Beyond Lexical and Syntactic Complexity : Contributions of N-grams and Cohesion to L2 Writing Proficiency

다국어 초록 (Multilingual Abstract)

Beyond Lexical and Syntactic Complexity: Contributions of N-grams and Cohesion to L2 Writing Proficiency Xuemei Liang Dept. of English Language and Literature The Graduate School Hanyang University The primary objective of this research is to examine the relationship between various linguistic features, such as lexical complexity, syntactic complexity, n-grams, and cohesion, and second language (L2) writing proficiency. An examination of a dataset comprising 1,200 English essays authored by Korean learners of English aims to investigate which measures of linguistic features explain L2 writing proficiency, with a specific focus on the distinctive impact of n-grams and cohesion. The current study finds that in the model of lexical complexity in predicting L2 writing proficiency, CCTR contributes 11.8% to predicting L2 writing proficiency. In the model of syntactic complexity, CN/C and CP/C contribute 2.5%. As for the model of n-grams, academic trigram proportion (100K), spoken trigram 2 ΔP, and spoken bigram MI account for 9.1% of the variance in the evaluation of L2 writing proficiency. In the model of cohesion, LSA given/new, causative connectives, temporal connectives, and logical connectives collectively explained 6%. The results suggest that L2 essays rated high quality tend to contain more lexical variations and complex nominals, more n-grams that are frequently used and have stronger association strength, more given information and temporal connectives, and fewer causative and logical connectives. Moreover, the result of hierarchical regression shows that after controlling for the influence of lexical and syntactic complexity, n-gram measures explain 4.6% of the variance in the model of L2 predicting writing proficiency, confirming the distinctive contribution of multi-word units to L2 writing quality. The drop in variation from 9.1% to 4.6% indicates an overlap between the measurement of n-grams and lexico-syntactic complexity. Meanwhile, LSA given/new contributes 7.7% to L2 writing proficiency when controlling for the impact of lexical complexity, syntactic complexity, and n-grams, suggesting that EFL learners rely heavily on semantic overlap to achieve cohesion and coherence in their essays. Besides, the sharp rise of LSA given/new from 2.4% to 7.7% indicates a competition between the measurements of cohesion and lexico-syntactic complexity and n-grams. The current study supports previous research regarding the predictive nature of linguistic features in relation to L2 writing proficiency. Additionally, the findings pertaining to the distinctive contribution of n-grams provide further evidence supporting that multi-word units are a distinct linguistic construct. Furthermore, the results concerning cohesion highlight the importance of the proportion of given information in the assessment of L2 writing proficiency. Moreover, the drop in n-gram variance and the rise in cohesion variance from stepwise regression analysis to hierarchical regression analysis reveals the potential relationship among measurements of linguistic features. Also, the findings of this study have implications for pedagogical practices and the development of automatic essay scoring (AES) systems. They provide valuable insights for the design and implementation of automated tools for evaluating written compositions.

목차 (Table of Contents)

• Table of Contents i
• List of Tables iv
• ABSTRACT v
• Chapter 1 Introduction 1
• 1.1 Background to the Study 1
• 1.2 Research Questions 5
• 1.3 Overview of the Dissertation 6
• Chapter 2 Literature Review 8
• 2.1 Linguistic Features and L2 Writing Proficiency 8
• 2.2 Lexico-Syntactic Complexity and L2 Writing 11
• 2.2.1 Lexical Complexity 11
• 2.2.2 Syntactic Complexity 17
• 2.3.3 Studies on Lexico-Syntactic Complexity and L2 Writing Proficiency 19
• 2.3 N-grams and L2 Writing Proficiency 21
• 2.3.1 N-grams 22
• 2.3.2 Studies on N-grams and L2 Writing Proficiency 26
• 2.4 Cohesion and L2 Writing Proficiency 27
• 2.4.1 Cohesion 27
• 2.4.2 Studies on Cohesion and L2 Writing Proficiency . 30
• 2.5 Summary of Literature Review 31
• Chapter 3 Method 34
• 3.1 Data Collection 34
• 3.2 Tools for the Analysis 37
• 3.3 Variables 39
• 3.3.1 Variables of Lexical Complexity 40
• ii
• 3.3.2 Variables of Syntactic Complexity 43
• 3.3.3 Variables of N-grams 45
• 3.3.4 Variables of Cohesion 47
• 3.4 Statistical Analysis 50
• Chapter 4 Results 52
• 4.1 Correlation Analysis of Linguistic Features 52
• 4.1.1 Results of Correlation Analysis on Lexical Complexity 52
• 4.1.2 Results of Correlation Analysis on Syntactic Complexity 53
• 4.1.3 Results of Correlation Analysis on N-grams 54
• 4.1.4 Results of Correlation Analysis on Cohesion 55
• 4.2 Results of Stepwise Regression 56
• 4.3 Results of Hierarchical Regression 58
• 4.4 Summary of the Results 60
• Chapter 5 Discussion 62
• 5.1 Measures that significantly Correlated with L2 Writing Proficiency 62
• 5.1.1 Lexical Indices That Correlated with L2 Writing Proficiency 62
• 5.1.2 Syntactic Indices That Correlated with L2 Writing Proficiency 65
• 5.1.3 N-grams Indices That Correlated with L2 Writing Proficiency 67
• 5.1.4 Cohesion Indices That Correlated with L2 Writing Proficiency 68
• 5.2 Influence of Linguistic Features on L2 Writing Proficiency 70
• 5.2.1 Influence of Lexical-Syntactic Complexity on L2 Writing Proficiency . 70
• 5.2.2 Influence of N-grams on L2 Writing Proficiency 73
• 5.2.3 Influence of Cohesion on L2 Writing Proficiency 79
• 5.3 Unique Contribution of N-grams to L2 Writing Proficiency 82
• 5.4 Unique Contribution of Cohesion to L2 Writing Proficiency 83
• Chapter 6 Conclusion 85
• 6.1 Findings of the Study 85
• 6.2 Implications 87
• iii
• 6.3 Limitations 89
• 6.4 Recommendations for Future Research 91
• References 93
• 국문요지 105
• Acknowledgments 106
• iv