How domain-sensitive are existential constructions? Evidence from Swedish and Dutch weather reports

International audience[Introduction] Over the last two decades, linguists have shown considerable interest in the advantages of CxG-theories for various linguistic sub-disciplines. A particular question that has arisen within more “applied” frameworks, is whether or not constructions can have domain-specific manifestations regarding meaning and/or form (Fischer/Nikiforidou 2015; Bücker/Günther/Imo 2015; Gautier/Bach 2019). However, sufficient empirical analyses are currently missing, and this research question needs to be more systematically tested for constructions in various languages and different LSP-oriented domains. Beside the mere “applied” focus, such work can lead to interesting outputs regarding the traditional oppositions between ‘arguments of a predicate’ and ‘modifier’, as well as between projectionist and constructional models (Müller 2006; Boas 2010, 2014; Herbst 2014; Engelberg 2019).[Research questions] We therefore propose to look at domain-specific constructions in weather reports in Swedish and Dutch. Said texts are a classic genre within text-linguistics, but more modern approaches regarding their linguistic organisation – i.e. Frame Semantics and CxG – are missing. Nevertheless, previous research (Krycki 2009) illustrated that these texts show many fixed domain-specific structures, which cannot be explained in terms of lexicon only and need to be treated at the level of the syntax-semantic-(pragmatic) interface. Consequently, we will look at existential constructions – best exemplified through the English there is/are-construction (Bentley et al. 2013) – which are generally considered as belonging to the “general” language. We will then proceed to illustrate that pressure from both the domain and specialized discourse genre (Gautier 2009) transforms them into “specialized constructions”.[Corpus] Our corpus consists of newspaper-weather reports from Expressen (Swedish) and Het Laatste Nieuws (Dutch). We will work with a comparable pilot corpus (Teubert 1996) from around 50.000 words in every language. The study will look at the following 3 constructions in Swedish (1) and Dutch (2):(1) a. det blir-construction“Framemot helgen blir det återigen runt 20 grader på många håll i Götaland […]”b. det finns-construction“Men det finns en del osäkerheter kring söndagen i söder, […]”c. det är-construction“Det ser ut som att det är mest uppehåll och sol, men att det finns lite risk för skurar.”(Expressen, 11-19/06/2020)(2) a. het wordt-construction“Het wordt vandaag wisselend bewolkt met in de loop van de dag enkele buien […]”b. er zijn/is-construction“Morgen is er eerst vooral richting Franse grens veel bewolking met nog wat regen […]”c. het is-construction“In het noorden en het oosten is het droog met brede opklaringen.”(HLN, 10/06/2020)[Discussion] The constructions in (2) are the Dutch equivalents of the ones in (1), for which a contrastive analysis will be very interesting. Our aim is to illustrate that the above constructions can indeed be regarded as domain-specific constructions and that both their form and meaning aredependent on the cognitive knowledge structure of the domain, i.e. weather. More specifically, our discussion will focus on (i) the domain-specific level of coercion, especially regarding the slots and the potential lexical filling, as well as (ii) the argument-like status of locative and temporal adverbials influencing said constructions