Discourse markers (henceforth DMs, e.g. Schiffrin 1987) have been the focus of a strong – and still growing – trend in pragmatics investigating in particular their multifunctionality and context-sensitivity. These frequent expressions (such as you know, well or so) are characterized by their role as structuring devices, addressee-oriented cues for interpretation and flexible syntactic status (formally varied, optional). The paradox of DMs is that, while very frequent and essential for successful communication (e.g. Crystal 1988; House 2013), they often go unnoticed during an interaction or, on the other hand, can be perceived as superfluous and even detrimental if used in the wrong conditions: an example from outside academia is provided by an article from the LanguageLog website which reports on US Senator Caroline Kennedy who was receiving bad press during her campaign because of “some cringing verbal tics that showed her inexperience as a speaker”, pointing out that she produced more than 200 you knows and many ums in a 30-minute interview (http://languagelog.ldc.upenn.edu/nll/?p=964). Yet, authors in fluency research tend to agree on the positive effects of DMs, especially in second language acquisition where they are associated with naturalness, automaticity and efficient planning strategies (Hasselgren 2002; Götz 2013). DMs are indeed intrinsically linked to fluency: they constitute windows on the cognitive processes behind speech production and perception, with many of their functions being directly connected to (dis)fluency moves (e.g. reformulation, planning). DMs are also telling of the speakers’ sensitivity to interaction settings such as topic, participants, degree of formality, or in other words to context in general: it has been repeatedly shown (e.g. Verdonik et al. 2008) that DMs tend to vary in forms and functions across different registers such as casual conversations or academic seminars. Context can also be understood i) in a broader sense as the modality, language or variety used by the speakers (e.g. Dostie 2009 on regional variation of French DMs; Taboada & de los Ángeles Gómez-González 2012 on English-Spanish speech-writing comparison) and ii) in a more local sense as the specific linguistic co-text in which the DM occurs (e.g. Degand 2014 on the effect of syntactic position on DMs). Overall, this large body of research suggests that DMs are a complex, highly variable category which involves a good command of the impact of linguistic and extra-linguistic context in order to be used (and acquired) accurately. In this paper, we propose a fine-grained corpus-based analysis of the context-sensitivity of DMs in three languages, namely English, French and Spanish across two registers: casual conversations and broadcast interviews, taken from the Val.Es.Co 2.0 corpus for Spanish (Cabedo & Pons 2013) and the DisFrEn dataset for English and French (Crible forthc.). By integrating a large number of variables such as syntactic and discursive position, pragmatic function and co-occurring disfluencies, we intend to uncover recurring patterns of DM use which transcend crosslinguistic and register variation, thus recommending them as potentially universal discourse “constructions” (Fischer & Alm 2013). Our analysis will also identify context-specific uses which cannot be transferred from one language or register to another. We will briefly present our contrastive methodology for discourse annotation (focusing on functional taxonomies) before turning to major corpus findings and the resulting suggestions for language learning. This study stands in the line of recent works in contrastive register studies (Neumann 2014) and applied Construction Grammar (De Knop & Gilquin 2016), while filling a gap in crosslinguistic fluency research.
Crible, L., & et al. (2017). How to be (dis)fluent in English, French and Spanish: towards universal constructions of discourse markers and disfluencies across registers. International Pragmatics Association Conference, Belfast, Ireland. https://hdl.handle.net/2078.5/45785