Abstract
To improve the scientific rigor and relevance of the assessment of second language oral development, this study is guided by alignment with the psychological planning process of second language learners, the representation of the production characteristics of spoken Chinese as a second language, and having a large effect size in meta-analysis. Preferred indices are selected, and a longitudinal tracking study is conducted using a spoken corpus of six native English-speaking learners of Chinese as a second language, over the course of one academic year. It systematically investigates the sensitivity and stability of selected indices throughout the dynamic development process, aiming to identify effective indices for assessing the development of syntactic complexity of spoken Chinese as a second language. The findings indicate that the mean length of minimal terminable unit (T-unit), which demonstrates both sensitivity and stability, serves as a universal and reliable measure that can reflect the development trend of language. Additionally, the phrase type per T-unit effectively reflects the developmental trajectory of combinatorial competence and is an effective measure of syntactic complexity in Chinese as a second language. In contrast, indices such as the relative clause ratio, although highly sensitive, exhibit low stability and are therefore better suited to fine-grained analyses within specific developmental stages or task conditions. Furthermore, the study reveals that the validity of indices is influenced not only by their intrinsic properties but also by the learners’ proficiency levels, target language features, and assessment purposes. Consequently, in various application contexts, such as research, instruction, evaluation, and automated analysis, indices should be rigorously selected and interpreted, taking into account specific objectives and learners’ characteristics, to enhance the evaluation validity, explanatory power, and practical applicability of syntactic complexity measures.
Keywords
Get full access to this article
View all access options for this article.
