Semantics circle: Cessation inferences in languages without present tense

Abstract

Use of the past tense on a stative in English triggers the so-called cessation inference (‘Al was sick’; CI: ‘Al is not sick anymore’). It has been argued that this inference is a quantity implicature in English, derived through pragmatic competition with the stronger present tense alternative (Altshuler & Schwarzschild 2013). However, cessation inferences associated with past tense are also found in optional past languages – languages that lack present tense. It has been claimed that cessation inferences triggered by past marking in such languages, unlike English cessation inferences, cannot be fully cancelled (Plungian & van der Auwera 2006, Cable 2017). This phenomenon is called ‘discontinuous past’ (DP). Existing semantic accounts of ‘discontinuous past’ argue that there are past markers with the meaning “past and not present” (Plungian and van der Auwera 2006, Leer 1991). Recently, they’ve faced criticism, and the pragmatic view has gained momentum (Cable 2017; Bochnak 2016; Bochnak & Martinovic 2019). Building on original fieldwork on Tundra Nenets, we provide arguments against the pragmatic view and propose a semantic account where DP-effects arise from applying Exh to past tense sentences, rather than from the meaning of past tense itself. We argue that our account solves a complication in the pragmatic view and explains the variation in the obligatoriness of DP-effects across optional past (OP) languages (through the obligatoriness parameter of Exh).

We also discuss the role cessation inferences in deriving simultaneous readings of past tense in embedded contexts. We argue that simultaneous readings of past-under-past in Russian and Tundra Nenets are derived via a de re construal complemented by a cessation inference.