Synchronous Grammars and the Syntax-Semantics Interface

人类语言的话语有句法结构--短语是从子短语建立起来的--它们的意义也有结构--例如，用于论元的谓词。事实上，句法和语义结构的核心格是如此紧密地联系在一起，以至于它们之间存在着直接的对应关系。句法-语义接口，即定义句法结构和语义表征之间关系的形式体系，应该体现这种对应关系。本研究通过使用同步的形式概念来刻画句法和语义语法之间的关系，从而促进了对句法-语义界面的理解。语法同步方法形成了模型的基础，当适当校准时，该模型可以同时提供理论简单性、计算似是而非以及捕捉人类语言的句法-语义接口的错综复杂的能力。这项研究将允许对当前句法-语义界面的语言学理论进行正式比较，使语言学和计算机科学中分散的研究领域更加突出。通过构建和比较处理广泛语言结构的语法片段，研究人员将研究句法和语义语法的一系列形式结构选择及其同步所固有的计算效率和语言充分性之间的权衡。最终的结果将是对人类语言的结构和意义之间的关系的计算影响的更清楚的图景，以及一种能够实现有效计算的表达这种关系的方法。因此，它是朝着成功设计能够执行诸如翻译英语或其他语言的句子的任务的系统工程迈出的一步。

The utterances of human languages have syntactic structure -- phrases being built from subphrases -- and their meanings have structure as well -- predicates applying to arguments, for instance. In fact, core cases of syntactic and semantic structure are so closely related as to suggest a direct correspondence between them. The syntax-semantics interface, that is, the formal architecture defining the relation between syntactic structure and semantic representation, ought to manifest this correspondence. This research advances the understanding of the syntax-semantics interface by using the formal concept of synchronization to characterize the relationship between syntactic and semantic grammars. The grammar synchronization method forms the basis for a model that, when properly calibrated, may simultaneously offer theoretical simplicity, computational plausibility, and the ability to capture the intricacies of the syntax-semantics interface for human language. The research will permit formal comparison of current linguistic theories of the syntax-semantics interface, bringing a dispersed area of research in linguistics and computer science into sharper focus. By constructing and comparing grammar fragments that handle a wide range of linguistic constructions, the investigators will study the trade-off between computational efficiency and linguistic sufficiency inherent in a range of choices of formal structures for syntactic and semantic grammars and their synchronization. The end result will be a clearer picture of the computational implications of the relationship between the structure and the meaning of human languages and a method of expressing that relationship that permits efficient computation. As such, it serves as a step towards successful engineering of systems that can perform tasks such as interpreting sentences of English or other languages.