The focus of this paper is a PCC pattern from the Northwest Caucasian language Adyghe. In this language, a cislocative marker emerges as a repair within applicative intransitives if the object outranks the subject on the person scale. In ditransitives, however, the cislocative appears if the indirect object outranks the direct object on the person scale. This pattern is unusual in two respects: First, prominent argument combinations seem to require a repair in ditransitives, in contrast to the common assumption that these scenarios are morphologically less marked than their non-prominent counterparts. Second, the subject argument of ditransitives seems to be invisible wrt. to the emergence of the cislocative. In this paper, we explore the compatibility of existing PCC approaches with the patterns
found in Adyghe. We show that functional (Aissen 1999; Haspelmath 2004, 2020) and case-based approaches (Béjar and Rezac 2003; Anagnostopoulou 2003, 2005; Adger and Harbour 2007) fail to predict the reverse PCC patterns, while other accounts need additional assumptions to derive the distribution of the repair (Nevins 2007, 2011; Coon and Keine 2020). Moreover, we claim that the invisibility of the subject in ditransitives arises from the fact that the argument is indexed by phi-agreement rather than clitic doubling.