Unpacking the contribution of linguistic features to graph writing quality: An analytic scoring approach

Abstract

Previous research on L2 writing quality has mostly focused on independent writing, and there is little empirical work on integrated writing quality. Furthermore, the majority of studies on L2 writing quality have adopted holistic scoring for writing assessment. Therefore, the purpose of this exploratory research was to investigate the contribution of linguistic features, i.e. measures of syntactic complexity, lexical complexity, cohesion, and linguistic accuracy, to scores of content, organization, and language use in graph writing, which is far less researched. Two hundred and forty-three undergraduate Iranian students described a graph with few visual chunks and a familiar topic. The findings of stepwise regression revealed that content nominals per clause and lexical diversity contributed to content and language use scores, linguistic accuracy could predict the scores of organization and language use, and contrastive and temporal connectives and temporal cohesion were predictive of the scores in all the three subscales. Moreover, some important discrepancies were observed between the raters’ scores to a subscale and its linguistic predictors. Implications for writing research, pedagogy, and assessment are offered. • Linguistic features contribute to the subscales of graph writing in different ways. • Complex nominals per clause predicted content and language use scores. • Lexical diversity was the strongest predictor of content scores. • Cohesion contributed to all the three subscales. • Linguistic accuracy was the best predictor of language use scores.