The Syntactic Complexity of Learner Languages of Japanese from the Perspective of Dependency Distance

A Case Study Based on the International Corpus of Japanese as a Second Language (I-JAS)

Abstract

　In order to analyze the syntactic complexity (SC) of Japanese learnersʼ compositions, this study measured the SC of 150 compositions using both traditional SC measures and dependency distance. The results show: First, among the traditional SC measures adopted, mean length of sentence (MLS), mean length of T-unit (MLT), and dependent clauses per clause (DC/C), are the most effective ones that can accurately reflect the SC of Japanese learners with different proficiency levels. Second, dependency distance can further compensate for any insufficiencies of these three traditional SC measures; compared with MLS and MLT, dependency distance is less affected by lexical factors, while compared with DC/C, dependency distance can better measure the syntactic complexity of sentences without dependent clauses. In short, the use of traditional SC measures combined with dependency distance can represent the development of the SC in Japanese learnersʼ compositions more clearly and comprehensively.