LAUSR

Abstract This article discusses one of several reasons for the loss of augmentation in English absolute constructions (ACs): the emergence of the verbal gerund (VG) in -ing. The VG was from the start strongly associated with prepositional contexts (de Smet, 2008) and has been shown to have a history connected to the AC (Fanego, 2004). More specifically, the strong association between VGs and prepositions probably caused a reinterpretation of most prepositionally augmented ACs as VGs, resulting in a functional split (Fonteyn & van de Pol, 2016). Second, this paper illustrates how the what with-construction underwent a very similar development, albeit in the opposite direction. Indeed, it appears that the grammaticalization of with into an AC-marker (van de Pol & Hoffmann, 2016), and the resulting strong association between with and ACs, may have caused the formally similar what with-construction to be reanalyzed as an AC. As a result, this study serves as an example of how category borders are often gradient (see also Malchukov, 2004, 2006) and how such gradient boundaries may cause the (bidirectional) reanalysis of constructions close to the border, which in turn may impact the larger overarching constructions involved.