Explaining L2 writing performance through a chain of predictors: A SEM approach

Abstract

Renewed calls for identifying other variables and validating previous models that account for the complex interplay of mechanisms that are activated in the writing process have been made. In response to this call, the study investigates whether text production processes (idea generation, idea encoding, idea/text revision, and idea transcription) mediate the effect of topic knowledge, linguistic knowledge, writing approach, and writing experience on writing performance. Writer's performance was measured by timed essays, while the psycholinguistic and linguistic variables were measured by scales and tests. Results yielded by Structural Equations Modeling indicate that topic knowledge and linguistic knowledge have direct effects on writing performance and indirect effects on writing performance through text production process. Writing approach and writing experience do not directly affect writing performance, but they indirectly affect writing performance through text production processes.