ABSTRACT
Chapter 15 illustrates how the use of complexity features can be explored in a corpus of L2 writing, with data from the Test of English as a Foreign Language (TOEFL). The chapter first examines the 23 linguistic features referenced in Biber et al. (2011), showing how almost all of the features of speech and writing vary in expected ways across the two modes and tasks (integrated, informational task and independent, argumentative task), underscoring that L2 writers, even at lower levels of proficiency, are sensitive to differences in register. However, the individual linguistic features do not show many differences across score level. The paper then shows how multidimensional analysis can be used to capture the constellations of linguistic features that show differences across score level. In particular, Dimension 1 of the Multi-Dimensional (MD) analysis parallels the findings from many previous studies of speech and writing, illustrating how the complexity features associated with academic writing work together to distinguish score level in both written and spoken tasks. In addition, the integrated tasks, which draw on explicit informational content, notably contain more of the complexity features associated with academic writing.
