句法复杂度视角：雅思写作中的英语语言能力对比分析

Comparative Analysis of English Language Ability in IELTS Writing—From Perspective of Syntactic Complexity

摘要: 在句法复杂度视阈下，本量化研究旨在借助L2SCA分析器中的14项指标来系统呈现、深度对比雅思写作(学术类：任务1、2)内初级水平学生(4.5~5.5分)和中级水平学生(5.5~6.5分)在“不同命题”及“相同命题”条件下的句法复杂度特征，并进一步计算分析结果中次级分类间的皮尔逊相关系数，以对比、讨论不同层级学生的英语写作语言能力指标。研究发现主要分为三方面：1) 就对比分析中的相似性而言，首先，无论命题情况及任务异同，5大分类中，“输出长度”内三项指标均为中级水平值不同程度上高于初级水平值，尤其是平均单句长度(MLS)和平均T单位长度(MLT)，这证实了输出长度是区分不同语言水平学生写作句法复杂度的直观指标这一传统教学猜想。其次，分析显示众多相似性也体现在“句子复杂度”、“从属复杂度”、及“特定短语结构”三个维度，基本都和T单位、小句、复杂名词性短语、及从属子句的比率有关，所以我们有理由推测这四种句子结构单位也许是区分初级及中级水平学生雅思写作语言能力的重要标志。再次，无论何种任务及命题条件，4项比率(“从属复杂度”中的CT/T、DC/C，“并列复杂度”中的CP/C和CP/T)均小于1，在一定程度上说明此类使用是相对较少的，尤其“并列复杂度”中两项与并列短语有关的指标(CP/C, CP/T)。2) 就对比分析的差异性而言，首先,“相同命题”条件下，“并列复杂度”的两项指标“并列短语/小句比率”(CP/C)和“并列短语/T单位比率”(CP/T)在两种写作任务中呈现出相反的结果(任务2 初级值 < 中级值；任务1 初级值 > 中级值)，因此可以推断写作任务要求与设计要素会引发学生在并列复杂度上不同的表现。3) 就皮尔逊相关系数分析而言，具有统计学意义的关联主要体现在两个配对样本：“输出长度”与“并列复杂度”及“从属复杂度”与“特定短语结构”均呈现高度负相关(p < 0.05)。相关教学启示也在文中得到了讨论。

Abstract: Bringing syntactic complexity as a lens to compare different L2 writing language proficiency, the quantitative study has showcased and compared the fourteen indices in syntactic complexity generated by L2SCA in IELTS writing context (academic: including Task 1 & 2) where fundamental-level (4.5~5.5) and intermediate-level (5.5~6.5) students were involved with same writing prompt and different prompt respectively. The Person correlations among sub-categories were also calculated. The findings indicate: 1) regarding similarities, first, intermediate-level students may generate longer output in three indices of Output Length (MLS, MLT, MLC) regardless of different writing prompts and tasks, validating the conventional pedagogical presumption of valuing length of output in differentiating different proficiency levels. Second, the similarity also emerged in other three categories, Subordination Complexity, Sentence Complexity, as well as Certain Phrases and Syntactic Structures, primarily correlated to T-units, clauses, complex nominal phrases, and dependent clauses, which may be considered as key indices to differentiate IETLS writing proficiency levels in the study. Third, the figures of four indices (CT/T, DC/C, CP/C, CP/T) were all lower than 1.0, implying their deficiency in corresponding writing praxis, particularly the two in Coordination Complexity (CP/C and CP/T); 2) regarding discrepancy, the comparisons on CP/C and CP/T in Coordination Complexity under same writing prompts have elicited contradictory results (Task 2: intermediated level > fundamental-level; Task 1: intermediate-level < fundamental level). Accordingly the study posited task demands and key elements in task design may elicit different task performance in such domain; 3) regarding Pearson correlations, two pairs were found statistically significant (p < 0.05) and negatively correlated (Output Length and Coordination Complexity, Subordination Complexity and Certain Phrases and Syntactic Structures. Relevant pedagogical implications were also discussed.