Multiple sources of constructional contamination in different language varieties

Two formally similar constructions can affect each other’s realization through a process called constructional contamination. Pijpops & Van de Velde (2016) have shown that language users may store (frequent) instances of one construction as an exemplar chunk and later recycle them in a different but formally similar construction. Contamination effects have since been described for several other constructions, but AUTHOR (accepted) has shown that at least for one pair of Dutch constructions, the process of constructional contamination can have different outcomes in different varieties of the language, i.e. in Belgian vs. Netherlandic Dutch. In addition, it has been argued that the storing and recycling of chunks might not fully explain constructional contamination in all cases (Pijpops et al. 2018; AUTHOR accepted). In this paper, we further explore the mechanism(s) behind constructional contamination as well as the nature and extent of national variation in contamination outcomes, by focussing on the Dutch receptive construction. We show that different slots of this construction are contaminated by multiple constructions at the same time. However, these contamination effects are not identical across both varieties of Dutch. In Netherlandic Dutch, a higher degree of formal similarity is needed to trigger constructional contamination, which we hypothesize to be linked to the more exemplar-based nature of Netherlandic Dutch compared to the more rule-based nature of Belgian Dutch (Grondelaers et al. 2008). Our case studies also provide additional evidence for a direct role of horizontal links in constructional contamination.