Corpus linguistics to tackle multiword units: Necessary but not sufficient?

Gilquin, Gaëtanelle [UCL]

Corpus linguistics has made an essential contribution to the development of phraseology as a field of research (see, e.g., Granger & Meunier 2008) and has highlighted the large extent to which language relies on multiword units. Thanks to techniques such as collocational analysis, collostructional analysis or lexical bundle analysis, it has become possible to extract regular patterns that can tell us something about how people go about producing language. Much of the information that corpora provide about such patterns could not be gathered in any other way, especially when it comes to the study of learner language (cf. Ebeling & Hasselgård 2015), which will be the focus of this presentation. Yet, it will be argued that corpus linguistics is not sufficient to tackle the learning and teaching of multiword units: corpus methods should be combined with other methods such as experimentation, and the criteria used in corpus linguistics (like frequency and association) should be supplemented with other criteria such as pedagogical relevance. From a language learning perspective, experimental data make it possible to go beyond learners’ mere production to also approach competence (e.g. knowledge of collocations in Gilquin 2007) and processes involved in production. The latter will be illustrated by means of a study of lexical bundles in the Process Corpus of English in Education (PROCEED; Gilquin 2022). This corpus is made up of written texts as typically found in learner corpora, but in addition it includes keylogging and screencasting data, which show writing processes as they unfold and which can thus reveal how multiword units tend to be produced. Experimental methods such as dictation task, eye-tracking or self-paced reading have also been used to study more cognitive aspects of language learning, such as the mental processing of multiword units (cf. Schmitt et al. 2004, Siyanova-Chanturia et al. 2011, Kim & Kim 2012). As for language teaching, frequency of occurrence as attested in corpora undoubtedly represents valuable information to decide what multiword units to teach, but it is not sufficient (cf. Leech 2011). Pedagogical relevance, which covers aspects such as teaching worth (Ellis & Simpson-Vlach 2009) and teachability (Granger 2009), should be taken into account too. Experimentation also has a role to play, in particular to help determine how best to teach multiword units. An experiment can thus be set up to test the feasibility and/or efficiency of phraseology-focused instruction (see, e.g., De Knop & Mollica 2016, Gilquin 2021). The presentation will end with some concluding remarks about the advantages but also the challenges of combining information from various sources to tackle the learning and teaching of multiword units.