Effectiveness of MALL and AI in foreign language education. Error Dynamics in Duolingo : First 30 Days
- 주제(키워드) AI in education , artificial intelligence in education , language learning , mobile language learning , MALL , Duolingo , second language acquisition , cognitiveload , CLT , error rate , error monitoring
- 발행기관 고려대학교 국제대학원
- 지도교수 Kwak Do Won
- 발행년도 2026
- 학위수여년월 2026. 2
- 학위명 석사
- 학과 및 전공 국제대학원 국제학과
- 세부전공 국제통상전공
- 원문페이지 89 p
- 실제URI http://www.dcollection.net/handler/korea/000000307984
- UCI I804:11009-000000307984
- DOI 10.23186/korea.000000307984.11009.0301502
- 본문언어 영어
초록/요약
This study investigates the effectiveness of Mobile-Assisted Language Learning (MALL) and AI-driven apps like Duolingo in second language acquisition, using error rate as a proxy for progress via Cognitive Load Theory. Analyzing the open-access Duolingo SLAM dataset, a two-level mixed-effects linear regression models how session type, exercise format, lagged errors, time spent, word/sentence length, and days predict current error rates in beginner L2 learners using the app for the first 30 days. Results suggest that mean error rate reduces overall but learners go through a U-shaped curve experiencing initial increase in errors, signaling initial adaptation, mid-phase overload, and consolidation amid attrition. However, not all learners progress in the same way; there are four distinct learning trajectories. Reverse-tap exercises reduce errors by 12.9% vs. baseline (p<0.001), outperforming reverse-translate (3.2%; p<0.001); errors rise about 3.1% per session (p<0.001), 2.5% per word character, and 1.3% per sentence word. Verbs (+1.5%) and long pronouns (+1.3%) amplify length penalties, while frequent parts of speech (e.g., auxiliaries, -7.3%) attenuate them. Findings validate CLT—optimizing intrinsic/extraneous load via format sequencing and complexity grading and inform AI-MALL design: prioritize short, high-support exercises early; target verb/pronoun drills.
more목차
ABSTRACT i
PREFACE iii
ACKNOWLEDMENTS iv
TABLE OF CONTENTS v
LIST OF TABLES viii
LIST OF FIGURES viii
NOMENCLATURE ix
CHAPTER 1. INTRODUCTION 1
CHAPTER 2. BACKGROUND 4
CHAPTER 3. LITERATURE REVIEW 6
3.1 AI Applications and Limitations 6
3.2 AI Application and Effectiveness in Education 10
3.3 AI Adoption in Language Education, MALL 16
3.4 Duolingo and AI 23
CHAPTER 4. THE PRESENT STUDY 27
4.1 Rationale for Choosing Duolingo 27
4.2 Research Question and Significance of the Study 28
4.3 Theoretical Framework 29
4.4 Methodology 36
4.5 Findings 40
CHAPTER 5. DISCUSSION 50
CHAPTER 6. LIMITATIONS AND DIRECTIONS FOR FUTURE RESEARCH 54
CHAPTER 7. CONCLUSION 57
REFERENCES 58
APPENDIX 70
A. Table 4 Parts of Speech Effects On Error Rate 70
B. Table 5 Robustness Check 73
C. Table 6 Robustness Check with Interactions 75
