This study investigated how lexical and phraseological sophistication together contribute to the quality of vocabulary use by L2 English writers in Asian regions. The result indicated that phraseological sophistication predicted the rating after controlling for lexical sophistication, confirming the importance of the construct of phraseological complexity in L2 writing assessment.
Recent proliferation of lexical and phraseological sophistication measures has enabled fine-grained analyses of lexical dimensions in L2 complexity research (e.g., Kyle, Crossley & Berger, 2017; Paquot, 2019). Research indicated that proficient L2 users will produce more sophisticated words and phrases, the latter of which are typically investigated in terms of frequency and association strengths of bigrams and trigrams (i.e., contiguous two and three word sequences; Bestgen & Granger, 2014; Kyle, Crossley & Berger, 2017). More recently, studies have used refined methods of extracting collocations by specifying syntactic relationships between the target words (e.g., verb-direct object; Paquot, 2019). However, more research is needed to determine how the phraseological dimension adds further contribution to the quality of L2 writing beyond complexity measured with single-word indices. The current study investigates how lexical and phraseological sophistication indices (e.g., verb-direct object, subject-verb) together predict vocabulary proficiency ratings. It used International Corpus Network of Asian Learner English edited essays 2.1 (Ishikawa, 2018), which includes 640 argumentative essays written by 320 learners across 10 different regions on two different prompts (Smoking and Part-time job). Essays were analytically rated based on organization, vocabulary, language use, and mechanics. Because two essay prompts were nested within individuals, a multilevel linear regression analysis on the vocabulary score using a range of lexical sophistication indices (e.g., concreteness, Age of Acquisition, Frequency; Kyle et al., 2017) and phraseological sophistication indices (e.g., n-gram and dependency-based collocation indices; Bestgen & Granger, 2014; Paquot, 2019). The results of regression analyses indicated that while single-word indices (e.g., Age of acquisition) contributed to a large portion of vocabulary ratings, some phraseological sophistication indices (e.g., v-dobj MI) also predicted the vocabulary score. This would indicate that raters are sensitive to lexical features beyond the single words, speaking to the importance of phraseological complexity in assessing L2 writing.
