• Aucun résultat trouvé

The Potential of Interactive Speech-Enabled CALL in the Swiss Education System: A Large-Scale Experiment on the Basis of English CALL-SLT

N/A
N/A
Protected

Academic year: 2022

Partager "The Potential of Interactive Speech-Enabled CALL in the Swiss Education System: A Large-Scale Experiment on the Basis of English CALL-SLT"

Copied!
355
0
0

Texte intégral

(1)

Thesis

Reference

The Potential of Interactive Speech-Enabled CALL in the Swiss Education System: A Large-Scale Experiment on the Basis of English

CALL-SLT

BAUR, Claudia

Abstract

L'objectif principal de cette thèse est l'évaluation du potentiel de l'ELAO dans le système scolaire suisse. Dans ce but, nous avons développé un système de dialogue interactif (CALL-SLT) qui utilise la reconnaissance de la parole pour apprendre l'anglais. Nous avons adapté ce système au contenu enseigné dans les classes et l'avons conçu spécifiquement pour des élèves germanophones du degré secondaire I. Dans une expérience à grande échelle nous avons évalué d'une part si l'utilisation régulière du logiciel CALL-SLT aidait les élèves à améliorer leurs compétences communicatives en anglais et d'autre part si les élèves et les enseignants appréciaient l'intégration de l'ELAO dans l'enseignement traditionnel. Pour répondre à ces questions de recherche, nous avons pris en compte des mesures tant objectives que subjectives, comme un test d'évaluation écrit, les enregistrements des interactions orales avec le système, ainsi que plusieurs questionnaires récoltant des données personnelles sur les participants et le feedback qualitatif des élèves et des enseignants.

BAUR, Claudia. The Potential of Interactive Speech-Enabled CALL in the Swiss

Education System: A Large-Scale Experiment on the Basis of English CALL-SLT. Thèse de doctorat : Univ. Genève, 2015, no. FTI 26

URN : urn:nbn:ch:unige-818783

DOI : 10.13097/archive-ouverte/unige:81878

Available at:

http://archive-ouverte.unige.ch/unige:81878

(2)

The Potential of Interactive Speech-Enabled CALL in the Swiss Education System:

A Large-Scale Experiment on the Basis of English CALL-SLT

Thèse

Présentée à la Faculté de traduction et d’interprétation de l’Université de Genève

pour obtenir le grade de Docteur en Traitement Informatique Multilingue par

Claudia Baur

Jury :

Prof. Pierrette Bouillon, Faculté de traduction et d’interprétation, Université de Genève (Directrice de thèse)

Prof. Sonia Halimi, Faculté de traduction et d’interprétation, Université de Genève (Présidente du Jury)

Dr. Monica Masperi, Directrice du service LANSAD, Université Stendhal - Grenoble 3 (Jurée externe)

Dr. Emmanuel Rayner, Faculté de traduction et d’interprétation, Université de Genève (Juré)

Prof. Martin Russell, School of Electronic, Electrical and System Engineering, University of Birmingham (Juré externe)

Soutenue le 18 décembre 2015 à l’Université de Genève

(3)
(4)

Frank Smith

(5)
(6)

Communicative second language skills are becoming increasingly important in today’s fast-moving and global world. There are a number of approaches to attain these competences, such as conversation trainings with native speakers, language stays abroad, and traditional language courses. Whereas the first two approaches are time- and cost-intensive, the third alternative is the most commonly used approach to Second Language Acquisition for children of school age. However, one big challenge of this latter approach is the lack of time for conversational training during limited teaching hours. A promising approach to tackle this problem is Computer Assisted Language Learning (CALL). The central objective of this thesis is to evaluate the potential of CALL in the Swiss education system. For this purpose we de- veloped an interactive speech-enabled English dialogue course (CALL-SLT) for Germanophone lower secondary school students that is customized to the content taught in traditional school classes. In a large-scale experi- ment – including 15 school classes – we evaluated whether the regular use of CALL-SLT helped students improve their communicative second lan- guage skills and whether the integration of CALL in traditional teaching was well received by both students and teachers. To assess these questions, both objective and subjective measures were taken into account, such as written placement tests, recordings of spoken system interactions, and var- ious questionnaires providing information on participants’ personal back- grounds, and students’ and teachers’ qualitative feedback on CALL-SLT.

Results of this experiment support our hypothesis that students who reg- ularly trained their competences with CALL-SLT significantly improved their post-experimental competences. We demonstrated that the frequent use of CALL-SLT resulted in statistically significant improvement in sec- ond language (L2) oral competences, such as pronunciation, fluency, and the

(7)

could be transferred to written language by demonstrating that highly active CALL-SLT users statistically significantly improved their post-experimental written vocabulary competences. Qualitative feedback from students and teachers also suggested that CALL-SLT was a well-received supplement to traditional teaching methods. The evaluation of the potential of gamifica- tion elements, such as scores and badges, as a motivation enhancer could not be proven in this study. However, although gamification did not prove to enhance students’ motivation from an objective point of view, subjec- tive user feedback indicated that the gamified version of CALL-SLT was preferred by a majority of all participants.

(8)

Dans notre soci´et´e mondialis´ee et en rapide ´evolution, les comp´etences com- municatives en langues secondes prennent une importance croissante. Il existe plusieurs m´ethodes pour acqu´erir ces comp´etences, telles que des ex- ercices de conversation avec des personnes de langue maternelle, des s´ejours linguistiques `a l’´etranger ou des cours de langue traditionnels. Alors que les deux premi`eres approches sont souvent coˆuteuses et chronophages, la troisi`eme est celle la plus souvent adopt´ee pour l’acquisition des langues sec- ondes par des enfants en ˆage scolaire. N´eanmoins, cette approche pr´esente un d´efi majeur, car le temps qu’elle consacre `a des exercices de conversation est insuffisant du fait du nombre limit´e d’heures de cours. L’enseignement des langues assist´e par ordinateur (ELAO) constitue une solution promet- teuse pour corriger ce d´efaut. L’objectif principal de cette th`ese est l’´evaluation du potentiel de l’ELAO dans le syst`eme scolaire suisse. Dans ce but, nous avons d´evelopp´e un syst`eme de dialogue interactif (CALL-SLT) qui utilise la reconnaissance de la parole pour apprendre l’anglais. Nous avons adapt´e ce syst`eme au contenu enseign´e dans les classes et l’avons con¸cu sp´ecifiquement pour des ´el`eves germanophones du degr´e secondaire I. Dans une exp´erience `a grande ´echelle – comprenant 15 classes – nous avons ´evalu´e d’une part si l’utilisation r´eguli`ere du logiciel CALL-SLT aidait les ´el`eves

`

a am´eliorer leurs comp´etences communicatives en anglais et d’autre part si les ´el`eves et les enseignants appr´eciaient l’int´egration de l’ELAO dans l’enseignement traditionnel. Pour r´epondre `a ces questions de recherche, nous avons pris en compte des mesures tant objectives que subjectives, comme un test d’´evaluation ´ecrit, les enregistrements des interactions orales avec le syst`eme, ainsi que plusieurs questionnaires r´ecoltant des donn´ees personnelles sur les participants et le feedback qualitatif des ´el`eves et des enseignants. Les r´esultats de l’exp´erience soutiennent notre hypoth`ese :

(9)

fr´equente de CALL-SLT r´esulte en une am´elioration statistiquement signi- ficative des comp´etences orales en langue seconde (L2), comme la pronon- ciation, la fluidit´e, et l’utilisation d’un vocabulaire ad´equat en conversation orale. Concernant les comp´etences ´ecrites en L2, nous avons montr´e que des comp´etences en vocabulaire acquises oralement peuvent ˆetre transf´er´ees

`

a la langue ´ecrite. En effet, les ´el`eves tr`es actifs sur CALL-SLT pr´esentaient de meilleures comp´etences ´ecrites du vocabulaire apr`es l’exp´erience, et ce de mani`ere statistiquement significative. Par ailleurs, le feedback qualitatif des

´el`eves et des enseignants sugg`ere que CALL-SLT constitue un compl´ement bienvenu aux m´ethodes d’enseignement classiques. Notre travail n’a pas per- mis de d´emontrer le potentiel des ´el´ements visant `a rendre l’apprentissage plus ludique – tels les scores et les badges – pour renforcer la motivation des ´el`eves. Cependant, si l’utilisation d’´el´ements ludiques (gamification) n’a pas r´ev´el´e une augmentation de la motivation d’un point de vue objectif, le feedback subjectif des ´el`eves a quant `a lui indiqu´e que la plupart des participants ont pr´ef´er´e la version ludifi´ee de CALL-SLT.

(10)

First and foremost, I would like to express my special appreciation and thanks to my advisor Professor Pierrette Bouillon, for her continuous sup- port, her patience, motivation, enthusiasm, and her guidance throughout the process of writing this thesis. My profound gratitude also goes to Manny Rayner, whose experience and enthusiasm have been an invaluable source of inspiration for my research. Their ideas, suggestions and comments helped me move from an idea to a completed study. I am also very appreciative of Doctor Monica Masperi and Professor Martin Russell for having read- ily accepted to be part of my thesis committee; I feel deeply honored to have them in my jury. Special mention goes to Aur´elie Picton for accepting the role of president of the jury and for her motivating words along the way.

Many thanks go to the teachers and students whose participation in my experiments made this thesis possible in the first place, with special men- tion to M. Pflugshaupt for his assistance as a subject matter expert. I am furthermore grateful to the three annotators who took the time to carefully listen to thousands of non-native sound files. A warm thank you goes to Robert for his many favors throughout the past years. I would like to ex- tend my appreciation to my colleagues at the FTI/TIM department, Nikos, Victoria, Marianne, Luc´ıa, Silvia and Violeta, for their encouragement and friendship during the past years. Special mention goes to Johanna and To- bias for having been such great and supportive office colleagues and friends, and for bearing with me during all of my ups and downs.

I am furthermore grateful to the Swiss National Science Foundation (SNSF) for their support in the CALL-SLT 2 project “Designing and Evaluating

(11)

Last, but by no means least, I would like to thank my family and my partner for all of their love and encouragement. I would like to dedicate this thesis to them. I feel deeply grateful to my parents, Sylvia and Andreas, who raised me with a love of science and who have been an invaluable source of support, always encouraging me to strive towards my goals. I am greatly thankful to my brother Daniel for having had an open ear to all my doubts and complaints and for knowing how to cheer me up again. Finally, I would like to address a very special thanks to my loving, encouraging, patient, and supportive partner Florian.

(12)

List of Figures ix

List of Tables xiii

1 Introduction 1

1.1 Motivation . . . 1

1.2 Context . . . 6

1.3 Objectives . . . 11

1.4 Methodology . . . 13

1.5 Dissertation Structure . . . 14

1.6 Published Work . . . 16

I State of the Art 17 2 Second Language Acquisition 19 2.1 Introduction . . . 19

2.2 Historic Perspectives in SLA . . . 20

2.2.1 Behaviorism . . . 20

2.2.2 Contrastive Analysis Hypothesis . . . 23

2.2.3 Universal Grammar . . . 25

2.2.4 Error Analysis . . . 29

2.2.5 Monitor Model . . . 30

2.3 Recent Perspectives in SLA . . . 32

2.3.1 Cognitivist Approach . . . 32

2.3.1.1 Interaction Hypothesis . . . 33

(13)

2.3.1.2 Noticing Hypothesis . . . 34

2.3.2 Functional Approach . . . 35

2.3.3 Sociocultural Approach . . . 37

2.4 Motivation . . . 39

2.4.1 Social Psychological Period . . . 39

2.4.2 Cognitive Situated Period . . . 41

2.4.3 Process-Oriented Period . . . 42

2.4.4 Motivation Evaluation . . . 45

2.4.5 Motivation in CALL and Gamification . . . 47

2.5 Summary . . . 49

3 Technology in Second Language Acquisition 53 3.1 Introduction . . . 53

3.2 Structural CALL . . . 55

3.2.1 Programmed Instruction . . . 55

3.2.2 Audio-Lingual Approach . . . 58

3.2.3 Structural CALL Projects . . . 60

3.3 Communicative CALL . . . 62

3.3.1 Communicative CALL Projects . . . 63

3.4 Integrative CALL . . . 65

3.4.1 Natural Language Processing . . . 66

3.4.1.1 Speech Recognition . . . 67

3.4.2 Integrative CALL Projects . . . 69

3.4.2.1 Pronunciation Training . . . 70

3.4.2.2 Closed Response Conversational Training . . . 72

3.4.2.3 Open Response Conversational Training . . . 75

3.4.2.4 Translation Game . . . 76

3.5 Summary . . . 81

II CALL-SLT 85 4 CALL-SLT 87 4.1 Introduction . . . 87

(14)

4.2 CALL-SLT Architecture . . . 90

4.2.1 Grammar-based Speech Recognizer . . . 91

4.2.1.1 Statistical Language Models . . . 95

4.2.1.2 Linguistic Language Models . . . 95

4.2.2 Interlingua-based Machine Translation . . . 97

4.2.2.1 Grammar and Lexicon Rules . . . 99

4.2.2.2 Translation Rules . . . 100

4.3 Interface . . . 103

4.4 Help Function . . . 106

4.5 Prompts . . . 107

4.5.1 Text Prompts . . . 108

4.5.2 Multimedia Prompts . . . 109

4.5.2.1 Image Prompts . . . 109

4.5.2.2 Audio Prompts . . . 109

4.5.2.3 Video Prompts . . . 110

4.6 System Feedback . . . 111

4.7 Gamification Elements . . . 113

4.8 Previous CALL-SLT Evaluations . . . 115

4.9 Summary . . . 118

5 CALL-SLT in the SLA Classroom 121 5.1 Introduction . . . 121

5.2 Language Level . . . 123

5.2.1 Communicative Competences . . . 124

5.2.1.1 Linguistic Competences . . . 125

5.2.1.2 Sociolinguistic Competences . . . 128

5.2.1.3 Pragmatic Competences . . . 128

5.3 Pedagogical Approach . . . 130

5.4 Customization of CALL-SLT . . . 133

5.4.1 English Course “Ready for English” . . . 134

5.4.2 Lesson Development . . . 135

5.4.2.1 Dialogue Structure . . . 137

5.4.2.2 Prompts . . . 143

(15)

5.4.2.3 System Feedback . . . 144

5.4.2.4 Gamification Elements . . . 144

5.5 Summary . . . 146

III Experiments and Evaluations 149 6 Evaluation Methodology 151 6.1 Introduction . . . 151

6.2 Experiment Methodology . . . 152

6.2.1 Experiment Setup . . . 153

6.2.2 Information Policy . . . 155

6.2.3 Data Collection . . . 157

6.2.4 Participants . . . 162

6.2.5 Activity Groups . . . 163

6.2.6 Pilot Experiment . . . 166

6.2.6.1 Findings . . . 167

6.3 Evaluation Designs . . . 168

6.3.1 Control Group Experiment Design . . . 169

6.3.1.1 Historical Background . . . 169

6.3.1.2 Methodology . . . 171

6.3.1.3 Control Group Design in CALL-SLT . . . 172

6.3.2 Single-Case Experiment Design . . . 172

6.3.2.1 Historical Background . . . 174

6.3.2.2 Methodology . . . 175

6.3.2.3 Single-Case in CALL-SLT . . . 177

6.4 Statistical Tests . . . 178

6.5 Summary . . . 179

7 Evaluations and Results 183 7.1 Introduction . . . 183

7.2 Written Competences . . . 184

7.3 Oral Competences . . . 191

7.4 Influence of Personal Background on Results . . . 200

(16)

7.5 Difficulty . . . 208

7.6 Help . . . 217

7.7 Prompt Design . . . 220

7.8 Appreciation . . . 221

7.9 Evaluation of Teacher Questionnaires . . . 225

7.10 Evaluation Motivation and Gamification . . . 230

7.10.1 Objective Evaluation . . . 232

7.10.2 Subjective Evaluation . . . 236

7.11 Summary . . . 238

8 Conclusion 243 8.1 Achievements . . . 243

8.2 Limitations . . . 246

8.3 Future Work . . . 249

References 253 Appendices 265 A Language Level: CEFR Definitions 267 A.1 General Linguistic Range . . . 267

A.2 Vocabulary Range . . . 268

A.3 Grammatical Accuracy . . . 269

A.4 Phonological Control . . . 270

A.5 Sociolinguistic Appropriateness . . . 270

A.6 Spoken Fluency . . . 271

A.7 Propositional Precision . . . 272

B Comparison of English Textbook Units 273 C English Course – Shopping Lesson 279 C.1 Dialogue Script – Shopping Lesson . . . 279

C.2 Corpus – Shopping Lesson . . . 285 D Information Letter and Consent Form 295

E Instructions CALL-SLT 299

(17)

F Written Placement Test 303

G Pre-Experiment Questionnaire 307

H Post-Experiment Questionnaire 315

I Evaluation Written Placement Test 321

J Teacher’s Questionnaire 329

(18)

1.1 Mode of communication . . . 2

1.2 Error tolerance vs. competence in content . . . 2

1.3 The Swiss education system (EDK (2013)) . . . 8

1.4 Order of introduction of foreign languages in the Swiss school system (Hutterli (2012)) . . . 9

1.5 HarmoS participating cantons (EDK (2010)) . . . 10

3.1 Tell Me More phoneme pronunciation exercise . . . 73

3.2 Translation Game system-user-interaction . . . 77

3.3 Interface of the Translation Game prototype . . . 78

3.4 Paraphrasing of input utterances . . . 79

3.5 Translation Game processing . . . 79

4.1 CALL-SLT scheme . . . 92

4.2 Speech recognition process (Rayner et al. (2006)) . . . 93

4.3 The two levels of abstraction in Regulus (Rayner et al. (2006), p. 15) . . 98

4.4 Interface of a plain text-prompt course (L1 = FR, L2 = IT) . . . 104

4.5 Interface of a multimedia dialogue course (L1 = DE, L2 = EN) . . . 105

4.6 L1-independent image prompt (L1 = any, L2 = EN) . . . 110

4.7 L1-independent audio prompt (L1 = any, L2 = EN) . . . 110

4.8 Color-coded system feedback . . . 112

4.9 Informative system feedback . . . 112

4.10 Badge overview table . . . 115

5.1 Dialogue draft for the “Hotel” lesson (light grey = video prompt; white = text prompt; dark grey = user’s response) . . . 138

(19)

5.2 Badges for the “Restaurant” lesson . . . 146

6.1 CALL-SLT Internet page . . . 156

6.2 Gender distribution across classes . . . 163

6.3 Frequency distribution of activity groups across classes . . . 165

7.1 Comparison pre- vs. post-performance per exercise . . . 187

7.2 Class comparison of pre- vs. post-performance . . . 188

7.3 Students’ subjective perception of CALL-SLT’s impact on vocabulary and grammar skills . . . 190

7.4 Student count per class for oral competence evaluation . . . 192

7.5 Logged recognition rates per activity group . . . 199

7.6 Student’s subjective opinion on CALL-SLT’s potential to train oral L2 competences . . . 201

7.7 Students’ L1 per activity group . . . 203

7.8 Students’ bilingualism per activity group . . . 203

7.9 Students’ motivation for learning additional L2s per activity group . . . 204

7.10 Students’ subjective L2 assessment per activity group . . . 205

7.11 Students’ subjective L2 motivation per activity group . . . 206

7.12 Students’ tech-savviness per activity group . . . 207

7.13 Students’ previous experience with CALL and speech recognition per activity group . . . 208

7.14 Students’ perception of the tasks’ difficulty . . . 210

7.15 Students’ perception of how to respond . . . 211

7.16 Students’ perception of the receptive tasks’ difficulty . . . 211

7.17 Recognition rate and help usage per lesson . . . 212

7.18 Students’ perception of the written and spoken help function . . . 218

7.19 Usefulness of spoken help for training of pronunciation competences . . 218

7.20 Recognition rate depending on help consultation . . . 219

7.21 Preferred prompt type and design . . . 220

7.22 Prompt design options . . . 221

7.23 Students’ main reason to use CALL-SLT . . . 223 7.24 Mean logins per parallel class (white = not gamified, shaded = gamified) 233

(20)

7.25 Mean interactions per parallel class (white = not gamified), shaded =

gamified) . . . 234

7.26 Mean interactions per login per parallel class (white = not gamified), shaded = gamified) . . . 235

7.27 Students’ overall appreciation of CALL-SLT . . . 237

7.28 Appreciation of gamified vs. non-gamified CALL-SLT version . . . 238

I.1 Pre- vs. post-scores across classes and activity levels (1) . . . 322

I.2 Pre- vs. post-scores across classes and activity levels (2) . . . 323

I.3 Pre- vs. post-scores across classes and activity levels (3) . . . 324 I.4 Pre- vs. post-score range per student across classes and activity levels (1) 325 I.5 Pre- vs. post-score range per student across classes and activity levels (2) 326 I.6 Pre- vs. post-score range per student across classes and activity levels (3) 327

(21)
(22)

4.1 Language pairs available in CALL-SLT . . . 89 5.1 CEFR levels to be attained after lower secondary school for English as

an L2 . . . 124 5.2 CALL-SLT lesson overview . . . 136 5.3 Total number of different prompts per lesson . . . 139 5.4 Badge criteria for English dialogue game . . . 146 6.1 Written placement test exercise overview . . . 158 6.2 Overview of participating school classes . . . 162 6.3 Overall age distribution . . . 164 6.4 Activity group categorization criteria . . . 164 6.5 Frequency distribution of activity groups across all participants . . . 166 6.6 System recognition rate . . . 168 6.7 Research question 1: Can CALL-SLT help students improve their com-

municative L2 skills? . . . 181 6.8 Research question 2: Is CALL-SLT a justified supplement to traditional

L2 teaching methods that is well received by both teachers and students? 181 6.9 Research question 3: Does a student’s personal background have an

influence on his or her motivation? . . . 182 6.10 Research question 4: Does gamification help motivate students? . . . 182 7.1 Placement test exercise 3: response length per activity group . . . 186 7.2 Inter-rater agreement between three English native annotators . . . 194 7.3 Inter-rater agreement between three English native annotators and teacher

1 . . . 194

(23)

7.4 Inter-rater agreement between three English native annotators and teacher 2 . . . 194 7.5 Paired t-test for oral pronunciation competences . . . 196 7.6 Paired t-test for oral fluency competences . . . 197 7.7 Paired t-test for oral vocabulary competences . . . 197 7.8 Paired t-test for oral grammar competences . . . 197 7.9 Correlation system recognition with human annotations . . . 199 7.10 Development of students’ subjective motivation to learn English . . . . 206 7.11 Number of attempts per lesson . . . 213 7.12 Most frequently failed prompts per lesson . . . 215 7.13 Correct vs. incorrect utterances per lesson, based on human annotations 216 7.14 Students’ subjective appreciation of CALL and CALL-SLT . . . 222 7.15 Teachers’ main reason to participate in the CALL-SLT project . . . 230 7.16 Teachers’ feedback concerning organizational matters . . . 231 7.17 Teachers’ appreciation of CALL-SLT . . . 231 7.18 Average number of badges across gamified and non-gamified classes . . . 236 B.1 Units “Ready For English 1” . . . 275 B.2 Units “New Snapshot Starter” . . . 278

(24)

Introduction

“You can never understand one language until you understand at least two”– Geoffrey Willans

1.1 Motivation

Communicative second language skills are becoming increasingly important in today’s fast-moving and global world. There are a number of approaches to attain these com- petences, such as conversation trainings with native speakers, language stays abroad, and traditional language courses. Whereas the first two approaches are time- and cost-intensive, the third alternative is the most commonly used approach to Second Language Acquisition (SLA) for children of school age. However, one big challenge of this latter approach is the lack of time for conversational training during limited teaching hours. A promising approach to tackle this problem is Computer Assisted Language Learning (CALL). The central objective of this thesis is to evaluate the potential of CALL in the Swiss education system. For this purpose we developed an in- teractive speech-enabled English dialogue course (CALL-SLT) for Germanophone lower secondary school students that is customized to the content taught in traditional school classes.

According to Hutterli (2012), communicative language skills occupy the first tier of the hierarchy in the model of foreign language competences, as is illustrated in Figure 1.1. The model shows that spoken production, spoken interaction, spoken comprehen-

(25)

sion and speech-to-speech mediation are key competences that should be trained and evaluated during the acquisition of foreign languages, based on theCommon Framework of Reference for Languages(CEFR) (Council of Europe (2001)).

Figure 1.1: Mode of communication

Today’s cognitive psychology and functional language acquisition research promote the importance of communicative second language (L2) competences, next to formal aspects, such as grammar or spelling (Hutterli (2012)). Especially during the early phases of the L2 acquisition process, students seem to learn more effectively if a higher error tolerance is granted, as is the case in conversational or communicative language teaching. Figure 1.2 illustrates the relationship between error tolerance and competence in content across the CEFR competence levels (Westhoff (2007)).

Figure 1.2: Error tolerance vs. competence in content

In addition, Willis (1996) introduced a four-point framework to capture the most important aspects that favor successful language learning. The framework includes the concepts of exposure, use, motivation and instruction, of which at least the first three need to be fulfilled for successful language learning. Exposure to the language

(26)

indicates the necessity to be in contact with the L2 in order to obtain real input of both written and spoken language in real situations. Use of the language is the second step in which the student makes use of the language to communicate and understand meaning;motivationmust be at hand for all areas, such as listening, reading, speaking and writing; andinstructionin language learning covers the formal aspects of SLA. In reality, exposure to and active use of the L2 are rather difficult to provide and are often limited to the classroom scenario. Although traditional instruction is the most practical method to teach a foreign language, Willis (1996) states that language instruction is not absolutely necessary to learn an L2, which can also be acquired “naturally” in the given environment. However, she nevertheless stresses the positive effect of instruction on the quality of L2 command in the longer run. Willis (1996), p. 23, concludes that communicative tasks in a formal setting (classroom teaching) should always be used

“for a communication purpose (goal) in order to achieve an outcome,” as this helps students put themselves in the place of real-life language use and exposure. Hutterli et al. (2008) furthermore found that young learners mostly learn implicitly, whereas older learners (> eight years) most effectively learn from a combination of explicit and implicit instruction, hence from a combination of traditional instructional teaching methods and language in use.

As introduced above, today’s SLA research suggests that the training of commu- nicative L2 competences efficiently helps students acquire an L2. However, since in- structional classroom time is limited and is mostly intended to teach the formal aspects of language, new methods are needed to cover the recently arisen and elementary need to train communicative L2 skills. Information and Communication Technologies (ICT) is a domain that is becoming increasingly prominent in today’s world, and which con- cerns all areas of our daily life and therefore also the field of education and SLA. It is a promising approach to tackle the challenges outlined above. The Swiss Department for Information Technology in Education (SFIB)1 acknowledges ICT as an integral part of education, as stated in a relatively recent publication (B¨uchner et al. (2009), p. 3):

Der Umgang mit ICT wird als neue “Kulturtechnik” – neben Schreiben, Lesen und Rechnen – zur Grundausbildung geh¨oren m¨ussen und damit auch Teil des

1Schweizerische Fachstelle f¨ur Informationstechnologien im Bildungswesen, SFIB

(27)

lebenslangen Lernens werden.1

Furthermore they state that ICT-based educational material can better account for the various individual needs that exist within a school class, compared to traditional programs (B¨uchner et al. (2009), p. 8):

ICT-gest¨utzte Lernmaterialien und Lernprogramme zeichnen sich u.a. dadurch aus, dass sie die individuellen Lernvoraussetzungen in heterogenen Klassen oft um- fassender und wirksamer ber¨ucksichtigen k¨onnen als herk¨ommliche p¨adagogisch- didaktische Medien.2

The SFIB states that the added value of the use of new media is threefold (Biffi et al.

(2007)): added value can be found on a social, on an individual, and on a classroom level. For our study, the latter two are of particular interest, since the main advan- tage on the individual level is that the computer can lead to an intensification of the learning behavior through its interactive and adaptive nature, and that multimedia, dynamic and audio-visual content can help deepen the understanding of the taught content (Petko and Reusser (2005)). The most important point at the classroom level is the possibility to differentiate and individualize courses and therefore respect the learning pace of each individual within a heterogeneous school class.

Further advantages that we see in a CALL approach, are the following:

1. CALL can be used anytime

Since face-to-face instruction time is limited – in Switzerland as little as 15% of classroom teaching time is dedicated to L2 instruction which amounts to ca. five hours per week for all L2s together – a resource that can be used outside of the planned language lessons is advantageous.

2. CALL can be used anywhere

Another advantage is the fact that CALL applications require minimal additional

1The exposure to ICT will have to be part of basic education – next to writing, reading and mathematics – and thus also be part of lifelong learning.

2ICT-based learning material and programs distinguish themselves from traditional pedagogic- didactic media, as they are able to better respect the individual learning conditions in heterogeneous classes.

(28)

resources – commonly only an Internet connection and a microphone or headset are necessary – and can hence be used practically anywhere.

3. CALL helps to time-efficiently train communicative competences Given that the time dedicated to training interactional speech in traditional class- room teaching is often restricted and it is difficult to get every single student to talk proportionally, a computer-mediated environment that imitates communica- tion scenarios with native speakers is an ideal supplement to traditional teaching methods.

4. CALL is a low-cost alternative to interactions with human native speak- ers

The practice of communicative skills in real interactions with native speakers is often regarded as one of the most effective ways of acquiring interactional skills;

however, language stays abroad are (both time- and cost-wise) not always an option. CALL applications are in this regard a low-cost and easily manageable alternative to real interactions with human native speakers of the L2.

5. CALL is easily customizable

An important advantage of dynamic CALL applications is that their content can easily be customized to the target audience and the requirements of the participants and teachers. This allows the course designer to create meaningful lessons that help students train their L2 competences.

6. CALL can help reduce anxiety

Another advantage of CALL applications is that they can help reduce students’

anxiety to engage in L2 productive speech thanks to the absence of direct judg- ment. When engaging in a CALL course, the student is solely judged by the application via system feedback, which is an efficient yet neutral way of judging a student’s performance. In a traditional school setting, both the teacher as well as fellow classmates can add to the anxiety and negatively influence students’ L2 production.

One important purpose of introducing a CALL application in the Swiss curriculum is to dedicate precious but limited classroom time to topics that need the immediate

(29)

presence of and interaction with a human teacher. Such topics are traditionally formal aspects of the L2, such as new grammatical constructions that ask for a comprehensive introduction and explanation by a teacher who can clarify any of the students’ questions or doubts. Some types of routine exercises, such as grammar exercises or vocabulary acquisition, are already outsourced to the home-study environments by means of home- work assignments. A component of L2 acquisition that is currently primarily seen as a classroom activity is the training of communicative or interactional L2 speech. Since the training of those oral productive competences is very time-consuming, it would be desirable to outsource at least part of this training to the home-study environment.

The purpose of introducing CALL in traditional L2 teaching is therefore to create an opportunity to train not only traditional competences, such as grammar and vocabu- lary, at home, but also the increasingly important communicative skills.

1.2 Context

On the basis outlined in section 1.1, the advantages of a computer assisted approach to teaching communicational competences in an L2 are evident. In this thesis we introduce such a CALL tool (CALL-SLT, c.f. for example Rayner et al. (2010b)), which is adapted for use in a blended learning scenario, therewith taking advantage of both traditional instruction and computer assisted home-study learning. The tool is specially designed to provide young beginner learners of the L2 with a platform with which they can train their oral productive communication competences in real-life dialogue scenarios.

Before addressing the objectives of this thesis and the experiment methodology in more detail in sections 1.3 and 1.4, the context in which the work of this thesis was conducted is of importance, since many aspects of the educational environment could influence our study. In this section the Swiss education system is introduced in order to set the basis for the research presented in later chapters.

Our study was conducted in the Swiss school system, where the CALL-SLT tool was integrated into lower secondary school classes. The target audience was beginner or low intermediate learners of English, with German as their native language. Figure

(30)

1.3 provides an overview of the Swiss education system, starting from Kindergarten up to tertiary education. The graph nicely indicates where our target group – the lower secondary level – is situated within the system (red frame).

The definition of the main characteristics of the lower secondary level of educa- tion (levels 2A, 2B and 2C) provided by the International Standard Classification of Education outlines the main goals of this final level of compulsory schooling (ISCED):

The lower secondary level of education generally continues the basic programmes of the primary level, although teaching is typically more subject-focused, often em- ploying more specialised teachers who conduct classes in their field of specialisation.

ISCED 2A: Programmes designed to prepare students for direct access to level 3 in a sequence which would ultimately lead to tertiary education, that is, entrance to ISCED 3A or 3B.

ISCED 2B: Programmes designed to prepare students for direct access to pro- grammes at level 3C.

ISCED 2C: Programmes primarily designed for direct access to the labour market at the end of this level (sometimes referred to as terminal’ programmes).

When students enter the lower secondary level of schooling in Switzerland, they have typically already attended two years of pre-school and four to six years of primary school. The goal of lower secondary school is then to further deepen the already ac- quired skills and knowledge and to expand on new subject matters in order to prepare students for higher school levels or to prepare them for the labour market.

In Switzerland, children start being exposed to a first foreign language as early as primary school, although usually by means of playful approaches. In lower secondary school, pupils then start to learn L2s in a more structured and performance-oriented way. The order in which foreign languages are introduced varies from canton to can- ton, as can be seen in Figure 1.4. The cantons highlighted in light and dark grey teach another national language as their first L21, whereas the cantons highlighted in white teach English as the first L2 and only introduce another national language as a second L2.

1German, French or Italian, depending on the cantons’ official first language

(31)

Figure 1.3: The Swiss education system (EDK (2013))

(32)

Figure 1.4: Order of introduction of foreign languages in the Swiss school system (Hutterli (2012))

As indicated by theSwiss Education Report 2010(Annen et al. (2011)), Switzerland is rather well situated as far as class time dedication to foreign languages is concerned in the lower secondary level. In German-speaking Switzerland, 15% of total class time is dedicated to foreign language teaching, compared to only 11%-12% in similar Western European countries, such as Austria, Germany or France. However, it is evident both from the graph in Figure 1.3 as well as from the map in Figure 1.4 that the Swiss school system is anything but uniform. This makes it difficult to compare performance on a national level and to introduce uniform teaching material and pedagogical approaches across cantons.

A big national project – which started in 2006 after having been approved by a popular vote with a ratio of 86% “yes”-votes – tries to tackle this problem by intro- ducing standards to harmonize the school system and curriculum across all 26 Swiss cantons. The HarmoS1 concordat has the goal of harmonizing the schooling system for

1Intercantonal Agreement on Harmonisation of Compulsory Education

(33)

compulsory education in all participating cantons, mainly accounting for the current inconsistencies in duration of the various school levels (c.f. Figure 1.3) and teaching content. However, since participation in the project is not mandatory, a certain lack of uniformness will continue to exist in the future. Figure 1.5 shows the participating cantons in green (58%), the seven cantons who refused the agreement in red (27%) and the four cantons whose decisions are still pending (15%). With regard to popu- lation ratio, the decisions can be broken down to 76.2% who are participating, 13.5%

who refused and 10.3% whose decisions are still pending (EDK (2010)), indicating that mainly rural cantons with fewer residents refused the concordat.

Figure 1.5: HarmoS participating cantons (EDK (2010))

The main goal of the HarmoS agreement is to harmonize the duration of compulsory education to two years of pre-school, six years of primary school and three years of lower secondary school, as well as to introduce national schooling standards as far as the taught content is concerned. The HarmoS agreement officially came into effect on August 1st, 2009 and some cantons have already introduced the new harmonized standards proposed by the project. The latest start date is set for the school year 2015/2016 (EDK), meaning that many schools in the participating cantons are currently

(34)

undergoing a major reorganization, which affects both teachers’ capacity, as well as the used teaching material.

This is currently an important factor affecting our target group, since the introduc- tion of new teaching aids and project-based evaluations are difficult to organize during this time of change. It was also one of the most important obstacles to finding teachers who are willing to participate in our experiments with their classes during the ongoing reorganization of the school system.

1.3 Objectives

The central objective of the thesis was to develop a CALL tool that is adapted to the requirements of lower secondary level language instruction and that puts a main focus on training communicative L2 competences. In a second step, the objective was to evaluate the previously customized tool. With the developed tool we aimed to inves- tigate how customized CALL was perceived by both students and teachers, as well as how its use influenced students’ L2 competences. Furthermore we aimed to analyze the influence that gamification had on students’ motivation. Our hypotheses and research questions are presented below, introducing the above-mentioned aspects in more detail.

The methodology of our experiments will be addressed in section 1.4.

Can CALL-SLT help students improve their communicative second language skills? The first research question that we will address in this thesis is whether the regular use of CALL-SLT helped students improve their L2 competences. To investigate this question we took into consideration both objective and subjective components.

On the one hand, we measured pre- and post-experimental performance in a number of written exercises, which served to measure students’ L2 competences in vocabulary, grammar and written conversational skills. On the other hand, we analyzed students’

spoken L2 production at the beginning and at the end of the experiment phase in order to observe how their oral competences, such as pronunciation and fluency, developed with the use of CALL-SLT. For the subjective evaluation of the first research question we asked students whether they (subjectively) felt that the use of CALL-SLT helped

(35)

them improve various L2 competences, such as vocabulary, grammar, pronunciation and fluency or overall L2 communicative skills. We hypothesized that regular use of CALL-SLT over a period of four weeks helps students improve their L2 communicative competences, most noticeably in terms of vocabulary, pronunciation and fluency.

Is CALL(-SLT) a justified supplement to traditional second language teach- ing methods that is well received by both teachers and students? The second research question that we investigated, is the teachers’ and students’ subjective percep- tion of the CALL approach to L2 teaching/learning in general, as well as the CALL-SLT tool in particular. We were interested in analyzing whether teachers found CALL-SLT to be a valuable supplement to traditional teaching and whether they felt that it helped students improve their communicative competences in the L2. We were furthermore interested in evaluating students’ appreciation of the tool, taking into account their overall motivation to engage in CALL, their appreciation of various components of CALL-SLT, such as the help function or the recognition feedback, as well as their subjective impression of whether the tool helped them improve their communicative L2 competences. Our hypotheses were that teachers appreciate a customized CALL tool (i.e. CALL-SLT) to vary their traditional teaching methodsand thatstudents like training their communicative L2 competences with the help of a computer program.

Does a student’s personal background have an influence on his or her motiva- tion? A third research question that we tackled was whether students’ personal back- ground – such as his or her first language (L1), previous experience with technology- enhanced language learning, or his or her overall interest in learning the L2 – had an observable influence on his or her motivation to use the CALL-SLT tool. Our hypoth- esis was that students who have a generally positive mindset to learning the L2, and who fit the pattern of the targeted audience, will be more motivated to engage in the CALL-SLT dialogue game.

Does gamification help motivate students? The fourth research question con- cerned the influence of gamification elements on students’ motivation. We evaluated whether students using a gamified version of CALL-SLT, containing scores and badges,

(36)

were more motivated to use the dialogue game than students using a non-gamified ver- sion. This question was analyzed by means of objective and subjective measures. The objective influence of gamification was measured in terms of number of total logins (re- turn rate), as well as total number of interactions and average number of interactions per login (perseverance). Concerning students’ subjective appreciation of the proposed serious game, we took into account qualitative feedback given at the end of the exper- iment phase, in which students indicated which version of CALL-SLT they preferred.

Our hypothesis was thatstudents using the gamified version of CALL-SLT will be more motivated to engage in the dialogue game, which will be visible in both objective and subjective measures.

1.4 Methodology

In order to address the objectives outlined in section 1.3, we developed a customized course for a CALL application, CALL-SLT, which allows German-speaking secondary school students in Switzerland to train their communicative competences in English.

The CALL-SLT course comprised eight lessons, which are based on a commonly used English manual in Swiss lower secondary schools and that imitate a virtual trip to Lon- don. The application’s goal is to engage students in a virtual dialogue with a cartoon character, in which students have to produce spoken output to respond to the questions and requests uttered by the cartoon dialogue partner. After every turn, students re- ceive color-coded feedback from the system concerning the correctness of their spoken input.

Once the lessons were created and customized to our target group in collabora- tion with a subject matter expert, the CALL-SLT course was introduced in 15 school classes across German-speaking Switzerland. The classes were asked to regularly use the CALL-SLT lessons in a home-study environment during a four-week experiment phase, in order to train their spoken productive competences in English.

To evaluate the impact that the regular use of the CALL-SLT course had on stu- dents’ L2 competences, a series of data was collected. On the one hand, students’

(37)

written L2 competences were measured by means of a written placement test, both pre- and post-experimentally. This test primarily served as an indicator of whether or not students’ written vocabulary and grammar competences improved with the use of CALL-SLT. On the other hand, students’ spoken responses were recorded and served as a basis for the evaluation of a possible improvement of their spoken L2 competences, such as pronunciation and fluency, but also the correct use of vocabulary and gram- mar. In addition, we collected qualitative data in the form of two questionnaires. A questionnaire administered pre-experimentally provided information on students’ per- sonal background, and a second questionnaire that was filled in post-experimentally provided students’ subjective perception and feedback on the CALL-SLT course and its usefulness in training various L2 competences.

1.5 Dissertation Structure

The body of the thesis is divided into three main parts. The first part – Chapters 2 and 3 – covers the theoretical background of SLA and CALL. The second part describes the development and customization of CALL-SLT (Chapters 4 and 5), and part three addresses the evaluation of the CALL-SLT tool in Chapters 6 and 7. In more detail, the content of the individual chapters is as follows:

Chapter 2 provides a background on SLA theories. Various approaches, as they have been applied over the past 60 years, are introduced and their influence on our research and on CALL-SLT is discussed. The chapter also introduces a number of common motivation theories in the educational context, as well as an overview of how motivation can be evaluated.

Chapter 3 addresses how technology can be introduced into second language learn- ing. In addition to a historical background, more recent research in CALL is presented, forming the basis for the principles used in CALL-SLT. A special focus is given to speech recognition technology, which is one of the main functionalities of CALL-SLT.

(38)

Chapter 4 describes the CALL-SLT system in detail. Next to the underlying pro- cessing, which mainly consists of a machine translation and speech recognition module, the different components are introduced, such as the interface, help function, prompt design, system feedback, and gamification elements. The chapter also provides an overview of previous evaluations that have been conducted with various versions of CALL-SLT.

Chapter 5 addresses the customization of the CALL-SLT application. The chap- ter first provides an overview of the target group’s expected language level and then discusses the pedagogical approach that was chosen for the integration of CALL-SLT in the Swiss English classroom, which is the blended learning approach. The second part of the chapter discusses how the CALL-SLT tool was customized in order to meet the requirements of the target group. The lesson development is discussed in detail, focusing on the applied dialogue structures, prompt design, system feedback and the developed gamification elements.

Chapter 6 discusses the experiment setup and evaluation methodology. The chapter provides an overview of how the experiments were organized, discussing the setup of the experiment phases, our information policy, and the data collection. It also gives information on the participants and how they were classified into four activity groups, depending on how frequently they interacted with CALL-SLT. In terms of evaluation methodology, the chapter introduces the control group as well as the single- case experiment designs and illustrates the respective advantages of both approaches for our research.

Chapter 7 presents and discusses the results of the evaluations conducted to answer our research questions. The students’ pre- and post-experimental performance in both written and spoken L2 competences are compared and analyzed in correlation to their activity level. Another aspect discussed in this chapter is the influence of students’

personal background on their performance in CALL-SLT and their motivation to use the tool. The correlation of students’ motivation with their perception of the exer- cises’ difficulty, their appreciation of the help function and their overall appreciation of the CALL-SLT tool are also addressed. Next to the students’ qualitative feedback

(39)

on CALL-SLT, the teachers’ opinions are discussed. Additionally, the influence of gamification elements on students’ motivation is evaluated on both an objective and subjective level.

Chapter 8 concludes, presents limitations of this research, and outlines future work.

1.6 Published Work

Parts of the work described in this thesis have been discussed in previous publications.

The CALL-SLT project was described in Rayner et al. (2010a), Rayner et al. (2010b), and Rayner et al. (2014b). The integration of CALL-SLT in lower secondary school, and the customized CALL-SLT course used in this study were discussed in Baur (2012), Baur et al. (2013) and Baur et al. (2014b). A publication on the pedagogical blended learning approach was presented in Bouillon et al. (2012), and the collected speech corpus and its annotation was discussed in Baur et al. (2014a). Finally, a first evaluation of students’ motivation to engage in the CALL-SLT serious game was presented in Baur et al. (2015)

(40)

State of the Art

(41)
(42)

Second Language Acquisition

“If you talk to a man in a language he understands, that goes to his head. If you talk to him in his own language, that goes to his heart”– Nelson Mandela

2.1 Introduction

Chapters 2 and 3 will lay the foundation for the research questions which are concerned with whether or not the use of Computer Assisted Language Learning (CALL) can help students improve their skills in a second language (L2) and whether CALL applications are an effective supplement to traditional L2 teaching methods. The chapter begins by discussing the various trends of SLA research and their development through history.

This overview provides a basis for Chapter 3 in which we will discuss how the vari- ous SLA theories have influenced the history and development of technology in SLA.

As we will see in later chapters, this background knowledge on both traditional SLA theories and the history of CALL was used to develop and customize our CALL-SLT tool (discussed in Chapters 4 and 5); a tool that was then used to address the research questions mentioned above (c.f. Chapters 6 and 7).

In SLA research second languages are defined as “any languages other than the learner’s native language or mother tongue” (Mitchell and Myles (1998), p. 5), and SLA itself is defined as “the study of how learners create a new language system with only limited exposure to a second language” (Gass and Selinker (2008), p. 1). SLA is a relatively young field of research and has roots in the theories of first (or child)

(43)

language acquisition. Although research in first and second language acquisition are closely related, there are some important differences that need to be addressed when studying L2 acquisition processes. In this chapter we will see that each school of thought attached different degrees of importance to the various differences such as the importance of native language structures in the L2 or the cognitive tools available at time of acquisition.

The first attempts of teaching an L2 date back to when the Romance language had diverged too far from standard Latin that schoolchildren had to actively learn Latin in order to read classic literature (Herschensohn and Young-Scholten (2013)). How- ever, SLA was not recognized as a scientific discipline until the mid-20th century. This chapter demonstrates that the roots of SLA are not originally in the linguistic field but mainly in the domains of pedagogy, sociology, and psychology, whose findings were then adapted to the field of language learning.

This chapter provides an overview of the various approaches to SLA during the past 60 years. The overview is in chronological order from early theories that started emerging in the 1950s – focusing mostly on first language acquisition – to more recent hypotheses and approaches to SLA as they still influence research today. The approach to SLA followed in this research, which is the functional approach, will be discussed in subsection 2.3.2. At the end of the chapter the importance of motivation in SLA will be introduced, as well as the concept of gamification, which is a relatively recent approach to enhance motivation in second language learners.

2.2 Historic Perspectives in SLA

2.2.1 Behaviorism

In the United States of America, the predominant learning theory of the 1950s was Skin- ner’s school of behaviorism, which was based on the notions of the stimulus-response theory and classical conditioning. The behaviorist approach to language acquisition is described in Herschensohn and Young-Scholten (2013), p. 705, as follows:

A school of psychology that flourished in the United States in the early twentieth century. Behaviorists (e.g. Skinner) stressed the study of observable behavior to

(44)

the exclusion of concepts of mind, emotion or consciousness. Extended to the study of language, behaviorists aimed to account for the acquisition of language as the outcome of conditioning shaped by external stimuli.

Behaviorism has its roots in the theory of stimulus and response, which is based on the three laws of effect, readiness, and exercise (Thorndike (1913)). The central idea of behaviorism is that human behavior (response) is reinforced in connection to a given stimulus if the outcome is positive or successful. One of the best known experiments in the field of stimulus-response was conducted by Pavlov in the early 20th century to demonstrate the connection between stimulus and response or the so-calledconditioned reflex (Pavlov (1927)). In Pavlov’s experiment a dog was presented food as a consecu- tive consequence to the ring of a bell. The procedure was repeated a number of times until the dog’s salivation to the ring of the bell (stimulus) in anticipation of the food (response) was produced even without any food being presented to the dog.

Decades later, B. F. Skinner introduced and elaborated the idea of conditioning in the field of language acquisition (e.g. Skinner (1957)). Skinner observed that children form habits of correct language use, by imitating the language as is used in their environment, and by adapting sounds and patterns (Lightbown and Spada (2006)).

Bloomfield’s work on the behaviorist view of language acquisition describes language as a succession of three steps: (1) a practical event, (2) a speech event, and (3) a reaction or response from the hearer. An example of these steps is given in Bloomfield (1933), p. 22-23:

Suppose that Jack and Jill are walking down a lane. Jill is hungry. She sees an apple in a tree. She makes a sound with her larynx, tongue and lips. Jack vaults the fence, climbs the tree, takes the apple, brings it to Jill and places it in her hand. Jill eats the apple.

The practical event is Jill being hungry (stimulus), which is followed by Jill’s speech act (“a sound with her larynx, tongue and lips”) (response / stimulus), and finally Jack’s response of fetching her an apple (response). On the one hand, Jill’s speech act is a response to the stimulus of being hungry, and on the other hand Jack’s reaction (which could also be of a linguistic nature, such as the response “Do you want me to fetch

(45)

you an apple?”) is a response to the stimulus of Jill’s speech act, demonstrating two stimulus-response pairs evoking a speech act.

As discussed in Gass and Selinker (2008), in the first language acquisition scenario, the stimulus is often an object presented to a child accompanied by the correct denom- ination, such as giving an infant a doll and saying “doll”. The child will soon start to imitate the sound of doll (in an imperfect way, such as “da”). If the stimulus is given repeatedly, the child will start to form a habit by uttering the response “da” and later

“doll”, and hence associating the stimulus-response set.

When Bloomfield’s theory of language learning was adapted to SLA in the 1960s, it resulted in putting an emphasis on mimicry and memorization exercises that imitated the first language acquisition process (Lightbown and Spada (2006)). During that same period, the behaviorist idea of transfer became important. The behaviorist approach suggests that learners try to transfer the linguistic structures that they already know from their mother tongue to the L2. An example given in Gass and Selinker (2008) suggests that an L1 Italian-speaker is expected to translate the Italian question“Mangia bene il bambino?” with “Eats well the baby?” (direct transfer from the L1 equivalent) rather than producing the correct English translation, “Does the baby eat well?” This direct transfer from L1 to L2 is parallel to behavior patterns that are shaped by previous knowledge and experiences; therefore, the known language structures or habits of the L1 are directly transferred to the L2 (Postman (1971)).

The idea of transfer can be categorized as positive transfer (or facilitation) and negative transfer (or interference). The above-cited example of the direct transfer of Italian to its English equivalent is an example of a negative transfer (interference), where the transfer of the Italian “Mangia bene il bambino?” to the English “Eats well the baby?” is obstructive. A positive transfer (facilitation) would be if an Italian- speaker were to learn Spanish and would similarly transfer the Italian “Mangia bene il bambino?” to the Spanish “¿Come bien el ni˜no?”, where the same word order is used in questions (Gass and Selinker (2008)). In SLA, the prevailing type of negative transfer is proactive inhibitionwhich occurs when a learner applies previously learned patterns of the L2 to a new situation in which they are not compatible. In a foreword to Lado (1957), p. vii, Fries states:

(46)

Learning a second language [...] constitutes a very different task from learning the first language. The basic problems arise not out of any essential difficulty in the features of the new language themselves but primarily out of the special “set”

created by the first language habits.

Fries’ statement, as well as the idea of negative transfer, indicate that the behaviorist school perceives the L1 as an obstacle to learning an L2 if the linguistic structures are not identical; suggesting that successful language acquisition of L2 requires new habits to be developed.

The two findings of the behaviorist approach to SLA that are most important for the study at hand are the positive effect of mimicry and the negative influence of direct (negative) transfer from the L1 to the L2. As we will see in Chapter 4, both find- ings were considered in the development and customization of our CALL-SLT dialogue game. With a spoken help option, we gave students the possibility to imitate native L2 speech and, by using an L1-independent form for prompting the students, we tried to avoid a possible negative effect that could occur if students translated word-for-word to the L2.

2.2.2 Contrastive Analysis Hypothesis

As a logical next step in the development of SLA theories, the Contrastive Analysis Hypothesis (CAH) established the rules for comparing a set of habits of the L1 to another set in the L2 (Gass and Selinker (2008)). Herschensohn and Young-Scholten (2013), p. 708-709, define CAH as follows:

The view that second language acquisition begins with a learner’s assumption that properties of the L1 hold for the L2. Where the L1 and the target language are similar, acquisition is predicted to be facilitated, and where they are different, acquisition is predicted to be more difficult.

The CAH approach emphasizes the learners’ L1, suggesting that learners presuppose that not only the structural properties, but also properties of the sound system, lexicon, and orthographic and cultural conventions hold for the L2. Fries and Lado argue that

(47)

if there are important differences between the learners’ L1 and L2, learners have to be taught to abandon the presupposition of L1/L2 identity (Herschensohn and Young- Scholten (2013)). Their approach to SLA focuses on the similarities and differences in the two languages to teach the L2 more effectively (Lado (1957), p. 59):

Those structures that are similar will be easy to learn because they will be trans- ferred and may function satisfactorily in the foreign language. Those structures that are different will be difficult because when transferred they will not function satisfactorily in the foreign language and will therefore have to be changed.

Since both Lado and Fries came from a background of structural linguistics, they de- scribed languages in a structural way when comparing them and finding the similarities and differences between them. They identified four steps to facilitate a comparison be- tween L1 and L2 (Lado (1957)): (1) comparison of the forms of pairs of words, (2) comparison of meaning, (3) comparison of distribution, and (4) comparison of conno- tations. Mitchell and Myles (1998) illustrate the challenge that lies in the difference of habits in the learner’s L1 and L2: According to the contrastive school, it will be much easier for an anglophone person to learn the German equivalent “Ich bin zw¨olf Jahre alt” to the English sentence “I am twelve years old” than the French equivalent “J’ai douze ans”, since the English grammatical sentence structure can be adopted one-to- one in German. In order to develop the same habits for French, it would be necessary to practice the structure many times by imitating and repeating it. As a result of this idea, CAH emphasizesdrilling exercisesto practice the differences between the L1 and L2. This concept of drilling exercises has also influenced some CALL-SLT courses (e.g. Jolidon (2013)) that put a special emphasis on training aspects of the L2 that are particularly difficult for a given L1 by means of drilling exercises.

However, the strong focus on the differences and hence on the difficulties of the L2 (e.g. Fries (1945)) also lead to criticism. An examination of the produced errors revealed that not all errors could be logically explained by the habits being transferred from the L1 to the L2. For example*He comed yesterdayinstead ofHe came yesterdayseems to represent a previously learned pattern of the L2 being incorrectly applied by the learner and not a habit of the L1 being projected (Gass and Selinker (2008)). Additionally, not all predicted errors actually occurred across various L1/L2 pairs. In the 1960s,

(48)

the generative school furthermore argued that language does not solely consist of a memorizable set of patterns, but more importantly of a creative component that is based on an innate language faculty (Herschensohn and Young-Scholten (2013)).

This criticism suggests more factors involved in SLA than a simple transfer from L1 to L2. A theory that reacted to this criticism is Chomsky’s Universal Grammar, which moved away from the idea of L1-L2 transfer and introduced the idea of an innate language faculty that naturally allows every human being to construct (L2) language.

The following subsection will discuss this theory in depth.

2.2.3 Universal Grammar

A new approach to language learning was introduced by Chomsky in the 1960s with his influential publications on generative linguistics focusing on the rule-governed and creative nature of human language (Mitchell and Myles (1998)). Herschensohn and Young-Scholten (2013), p. 734, define Chomsky’s Universal Grammar theory (UG) as follows:

Universal Grammar is a theory of the necessary formal constraints on all human grammars, imposed in advance of experience by the structure of an innate human language faculty. Universal Grammar provides learners of every language with an abstract basis for the construction of a natural language grammar. The observ- able language-particular properties shape the expression of UG within any specific language.

Or in Chomsky’s own words (Chomsky (1995), p. 167):

The theory of a particular language is itsgrammar. The theory of language and the expressions they generate isUniversal Grammar (UG); UG is a theory of the initial state Soof the relevant component of the language faculty.

The main idea of UG is that a child’s innate linguistic properties are applied if input alone fails to learn the complexities of “adult” grammar (Gass and Selinker (2008)).

This trend evolved as a direct reaction to Skinner’s behaviorist view of language ac- quisition. Chomsky’s criticism of Skinner’s approach was openly visible in his review (Chomsky (1959)) of Skinner’s publication Verbal Behavior (Skinner (1957)), where

(49)

one of Chomsky’s main points of criticism concerned the lack of consideration of the creativity of natural language. He argued that children learn underlying rules to a given language rather than strings of words as proposed by Skinner and can hence create their “own” language out of a constrained set of principles and parameters. As an indicator for such an underlying framework of linguistic rules he named children’s use of standard forms in cases of irregular grammar, as can be seen in sentences such as

“*I fed the fishes” (instead of “I fed the fish”) or “*the radio breaked” (instead of “the radio broke”), which cannot be explained with the behaviorist input theory (Mitchell and Myles (1998)).

The three questions setting the basis for the theory of Universal Grammar were addressed by Chomsky (1986) as follows.

1. What constitutes knowledge of language?

Chomsky suggests that every individual has the same universal set of principles and parameters that shape the human language and that make it possible to detect similarities between different languages. According to UG, the principles apply to all natural languages, whereas the parameters have some open values that can vary between languages.

2. How is knowledge of language acquired?

According to Chomsky, every child has an innate language faculty that helps them learn their native language without too much difficulty. This means that children construct a mental picture of the structure and shape of language (ab- stract representation), based on the fragmentary input that they get, and they, therefore, have clear expectations towards language and the acquisition process.

3. How is knowledge of language put to use?

Chomsky’s Universal Grammar approach provides an abstract representation of language that all humans possess innately. It is all aboutknowledgeorcompetence and not aboutperformance that only comes into play when underlying cognitive language representation is applied in real-life scenarios. Conversely, competence is only one aspect of performance, since sociolinguistic and psycholinguistic com- ponents also influence performance.

Références

Documents relatifs

The Faculty of Translation and Interpreting, University of Geneva, has developed a multilingual Open Source CALL platform, CALL-SLT, based on speech recognition and language

We have evaluated 130 post mortem examination protocols from Pathology centre of Riga Eastern Clinical university hospital from persons who died without medical aid. We have

In this paper we describe how we used this tool in the context of a real French foreign language (FFL) training for employees of French companies.. For exploratory purposes, 10

Understanding exactly why it works would require further investigation, but most of the explanation might be that (i) it takes some time before the synchronization is lost and (ii)

While  there  is  blame  to  be  shared  by  both  sides  of  this  debate—those deploying patient eligibility measures and  those  unpleasant  patients 

In short, better prior knowledge can be paired with data collected in courses to better identify how students are learning the course content and improve the course.. 5

Given that on the one hand, learners are usually taught highly ambiguous language because this is the most useful type of language for communicative purposes, and that on the other

By documenting the first voice experiences of parents and their perception of their children’s experiences, we gain a better under- standing of the challenges experienced by