The Semantics of Gradable Modal Expressions

我们使用语言不仅是为了交流我们周围的事物和实际发生的事情，而且是为了考虑各种可能性，描述可能发生的事情，叙述什么是可取的，什么是必要的。诸如“permissible”、“probability”、“may”、“able”和“likely”等情态表达允许我们在语言上进入非现实的领域。在过去的四十年里，语言学家和哲学家在理解情态表达的意义方面取得了很大的进步。然而，人们对这一意义的一个方面理解不深，那就是情态意义的分级性：可能性可以根据其可能性、可取性或允许性来分级，而可能性、可取性或允许性的程度往往可以被衡量。我们不仅可以说“可能会下雪”，还可以说“很有可能会下雪”、“有60%的可能性会下雪”或“下雪的可能性大于下雨的可能性”。“这个项目的目的是提供一个完整的情态理论，以系统的方式解释可分级的情态表达的意义，理解基本情态表达的意义以及这种意义如何被语境调节。研究者将利用形式化模型和现代语料库的方法来研究语境和内容之间复杂的相互作用对分级情态意义的解释。通过坚定地接地我们的语言的分级模态的理解，并将这种理解融入当代工作的不确定性推理，这项研究有望有重要的应用，部分信息下的自动推理和风险管理。

We use language not only to communicate about what immediately surrounds us and what actually takes place, but also to entertain possibilities and to describe what may happen or what is likely to occur, to recount what is desirable and what is necessary. Modal expressions such as 'permissible', 'probability', 'may', 'able', and 'likely' allow us to linguistically enter the realm of the non-actual. In the past forty years linguists and philosophers have made great strides in understanding the meaning of modal expressions. One aspect of that meaning that is poorly understood, however, is the gradability of modal meaning: Possibilities can be ranked in terms of their likelihood, their desirability, or their permissibility, and the degrees of likelihood, desirability, or permissibility can often be measured. We can say not only "It may snow," but also "There is a high probability that it'll snow," "There is a 60% probability that it'll snow," or "It is more likely to snow than to rain." This project aims to offer an integrated theory of modality that accounts in a systematic way for the meaning of gradable modal expressions, understanding both the meaning of basic modal expressions as well as how that meaning is modulated by context. The researchers will make use of formal models as well as modern corpus-based methods to investigate the complex interactions between context and content in the interpretation of gradable modal meaning. By firmly grounding our understanding of the language of gradable modality and by integrating this understanding into contemporary work on uncertain inference, this research promises to have important applications to automatic reasoning under partial information and to risk management.