Discontinuous past: a semantic account

Abstract

In some languages, past tense markers give rise to nearly uncancellable cessation inferences - a phenomenon known as ‘discontinuous past’ (DP). In their original discussion of this phenomenon, (Plungian and van der Auwera, 2006) proposed a semantic account of DP-effects, arguing that certain past markers encode the meaning “past and not present”. They further suggested that in languages with optional past tense, the (‘idle’) past marker exhibits this property. More recently, the semantic approach has faced criticism. (Cable, 2017) showed that Tlingit, an optional past tense language exhibiting DP-effects, allows cessation inferences to be cancelled in certain cases (through a statement of ignorance). This finding contradicts the predictions of the semantic account. As a result, an alternative pragmatic explanation has gained traction (Cable, 2017; Bochnak, 2016; Bochnak and Martinović, 2019). Building on original fieldwork on Tundra Nenets, we present arguments against the pragmatic approach and propose a novel semantic account in which DP-effects arise not from the meaning of Past itself, but from the application of Exh to past tense sentences. We argue that our account not only captures the contrasts observed by (Cable, 2017) in Tlingit but also explains the cross-linguistic variation in DP-effects across optional past languages, which we attribute to differences in the obligatoriness of Exh.