The Implicit Content of Sluicing

在规划语言表达时，说话者和作者经常能够省略信息上多余的语法材料，例如在“Fred called, but Sheila didn't”中省略动词“call”。这一过程被称为省略，在世界语言中普遍存在，在非正式语言和对话中尤为常见。在许多类型的省略中有一种是sluicing，在这种情况下，省略的不是一个动词，而是整个句子。例如，说话者可能会在“他打电话，但我不知道为什么”这样的句子中，在“为什么”之后选择省略“他打电话”。省略提出了具有挑战性的科学和工程问题。过去50或60年的研究表明，允许省略的原则涉及许多不同类型的信息（语法结构、话语语境的动态和现实世界的知识），但这些原则的确切特征及其相互作用仍然是一个悬而未决的问题。由于缺乏一种关键的资源类型，这方面的进展受到了阻碍：即能够有效使用省略号的数据库，这些数据库的规模足够大，可以用来验证理论，也足够丰富，可以构成机器学习的基础。该项目的第一个目标是建立这样一个数据库，并使其免费提供给语言科学家和工程师。这个资源将是一个大型的，精心策划的自然发生的省略号语料库，在一定程度上进行注释，这将允许对一系列分析问题进行定量探索。随着整理和注释的进行，从数据中产生的模式将用于研究使省略成为可能的语法和上下文之间的相互作用。由于省略是人类语言的普遍特征，更好地理解它是如何工作的，就是更好地理解人类语言行为本身的本质。这个项目还应该在一个迫切需要的领域激发技术创新：设计更复杂的系统，与充满各种省略号的自然人类对话相连接。

In planning their linguistic expressions, speakers and writers are often able to leave out informationally redundant grammatical material, such as when the verb "call" is omitted in "Fred called, but Sheila didn't". This process, known as ellipsis, is widespread across the languages of the world, and is particularly common in informal language and dialogue. Among the many varieties of ellipsis is sluicing, where what is omitted is not a verb but an entire sentence. For example, a speaker may choose to leave out the understood sentence "he called" after "why" in a sentence like: "He called, but I don't know why".Ellipsis poses challenging scientific and engineering problems. Research over the past 50 or 60 years has demonstrated that the principles permitting ellipsis involve many different kinds of information (grammatical structure, the dynamics of the discourse context, and real-world knowledge), but the precise character of these principles and their interaction is still an open question. Progress has been hampered by the lack of a crucial resource type: databases of felicitous uses of ellipsis that are large enough to validate theories against, and rich enough to form the basis for machine learning.The first goal of this project is to build such a database for sluicing and to make it freely available to language scientists and engineers. This resource will be a large, curated corpus of naturally occurring ellipses, annotated at a level of sophistication that will allow a range of analytic questions to be probed quantitatively. As the curation and annotation proceed, patterns that emerge from the data will be used to investigate the interplay between grammar and context that makes ellipsis possible. Since ellipsis is a pervasive feature of human language, to better understand how it works is to better understand the nature of human linguistic behavior itself. This project should also stimulate technological innovation in an area of urgent need: in designing more sophisticated systems for interfacing with natural human conversation, which is replete with ellipses of every kind.