Exploring complexity at the lexis-grammar interface: Diversity and sophistication of verb-argument structures in L2 Dutch writing

Phraseological complexity has received increasing attention within L2 complexity research, with studies calculating a range of quantitative measures targeting various types of phraseological units. One of the main aims of this research has been to establish an empirical link between phraseological complexity and L2 proficiency. The present study expands upon existing research by investigating phraseological complexity in a corpus of L2 Dutch writing collected from the B1,
B2, and C1 levels of the Certificaat Nederlands als Vreemde Taal (CNaVT) exams. To this end, measures of verb-argument structure (VAS) diversity and sophistication, respectively operationalized as the type-token ratio (TTR) and mean mutual information (MI) score of these units, are calculated, alongside similar sophistication measures for verb + direct object dependencies. The
complexity measures are entered as predictors in multiple regression models predicting the assessed proficiency of the learners, as indicated by their numeric overall exam scores. Results indicate that the measure of VAS sophistication, in particular, demonstrates strong potential as a quantitative measure to include alongside measures of dependency-based phraseological units.
Not only does this measure exhibit a clear relationship with learner proficiency at the advanced stages, it also appears to supplement the predictive ability of existing phraseological complexity measures.