks were used in both Study 1 and Study 2. To build

schema for the descriptive task, students completed activities that were designed to help

them build knowledge of expressing spatial relations and describing people. The lessons

for spatial relations were adapted from Asano, Uruno, and Rost (1985, p. 7) and from

Richards, Gordon, and Harper (1995, pp. 7-9, 14). The lessons for describing people were

adapted from Madden and Reinhart (1987, p. 5), Asano et al. (1985, p. 30), and Richards

et al. (1995, p. 14). In addition, a task similar to those to be utilized during the treatment

sessions was implemented to introduce the students to the design of the tasks.

74 To prepare for the narrative task, the students completed activities adapted from

Asano et al. (1985) and an activity modified from a picture story in Dumicich (1978, pp.

61-69) in which one student of the pair listened to the story and took notes. This student then told the story to his or her partner, who held the story frames in an incorrect order.

The goal of this task was for the second student to put the story frames in the correct order while listening to his or her partner. After this, the students completed tasks from

Maley, Duff, and Grellet (1980, p. 72) and Ur (1981, p. 63), which were similar to the tasks used in the treatment sessions where instead of one partner listening to a tape of the story and taking notes, one partner had a paper with the story in the correct order and then told that story to his or her partner.

To prepare for the decision-making task, the students gave their opinions about photographs taken from various sources. The goal of this task was for the students to express their opinions freely. Next, an activity modified from Martin (1997) required that the students decide which Hawaiian islands they would visit if they were to go to Hawaii together. The goal of this lesson was for the students to come to a mutual decision about an activity that they would do together.

Task Materials Used for the Treatment Sessions

For the descriptive task with no choice and limited choice of topic, students completed activities from Nicholson and Sakuno (1982). For the session with no choice of topic, the first round task (Figure 5) was from page 35 (Appendix C) and the second round

75 task was from page 9 (Appendix D) of the same book. For the session with limited choice of topic, tasks from Nicholson and Sakuno were selected. Student A of the pair could choose the activity on page 55 (Appendix E), page 39 (Appendix F), or page 63 (Appendix

G) for the first round task, and Student B of the pair could choose the activity on page 25

(Appendix H), page 33 (Appendix I), or page 57 (Appendix J) for the second round of the task. For both rounds of the session with complete choice of topic, the students described a place that they chose while their partner sketched it (Appendix K).

For the narrative task with no choice and limited choice of topic, students told picture stories from Heaton (1966). Student B had each frame of the story printed on separate sheets of paper but in an incorrect order. Student A of the pair had the complete story on a single sheet of paper in the correct order. Student B needed to put the frames of the story into the correct order by interacting with Student A. In the treatment with no choice of topic, the first round task was from pages 35 and 36 (Appendix L) and the second round task was to tell the story on pages 25 and 26 (Appendix M) of the same book. For the session with limited choice of task topic, using tasks again from Heaton,

Student A of the pair could choose the story on pages 27 and 28 (Appendix N), the story on pages 31 and 32 (Appendix O), or the story on pages 43 and 44 (Appendix P) for the first-round task and Student B of the pair could choose the story on pages 23 and 24

(Appendix Q), the story on pages 47 and 48 (Appendix R), or the story on pages 37 and

38 (Appendix S) for the second round of the task. For both rounds of the session with complete choice of topic, the students told a personal story that they chose while their

76 partner listened to the story and outlined it. The task paper used for this session is shown in Appendix T.

All the decision-making task activities were made originally for this study. The paper for the no choice of topic session is in Appendix U. The task papers for the limited choice of topic session are in Appendices V to AA. For the session with complete student choice of topic, topics were gathered over two homework sessions. The reason for these homework sessions was to narrow down the topics that the students engaged in during the treatment session but still give them a sense that they had complete control over topic selection. For the first session, the students suggested any topic that they wanted to discuss in class. For the second session, students were asked for topics dealing with current environmental problems because that was a common theme proffered by the students from the first homework session. Lastly, the students were given a paper with all submitted topics printed on it (Appendix BB) and they were asked to choose a topic and to write a half-page in English on the topic that they chose for homework. The following week this homework was collected and then the task (Appendix CC) was conducted. This was the only session in which the students prepared for the topic before the class started.

For neither the descriptive nor the narrative task with complete choice of topic could the students prepare for the topic outside of class before the treatment session.

77 After-task Survey

A 12-item after-task survey (Appendix DD) was administered each time that the students finished the task for each round. Some of the items were from published sources and some were originally written for this study. The survey was piloted in the spring of 2004.

First, a native speaker of Japanese translated items that were originally written in English into Japanese. Next, the surveys were administered in sessions that replicated the actual research design. During this time, following a suggestion by Abe (2001), the surveys were passed out where instead of categories for responding to the items as normally would be done after the task, five response categories eliciting the level of the difficulty of understanding the wording of the items were used. Some items were re- worded as a result of the feedback that was received. Items gauging the respondents’ feelings towards making a choice were removed to reduce the length of the survey. In addition, items eliciting a response towards the topic of the task were removed because such items would not be reliably consistent across all data collection sessions due to the number of topics involved.

The revised items were back-translated into English by a Japanese high school

English teacher who did not have access to the original questions. After a conference with the teacher, the items in Japanese and the instructions were modified and the format of the survey was finalized.

English translations of these items and their sources are shown in Table 2. Some of the survey items were written originally for this study, some were taken from original

78 Japanese research (Takashima, 2000), and some were garnered from sources in English

(Julkunen, 1989; Robinson 2001b). The questions from Japanese sources were also slightly

modified for this study.

Table 2 After-task Survey Items and Their Sources Item 1. I liked this task. (original item)

Item 2. I learned from this task. (Julkunen, 1989)

Item 3. I told my feelings to my partner while doing this task. (Takashima, 2000)

Item 4. I talked with my partner without undue silence. (Takashima, 2000)

Item 5. I cooperated with my partner while doing this task. (Takashima, 2000)

Item 6. I enjoyed doing this task. (original item)

Item 7. I want to do more tasks like this. (Robinson, 2001b)

Item 8. This task was difficult. (Julkunen, 1989)

Item 9. I used a lot of time doing this task. (Julkunen, 1989)

Item 10. I did the task to the best of my ability. (Julkunen, 1989)

Item 11. I was able to concentrate while doing this task. (Julkunen, 1989)

Item 12. I am satisfied with my performance doing this task. (Julkunen, 1989)

Response formats were also piloted. Although different levels of responses were

experimented with, a five-level response category was selected: 1 = mattaku so omowanai

(I do not think so at all); 2 = dochiraka to ieba so omowanai (If I were to say, I do not

79 think so); 3 = dochira tomo ienai (I can not say either way); 4 = dochira to ieba so omou (If

I had to say, I think so); 5 = sono toori dato omou (That is {exactly} what I think).

Procedures

In this section the preliminary procedures that were followed before the actual task was distributed are described. Similar preliminary steps were used in each data collection session. When a data collection session was about to begin, the cooperation of the students was requested. Then, the students were given a small piece of disinfectant cloth to sterilize and clean the headphones before putting them on. Next, the student pairs were given a cassette tape with their names and student numbers already affixed to it and asked to put the tape in a machine with a red dot facing up. Next, the teacher held up for all the class to see a task that they had done during the schema-building sessions that was similar to the one that they were about to commence. The teacher asked them if they remembered it, in order to bring the task to the forefront of their minds. Group

B students were asked not to use dictionaries while they were doing the task. These preliminary steps were similar to those before the second round with the small difference that the students were asked to turn the cassette tape over so a blue dot was facing up.

Figure 5 shows the data collection procedures that occurred after the above preliminary steps were completed. On the left of Figure 5 are the procedures for the no choice of topic treatment sessions and on the right are the procedures for the treatment sessions with limited and complete choice of topic. In the first round class sessions for the

80 XPSLĘPX 4UVEFOU"4UVEFOU# 4UVEFOU"4UVEFOU#

'*345306/% '*345306/%

(FU5BTL 4FMFDU5BTL

$PNQMFUF5BTL (FU5BTL

"ęFSUBTL4VSWFZ $PNQMFUF5BTL

4&$0/%306/% "ęFSUBTL4VSWFZ

1SPEVDUJPO (FU5BTL 4&$0/%306/% EBUB ĕSTUh $PNQMFUF5BTL 4FMFDU5BTL 1SPEVDUJPO ^ "ęFSUBTL4VSWFZ EBUB (FU5BTL ĕSTUh $PNQMFUF5BTL ^ "ęFSUBTL4VSWFZ 1SPDFEVSFTGPS 1SPDFEVSFTGPS-JNJUFEBOE /P$IPJDFPG5PQJD5SFBUNFOU4FTTJPOT $PNQMFUF$IPJDFPG5PQJD5SFBUNFOU4FTTJPOT 4VSWFZTVTFEGPS4UVEZBOBMZTJT

Figure 5. Graphical representation of the task and data collection procedures.

descriptive task with no choice of task, each Student A was given a folder containing the task. Student A was then asked to take the papers out of the folder, keep the paper with the complete information, and give their partner the papers with the missing pictures and answer sheet (descriptive task). The same process was followed for the narrative task. In the decision-making task, papers with the task printed on it were passed out and

81 each student wrote his or her name and student number at the top of the page. A short explanation of the task (or a reading of the instructions by the teacher in the case of the decision-making task) followed.

In the case of the data sessions with a limited choice of topic for the first round, a paper with the three task topics printed on it was distributed to Student A. An example from the first round of the descriptive task session is in Appendix EE, an example from the first round of the narrative task is in Appendix FF, and an example from the first round of the descriptive task is in Appendix GG. Student A then chose the topic and wrote his or her student number to the right of the task. After a short time, the teacher went down the rows and asked each Student A what task they had chosen and the teacher gave it to them in a large envelope. When all of the students received the task, they were asked to take it out of the folder, keep the paper with the complete information, and give their partners the papers with the missing pictures. The same process was followed for the narrative task.

In the case of the decision-making task, after asking which topic Student A wanted to do, papers with the task printed on them were given to that student and each student wrote his or her name and student number at the top of the page. A short, fifteen- second explanation of the task (or a reading of the instructions by the teacher in the case of the decision-making task) followed.

In the case of the complete choice of topic sessions of the descriptive task

(Appendix K) and the narrative task (Appendix T), a paper with the task instructions

82 printed on it was passed out to Student A, stapled together with a paper for Student B to sketch the place being described. Student A was asked to separate the papers and to give Student B the appropriate paper. Then, Student A was given about one minute to decide the topic and to write it on the paper. When the students seemed ready, a short explanation of the student roles followed.

In the case of the decision-making task with complete choice of topic, the students had already decided the topic and written a half-page about the topic. Before this task commenced, the homework (Appendix BB) was collected. Then the students were given a paper (Appendix CC) with all the topics and Student A was asked to circle the topic he or she chose. After this, the teacher read the instructions for the task. When the students were ready to commence the task, they were asked to put on the headphones.

When all the students were ready, the teacher started the recording machine.

Student A was asked to say his or her student number and then they started the task with a “Go” from the teacher. There was no time limit to complete the task, When they had finished the task, the students took off their headphones and the after-task survey was distributed. When the students had finished the surveys, the second round was initiated in the same sequence as the first round. After both rounds were finished, the materials were collected and a short break ensued if there was another task to do or the class was adjourned if it was the last task of the class session.

83 The Design of this Study

The data collection session are detailed in Table 3. The tasks for Group A for each class were presented serially, with the same task type implemented for three consecutive data sessions but with the choice sessions changing each time. This sequencing, both in the order of the task-type and the order of the level of choice, matches a 3 x 3 orthogonal latin square design shown in Fisher and Yates (1953, p. 72, Table 16). The tasks for Group B were implemented on a random schedule.

Table 3 Task Sequence for Groups A and B Group Aa Group Bb Monday Thursday Friday Monday Thursday 1A 5/09 2A 5/26 3A 5/27 1B 12/04 2A 12/07

1B 5/23 2B 6/02 3B 6/03 2C 12/11 1B 12/14

1C 5/30 2C 6/09 3C 6/10 1C 12/18 3B 12/21

2B 6/13 3B 6/16 1B 6/24 3C 12/18 1C 1/11

2C 6/20 3C 6/23 1C 7/01 2A 1/15 2B 1/11

2A 6/27 3A 7/07 1A 7/08 1A 1/15 2C 1/18

3C 7/04 1C 7/14 2C 7/15 2B 1/22 3A 1/18

3A 7/11 1A 7/21 2A 7/22 3A 1/22 3C 1/25

3B 7/11 1B 7/21 2B 7/22 3B 1/29 1A 1/25 Note. 1 = Descriptive task; 2 = Narrative task; 3= Decision-Making task; A = No topic choice; B = Limited topic choice; C = Complete topic choice. aData were gathered in May, June, and July, 2005. bData were gathered in December 2006 and January 2007.

84 Data Analyses

A 3 X 3 repeated-measures ANOVA was run to answer the research questions in both studies. Repeated-measures analyses have two main advantages. First, random variability is reduced and second, there is greater power to detect effects (Field, 2005).

This power occurs because the error-term in a repeated-measures analysis is frequently smaller than the error term for a corresponding randomized group design, which makes the overall ANOVA more powerful (Tabachnick & Fidell, 2007a, p. 243).

However, there is also an important statistical assumption associated with repeated-measures ANOVAs, sphericity. According to Field (2005), sphericity includes the assumption that “the variances of the differences between the data taken from the same participant are equal” (p. 745). That is, when scores for several different treatments are taken from the same participants, the scores are likely to be related because they come from the same participants. In this case, the conventional F-test will lack accuracy. A violation of sphericity means a loss of power and F-ratios that cannot be compared to the values in the F-distribution.

Sphericity is usually checked using Mauchly’s Test of Sphericity (ε). In this test, if ε is significant (p < .05), then the assumption of sphericity has been violated. There are two corrections available that use an adjustment to the degrees of freedom in case there are violations of sphericity. According to Girden (1992, p. 21), if the sphericity estimate (ε) is over .75, then the Huynh-Feldt correction (ε~) should be used. However, if the sphericity estimate is below .75, then the Greenhouse-Geisser correction (ε^) should be used.

85 Using the Rasch Model for Interval Scaling of the Variables

Based on the factor analysis results, the data were prepared for final statistical analysis. First, the items with the original item scores that comprised each factor were copied from the original file and pasted into a new SPSS file only for that variable (e.g.,

Task Interest) for that treatment (e.g., descriptive task with no choice of topic). For Task

Self-efficacy, a reverse coded version of Item 8 was used. Next, these files were prepared for use in Winsteps (Linacre, 2007), a computer program that analyzes data according to the principles developed by Georg Rasch (1960). This model of data analysis is called

Rasch analysis (or by some the one-parameter logistic item response model).

According to Yang and Kramer (2007), the Rasch model is a “mathematical probability model that allows for the investigation of dimensionality and the ordering of items on a measurement continuum” (p. 163). In this model, items and persons are measured on a common interval scale, and the estimates of persons and items are independent of one another, which makes for objective measurement (Yang & Kramer, 2007, p. 163).

In countering criticisms of Rasch analysis as being too simplistic, Bond and Fox

(2007) wrote a powerful retort:

The Rasch approach is simple, not simplistic: The aim is to develop fundamental measures that can be used across similar appropriate measurement situations, not merely to describe the data produced by administering Test a to Sample b on Day c. Rasch modeling addresses itself to estimating properties of persons and tests that go beyond the particular observations made during any testing situation. (p. 143)

86 According to Bond and Fox (2001), the Rasch model transforms raw data into equal- interval scales through log transformations of raw data odds and probabilistic equations

(p. 7). Through this, the measure of a person is separated from the scale to which he or she is measured. The person measure, called the ability estimate, is reported in logits.

The entire sample is placed on a logit scale, which is an interval scale where the interval between two and three, for example, is the same distance as that between three and four.

The Rasch model is often compared to classical test theory, which is a method of assessment with measurement data consisting of a true score and error. Classical test theory relies on the use of correlational estimates as a statistical procedure (Henning,

1987, p. 107). One weakness of classical test theory is that the estimates of item difficulty are dependent on the group that was measured (Hambleton, Swaminathan, & Rogers,

1991, p. 3). Therefore, the difficulty, reliability, and validity of the items are viable only for the particular group to whom they were administered. Henning (1987, pp. 108-116) listed several advantages of the Rasch measurement model over classical test theory, including sample-free item calibration, test-free person measurement, and item and person fit measures. Beck and Gable (2001, p. 8) added that, in comparison to classical test theory, the Rasch model allows ordinal data to be converted to interval data, predictions to be made about how a person will score on a given item, person ability and item difficulty estimates to be calculated.

There are also competing models within the item-response theory paradigm.

The single parameter model is the Rasch model that measures item difficulty. The two-

87 parameter model measures item difficulty and item discrimination. The three-parameter

model measures these two as well as the amount of guessing that may occur in the

responses to the items at the low end of the ability continuum (Hambleton et al., 1991,

p. 17). Henning (1987) suggested using the Rasch model in the language assessment field

because of, amongst others reasons, the ease of interpretation of the results and the low

number of participants needed for the model to be operative (p. 116).

Although Rasch originally proposed an equation for dichotomous responses,

Andrich (1978) developed a rating formula for ordered response categories, such as

polytomous, Likert-style, response categories. Prieto and Delgado (2007) explain the

equation for this rating scale model as log(Pnik/Pni(k-1) = Bn - Di - Fk), where Pnik is the probability that person n, on encountering item i would respond in category k, where

Pni(k-1) is the probability that the response would be in category k - 1, where Bn is the

ability of person n, where Di is the difficulty of item i, and where Fk is the impediment to

being measured in category k relative to k - 1, or simply, the kth category step calibration

(p. 151). In this last case, a series of thresholds between the response categories are created.

The threshold is the point where there is a likelihood of a lower response turning to a

higher response, and vice versa.

An important assumption of most Rasch models is that the data are

fundamentally unidimensional. One way in which this assumption is often checked is

through the inspection of Rasch fit indices. An item that does not fit the Rasch model

may not be measuring the primary trait (Yang & Kramer, 2007, p. 163).

88 According to Bond and Fox (2001, p. 77), items in a Likert response format are likely to vary in their difficulty, and the items are also unlikely to contribute to the construct equally (p. 85), an assumption of Likert scaling. In this case, data are regarded as ordinal data and the Rasch model is used to transform the ordered Likert categories into interval scales. Linacre (2004) suggested guidelines for the validation of rating scale categories. While not all guidelines are relevant for all data analyses (p. 276), the guidelines most important for verifying the validity of the rating scale for this study are (a) the outfit mean squares, (b) the frequency of the observations for each response category, and (c) minimum and maximum step distances between the categories.

Procedures for Calculating Production Data

The production data were calculated using the output of both students in the pair, rather than only one of the students. The first two usable minutes of the transcript were chosen for calculating the production data. Two minutes was selected because it was considered long enough to gather reliable data and because many pairs finished some tasks in less than three minutes.

In this study, the recording from the second round was used for the transcriptions because the students were more used to doing the tasks, there was less talk about how to do the tasks, and more talk was focused on the task itself, especially for the treatments with complete choice of task topic. Possible drawbacks from using the second round are

89 that some pairs did the task more quickly than during the first-round and they may have been less interested in doing the same task a second time.

The start of the transcript was marked when the students started to do the task in

English in a consistent way, that is, without a false start in commencing the task. Some students talked about the task (or other things) in Japanese and then started the task in

English while many pairs started the task immediately after the teacher gave the direction to start the task. The end of the transcript was the two-minute mark or when the students had completed the task.

Calculating Accuracy

Two methods were used when calculating Accuracy. The first method involved calculating the ratio of correct verb forms. This method was utilized by Ellis and Yuan

(2003) for assessing written production, by Yuan and Ellis (2003) for assessing oral production, and by Ellis and Yuan (2005) for comparing both types of production.

Verbs were checked for correctness in terms of tense, aspect, modality, and subject-verb agreement (Ellis & Yuan, 2005, p. 182). The following guidelines were used in calculating the ratio of correct verb forms (guidelines with an asterisk are shown in Appendix HH):

1. The verb had to be in a phrase.* 2. Non-existent but required verbs were not considered in the total.* 3. Uncorrected repetitions by the same person in the same turn were not considered in the total (i.e., counts as one verb).* 4. Progressives and perfects were counted wrong if an element (either the be-verb or the V-en/V-ing) was missing but as one verb if correct.* 5. Apostrophized verbs were considered.*

90 6. Auxiliary verbs, modal auxiliaries, and main verbs in the same clause were counted separately.* 7. Verb compliments, such as want (to V), start (to V), and begin (to V) were also counted as a verb and coded for correctness. 8. Unless there were special communication requirements, the descriptive task for all treatments of choice should be in the simple present tense, the narrative task for the no and limited treatments of choice should be in the present progressive tense, and the narrative task for the complete treatment of choice should be in the simple past tense. There were no special verb tense requirements for the decision-making task for all treatments of choice.

The second method involved calculating the ratio of error-free clauses in the transcript. This method was used by Foster and Skehan (1996, 1999) and by Skehan and Foster (1997, 1999).

Phrases were examined for correctness in terms of syntax, morphology, context, and lexical choices (Ellis & Yuan, 2005, p. 182). The following guidelines were used when assessing the ratio of correct phrases (guidelines with an asterisk are shown in Appendix II):

1. There had to be an English verb. Verbless clauses (Quirk, Greenbaum, Leech, & Svartvik, 1985, pp. 996-997) were not counted.* 2. Some Japanese nouns that were included in the word count were considered acceptable if they were in a correct phrase. 3. Relative clauses were separated and counted separately as correct or incorrect.* 4. Phrases connected by conjunctions were separated and counted right or wrong separately.* 5. The student was given the benefit of the doubt if there was an untranscribable segment in the phrase and if it would have made a correct phrase.* 6. The phrase was counted as incorrect if Japanese ((j) in the transcript) was used without English correction soon afterwards, except some Japanese words printed on the descriptive tasks that were integral to completing the task.* 7. Phrases that included repetitions were counted as correct if a correct form eventually was used.* 8. A phrase was counted as correct even if it required more than one turn to complete.* 9. Exact repetitions of one student by the other in the next turn were not counted.* 10. Other-initiated corrections were counted in the total.* 11. Imperatives and indirect speech were counted in the total.* 91 12. English phrases out of context, even if grammatically correct, were counted as incorrect.* 13. Improper nouns were counted as incorrect.* 14. In a clause utterance, if the proper words that made a correct clause were supplied in the correct order, no matter the surrounding words, that clause was counted as correct.* 15. One error per clause was counted. Multiple errors in the same clause were not added. 16. Clauses without articles in obligatory contexts were counted as incorrect but errors of definiteness or indefiniteness were not counted as incorrect if an article was supplied in an obligatory context. 17. “To-Infinitive” clauses (to-V) were not counted. 18. Exact repetitions or repetitions not substantially changed after a request for clarification or for a request in some other manner for a repetition were not counted.*

Two native speakers of English who are applied linguists examined approximately 20% of the transcripts for each treatment. The guidelines for calculating Accuracy were finalized as a result of extensive consultation.

Calculating Complexity

Three assessment methods were utilized for calculating Complexity. Two were from what Ellis and Barkhuizen (2005) label as interactional attributes of complexity, turns and words per turn, and another was what Ellis and Barkhuizen categorized as a lexical attribute of complexity, the type-token ratio.

The first method of calculating Complexity, counting turns, was used by Duff

(1986). van Lier (1988, p. 109) defined three categories of turns as (a) prospective, which is the way the turn is linked to subsequent turns, (b) retrospective, which is the way the turn is linked to preceding turns, and (c) concurrent, which consists of backchannels and listening responses.

92 In this study, a turn is an utterance that changes the proceeding conversation or actions by the partner in the pair. At times, a concurrent turn was counted as a turn if it changed the conversation, such as a signal of non-understanding (“Uh?”), which required the partner to respond. Also, a turn that was not interrupted verbally for the entire two minutes was counted as a single turn because it caused a change in the partner, either by causing the person to write, place a picture in the proper order, or, as proposed by van

Lier (1988, p. 106), by providing an intrinsic motivation for listening. In deciding turns, I listened to the entire two-minute segment using the following guidelines (guidelines with an asterisk are shown in Appendix JJ):

1. An utterance was not counted as a turn if Japanese was used for that entire utterance.* 2. An utterance was not counted as a turn if only hesitation or backchanneling, such as ‘um,’ ‘uhuh,’ ‘yea,’ or ‘yeah,’ or Japanese equivalents were used.* 3. Proper nouns constituted a turn.* 4. The words ‘OK’ and ‘Yes’ constituted a turn in themselves.* 5. Onomatopoeia, sound effects, or animal sounds were not counted as a turn in themselves.* 6. Incidents of reading English from the task (or Japanese words printed on the task paper integral to the completion of the task) constituted a turn. 7. Incidents of spelling interaction were counted as turns.*

The second method of assessing complexity was the type-token ratio, which is the total number of different words, or types, divided by the total word count. This is a very commonly used statistic, but it has one weakness in that it is influenced by text length: the shorter the text is, the higher the ratio is likely to be (e.g., Malvern & Richards, 1997;

Wolfe-Quintero, Inagaki, & Kim, 1998). The final figure depends on what is counted

93 as word as explained in the section on word count. However, this method of assessing output has a long history in both first and second or foreign language situations

(Johnson, 1944; Wolfe-Quintero et al. 1998). In addition, as Samuda (2001) has shown, low proficiency learners often overcome communication difficulties lexically rather than grammatically. Only when a form has been explicitly taught, in the case of Samuda’s research, will the learner start to utilize grammatical knowledge. Considering the proficiency level of the participants in this study, the use of the type-token ratio to gauge complexity was considered appropriate.

The third method of assessing complexity, words per turn, was also used by

Duff (1986). Words per turn was calculated by dividing the total word count for the two minutes of the transcript by the total number of turns.

Calculating Fluency

Ellis and Barkhuizen (2005) listed two categories of fluency measures, temporal variables and hesitation phenomena (p. 157). For the hesitation phenomena, unaltered repetitions were used in this study. This method was used by Skehan and Foster (Foster &

Skehan, 1996, 1999; Skehan & Foster, 1997, 1999, 2005). To calculate unaltered repetitions, the following guidelines were followed (guidelines with an asterisk are shown in

Appendix KK):

1. The repetition was the exact same word or phrase, including proper nouns, in the same turn.* 2. Repetitions separated by hesitations were not counted.*

94 3. False-starts, ‘yes,’ ‘yeah,’ ‘yea,’ ‘OK,’ ‘Oh,’ ‘no’, onomatopoeia, animal sounds, words such as ‘wow’, repetitions of the teacher’s directions (e.g., ‘Go, go, go’ and ‘English, English, English’ often used by the teacher for pep reasons), or words repeated at the end of the task to denote the finish of the task (e.g., ‘finish finish,’ or ‘thank you thank you’) were not counted.* 4. The occurrence of ‘very very’ was counted as a repetition. 5. The same word could be at the end of a phrasal repetition as well as the start of another.*

For the temporal phenomena, the total word count for the two-minute segment was used.

This method was also used by Duff (1986). To calculate this total, the appropriate segment of the transcript was copied into an Internet site (http://www.lextutor.ca/vp/eng/) that counts the number of English words, automatically rejecting unfinished words and

Japanese words, and provides a type count for the segment as well. The resulting printout from this website was examined carefully for any additional words (and types for type- token ratio analysis) that should be rejected or included into the count. Examples of this process are detailed in Appendices LL to TT. For this analysis, the following guidelines were applied.

1. English words were counted. 2. One, and only one, occurrence of a proper name per turn was included in the word count. If the proper name was a person’s name, both the first and last names were counted together as one proper noun for the turn. Country names that were in two or more words (e.g., ‘North Korea’) were counted as one word. If a proper noun was said in the English form (‘Nijo Castle’) and the Japanese form (‘Nijo-jo’) in the same turn, it was counted only once. 3. Words such as ‘Ok,’ ‘Oh,’ and ‘Uhuh’ were not counted. 4. Nonsense words, such as ‘wow,’ ‘yea,’ and ‘yeah,’ and animal sounds, such as ‘Bowwow,’ were not counted. 5. Japanese style miscommunication cues such as ‘e?’ and ‘m?’ were not counted. 6. Incomplete words from false starts were not counted, unless they made a complete English word. 7. Incidents of spelling or reading English from the task paper were not counted as words. 95 8. Apostrophized words, such as ‘I’ll,’ and ‘He’s’ were separated and counted as two words. Possessives and the word ‘Let’s’ were counted as a single word. 9. Names of chemical compounds, such as ‘CO2,’ were counted for each occurrence. 10. Some Japanese words that are common in current English were counted. These words were futon, geisha, judo, kabuki, karaoke, kimono, sushi, and ukiyoe. 11. Some English loan-words in Japanese were counted if they made sense in English. However, three phrases, ‘color box’ (for a small book case) ‘punch perm’ (for a certain hairstyle), and ‘salary man’ (businessman) were not counted. 12. Other words accepted were etcetera, hot carpet, MD, PC, permed, TV, and versus.

To further aid the reader, one example transcript from each of the nine different treatments is in Appendices KK to SS. Each transcript in the appendices is coded for the seven different assessments used in this study for Accuracy, Complexity, and Fluency detailed above.

Why Some Measures Were Not Used in This Study

Some measures that are commonly used for assessing output production were not used in this study. Ellis and Barkhuizen (2005), for example, stated that instead of counting words, counting syllables is common. However, considering the proficiency level of the students, most of the words were monosyllabic. Also, Skehan and Foster counted the number of c-units per transcript to measure complexity. Again, the low

English proficiency level of the students suggested that there would be so few c-units that other measures would be more reliable.

Lastly, many scholars engaged in output research measure fluency by calculating pauses and the amount of silence between utterances. However, there were two reasons

96 for not using pauses and the amount of silence as fluency measures for in this study. First,

Japanese anthropological researchers have claimed that silence is common in Japanese discourse and is not perceived negatively (e.g., Barnlund, 1989; Gudykunst & Nishida,

1994; Ishii & Bruneau, 1988; Mare, 1990; Yamada, 1997). For example, Lebra (1987) stated,

“I speculate that the prevalence of silence among the Japanese has to do with their awareness of individuals being interdependent and interconnected, which inhibits vocal self-assertion” (p. 354). Harumi (2002) stated that there may be four reasons for silence by Japanese students in English as a foreign language classes: (a) linguistic problems, (b) problems with time, (c) problems of confidence, and (d) problems with turn-taking (p.

31). For these reasons and the cultural tendency to accept sustained silence in discourse,

Japanese students can display many occurrences of silence in student conversations.

An additional reason for not using silence as a measure was the requirement that I balance research with teaching. Referring to the descriptions of the eighteen tasks used in the study, the students needed to write in many of the tasks, especially for the decision-making tasks and the complete choice of topic session for the descriptive and the narrative tasks. Writing may introduce silence that was not an indication of disfluency.

97 CHAPTER 4

PRELIMINARY RESULTS OF STUDY 1

In this chapter, the preliminary statistical results of Study 1 are presented. Study

1 is based on after-task survey data. The data from Student A for the first round (Figure

5) and from Student B for the second round were used. First, the data were checked for missing data, outliers, and skewed variables, and transformations were applied in the case of extremely skewed variables. Following this is an overview of the factor analyses and Rasch analyses that were used to determine the dimensionality of the after-task survey data and to construct interval measures of the participants’ responses to the survey items. The procedures used in this study to prepare the data for final analysis are shown in Appendix UU.

Missing Cases and Removing Univariate Outliers

In preparing the data for analysis, Tabachnick and Fidell (2007b) suggest cleaning up the data so as to meet the assumptions of multivariate data analysis (p. 60). The progression goes from replacing missing values, to examining the data for univariate outliers, to transforming the data, and then to examining the data for multivariate outliers (p. 91).

Before examining that data for outliers, nine missing values (0.09% of the total number of values) in the data set were replaced with the mean for the item. According

98 to Tabachnick and Fidell (2007b, p. 63), as long as the number of missing values is under

5% of the total number of values, any replacement strategy can be used. Next, the data

were examined for univariate outliers. For univariate outlying cases, a participant with

a z-score greater than +/- 3.29 for any item for any treatment was removed from the

following analyses. Ten univariate outliers were identified and removed, and when the

data were checked a second time, one more outlier was identified and removed. A third

round revealed no remaining univariate outliers. Eleven univariate outliers were removed

in all.

Descriptive Statistics

Next, descriptive statistics were calculated. Field (2005, p. 72) suggested using the z-score of the skewness to gauge the severity of an item’s skewness, which is calculated by dividing the skewness score by the standard error of the skewness. According to Field, a z-score greater than +/-1.96 indicates the skewness to be statistically significant at the p

< .05 level. The results of the descriptive analyses for the nine tasks (i.e., three task types with three levels of choice) are shown in Tables 4 to 12.

In the descriptive task with no choice of topic treatment (Table 4), four items

(Items 3, 4, 6 and 11) were negatively skewed at the p < .05 significance level. No statistically significant kurtosis was evident.

99 Table 4 Descriptive Statistics for the Descriptive Task with No Choice of Topic

M SE SD Skewness SES zskewness Kurtosis SEK Item 1 4.12 .08 .73 -.37 .26 -1.42 -.42 .52

Item 2 3.69 .08 .76 .07 .26 .26 -.48 .52

Item 3 3.91 .10 .96 -.97 .26 -3.73* .87 .52

Item 4 3.67 .12 1.13 -.64 .26 -2.46* -.31 .52

Item 5 4.28 .07 .61 -.23 .26 -.88 -.57 .52

Item 6 3.95 .09 .87 -.57 .26 -2.19* -.24 .52

Item 7 3.55 .11 1.01 -.36 .26 -1.38 -.38 .52

Item 8 3.28 .12 1.08 -.18 .26 -.69 -.65 .52

Item 9 3.51 .10 .96 -.02 .26 -.08 -.91 .52

Item 10 4.08 .07 .64 -.07 .26 -.26 -.51 .52

Item 11 4.26 .08 .71 -.83 .26 -3.19* .91 .52

Item 12 3.41 .11 1.03 -.36 .26 -1.38 -.45 .52 Note. N = 85; α = .80. *p < .05.

In the descriptive task with limited choice of topic treatment (Table 5) six items

(Items 1, 3, 5, 6, 10 and Item 11) were negatively skewed at the p < .05 significance level.

Items 4 and 12 were borderline leptokurtic at the p < .05 level.

In the descriptive task with complete choice of topic treatment (Table 6), five items (Items 1, 3, 5, 6 and 10) were negatively skewed at the p < .05 significance level. No significant kurtosis was evident.

100 Table 5 Descriptive Statistics for the Descriptive Task with Limited Choice of Topic

M SE SD Skewness SES zskewness Kurtosis SEK Item 1 4.22 .09 .85 -.93 .26 -3.58* .22 .52

Item 2 3.68 .10 .93 -.23 .26 -.88 -.34 .52

Item 3 3.69 .13 1.18 -.68 .26 -2.62* -.49 .52

Item 4 3.47 .14 1.32 -.49 .26 -1.88 -1.02 .52

Item 5 4.33 .07 .68 -.52 .26 -2.00* -.75 .52

Item 6 4.15 .09 .87 -.87 .26 -3.34* .17 .52

Item 7 3.74 0 .93 -.19 .26 -.73 -.84 .52

Item 8 3.49 .11 1.05 -.30 .26 -1.15 -.70 .52

Item 9 3.44 .11 1.04 -.12 .26 -.46 -.67 .52

Item 10 4.05 .10 .91 -.67 .26 -2.57* -.35 .52

Item 11 4.21 .08 .76 -.71 .26 -2.73* .17 .52

Item 12 3.31 .13 1.24 -.26 .26 -1.00 -1.02 .52 Note. N = 85; α = .87. *p < .05.

In the narrative task with no choice of topic treatment (Table 7), five items (Items

1, 2, 6, 8, and 11) were negatively skewed at the p < .05 significance level. Items 3, 4, and 12 were significantly leptokurtic at the p < .05 level.

In the narrative task with limited choice of topic treatment (Table 8), six items

(Item 1, 3, 5, 10, 11, and 12) were negatively skewed at the p < .05 significance level. Item 5 was significantly platykurtic at the p < .05 level.

101 Table 6 Descriptive Statistics for the Descriptive Task with Complete Choice of Topic

M SE SD Skewness SES zskewness Kurtosis SEK Item 1 4.01 .09 .79 -.61 .26 -2.34* .17 .52

Item 2 3.76 .10 .90 -.33 .26 -1.27 -.58 .52

Item 3 3.66 .13 1.17 -.58 .26 -2.23* -.74 .52

Item 4 3.42 .14 1.27 -.35 .26 -1.35 -.94 .52

Item 5 4.21 .08 .76 -.88 .26 -3.38* .83 .52

Item 6 3.93 .10 .94 -.75 .26 -2.88* .25 .52

Item 7 3.56 .11 1.05 -.43 .26 -1.65 -.33 .52

Item 8 3.65 .12 1.13 -.48 .26 -1.84 -.82 .52

Item 9 3.65 .11 1.03 -.24 .26 -.92 -.79 .52

Item 10 3.99 .10 .92 -.82 .26 -3.15* .48 .52

Item 11 4.15 .08 .72 -.43 .26 -1.65 -.24 .52

Item 12 3.33 .12 1.08 -.12 .26 -.46 -.79 .52 Note. N = 85; α = .84. *p < .05.

In the narrative task with complete choice of topic treatment (Table 9), three items (Items 1, 3 and 5) were negatively skewed at the p < .05 significance level. Item 4 was significantly leptokurtic at the p < .05 level.

In the decision-making task with no choice of topic treatment (Table 10), six items

(Items 1, 3, 5, 6, 9, and 11) were negatively skewed at the p < .05 significance level. No significant kurtosis was evident.

102 Table 7 Descriptive Statistics for the Narrative Task with No Choice of Topic

M SE SD Skewness SES zskewness Kurtosis SEK Item 1 3.94 .11 .98 -.97 .26 -3.73* .74 .52

Item 2 3.61 .12 1.10 -.43 .26 -1.65 -.56 .52

Item 3 3.42 .14 1.27 -.27 .26 -1.04 -1.16 .52

Item 4 3.24 .15 1.35 -.12 .26 -.46 -1.28 .52

Item 5 4.14 .08 .73 -.41 .26 -1.58 -.35 .52

Item 6 3.96 .10 .96 -.85 .26 -3.37* .67 .52

Item 7 3.54 .12 1.06 -.26 .26 -1.00 -.46 .52

Item 8 3.78 .11 1.03 -.95 .26 -3.65* .77 .52

Item 9 3.45 .10 .96 -.01 .26 -.04 -.92 .52

Item 10 3.89 .10 .91 -.46 .26 -1.77 -.57 .52

Item 11 4.14 .08 .73 -.60 .26 -2.31* .36 .52

Item 12 2.81 .14 1.25 .10 .26 .38 -1.16 .52 Note. N = 85; α = .85. *p < .05.

In the decision-making task with limited choice of topic treatment (Table 11), eight items (Items 1, 3, 5, 6, 8, 9, 10, and 11) were negatively skewed at the p < .05 significance level. No significant kurtosis was evident.

In the decision-making task with complete choice of topic treatment (Table 12), three items (Item 1, Item 8 and Item 10) were negatively skewed at the p < .05 significance level. No significant kurtosis was evident.

103 Table 8 Descriptive Statistics for the Narrative Task with Limited Choice of Topic

M SE SD Skewness SES zskewness Kurtosis SEK Item 1 4.18 .09 .79 -.62 .26 -2.38* -.25 .52

Item 2 3.72 .10 .89 -.43 .26 -1.65 .02 .52

Item 3 3.80 .11 1.02 -.75 .26 -2.88* .06 .52

Item 4 3.72 .12 1.08 -.53 .26 -2.04 -.74 .52

Item 5 4.27 .08 .70 -.86 .26 -3.31* 1.14 .52

Item 6 4.11 .08 .77 -.50 .26 -1.92 -.24 .52

Item 7 3.71 .11 .97 -.48 .26 -1.85 -.01 .52

Item 8 3.41 .10 .90 -.42 .26 -1.62 -.02 .52

Item 9 3.44 .10 .88 .09 .26 .35 -.65 .52

Item 10 4.16 .08 .77 -.61 .26 -2.35* -.08 .52

Item 11 4.27 .08 .73 -.85 .26 -3.27* .68 .52

Item 12 3.56 .11 .99 -.52 .26 -2.00* .05 .52 Note. N = 85; α = .86. *p < .05.

In a review of all the items that exhibited skewness, only Item 7 showed no skewness through all nine treatments. Item 1 was significantly skewed at the p < .05 level seven times. Item 2 was significantly skewed at the p < .05 level one time. Item 3 was significantly skewed at the p < .05 level six times. Item 4 was significantly skewed at the p

< .05 level one time. Item 5 was significantly skewed at the p < .05 level six times. Item 6 was significantly skewed at the p < .05 level six times. Item 8 was significantly skewed at the p < .05 level two times. Item 9 was significantly skewed at the p < .05 level two times.

104 Table 9 Descriptive Statistics for the Narrative Task with Complete Choice of Topic

M SE SD Skewness SES zskewness Kurtosis SEK Item 1 3.99 .09 .87 -.54 .26 -2.08* -.35 .52

Item 2 3.67 .10 .88 -.05 .26 -.19 -.72 .52

Item 3 3.68 .12 1.07 -.64 .26 -2.46* -.43 .52

Item 4 3.52 .13 1.23 -.34 .26 -1.31 -1.05 .52

Item 5 4.12 .08 .78 -.68 .26 -2.62* .24 .52

Item 6 3.89 .09 .79 -.41 .26 -1.58 -.09 .52

Item 7 3.41 .11 1.00 -.19 .26 -.73 -.24 .52

Item 8 3.71 .10 .92 -.40 .26 -1.54 -.60 .52

Item 9 3.64 .10 .94 -.09 .26 -.35 -.86 .52

Item 10 4.12 .07 .68 -.13 .26 -.58 -.80 .52

Item 11 4.29 .06 .57 -.10 .26 -.38 -.53 .52

Item 12 3.13 .11 1.03 -.20 .26 -.77 -.87 .52 Note. N = 85; α = .81. *p < .05.

Item 10 was significantly skewed at the p < .05 level four times. Item 11 was significantly skewed at the p < .05 level seven times. Item 12 was significantly skewed at the p < .05 level one time. top

Out of 12 items, the descriptive task with no choice of topic treatment survey had four skewed items, the descriptive task with limited choice of topic treatment survey had five skewed items, the descriptive task with complete choice of topic treatment survey

105 Table 10 Descriptive Statistics for the Decision-Making Task with No Choice of Topic

M SE SD Skewness SES zskewness Kurtosis SEK Item 1 3.96 .10 .94 -.71 .26 -2.73* .10 .52

Item 2 3.74 .09 .85 -.07 .26 -.27 -.69 .52

Item 3 3.75 .11 1.03 -.61 .26 -2.35* -.21 .52

Item 4 3.52 .12 1.15 -.36 .26 -1.38 -.88 .52

Item 5 4.12 .08 .76 -.70 .26 -2.69* .44 .52

Item 6 3.85 .11 .98 -.69 .26 -2.65* .23 .52

Item 7 3.51 .11 .97 -.14 .26 -.54 -.60 .52

Item 8 3.74 .11 .99 -.21 .26 -.81 -1.01 .52

Item 9 3.92 .10 .95 -.59 .26 -2.27* -.51 .52

Item 10 4.05 .09 .83 -.47 .26 -1.81 -.48 .52

Item 11 4.14 .09 .79 -.56 .26 -2.15* -.33 .52

Item 12 3.38 .12 1.09 -.24 .26 -.92 -.59 .52 Note. N = 85; α = .88. *p < .05.

had five skewed items, the narrative task with no choice of topic treatment survey had five skewed items, the narrative task with limited choice of topic treatment survey had six skewed items, the narrative task with complete choice of topic treatment survey had three skewed items, the decision-making task with no choice of topic treatment survey had six skewed items, the decision-making task with limited choice of topic treatment survey had

106 Table 11 Descriptive Statistics for the Decision-Making Task with Limited Choice of Topic

M SE SD Skewness SES zskewness Kurtosis SEK Item 1 4.13 .11 1.00 -1.08 .26 -4.15* .48 .52

Item 2 3.75 .10 .90 -.09 .26 -.35 -.86 .52

Item 3 3.68 .12 1.14 -.54 .26 -2.08* -.78 .52

Item 4 3.48 .12 1.13 -.34 .26 -1.31 -.84 .52

Item 5 4.27 .08 .71 -.65 .26 -2.50* -.02 .52

Item 6 4.00 .10 .94 -.80 .26 -3.08* .28 .52

Item 7 3.62 .11 1.01 -.38 .26 -1.46 -.35 .52

Item 8 3.61 .11 1.02 -.58 .26 -2.23* -.05 .52

Item 9 3.66 .11 1.05 -.59 .26 -2.27* -.15 .52

Item 10 4.12 .08 .78 -.52 .26 -2.00* -.27 .52

Item 11 4.20 .08 .74 -.70 .26 -2.69* .38 .52

Item 12 3.49 .13 1.18 -.47 .26 -1.81 -.72 .52 Note. N = 85; α = .88. *p < .05.

eight skewed items, and the decision-making task with limited choice of topic treatment survey had three skewed items.

Data Transformation

Next, transformation of the skewed data was commenced. According to

Tabachnick and Fidell (2007b, p. 89), if the data are moderately negatively skewed, the

107 Table 12 Descriptive Statistics for the Decision-Making Task with Complete Choice of Topic

M SE SD Skewness SES zskewness Kurtosis SEK Item 1 3.73 .10 .93 -.52 .26 -2.00* -.09 .52

Item 2 3.69 .09 .87 -.45 .26 -1.73 -.36 .52

Item 3 3.44 .12 1.11 -.45 .26 -1.73 -.43 .52

Item 4 3.28 .13 1.17 -.25 .26 -.96 -.85 .52

Item 5 3.95 .09 .80 -.48 .26 -1.85 -.07 .52

Item 6 3.73 .10 .92 -.38 .26 -1.46 -.60 .52

Item 7 3.35 .12 1.08 -.05 .26 -.19 -.67 .52

Item 8 3.81 .11 1.05 -.55 .26 -2.31* -.60 .52

Item 9 3.59 .10 .95 -.26 .26 -1.00 -.46 .52

Item 10 3.93 .09 .84 -.60 .26 -2.31* -.01 .52

Item 11 3.98 .09 .82 -.49 .26 -1.88 -.18 .52

Item 12 2.96 .12 1.14 -.18 .26 -.69 -.94 .52 Note. N = 85; α = .85. *p < .05.

SPSS square root transformation (NEW ITEM VALUE = SQRT(6 - OLD ITEM VALUE)) should be used and if the data are substantially negatively skewed, the SPSS log10 transformation should be (NEW ITEM VALUE = LG10(6 - OLD ITEM VALUE)). However, to keep as much of the original data as possible, a transformed set of data was chosen where skewness was less than z = +/-1.96, even if a more powerful transformation would transform the data closer to z = 0.

108 After an examination of the transformed data using the SPSS square root

transformation, two items needed further transformation using the SPSS log10

transformation. Both of these items were Item 1 for the descriptive and decision-making

tasks with limited choice of topic treatment. All other skewed items were transformed

using the square root transformation.

Removing Multivariate Outliers

Next, Mahalanobis distance was calculated to identify multivariate outliers with

the item total for each case as the dependent variable. The examination for multivariate

outliers was conducted twice, the second time after the outliers from the first round

were removed. A third round revealed no further multivariate outliers. Participants over

the maximum level of χ2(11, N = 85) = 32.91 p = .001 for the first round or χ2(11, N = 80)

= 32.91 p = .001 for the second round were removed from the analysis for all treatments.

Seven outliers were removed in all. After the univariate and the multivariate outliers were removed, 78 students remained for Study 1. The details of the removal of multivariate outliers is in Table 13.

Factor Analysis of the Data

The next step was to extract factors (using the transformed data) in order to reduce the number of variables used in the final analyses and thereby decrease the probability of making a Type I error. A two-factor solution was requested because it was

109 Table 13 Multivariate Outliers Removed From the Analysis Treatment Group/Gender χ2 DT-LC B/Male 37.11

B/Male 34.30

B/Female 36.39

DT-CC A/Female 44.24

NT-LC B/Male 34.66

NT-LC A/Female 35.55

NT-CC B/Male 35.58 Note. DT = Descriptive task; NT = Narrative task; LC = Limited topic choice; CC = Complete topic choice.

hypothesized that the survey measured two factors, Task Interest and Task Self-efficacy.

After using various extraction methods with a combination of rotations, the combination

that yielded the most consistent results across all the treatments was the maximum

likelihood extraction method using a varimax rotation. An item was considered to be

included in a factor (for all treatments) when it loaded on that factor in five or more out

of the nine treatments.

However, some sacrifices to consistency were needed. In some cases, especially for

the decision-making task treatments for no choice of topic and limited choice of topic,

some items loaded on the opposite factor with much greater Eigenvalues than the factor

it was put into. In addition, although Kline (1994, p. 6) suggested a cut-off point of .30 for

the factor loading and Tabachnick and Fidell (2007b, p. 649) suggested that a loading less

110 than .32 not be considered for interpretation, in this study there were some cases where an item had a very low loading value for the construct that it was hypothesized to measure.

This was especially true of Item 9, which was expected to load on Factor 1.

Table 14 The Constitution of the Dependent Variables Task Interest (Factor 1) Task Self-efficacy (Factor 2) Item 1 (in 8 of 9 treatments) Item 3 (6 of 9)

Item 2 (6 of 9) Item 4 (6 of 9)

Item 5 (8 of 9) Item 8 (7 of 9)

Item 6 (9 of 9) Item 12 (6 of 9)

Item 7 (9 of 9)

Item 9 (5 of 9)

Item 10 (7 of 9)

Item 11 (6 of 9)

Be that as it may, an examination of the items for each factor revealed that Factor

1 was the Task Interest factor and Factor 2 was the Task Self-efficacy factor. The items that constituted each variable and the number of times that the item loaded on the factor above are listed in Table 14. Tables 15 to 23 show the Eigenvalues, the percentages of variance accounted for, and the communalities for the factor analyses for each variable.

Variables are ordered by size of loading for ease of interpretation. The communalities for the items ranged between .07 and .95.

111 Table 15 Factor Loadings for the Descriptive Task with No Choice of Topic Factor Loading Item 1 2 h2 Item 6 -.83 .22 .74

Item 7 .79 .01 .63

Item 1 .72 -.25 .58

Item 11 -.72 .05 .52

Item 10 .70 -.22 .53

Item 2 .68 .12 .47

Item 5 .63 -.35 .51

Item 9 .24 .34 .18

Item 4 -.28 .92 .93

Item 3 -.35 .77 .71

Item 8 .05 .56 .32

Item 12 .26 -.50 .32

Eigenvalues 4.01 2.42

% of variance 33.41 20.13

Cumulative % 33.41 53.54

For all its usefulness, factor analysis has problems. Wright (1996, p. 10) listed several problems with factor analysis. For one, as with all analyses, raw data are not linear so the factor scores are nonlinear. For another, there is no solid objective method to know when to stop factoring and the residuals that remain after each factor is extracted are

112 Table 16 Factor Loadings for the Descriptive Task with Limited Choice of Topic Factor Loading Item 1 2 h2 Item 2 -.82 -.11 .68

Item 11 .79 .20 .66

Item 6 .71 .46 .72

Item 10 .71 .42 .69

Item 1 .69 .44 .67

Item 9 -.62 -.04 .38

Item 5 .61 .46 .58

Item 7 -.60 -.34 .48

Item 4 -.16 -.89 .81

Item 3 .21 .79 .67

Item 12 -.41 -.72 .68

Item 8 -.41 .38 .31

Eigenvalues 4.30 3.04

% of variance 35.86 25.32

Cumulative % 35.86 61.18

part of the data for the smaller factors, a problem that may cause smaller factors in the data to go unnoticed. According to Wright, smaller factors are “adrift in the turbulence left behind by the preceding factors” (p. 10). However, most problematic is the fact that the same items will seldom reproduce similar factor loadings and sizes when a different

113 Table 17 Factor Loadings for the Descriptive Task with Complete Choice of Topic Factor Loading Item 1 2 h2 Item 1 -.83 .35 .81

Item 6 -.82 .36 .81

Item 7 .80 -.30 .73

Item 11 .69 -.11 .49

Item 10 -.65 .30 .52

Item 2 .63 -.07 .40

Item 9 .43 -.08 .19

Item 5 -.42 .57 .50

Item 3 -.16 .96 .95

Item 4 .17 -.87 .78

Item 12 .44 -.46 .41

Item 8 .34 .27 .19

Eigenvalues 4.03 2.74

% of variance 33.56 22.86

Cumulative % 33.56 56.42

sample is used. The result of this is that researchers are forced to come to nominal conclusions from the numerical instability. However, factor analysis is still a viable option readily available and commonly understood by researchers as a method to reduce the chance for committing a Type I error by reducing a set of items to a common factor.

114 Table 18 Factor Loadings for the Narrative Task with No Choice of Topic Factor Loading Item 1 2 h2 Item 1 .84 -.23 .76

Item 7 -.82 .13 .69

Item 2 .78 .00 .61

Item 6 .77 -.43 .77

Item 5 -.56 .43 .49

Item 11 .50 -.10 .26

Item 10 -.49 .22 .29

Item 9 -.26 .02 .07

Item 4 -.17 .90 .84

Item 3 -.17 .85 .76

Item 12 -.28 .61 .45

Item 8 .00 .36 .13

Eigenvalues 3.59 2.53

% of variance 29.93 21.06

Cumulative % 29.93 51.00

Review of the Data Using Rasch Analysis

In a check for misfitting items (Infit Mean Square > 2.00), Item 9 (of the Task

Interest factor) misfit twice, for the descriptive task with no choice of topic (Infit Mean

Square = 2.28) and for the narrative task with limited choice of topic (Infit Mean Square =

115 Table 19 Factor Loadings for the Narrative Task with Limited Choice of Topic Factor Loading Item 1 2 h2 Item 1 .83 .30 .77

Item 6 -.81 -.40 .81

Item 7 -.81 -.11 .67

Item 11 .75 .11 .58

Item 2 -.73 .83 .56

Item 10 .65 .28 .50

Item 5 .64 .42 .58

Item 9 -.33 .06 .11

Item 3 .38 .87 .90

Item 4 -.34 -.83 .81

Item 12 .39 .44 .34

Item 8 -.10 .26 .08

Eigenvalues 4.45 2.26

% of variance 37.10 18.84

Cumulative % 37.10 55.94

2.02). This item was removed (with the IDFILE command (Linacre, 2007, pp. 110-111)) from these two treatments and the Winsteps analysis was run again and new person measures were produced.

116 Table 20 Factor Loadings for the Narrative Task with Complete Choice of Topic Factor Loading Item 1 2 h2 Item 1 -.60 -.39 .51

Item 5 -.59 -.31 .45

Item 7 .52 .44 .47

Item 2 .33 .61 .48

Item 10 .24 .80 .70

Item 11 .23 .78 .66

Item 9 .18 .69 .50

Item 8 -.16 .43 .21

Item 4 .69 .07 .48

Item 3 -.86 .04 .74

Item 12 .70 .04 .49

Eigenvalues 3.33 2.86 .32

% of variance 27.72 23.82

Cumulative % 27.72 51.53

The response categories on the after-task survey, from 1 (I do not think so at all) to

5 (That is {exactly} what I think) were examined with Winsteps. As written in Chapter 3, the guidelines most important for verifying the validity of the rating scale for this study are: (a) the outfit mean squares, (b) the frequency of the observations for each response category, and (c) minimum and maximum step distances between the categories. As

117 Table 21 Factor Loadings for the Decision-Making Task with No Choice of Topic Factor Loading Item 1 2 h2 Item 5 .73 .31 .63

Item 6 .73 .35 .65

Item 1 .65 .37 .57

Item 2 -.62 -.47 .60

Item 7 -.62 -.43 .57

Item 10 -.48 -.66 .66

Item 11 .47 .67 .67

Item 9 .15 .73 .56

Item 8 .26 -.62 .45

Item 4 -.88 .08 .79

Item 3 .89 .04 .79

Item 12 -.74 -.03 .55

Eigenvalues 4.92 2.57

% of variance 40.96 21.42

Cumulative % 40.96 62.38

in the factor analysis of all the surveys, it would have been impossible to expect that all categories for all treatments for both dependent variables to meet all of the guidelines; therefore, if, as in the factor analysis, more than half the surveys (i.e., five or more of the nine surveys) for each of the two dependent variables met the minimum of each

118 Table 22 Factor Loadings for the Decision-Making Task with Limited Choice of Topic Factor Loading Item 1 2 h2 Item 7 -.67 -.52 .73

Item 5 .66 .31 .53

Item 6 .65 .58 .75

Item 1 .59 .60 .71

Item 10 .49 .73 .77

Item 11 .46 .63 .61

Item 2 -.44 -.61 .57

Item 9 .18 .69 .51

Item 8 -.09 .60 .36

Item 12 -.49 -.38 .38

Item 3 .84 .23 .77

Item 4 -.79 .11 .63

Eigenvalues 3.91 3.41

% of variance 32.61 28.39

Cumulative % 32.61 61.00

guideline, it was considered sufficient to continue the analysis without modification to the data.

For the first guideline, Linacre (2004) suggested that an outfit mean square of less that 2.0 is required. The outfit mean square statistic is more sensitive to unexpected

119 Table 23 Factor Loadings for the Decision-Making Task with Complete Choice of Topic Factor Loading Item 1 2 h2 Item 1 -.87 -.09 .76

Item 6 .81 .28 .73

Item 2 .79 .20 .67

Item 7 .75 .27 .64

Item 11 .65 .15 .45

Item 10 -.59 -.20 .39

Item 9 .53 -.12 .30

Item 5 .51 .36 .39

Item 4 .11 .88 .79

Item 3 .17 .86 .77

Item 12 .30 .52 .36

Item 8 .03 .43 .18

Eigenvalues 4.05 2.38

% of variance 33.72 19.82

Cumulative % 33.72 53.55

responses than the infit mean square statistic. An outfit mean higher than 2.0 would, according to Linacre, indicate that there is too much randomness in the participants’ response patterns. There were no occurrences where the outfit mean square statistic was greater than 2.0.

120 For the second guideline, Linacre suggested that there be at least ten observations in a response category. Low category frequency may cause imprecise step calibrations.

For the Task Interest variable, there was only one survey (the narrative task with no choice of topic) that had more than ten responses for all categories. All occurrences of less than ten responses were for response category 1 (I do not think so at all). There were no responses for this category at all for the complete choice of topic treatment for the narrative task survey. Therefore, the lowest category for this survey was response 2. In this case, Linacre terms this as an incidental zero, a zero that occurs particular to that data set, and suggests treating this as a structural zero, a category of the rating scale that will never be observed (p. 265).

Lastly, Linacre suggested that no category should be more than five logits distant from a neighboring category, but that there be at least one logit of a step distance between the categories for a rating scale with five categories. For this data set, there were no occurrences where a category was more than five logits apart from the neighboring category. However, the guideline that there be a minimum of one logit between categories was more difficult to meet. In this case, a category becomes subsumed under categories next to it when this criterion is not met. An example of this is in Figure 6, where category

3 is barely discernable.

In this study, there were five occurrences, all for the Task Self-efficacy dependent variable, where the step distance between a category was less than one logit. These were all for response category 3. Even though the step distance was low for these five

121 Figure 6. Category probabilities from Rasch analysis.

cases, there was no case where the response category was completely subsumed under neighboring categories. The surveys in which these small step distances occurred were the no choice of topic treatment for the narrative task, the limited choice of topic treatments for the descriptive (Figure 6) and decision-making tasks, and the complete choice of topic treatments for the descriptive and narrative tasks.

Because more than half the surveys for the Task Self-efficacy dependent variable did not meet the third guideline, categories were collapsed so that the problematic category can be included in either the category above or the category below, resulting in four categories instead of five. However, whether to collapse category 3 into category

2 or category 4 is not arbitrary. Bond and Fox (2007, pp. 230-231) suggest using person and item separation statistics as well as fit statistics in order to decide which direction

122 category 3 should be collapsed. To do this, the data in each of the nine surveys were re-coded twice in order to compare the above-mentioned statistics of the re-coded data with category 3 collapsed into category 2 with another set of re-coded data with category

3 collapsed into category 4. A comparison of the separation statistics after this re-coding revealed that category 3 should be collapsed into category 2 for all surveys. This was completed and further analyses were run using a 4-category scale for this dependent variable.

In the final step using Winsteps to prepare the data for analysis, the person ability logit score produced by Winsteps was transformed into a measure where the lowest possible number is zero and the highest possible score is 100 (Linacre, 2007, pp. 352-354).

These new person scores were then pasted into a single new SPSS file containing both the variables for all treatments

123 CHAPTER 5

RESULTS OF STUDY 1

Research Question 1

The first research question concerned the effect of choice on Task Interest.

Research Question 1 was: To what degree does the level of Task Interest change across three levels of choice (i.e., no choice, limited choice, and complete choice)? This question will be answered by showing the descriptive statistics for each treatment, the results of the repeated-measures ANOVA, profile plots from the analysis, pairwise comparisons with a Bonferroni correction, and t-tests showing significant contrasts between the

individual treatments.

Tables 24 to 26 show the descriptive statistics for each of the levels of choice

according to the different types of tasks. The descriptive statistics in Tables 24 to 26 are

based on Rasch person measures. The descriptive statistics reveal that the means for the

descriptive and narrative tasks were higher for the limited choice of task topic treatments

(M = 70.09 and M = 69.19, respectively) than for the no choice of task topic treatment (M

= 66.44 and M = 61.28, respectively). The means for the descriptive and narrative tasks,

however, were lower for the complete choice of topic treatment (M = 66.35 and M = 59.19,

respectively). The means for the decision-making task for the no M( = 66.14) and limited

choice (M = 66.86) of topic treatments were similar and the complete choice (M = 61.96)

of topic treatment was lowest.

124 The Rasch Item Reliability (RIR) estimate is analogous to the KR-20 or the

Cronbach alpha reliability coefficient for assessing reliability, but, according to Linacre

(1997), it is more conservative and less misleading. A low estimate may indicate that a

larger sample from the population is needed.

Table 24 Descriptive Statistics for Task Interest for the No Choice of Topic DT NT DMT M 66.44 61.28 66.14

SE 1.81 1.74 1.88

95% CI Low 62.84 57.81 62.41 High 70.03 64.75 69.88 SD 15.94 15.39 16.56

Minimum 42 32 32

Maximum 100 100 100

Skewness .71 .98 .68

SES .27 .27 .27

Kurtosis -.27 .78 -.17

SEK .54 .54 .54 Rasch Item .94 .89 .86 Reliability N 78 78 78 Notes. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

125 Table 25 Descriptive Statistics for Task Interest for the Limited Choice of Topic DT NT DMT M 70.09 69.19 66.86

SE 1.69 2.02 1.99

95% CI Low 66.73 65.17 62.89 High 73.45 73.21 70.83 SD 14.92 17.82 17.59

Minimum 37 32 32

Maximum 100 100 100

Skewness .43 .26 .41

SES .27 .27 .27

Kurtosis -.06 -.77 -.50

SEK .54 .54 .54 Rasch Item .95 .95 .92 Reliability N 78 78 78 Notes. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

Each new variable compiled from each factor for each treatment was applied to a repeated-measures ANOVA analysis. The results from the analysis for the Task

Interest variable are shown in Table 27. Mauchly’s test indicated that the assumption of sphericity had not been violated for the main effect of Task (ε = .93), the main effect of Choice (ε = .94), nor for the interaction effect between Task and Choice (ε = .82).

126 Table 26 Descriptive Statistics for Task Interest for the Complete Choice of Topic DT NT DMT M 66.35 59.19 61.96

SE 1.74 1.96 1.75

95% CI Low 62.88 55.29 58.48 High 69.81 63.09 65.44 SD 15.35 17.29 15.44

Minimum 36 32 36

Maximum 100 100 100

Skewness .71 1.09 .99

SES .27 .27 .27

Kurtosis .19 .65 .66

SEK .54 .54 .54 Rasch Item .88 .93 .85 Reliability N 78 78 78 Notes. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

Therefore, no correction of degrees of freedom for sphericity was needed. Correcting for

the number of comparisons in Study 1, interactions at the p < .025 (.05/2) significance

level were considered statistically significant. Indicators of effect size 2(η ) and observed power (β) are also provided. According to Tabachnick and Fidell (2007b, p. 54), effect size represents the degree to which the independent variables and the dependent variables

127 Table 27 Repeated-Measures ANOVA Results for Task Interest Source df SS MS F η2 β Task 2 2296.52 1148.26 10.72* .12 .99

Error (Task) 154 16494.37 107.11

Choice 2 4669.37 2334.68 23.99* .24 .99

Error (Choice) 154 14989.52 97.34

Task x Choice 4 1473.93 368.48 4.13* .05 .92 Error 308 27469.85 89.19 (Task x Choice) Notes. η2 = effect size (partial eta squared). β = observed power. *p < .025.

are related. According to these authors (p. 55), an effect size of 2η = .01 is small, an effect

size of η2 = .09 is medium, and an effect size of 2η = .25 is large. The observed power

takes the given data as representative of the population and estimates the proportion of

times a significant result would be obtained in a random sample of this size. In this case,

observed power of .90 or better is desired.

There was a significant main effect for Choice, F(2, 154) = 23.99 (Table 27), and a

significant interaction effect between Task and Choice, F(4, 308) = 4.13. This indicates that

Choice had different effects on the ratings of Task Interest depending on the type of task.

For all three tests, the observed power is quite high, suggesting that the probability of

avoiding a Type I error is high. The effect size for Choice is also large (.24). This indicates

that Task Interest and Choice are strongly related. A medium effect size was found for the

main effect of Task and the interaction effect between Task and Choice.

128 ţŞ

ţŜ

ŢŤ

ŢŢ -FWFMPG$IPJDF ŢŠ /POF

ŢŞ -JNJUFE

ŢŜ $PNQMFUF 3BTDI.FBTVSFPG5BTL*OUFSFTU šŤ %FTDSJQUJWF/BSSBUJWF%FDJTJPO .BLJOH 5BTLUZQF

Figure 7. The degree of Task Interest for each level of choice.

Examining the profile plot for the level of choice of task topic (Figure 7), the limited level of choice showed the highest level for all task types. However, interest to engage in the task was much lower for all task types when the complete choice of topic was introduced. In fact, these levels of Task Interest were much lower than those for the no choice of task topic levels for the narrative and decision-making tasks. The no choice of topic level maintained a level between the two other levels of choice, but the participants showed a high degree of interest on the decision-making task for this level of choice.

To break down this interaction, pairwise comparisons with a Bonferroni correction were requested. In this comparison, the limited level of topic choice was significantly higher than both the no choice and the complete choice of topic treatments

129 (p < .05). However, there was no statistically significant difference between the no choice

of topic and complete choice of topic treatments.

Within-samples t-tests between pairs of treatments were conducted in order

to determine where specific differences were. These tests were conducted in sets of

three, with the same type of task but with different levels of choice. To protect against

committing a Type I error, a Bonferroni correction with a probability level of p < .017 (α =

.05/3) was used. For the descriptive task with different levels of choice, the limited choice

of topic was significantly higher than the no choice of topic M( diff = 3.65, SD = 11.71, t(77)

= 2.76) and the complete choice of topic (Mdiff = 3.74, SD = 10.48, t(77) = 3.16). The same pattern held for the narrative task as well: the limited choice of topic was significantly higher than the no choice of topic (Mdiff = 7.91, SD = 12.24, t(77) = 5.71) and the complete

choice of topic (Mdiff = 10.00, SD = 12.58, t(77) = 7.02). For the decision-making task, the limited choice of topic was significantly higher than the complete choice of topic M( diff =

4.90, SD = 17.56, t(77) = 2.46) and the no choice of topic was significantly higher than the

complete choice of topic (Mdiff = 4.18, SD = 14.53, t(77) = 2.54).

Research Question 2

The second research question concerned the effect of task type on Task Interest.

Research Question 2 was: To what degree does the level of Task Interest change between the three types of tasks (i.e., the descriptive, narrative, and decision-making tasks)?

This question will be answered by showing the descriptive statistics for each treatment,

130 Table 28 Descriptive Statistics for Task Interest for the Descriptive Task DT NT DMT M 66.44 70.09 66.35

SE 1.81 1.69 1.74

95% CI Low 62.84 66.73 62.88 High 70.03 73.45 69.81 SD 15.94 14.92 15.35

Minimum 42 37 36

Maximum 100 100 100

Skewness .71 .43 .71

SES .27 .27 .27

Kurtosis -.27 -.06 .19

SEK .54 .54 .54 Rasch Item .94 .95 .88 Reliability N 78 78 78 Notes. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

the results of the repeated-measures ANOVA, profile plots from the analysis, pairwise

comparisons with a Bonferroni correction, and t-tests showing significant contrasts between the individual treatments.

Tables 28 to 30 show the descriptive statistics for each of the tasks according to the level of choice. The descriptive statistics in Tables 28 to 30 are based on Rasch

131 Table 29 Descriptive Statistics for Task Interest for the Narrative Task DT NT DMT M 61.28 69.19 59.19

SE 1.74 2.02 1.96

95% CI Low 57.81 65.17 55.29 High 64.75 73.21 63.09 SD 15.39 17.82 17.29

Minimum 32 32 32

Maximum 100 100 100

Skewness .98 .26 1.09

SES .27 .27 .27

Kurtosis .78 -.77 .65

SEK .54 .54 .54 Rasch Item .89 .95 .93 Reliability N 78 78 78 Notes. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

person measures. The descriptive statistics reveal that the mean for the no choice of topic treatment was highest on the descriptive task (M = 66.44) compared to the decision- making task (M = 66.14) and the narrative task (M = 61.28). The mean for the limited choice of topic was similar for the descriptive (M = 70.09) and the narrative tasks (M =

69.19) and was lower for the decision-making task (M = 66.86). The mean for the complete

132 Table 30 Descriptive Statistics for Task Interest for the Decision-making Task DT NT DMT M 66.14 66.86 61.96

SE 1.88 1.99 1.75

95% CI Low 62.41 62.89 58.48 High 69.88 70.83 65.44 SD 16.56 17.59 15.44

Minimum 32 32 36

Maximum 100 100 100

Skewness .68 .41 .99

SES .27 .27 .27

Kurtosis -.17 -.50 .66

SEK .54 .54 .54 Rasch Item .86 .92 .85 Reliability N 78 78 78 Notes. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

choice of topic treatment was highest on the descriptive task (M = 66.35), and it was

followed by the decision-making task (M = 61.96) and the narrative task (M = 59.19).

In regards to Research Question 2, there was a significant main effect for Task,

F(2, 154) = 10.72 as well as a significant interaction effect between Task and Choice, F(4,

308) = 4.13 (Table 27). This indicates that Choice had different effects on the Task Interest

133 ţŞ

ţŜ

ŢŤ

ŢŢ 5BTLUZQF

ŢŠ %FTDSJQUJWF

ŢŞ /BSSBUJWF

3BTDI.FBTVSFPG5BTL*OUFSFTU ŢŜ %FDJTJPO .BLJOH šŤ /POF-JNJUFE$PNQMFUF -FWFMPG$IPJDF

Figure 8. The degree of Task Interest for each task-type.

ratings depending on the type of task. Also, as shown in Table 27, there was a medium effect size for Task Interest and Task (.12) and high observed power (.99).

Examining the profile plot for Task Interest by task-type (Figure 8), the participants maintained a high level of Task Interest across the different levels of choice for the descriptive task. Indeed, Task Interest was sharply lower for the no choice and complete choice of topic for the narrative task, but was higher again for the decision- making task. Overall, the lowest scores for Task Interest occurred with the narrative task, while the highest occurred with the descriptive task.

To break down the interaction between task types, pairwise comparisons with a Bonferroni correction were requested. In this comparison, the descriptive task was

134 significantly higher than both the narrative and the decision-making tasks at the p < .05

significance level. However, there was no statistically significant difference between the

narrative and decision-making tasks.

Within-samples t-tests between pairs of treatments were conducted in order

to determine where specific differences were. These tests were conducted in sets of

three, with the same type of task but with different levels of choice. To protect against

committing a Type I error, a Bonferroni correction with a probability level of p < .017 (α =

.05/3) was used. For the no choice of topic with different types tasks, the descriptive task

was significantly higher than the narrative task M( diff = 5.15, SD = 13.16, t(77) = 3.46) and the decision-making task was significantly higher than the narrative task M( diff = 4.86,

SD = 15.02, t(77) = 2.86). For the complete choice of topic, the descriptive task was higher

than the narrative task (Mdiff = 7.15, SD = 11.98, t(77) = 5.27) and the decision-making task

(Mdiff = 4.38, SD = 12.95, t(77) = 2.99).

Research Question 3

The third research question concerned the effect of choice on feelings of Task

Self-efficacy. Research Question 3 was: To what degree does the level of Task Self-efficacy

change across the three levels of choice? This question will be answered by showing the

descriptive statistics for each treatment, the results of the repeated-measures ANOVA,

profile plots from the analysis, pairwise comparisons with a Bonferroni correction, and

t-tests showing significant contrasts between the individual treatments.

135 Table 31 Descriptive Statistics for Task Self-efficacy for the No Choice of Topic DT NT DMT M 57.35 51.62 52.77

SE 1.74 2.01 1.69

95% CI Low 53.89 47.61 49.41 High 60.80 55.62 56.13 SD 15.32 17.75 14.88

Minimum 12 14 0

Maximum 100 100 86

Skewness -.20 -.13 -.63

SES .27 .27 .27

Kurtosis 1.05 .56 1.48

SEK .54 .54 .54 Rasch Item .97 .97 .98 Reliability N 78 78 78 Notes. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

Tables 31 to 33 show the descriptive statistics for each of the levels of choice

according to the three task types. The descriptive statistics in Tables 31 to 33 are based

on Rasch person measures. A perusal of the descriptive statistics reveals that the mean

for the descriptive task was highest for the no choice of topic treatment (M = 57.35), the

limited choice of topic (M = 56.60) was the next highest, and the complete choice of

136 Table 32 Descriptive Statistics for Task Self-efficacy for the Limited Choice of Topic DT NT DMT M 56.60 59.03 55.99

SE 1.75 1.38 1.28

95% CI Low 53.11 56.28 53.43 High 60.09 61.77 58.54 SD 15.48 12.19 11.34

Minimum 0 22 25

Maximum 86 100 86

Skewness -.74 -.05 -.57

SES .27 .27 .27

Kurtosis 1.35 1.35 .88

SEK .54 .54 .54 Rasch Item .96 .97 .96 Reliability N 78 78 78 Notes. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

topic treatment (M = 54.42) was the lowest. The narrative and decision-making tasks were higher on the limited choice of task topic treatments (M = 59.03 and M = 55.99, respectively) than on the no choice of task topic treatments (M = 51.62 and M = 52.77, respectively). However, the means for the narrative and the decision-making tasks were the lowest for the complete choice of topic treatment (M = 54.74 and M = 47.46, respectively).

137 Table 33 Descriptive Statistics for Task Self-efficacy for the Complete Choice of Topic DT NT DMT M 54.42 54.74 47.46

SE 1.34 1.56 1.85

95% CI Low 51.76 51.63 43.78 High 57.08 57.85 51.14 SD 11.80 13.80 16.33

Minimum 27 23 0

Maximum 85 87 100

Skewness .29 -.41 -.26

SES .27 .27 .27

Kurtosis .63 .11 1.48

SEK .54 .54 .54 Rasch Item .95 .97 .97 Reliability N 78 78 78 Notes. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

The results from the repeated-measures ANOVA analysis for Task Self-efficacy

are shown in Table 34. Both the Task (ε = .97) and the Choice (ε = .99) independent

variables passed the assumption of sphericity test. However, Mauchly’s test indicated that

the assumption of sphericity had been violated for the interaction effect between Task

and Choice (χ2(9) = 22.09, p < .05). Because the interaction effect of Task and Choice (ε

138 Table 34 Repeated-Measures ANOVA for Task Self-efficacy Source df SS MS F η2 β Task 2 2085.78 1042.89 8.16* .10 .96

Error (Task) 154 19691.33 127.87

Choice 2 3019.10 1509.55 9.55* .11 .98

Error (Choice) 154 24336.68 158.03

Task x Choicea 3.54 2391.35 674.87 5.02* .06 .95 Error 272.85 36661.53 134.36 (Task x Choice)a Notes. η2 = effect size (partial eta squared). β = observed power. a df and MS from Greenhouse-Geisser correction for degrees of freedom. *p < .025.

ŢŜ

šŤ

šŢ

šŠ

šŞ -FWFMPG$IPJDF /POF šŜ -JNJUFE ŠŤ

3BTDI.FBTVSFPG5BTL4FMG&ďDBDZ $PNQMFUF ŠŢ %FTDSJQUJWF/BSSBUJWF%FDJTJPO .BLJOH 5BTLUZQF

Figure 9. The degree of Task Self-efficacy for each level of choice.

139 = .74) was < .75 for Mauchly’s test of sphericity, the degrees of freedom was corrected

using Greenhouse-Geisser estimates of sphericity (ε^ = .89). Correcting for the number of comparisons in Study 1, interactions at the p < .025 significance level (.05/2) will be considered statistically significant.

There was a significant main effect for Choice, F(2, 154) = 9.55 (Table 34) as well

as a significant interaction effect between Task and Choice, F(3.54, 272.85) = 5.02. This

indicates that Choice had different effects on ratings of Task Self-efficacy depending

on the type of task. Again, the observed power for all three tests was high. There was a

medium effect size (.11) for the Choice main effect, and a small effect size (.06) for the

Task and Choice interaction effect.

Examining the profile plot for Task Self-efficacy by level of choice (Figure 9),

it can be seen that the limited choice of topic treatment engendered higher levels of

perceived self-efficacy. However, while the self-efficacy that the students felt when

conducting the descriptive task with no choice was very high, this was also true of the

narrative task with a limited choice of topic. However, the no choice of topic and the

complete choice of topic treatments showed mixed results and mirrored each other

across the different types of tasks.

To break down the interaction between the different levels of choice, pairwise

comparisons with a Bonferroni correction were requested. In this comparison, the

limited level of topic choice was significantly higher than both the no choice and the

complete choice of topic treatments at the p < .05 significance level. However, there was

140 no statistically significant difference between the no choice of topic and complete choice

of topic treatments.

Within-samples t-tests between pairs of treatments were conducted in order

to determine where specific differences were. These tests were conducted in sets of

three, with the same type of task but with different levels of choice. To protect against

committing a Type I error, a Bonferroni correction with a probability level of p < .017 (α

= .05/3) was used. For the narrative task with different levels of choice, the limited choice

of topic was significantly higher than both the no choice of topic M( diff = 7.41, SD = 16.19, t(77) = 4.04) and the complete choice of topic (Mdiff = 4.28, SD = 12.78, t(77) = 2.96). For

the decision-making task, the limited choice of topic was significantly higher than the

complete choice of topic (Mdiff = 8.53, SD = 18.16, t(77) = 4.15), and the no choice of topic

was higher than the complete choice of topic (Mdiff = 5.31, SD = 17.86, t(77) = 2.63).

Research Question 4

The fourth research question concerned the effect of task-type on Task Self-

efficacy. Research Question 4 was: To what degree does the level of Task Self-efficacy

change significantly between the three types of tasks (i.e., the descriptive, narrative,

and decision-making tasks)? This question will be answered by showing the descriptive

statistics for each treatment, the results of the repeated-measures ANOVA, profile

plots from the analysis, pairwise comparisons with a Bonferroni correction, and t-tests

showing significant contrasts between the individual treatments.

141 Table 35 Descriptive Statistics for Task Self-efficacy for the Descriptive Task DT NT DMT M 57.35 56.60 54.42

SE 1.74 1.75 1.34

95% CI Low 53.89 53.11 51.76 High 60.80 60.09 57.08 SD 15.32 15.48 11.80

Minimum 12 0 27

Maximum 100 86 85

Skewness -.20 -.74 .29

SES .27 .27 .27

Kurtosis 1.05 1.35 .63

SEK .54 .54 .54 Rasch Item .97 .96 .95 Reliability N 78 78 78 Notes. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

Tables 35 to 37 show the descriptive statistics for each of the tasks according to the different levels of choice. The descriptive statistics in Tables 35 to 37 are based on Rasch person measures. A perusal of the descriptive statistics reveals that the mean for the no choice of topic for the descriptive task (M = 57.35) was higher that the means on the narrative task (M = 51.62) and the decision-making task (M = 52.77). The means for the

142 Table 36 Descriptive Statistics for Task Self-efficacy for the Narrative Task DT NT DMT M 51.62 59.03 54.74

SE 2.01 1.38 1.56

95% CI Low 47.61 56.28 51.63 High 55.62 61.77 57.85 SD 17.75 12.19 13.80

Minimum 14 22 23

Maximum 100 100 87

Skewness -.13 -.05 -.41

SES .27 .27 .27

Kurtosis .56 1.35 .11

SEK .54 .54 .54 Rasch Item .97 .97 .97 Reliability N 78 78 78 Notes. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

limited level of topic treatment mirrors this as the mean for the narrative task (M = 59.03)

was higher than the means for the descriptive task (M = 56.60) and the decision-making task (M = 55.99). The means for the descriptive and narrative tasks M( = 54.42 and M =

54.74, respectively) were nearly identical for the complete choice of topic and lowest for the decision-making task (M = 47.46).

143 Table 37 Descriptive Statistics for Task Self-efficacy for the Decision-Making Task DT NT DMT M 52.77 55.99 47.46

SE 1.69 1.28 1.85

95% CI Low 49.41 53.43 43.78 High 56.13 58.54 51.14 SD 14.88 11.34 16.33

Minimum 0 25 0

Maximum 86 86 100

Skewness -.63 -.57 -.26

SES .27 .27 .27

Kurtosis 1.48 .88 1.48

SEK .54 .54 .54 Rasch Item .98 .96 .97 Reliability N 78 78 78 Notes. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

There was a significant main effect for Task, F(2, 154) = 8.16 as well as a significant interaction effect between Task and Choice, F(3.54, 272.85) = 5.02 (Table 34). This indicates that Choice affected the ratings of Task Self-efficacy differently depending on the type of task. As shown in Table 34, the observed power for the Task main effect was high (.96) and there was a medium effect size (.10).

144 ŢŜ

šŤ

šŢ

šŠ 5BTLUZQF

šŞ %FTDSJQUJWF

šŜ /BSSBUJWF

ŠŤ %FDJTJPO

3BTDI.FBTVSFPG5BTL4FMG&ďDBDZ .BLJOH ŠŢ /POF-JNJUFE$PNQMFUF -FWFMPG$IPJDF

Figure 10. The degree of Task Self-efficacy for each task-type.

Examining the profile plot for Task Self-efficacy by the type of task (Figure 10), the results are mixed. Task Self-efficacy was higher for the narrative and decision-making tasks when limited choice is implemented.

To break down the interaction between task types, pairwise comparisons with a

Bonferroni correction were requested. In this comparison, the decision-making task was lower than both the descriptive and the narrative tasks at the p < .05 significance level.

However, there was no significant difference between the descriptive and the narrative tasks.

Within-samples t-tests between pairs of treatments were conducted in order to determine where specific differences were. These tests were conducted in sets of three, with the same type of task but with different levels of choice. To protect against committing a Type I error, a Bonferroni correction with a probability level of p < .017 (α

145 = .05/3) was used. For the no choice of topic with different types tasks, the descriptive

task was significantly higher than the narrative task (Mdiff = 5.73, SD = 19.85, t(77) = 2.55).

For the limited choice of topic with the different types of tasks, the narrative task was significantly higher than the decision-making task (Mdiff = 3.04, SD = 10.72, t(77) = 2.50).

For the complete choice of topic with the different types of tasks, the descriptive task was significantly higher than the decision-making task (Mdiff = 6.96, SD = 15.89, t(77) = 3.87)

and the narrative task was significantly higher than the decision-making task as well

(Mdiff = 7.28, SD = 14.73, t(77) = 4.37).

146 CHAPTER 6

RESULTS OF STUDY 2

Production data garnered from student conversations comprised Study 2. By the end of the data collection sessions, 21 pairs of students had participated in all the data collection sessions and remained for further analysis. These students are described in Chapter 3. The production data gathered during the first two minutes of the second round of the task implementation were used for data analysis in order to control for planning, as mentioned in Chapter 3. The production data are of both students in the pair combined.

As stated in Chapter 3, there are three dependent variables for Study 2; Accuracy,

Complexity, and Fluency. Accuracy was operationalized as the ratio of correct verb

forms (CVF) and the ratio of error-free clauses (EFC). Complexity was operationalized as

turns (TURNS), type-token ratio (TTR), and words per turn (WPT). Lastly, Fluency was

operationalized as unaltered repetitions (UNREP) and total word count for the selected

portion of the transcript (WC).

Descriptive Statistics

The data were first examined for skewness, using the z-score of the skewness

to gauge the severity of each item’s skewness (Field, 2005, p. 72). The skewness statistic

is divided by the standard error of the skewness to determine the severity of the skew.

147 According to Field, a z-score greater than +/- 1.96 indicates the skewness to be significant at the p < .05 level.

Table 38 Descriptive Statistics for the Descriptive Task with No Choice of Topic

M SE SD Min Max Skew SES zsk Kur SEK CVF .92 .02 .11 .60 1.00 -1.40 .50 -2.80* 1.46 .97

EFC .48 .05 .23 .00 .80 -.38 .50 -.76 -.73 .97

TURNS 19.00 2.04 9.33 6 38 1.01 .50 2.02* .30 .97

TTR .34 .01 .05 .24 .46 -.11 .50 -.22 .26 .97

WPT 4.73 .35 1.59 2.77 9 .99 .50 1.98* 1.04 .97

UNREP 3.57 .49 2.23 0 9 .57 .50 1.14 .28 .97

WC 79.29 4.91 22.48 41 145 1.41 .50 2.82* 2.97 .97 Notes. N = 21. Skew = Skewness; Kur = Kurtosis. CVF = ratio of correct verb phrases (max = 1.00); EFC = ratio of error-free clauses (max = 1.00); TURNS = number of turns; TTR = type-token ratio (max = 1.00); WPT = word per turn; UNREP = unaltered repetitions; WC = word count. *p < .05.

In the descriptive task with no choice of topic treatment (Table 38), turns, words-

per-turn, and word count were positively skewed at the p < .05 significance level, and

correct verb forms was negatively skewed at the p < .05 level. Correct verb forms and word count were significantly platykurtic at the p < .05.

In the descriptive task with limited choice of topic treatment (Table 39), unaltered repetitions and words-per-turn were positively skewed at the p < .05 significance level,

148 Correct verb forms was negatively skewed at the p < .05 level. Words-per-turn was

significantly leptokurtic at the p < .05 level and word count was significantly platykurtic

at the p < .05. In addition, correct verb forms and turns were bimodal.

Table 39 Descriptive Statistics for the Descriptive Task with Limited Choice of Topic

M SE SD Min Max Skew SES zsk Kur SEK CVF .85 .04 .17 .40 1.00 -1.36 .50 -2.72* 1.22 .97

EFC .42 .03 .17 .00 .71 -.35 .50 -.70 .38 .97

TURNS 19.48 2.37 10.87 5 44 .48 .50 .96 -.52 .97

TTR .41 .02 .09 .26 .54 -.09 .50 -.18 -1.12 .97

WPT 5.42 .64 2.94 2.71 14.80 1.94 .50 3.88* 4.24 .97

UNREP 2.38 .50 2.31 0 8 1.09 .50 2.18* .52 .97

WC 83.57 5.63 25.80 38 123 .14 .50 .28 -1.20 .97 Notes. N = 21. Skew = Skewness; Kur = Kurtosis. CVF = ratio of correct verb phrases (max = 1.00); EFC = ratio of error-free clauses (max = 1.00); TURNS = number of turns; TTR = type-token ratio (max = 1.00); WPT = word per turn; UNREP = unaltered repetitions; WC = word count. *p < .05.

In the descriptive task with complete choice of topic treatment (Table 40), correct

verb forms was negatively skewed at the p < .05 significance level and words-per-turn was positively skewed at the p < .05 level. Correct verb forms was significantly platykurtic at the p < .05 level. Error-free clauses was bimodal.

149 Table 40 Descriptive Statistics for the Descriptive Task with Complete Choice of Topic

M SE SD Min Max Skew SES zsk Kur SEK CVF .96 .01 .08 .67 1.00 -2.41 .50 -4.82* 6.10 .97

EFC .34 .05 .25 .00 1.00 .69 .50 1.38 .83 .97

TURNS 14.29 2.33 10.68 1 33 .30 .50 .60 -1.17 .97

TTR .42 .02 .09 .29 .63 .61 .50 1.22 -.18 .97

WPT 15.70 4.85 22.21 2.64 69 1.76 .50 3.52* 1.70 .97

UNREP 1.62 .32 1.47 0 4 .53 .50 1.06 -.96 .97

WC 68.57 4.81 22.06 31 102 -.13 .50 -.26 -1.09 .97 Notes. N = 21. Skew = Skewness; Kur = Kurtosis. CVF = ratio of correct verb phrases (max = 1.00); EFC = ratio of error-free clauses (max = 1.00); TURNS = number of turns; TTR = type-token ratio (max = 1.00); WPT = word per turn; UNREP = unaltered repetitions; WC = word count. *p < .05.

Table 41 Descriptive Statistics for the Narrative Task with No Choice of Topic

M SE SD Min Max Skew SES zsk Kur SEK CVF .39 .05 .25 .00 .73 -.46 .50 -.92 -.92 .97

EFC .11 .03 .13 .00 .40 .92 .50 1.84 -.50 .97

TURNS 16.57 22.38 10.91 1 36 .41 .50 .82 -.72 .97

TTR .42 .02 .09 .27 .58 .20 .50 .40 -.73 .97

WPT 7.84 2.29 10.49 2.31 42 2.71 .50 5.42* 6.74 .97

UNREP 1.95 .36 1.63 0 5 .47 .50 .94 -.74 .97

WC 62.43 4.54 20.80 34 105 .38 .50 .76 -1.06 .97 Notes. N = 21. Skew = Skewness; Kur = Kurtosis. CVF = ratio of correct verb phrases (max = 1.00); EFC = ratio of error-free clauses (max = 1.00); TURNS = number of turns; TTR = type-token ratio (max = 1.00); WPT = word per turn; UNREP = unaltered repetitions; WC = word count. *p < .05. 150 In the narrative task with no choice of topic treatment (Table 41), words-per-

turn was positively skewed at the p < .05 significance level. Words-per-turn was also

significantly platykurtic at the p < .05 level. Unaltered repetitions was bimodal.

Table 42 Descriptive Statistics for the Narrative Task with Limited Choice of Topic

M SE SD Min Max Skew SES Zsk Kur SEK CVF .41 .05 .23 .00 .79 -.98 .50 -1.96* -1.15 .97

EFC .13 .04 .19 .00 .67 1.65 .50 3.30* 2.43 .97

TURNS 14.95 2.13 9.74 1 43 1.00 .50 2.00* 2.10 .97

TTR .49 .03 .14 .28 .76 .07 .50 .14 -.93 .97

WPT 10.00 3.32 15.23 2.67 64 2.98 .50 5.96* 8.71 .97

UNREP 1.52 .34 1.57 0 6 1.24 .50 2.48* 1.92 .97

WC 68.00 4.18 19.15 43 117 1.11 .50 2.22* 1.19 .97 Notes. N = 21. Skew = Skewness; Kur = Kurtosis. CVF = ratio of correct verb phrases (max = 1.00); EFC = ratio of error-free clauses (max = 1.00); TURNS = number of turns; TTR = type-token ratio (max = 1.00); WPT = word per turn; UNREP = unaltered repetitions; WC = word count. *p < .05.

In the narrative task with limited choice of topic treatment (Table 42), error-free

clauses, turns, words-per-turn, unaltered repetitions, and word count were positively

skewed at the p < .05 significance level. Correct verb forms was negatively skewed. Turns,

words-per-turn, and unaltered repetitions were significantly platykurtic at the p < .05

level. In addition, type-token ratio was bimodal.

151 In the narrative task with complete choice of topic treatment (Table 43), turns

and words-per-turn were positively skewed at the p < .05 significance level, and unaltered

repetitions was negatively skewed at the p < .05 level. Unaltered repetitions was also

significantly platykurtic at the p < .05 level.

Table 43 Descriptive Statistics for the Narrative Task with Complete Choice of Topic

M SE SD Min Max Skew SES zsk Kur SEK CVF .64 .05 .25 .20 1.00 -.10 .50 -.20 -1.28 .97

EFC .39 .04 .18 .00 .78 .17 .50 .34 .28 .97

TURNS 12.71 3.17 14.52 1 51 1.45 .50 2.90* 1.46 .97

TTR .51 .03 .14 .27 .72 -.20 .50 -.41 -.91 .97

WPT 23.32 6.16 28.20 2.39 98 1.34 .50 2.67* .86 .97

UNREP 3.38 .95 4.36 0 18 2.06 .50 4.12* 5.39 .97

WC 68.52 6.15 28.17 31 122 .79 .50 1.57 -.60 .97 Notes. N = 21. Skew = Skewness; Kur = Kurtosis. CVF = ratio of correct verb phrases (max = 1.00); EFC = ratio of error-free clauses (max = 1.00); TURNS = number of turns; TTR = type-token ratio (max = 1.00); WPT = word per turn; UNREP = unaltered repetitions; WC = word count. *p < .05.

In the decision-making task with no choice of topic treatment (Table 44), correct

verb forms was negatively skewed at the p < .05 significance level, and words-per-turn was positively skewed at the p < .05 significance level. Correct verb forms and words-per-turn were significantly platykurtic at the p < .05 level.

152 Table 44 Descriptive Statistics for the Decision-Making Task with No Choice of Topic

M SE SD Min Max Skew SES zsk Kur SEK CVF .92 .04 .22 .00 1.00 -3.87 .50 -7.74* 15.95 .97

EFC .48 .06 .28 .00 1.00 .13 .50 .26 -.17 .97

TURNS 14.10 1.54 7.07 3 28 .28 .50 .56 -.68 .97

TTR .42 .02 .08 .27 .56 -.29 .50 -.58 -.60 .97

WPT 5.47 .82 3.77 2.21 15.67 1.72 .50 3.44* 2.50 .97

UNREP 2.10 .37 1.70 0 5 .51 .50 1.02 -.88 .97

WC 57.86 3.47 15.91 33 94 .32 .50 .64 -.20 .97 Notes. N = 21. Skew = Skewness; Kur = Kurtosis. CVF = ratio of correct verb phrases (max = 1.00); EFC = ratio of error-free clauses (max = 1.00); TURNS = number of turns; TTR = type-token ratio (max = 1.00); WPT = word per turn; UNREP = unaltered repetitions; WC = word count. *p < .05.

Table 45 Descriptive Statistics for the Decision-Making Task with Limited Choice of Topic

M SE SD Min Max Skew SES zsk Kur SEK CVF .83 .06 .28 .00 1.00 -2.34 .50 -4.68* 5.06 .97

EFC .49 .05 .27 .00 1.00 -.37 .50 -.74 .18 .97

TURNS 17.29 1.88 8.61 2 38 .34 .50 .68 .26 .97

TTR .48 .03 .14 .24 .72 .51 .50 1.02 -.64 .97

WPT 5.20 1.18 5.41 1.71 25.50 2.99 .50 5.98* 10.13 .97

UNREP 2.43 .41 1.86 0 7 .64 .50 1.28 .09 .97

WC 62.33 5.48 25.12 26 115 .29 .50 .58 -.89 .97 Notes. N = 21. Skew = Skewness; Kur = Kurtosis. CVF = ratio of correct verb phrases (max = 1.00); EFC = ratio of error-free clauses (max = 1.00); TURNS = number of turns; TTR = type-token ratio (max = 1.00); WPT = word per turn; UNREP = unaltered repetitions; WC = word count. *p < .05.

153 In the decision-making task with limited choice of topic treatment (Table 45),

correct verb forms was negatively skewed at the p < .05 significance level, and words-per- turn was positively skewed at the p < .05 significance level. Correct verb forms and words- per-turn were also significantly platykurtic at the p < .05 level. In addition, error-free clauses, turns, unaltered repetitions, and word count were bimodal.

Table 46 Descriptive Statistics for the Decision-Making Task with Complete Choice of Topic

M SE SD Min Max Skew SES zsk Kur SEK CVF .89 .03 .14 .60 1.00 -.96 .50 -1.92 -.34 .97

EFC .28 .04 .03 .00 .71 .28 .50 .56 .16 .97

TURNS 16.00 1.56 7.14 2 31 -.01 .50 -.02 -.25 .97

TTR .45 .02 .01 .09 .32 1.43 .50 2.86* 2.22 .97

WPT 6.27 .75 3.45 2.52 14.50 1.23 .50 2.46* .51 .97

UNREP 2.90 .46 2.10 0 6 .10 .50 .20 -1.39 .97

WC 80.00 4.40 20.16 29 121 -.54 .50 -1.08 1.13 .97 Notes. N = 21. Skew = Skewness; Kur = Kurtosis. CVF = ratio of correct verb phrases (max = 1.00); EFC = ratio of error-free clauses (max = 1.00); TURNS = number of turns; TTR = type-token ratio (max = 1.00); WPT = word per turn; UNREP = unaltered repetitions; WC = word count. *p < .05.

In the decision-making task with complete choice of topic treatment (Table 46), type-token ratio and words-per-turn were positively skewed at the p < .05 significance level. Type-token ratio was also significantly platykurtic at the p < .05 level.

154 In review, all variables showed skewness through the nine treatments. Correct

verb forms was significantly skewed at the p < .05 level six times. Error-free clauses

was significantly skewed at the p < .05 level once. Turns was significantly skewed at the

p < .05 level three times. Type-token ratio was significantly skewed at the p < .05 level

once. Words-per-turn was significantly skewed at the p < .05 level nine times. Unaltered

repetitions was significantly skewed at the p < .05 level three times. Word count was significantly skewed at the p < .05 level two times.

Out of the nine treatments, the descriptive task with no choice of topic treatment survey had four skewed items, the descriptive task with limited choice of topic treatment survey had three skewed items, the descriptive task with complete choice of topic treatment survey had two skewed items, the narrative task with no choice of topic treatment survey had one skewed item, the narrative task with limited choice of topic treatment survey had six skewed items, the narrative task with complete choice of topic treatment survey had three skewed items, the decision-making task with no choice of topic treatment survey had two skewed items, the decision-making task with limited choice of topic treatment survey had two skewed items, and the decision-making task with limited choice of topic treatment survey had two skewed items.

Final Data Analysis

The non-transformed data were entered into an SPSS file for the repeated- measures ANOVA analysis. After analyzing the data using the repeated-measures

155 general linear model in SPSS, words per turn violated the assumption of sphericity for

the interaction effect (ε = .011, p < .05). This variable was removed from the final analysis.

Also, the ratio of correct verb forms was severely non-normal, even though it passed the

assumption of sphericity (ε = .644). Therefore, this variable was removed for the final

analysis. This left the ratio of error-free clauses (ε = .443) to gauge Accuracy, the number

of turns in the interaction (ε = .469) and type-token ratio (ε = .637) to gauge Complexity,

and unaltered repetitions (ε = .696) and word count to gauge Fluency.

For Complexity, even though neither type-token ratio nor the number of turns

violated the assumption of sphericity, according to the Greenhouse-Geisser correction,

type-token ratio (ε^ = .817) was closer to sphericity than was turns (ε^ = .696). Therefore, type-token ratio was used to gauge Complexity.

For Fluency, according to the Greenhouse-Geisser correction, word count (ε^ =

.869) was closer to sphericity than was unaltered repetitions (ε^ = .696). Therefore, word count was used to gauge Fluency. Reflecting back to the summary of the descriptive statistics for the individual measures for Study 2, the ratio of error-free clauses, type- token ratio, and word count were also the least skewed compared to the other measures.

Research Question 1

The first research question concerned the effect of choice on Accuracy (ratio of

error-free clauses). Research Question 1 was: To what degree does the level of accuracy

change across the three levels of choice (i.e., no choice, limited choice, and complete

156 choice)? This question will be answered by showing the descriptive statistics, the results of the repeated-measures ANOVA, profile plots from the analysis, pairwise comparisons with a Bonferroni correction to investigate differences in the levels of choice, and t-tests showing significant contrasts between the individual treatments.

Tables 47 to 49 display the descriptive statistics for Accuracy for each of the three levels of choice according to the different tasks. The mean for Accuracy on the descriptive

Table 47 Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the No Choice of Topic Treatment DT NT DMT M .48 .11 .48

SE .05 .03 .06

95% CI Low .37 .04 .35 High .58 .17 .60 SD .23 .13 .28

Minimum .00 .00 .00

Maximum .80 .40 1.00

Skewness -.38 .92 .13

SES .50 .50 .50

Kurtosis -.72 -.50 -.17

SEK .97 .97 .97

N 21 21 21 Note. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

157 Table 48 Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the Limited Choice of Topic Treatment DT NT DMT M .42 .13 .49

SE .04 .04 .06

95% CI Low .34 .04 .37 High .50 .22 .62 SD .17 .19 .27

Minimum .00 .00 .00

Maximum .71 .67 1.00

Skewness -.35 1.65 -.37

SES .50 .50 .50

Kurtosis .38 2.43 .18

SEK .97 .97 .97

N 21 21 21 Note. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

task was higher for the no choice of topic treatment (M = .48) than for the limited choice

of topic treatment (M = .42), and the complete choice of topic treatment (M = .34) was the

lowest. The mean for Accuracy on the narrative task was highest for the complete choice of topic treatment (M = .38) and this was followed by the limited choice (M = .13) and the no choice of topic treatment (M = .11). The mean for Accuracy on the decision-making task

158 Table 49 Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the Complete Choice of Topic Treatment DT NT DMT M .34 .38 .28

SE .06 .04 .04

95% CI Low .22 .30 .20 High .45 .47 .37 SD .25 .18 .18

Minimum .00 .00 .00

Maximum 1.00 .78 .71

Skewness .68 .17 .29

SES .50 .50 .50

Kurtosis .83 .28 .16

SEK .97 .97 .97

N 21 21 21 Note. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

was highest for the limited choice of topic treatment (M = .49). This was followed by the no choice of topic treatment (M = .48) and the complete choice of topic treatment (M = .28).

For the repeated-measures ANOVA analysis, Mauchly’s test indicated that the assumption of sphericity had been met for the main effect of Choice (ε = .884) and the interaction effect between Task and Choice (ε = .443), but the main effect of Task (ε =

.616) failed to meet sphericity (χ2(2) = 9.20, p < .05). Therefore, the Greenhouse-Geisser

159 Table 50 Repeated-Measures ANOVA for Accuracy (Ratio of Error-Free Clauses) Source df SS MS F η2 β Taska 1.44 .01 .005 .11 .05 .06

Error (Task) 28.91 2.11 .07

Choice 2 1.81 1.01 15.37* .43 .99

Error (Choice) 40 2.36 .05

Task x Choice 4 1.75 .44 11.90* .37 .99 Error 80 2.94 .04 (Task x Choice)a Notes. η2 = effect size (partial eta squared). β = observed power. a df and MS from Greenhouse-Geisser correction for degrees of freedom. *p < .017.

correction was used (ε^ = .723) because the test of sphericity was under .75 (Girden,

1992). Correcting for the number of comparisons in Study 2, interactions at the p <

.017 significance level were considered statistically significant. In relation to Research

Question 1, there was a statistically significant interaction effect between Task and

Choice, F(4, 80) = 11.90 (Table 50). This indicates that Choice had different effects on

Accuracy depending on the type of task. The main effect of Choice also was statistically

significant, F(2, 40) = 15.37. In addition, the observed power and the effect size for the

Choice main effect, which indicates that the relationship between the independent variable of Choice and Accuracy as well as the interaction effect of Task and Choice were

both very strong. This indicates that there was a strong relationship between the two

independent variables and Accuracy.

160 Ţ

ş -FWFMPG$IPJDF /POF Ş -JNJUFE ŝ $PNQMFUF "DDVSBDZ 3BUJPPG&SSPS'SFF$MBVTFT ŜŜ %FTDSJQUJWF/BSSBUJWF%FDJTJPO .BLJOH 5BTLUZQF Figure 11. Profile plot of Accuracy (ratio of error-free clauses) by level of choice.

The results were mixed for the profile plot for Accuracy by level of choice (Figure

11). The complete and no choice of topic treatments engendered greater accuracy across the different tasks, but the limited choice treatment engendered greater accuracy only for the decision-making task. It appears that in general, providing a choice of topic exerted a positive influence on the accuracy of the students’ spoken production.

To break down this interaction, pairwise comparisons with a Bonferroni correction were requested. Both the no choice and the complete choice of topic treatments were significantly higher than the limited choice of topic treatment at the p < .05 significance level. However, there was no significant difference between the no choice and the complete choice of topic treatments.

161 Within-samples t-tests between pairs of treatments were conducted in order

to determine where specific differences were. These tests were conducted in sets of

three, with the same type of task but with different levels of choice. To protect against

committing a Type I error, a Bonferroni correction with a probability level of p < .017 (α =

.05/3) was used. For the narrative task with the different levels of choice, complete choice

was significantly higher than limited choice M( diff = 0.25, SD = .26, t(20) = 4.36) as well

as no choice (Mdiff = 0.28, SD = .23, t(20) = 5.53). For the decision-making task with the different levels of choice, limited choice was significantly higher than complete choice

(Mdiff = 0.21, SD = .28, t(20) = 3.37) and no choice was significantly higher than complete

choice (Mdiff = 0.19, SD = .33 t(20) = 2.70). There were no differences amongst the three levels of choice for the descriptive task.

Research Question 2

The second research question concerned the effect of task on Accuracy (the ratio of error-free clauses). Research Question 2 was: To what degree does the level of accuracy change between the three types of task (i.e., the descriptive, narrative, and decision- making tasks)? This question was answered by inspecting the descriptive statistics, the results of the repeated-measures ANOVA, profile plots from the analysis, pairwise comparisons with a Bonferroni correction to investigate differences in the levels of choice, and t-tests showing significant contrasts between the individual treatments.

162 Table 51 Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the Descriptive Task Treatment DT NT DMT M .48 .42 .33

SE .05 .04 .06

95% CI Low .37 .34 .22 High .58 .50 .45 SD .23 .17 .25

Minimum .00 .00 .00

Maximum .80 .71 1.00

Skewness -.38 -.35 .69

SES .50 .50 .50

Kurtosis -.73 .38 .83

SEK .97 .97 .97

N 21 21 21 Note. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

The descriptive statistics for each of the tasks according to the level of choice

are displayed in Tables 51 to 53. The means for the no choice of topic treatment for the

descriptive task (M = .48) and the decision-making task (M = .48) were considerably higher than the mean for the narrative task (M = .11). This pattern was repeated for the limited choice of topic treatments as the means for the decision-making task (M = .49)

163 Table 52 Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the Narrative Task Treatment DT NT DMT M .11 .13 .39

SE .03 .04 .04

95% CI Low .04 .05 .30 High .17 .22 .47 SD .13 .19 .18

Minimum .00 .00 .00

Maximum .40 .67 .78

Skewness .92 1.65 1.04

SES .50 .50 .50

Kurtosis -.50 2.43 .28

SEK .97 .97 .97

N 21 21 21 Note. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

and the descriptive task (M = .42) were higher than the mean for the narrative task (M =

.13). For the complete choice of topic treatment, this pattern was reversed as the mean for the narrative task (M = .39) was higher than the mean for either the descriptive task (M =

.33) or the mean for the decision-making task (M = .28).

164 Table 53 Descriptive Statistics for Accuracy (Ratio of Error-free Clauses) for the Decision-making Task Treatment DT NT DMT M .48 .49 .28

SE .06 .06 .04

95% CI Low .35 .37 .20 High .60 .62 .36 SD .28 .27 .18

Minimum .00 .00 .00

Maximum 1.00 1.00 .71

Skewness .13 -.37 .29

SES .50 .50 .50

Kurtosis -.17 .17 .16

SEK .97 .97 .97

N 21 21 21 Note. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

There was no statistically significant main effect for Task, F(1.45, 28.90) = .11

(Table 50). However, there was a statistically significant interaction effect between Task and Choice, F(4, 80) = 8.95. This indicates that Choice had different effects on the ratings of Accuracy depending on the type of task. In addition, as shown in Table 50, observed

165 power (.99) was high and the effect size (.37) was strong for the interaction effect, and there was low power and a very weak effect size for the Task independent variable.

Š 5BTLUZQF ş %FTDSJQUJWF Ş /BSSBUJWF

ŝ %FDJTJPO .BLJOH

"DDVSBDZ 3BUJPPG&SSPS'SFF$MBVTFT ŜŜ /POF-JNJUFE$PNQMFUF -FWFMPG$IPJDF

Figure 12. Profile plot of Accuracy (ratio of error-free clauses) by task type.

The profile plot for Accuracy by task type (Figure 12) indicates that the participants maintained a more even level of accuracy across the different levels of choice on the decision-making task. The results are mixed for the descriptive and narrative tasks, as the participants were much less accurate for the limited choice of topic treatment for these two tasks.

166 To break down the interaction between task types, pairwise comparisons with a

Bonferroni correction were requested. There were no significant differences between the

different tasks.

Within-samples t-tests between pairs of treatments were conducted in order

to determine where specific differences were. These tests were conducted in sets of

three, with the same type of task but with different levels of choice. To protect against

committing a Type I error, a Bonferroni correction with a probability level of p < .017 (α

= .05/3) was used. For the no choice of topic, the descriptive task was significantly higher

than the narrative task (Mdiff = 0.37, SD = .24, t(20) = 7.14), and the decision-making task was also significantly higher than the narrative task M( diff = 0.37, SD = .30, t(20) = 5.71).

For the limited choice of topic, the descriptive task was significantly higher than the narrative task (Mdiff = 0.29, SD = .27, t(20) = 4.80), and the decision-making task was also

significantly higher than the narrative task M( diff = 0.36, SD = .36, t(20) = 4.67). Lastly,

the decision-making task was significantly higher than the descriptive task M( diff = 0.41,

SD = .26, t(20) = 7.29). There were no differences amongst the three different tasks for complete choice.

Research Question 3

The third research question concerned the effect of choice on Complexity (type- token ratio). Research Question 3 was: To what degree does the level of complexity change across the three levels of choice (i.e., no choice, limited choice, and complete choice)?

167 This question was answered by inspecting the descriptive statistics, the results of the

repeated-measures ANOVA, profile plots from the analysis, pairwise comparisons with a

Bonferroni correction to investigate differences in the levels of choice, and t-tests showing

significant contrasts between the individual treatments.

Each operationalization of the Complexity dependent variable was applied to a

repeated-measures ANOVA analysis. As noted above, words per turn (ε = .011, p < .05)

Table 54 Descriptive Statistics for Complexity (Type-token Ratio) for the No Choice of Topic Treatment DT NT DMT M .34 .42 .42

SE .01 .02 .02

95% CI Low .32 .38 .38 High .37 .46 .45 SD .05 .09 .08

Minimum .24 .27 .27

Maximum .46 .58 .56

Skewness -.11 .20 -.29

SES .50 .50 .50

Kurtosis .26 -.73 -.60

SEK .97 .97 .97

N 21 21 21 Note. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

168 Table 55 Descriptive Statistics for Complexity (Type-token Ratio) for the Limited Choice of Topic Treatment DT NT DMT M .41 .49 .48

SE .02 .03 .03

95% CI Low .37 .43 .42 High .45 .55 .54 SD .09 .14 .14

Minimum .26 .28 .24

Maximum .54 .76 .72

Skewness -.09 .07 .51

SES .50 .50 .50

Kurtosis -1.12 -.93 -.64

SEK .97 .97 .97

N 21 21 21 Note. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

was in violation of the sphericity assumption, so this variable was removed from the

final analysis. In addition, although the number of turns in the interaction (ε = .469) and

type-token ratio (ε = .637) both met the sphericity assumption, type-token ratio (ε^ = .817)

was closer to sphericity than was turns (ε^ = .696), according to the Greenhouse-Geisser correction. Therefore, type-token ratio was used to gauge Complexity.

169 Table 56 Descriptive Statistics for Complexity (Type-token Ratio) for the Complete Choice of Topic Treatment DT NT DMT M .42 .51 .45

SE .02 .03 .02

95% CI Low .38 .44 .41 High .46 .57 .49 SD .09 .14 .09

Minimum .29 .27 .32

Maximum .63 .72 .72

Skewness .61 -.20 1.43

SES .50 .50 .50

Kurtosis -.18 -.91 2.22

SEK .97 .97 .97

N 21 21 21 Note. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

The descriptive statistics for Complexity for each of the different levels of choice

according to the different types of tasks are displayed in Tables 54 to 56. An examination

of the descriptive statistics reveals that the mean for the descriptive task was higher for

the complete choice of topic treatment (M = .42) and the limited choice of topic treatment

(M = .41) than the mean for the no choice of topic treatment (M = .34). The same pattern

held for the narrative task, as the mean for the complete choice was the highest (M =

170 .51), the mean for the limited choice of topic treatment the next highest (M = .49), and

the mean for the no choice of topic treatment was the lowest (M = .42). Lastly, the mean

for the decision-making task was highest with the limited choice of topic treatment (M

= .48), the mean for the complete choice of topic was the next highest (M = .45), and the

mean for the no choice of topic was the lowest (M = .42).

Table 57 Repeated-Measures ANOVA for Complexity (Type-Token Ratio) Source df SS MS F η2 β Task 2 .20 .10 7.74* .28 .93

Error (Task) 40 .53 .01

Choice 2 .18 .09 9.98* .33 .98

Error (Choice) 40 .37 .01

Task x Choice 4 .02 .01 .74 .04 .23 Error (Task x 80 .59 .01 Choice) Notes. η2 = effect size (partial eta squared). β = observed power. *p < .017.

The results from the analysis for the Complexity variable are shown in Table 57.

Mauchly’s test indicated that the assumption of sphericity had been met for the main

effect of Task (ε = .966), the main effect of Choice (ε = .927), and the interaction effect

between Task and Choice (ε = .637). Therefore, no correction of degrees of freedom for

sphericity was needed. Correcting for the number of comparisons in Study 2, interactions

171 at the p < .017 significance level were considered to be statistically significant. The main effect of Choice, F(2, 40) = 9.98 was significant at p < .017. However, the interaction effect between Task and Choice F(4, 80) = .74 was not statistically significant. Although the power and effect size for the interaction effect of Task and Choice were both very low, there was high observed power for the Choice main effect as well as a very strong effect size. This indicates that Choice and Complexity were strongly related.

-FWFMPG$IPJDF

Š /POF -JNJUFE $PNQMFYJUZ 5ZQF5PLFO3BUJP $PNQMFUF ş %FTDSJQUJWF/BSSBUJWF%FDJTJPO .BLJOH 5BTLUZQF

Figure 13. Profile plot of Complexity (type-token ratio) by level of choice.

The profile plot for Complexity by level of choice (Figure 13) indicates that the limited and complete choice of topic treatments resulted in relatively high levels of

172 complexity across all task types. The no choice of topic showed consistently lower levels of complexity than both choice treatments across all task types.

To break down the interaction between the different levels of choice, pairwise comparisons with a Bonferroni correction were requested. In this comparison, the limited level of topic choice was higher than the no choice of topic treatment to a statistically significant degree (p < .05). The complete choice of topic treatment was also significantly higher then the no choice of topic treatment (p < .05). However, there was no statistically significant difference between the limited choice of topic and complete choice of topic treatments.

Within-samples t-tests between pairs of treatments were run in order to determine where the specific differences were. These tests were conducted in sets of three, with the same type of task but with different levels of choice. To correct for Type

I error, a Bonferroni correction with a probability level of p < .017 (α = .05/3) was used.

For the descriptive task with different levels of choice, the limited choice of topic was significantly higher than the no choice of topic M( diff = 0.07, SD = .07, t(20) = 4.53). In addition, the complete choice of topic was significantly higher than the no choice of topic

(Mdiff = 0.08, SD = .08, t(20) = 4.52). For the narrative task with different levels of choice, the same pattern emerged. The limited choice of topic was significantly higher than the no choice of topic (Mdiff = 0.07, SD = .11, t(20) = 3.05), and the complete choice of topic was significantly higher than the no choice of topic M( diff = 0.09, SD = .13, t(20) = 3.11). There

173 were no statistically significant differences for the decision-making task amongst the different levels of choice.

Research Question 4

The fourth research question concerned the effect of task on Complexity (type- token ratio). Research Question 4 was: To what degree does the level of complexity

Table 58 Descriptive Statistics for Complexity (Type-token Ratio) for the Descriptive Task Treatment DT NT DMT M .34 .41 .42

SE .01 .02 .02

95% CI Low .32 .37 .38 High .37 .45 .46 SD .05 .09 .09

Minimum .24 .26 .29

Maximum .46 .54 .63

Skewness -.11 -.09 .61

SES .50 .50 .50

Kurtosis .26 -1.12 -.18

SEK .97 .97 .97

N 21 21 21 Notes. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

174 change between the three types of tasks (i.e., the descriptive, narrative, and decision-

making tasks)? This question was answered by inspecting the descriptive statistics,

the results of the repeated-measures ANOVA, profile plots from the analysis, pairwise

comparisons with a Bonferroni correction to investigate differences in the levels of choice

and, t-tests showing significant contrasts between the individual treatments.

Table 59 Descriptive Statistics for Complexity (Type-token Ratio) for the Narrative Task Treatment DT NT DMT M .42 .49 .51

SE .02 .03 .03

95% CI Low .38 .43 .44 High .46 .55 .57 SD .09 .14 .14

Minimum .27 .28 .27

Maximum .58 .76 .72

Skewness .20 .07 -.20

SES .50 .50 .50

Kurtosis -.73 -.93 -.91

SEK .97 .97 .97

N 21 21 21 Notes. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

175 Tables 58 to 60 show the descriptive statistics for Complexity for each of the tasks according to the different levels of choice. The mean for the descriptive task was lowest for the no choice of topic treatment (M = .34) and 0the means for the narrative task and the decision-making task for the no choice of topic treatment were the same (M = .42).

For the limited choice of topic treatment, the mean for the narrative task (M = .49) was

Table 60 Descriptive Statistics for Complexity (Type-token Ratio) for the Decision-making Task Treatment DT NT DMT M .42 .48 .45

SE .02 .03 .02

95% CI Low .38 .42 .41 High .45 .54 .49 SD .08 .14 .09

Minimum .27 .24 .32

Maximum .56 .72 .72

Skewness -.29 .51 1.43

SES .50 .50 .50

Kurtosis -.60 -.64 2.22

SEK .97 .97 .97

N 21 21 21 Notes. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

176 the highest, the next highest mean was for the decision-making task (M = .48), and the lowest mean was for the descriptive task (M = .41). Lastly, for the complete choice of topic treatment, the same pattern was repeated; the mean for the narrative task (M = .51) was highest, the mean for the decision-making task (M = .45) was the next highest, and the mean for the descriptive task (M = .42) was the lowest.

The main effect of Task, F(2, 40) = 7.74 was statistically significant (p < .017) (See

Table 57). However, the interaction effect between Task and Choice F(4, 80) = .74 was not statistically significant. Also, as shown in Table 57, there was high observed power (.93) for the Task main effect as well as a strong effect size (.28).

5BTLUZQF

%FTDSJQUJWF Š /BSSBUJWF

$PNQMFYJUZ 5ZQF5PLFO3BUJP %FDJTJPO .BLJOH ş /POF-JNJUFE$PNQMFUF -FWFMPG$IPJDF

Figure 14. Profile plot of Complexity (type-token ratio) by task type.

177 Examining the profile plot for Complexity by task type (Figure 14), the narrative

and decision-making tasks showed high levels of complexity when compared to the

descriptive task. Although both the descriptive and the narrative tasks maintained higher

levels of complexity with more choice, the decision-making task was lower in complexity

when there was a complete choice of topic.

To break down the interaction between task types, pairwise comparisons with

a Bonferroni correction were requested. In this comparison, the narrative task and

decision-making task were significantly higher than the descriptive task, at the p < .05

significance level. However, there was no statistically significant difference between the

narrative and decision-making tasks.

Within-samples t-tests between pairs of treatments were conducted in order

to determine where specific differences were. These tests were conducted in sets of

three, with the same type of task but with different levels of choice. To protect against

committing a Type I error, a Bonferroni correction with a probability level of p < .017 (α

= .05/3) was used. For the no choice of topic with different types tasks, the narrative task

was significantly higher than the descriptive task M( diff = 0.07, SD = .09, t(20) = 3.67), and the decision-making task was significantly higher than the descriptive task M( diff = 0.07,

SD = .09, t(20) = 3.60). For the limited choice of topic with the different types of tasks, the narrative task was again significantly higher than the descriptive task M( diff = 0.79, SD =

.13, t(20) = 2.67). For the complete choice of topic with the different types of tasks, there

were no statistically significant differences between the tasks, but the narrative task was

178 close to being significantly higher than the descriptive task M( diff = .08, SD = .15, t(20) =

2.55, p = 0.19 (2-tailed)). There were no statistically significant differences for the complete level of topic choice between the decision-making task and the descriptive task.

Research Question 5

The fifth research question concerned the effect of choice onFluency (word count). Research Question 5 was: To what degree does the level of fluency change across the three levels of choice (i.e., no choice, limited choice, and complete choice)?

This question was answered by inspecting the descriptive statistics, the results of the repeated-measures ANOVA, profile plots from the analysis, pairwise comparisons with a

Bonferroni correction to investigate differences in the levels of choice and, t-tests showing significant contrasts between the individual treatments.

Tables 61 to 63 show the descriptive statistics for Fluency for each of the different levels of choice according to the different tasks. A perusal of the descriptive statistics shows that for the descriptive task, the mean for the limited choice of topic treatment (M

= 83.57) was higher than either the mean for the no choice of topic treatment (M = 79.29) or the complete choice of topic treatment (M = 68.57). The mean for the narrative task was highest for the complete choice of topic treatment (M = 68.52) followed by the mean for the limited choice of topic treatment (M = 68.00) and the mean for the no choice of topic treatment (M = 62.43). The same pattern followed for the decision-making task as the mean for the complete choice of topic treatment (M = 80.00) was the highest, followed

179 Table 61 Descriptive Statistics for Fluency (Word Count) for the No Choice of Topic Treatment DT NT DMT M 79.29 62.43 57.86

SE 4.91 4.54 3.47

95% CI Low 69.05 52.96 50.61 High 89.52 71.90 65.10 SD 22.48 20.80 15.91

Minimum 41 34 33

Maximum 145 105 94

Skewness 1.41 .38 .32

SES .50 .50 .50

Kurtosis 2.97 -1.06 -.20

SEK .97 .97 .97

N 21 21 21 Note. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

by the mean for the limited choice of topic treatment (M = 62.33) and the mean for the no choice of topic treatment (M = 57.86).

Each operationalization for the Fluency dependent variable was applied to a repeated-measures ANOVA analysis. As written above, unaltered repetitions (ε = .469) and word count (ε = .682) were used to gauge fluency. Of these two, even though neither violated the assumption of sphericity, according to the Greenhouse-Geisser correction,

180 Table 62 Descriptive Statistics for Fluency (Word Count) for the Limited Choice of Topic Treatment DT NT DMT M 83.57 68.00 62.33

SE 5.63 4.18 5.48

95% CI Low 71.83 59.28 50.90 High 95.32 76.72 73.77 SD 25.80 19.15 25.12

Minimum 38 43 26

Maximum 123 117 115

Skewness .14 1.11 .29

SES .50 .50 .50

Kurtosis -1.20 1.19 -.89

SEK .97 .97 .97

N 21 21 21 Note. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

word count (ε^ = .869) was closer to sphericity than unaltered repetitions (ε^ = .696).

Therefore, word count was used to gauge fluency.

The results from the analysis for Fluency are shown in Table 64. Mauchly’s test indicated that the assumption of sphericity had been met for the main effect of Task

(ε = .922), the main effect of Choice (ε = .972), and the interaction effect between Task and Choice (ε = .682). Therefore, no correction of degrees of freedom for sphericity was

181 Table 63 Descriptive Statistics for Fluency (Word Count) for the Complete Choice of Topic Treatment DT NT DMT M 68.57 68.52 80.00

SE 4.81 6.15 4.40

95% CI Low 58.53 55.70 70.82 High 78.61 81.35 89.18 SD 22.06 28.17 20.16

Minimum 31 31 29

Maximum 102 122 121

Skewness -.13 .79 -.54

SES .50 .50 .50

Kurtosis -1.09 -.60 1.13

SEK .97 .97 .97

N 21 21 21 Note. DT = Descriptive Task; NT = Narrative Task; DMT = Decision-making Task. 95% CI = 95% Confidence Interval for the mean.

needed. Correcting for the number of comparisons in Study 2, interactions at the p < .017 significance level were considered significant.

There was a statistically significant interaction effect between Task and Choice,

F(4, 80) = 7.43 (Table 64). This indicates that Choice had different effects on the ratings of Fluency depending on the type of task. However, there was no statistically significant main effect for Choice, F(2, 40) = 1.84. Power was low and the effect size was small for the

182 Table 64 Repeated-Measures ANOVA for Fluency (Word Count) Source df SS MS F η2 β Task 2 4741.46 2370.73 5.34* .21 .81

Error (Task) 40 17774.32 444.36

Choice 2 1219.65 609.83 1.84 .08 .36

Error (Choice) 40 13286.79 332.17

Task x Choice 4 7523.94 1880.98 7.43* .27 .99 Error (Task x 80 20254.95 253.19 Choice) Notes. η2 = effect size (partial eta squared). β = observed power. *p < .017.

Choice main effect. However, for the interaction effect of Task and Choice, the observed

power was high and the effect size was large. The large effect size indicates that the two

independent variables of Task and Choice were strongly related to Fluency.

Examining the profile plot for Fluency by level of choice (Figure 15), a noticeable

lower level of fluency across the tasks for the no choice of topic treatment as well as the

limited choice of topic treatment was evident, although this last level of choice was the

highest of the group for the descriptive task. In general, the treatments with choice were

again higher than the no choice of topic treatment across the different types of tasks.

Despite the very low level of fluency for the decision-making task, choice seemed to

promote fluency in the students’ oral production.

To break down the interaction between the different levels of choice, pairwise

comparisons with a Bonferroni correction were requested. No statistically significant

183 ťŜ

ŤŜ

ţŜ -FWFMPG$IPJDF /POF

'MVFODZ 8PSE$PVOU ŢŜ -JNJUFE

$PNQMFUF šŜ %FTDSJQUJWF/BSSBUJWF%FDJTJPO .BLJOH 5BTLUZQF

Figure 15. Profile plot of Fluency (word count) by level of choice.

differences were found between the different levels of choice amongst the different types of tasks.

.05/3) was used. For the descriptive task with different levels of choice, the limited choice of topic level was significantly higher than the complete choice of topic level M( diff = 15.00,

SD = 21.31, t(20) = 3.23). For the decision-making task with different levels of choice, the complete choice of topic level was significantly higher than the limited choice of topic level (Mdiff = 17.67, SD = 28.85, t(20) = 2.81) as well as the no choice of topic level (Mdiff =

184 22.14, SD = 24.25, t(20) = 4.19). There were no significant differences between the different

levels of choice for the narrative task.

Research Question 6

The sixth and final research question for Study 2 concerned the effect of task on

Fluency (word count). Research Question 6 was: To what degree does the level of fluency

Table 65 Descriptive Statistics for Fluency (Word Count) for the Descriptive Task Treatment DT NT DMT M 79.29 83.57 68.57

SE 4.91 5.63 4.81

95% CI Low 69.05 71.83 58.53 High 89.52 95.32 78.61 SD 22.48 25.80 22.06

Minimum 41 38 31

Maximum 145 123 102

Skewness 1.41 .14 -.13

SES .50 .50 .50

Kurtosis 2.97 -1.20 -1.09

SEK .97 .97 .97

N 21 21 21 Note. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

185 Table 66 Descriptive Statistics for Fluency (Word Count) for the Narrative Task Treatment DT NT DMT M 62.43 68.00 68.52

SE 4.54 4.18 6.15

95% CI Low 52.96 59.28 55.70 High 71.90 76.72 81.35 SD 20.80 19.15 28.17

Minimum 34 43 31

Maximum 105 117 122

Skewness .38 1.11 .79

SES .50 .50 .50

Kurtosis -1.06 1.19 -.60

SEK .97 .97 .97

N 21 21 21 Note. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

change between the three types of tasks (i.e., the descriptive, narrative, and decision-making

tasks)? This question was answered by inspecting the descriptive statistics, the results of the

repeated-measures ANOVA, profile plots from the analysis, pairwise comparisons with a

Bonferroni correction to investigate differences in the levels of choice; and t-tests showing

significant contrasts between the individual treatments.

186 Table 67 Descriptive Statistics for Fluency (Word Count) for the Decision-making Task Treatment DT NT DMT M 57.86 62.33 80.00

SE 3.47 5.48 4.40

95% CI Low 50.61 50.90 70.82 High 65.10 73.77 89.18 SD 15.91 25.12 20.16

Minimum 33 26 29

Maximum 94 115 121

Skewness .32 .29 -.54

SES .50 .50 .50

Kurtosis -.20 -.89 1.13

SEK .97 .97 .97

N 21 21 21 Note. NC = No Choice of Topic; LC = Limited Choice of Topic; CC = Complete Choice of Topic. 95% CI = 95% Confidence Interval for the mean.

Tables 65 to 67 show the descriptive statistics for Fluency for each level of choice

according to the different task types. A review of the descriptive statistics reveals that the

mean for the no choice of topic treatment was highest for the descriptive task (M = 79.29)

followed by the narrative task (M = 62.43) and the decision-making task (M = 57.86). The same

pattern was found for the limited choice of topic treatment. The mean for the descriptive

task was the highest (M = 83.57) followed by the narrative task (M = 68.00) and the decision-

187 making task (M = 62.33). Lastly, the means for the complete choice of topic treatments were

checked. The highest mean was for the decision-making task M( = 80.00) followed by the

descriptive task (M = 68.57) and the narrative task (M = 68.52).

ťŜ

ŤŜ

5BTLUZQF ţŜ %FTDSJQUJWF

'MVFODZ 8PSE$PVOU ŢŜ /BSSBUJWF %FDJTJPO .BLJOH šŜ /POF-JNJUFE$PNQMFUF -FWFMPG$IPJDF

Figure 16. Profile plot of Fluency (word count) by task type.

The main effect of Task, F(2, 40) = 5.34 was significant at the p < .017 significance level

(Table 64). There was also a significant interaction effect between Task and Choice, F(4, 80) =

7.43. This indicates that Choice had different effects on ratings of Fluency depending on the type of task. Also, as shown in Table 64, the interaction effect of Task and Choice had low observed power and a small effect size. However, as with the previous dependent variable, the main effect of Task had high observed power (.81) and a large effect size (.21).

188 Examining the profile plot for Fluency by task type (Figure 16), the descriptive task

maintained high levels of fluency across the different levels of choice, even though fluency

was lower for the descriptive task when there was complete choice of topic. Conversely, the

decision-making task displayed higher levels of fluency as more choice was available and was

highest for the complete level of choice. The narrative task was at a level somewhat between

the other two types of tasks. Each of the tasks was higher in fluency with a limited level of

topic choice compared to the no choice of topic level.

To break down the interaction between task types, pairwise comparisons with a

Bonferroni correction were requested. The descriptive type of task was significantly higher

than the narrative task at p < .05. However, there were no other significant differences between

the different task types.

Within-samples t-tests between pairs of treatments were conducted in order to

determine where specific differences were. These tests were conducted in sets of three, with

the same type of task but with different levels of choice. To protect against committing a Type

I error, a Bonferroni correction with a probability level of p < .017 (α = .05/3) was used. For the

no choice of topic, the descriptive task was significantly higher than the narrative task M( diff

= 16.86, SD = 22.54, t(20) = 3.43) as well as significantly higher than the decision-making task

(Mdiff = 21.43, SD = 18.99, t(20) = 5.17). For the limited choice of topic, the descriptive task was

significantly higher than both the narrative task M( diff = 15.57, SD = 23.94, t(20) = 3.05) and the decision-making task (Mdiff = 21.24, SD = 33.06, t(20) = 2.94). Lastly, there were no statistically

significant differences for the complete level of topic choice among the different types of tasks.

189 CHAPTER 7

DISCUSSION

This chapter is a review of the findings as related to the research questions and the research hypotheses. Following a summary of the results of each research question, interpretations of the results are offered.

A unique contribution of this study was that two influential areas from two fields, motivation and task-based language teaching, were incorporated in the design in order to study the complementary nature between them. From the area of human motivation, the theory of self-determined behavior was integrated through the use of choice. From the area of foreign or second language teaching, the method of teaching that uses tasks as the central unit of analysis, task-based language teaching, was the second pillar of this study.

The interaction between the two provides useful benefits for teaching and motivating students learning a foreign language. To the best of my knowledge, no other study has effectively combined the two.

Study 1

Research Question 1

Research question 1 asked to what degree the level of Task Interest changed

across three levels of choice (i.e., no choice, limited choice, and complete choice)? The

hypothesis accompanying this research question was that Task Interest will increase

190 significantly when limited or complete choice are available. This hypothesis is based on

studies comparing the presence and absence of choice when adults are engaged in a task.

However, motivation may decrease when complete choice is implemented. (e.g., Iyengar &

Lepper, 2000; Schwartz, 2004a, 2004b)

The hypothesis was supported. The degree of Task Interest shown for the limited

level of topic choice was significantly higher than that for either the no or the complete

levels of topic choice. This result is in agreement with previous studies showing that when

adults are able to choose an aspect of the task, such as the order of the sub-tests of a larger

test (Stotland & Blumenthal, 1964), the choice of a puzzle to complete (Zuckerman et al.,

1978), or when elderly participants in a home could choose the conditions of a university

student’s visit (Schulz, 1976), there was an increase in the individuals’ motivation to

engage in the task. The common thread running through these studies is that when

people feel in control of their environment, they typically have greater motivation to

complete the task. One effective way to induce this sense of control, as these and other

studies have shown (e.g., Corah & Boffah, 1970; Geer et al., 1970; Monty et al., 1973; Rodin

& Langer, 1977), is through the introduction of choice. In this study, the participants

evidenced feelings of increased control through the limited level of task topic choice.

While limited choice can produce positive effects, too much choice exerted negative

effects on Task Interest; the complete level of topic choice showed the lowest level of Task

Interest across the three tasks. Although this result may be counter to the belief that “more is better”, it may arise from at least two sources. The first concerns increased task demands

191 resulting from the availability of a large number of choices. The complete level of task topic asked more of the students in completing the task, especially the narrative task, which required Student A to think of a story, decide on the order of the events in the story, and then tell that story in English. Also, because Student B’s task was to write an outline of the story, Student A had to ensure that Student B was writing it properly.

A second explanation for the low level of interest for the complete choice of topic level may be psychological. This assertion emanates from the literature on choice by researchers such as Iyengar and Lepper (2000) and Schwartz (2004a, 2004b). Iyengar and Lepper discovered that people with many items to choose from experienced more frustration in the decision-making process and found the task more difficult, compared with those who had fewer choices, even though they enjoyed having a large amount of choice available. In their study, there was a more positive outcome (e.g., a participant buying a jar of jam) for limited choice (six possible choices) than for a large number of choices (30 possible choices). These authors concluded that too much choice can be demotivating and they speculated that there is a greater probability of people feeling unhappy with their choice when many options are available. Moreover, people may feel burdened by the responsibility of distinguishing good from bad decisions when that decision is based on a large number of choices (p. 1004).

Schwartz also suggested that too much choice could be accompanied with regret because as the number of choices increases, people’s motivation to choose may decrease

192 due of their lack of skill in dealing with the large number of choices. According to

Schwartz (2004b), greater choice:

1. Increases the burden of gathering information to make a wise choice. 2. Increases the likelihood that people will regret the decisions they make. 3. Increases the likelihood that people will anticipate regretting the decisions they make, with the result that they can’t make a decision at all. 4. Increases the feeling of lost opportunity. 5. Increases the expectation about how good a chosen option will be. 6. Increases the chances that people will blame themselves when their choices fail to live up to expectations. (p. 6)

The results of this study support Schwartz’s hypotheses. When faced with the complete

choice of treatment, the participants' desire to make a wise choice could have increased

the pressure on them to, for example, produce a story that they felt was good enough to

tell. In addition, although all students chose a story to tell, it is possible that some of them

regretted their choice and that this feeling decreased their interest to engage in the task.

Lastly, it is also possible that some students could have blamed themselves afterwards if

their story failed to meet their or their partner’s expectations.

In summary, choice was very influential in promoting the participants’ interest

in engaging in the tasks used in this study. The level of choice with the most teacher

guidance, limited choice, resulted in the highest level of Task Interest. This finding may provide useful guidance for task designers, teachers, and cross-cultural motivational researchers.

193 Research Question 2

Research Question 2 concerned the degree that the level of Task Interest changed for the three task types (i.e., the descriptive, narrative, and decision-making tasks).

The hypothesis accompanying this research question was that the descriptive task may engender more interest because it is a more familiar task that better matches the participants’ English proficiency.

This hypothesis was supported. The pair-wise comparisons indicated that

Task Interest for the descriptive task was significantly higher than that for either the narrative or the decision-making tasks. The descriptive task was significantly higher than either of the other types of tasks for both the no and complete choice of topic, but not for the limited choice of topic. This result may partly be a reflection of the students’ familiarity with descriptive tasks. The vocabulary and the language functions used in this task, describing the locations of objects and describing people, had been taught to the participants in past secondary school classes. First, as required by government standards (Ministry of Education, Culture, Sports, and Technology, 1998), Japanese students are exposed to vocabulary and schema to complete the descriptive task as early as the first year of their study of English in junior high school (e.g., Sano et al., 2006).

In contrast, the other two tasks, telling a story and making a decision, may have been relatively unfamiliar, even though schema-building activities and vocabulary tasks were provided. Because familiarity is a feature of a task that generally decreases task difficulty,

194 the participants’ familiarity with the description task may have positively affected their interest to engage in the task.

Second, Brown and Yule (1983) claimed that descriptive tasks are simpler than narrative and decision-making tasks. If this is true, they may have better matched the proficiency level of the participants in this study, and as a result, they may have felt more interest while engaging in the descriptive task. For individuals with limited English oral proficiency, task difficulty is an important issue. The descriptive task was possibly a task in which task difficulty and task familiarity were at optimal levels.

Summary of the Discussion for Task Interest

In summary, Task Interest was greater for all three types of tasks when limited choice was part of the design than when it was not. This was particularly true for the descriptive task and the narrative task, which showed statistically significant increases in Task Interest from the no choice of topic to the limited choice of topic. This finding is important because these tasks were the most similar in terms of their design; the materials were taken from the same book and calibrated to have equal levels of difficulty.

By removing moderating variables as much as possible, topic and degree of choice were the only remaining differences between the two treatments. Thus, it is likely that simply by introducing a limited degree of choice, the students felt more interest in completing a task.

One of the aims of this study was to examine the effects of choice upon Task

Interest from a cross-cultural perspective. As stated above, the research of Iyengar and

195 colleagues (e.g., Iyengar & DeVoe, 2003; Iyengar & Lepper, 1999, 2002; Iyengar et al., 1999) revealed that some children from Asian cultures defer to or value highly the opinions of others when making a choice and therefore have less motivation for completing tasks that they chose themselves. This finding implies that many children from Asian cultures value guidance from another person more than autonomy. This idea has also been extended to

Asian university students by researchers such as Nakata (2004).

The results of this research, that choice can cause higher levels of Task Interest for some individuals from Asian cultures and for young adults, shed new light on a more dynamic value of choice demonstrated by Asian students. The participants in this study were first-year students at a university in a rural area of Japan. Many of these students are living away from home for the first time. With this new experience, these participants may be exhibiting greater feelings of autonomy because of the requirement that they make more decisions and choices on their own to meet the daily demands imposed by school work and living alone.

Stotland and Blumenthal (1964) and Zuckerman et al. (1978) also studied university-aged participants. Stotland and Blumenthal found that a group of 66 university students displayed significantly less anxiety (measured by palmar sweat) when they could choose the order of a series of tests than those who had no choice. Indeed,

Noels et al. (1999) found that, of 78 Anglophone participants learning French in an immersion course in Canada for six weeks, those with identified regulation and intrinsic motivation showed a statistically significant negative correlation with class anxiety.

196 Zuckerman et al. (1978) found that 80 undergraduates in a university in New York had

significantly higher levels of intrinsic motivation (measured as longer time on task) when

they could choose three (of six) puzzles to complete than participants who could not

choose the three puzzles.

Research Question 3

Research question 3 asked to what degree the level of Task Self-efficacy changed

across the three levels of choice. The hypothesis accompanying this research question was

that Task Self-efficacy would increase significantly when more choice is available. This is

based on studies comparing the presence and absence of choice when adults are engaged

in a task and that more control of the environment increases the ability to do a task (e.g.,

Monty et al., 1973; Stotland & Blumenthal, 1964). However, because Task Self-efficacy is an affective construct, as is Task Interest, it was hypothesized that the complete level of

choice would be a lower level of feelings of Task Self-efficacy.

This hypothesis was supported. Similar to the results for Research Question 1, the participants felt significantly more Task Self-efficacy to complete a task when choice

was introduced, according to the pair-wise comparisons. In this case as well, a limited

level of choice resulted in increased levels of positive affect. Previous research involving

choice had similar results. In Monty et al. (1973), the participants in the choice treatment

remembered words better on a word association task than the group where choice was

not part of the treatment. In Stotland and Blumenthal (1964), the participants showed

197 greater signs of competence when they could choose the order of some subtests of a larger

test. Each of these early studies showed that some individuals display higher degrees of

competence when choice is introduced.

For the complete level of topic choice, the results were more equivocal. There

was a lower level of Task Self-efficacy for the decision-making task, which may have

been an influence of the design of the task. For this task, the students chose a topic on

an environmental problem, that required a relatively high level of abstract thinking

compared to the other two task types. Thus, the decision-making task with this level of

choice was probably difficult for many of the low-proficiency participants in this study.

For the no choice of topic level, Task Self-efficacy was slightly higher for the

descriptive task than for the other two levels of choice, possibly because of a negative

influence caused by the combination of the presence of choice and limited proficiency. As

Burger (1987) wrote, there is some degree of pressure on the one doing the choosing when

choice is introduced, and this may lead to increased anxiety. This anxiety may be reduced

when choice of a topic is not part of the task implementation, and a lessening of anxiety

may increase feelings of Task Self-efficacy.

The effects of choice on anxiety is an important issue where self-efficacy is concerned, as language learning anxiety is one of the key individual differences that students bring to the foreign language classroom (e.g., Horowitz, Horowitz, & Cope, 1986;

MacIntyre, 1998). MacIntyre, Noels, and Clément (1997) found that anxiety to use the language and quality of speaking performance (accuracy, complexity, fluency, native-like

198 pronunciation, elaboration, and usage of colloquial expressions) mirrored each other.

When anxiety was low, the quality of the participants’ speaking performance improved

to a statistically significant degree. In addition, the participants in the MacIntyre et

al. (1997) study felt more proficient when they felt less anxious. However, a limitation

of MacIntyre et al’s (1997) study was that the participants who were more anxious

underrated their proficiency and those who were less anxious overrated their proficiency;

thus, the students’ perception of their own competence was biased to a degree.

Lastly, there was a statistically significant increase in feelings of Task Self-efficacy for the narrative task with the limited level of topic choice. This finding may have been caused in part by the inclusion of a particular topic used with this task. The topic of children going on a picnic (Heaton, 1966, pp. 37-38, Appendix S) was chosen frequently by the students. This topic seemed to be easier for them to talk about than the other topics, and it is possible that the ease with which the students dealt with this topic was reflected in increased Task Self-efficacy. In this case, the availability of choice allowed the

participants to select the topic that was most attractive to them.

In summary, the participants felt more Task Self-efficacy when a limited level of

choice was incorporated into the syllabus. Although there have been studies examining

the influence choice has upon interest, this study uniquely incorporates self-efficacy as

well. With the limited choice level, not only did the participants feel more Task Interest,

they also felt more Task Self-efficacy when engaging in the task, and perceptions of

increased Task Self-efficacy typically lead to lower levels of anxiety. The potential of

199 limited choice to produce a heightened sense of self-efficacy and reduced anxiety can be exploited by teachers in a wide variety of instructional contexts.

Research Question 4

Research question 4 asked to what degree the level of Task Self-efficacy changed among the three types of tasks. The hypothesis accompanying this research question was that the descriptive task would engender more self-efficacy because it may be a more familiar task that may better match the participants' proficiency level.

This hypothesis was supported. In this case, the participants’ Task Self-efficacy was significantly higher for the descriptive than for the decision-making task across all levels of choice. This may have occurred for at least two reasons. First, as stated above, the descriptive task may be a simpler task to complete than either the narrative task or the decision-making task (Brown & Yule, 1983) and, as a result, this task may have matched the proficiency level of the students better. Second, familiarity with the descriptive task and the schema needed to complete the task may have engendered a relatively high level of Task Self-efficacy. This is an area of language knowledge that the students more than likely studied before entering the university. As noted above in the discussion for

Research Question 2, Japanese Ministry of Education standards (Ministry of Education,

Culture, Sports, and Technology, 1998) dictate that students should have been exposed to the vocabulary and schema needed to complete the descriptive task in junior high school as early as the first year of their study of English (e.g., Sano et al., 2006).

200 The fact that the decision-making task engendered the lowest level of Task Self- efficacy may stem from the design of the task itself. This task required the participants to express their feelings to another student. For some of these students, this may have been an anxiety-provoking experience, and the feelings of anxiety may have caused feelings of reduced self-efficacy. In sum, the combination of a high degree of task difficulty and the anxiety-provoking feature of the decision-making task may have had detrimental effects on the students’ feelings of Task Self-efficacy.

Summary of the Discussion for Task Self-efficacy

In summary, a limited level of choice positively affected the participants’ feelings of Task Self-efficacy. The descriptive task also engendered high feelings of Task Self- efficacy, possibly because of the familiarity of the task. When feelings of Task Self-efficacy are important, choice, in combination with task types that are more familiar to students, may help increase those feelings. As a group, the students felt most self-efficacious when conducting the descriptive task, the least self-efficacious when conducting the decision- making task, and experienced mixed feelings of self-efficacy when engaging in the narrative task.

201 Study 2

The primary purpose of Study 2 was to examine the students’ language production using three output variables of Accuracy, Complexity, and Fluency. In the

following section, the results of Study 2 will be discussed.

Research Question 1

Research Question 1 asked to what degree the level of Accuracy changed across the

three levels of choice (i.e., no choice, limited choice, and complete choice). The hypothesis

accompanying this research question was that Accuracy would increase significantly

when choice is available. This hypothesis was based on studies comparing the presence

and absence of choice when adults are engaged in a task requiring high levels of attention

(e.g., Dember et al., 1992).

This hypothesis received limited support. Although the results were mixed, the

students produced more accurate output when choice was a part of the treatment. For the

descriptive and narrative tasks, accuracy was higher for the complete level of topic choice.

For the decision-making task, accuracy was highest for the limited level of choice.

In the psychological literature, Dember et al. (1992) found that participants were

more vigilant (in detecting bar flashes on a computer screen) when they were told that

they had a choice of a difficult or easy task compared to those who had no choice of the

difficulty of the task. In actuality, the participants were randomly assigned in that study.

Vigilance requires a high level of attention. In this study, if the participants were paying

202 attention more closely when engaging in a self-selected task, they may have monitored the language forms in their output more carefully. Despite this potential increase in monitoring, the participants may have been unable to mobilize their linguistic resources

(e.g., morpho-syntactic, collocational, and pragmatic knowledge) to a degree that was high enough to increase the accuracy of their spoken output compared to when there was no choice of topic. This supposition is supported by the fact that the score for the ratio of error-free clauses exceeded .50 on only one treatment. Thus, the introduction of choice had little effect on spoken accuracy.

Accuracy was highest for the narrative task with complete choice. This may have occurred because one student had to write an outline of the story that the other student told. This feature of that task may have encouraged the speakers to monitor the accuracy of their output. This finding lends limited support to the contention of Willis (1996) and

Skehan and Foster (2001), who suggested that if students are required to make a final product, then they will be pushed to focus more on spoken accuracy.

In summary, greater spoken Accuracy was in evidence for both the limited and the complete choice of topic. The answer for this research question showed mixed results for this level of English proficiency. It is possible that many students at this proficiency level are more concerned with completing the task than concentrating on Accuracy.

203 Research Question 2

Research Question 2 asked to what degree the participants’ level of Accuracy changed between the three types of tasks (i.e., the descriptive, narrative, and decision- making tasks). The hypothesis accompanying this research question was that the participants would produce greater Accuracy with the decision-making task, less accuracy with the descriptive task, and the least accuracy with the narrative task (e.g.,

Skehan & Foster, 1997).

This hypothesis was not supported. The pairwise comparisons indicated that the participants’ Accuracy was not statistically significantly greater for one task over the other. The profile plots also indicated that the results for this research question were mixed. This finding stands in contrast to Foster and Skehan’s (1996) results in which the decision-making task had the second lowest level of accuracy. However, in a follow- up study, Skehan and Foster (1997) reported that their participants produced greater accuracy on the decision-making task than on the narrative task and the personal task, in which the students described situations in the host country (England) that pleasantly or unpleasantly surprised them. The results from Skehan and Foster (1997) more closely agree with the results of this study perhaps because the tasks are more similar.

Be that as it may, the decision-making task had statistically significant greater

Accuracy for the treatment with limited choice, according to the t-test results. This suggests that the introduction of choice provides affective support for producing greater spoken Accuracy with a difficult task, in that when students are more interested in

204 engaging in a task, they may concentrate more on the accuracy of their output. However, this was not evident in an equally difficult task, the narrative task, as well as the simpler descriptive task.

Another reason that the decision-making task evidenced a higher level of

Accuracy may be attributed to the design of the task. This was a two-way task in which the students interacted with each other; both students started with equal levels of knowledge and contributed equally to the completion of the task. The successful completion of two-way tasks may require more accuracy, as it would contribute to the establishment of mutual understanding.

The extremely low level of Accuracy for the narrative task with no choice of topic level may be an effect of either a lack of self-efficacy felt by the students in completing the task or a lack of interest to engage in the task. In Study 1, there were somewhat low feelings of Task Interest and Task Self-efficacy for the narrative task with no choice of topic. The lack of Task Interest for this treatment is evident in the transcripts, which revealed that the students used more sentence fragments and short phrases and fulfilled only the minimum requirements needed to complete the task. The participants did not seem to be motivated enough to expend the effort to speak more accurately.

Research Question 3

Research Question 3 asked to what degree the level of Complexity changed across the three levels of choice. The hypothesis accompanying this research question was

205 that Complexity would increase when more choice is available because when choice is

introduced in the implementation stage of a task, attentional resources may be freed up

and allocated towards producing more complex output (e.g., Dember et al., 1992).

This hypothesis was supported. By the simple inclusion of choice, the students

produced significantly more complex output. As written previously, more complex output

is a possible signal that students are stretching their interlanguage more to meet the

demands of the task.

The reason that Complexity increased when choice was introduced may stem from

an increased utilization of attentional resources in the limited choice treatment. As noted

above, Dember et al. (1992) found that their participants were more vigilant (in detecting

bar flashes on a computer screen) when they were told that they had a choice of a difficult

or easy task. Vigilance also requires a high level of attention. Therefore, in this study, the

attentional resources of the students may have been stimulated to some degree when they

had choice.

Another possible influence upon Complexity may have been an affective component, the participants’ willingness to communicate. MacIntyre et al. (1998) defined the willingness to communicate in a second language as “the readiness to enter into discourse at a particular time with a specific person or persons, using a L2” (p.

547). In Study 1, Task Self-efficacy was higher for treatments with choice. An important component for self-efficacy is self-confidence, which is also an important determinant of willingness to communicate. A higher level of self-confidence may have promoted

206 a willingness to take linguistic risks and to attempt to produce more complex speech.

MacIntyre et al. also proposed that motivation plays a role in willingness to communicate in that more motivated individuals feel a greater sense of self-confidence as well as a greater desire to communicate. Such feelings may underlie greater willingness to use more complex language because of the element of risk that is a part of using more complex language.

More recently, Yashima (2007) has examined willingness to communicate through the self-determined model of motivation. She found correlations of willingness to communicate and intrinsic motivation (.28), as well as the identified (.29) and integrated (.35) types of extrinsic motivation. Because choice is an important part of internalized regulation and intrinsic motivation (e.g., Deci & Ryan, 1985), the provision of choice in this study may have induced higher levels of internally regulated extrinsic motivation and intrinsic motivation, which in turn may have increased the students’ willingness to communicate and use more complex language. Yashima found that the frequency of communication also had positive correlations with intrinsic motivation

(.28), identified extrinsic motivation (.41), and integrated extrinsic motivation (.42).

Enhanced by choice, higher levels of willingness to communicate may have led to more complex output.

Based on the definition provided by MacIntyre et al. (1998), it may be assumed that willingness to communicate affects language learners' tendency to initiate conversation in the L2. According to research conducted by Kormos and Dörnyei

207 (2004), willingness to communicate may also engender more complex output. Kormos and Dörnyei found that the learners in their study with positive task attitudes displayed highly significant correlations between willingness to communicate and the number of turns (r = .91, p < .01), a measure of complexity. This result was an affirmation of the results found by Dörnyei and Kormos (2000) and Dörnyei (2002). With choice as part of the treatment, students’ willingness to communicate may be positively influenced and they may therefore be more willing to take risks, which is, as stated by Skehan and Foster

(2001), important for improving complexity.

In summary, spoken Complexity was positively influenced by choice to a significant degree. The finding that choice affected both affective variables, such as interest and self-efficacy, as well as linguistic variables, such as complexity, which require a manipulation of cognitive resources, is unique to this study.

Research Question 4

Research Question 4 asked to what degree the level of Complexity changed between the three types of tasks. The hypothesis accompanying this research question was that learners would produce the greatest Complexity with the narrative task, followed by the decision-making task, and finally the descriptive task (e.g., Foster & Skehan, 1996).

This hypothesis was partly supported. The profile plot indicated that the participants produced the greatest Complexity when engaged in the narrative task. Both

208 the narrative task and the decision-making task were higher than the descriptive task to a statistically significant degree.

The findings from Foster and Skehan (1996), Skehan and Foster (1997), and

Robinson (2001a) were supported to a certain extent by the above results. Foster and

Skehan (1996) found that the narrative task had the highest level of complexity, the decision-making task the next highest, and the personal information task the lowest.

However, Skehan and Foster (1997) found that while the decision-making task had the highest complexity, the descriptive task had the next highest complexity, and the narrative task had the least complexity.

Robinson’s (2001a) hypotheses were also partially supported by the results of this study. The decision-making task was quite high in complexity throughout the different levels of choice. As Robinson wrote, this task should be higher in complexity because of the added reasoning demands. It was a part of the design of this task in this study that the students interact not in the here-and-now (i.e., the students are looking at the task, solving a problem before them), such as with the narrative task, but in the there-and-then

(i.e., the problem to be solved is a possible future situation or a past problem). In the case of the decision-making tasks in this study, the students were asked to imagine situations that might happen, such as making a home page of a famous person. As these tasks involved there-and-then situations, they were hypothesized by Robinson to facilitate higher levels of complex output.

209 A common theme for this type of task in past studies has been that students are

required to make moral judgments for dilemmas, such as judging offenders (Foster &

Skehan, 1996), or deciding who to throw out of a balloon so that it will not crash as it

is losing air (Foster & Skehan, 1999). The difference between the decision-making tasks

used in Foster and Skehan’s studies and the tasks in this study is that the tasks used in

this study did not ask moral questions of the participants. For this reason, it is possible

that the tasks used in this study were perceived as less personal and therefore may have

promoted less complex oral output. More personal topics may create greater personal

investment and this in turn may push learners to produce more complex output, as in

Foster and Skehan (1996, 1999).

Research Question 5

Research Question 5 asked to what degree the level of Fluency changed across

the three levels of choice. The hypothesis accompanying this research question was that

Fluency would increase significantly when more choice was available because increases in

Task Interest caused by the introduction of choice can positively affect Fluency.

This hypothesis was partly supported. Although there were no statistically significant differences between the levels of choice, the limited choice of topic was accompanied by degrees of Fluency higher than the no choice of topic across all types of tasks. Overall, there was a slight increase in the number of words that the students used to complete the task when choice was available. This is desirable because by producing

210 more output, language learners can experiment more with the language, test hypotheses, and possibly learn the target language more efficiently.

Affective factors may have exerted an effect on the increase in Fluency. For one, in this study, Task Interest increased when choice was available. A higher level of interest in conducting the task may have influenced the total number of words that the students produced when conducting the task. In addition, Kormos and Dörnyei (2004) found that the learners with positive task attitudes displayed statistically significant correlations between willingness to communicate and the number of words produced (r = .93, p <

.001), a measure of fluency also used in this study. Although Dörnyei and Kormos (2000) claimed that the construct of willingness to communicate did not presume that a person would speak more, but rather that the person would more likely initiate communication, this is in fact what happened in their study. Interest and willingness to communicate are influential determinants of various aspects of task performance and when choice is available, Task Interest and students’ willingness to communicate may increase and this in turn may positively influence linguistic variables such as spoken fluency.

For the decision-making task with the complete choice of topic there was a statistically significant increase in Fluency. This may have been an effect of the task design because the students needed more preparation in choosing a topic, compared to the complete level of topic choice for the descriptive and narrative tasks. As described above, this task had a written component because it seemed that the topics were too difficult without some preparation by the students. It is possible that the influence of the writing

211 assignment increased the word count in subtle ways. A review by Ellis (2005, p. 20) on the effects of strategic planning (i.e., planning where the students can prepare for the task with access to the actual task materials) on fluency revealed that strategic planning can lead to increases in fluency (i.e., more words in the output with fewer pauses).

In summary, choice exerted a limited effect on the participants’ spoken Fluency.

It may be the case that Fluency is less amenable than Accuracy or Complexity to changes due to the introduction of choice and increased Task Interest. However, if a teacher desires greater word count, a descriptive task with limited choice of topic may be beneficial in producing greater oral fluency. Be that as it may, open-ended tasks, such as the narrative and descriptive tasks for the complete choice of topic, should encourage students to produce relatively high word counts when compared to the limited and no choice of topic treatments for these same tasks.

Research Question 6

Research Question 6 asked to what degree the level of Fluency changed among the three types of tasks. The hypothesis accompanying this research question was that

Fluency would be highest for the decision-making task because it is a two-way task in which both parties contribute equally to the completion of the task.

This hypothesis was not supported. According to the t-test results, the descriptive task was significantly higher than both the narrative task and the decision-making task

212 when the no and limited levels of topic choice were implemented. However, there were no differences amongst the three tasks for the complete choice of topic treatment.

It is difficult to compare the measure of fluency used in this study, total word count, with previous research results. As stated previously, pauses and silence are common measures for assessing spoken fluency (e.g., Foster & Skehan, 1996; Skehan

& Foster, 1997). Foster and Skehan (1996) found that, using pauses, their participants produced the most fluent language when they engaged in the personal task, and they produced the lowest levels of fluency when engaged in the decision-making task. Skehan and Foster (1997) found that the most fluent output was produced in the narrative task, and the least fluent output was produced in the decision-making task.

It was hypothesized that the word count for the decision-making task would be highest because that task required both participants to speak (e.g., Long, 1989).

However, this was not supported possibly because topic familiarity again played a role in creating higher degrees of fluency for the descriptive task. Indeed, Foster and Skehan

(1996), Skehan (2001), and Skehan and Foster (1997) all found that fluency was positively affected by the participants’ familiarity with the task and the content of the task, but these differences were not statistically significant. Skehan (2001), for example, found that familiar information (operationalized as a personal task) led his participants (32 pairs in Study 1) to produce significantly more fluent output than unfamiliar information, but this was not the case for the participants (40 pairs) in Study 2. In summary, topic

213 familiarity may have played a part in increasing fluency for the descriptive task, but past research does not support this claim.

General Discussion of Study 2

In summarizing the research by Skehan and Foster (e.g., Foster & Skehan, 1996,

1999; Skehan & Foster, 1997, 1999, 2001, 2005), and Robinson (e.g., 1995, 2001a, 2001b,

2003) on the differences of accuracy, complexity and fluency in relation to task types, I wrote:

1. The less difficult a task is, the more fluent the performance will be. 2. More difficult tasks increase the complexity of learners’ utterances. 3. Tasks that promoted learner accuracy were less difficult, so accuracy seems to be more dynamic and unpredictable.

First, the results for the descriptive task, which is often presumed to be a relatively easy task (e.g., Foster & Skehan, 1996; Skehan & Foster, 1997), generally supported the first and third suppositions. The participants produced the most fluent performance on the descriptive task for two out of the three choice treatments. In addition, choice helped to promote Accuracy for the more difficult decision-making task. Second, in the case of

Complexity, the descriptive task was also the least complex for all choice treatments in this study. This result matched the above findings from previous research as well. Lastly,

Foster and Skehan (1996) and Skehan and Foster (2001) stated these relations between tasks and output performance:

1. The least personal exchange (i.e., the least difficult task in this study) task has the highest degree of accuracy but little complexity. 2. The narrative task has a high level of complexity but less accuracy.

214 3. The decision-making task is between the personal exchange and the narrative tasks in the case of accuracy, complexity, and fluency. 4. There is a trade-off between accuracy and complexity dependent on the difficulty of the task.

Concerning the first point, the results of this study for Accuracy were too mixed to arrive at any strong conclusions. Concerning the second point, the narrative task and the decision-making task engendered high levels of Complexity across all levels of choice.

Concerning the third point, the decision-making task was very high or very low for each of the performance measures and, considered together, the decision-making task seems to occupy a middle area between the descriptive task and the narrative task in regards to Accuracy, Complexity, and Fluency, as Skehan and Foster (2001) suggest. Lastly, there was limited trade-off between Accuracy and Complexity for the narrative task only.

Examining the profile plots for accuracy (Figure 11) and complexity (Figure 13), the narrative task is almost a mirror of itself between the two dependent variables. Where

Accuracy is lower in these plots, Complexity is greater. For instance, Accuracy increased for the decision-making task and there was a decrease in Complexity. Therefore, Skehan and Foster’s contention was partially supported. In this case, choice was influential in supporting higher levels of Accuracy, Complexity, and Fluency and tasks hypothesized to be less complex evidenced greater levels of Complexity when choice was part of the treatment. This shows that choice is an influential factor for creating higher levels of

Complexity, regardless of the type of task.

215 CHAPTER 8

CONCLUSION

Choice consistently promoted higher levels of Task Interest and Task Self-efficacy, and Accuracy, Complexity, and Fluency in this study. Task Interest and Task Self-efficacy play important roles in foreign language acquisition and they are important issues faced by many foreign language teachers. With the simple introduction of choice, Task

Interest and Task Self-efficacy can be increased. Additionally, Complexity was positively influenced by choice to a statistically significant degree and Accuracy and Fluency were

positively influenced as well. These findings are important not only to teachers, but

also to researchers who are interested in the cognitive processes that students engage

in and how attention can be freed up for use in other areas of output. With choice, a

wide spectrum of concerns of both teachers and researchers can be manipulated and

improved.

Integrating Task-type and Choice

This section concerns finding the intersection of choice and task-type. Where

do the twain meet for the best results? Which task with what level of choice promotes

increased Task Interest and Task Self-efficacy as well as increased Accuracy, Complexity,

and Fluency?

216 With regard to the types of tasks used in this study, the narrative task was often positively influenced by choice. When engaged in the narrative task with limited choice of topic, the participants displayed the highest level of Task Interest, very high Task Self- efficacy, the greatest Accuracy when combined with complete choice of topic, and the greatest Complexity when combined with limited choice of topic. In addition, Accuracy was strongly affected by the introduction of choice with this task. The narrative task seems to be a very vibrant task type and when combined with choice, can provide an efficient method for improving students’ foreign language speaking abilities as well as their motivation. Other promising task-autonomy combinations were the descriptive task with no choice of topic for Task Self-efficacy, the decision-making task with limited topic choice for Complexity, and the decision-making task with complete choice of topic for Fluency.

Limitations of This Study

There were four main limitations of this study: the number of participants, the assignment of the participants to the classes, the lack of randomization in some cases of task implementation, the design of the tasks, and the procedures used during factor analysis for Study 1.

First, the number of the participants was minimal for some of the analyses used in this study. A larger number of participants would have been desirable for conducting the factor analysis in Study 1. According to Good and Hardin (2003), small samples may

217 give a distorted view of the population (p. 6). As sample size increases, the sample tends to more closely resemble the population.

Second, the participants were in intact groups. Although Adams (2007) stated that intact groups are more pedagogically realistic, randomized groups would have been preferable because randomization is an assumption of true experimental designs.

Without randomization, generalizations are difficult to make with confidence (Hatch

& Lazaraton, 1991), and it is difficult to attribute causality to an independent variable

(Tabachnick & Fidell, 2007b).

Third, randomization could have been introduced for the implementation of the tasks. For Group A in Study 1, the treatments were introduced using a Latin square design. That is, the same types of tasks were implemented over three consecutive treatment sessions but the implementation of choice was staggered. Some students recognized that choice was the focus of this study and this realization could have affected their answers on the survey. This design was unavoidable, however, for Group A because syllabus restrictions required that schema-building activities for one type of task be conducted ongoing with data collection treatments for other types of tasks. That is, one half of the class time was set aside for schema-building activities for one type of task and the other half of the class was devoted to data collection for another type of task.

In contrast, the treatments were randomly implemented with Group B, which made up about half of the participants in Study 1 and all of the participants in Study 2.

218 The fourth limitation was the design of the tasks. Although the descriptive task and the narrative task for the no choice of topic and limited choice of topic treatments were taken from the same sources, there is no definite assurance that there were no differences in difficulty between them. This might especially be true of the limited choice of topic treatment for the narrative task (i.e., children going on a picnic (Heaton, 1966, pp. 37-38,

Appendix S)), which was overwhelmingly chosen by students in both studies, but which was especially ubiquitous in Study 2. This seemed to be an easier topic to complete and this may have affected the results of Study 2. Randomizing the selection of task topics might have helped alleviate this limitation. However, to compare the groups more consistently, I felt it better that each group had the chance to choose from the same set of topics.

The fifth and last limitation concerns some procedures used for the data analysis, especially those used for the factor analysis for Study 1. With nine treatments and so few participants, it would have been unrealistic to expect the item loadings for both the Task

Interest factor and the Task Self-efficacy factor to strongly and consistently load onto the same factor for all nine treatments. The criterion for including an item on one or the other factor was if that item loaded on that factor in five out of the nine treatments. Considering the criterion, most of the items performed reasonably well. However, it also meant including an item into a factor for treatments in which it had a poor loading. For example,

Item 3, an item included as part of the Task Self-efficacy factor, loaded on that factor at

.04 for the decision-making task with no choice of topic treatment. On the other hand, at certain times an item was not included as part of a factor even when it loaded strongly on

219 that factor. The above-mentioned Item 3 for the same treatment loaded on the Task Interest

factor at .89. In this case, this item contributed little information about Task Self-efficacy but could possibly have contributed a great deal of information about Task Interest.

Suggestions for Further Research

Five suggestions for further research can be made based on the results of this study. First, further research on the effect of choice should be conducted. Would students from different cultures or with differing levels of proficiency compared to the participants in this study, experience the same increases in Task Interest as these students evidenced when choice was introduced? Cross-cultural research into this question has implications not only for foreign language curriculum design, but also for cross-cultural motivational research and educational motivational research. In addition, research investigating how students feel about choosing and the process that they use to choose would be fruitful.

Second, the results concerning Task Interest for the complete choice of topic level suggest that qualitative approaches are needed in order to illuminate the processes that students use when choosing topics and the feelings that they have towards their choices.

For instance, while Iyengar and Lepper (2000) found that their participants enjoyed having more choices to select from, feelings such as regret may also influence Task

Interest. Qualitative research methods, such as participant interviews might shed light on such issues.

220 Third, research designs combining different types of planning with choice may prove fruitful. For example, with planning, accuracy is often improved (e.g., Ellis, 2005).

Would choice lead to further increases in accuracy? Or would there be no difference between, for example, a treatment with planning and no choice and a treatment with choice and no planning? What influence affect has upon other frequently researched aspects of task-based language teaching, such as planning, could provide an additional dimension to task-based language teaching.

Fourth, this study dealt with oral communication. However, research into the effect of choice on other modes of communication in a foreign language, such as listening, reading, and writing, is also needed, as motivating students in all language skills would be beneficial to everyone involved in the educational process.

Fifth, further research is needed in order to more efficiently bring choice to the task-based syllabus. This would help teachers to improve students’ motivation and oral output. Some of the tasks in this study required a great amount of preparation on the part of the teacher. This was especially true for the narrative task with the limited choice of topic. For busy teachers, long preparation for a class is sometimes impossible. Research in developing a format for the efficient introduction of choice into the class is needed. It may be that methods of introducing choice into the classroom via the computer may be more viable than methods utilizing printed formats.

221 Implications of This Study

First, no matter what the task, if a modicum of choice was introduced, the students felt more Task Interest and Task Self-efficacy. Thus, the introduction of choice provides teachers with a simple method of improving students’ Task Interest and Task

Self-efficacy. However, this is only true up to a certain point. Students are students and more work for them, as in the complete choice of topic treatments, may lead to lower levels of Task Interest. The implication of this is that students, through a limited set of pre-selected topics, should be given guidance in their choices and through this guidance,

Task Interest and Task Self-efficacy can be increased.

A second implication of this study is that it offers further support for the self- determination theory (SDT) of motivation as it applies to language learning (e.g.,

Noels, 2001; Noels et al., 2000, 2001). A core aspect of self-determined motivation, the internalized locus of causality, was used in this study. Choice can be a powerful promoter that leads individuals to internalize their behavior and choice is required for promoting autonomy. Moreover, the type of autonomy used in this study was student-based at the task level. Noels’, for example, has looked at the larger picture, comparing SDT with other hypotheses of language motivation (Noels et al. 2000) or a participant’s level of intrinsic motivation, for which autonomy is important, with class outcomes, such as grades, self- evaluation of ability, or continuing the course, as well as students’ perceptions of the teacher having a controlling or autonomy-supporting teaching style, a common theme in the educational psychological literature (Noels et al. 1999). Additional research focused

222 on the students’ perspectives and how they feel about choice, could also provide support for the viability of SDT in language learning motivational research.

A third implication of this study concerns task design. If increased Task Interest created through the provision of choice increases learners' attentional resources, task designers can manipulate these resources so that they can be used for other forms of output. With the same task but with differing levels of choice, task designers can add a new dimension when the consideration is, for example, task complexity. Through choice, complexity can be increased with no loss to Task Interest. Robinson (2001a) stated that more complex tasks are more difficult, but some teachers may not wish to increase complexity through manipulating the task design if it would increase the task’s difficulty.

However, with the provision of choice, even relatively difficult tasks may not adversely affect students’ interest in the task or their sense of self-efficacy.

The fourth implication concerns adding a new dimension to task implementation, in addition to that of pre-task planning (Ellis, 2003). There are various types of pre-task planning, such as strategic planning (Ellis, 2005), detailed planning, and undetailed planning (Foster & Skehan, 1996). As demonstrated in this study, a new procedure involving task topic choice can be utilized. The results of this study suggest that complexity can be increased through choice and students’ interest in the task and feelings of self-efficacy may also be improved at the same time. Being able to improve linguistic and affective factors simultaneously would be beneficial for teachers and students.

223 The fifth implication concerns introducing choice into the foreign language curriculum. As in this study, the introduction of choice can be introduced gradually, possibly preventing a backlash due to conflicts with cultural norms (e.g., Jones, 1995).

Introducing choice at the task implementation stage is a subtle method of promoting autonomy and may be simple to implement and non-intrusive. However, as implied through the low level of Task Interest and Task Self-efficacy for the compete choice of topic treatment, students will require some guidance in their choices. This guidance may help relieve the burden of choosing when a large number of choices are available and reduce the regret students may feel with their choices. Choice can be offered in the classroom as well as the self-access center.

The sixth implication is in the area of cross-cultural motivational studies. Iyengar and her colleagues have stated that children from Asian cultures may not value the concept of choice very highly and when choice is introduced via a task, there is little or no increase in motivation and the participants will sometimes defer the choice to a caretaker or a friend (e.g., Iyengar, & DeVoe, 2003; Iyengar & Lepper, 1999, 2002; Iyengar et al, 1999; Sethi,

1997). However, an implication of this study is that mature teenage Asian students value choice and will be more interested in the task if choice is made available in the curriculum.

A final implication concerns a possible connection between affect and its role in task-based language teaching. Up to now, research into motivation in the task-based language teaching area has been sparse and somewhat speculative; some researchers claim that task-based language teaching is more motivating compared to other methods

224 (e.g., Dörnyei, 2002; Dörnyei & Kormos, 2000; Julkunen, 1989, 2001; Kormos & Dörnyei,

2004). According to Dörnyei (2002), with a good work plan based on the concepts of task- based language teaching, students will exhibit more motivation in comparison with other teaching methods. I believe that this study adds an additional dimension to the study of affect in task-based language teaching through its examination of the role of interest and self-efficacy and the spoken output produced in particular tasks. From this, a more dynamic picture of task-based language teaching is possible.

Conclusion

The primary goal of this study was to examine the effect of choice upon (a) affective variables taken from the psychological literature on human motivation as proposed in the self-determination model and (b) linguistic variables that are concerned with the language that students produce (i.e., Accuracy, Complexity, and Fluency), taken from the task-based language teaching research literature. In essence, this study was broad in scope, as it incorporated both affective and cognitive variables. Additionally, three task types were compared with each other in terms of their effect on two affective variables, Task Interest and Task Self-efficacy, and three cognitive variables, Accuracy,

Complexity, and Fluency. Researchers such as Robinson, Skehan, and Skehan and Foster have examined the last three variables, and Dörnyei has examined task motivation.

However, this study is an original examination of the effect different types of tasks exert

225 upon Task Interest and is original as well in incorporating perceived Task Self-efficacy

into the design.

Design is one thing but results are another, and the results in this study are

substantial. Choice was beneficial for students in terms of their feelings of Task Interest

and Task Self-efficacy, and for increased Accuracy, Complexity, and Fluency. This result has positive implications for researchers and teachers. With choice, teachers can increase their repertoire of materials for not only improving the language produced by the students, but also for generating more positive levels of affect and for rejuvenating motivation lost in the system.

226 REFERENCES

Abe, N. (2001). Nihongoban SDLRS no kaihatsu to kangokyoiku heno uoyou [Development of Japanese-SDLRS for application in nursing education] (Heisei 11•12nen Kagakukenkyuhijoseikin Kibankenkyu (C) (2) Publication No. 11672317)). Asahikawa, Japan: Author.

Adams, R. (2006). L2 tasks and orientation to form: A role for modality? International Journal of Applied Linguistics, 152, 7-33.

Adler, A. (1930). Individual psychology. In C. Murchison (Ed.), Psychologies of 1930 (pp. 395-405). Worcester, MA: Clark University Press.

Andrich, D. (1978). A rating formulation for ordered response categories. Psychometrika, 43(4), 561-573.

Aoki, N. (1998). Affect and the role of teachers in the development of learner autonomy. In J. Arnold (Ed.), Affect in language learning (pp. 142-154). Cambridge: Cambridge University Press.

Asano, H., Uruno, M., & Rost, M. (1985). Basics in listening: Short tasks for listening development. Tokyo: Lingual House.

Barfield, A., & Nix, M. (Eds.). (2003). Learner and teacher autonomy in Japan 1: Autonomy you ask! Tokyo: The Learner Development Special Interest Group of JALT.

Barnlund, D. C. (1989). Communicating styles of Japanese and Americans: Images and realities. Belmont, CA: Wadsworth.

227 Beck, C. T., & Gable, R. K. (2001). Item response theory in affective instrument development: An illustration. Journal of Nursing Measurement, 9(1), 5-22.

Benson, P. (1997). The philosophy and politics of learner autonomy. In P. Benson & P. Voller (Eds.), Autonomy and independence in language learning (pp. 18-34). Harlow, UK: Longman.

Benson, P. (2001). Teaching and researching autonomy in language learning. Harlow, UK: Longman.

Benson, P. (2006). Autonomy in language teaching and learning. Language Teaching, 40, 21-40.

Benson, P., & Voller, P. (Eds.). (1997). Autonomy and independence in language learning. Harlow, UK: Longman.

Bond, T. G., & Fox, C. M. (2001). Applying the Rasch model: Fundamental measurement in the human sciences. Mahwah, NJ: Lawrence Erlbaum.

Bond, T. G., & Fox, C. M. (2007). Applying the Rasch model: Fundamental measurement in the human sciences (2nd ed.). Mahwah, NJ: Lawrence Erlbaum.

Breen, M. P. (1987). Learner contributions to task design. In C. N. Candlin & D. F. Murphy (Eds.), Language learning tasks (pp. 23-46). Englewood Cliffs, NJ: Prentice-Hall.

Brown, G., & Yule, G. (1983). Teaching the spoken language. Cambridge: Cambridge University Press.

Brown, J. D. (1980). Relative merits of four methods for scoring cloze tests. Modern Language Journal, 64(3), 311-317.

Burger, J. M. (1987). Increased performance with increased personal control: A self- presentation interpretation. Journal of Experimental Social Psychology, 23, 350-360.

228 Bygate, M., Skehan, P., & Swain, M. (2001). Introduction. In M. Bygate, P. Skehan, & M. Swain (Eds.), Researching pedagogic tasks: Second language learning, teaching and testing (pp. 1-20). Harlow, UK: Longman.

Candlin, C. N. (1987). Towards task-based language learning. In C. N. Candlin & D. F. Murphy (Eds.), Language learning tasks (pp. 5-22). Englewood Cliffs, NJ: Prentice- Hall.

Caudill, W. (1963). Patterns of emotion in modern Japan. In R. J. Smith & R. K. Beardsley (Eds.), Japanese culture: Its development and characteristics (pp. 115-131). Chicago: Aldine.

Chirkov, V., Ryan, R. M., Kim, Y., & Kaplan, U. (2003). Differentiating autonomy from individualism and independence: A self-determination theory perspective on internalization of cultural orientations and well-being. Journal of Personality and Social Psychology, 84(1), 97-110.

Corah, N. L., & Boffa, J. (1970). Perceived control, self-observation, and response to aversive stimulation. Journal of Personality and Social Psychology, 16(1), 1-4.

Courtney, M. (2001). Tasks, talk, and teaching: Task-based language learning and the negotiation of meaning in oral interaction (Research Report, Volume 1). Hong Kong: The Hong Kong University of Science and Technology, Language Centre.

Crookes, G. (1986). Task classification: A cross-disciplinary review (Technical Report, Number 4). Honolulu, HI: University of Hawai’i. de Charms, R. (1968). Personal causation: The internal affective determinants of behavior. New York: Academic Press.

Deci, E. L. (1971). Effects of externally mediated rewards on intrinsic motivation. Journal of Personality and Social Psychology, 18(1), 105-115.

Deci, E. L. (1972). Intrinsic motivation, extrinsic reinforcement, and inequity. Journal of Personality and Social Psychology, 22(1), 113-120.

229 Deci, E. L. (1975). Intrinsic motivation. New York: Plenum Press.

Deci, E. L. (1987a). Intrinsic motivation. In R. J. Corsini (Ed.), Concise encyclopedia of psychology (pp. 619-620). New York: John Wiley & Sons.

Deci, E. L. (1987b). Self-determination. In R. J. Corsini (Ed.), Concise encyclopedia of psychology (p. 1015). New York: John Wiley & Sons.

Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human behavior. New York: Plenum.

Deci, E. L., & Ryan, R. M. (2000). The “what” and “why” of goal pursuits: Human needs and the self-determination of behavior. Psychological Inquiry, 11(4), 227-268.

Dember, W. N., Galinsky, T. L., & Warm, J. S. (1992). The role of choice in vigilance performance. Journal of the Psychonomic Society, 30(3), 201-204.

Dickinson, L. (1995). Autonomy and motivation: A literature review. System, 23(2), 165-174.

Doi, T. (1973). The anatomy of dependence. Tokyo: Kodansha International.

Doi, T. (1993). Forward. In F. A. Johnson, Dependency and Japanese socialization: Psychoanalytic and anthropological investigations into amae (pp. ix-xi). New York: New York University Press.

Dörnyei, Z. (1994). Motivation and motivating in the foreign language classroom. The Modern Language Journal, 78(3), 273-284.

Dörnyei, Z. (2001). Teaching and researching motivation. Harlow, UK: Longman.

Dörnyei, Z. (2002). The motivational basis of language learning tasks. In P. Robinson (Ed.), Individual differences and instructed language learning (pp. 137-157). Amsterdam: John Benjamins.

230 Dörnyei, Z. (2003). Attitudes, orientations, and motivations in language learning: Advances in theory, research, and applications. In Z. Dörnyei (Ed.), Attitudes, orientations, and motivations in language learning (pp. 3-32). Malden, MA: Blackwell Publishing.

Dörnyei, Z., & Kormos, J. (2000). The role of individual and social variables in oral task performance. Language Teaching Research, 4, 275-300.

Dörnyei, Z., & Ottó, I. (1998). Motivation in action: A process model of L2 motivation. Working Papers in Applied Linguistics, Thames Valley University, ,4 43-69.

Duff, P. (1986). Another look at interlanguage talk: Taking task to task. In R. R. Day (Ed.), Talking to learn: Conversation in second language acquisition (pp. 147-181). Rowley, MA: Newbury House.

Duff, P. (1993). Tasks and interlanguage performance: An SLA perspective. In G. Crookes & S. Gass (Eds.), Tasks and language learning: Integrating theory and practice (pp. 57-95). Clevedon, UK: Multilingual Matters.

Dumicich, J. (1979). Picture It! Sequences for conversation. New York: Regents Publishing Company.

Dworkin, G. (1988). The theory and practice of autonomy. Cambridge: Cambridge University Press.

Ellis, N. C. (2001). Memory for language. In P. Robinson (Ed.), Cognition and second language instruction (pp. 33-68). Cambridge, UK: Cambridge University Press.

Ellis, R. (2003). Task-based language learning and teaching. Oxford: Oxford University Press.

Ellis, R. (2005). Planning and task-based performance: Theory and research. In R. Ellis (Ed.), Planning and task performance in a second language (pp. 3-34). Amsterdam: John Benjamins.

231 Ellis, R., & Barkhuzien, G. (2005). Analyzing learner language. Oxford: Oxford University Press.

Ellis, R., & Yuan, F. (2003). The effects of planning on fluency, complexity, and accuracy in second language narrative writing. Studies in Second Language Acquisition, 26, 59-84.

Ellis, R., & Yuan, F. (2005). The effects of careful within-task planning on oral and written task performance. In R. Ellis (Ed.), Planning and task performance in a second language (pp. 167-192). Amsterdam: John Benjamins.

Field, A. (2005). Discovering statistics using SPSS. London: Sage.

Fisher, R. A., & Yates, F. (1953). Statistical tables for biological, agricultural and medical research (4th ed.). New York: Hafner.

Foster, P., & Skehan, P. (1996). The influence of planning and task type on second language performance. Studies in Second Language Acquisition, 18, 299-323.

Foster, P., & Skehan, P. (1999). The influence of source of planning and focus of planning and task-based performance. Language Teaching Research, 3(3), 215-247.

Gardner, R. C. (1985). Social psychology and second language learning: The role of attitudes and motivation. London: Edward Arnold.

Gardner, R. C., & Lambert, W. E. (1959). Motivational variables in second language acquisition. Canadian Journal of Psychology, 13(4), 266-272.

Gass, S., & Varonis, E. M. (1984). The effect of familiarity on the comprehensibility of nonnative speech. Language Learning, 34(1), 65-89.

Geer, J. H., Davison, G. C., & Gatchel, R. I. (1970). Reduction of stress in humans through nonverdical perceived control of aversive stimulation. Journal of Personality and Social Psychology, 16(4), 731-738.

232 Geer, J. H., & Maisel, E. (1972). Evaluating the effects of the prediction-control confound. Journal of Personality and Social Psychology, 23(3), 314-319.

Girden, E. R. (1992). ANOVA: Repeated measures. Newbury Park, CA: Sage.

Glass, D. C., Singer, J. E., & Friedman, L. N. (1969). Psychic cost of adaptation to an environmental stressor. Journal of Personality and Social Psychology, 12(3), 200-210.

Good, P. I., & Hardin, J. W. (2003). Common errors in statistics (And how to avoid them). Hoboken, NJ: John Wiley & Sons.

Gorsuch, G. J. (1998). Yakudoku EFL instruction in two Japanese high school classrooms: An exploratory study. JALT Journal, 20(1), 6-32.

Gudykunst, W. B., & Nishida, T. (1994). Bridging Japanese/North American differences. Thousand Oaks, CA: Sage.

Hambleton, R. K., Swaminathan, H., & Rogers, J. H. (1991). Fundamentals of item response theory. Newbury Park, CA: Sage.

Hara, N. (2005, June 10). Reforms cause conflict. The Times Higher Education Supplement, p. 12.

Harumi, S. (2002). The use of silence by Japanese EFL learners. In M. Swanson & D. McMurray (Eds.), On PAC3 at JALT 2001: A language 0dyssey (pp. 27-34). Tokyo: Japan Association for Language Teaching.

Hatch, E., & Lazaraton, A. (1991). The research manual: Design and statistics for applied linguistics. Boston: Heinle and Heinle.

Heaton, K. B. (1966). Composition through pictures. London: Longman.

233 Henning, G. (1987). A guide to language testing: Development, evaluation, research. Boston: Heinle & Heinle.

Ho, J., & Crookall, D. (1995). Breaking with Chinese cultural traditions: Learner autonomy in English language teaching. System, 23(2), 235-243.

Holec, H. (1981). Autonomy and foreign language learning. Oxford: Pergamon Press.

Horowitz, E. K., Horowitz, M. B., & Cope, J. (1986). Foreign language classroom anxiety. The Modern Language Journal, 70(2), 125-132.

Ishii, S., & Bruneau, T. (1988). Silence and silences in cross-cultural perspective: Japan and the United States. In L. Samovar & R. Porter (Eds.), Intercultural communication: A reader (5th ed., pp. 310-315). Belmont, CA: Wadsworth.

Iyengar, S., & DeVoe, S. E. (2003). Rethinking the value of choice: Considering cultural mediators of intrinsic motivation. In R. A. Dienstbier (Ed.), Nebraska symposium on motivation (pp. 129-174). Lincoln, NE: University of Nebraska Press.

Iyengar, S., & Lepper, M. (1999). Rethinking the value of choice: A cultural perspective on intrinsic motivation. Journal of Personality and Social Psychology, 76(3), 349-366.

Iyengar, S., & Lepper, M. (2000). When choice is demotivating: Can one desire too much of a good thing? Journal of Personality and Social Psychology, 79(6), 995-1006.

Iyengar, S., & Lepper, M. (2002). Choice and its consequences: On the costs and benefits of self-determination. In A. Tesser, D. A. Stapel, & J. V. Wood (Eds.), Self and motivation: Emerging psychological perspectives (pp. 71-96). Washington. DC: American Psychological Association.

Iyengar, S., Lepper, M., & Ross, L. (1999). Independence from whom? Interdependence with whom? Cultural perspectives on ingroups versus outgroups. In D. A. Prentice & D. T. Miller (Eds.), Cultural divides: Understanding and overcoming group conflict (pp. 273-301). New York: Russell Sage Foundation.

234 Johnson, F. A. (1993). Dependency and Japanese socialization: Psychoanalytic and anthropological investigations into amae. New York: New York University Press.

Johnson, W. (1944). Studies in language behavior: I. A program of research. Psychological Monographs, 56(1-15).

Jones, J. (1995). Self-access and culture: Retreating from autonomy. ELT Journal, 49(3), 228-234.

Julkunen, K. (1989). Situation- and task-specific motivation in foreign language learning and teaching. Joensuu: University of Joensuu.

Julkunen, K. (2001). Situation- and task-specific motivation in foreign language learning. In Z. Dörnyei & R. Schmidt (Eds.), Motivation and second language acquisition (pp. 29-41). Honolulu: University of Hawai’i Press.

Kamada, L. D. (1987, September). Intrinsic and extrinsic motivation learning processes: Why Japanese can’t speak English. Paper presented at the Japan Association of Language Teachers International Conference on Language Teaching and Learning, Hamamtsu, Japan. (ERIC Document Reproduction Service No. ED285408)

Kelly, L. G. (1969). 25 centuries of language teaching. Rowley, MA: Newbury House.

Kline, P. (1994). An easy guide to factor analysis. London: Routledge.

Kormos, J., & Dörnyei, Z. (2004). The interaction of linguistic and motivational variables in second language task performance, Zeitschrift für Interkulturellen Fremdsprachenunterricht, 9(2). Retrieved October 16, 2005 from http://www.spz. tu-darmstadt.de/projekt_ejournal/jg-09-2/beitrag/kormos2.htm.

Krahnke, K. (1987). Approaches to syllabus design for second language teaching. Englewood Cliffs, NJ: Prentice-Hall.

235 Krashen, S. D. (1982). Principles and practice in second language acquisition. New York: Prentice Hall MacMillan.

Kumaravadivelu, B. (1993). The name of the task and the task of naming: Methodological aspects of task-based pedagogy. In G. Crookes & S. Gass (Eds.), Tasks in a pedagogical context: Integrating theory and practice (pp. 69-96). Clevedon, UK: Multilingual Matters.

Kuwayama, T. (1989). The Japanese conception of self: The dynamics of autonomy and heteronomy. Unpublished Doctoral Dissertation, University of California, Los Angeles.

Langer, E. J. (1975). The illusion of control. Journal of Personality and Social Psychology, 32(2), 311-328.

Langer, E. J., & Rodin, J. (1976). The effects of choice and enhanced personal responsibility for the aged: A field experiment in an institutional setting. Journal of Personality and Social Psychology, 34(2), 191-198.

Larsen-Freeman, D. (2000). Techniques and principles in language teaching (2nd ed.). Oxford: Oxford University Press.

Lebra, T. S. (1987). The cultural significance of silence in Japanese communication. Multilingua, 6(4), 343-357.

Lee, J. F. (2000). Tasks and communicating in language classrooms. Boston: McGraw-Hill.

Lennon, P. (2000). The lexical elements in spoken second language fluency. In H. Riggenbach (Ed.), Perspectives on fluency (pp. 25-42). Ann Arbor, MI: University of Michigan Press.

Linacre, J. M. (1997). KR-20 / Cronbach alpha or Rasch reliability: Which tells the “truth”? Rasch Measurement Transactions, 11(3), 580-581.

236 Linacre, J. M. (2004). Optimizing rating scale category effectiveness. In E. V. Smith & P. M. Smith (Eds.), Introduction to Rasch measurement (pp. 258-278). Maple Grove, MN: JAM Press.

Linacre, J. M. (2007). Winsteps (Version 3.63.2) [Computer software]. Chicago: Winsteps.

Little, D. (1991). Learner autonomy 1: Definitions, issues and problems. Dublin: Authentik Language Learning Resources.

Little, D. (1999). Learner autonomy is more than a western cultural construct. In S. Cotterall & D. Crabbe (Eds.), Learner autonomy in language learning: Defining the field and effecting change (pp. 11-27). Frankfurt am Main, Germany: Peter Lang.

Littlewood, W. (1996). “Autonomy”: An anatomy and a framework. System, 24(4), 427-435.

Littlewood, W. (1999). Defining and developing autonomy in East Asian contexts. Applied Linguistics, 20(1), 71-94.

Long, M. H. (1985). A role for instruction in second language acquisition: Task-based language teaching. In K. Hyltenstam & M. Pienemann (Eds.), Modelling and assessing second language acquisition (pp. 77-99). Clevedon, UK: Multilingual Matters.

Long, M. H. (1989). Task, group, and task-group interactions. University of Hawai’i Working Papers in ESL, 8(2), 1-26.

Loschky, L., & Bley-Vroman, R. E. (1993). Grammar and task-based methodology. In G. Crookes & S. Gass (Eds.), Tasks and language learning: Integrating theory and practice (pp. 123-167). Clevedon, England: Multilingual Matters.

MacIntyre, P. D. (1998). Language anxiety: A review of research for language teachers. In D. J. Young (Ed.), Affect in foreign language and second language learning: A practical guide to creating a low-anxiety classroom atmosphere (pp. 24-45). Boston: McGraw-Hill.

237 MacIntyre, P. D., Clément, R., Dörnyei, Z., & Noels, K. (1998). Conceptualizing willingness to communicate in a L2: A situational model of L2 confidence and affiliation. The Modern Language Journal, 82(4), 545-562.

MacIntyre, P. D., Noels, K., & Clément, R. (1997). Biases in self rating of second language proficiency: The role of language anxiety. Language Learning, 47(2), 265-287.

Mackenzie, A. S., & McCafferty, E. (Eds.). (2002). Developing autonomy. Tokyo: Japan Association for Language Teaching.

Madden, C., & Reinhart, S. (1987). Pyramids: Structurally based tasks for ESL learners. Ann Arbor: University of Michigan.

Maley, A., Duff, A., & Grellet, F. (1980). The mind’s eye: Using pictures creatively in language learning. Cambridge: Cambridge University Press.

Malvern, D. D., & Richards, B. J. (1997). A new measure of lexical diversity. In A. Ryan & A. Wray (Eds.), Evolving models of language (pp. 58-71). Clevedon, UK: Multilingual Matters.

Mare, L. (1990). Ma and Japan. The Southern Communication Journal, 55, 319-328.

Markus, H. R., & Kitayama, S. (1991). Cultural variation and self-concept. In J. Strauss & G. R. Goethals (Eds.), The self: Interdisciplinary perspectives (pp. 18-48). New York: Springer.

Martin, D. (1997). Talk a lot: Starting out. Okegawa, Japan: EFL Press.

McIntosh, C. N., & Noels, K. (2004). Self-determined motivation for language learning: The role of need for cognition and language learning strategies, Zeitschrift für Interkulturellen Fremdsprachenunterricht, 9(2). Retrieved October 16, 2005 from http://zif.spz.tu-darmstadt.de/jg-09-2/beitrag/Mcintosh2.htm.

238 Ministry of Education, Culture, Sports, and Technology. (1998). Gakushu Shido Yoryo [Course of Study in Lower Secondary School]. Tokyo: Zaimusho Insatsukyoku.

Ministry of Education, Culture, Sports, and Technology. (2003). Legislation of “National university corporation law”. Retrieved May 16, 2007 from www.mext.go.jp/english/ news/2003/07/03120301.htm.

Moloney, J. C. (1953). Understanding the paradox of Japanese psychoanalysis. The International Journal of Psycho-Analysis, 34(4), 291-303.

Monty, R. A., Rosenberger, M. A., & Perlmuter, L. C. (1973). Amount of locus of choice as sources of motivation in paired-associates learning. Journal of Educational Psychology, 97(1), 16-71.

Nakane, C. (1970). Japanese society. Berkeley, CA: University of California Press.

Nakata, Y. (2004). Motivation and the learning environment: A study of Japanese learners of English. Unpublished Doctoral Dissertation, University of Dublin, Trinity College, Dublin, Ireland.

Nicholson, P., & Sakuno, R. (1982). Explain yourself! An English conversation book for Japan. Tokyo: PAL Limited.

Noels, K. A. (2001). Learning Spanish as a second language: Learners’ orientations and perceptions of their teachers’ communication style. Language Learning, 51(1), 107-144.

Noels, K. A., Clément, R., & Pelletier, L. C. (1999). Perceptions of teachers’ communicative style and students’ intrinsic and extrinsic motivation. The Modern Language Journal, 83(1), 23-34.

Noels, K. A., Clément, R., & Pelletier, L. C. (2001). Intrinsic, extrinsic, and integrative orientations of French Canadian learners of English. The Canadian Modern Language Review, 57(3), 424-442.

239 Noels, K. A., Pelletier, L. G., Clément, R., & Vallerand, R. J. (2000). Why are you learning a second language? Motivational orientations and self-determination theory. Language Learning, 50(1), 57-85.

Norris, J. M., Brown, J. D., Hudson, T., & Yoshioka, J. (1998). Designing second language performance assessments. Honolulu, HI: University of Hawai’i Press.

Nunan, D. (2004). Task-based language teaching. Cambridge: Cambridge University Press.

Oxford, R. (1990). Language learning strategies: What every teacher should know. Boston, MA: Heinle & Heinle.

Oxford, R. (2003). Toward a more systematic model of L2 learner autonomy. In D. Palfreyman & R. C. Smith (Eds.), Learner autonomy across cultures: Language education perspectives (pp. 75-91). Basingstoke, UK: Palgrave MacMillan.

Oxford, R. L. (2006). Task-based language teaching and learning: An overview. Asian EFL Journal, 8(3), 94-121.

Pemberton, R., Li, E. S. L., Or, W. W. F., & Pierson, H. D. (Eds.). (1996). Taking control: Autonomy in language learning. Hong Kong: Hong Kong University Press.

Pennycook, A. (1997). Cultural alternatives and autonomy. In P. Benson & P. Voller (Eds.), Autonomy and independence in language learning (pp. 35-53). Harlow, UK: Longman.

Perls, F. (1973). The Gestalt approach and eyewitness to therapy. Palo Alto, CA: Science and Behavior.

Pervin, L. A. (1963). The need to predict and control under conditions of threat. Journal of Personality, 31, 570-585.

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

Pierson, H. D. (1996). Learner culture and learner autonomy in the Hong Kong Chinese context. In R. Pemberton & E. S. L. Li & W. W. F. Or & H. D. Pierson (Eds.), Taking control: Autonomy in language learning (pp. 49-58). Hong Kong: Hong Kong University Press.

Plough, I., & Gass, S. (1993). Interlocutor and task familiarity: Effects on interactional structure. In G. Crookes & S. Gass (Eds.), Tasks and language learning: Integrating theory and practice (pp. 35-56). Clevedon, England: Multilingual Matters.

Prabhu, N. S. (1987). Second language pedagogy. Oxford: Oxford University Press.

Prieto, G., & Delgado, A. R. (2007). Measuring math anxiety (in Spanish) with Rasch rating scale model. Journal of Applied Measurement, 8(2), 149-160.

Quirk, R., Greenbaum, S., Leech, G., & Svartvik, J. (1985). A comprehensive grammar of the English language. Essex, England: Longman.

Ramage, K. (1990). Motivational factors and persistence in foreign language study. Language Learning, 40(2), 189-219.

Rasch, G. (1960). Probabilistic models for some intelligence and attainment tests. Copenhagen: Danish Institute for Educational Research (Expanded edition, 1980. Chicago: University of Chicago Press).

Reeve, J. (1997). Understanding motivation (2nd ed.). Forth Worth, TX: Harcourt Brace College Publishers.

Reim, B., Glass, D. C., & Singer, J. E. (1971). Behavioral consequences of exposure to uncontrollable and unpredictable noise. Journal of Applied Social Psychology, 1(1), 44-56.

241 Richards, J. C., Gordon, D., & Harper, A. (1995). Listen for it: A task-based listening course. New edition. Oxford: Oxford University Press.

Richards, J. C., Schmidt, R., Kendricks, H., & Kim, Y. (2002). Longman dictionary of language teaching & applied linguistics (3rd ed.). London: Longman.

Robinson, P. (1995). Task complexity and second language narrative discourse. Language Learning, 45(1), 99-140.

Robinson, P. (2001a). Task complexity, cognitive resources, and syllabus design: A triadic framework for examining task influences on SLA. In P. Robinson (Ed.), Cognition and second language instruction (pp. 287-318). Cambridge, UK: Cambridge University Press.

Robinson, P. (2001b). Task complexity, task difficulty, and task production: Exploring interactions in a componential framework. Applied Linguistics, 22(1), 27-57.

Robinson, P. (2003). The cognition hypothesis, task design, and adult task-based language learning. Second Language Studies, 21(2), 45-105.

Rodin, J., & Langer, E. J. (1977). Long-term effects of a control-relevant intervention with institutionalized aged. Journal of Personality and Social Psychology, 35(12), 897-902.

Ryan, R. M., & Connel, J. P. (1989). Perceived locus of causality and internalization: Examining reasons for acting in two domains. Journal of Personality and Social Psychology, 57(5), 749-761.

Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. American Psychologist, 55(1), 68-78.

Ryan, R. M., & Deci, E. L. (2002). Overview of Self-Determination Theory: An organismic dialectical perspective. In E. L. Deci & R. M. Ryan (Eds.), Handbook of self- determination research (pp. 3-33). Rochester, NY: University of Rochester Press.

242 Samuda, V. (2001). Guiding relationships between form and meaning during task performance: The role of the teacher. In M. Bygate, P. Skehan & M. Swain (Eds.), Researching pedagogic tasks: Second language learning, teaching and testing (pp. 119-140). Harlow, UK: Longman.

Sano, M., Yamaoka, T., Matsumoto, S., & Sato, Y. (2006). Sunshine English course: Book 1. Tokyo: Kairyudo.

Schmidt, R. W. (1990). The role of consciousness in second language learning. Applied Linguistics, 11(2), 129-158.

Schmidt, R., & Frota, S. (1986). Developing basic conversational ability in a second language: A case study of an adult learner of Portuguese. In R. Day (Ed.), Talking to learn: Conversation in second language acquisition (pp. 237-326). Rowley, MA: Newbury House.

School menus. (2004, April 14). Jupiter Courier, p. B6.

Schulz, R. (1976). Effects of control and predictability on the physical and psychological well-being of the institutionalized aged. Journal of Personality and Social Psychology, 33(3), 563-573.

Schwartz, B. (2000). Self-determination: The tyranny of choice. American Psychologist, 55(1), 79-88.

Schwartz, B. (2004a). The paradox of choice: Why more is less. New York: Harper-Collins.

Schwartz, B. (2004b, January 23). The tyranny of choice. The Chronicle of Higher Education, p. 6.

Schwartz, B., Ward, A., Monterosso, J., Lyubomirsky, S., White, K., & Lehman, D. R. (2002). Maximizing versus satisficing: Happiness is a matter of choice. Journal of Personality and Social Psychology, 83(5), 1178-1197.

243 Sethi, S. (1997). Choice and its discontents: A new look at the role of choice in intrinsic motivation. Unpublished Doctoral Dissertation, Stanford University.

Shimizu, K. (Ed.). (2006). 2007 nenban daigaku rankingu [University Ranking 2007]. Tokyo: Asahi Shinbunsha.

Simpson, J. A., & Weiner, E. S. C. (1989). The Oxford English Dictionary (Vol. 17, pp. 655-656). Oxford: Oxford University Press.

Skehan, P. (1995). Analysability, accessibility, and ability for use. In G. Cook & B. Seidlhofer (Eds.), Principle and practice in applied linguistics: Studies in honor of H. G. Widdowson (pp. 91-106). Oxford: Oxford University Press.

Skehan, P. (1996). A framework for the implementation of task-based instruction. Applied Linguistics, 17(1), 38-62.

Skehan, P. (1998). A cognitive approach to language learning. Oxford: Oxford University Press.

Skehan, P. (2001). Tasks and language performance assessments. In M. Bygate & P. Skehan & M. Swain (Eds.), Researching pedagogic tasks: Second language learning, teaching and testing (pp. 167-185). Harlow, UK: Longman.

Skehan, P. (2003a). Focus on forms, tasks, and technology. Computer Assisted Language Teaching, 16(5), 391-411.

Skehan, P. (2003b). Task-based instruction. Language Teaching, 36, 1-14.

Skehan, P. (2007, September). Tradeoff and Cognition: Two hypotheses regarding attention during task-based performance. Paper presented at the International Conference on Task-Based Language Teaching, Honolulu, HI.

Skehan, P., & Foster, P. (1997). Task type and task processing conditions as influences on foreign language performance. Language Teaching Research, 1(3), 185-211.

244 Skehan, P., & Foster, P. (1999). The influence of task structure and processing conditions on narrative retellings. Language Learning, 49(1), 93-120.

Skehan, P., & Foster, P. (2001). Cognition and tasks. In P. Robinson (Ed.), Cognition and second language instruction (pp. 183-205). Cambridge: Cambridge University Press.

Skehan, P., & Foster, P. (2005). Strategic and on-line planning: The influence of surprise information and task time on second language performance. In R. Ellis (Ed.), Planning and task performance in a second language (pp. 193-216). Amsterdam: John Bejamins.

Solomon, R. (2001). In defense of freedom. American Psychologist, 56(1), 79-80.

Stotland, E., & Blumenthal, A. L. (1964). The reduction of anxiety as a result of the expectation of making a choice. Canadian Journal of Psychology, 18(2), 139-145.

Swain, M. (1985). Communicative competence: Some roles of comprehensible input and comprehensible output in its development. In S. Gass & C. Madden (Eds.), Input in second language acquisition (pp. 235-253). Cambridge, MA: Newbury House.

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

Swain, M. (2000). The output hypothesis and beyond: Mediating acquisition through collaborative dialogue. In J. P. Lantolf (Ed.), Sociocultural theory and second language learning (pp. 97-114). Oxford: Oxford University Press.

Tabachnick, B., & Fidell, L. (2007a). Experimental designs using ANOVA. Belmont, CA: Thomson.

Tabachnick, B. G., & Fidell, L. S. (2007b). Using multivariate statistics (5th ed.). Boston: Allyn and Bacon.

245 Takashima, H. (2000). Jissenteki komyunike-shonnoryoku no tame no Eigo no tasukukatsudo to bunposhido [Structure-based tasks and grammar instruction for practical communicative competence]. Tokyo: Taishukan.

Tanaka, K., & Yamaguchi, H. (2000). Influence of autonomy on perceived control beliefs and self-regulated learning in Japanese undergraduate students. North American Journal of Psychology, 2(2), 255-272.

Tobin, J. J., Wu, D. Y. H., & Davisdson, D. H. (1989). Preschool in three cultures: Japan, China, and the United States. New Haven, CT: Yale University Press.

Tremblay, P. F., & Gardner, R. C. (1995). Expanding the motivation construct in language learning. The Modern Language Journal, 79(4), 505-520.

Tsuchida, T. (1992). From ethos to ethics: Japanese views on life and death. In Institute of Medical Humanities, Kitasato University (Ed.), Toward a new replenishment of medical education and hospital service (pp. 319-325). Tokyo: Kitasato University.

Tyler, W. J. (1983). Amae. In Kodansha encyclopedia of Japan (Vol. 1, pp. 48-49). Tokyo: Kodansha.

Ur, P. (1981). Discussions that work: Task-centered fluency practice. Cambridge: Cambridge University Press.

Ushioda, E. (1996). Learner autonomy 5: The role of motivation. Dublin: Authentik Language Learning Resources, Ltd.

Ushioda, E. (2004, June). Motivation, autonomy and sociocultural theory. Paper presented at the Autonomy and Language Learning: Maintaining Control, Hong Kong and Hangzhou, China.

Vallerand, R. J., Pelletier, L. G., Blias, M. R., Brière, N. M., Senécal, C., & Vallières, E. F. (1992). The academic motivation scale: A measure of intrinsic, extrinsic and amotivation in education. Educational and Psychological Measurement, 52, 1003-1017.

246 Van Ek, J. A., & Alexander, L. G. (1975). Threshold level English. Oxford: Pergamon Press.

Van Ek, J. A., & Alexander, L. G. (1976). The threshold level for modern language learning in schools. London: Longman. van Lier, L. (1988). The classroom and the language teacher: Ethnography and second- language classroom research. London: Longman. van Lier, L. (1996). Interaction in the language curriculum: Awareness, autonomy & authenticity. Harlow, UK: Longman.

Wagatsuma, H. (1983). Psychology of the Japanese people. In Kodansha encyclopedia of Japan (Vol. 4, pp. 38-40). Tokyo: Kodansha.

Walker, E. W. (2004, January 26). So much choice. How do I choose? [Letter to the editor]. The New York Times, p. 22.

Weisz, J. R., Rothbaum, F. M., & Blackburn, T. C. (1984). Standing out and standing in: The psychology of control in America and Japan. American Psychologist, 39(9), 955-969.

White, R. W. (1959). Motivation reconsidered: The concept of competence. Psychological Review, 66(5), 297-333.

Wilkins, D. A. (1976). Notional syllabuses. Oxford, England: Oxford University Press.

Willis, J. (1996). A framework for task-based learning. Harlow, England: Longman.

Willis, J. (2004). Perspectives on task-based instruction: Understanding our practices, acknowledging different practitioners. In B. L. Leaver & J. Willis (Eds.), Task- based instruction in foreign language education (pp. 3-44). Washington, D. C.: Georgetown University Press.

247 Wolfe-Quintero, K., Inagaki, S., & Kim, H. Y. (1998). Second language development in writing: Measures of fluency, accuracy & complexity. Honolulu, HI: University of Hawai’i Press.

Wright, B. D. (1996). Comparing Rasch measurements and factor analysis. Structural Equation Modeling, 3(1), 3-24.

Yamada, F. (1999). Task-based instruction in second language learning: An experimental study focusing on the progressive form. Unpublished Master’s Thesis, Hyogo University of Education, Yashiro, Japan.

Yamada, H. (1997). Different games different rules: Why Americans and Japanese misunderstand each other. New York: Oxford University Press.

Yamaguchi, S. (2004). Further clarifications of the concept of amae in relation to dependence and attachment. Human Development, 47(1), 28-33.

Yang, C.-L., & Kramer, G. A. (2007). Using Rasch analysis to construct a clinical problem-solving inventory in the dental clinic: A case study. Journal of Applied Measurement, 8(2), 161-174.

Yashima, T. (2002). Willingness to communicate in a second language: The Japanese EFL context. The Modern Language Journal, 86(1), 54-66.

Yashima, T. (2007, October). Autonomy and willingness to communicate: Development of an English-using ideal self. Paper presented at The Independent Learning Association 2007 Japan Conference, Chiba, Japan.

Yuan, F., & Ellis, R. (2003). The effects of pre-task planning and on-line planning on fluency, complexity and accuracy in L2 monologic oral production. Applied Linguistics, 24(1), 1-27.

Zimbardo, P. G., Weisenberg, M., Firestone, I., & Levy, B. (1965). Communicator effectiveness in producing public conformity and private attitude change. Journal of Personality, 33, 233-255.

248 Zuckerman, M., Porac, J., Lathin, D., Smith, R., & Deci, E. L. (1978). On the importance of self-determination for intrinsically-motivated behavior. Personality and Social Psychology Bulletin, 4(3), 443-446.

Zuengler, J., & Bent, B. (1991). Relative knowledge of content domain: An influence on native non-native conversations. Applied Linguistics, 12(4), 397-415.

249 APPENDICES

250 APPENDIX A CLOZE TEST

DIRECTIONS: 1. Read the passage quickly to get the general meaning. 2. Write only one word in each blank in the column to the right. Contractions (example: don’t) and possessives (John’s bicycle) are one word. 3. Check your answers. NOTE: Spelling will not count against you as long as the scorer can read the word. EXAMPLE: The boy walked up the street. He stepped on a piece of ice. He fell (1) but he didn’t hurt himself. 1. down

MAN AND HIS PROGRESS

Man is the only living creature that can make and use tools. He is the most teachable of living beings, earning the name of Homo sapiens. (1) ever restless brain has used the (2) and the wisdom of his ancestors (3) improve his way of life. Since (4) is able to walk and run (5) his feet, his hands have always (6) free to carry and to use (7) . Man’s hands have served him well (8) his life on earth. His development, (9) can be divided into three major (10) , is marked by several different ways (11) life. Up to 10,000 years ago, (12) human beings lived by hunting and (13) . They also picked berries and fruits, (14) dug for various edible roots. Most (15) the men were the hunters, and (16) women acted as food gatherers. Since (17) women were busy with the children, (18) men handled the tools. In a (19) hand, a dead branch became a (20) to knock down fruit or to (21) for tasty roots. Sometimes, an animal (22) served as a club, and a (23) piece of stone, fitting comfortably into (24) hand, could be used to break (25) or to throw at an animal. (26) stone was chipped against another until (27) had a sharp edge. The primitive (28) who first thought of putting a (29) stone at the end of a (30) made a brilliant discovery: he (31) joined two things to make a (32) useful tool, the spear. Flint, found (33) many rocks, became a common cutting (34) in the Paleolithic period of man’s (35) . Since no wood or bone tools (36) survived, we know of this man (37) his stone implements, with which he (38) kill animals, cut up the meat, (39) scrape the skins, as well as (40) pictures on the walls of the (41) where he lived during the winter. (42) the warmer seasons, man wandered on (43) steppes of Europe without a fixed (44) , always foraging for food. Perhaps the (45) carried nuts and berries in shells (46) skins or even in light, woven (47) . Wherever they camped, the primitive people (48) fires by striking flint for sparks (49) using dried seeds, moss, and rotten (50) for tinder. With fires that he kindled himself, man could keep wild animals away and could cook those that he killed, as well as provide warmth and light for himself.

251 CLOZE TEST: ACCEPTABLE ANSWERS

1. his, man’s, our, the 2. accomplishments, culture, cunning, examples, experience(s), hands, ideas, information, ingenuity, instinct, intelligence, knowledge, mistakes, nature, power, skill, skills, talent, teaching, technique, thought, will, wit, words, work 3. to 4. he, man 5. on, upon, using, with 6. been, felt, hung, remained 7. adequately, carefully, conventionally, creatively, diligently, efficiently, freely, implements, objects, productively, readily, them, things, tools, weapons 8. all, during, for, improving, in, through, throughout, with 9. also, basically, conveniently, easily, historically, often, however, since, that, thus, which 10. areas, categories, divisions, eras, facets, groups, parts, periods, phases, sections, stages, steps, topics, trends 11. for, in, of, through, towards 12. all, early, hungry, many, most, only, primitive, the, these 13. farming, fishing, foraging, gathering, killing, scavenging, scrounging, sleeping, trapping 14. and, often, ravenously, some, the 15. always, emphatically, important, nights, normally, of, often, times, trips 16. all, house, many, most, older, the, their, younger 17. all, married, often, many, most, older, primate, the, these 18. all, constructive, many, most, older, primate, the, tough, younger 19. able, big, closed, coordinated, creative, deft, empty, free, human(‘s), hunter’s, learned, man’s, needed, needy, person’s, right, single, small, skilled, skillful, strong, trained 20. club, device, instrument, pole, rod, spear, stick, tool, weapon 21. burrow, excavate, dig, probe, search, test 22. arm, bone, easily, had, hide, horn, log, skull, tail, toot, tusk 23. big, chipped, fashioned, flat, hard, heavy, large, rough, round, shaped, sharp, sizeable, small, smooth, soft, solid, strong, thin 24. a, his, man’s, one(‘s), the 25. apart, bark, bones, branches, coconuts, down, firewood, food, heads, ice, items, meat, nuts, objects, open, rocks, sticks, shells, stone, things, tinder, trees, wood *something 26. a, each, flat, flint, glass, hard, obsidian, one, shale, softer, some, the, then, this 27. each, it, one, they 28. being, creature, human’s, hunter, man, men, owner, people, person *human 29. glass, hard, jagged, large, lime, pointed, sharp, sharpened, small 30. bone, branch, club, log, pole, rod, shaft, stick 31. accidentally, cleverly, clumsily, conveniently, creatively, dexterously, double, easily, first, had, ingeniously, securely, simply, soon, suddenly, tastefully, then, tightly, would 32. bad, extremely, hunter’s, incredibly, intelligent, good, long, modern, most, necessarily, new, portentously, quite, tremendously, useful, very 252 33. all, among, amongst, by, in, inside, on, that, using, within 34. device, edge, implement, instrument, item, material, method, object, piece, practice, stone, tool, utensil *knife 35. age, ancestry, development, discoveries, era, evolution, existence, exploration, history, lite, time 36. actually, apparently, ever, have 37. and, by, for, from, had, made, through, used, using 38. could, did, would 39. and, carefully, help, or, skillfully, then, would 40. carve, create, draw, drawing, engrave, hang, paint, painting, place, sketch, some, the 41. animals, cave(s), place(s), room 42. and, during, in, with 43. all, barren, across, aimless, dry, flat, high, in, long, many, plain, stone, the, through, to, toward, unknown, various 44. appetite, camp, course, destination, destiny, diet, direction, domain, foundation, habitat, home, income, knowledge, location, lunch, map, meal, path, pattern, place, plan, route, supplement, supply, time, weapons 45. children, man, men, families, group, human, hunter, people, primitives, voyager, wanderers, woman, women 46. and, animal, animal’s, covered, in, of, like, or, on, their, using, with 47. bags, baskets, blankets, chests, cloth(s), clothes, fabric, garments, hides, material, nets, pouches, sacks 48. began, built, lighted, lit, made, produced, started, used 49. also, and, by, occasionally, or, then, together, while 50. bark, branches, dung, forage, grass, leaves, lumber, roots, skin, tree(s), timber, wood

(Original passage from Kurilecz, M. (1969). Man and his world: A structured reader. New York: Crowell.) UNIT 6 MAN AND HIS MATERIAL PROGRESS (pp. 58-61)

Man is the only living creature that can make and use tools. He is the most teachable of living beings, earning the name of Homo sapiens. His ever restless brain has used the knowledge and the wisdom of his ancestors to improve his way of life. Since man is able to walk and run on his feet, his hands have always been free to carry and to use tools. Man’s hands have served him well during his life on earth. This life story can be divided into three major periods, each marked by a different way of life.

Up to 10,000 years ago, all human beings lived by bunting and fishing, by picking berries, fruits, and nuts, and by digging for edible roots. Most likely, the men were the hunters, and the women acted as food gatherers. Since the women were busy with the children, the man handled the tools. In a man’s hand, a dead branch became a stick to knock down fruit or to dig for tasty roots. An animal thigh bone served as a club, and a rounded piece of stone, fitting comfortably into the hand, could be used to break nuts or to throw at an animal. This stone was chipped against another until it had a sharp 253 edge. The primitive man who first thought of putting a sharp stone at the end of a stick had made a brilliant discovery: he had joined two things to make a more useful whole, a spear. Flint, found in many rocks, became a common cutting tool in this Paleolithic period of man’s life. Since no wood or bone tools have survived, we know this man only by his stone implements, with which he could kill animals, cut up the meat, and scrape the skins, as well as carve pictures on the walls of the caves where be lived during the winter.

In the warmer seasons, man wandered on the steppes of Europe, without a fixed home, always foraging for food. Perhaps the women carried nuts or berries in shells or skins or even in light, woven baskets. Wherever they camped, the primitive people built fires by striking flint for sparks and using dried seeds, moss, and rotten wood for tinder. Man had probably realized that the sparks from flint looked like the natural fire that sometimes spread after lightning had struck a tree. With the fire that he kindled himself, man could keep wild animals away and could cook those that he killed, as well as provide warmth and light for himself.

When the vast glacier that had covered Europe in the ice age began to melt, forests began to creep over the continent and animals had to migrate to more open spaces. Men moved with the animals to the seashores of western and northern Europe, living poorly on fish, berries, and seeds during this Mesolithic period.

Perhaps it was the women who first noticed that grasses grew where seeds had fallen. The deliberate planting of seeds brought a tremendous change in the life of man. He stopped his wandering, he built himself a house which he furnished with utensils of his own making, and he began to keep animals near him to use for food and work. Man first raised wheat and barley, probably in the valley between the Tigris and Euphrates rivers. Here the rich soil, abundant water, and warm climate created a “Fertile Crescent” by the year 3000 B.C. Man’s digging stick, fitted with a sharp flint point, became his first pick. With his great skill in shaping and polishing wood and stone, the Neolithic man developed the stick into a plow. His important discovery of metals (copper, bronze, and iron) gave him an iron-tipped plow and a sickle for harvesting. When he harnessed oxen to his plow, he had a tool that lasted him until the invention of the engine. Oxen could also drag the heavy sledge set with stones that beat the grain. Oxen could even turn the stones to grind the grain.

Probably animals dragged loads for man until he thought of putting two circles of wood at the ends of an axle. These circles became the wheels of a cart, which later developed into a light, fast chariot drawn by a horse. This, with only minor changes, was man’s way of traveling on land until the locomotive was invented.

Whoever first built a boat did as much for Neolithic man as the Wright brothers did for the modern world with their airplane. Perhaps a floating tree trunk first gave man the idea for a canoe which he shaped out of a tree. With his better tools, the Neolithic carpenter could make a keeled boat, the prototype of a modern ocean-going vessel. 254 In his house, the farmer had clay pots made on the potter’s wheel and baked hard to bold food and water. He also had floor mats and clothing woven of thread on hand looms. The Neolithic man was a farmer, fisherman, woodcutter, and carpenter. His pattern of life has continued to the present time, although today in progressive countries, only a quarter of the people are needed to raise food for the whole population.

No important change in this way of life occurred until the revolutionary invention of the steam engine in the nineteenth century showed man how to use heat to create power. Having learned to create power, man was able to operate the locomotive, the automobile, the radio, air conditioning, the telephone. Power may be the basic fact of the modern world. When man first understood the power of fire and used it on metals, he began: to climb up the stairway of civilization. Now man has discovered a new form of power, atomic energy. As a result, he may be at the beginning of a new period of his history, a period of change even in the form of life.

255 APPENDIX B SYLLABUSES OF THE JAPANESE PROFESSORS FOR THE TREATMENT CLASSES

Study 1, Group A, Monday The textbook this class used was Tsuda, A. (2003). Primary listening: 20 tips for better communication. Tokyo: Kinseido. This class was mostly a listening class. The students also memorized the model dialogs before class and presented them in class. The class followed along the plan below: “Orientation Content Words Function Words Understanding Situation and Content Developing Conversation Intonation Numbers Question Sentences Vocal Changes Sense Groups/Thought Groups Transitioning Context from Meaning Talking About Japanese Culture Role Play Group Presentations”

Study 1, Group A, Thursday The textbook this class used was Miyamoto, S., Shimada, H., Imai, H., Nishio, Y., Yanagi., Y., Yanagihara, Y., & Hayashi, C. (2004). Inside stories U.S.A. with multimedia. Tokyo: Seibido. “This class will use the CD enclosed in the book to engage in themes from the text and to elicit the needs of the students to build experience in reading, speaking, writing, and listening skills through the program. The central point of the class is to study through four-person groups. My hope is that the students will build confidence and trust in the other members of the class. The lessons we do and the schedule we follow will be decided with the students the first day of class.”

256 Study 1, Group A, Friday The textbooks this class used were Hagen, S. A. (2000) Journeys: Listening and speaking, 2. Hong Kong: Longman Asia ELT., Yamaoka, T., & Takahashi, Y. (2003). Sunshine kids: Book 1. Tokyo: Kairyudo., and Takahashi, Y., & Yamaoka, T. (2003). Sunshine kids: Book 2. Tokyo: Kairyudo.

“This class will improve your listening and speaking using listen and repeat, conversation, listen, listen and understand, pair-work, pronunciation rhythm, and etc. tasks. “1. The class will use English conversation content for use in elementary schools from the text, such as: (1) Could you please repeat that? (2) Tell me about yourself. (3) I’ll have the fried shrimp. (4) That dress looks great on you. (5) How was your weekend? (6) What are you going to do? (7) I’ll be back the day after tomorrow. 2. The class will introduce to the students activities that can be used for elementary school English activities such as games, picture books, projects and 50 rhythm and chants culled from actual research experience. 3. The class will conduct practical activities, performances and guidance for the elementary school English activities class.”

Study 1 & Study 2, Group B, Monday The textbook for this class was Viney, P. (2004). New survival English. Tokyo: Macmillan Language House. “This class will improve the listening skills of students through situations close to the students. The goal is to survive in English through real-life situations. Students will study basic listening improvement, expression, and conversation skills, through listening activities and role-play. Introductions Making appointments Describing jobs Buying things Making a reservation Conversation strategies Thanking Discussing”

257 Study 1 & Study 2, Group B, Thursday MacLean, P. (2005). My opinion, your opinion (2nd ed.). Tokyo: Macmillan Language House. “The goal of this class is to improve the students’ writing, listening, and expression skills but will concentrate on listening skills. Study will be from paragraph writing, listening, and conversation. Topics will be from: (1) Eating Well; (2) Personality Types; (3) Sports/Music; (4) Animal Rights; (5) Lifestyles; (6) Drinking/Smoking; (7) Executive Salaries; (8) Endangered Species; (9) Abstract Art/Movies; (10) Man’s Best Friend; (11) Gun Control/Death Penalty; (12) Population Control; (13) The Influence of Television; (14) Summer or Winter?

258 APPENDIX C descriptive task-no choice of topic first round

259 APPENDIX D descriptive task-no choice of topic second round

260 APPENDIX E descriptive task-limited choice of topic first round, task a

261 APPENDIX F descriptive task-limited choice of topic first round, task b

262 APPENDIX G descriptive task-limited choice of topic first round, task c

263 APPENDIX H descriptive task-limited choice of topic second round, task a

264 APPENDIX I descriptive task-limited choice of topic second round, task b

265 APPENDIX J descriptive task-limited choice of topic second round, task c

266 APPENDIX K Descriptive task, complete choice of topic

Partner A:

The topic for today: For the task today, please think of any place you like. This could be any place at all, your room, a building, or, for example, a theme park or a school. Any place at all. For your task, what I would like you to do is to describe this place. Please use English as much as you can. Your partner will make a drawing of a place you describe. PLEASE DO NOT POINT TO YOUR PARTNER’S PAPER TO HELP. Try explaining in English as much as you can. The place I want to describe is:

______

Partner B:

Your partner will describe a place to you. Please make a drawing of the place your partner describes below. DO NOT LET YOUR PARTNER POINT TO YOUR PAPER TO HELP. Try using English as much as you can.

267 appendix L narrative task-no choice of topic first round

268 appendix M narrative task-no choice of topic second round

269 appendix N narrative task-limited choice of topic first round task a

270 appendix O narrative task-limited choice of topic first round task b

271 appendix P narrative task-limited choice of topic first round task c

272 appendix Q narrative task-limited choice of topic second round task a

273 appendix R narrative task-limited choice of topic second round task b

274 appendix S narrative task-limited choice of topic second round task c

275 appendix T Narrative task, complete choice of topic

Partner A:

The topic for today: For the task today, please think of any story you want. This can be something that happened to you or it can be a story you heard. This can be any story at all. For your task, what I would like you to do is to tell this story to your partner. Please use English as much as you can.

Partner B:

For the task today, your partner will tell you a story. After your partner has told you the topic of the story, please answer the yellow survey. After that, please listen to your partner tell you the story. As you listen, please write an outline of the story.

276 Appendix U TOPICS FOR DECISION-MAKING TASK, NO CHOICE OF TOPIC

No Choice of Topic, First-Round Task: You and your partner have won a prize to visit three foreign countries. You can visit any three foreign countries but you only can spend one day in each country. The rest of the time you will be traveling in the plane. What three countries would you and your partner like to visit? Why do you two want to go to that country? Please discuss and decide with your partner which countries you would like to visit. Please discuss and decide with your partner which countries you would like to visit. 1. Country: ______Reason: ______2. Country: ______Reason: ______3. Country: ______Reason: ______

No Choice of Topic, Second-Round Task: Please decide the following. You and your partner will be able to visit six world leaders of today. What questions would you like to ask them? Please write a question for each world leader. 1. Junichiro Koizumi (Group A)/Shinzo Abe (Group B), Prime Minster of Japan- Question: ______2. Vladimir Putin, President of Russia- Question: ______3. George Bush, President of the United States- Question: ______4. Kim Jong Il, President of North Korea- Question: ______5. Jacques Chirac, President of France- Question: ______6. Wen Jiabao, Premier of China- Question: ______

277 appendix V decision-making task-limited choice of topic first round task a

You and your partner will have a visitor from the United States. You and your partner have one day to take him to Nara. You and your partner have enough time to take this person to six (6) places. Which places do you want to go to? Please put a check next to the places you want to go to. Good Luck! (adapted from http://www.pref.nara.jp/ nara_e/index.html).

Nara Park and its Outskirts ___ Nara Park (Deer Park) ___ Todaiji Temple Great Image of Buddaha ___ Shoso-in Hall An ancient treasure house ___ Kasuga Taisha Shrine Stone-lanterns line the road ___ Nara-Machi

Nishinokyo, Saho and Sakiji Area ___ Heijo Palace Site The ancient capital of Japan ___ Saidaiji Temple Try powdered green tea ___ Yakushiji Temple End of the Silk Road ___ Toshodaiji Temple Uchiwamaki - a ceremony

Ikaruga and Mt. Shigi Area ___ Horyuji Temple ___ Fujinoki Burial Mound ___ Hokiji Temple The oldest three-storied pagoda in Japan.

Asuka and Tonomine Area ___ Asukadera Temple ___ Takamatsuzuka Burial Mound and Wall Painting Museum ___ Ishibutai Burial Mound An ancient tomb

Yamanobe-no-Michi, Hase and Murou Area ___ Sumo Shrine ___ Isonokami Jingu Shrine

278 appendix W decision-making task-limited choice of topic first round task b

You and your partner will go on a camping trip. What will you and your partner take? You will already have a tent, a sleeping bag, and a backpack. What ten (10) things will you take?

1. ______

2. ______

3. ______

4. ______

5. ______

6. ______

7. ______

8. ______

9. ______

10. ______

279 appendix X decision-making task-limited choice of topic first round task c

The university will make a time capsule. This is a box where you put personal things and then the time capsule is put in the ground. This time capsule will removed from the ground in 100 years. What four (4) things will you put in this time capsule? Please choose four things with your partner and the reason for putting them in the time capsule. Good luck!

Item 1: ______

Reason: ______

Item 2: ______

Reason: ______

Item 3: ______

Reason: ______

Item 4: ______

Reason: ______

280 appendix Y decision-making task-limited choice of topic second round task a

You and your partner will have a visitor from the United States. You and your partner have one day to take him to Kyoto. You and your partner have enough time to take this person to six (6) places. Which places do you want to go to? Please put a check next to the places you want to go to. Good Luck! (adapted from http://www.japan-guide. com/e/e2155.html).

Central Kyoto ___ Kyoto Gosho Palace The emperor’s residence until 1868. ___ Nijo Castle Kyoto residence of the Tokugawa shogun. ___ Honganji Head temple of the Shin-Jodo sect.

Eastern Kyoto ___ Sanjusangendo Hall exhibiting 1001 Kannon statues. ___ Kiyomizudera Temple famous for its large terrace. ___ Gion Kyoto’s most famous geisha district. ___ Yasaka Shrine Shrine famous for its Gion Festival. ___ Heian Shrine Shrine of the ancient Imperial Palace. ___ Ginkakuji Silver Pavilion.

Northern Kyoto ___ Kinkakuji Golden Pavilion - actually covered in gold. ___ Ryoanji Zen Temple most famous for its rock garden. ___ Enryakuji Head temple of the Tendai sect.

Western Kyoto ___ Arashiyama Famous for the Togetsukyo Bridge. ___ Katsura Villa Extremely beautiful imperial villa.

Southern Kyoto and Uji ___ Fushimi Inari Shrine The ultimate torii experience. ___ Daigoji Famous temple in the southeast of Kyoto. ___ Byodoin Temple with a beautiful Pure Land Garden.

281 appendix Z decision-making task-limited choice of topic second round task B You and your partner will go America. You and your partner only have enough space to take ten personal items between you. What will you and your partner take in your luggage? Please choose ten (10) things to take. What ten (10) things will you take?

1. ______

2. ______

3. ______

4. ______

5. ______

6. ______

7. ______

8. ______

9. ______

10. ______

282 appendix AA decision-making task-limited choice of topic second round task B You will make a home page of famous Japanese people of today. You and your partner only have enough space to write about four (4) people. Please choose four people and the reason you chose that person.

Person 1: ______

Reason: ______

Person 2: ______

Reason: ______

Person 3: ______

Reason: ______

Person 4: ______

Reason: ______

283 APPENDIX BB Decision-Making Task, Complete Choice of Topic: Homework Assignment Homework:

You recently gave me many topics you want to do in this class. Thank you very much for that. Many people wrote topics on the environment. What I did below was to take the environment topics you gave me last week and put them in the list below. You can also add a topic if the topic you want to do is not below. Please choose any of the topics below and study it for homework for next week, January 25, 2007. Please be ready to discuss the problem with a student. You should be able to understand what causes the problem and what solutions there are and how you think you can make it better and decide with your partner on a solution to fix the problem. Write out at least a half page beforehand about the topic and the requirements above. I WILL COLLECT THE PAPERS AT THE START OF THE CLASS. THERE WILL BE A GRADE FOR THIS. Here are the topics: The garbage problem Soil pollution Global warming UV radiation Bad smells Deforestation Dwindling resources Freon gas Rising sea levels The ozone hole Fish depletion CO2 Nuclear waste Sinking land Golf course construction Heat islands Desertification Kitchen waste Dirty air Noise Dirty water Dioxin Endangered species Bird influenza Population increase Deforestation Food additives Dirty Oceans Acid rain

Your own topic: ______

284 APPENDIX CC DECISION-MAKING TASK, COMPLETE CHOICE OF TOPIC: CLASS MATERIAL

A week ago, I gave you the list of topics to do for today. Please circle the topic you want to do today. Here are the topics: The garbage problem Soil pollution Global warming UV radiation Bad smells Deforestation Dwindling resources Freon gas Rising sea levels The ozone hole Fish depletion CO2 Nuclear waste Sinking land Golf course construction Heat islands Desertification Kitchen waste Dirty air Noise Dirty water Dioxin Endangered species Bird influenza Population increase Deforestation Food additives Dirty Oceans Acid rain

Your own topic: ______

Now, please discuss with your partner the topic. You should be able to discus

1. What the problem is.

2. What the cause of the problem is.

3. How the problem can be made better.

285 appendix dd After-task SURVEY

286 APPENDIX ee example of the choice paper for limited choice session- descriptive task

287 APPENDIX ff example of the choice paper for limited choice session-narrative task

288 APPENDIX gg example of the choice paper for limited choice session-DECISION- MAKING task

Please choose a topic to discuss from the list below. Please circle the topic you want to do.

Topic 1

Topic 2

You and your partner will go on a camping trip. What will you and your partner take? You will already have a tent, a sleeping bag, and a backpack. What ten (10) things will you take?

Topic 3

289 APPENDIX hh examples from the transcript data of marking for correct verb forms

Guideline #1: The verb had to be in a phrase. (Example from narrative task, complete choice of topic (Appendix T).) counted as correct First, we went to Suikoden

Guideline #2: Non-existent but required verbs were not considered in the total. (Example from decision-making task, complete choice of topic (Appendix CC).) non-existent verb—not counted Human human eighty-seven percent

Guideline #3: Uncorrected repetitions by the same person in the same turn were not considered in the total (i.e., counts as one verb). (Example from narrative task, limited choice of topic (Appendix S). Acceptable answers: told, is telling, or tells.) counted as one verb—error Yeah Bread. Second. Her, aa their mother tell tell them Guideline #4: Progressives and perfects were counted wrong if an element (either the be-verb or the V-en/V-ing) was missing but as one verb if correct. (Example from narrative task, no choice of topic (Appendix M). Acceptable answers: walked, is walking, or walked.) error Yes, fourth, the man walking with the le- leaf Yes Yes, Then win-, wing? wing

Guideline #5: Apostrophized verbs were considered. (Example from decision- making task, limited choice of topic (Appendix AA).) counted as correct that’s good

290 Guideline #6: Auxiliary verbs, modal auxiliaries, and main verbs in the same clause were counted separately. (Example 1 from decision-making task, no choice of topic (Appendix U).) (incorrect—should be "will Iraq have" or "will there be") Iraq war will do next year

(Example 2 from decision-making task, no choice of topic (Appendix U).

Japan will do do future

(Example 3 from descriptive task, limited choice of topic (Appendix H).)

she is sit- sitting right side

291 appendix ii examples from the transcript data of marking for error-free clauses

Guideline #1: There had to be an English verb. Verbless clauses were not counted in the total. (Example from descriptive task, complete choice of topic (Appendix K).) missing a verb so not counted And PC on the desk

Guideline #3: Relative clauses were separated and counted separately as correct or incorrect. (Example from descriptive task, limited choice of topic (Appendix H). In this example, the clause on the right is an error but it would be counted as correct according to Guideline #14.)

correct correct and where is a conductor wearing , wear glasses man

Guideline #4: Phrases connected by conjunctions were separated and counted right or wrong separately. (Example of both correct and incorrect from narrative task, no choice of topic (Appendix M).) counted as correct error A man has a brush and go to the park

Guideline #5: The student was given the benefit of the doubt if there was an untranscribable segment in the phrase and if it would have made a correct phrase. (Example from descriptive task, complete choice of topic (Appendix K).) counted as correct my room is xxx, xxx,

Guideline #6: The phrase was counted as incorrect if Japanese ((j) in the transcript) was used without English correction soon afterwards, except some Japanese words printed on the descriptive tasks that were integral to completing the task. (Example of no correction from descriptive task, complete choice of topic (Appendix K).) error and, er, (j) on the right side, er, right side is , er (j)

292 Guideline #6: (Example of Japanese words printed on the task integral to completing the task from descriptive task, no choice of topic (Appendix D).) counted as correct The lamp is next to the (paper in shop in Japanese–upper right) poster ee

Guideline #7: Phrases that included repetitions were counted as correct if a correct form eventually was used. (Example from descriptive task, complete choice of topic (Appendix K).) counted as correct Indus Many people is are are Hindu Hindu

Guideline #8: A phrase was counted as correct even if it required more than one turn to complete. (Example from narrative task, limited choice of topic (Appendix S).) Student A: Last picture, there there is nothing

Student B: Oh, OK counted as correct—one clause

Student A: on the in the basket

(Example from narrative task, limited choice of topic (Appendix T).) Student A: I went to eat dinner counted as correct Student B: Yeah OK

Student A: Dinner at P.M. fifty thirty

Student B: mmm aa time?

Student A: Time, umm yes

Student B: Fifty thirty?

Student A: Fifty thirty

Student B: Fifty thirty? Fifty thirty?

Student A: Yes fif-. five thirty

293 Guideline #9: Exact repetitions of one student by the other in the next turn were not counted in the total. (Example from decision-making task, no choice of topic (Appendix U).) Student A: What do you like? —counted as correct

Student B: What do you like? —not counted

Guideline #10: Other-initiated corrections were counted in the total. (Example from decision-making task, limited choice of topic (Appendix Z). An incorrect mass noun was said. Acceptable answer: I want to take something to drink.) Student A: I want to take drink —counted as error

Student B: Drink. I want to take food —counted as correct

Guideline #11: Imperatives and indirect speech were counted in the total. (Example from decision-making task, no choice of topic (Appendix U). The phrase didn’t make sense in the context.) error eeeee France? France France please shake your hand

Guideline #12: English phrases out of context, even if grammatically correct, were counted as incorrect. (Example from decision-making task, no choice of topic (Appendix U). The students are making questions to ask to world leaders, not to themselves.) error—out of context Student A: I will be good

Student B: Japan will be good? eeee

Student A: WIll Japan

Guideline #13: Improper nouns were counted as incorrect. (Example from descriptive task, no choice of topic (Appendix D).) error aa He is grandmother

294 Guideline #14: In a clause utterance, if the proper words that made a correct clause were supplied in the correct order, no matter the surrounding words, that clause was counted as correct. (Example from the narrative task, no choice of topic (Appendix M).) Student A: Ride on (j: not) the man is carrying carry

Student B: mm carry

Student A: carry in the

Student B: carry

Student A: basket

Guideline #17: “To-Infinitive” clauses (to-V) were not counted. (Example 1 from decision-making task, complete choice of topic (Appendix CC).) correct error but it is difficult for us to reduce the use of car

(Example 2 from decision-making task, no choice of topic (Appendix U).) what do you want to ask Shinzo Abe

295 appendix jj examples from the transcript data of marking for turns

Guideline #1: An utterance was not counted as a turn if Japanese was used for that entire utterance. (Example from descriptive task, no choice of topic (Appendix D).) Note: Each solid, curved line on the left equals one turn. Student A: ok On the left top um

Student B: hai

Student A: wall

Student B: wall?

Student A: Wall

Guideline #2: An utterance was not counted as a turn if only hesitation or back channeling, such as ‘um,’ ‘uhuh,’ ‘yea,’ or ‘yeah,’ or Japanese equivalents were used. (Example of hesitation from descriptive task, no choice of topic (Appendix D).) Note: Each solid, curved line on the left equals one turn. Student A: Where is the picture

Student B: Picture? Oh lady picture

Student A: la-

Student B: lady?

Student A: -dy oh yeah

Student B: oh aa

Student A: She is she

296 Guideline #2: (Example of back channeling from descriptive task, no choice of topic (Appendix D).) Note: Each solid, curved line on the left equals one turn. Student A: wall mm aa mm

Student B: mm

Student A: hai where is

Student B: mm

Student A: the girl

Student B: mm

Student A: mmm eee aa?

Student B: aa?

Student A: black hair and the long

Student B: lo- long? long

Student A: long

Guideline #3: Proper nouns can make a turn. (Example from decision-making task, no choice of topic (Appendix U).) Note: Each solid, curved line on the left equals one turn. Student A: OK. Ok. Next, do you have a question for Vladi-

Student B: Putin

Student A: Putin?

Student B: I have no questions

297 Guideline #4: The words ‘OK’ and ‘Yes’ constituted a turn in themselves. (Example from narrative task, complete choice of topic (Appendix T).) Note: Each solid, curved line on the left equals one turn. Student A: I met many children there

Student B: OK

Student A: And

Student B: mm

Student A: I play soccer

Student B: mm

Student A: Together. It is

Student B: mm

Student A: very

Student B: mm

Student A: fun for me

Student B: OK

Student A: OK

298 Guideline #5: Onomatopoeia, sound effects, or animal sounds were not counted as a turn in themselves. (Example from narrative task, limited choice of topic (Appendix S).) Note: Each solid, curved line on the left equals one turn. Student A: In the basket the dog

Student B: Bowwow

Student A: Bowwow

Student B: Bowwow

Student A: Bowwow

Student B: Bowwow dog bowwow

Guideline #7: Incidents of spelling interaction were counted as turns. (Example from decision-making task, no choice of topic (Appendix U).) Note: Each solid, curved line on the left equals one turn. Student A: borscht b-o-r-s-h-t

Student B: e-o

Student A: b- b-o-r

Student B: r-

Student A: s-h-t

Student B: s-

Student A: h-

Student B: h-t

Student A: borscht

299 appendix kk examples from the transcript data of marking for unaltered repetitions

Guideline #1: The repetition was the exact same word or phrase, including proper nouns, in the same turn. (Example from decision-making task, no choice of topic (Appendix U).) Student A: Sports

Student B: Sports, Sports

Guideline #2: Repetitions separated by hesitations were not counted. (Example from decision-making task, limited choice of topic (Appendix AA).) These two pronouns counted as a repetition I I choose aaa aa I choose mmm I choose tomonori jinnai and norika fujiwara The phrase “I choose” said repeatedly not counted as a repetition because it is separated by hesitations

Guideline #3: False-starts, ‘yes,’ ‘yeah,’ ‘yea,’ ‘OK,’ ‘Oh,’ ‘no’, onomatopoeia, animal sounds, words such as ‘wow’, repetitions of the teacher’s directions (e.g., ‘Go, go, go’ and ‘English, English, English’ often used by the teacher for pep reasons), or words repeated at the end of the task to denote the finish of the task (e.g., ‘finish finish,’ or ‘thank you thank you’) were not counted. (Example of a false start from decision-making task, complete choice of topic (Appendix CC).) yes yes yes yes the case of is deforest- deforestation is human being not counted as a repetition

Guideline #3 (Example of repeating after the teacher.) Teacher: English, English, English! You can do it!

Student: English, English, English. not counted as a repetition

Guideline #3 (Example of ending the conversation.): finish thank you thank you thank you not counted as a repetition

300 Guideline #5: The same word can be at the end of a phrasal repetition as well as the start of another. (Example from descriptive task, no choice of topic (Appendix D).) first phrase second phrase boy is mmm ooo ooo ooo where is the picture the picture is woman counted as a repetition

301 APPENDIXAPPENDIX LL DD EXAMPLEEXAMPLE OF OF TRANSCRIPT TRANSCRIPT FROM FROM THE THE DESCRIPTIVE DESCRITPTIVE TASK-NOTASK-NO CHOICE CHOICE OF OF TOPIC TOPIC

Group B, Class 1, female-female pair. (Topic in Appendix D.)

(student says student number here) (teacher: ok please go) one x (no article) turn \where is Where is picture? x (the preposition “on” is missing) The picture. (off mic on headphones) The picture is left, left side of the wall next to the beerbeer poster x (no article) Where is child eee girl?

She, she is playing on the left side, left side x (no article; the Where is a lamp? preposition “on” is missing) The lamp is next to the (paper in shop in japanese) poster. ee The light is right x (no article) side (“smoking” seems to be a modifier so ee Where is man ee smoking, smoking man?not counted as a clause because there is no verb) He, he is cen-, he is next to the man who is cooking the fish

Correct verb forms: 12 of 12 correct Error-free clauses: 5 of 10 correct Unaltered repetitions: 6 Turns: 7 Word count: 72 (word added: beer {+ 2 types | + 1 token}) Words per turn: 10.29 Type-token ratio: 72 types ÷ 26 tokens = .32

Key: = Error-free clauses = Correct verb forms = Unaltered repetitions x = entity is incorrect \ = Start of transcript (j) = Japanese words (solid line connects an entity separated in the transcript)

302 APPENDIXAPPENDIX MM EE EXAMPLEEXAMPLE OF TRANSCRIPTOF TRANSCRIPT FROM FROM THE THE DESCRIPTIVE DESCRIPTIVE TASK-LIMITEDTASK-LIMITED CHOICE CHOICE OF TOPICOF TOPIC

Group B, Class 1, female-female pair. (Topic in Appendix H.)

(teacher: please say your student number) (student says student number) (teacher: sound check ok?) (teacher: ok go go go) one turn \where is where is

the woman

mm x (should be “has”)

mmm having a bag x (verb tense error)

having a bag

and wearing boots

aaa

and short cut

aaa

hair

stripe suits?

mmm yes

aaa aaa

boots

303 x (should be “sitting”) eeto she is sit on seat x (verb tense error) mmm and eee right side of mmm x (missing articles, auxiliary verb) eeto right side of parma look at eee right side of woman mmm x (should be “looking”) aa (j) between mm woman and woman woman wo- woman (j: not) (j) eeeto left side of mmm woman aaaa skirt and (j) eeeto right side of mmm man? man? woman? panchi parma ok ok

304 Correct verb forms: 2 of 5 correct Error-free clauses: 1 of 4 correct Unaltered repetitions: 0 Turns: 8 Word count: 59 (word added: boots {+ 2 types | + 1 token}) Words per turn: 7.38 Type-token ratio: 59 types ÷ 29 tokens = .49

Key: = Error-free clauses = Correct verb forms = Unaltered repetitions x = entity is incorrect \ = Start of transcript (j) = Japanese words (solid line connects an entity separated in the transcript)

305 APPENDIXAPPENDIX NN FF EXAMPLEEXAMPLE OF OFTRANSCRIPT TRANSCRIPT FROM FROM THE THE DESCRIPTIVE DESCRITPTIVE TASK-COMPLETETASK-COMPLETE CHOICE CHOICE OF OFTOPIC TOPIC

Group B, Class 1, female-female pair. (Topic in Appendix K.)

(t: please say your student number)

u zero six one five zero e

mm one turn \my room is mmm square

square

oh square

eee

square

mmm eeto eto the door is

is (There was grammaticality in this utterance and the student down not here under was given the benefit of the doubt. Therefore the two words under und- “down” and “under” were not considered.) down down mmm

right or left

306 middle

middle oh middle

enter the door

mm x (missing adverbial phrase of position and article; should be “on the right side is a/the kitchen”) enter right side is kitchen

right side is kitchen

kitchen (Exact repetition from the last turn so not counted as a clause or a verb) door’s right

right side

right right

right kitchen x (missing adverbial phrase of position and article; should be “on the right side is a/the bath and kitchen (j) kitchen’s right side is bath and toilet toilet”)

mmm x (hard to understand)

the right it’s right is floor floor

bath and toilet right (a mistake for a possessive so not considered as a verb) last countable turn yes enter the my room is left side is my bed and x (should be “on the left side of my room is my bed”) mm?

my bed right side is table x (should be “on the right side of my bed is a table)

307 mmm x (should be “in front of my table is a desk”)

and table in front of my table is study’s desk and study’s desk left is tv x (should be “on the left side of the desk is a/my tv”) Correct verb forms: 9 of 9 correct Error-free clauses: 2 of 9 correct Unaltered repetitions: 2 Turns: 22 Word count: 102 (words added: toilet {+ 2 types | + 1 token}; tv {+ 1 type | + 1 token}) Words per turn: 4.64 Type-token ratio: 102 types ÷ 33 tokens = .32

Key: = Error-free clauses = Correct verb forms = Unaltered repetitions x = entity is incorrect \ = Start of transcript (j) = Japanese words (solid line connects an entity separated in the transcript)

308 APPENDIX OO EXAMPLE OF TRANSCRIPT FROM THE NARRATIVE TASK-NO CHOICE OF TOPIC

Group B, Class 1, female-female pair. (Topic in Appendix M.)

(teacher: please say your student number again) (student says student number) (teacher: sound check ok?) (teacher: ok please go) x (verb tense error) one turn \one the man is aa (j) the man walking walking in the park shor- the man aa ee kick a dog? x (wrong verb) x (should be aa yea kick a dog “kicking”) x (should be “kicking”) ok

two ee mm the man cleaning

aa ok x (missing an auxiliary)

in the park

mm ok x (wrong verb) x (should be “has”) the man mm? xxx xxx mmm have a have a lot of leaf?

ok ok ok (j) ok aaaa ooo ok x (wrong verb) the man carry carry x (should be “carrying”) mm

eeto basket

in the many leaf window?

mm wind ok ok oh aa

309 now man very very the man is very angry? angry angry ok finish fine

Correct verb forms: 4 of 8 correct Error-free clauses: 2 of 7 correct Unaltered repetitions: 4 Turns: 15 Word count: 59 Words per turn: 3.93 Type-token ratio: 59 types ÷ 26 tokens = .44

Key: = Error-free clauses = Correct verb forms = Unaltered repetitions x = entity is incorrect \ = Start of transcript (j) = Japanese words (solid line connects an entity separated in the transcript)

310 APPENDIX PP EXAMPLE OF TRANSCRIPT FROM THE NARRATIVE TASK-LIMITED CHOICE OF TOPIC

Group B, Class 2, male-male pair. (Topic in Appendix S.)

Yeah, OK, sound check

U zero sc: oh

Check OK sc: (teacher: ok please go) one turn \First

Yes

aa The boy into pac-, a bucket into sandwich (no verb so not counted) Yeah x (missing an auxilairy verb and an article) And girl cut bread x (should be “cutting”) OK

and Next.

next x (wrong selction of verb and hard to understand) eeee mm, the mmm their mother uh, ah? teach map

OK x (should be “teaching”)

OK. Next.

Into the baggage, ba- dog (no verb so not counted)

311 x (should be “going”) (out of context) x OK. Yes, yes. Next, I’m go home x (out of context, should be in the third person) Go home?(*)

Go home(*), aaa this this children

OK go home(*) (j) (j) x (out of context) I’m home (j) bye-bye ok x (out of context) yeah x (missing an article) Next, mmm they climb mountain x (should be “are climbing”) Oh, climb mountain(*) x x And next, the- aa they surprise uh, because aa see it small dog ah, (j) aa (j) x (The student seems to be saying Basket “They see it” or “They see a small dog” so the verb is correct but the sentence is incorrect.) mm bas- ok x (wrong verb and missing an article)

Basket And next, aaa into sandwich, is eee eaten by this dog

OK (j) x (should be “was” or “has been”) sad story

(*) = Exact repetition of utterance by partner in previous turn; therefore not counted (Appendix II; Guideline #9).

312 Correct verb forms: 2 of 10 correct Error-free clauses: 0 of 8 correct Unaltered repetitions: 1 Turns: 20 Word count: 74 (word added: sandwich{+ 1 type | + 1 token}) Words per turn: 3.70 Type-token ratio: 74 types ÷ 39 tokens = .53

Key: = Error-free clauses = Correct verb forms = Unaltered repetitions x = entity is incorrect \ = Start of transcript (j) = Japanese words (solid line connects an entity separated in the transcript)

313 APPENDIX QQ EXAMPLE OF TRANSCRIPT FROM THE NARRATIVE TASK-COMPLETE CHOICE OF TOPIC

Group B, Class 1, female-female pair. (Topic in Appendix T.)

(teacher: please go) please go (student says student number) one h (j)

OK sc: (teacher: ok please go) x (wrong verb) one turn \My, my (j) , I talk my, I talk you happy, happy story

Happy story x (should be “tell”)

Happy story.

aaa Yesterday,

eeto oh, i I

went to went to Kobe (a university in Kobe). (j) aaa There,

mmm

there is,

mm (This repetition is counted because it is in the same turn.) From here to the next full clause, “There were aaa soccer there is game,” there are many false starts and other-supplied corrections. These are not considered in the total for either correct mm verbs forms or error-free clauses.

314 (j)

There are

There was mm?

There were

There

Were x (should be “was”) There were aaa soccer game soccer game x (wrong verb) Soccer game

Soccer game (j)

Soccer, soccer game, soccer game

eee last countable turn mmm

eee Hyogo u- (a university in Hyogo) v-s Osaka Osaka (a university in Osaka) eee (j)

(j)

aaa aaaa aaaato eeeto eeeto (j)

315 Correct verb forms: 1 of 3 correct Error-free clauses: 1 of 3 correct Unaltered repetitions: 8 Turns: 11 Word count: 50 (words added: university in Hyogo in the same turn {proper noun, + 2 type | + 1 token}; university in Kobe in the same turn {proper noun, + 1 type | + 1 token}; university in Osaka in the same turn {proper noun, + 1 type | + 1 token}; soccer {+ 7 type | + 1 token}) Words per turn: 5.55 Type-token ratio: 50 types ÷ 19 tokens = .38

Key: = Error-free clauses = Correct verb forms = Unaltered repetitions x = entity is incorrect \ = Start of transcript (j) = Japanese words (solid line connects an entity separated in the transcript)

316 APPENDIX RR EXAMPLE OF TRANSCRIPT FROM THE DECISION- MAKING TASK-NO CHOICE OF TOPIC

Group B, Class 1, male-female pair. (Topic in Appendix U.)

OK, first question, Shinzo Abe (teacher: please say your student number) student number

(male student says student number)

(female student says student number)

(teacher: sound check ok?) ok (teacher: everybody ok ok please go) ok

\first shinzo abe abe abe shinzo mmmm mmm

Japan x Japan will do do future (In this task, the students are making questions to ask world leaders. The student seems to be saying, “What will Japan Oh do for the future.” The student also seems to be making a question so there is the modal auxiliary “will” and the primary verb “do” (Quirk et al., 1985, p. 136). In a question Oh format the verbs should have been separated by the noun “Japan” and therefore the clause is incorrect. However, ac- Good question cording to the guidelines, auxilary verbs and primary verbs in questions are separated and counted individually. In the format of this question, asking about the future, both verbs Heaviest are correct so there are two correct verbs in this clause.)

Heaviest heaviest question, mmm? mmm

317 oh second Putin

Putin

Russia

This winter is warmer winter

Oh x (should be “was”) Ru- Russia is very cold

Mmm x (wrong verb tense) last year, mmmm but this year is very hot

Oh, mmm global warming?

Global warming

Global warming mmmm Bush

Bush, George Bush (Same as on previous page about auxiliary Bush, um, Bush? verbs.)

Iraq

Oh, Iraq war

Iraq war x (should be “will Iraq have” or “will there be”)

Iraq war will do next year x (difficult to understand)

318 Correct verb forms: 5 of 7 correct Error-free clauses: 2 of 5 correct Unaltered repetitions: 2 Turns: 21 Word count: 57 (words added: George Bush or Bush in the same turn {proper noun, + 3 types | + 1 token}; Putin {proper noun, + 2 types | + 1 token}; Russia {proper noun, + 3 types | + 1 token}; Iraq {+ 4 types | + 1 token}; Shinzo Abe, Abe Shinzo, or Abe in the same turn {+ 2 types | + 1 token}; Japan {+ 2 types | + 1 token}) Words per turn: 2.71 Type-token ratio: 57 types ÷ 28 tokens = .49

Key: = Error-free clauses = Correct verb forms = Unaltered repetitions x = entity is incorrect \ = Start of transcript (j) = Japanese words (solid line connects an entity separated in the transcript)

319 APPENDIX SS EXAMPLE OF TRANSCRIPT FROM THE DECISION- MAKING TASK-LIMITED CHOICE OF TOPIC

Group B, Class 1, female-female pair. (Topic in Appendix AA.)

(teacher: last time {for the tasks})

yeah

(j) (teacher: please say your student number)

zero # # # # k

u # # # # e (teacher: sound check ok) ok (teacher: please go) one (The present tense in this case is turn \what people do you choose? correct because this is an activity that is in progress.) Wha- what

Mm xxx

I I choose aaa aa I choose mmm I choose tomonori jinnai and norika fujiwara

And why x (used Japanese to finish the phrase) Aa tomonori jinnai he is (j: comedian)

(j: I can’t write)

but

320 he marries

mm x (wrong verb tense) he he marry actor actress x (should be “married”) actress

actress

so ok

(j: I can’t believe it)

(j)

aa ok what people do you choose? (The present tense in this case is correct because this is an activity that is in progress.) Mmm I I choose daisuke matsuzaka

Mm why? last countable turn Because he is he is a great baseball player aaa and he is a little a little aa he is a clever Mm x (should be “chose”) Clever so he choose (Although the two minutes were up at this point and a phrase was not counted, this verb was included in the correct verb form count.)

321 Correct verb forms: 10 of 12 correct Error-free clauses: 7 of 9 correct Unaltered repetitions: 5 Turns: 10 Word count: 66 (words added: Tomonori Jinnai {proper noun, + 2 types | + 1 token}; Norika Fujiwara {proper noun, + 1 type | + 1 token}; Daisuke Matsuzaka {proper noun, + 1 type | + 1 token}) Words per turn: 6.60 Type-token ratio: 66 types ÷ 26 tokens = .39

Key: = Error-free clauses = Correct verb forms = Unaltered repetitions x = entity is incorrect \ = Start of transcript (j) = Japanese words (solid line connects an entity separated in the transcript)

322 APPENDIX TT EXAMPLE OF TRANSCRIPT FROM THE DECISION- MAKING TASK-COMPLETE CHOICE OF TOPIC

Group B, Class 2, male-male pair. (Topic in Appendix CC.)

(teacher: please say your student number)

U zero six zero three zero d

Yeah

(teacher: sound check ok?)

Yeah

(teacher: everybody ok)

Everybody OK oh

Aaa

(teacher: do your best) one turn \My my own topic is the ozone hole

The ozone hole

Ozone hole

Oh yeah x (used Japanese to finish the phrase) Aaa ozone hole is (j) aaaa (j)

(j)

323 The cause of the ozone hole ozone hole is aaa (j) freon gas freon gas freon gas freon gas gas gas gas (j) gas x (missing an article; should be freon is aaa (j) generated spray “generated as a spray”) spray? spray

(sound effect) spray yeah so cooler (j)

Ooo cooler (j) refrigerator refrigerator (j)

324 Mm (j)

Maa (j)

(j) Ok

Aaa ozone ozone (j: what did you say?)

Aaa ozone lay- lay-

Ray (j: is a ray)

(j: a ray?) x (missing an article) Ozone hole is aa if ozone hole is big

Big oo big big

Hmm (noise from another pair and teacher) Oooo

equals

(j)

(j) Violet

Violet last countable turn Ray (j: mistake)

Ooo

(j) (j) ray (j)

(j)

(j) (j) wait wait

325 Correct verb forms: 5 of 5 correct Error-free clauses: 2 of 5 correct Unaltered repetitions: 5 Turns: 24 Word count: 63 (word added: Freon {+ 5 types | + 1 token}; ozone {+ 11 types | + 1 token}; refrigerator {+ 2 types | + 1 token}; spray {+ 4 types | + 1 token}; violet {+ 2 types | + 1 token}; Words per turn: 2.63 Type-token ratio: 72 types ÷ 22 tokens = .35

Key: = Error-free clauses = Correct verb forms = Unaltered repetitions x = entity is incorrect \ = Start of transcript (j) = Japanese words (solid line connects an entity separated in the transcript)

326 APPENDIX UU PROCEDURES TO PREPARING DATA FOR FINAL ANALYSIS

Original participants who completed all treatment sessions (N = 96)

Missing values replaced

Univariate outliers removed (N = 11)

Descriptive statistics

Data transformation

Multivariate outliers removed (N = 7)

Factor analysis (N = 78)

Separation of items into each factor

Check of fit (Rasch)

Removal of item not fitting (Rasch)

Check of validity of response categories (Rasch)

Collapsing of response categories (Rasch)

Final set of person measures in logit scale (Rasch)

Data transformation (Rasch)

New measures entered into separate SPSS files for each variable

327