Back to list

Detail of contribution

Auteur: Rebekah BAGLINI

The lexical semantics of derived states

Abstract/Résumé: It has been noted in the literature that stative passives don’t form a homogenous class (Nedjalkov and Comrie, 1988; Kratzer, 2000): two subclasses, target state passives (TSPs) and resultant state passives (RSPs), can be identified by their different behavior w.r.t the adverbial ‘still’ (1). TSPs (1a) describe states that are in principle reversible, thus satisfying the presuppositional requirement of ‘still’; RSPs (1b) convey that an event of the kind described by the participle is over by the reference time, and are therefore incompatible with ‘still’ (Parsons, 1990; Kratzer, 2000). (1) a. The door is (still) closed. b. The car is (#still) washed. To capture these two types of stative passives, Kratzer posits a semantic type difference between the verbal expressions in (1a)-(1b) and two distinct stativizing operators: one for TSPs, one for RSPs. However, this dual approach is not only theoretically undesirable, but empirically dubious, since TSPs and RSPs are not known to be morphologically distinguished in any language (Nedjalkov and Comrie, 1988). The present paper draws on recent work in degree semantics to derive the observed contrast between TSPs and RSPs in a unified way. Stativization targets verbs of scalar change, and the TSP/RSP contrast reflects the type of scale associated with the verbal expression: specifically, property scales yield TSPs, quantity scales yield RSPs. The semantic core of a scalar change verbal expression is a Measure of Change function (Kennedy and Levin, 2008) (2), where M may correspond to a lexically encoded property scale (e.g. closed) or a quantity scale associated with the mereological properties of an incremental theme (e.g. quant(the-car)). In the first case, (2) returns the amount of change in the property of closed-ness as a result of participating in the event. In the second, (2) returns the amount of change in the part structure of the car as a result of participating in the event. (2) [M] = λdλxλe.m↑m(x)(init(e))(x)fin(e)) = d I then posit a single stativizing operator which targets gradable event descriptions like (2) and existentially quantifies over the event argument. The acceptability pattern with ‘still’ falls out naturally from this approach. Like gradable adjectives with closed scales, without overt degree morphology the degree argument of a stative passive receives the maximum value by default (see Kennedy and Levin 2008 for details). Thus, in (1b) every subpart of the car is understood to have undergone a washing event; at no time in the future will this maximal value change, since no subpart of the car can un-participate in the event—hence, the infelicity with ‘still’. Since lexical properties associated with verbs like closed are typically reversible, TSPs are correctly predicated to be felicitous with ‘still’.