The paradox of Discourse Markers: Evidence from production under cognitive load.

Degand, Liesbeth [UCL]

On the basis of my (mainly) functionally oriented corpus-based work on Discourse Markers (both onomasiological and semasiological) (see e.g. Bolly et al., 2017; Crible & Degand, 2019, in press; Degand, Cornillie & Pietrandrea, 2013; Degand & Fagard, 2011; Pander Maat & Degand, 2001), I started developing the idea of “the paradox of discourse markers”. The paradox consists in the observation that DMs are highly frequent in language in use, thus informing us of a crucial inherent aspect of human language. At the same time, DMs appear to be optional at the sentence level (syntax and semantics), thus challenging grammatical description. It follows that we are facing the paradox of an inescapable linguistic category without consensual linguistic description. To resolve this paradox, I believe we need more (fine-grained) knowledge about the ways DMs are used in different communicative contexts. Despite an impressive quantity of work starting in the early 1980s, the linguistic description of DMs remains scattered, most and foremost because the bulk of research on this linguistic class consists of a multitude of case studies of particular DMs in a diversity of languages (for a recent overview, see Blühdorn, Foolen, and Loureda 2017; Fischer 2014; Maschler and Schiffrin 2015). Much less of the work has been concerned with the categorical description of DMs as a linguistic class (notable exceptions are Crible, 2017; Fischer, 2014; Fraser, 1999). It is this latter line of research that I will follow here to illustrate how the study of DMs in spoken language, in our case French, can bring us closer to descriptively and explanatorily adequate language theory. The second goal of this talk is to show how such a systematic study of DMs can learn us more about the underlying cognitive and functional principles of human communication, as I assume that DMs are (also) indices of fundamental cognitive processing during (spoken) language production. Here too, we need insight in the ways speakers use DMs in different contexts, including contexts with high and low cognitive load.