Phraseological Complexity as an Index of L2 French Writing Proficiency

Vandeweerd, Nathan [UCL] Housen, Alex [Vrije Universiteit Brussel] Paquot, Magali [UCL] et al.

This study aims to expand the construct of L2 complexity beyond purely syntactic or lexical measures to include the ways in which words combine to form meaningful word combinations. We define phraseological complexity as the “range of phraseological units that surface in learner production and the degree of sophistication of such phraseological units” (Paquot, 2019, p. 124). For L2 English, previous research has shown that measures of phraseological complexity can predict proficiency level better than measures which target solely syntactic or lexical characteristics of a text (Paquot, 2019). Although research in L2 French has shown that learners use more phraseological units overall as they increase in proficiency (Forsberg & Bartning, 2010), thus far no study has compared phraseological complexity (i.e. diversity and sophistication of phraseological units) across proficiency levels in L2 French. The current study attempts to fill this gap by providing cross-linguistic validation of the results of Paquot (2019) for L2 written French. The data for this study come from the Leerdercorpus Frans, a 100,000 word corpus of argumentative essays written by L1 Dutch parser. Phraseological sophistication is operationalized as the mutual information score of those units. In addition to phraseological complexity, we also calculate several measures of syntactic and lexical complexity. Measures of syntactic complexity include measures of length (mean length of clause), subordination (clauses per T-unit, dependent clauses per T-unit, dependent clauses per clause) and part-of-speech based structures (verb phrases per T-unit, complex nominals per T-unit and complex nominals per clause). Measures of lexical complexity include measures of diversity (transformations of type-token-ratio) and sophistication (proportions of low-frequency words). Mixed effect regression analyses are used to determine which of the phraseological, syntactic and lexical complexity measures best account for human raters’ overall proficiency assessment (in terms of CEFR levels) of the same texts. In line with Paquot (2019), we expect to find that measures of phraseological complexity will better predict L2 proficiency level than syntactic and lexical complexity measures, especially for highly advanced learners. However, the burden of learning a relatively richer inflectional system may mean that our learners of L2 French do not exhibit the same degree of phraseological complexity as learners of L2 English (cf. Stengers, Boers, Housen, & Eyckmans, 2011).