Phraseological Complexity and Academic Writing Proficiency in Abstracts Authored by Student and Expert Writers

The construct of linguistic complexity is widely used in academic writing as an index of writing development. However, linguistic complexity has been under-explored in its multidimensionality. In line with the aim of extending the scope of existing literature, the present study examined phraseological complexity, which is an important, under-investigated dimension of the construct. In order to do so, we drew on the related literature to operationalize phraseological complexity and used its proposed measures to compare research article (RA) abstracts authored by expert and student writers. Phraseological features of interest were extracted using an automated tagging program (Stanford Core NLP) and a programming script which was written in a Python environment. The extracted word pairs were subsequently analyzed for the strength of association using the index of mutual information (MI). The results indicated that as writers advance in their studies, their writing is characterized by a heavier reliance on the number of types and tokens of two-word combinations. It was also shown that phraseological complexity can be served as a reliable index of writing development and may be used as a complementary measure to other existing measures of linguistic complexity. The implications of the study are discussed in relation to language assessment and L2 writing pedagogy.