英语论文网

摘要：本文是一篇英语语法论文，本研究从中国英语学习者语料库（ICNALE）和英语本族语者语料库（ICNALE、LOCNESS）中提取含有目标情态动词的样例，共计4982条，并对其进行手工标注；开展多因素预测与偏差回归分析（MuPDAR），探讨影响中国英语学习者在使用情态动词can、may和could时偏离本族语者的主要因素，并对比三个情态动词在各显著因素下的差异程度。

lity. Perkin argues that every modal verb has a core meaning, and different modal meanings are derived from specific contextual conditions.

Several studies (e.g., Collins, 2007; Palmer, 2001) have investigated modal verbs from the perspective of epistemic, deontic and dynamic modality and found that it has a significant impact on the choice of modal verbs. This is generally manifested in the fact that dynamic modality is predominately expressed by can and could, while epistemic modality is dominated by modal verbs like may. Zhang (2007a) discussed dynamic modality from the perspective of subjectivity. She provided a brief overview of the historical study of dynamic modality in linguistics and interpreted its subjectivity from various grammatical perspectives, including tradition, function and cognition. Additionally, in her study of the subjectivity of deontic modality, Zhang (2007b) emphasized the importance of examining modality and modal verbs from multiple grammatical perspectives to gain a more comprehensive understanding of modality.

Chapter 3 Theoretical Framework ........................ 16

3.1 Usage-based Theory ........................... 16

3.2 Entrenchment-and-Conventionalization Model ............................... 18

3.3 Adapted Usage-based EC-Model in this Study ............................... 20

Chapter 4 Research Design ........................ 22

4.1 Research Questions ............................ 22

4.2 Corpora and Data Retrieval ............................. 22

4.3 Variables and Data Annotation .......................... 23

Chapter 5 Results and Discussion ......................... 29

5.1 Factors Contributing to the Deviation .............................. 29

5.2 Deviation under Each Significant Factor ............................ 32

5.3 General Discussion ................................. 46

Chapter 5 Results and Discussion

5.1 Factors Contributing to the Deviation

MuPDAR was carried out to explore the factors that condition the deviation scores. The multinomial logistic regression was used in R1 because logistic regression is applicable to the case where the dependent variable is a categorical variable, and is often used to analyze the choice of near-synonymous constructions in the field of linguistics. When the dependent variable has more than two levels (e.g., the three modals can, may and could in the present study), multinominal logistic regression is needed (Levshina, 2015). The logistic regression model established in this study is to compare the reference level of the dependent variable with other levels one by one. Specifically, may was taken as the reference level, and the coefficients of can and could were compared to may respectively. The multinom() function in the {nnet} R package was used to build the regression. After constructing the regression model with the maximum number of predictors, an independent variable was dropped each time, and the anova() function was used to compare whether there was a significant difference when decreasing a predictor in the model. Then, the best model was constructed as follows:

model.best = multinom(Form~Neg + SentType + Cl本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。