
Full text loading...
Abstract
L2 syntactic complexity research has shown that employing multidimensional measures to assess writing proficiency levels may yield more accurate results compared to one-dimensional measures. However, the optimal number and combination of measures have not been thoroughly explored. This study addresses these gaps by employing a novel machine learning-based approach to quantitatively investigate the effectiveness of multidimensional measures and determine the optimal combination of measures. Through the analysis of a dataset comprising 36 L2 Chinese learners, we found that multidimensional syntactic complexity measures outperformed one-dimensional measures in accurately differentiating learners. Specifically, a combination of three measures (Mean Length of Sentence, Mean Length of T-unit, and Mean Length of Clause) achieved the highest accuracy in classifying the learners. The implications of this research include the development of computer-assisted placement tests and proficiency evaluations, as well as providing practical guidance for language teachers in designing complexity-based instruction.