Syntactic complexity measures and academic writing proficiency: A corpus-based study of professional and students’ prose

Abstract

This study examines large-grained, medium-grained, and fine-grained measures, as different dimensions of syntactic complexity, and their relationship with academic writing proficiency in the abstract sections of academic prose. We sampled 237 texts written by expert academic writers of research articles, PhD level L1 Persian writers, and MA level L1 Persian writers. Using one-way multiple analysis of variance (MANOVA) and ordinal logistic regression, we compared the abstracts in terms of their syntactic complexity and checked the predictability of large-grained, medium-grained, and fine-grained measures in academic writing proficiency. The findings revealed that the differences among the groups were not large enough for clausal and sentential measures of dependent clauses per T-unit, and clauses per sentence to show statistical significance. However, the differences among the groups on length-based measures of mean length of T-unit, and mean length of clause along with noun phrase measures of clauses per nominals, noun premodifiers, and prepositional phrases reached significance. Analysis of fine-grained measures of noun phrase complexity shows that four indices that measured the number and variation of dependents in the texts made unique statistically significant contribution to the prediction of academic writing proficiency. These findings suggest the importance of incorporating complementary measures of syntactic complexity (i.e., large-grained measures) as integral parts of L2 writing instruction practices. The implications of the findings in relation to L2 academic writing pedagogy are discussed.