Effects of D-linking on the real-time processing of Italian wh-questions: Evidence from self-paced listening

In Italian, an asymmetry in wh-dependency resolution holds between questions with bare wh-items, like chi (who) (1) and those with complex, lexically-restricted wh-items, like quale N (which N) (2) with greater processing cost for the latter (De Vincenzi, 1991; Pagliarini et al., 2025): (1) Chi ha abbracciato il ragazzo? Who has hugged the boy (2) Quale studente ha abbracciato il ragazzo? Which student has hugged the boy Such cost has been attributed cross-linguistically (i) to richer referential assumptions driven by D-linking (the property of complex wh-items requiring that the entities they quantify over belong to a contextually salient set; Pesetsky, 1987), or (ii) to a more pronounced set-restrictiveness of complex wh-items (Donkers et al., 2013). Stronger reliance on context, together with featural similarity between which N and the postverbal NP (e.g., il ragazzo, (1)-(2)), would burden dependency resolution regardless of wh-function, even concealing subject-object asymmetries commonly observed in filler-gap dependencies (De Vincenzi, 1991). However, this explanation is challenged by processing studies showing that filler integration costs are not significantly affected by its semantic richness (Gordon et al. 2004), and by intervention-based grammatical accounts postulating that only features triggering syntactic movement may modulate dependency resolution costs (Rizzi, 1990; Starke, 2001; Grillo, 2008; Villata et al., 2016).