The lexis-grammar interface in learner language: From learner corpus data to applications

Paquot, Magali [UCL]

The field of learner corpus research (LCR) emerged at the turn of the 1990s when academics and publishing houses started to realize the considerable potential of large computerized datasets of learner production to describe learner language and/or develop new pedagogical tools and methods that target language learners’ specific needs. Since its beginning, LCR has shown strong interest in the lexis-grammar interface in learner production (e.g. Paquot & Granger, 2012 for a recent survey) and relied on corpus linguistic techniques to explore multi-word combinations of all kinds (collocations, phrasal verbs, constructions, etc.). The talk will start with a short introduction to LCR and a brief overview of some of its major developments before quickly zooming in on corpus based approaches to the lexis-grammar interface in foreign language proficiency and development. I will report on two recent studies in which I examined EFL learners’ lexis-grammar interface via a co-occurrence analysis that relies on association measures to identify statistically salient collocations from Stanford-typed dependencies (e.g. verb + direct object relations, modifier + noun relations) extracted from two learner corpora: (1) A sub-component of the Varieties of English for Specific Purposes dAtabase (VESPA) that consists in 98 CEFR-rated research papers written by French EFL learners in the context of BA and MA linguistics courses. (2) A sub-set of the Longitudinal Database of Learner English (LONGDALE, Meunier 2013) that includes 417 argumentative essays written by French EFL learners followed over a period of three years. I will show that, depending on the dependency relations investigated and the association measures used, the phraseological or lexico-grammatical indices can be used to discriminate between proficiency levels and trace language development. I will also argue that lexico-grammatical indices are potentially better discriminators of proficiency level for academic texts written by higher-intermediate to advanced L2 writers than traditional measures of syntactic complexity and lexical diversity. I will round off my talk with broader applications of learner corpus research into the lexis-grammar interface, more particularly in language testing and assessment.