Abstract:

Predicates with meanings like ‘avoid’ or ‘refuse’ are said to denote implicitly negative meaning because they carry certain semantic properties that are typical in a negative context such as licensing negative polarity items (NPIs). In this thesis, I examine the syntactic and semantic properties of four major types of verbs of implicit negation (VINs) in Mandarin, namely, the REFUSE-type, REGRET-type, STOP-type, and REMOVE-type predicates, each taking a different type of phrase as complement and representing a specific type of implicit negation. I select several kinds of nominal, verbal, and adverbial NPIs as well as the polarity-sensitive ‘X, y ě Y’ ALSO construction to test the NPI-licensing ability of each type of predicates.

Based on the results of the NPI-licensing tests, I classify these predicates into super-strong, strong, weak, and super-weak VINs depending on the types and numbers of NPIs that one predicate can license. This reveals a fine-grained spectrum of verbs of implicit negation in terms of the ‘strength’ of effects of negation: at one end stand the REFUSE-type predicates such as jùjué (’refuse’) and lănde (’don’t feel like...’), which have the strongest ability to license various kinds of NPIs, and at the other end the REGRET-type predicates such as hòuhuǐ (’regret’), which do not license any NPIs. In between, STOP-type predicates like tíngzhǐ (’stop’) and kāishĭ (’start’) can license particular kinds of NPIs depending on their aspectual properties; REMOVE-type predicates such as shānchú (’delete’) can license NPI rènhé (’any’) and NPI yígài (’all’) since they entail the non-existence of their objects.

I further argue that the conceptual structure of these verbs of implicit negation relates to canonical negation in many different ways, which include entailing or presupposing a negative proposition, activating a negative proposition through counterfactual thinking, or even more broadly, expressing the confrontation between two volitional forces or countering a certain pre-assumption. This study also suggests that a strictly-defined logical relation such as downward entailment many not be sufficient to explain all the licensing conditions of various kinds of NPIs. Rather, many NPIs, especially verbal and adverbial NPIs, are licensed by a conceptual structure of negativity that is embedded in the lexical meaning of a predicate or in a certain context.