理论与方法专栏 | Minimalist Parsing
编者按
《理论语言学五道口站》（2020年第27期，总第91期）理论与方法专栏为大家推荐麻省理工学院计算机语言学教授Robert C. Berwick 和加州大学洛杉矶分校语言学荣誉教授Edward P. Stabler 2019年在Oxford University Press出版的著作：Minimalist Parsing。
Abstract
This book is the first dedicated to linguistic parsing—the processing of natural language according to the rules of a formal grammar—in the minimalist framework. While the Minimalist Program has been at the forefront of generative grammar for several decades, it often remains inaccessible to computer scientists and others in adjacent fields. In particular, minimalism reveals a surprising paradox: human language is simpler than we thought, and yet it cannot be processed by the machinery used by computer scientists.
In this volume, experts in the field show how to resolve this apparent paradox, and how to turn Chomsky’s abstract theories into working computer programs that can process sentences or make predictions about the time course of brain activity when dealing with language. The book will appeal to graduate students and researchers in:
formal syntax,
computational linguistics,
psycholinguistics,
computer science.
著作简介
本书以“最简方案”为理论框架，通过对自然语言分析处理，深入探究句子结构。“最简方案”为生成语法的前沿理论，但是其某些规则在计算机领域并不适用，换言之，用计算机处理语言时仍存在着一些悖论。如“最简方案”主张人类语言结构比想象中简单，但计算机科学家却未能用机器将语言结构清晰正确地呈现出来。
本书对该现象进行了探究，将乔姆斯基的抽象理论转化为电脑程序、深入分析句子结构的同时也预测了大脑在处理语言活动时的发展进程。本书对形式句法、计算机语言学、心理语言学和计算机科学等学科的研究生及学者都具有重要参考价值。
Preface
（引自Minimalist Parsing）
The chapters in this volume grew out of a workshop on Minimalist Parsing held at MIT on October 10–11, 2015. So far as we know, they represent the first workshop devoted solely to this topic. They draw on ongoing work originating from Noam Chomsky in the early 1990s through the present day, leading to perhaps the dominant approach in contemporary generative grammar, the Minimalist Program (MP). This approach left open a central question: how can we build computer programs that map from external sentences back to internal representations that follow MP linguistic theories, comprising part of the broader program of viewing linguistic theory as fully integrated into human computational cognitive (and biological) science? The aim of this book is to answer this question.
A central tenet of the MP is that accounts of human language ought to be grounded on the simplest possible logically necessary set of assumptions. For example, hierarchical syntactic structure need only be binary-branching, rather than ternary-branching, since empirically, binary-branching seems to suffice; further, this property itself can be seen to follow from more basic principles, as a corollary of the basic combinatory operations of Minimalist Grammars. However, despite this apparent simplicity, building parsers for such systems turns out not be not so simple. In fact, it is one of the great surprises of the modern study of human language that, as the Introduction that follows explains, the standard computational tool kit for parsing the structure of, say, programming languages fails outright when applied to the “simpler” case of human languages.
Most of the chapters that follow describe the ways in which we must extend our computational tool kit to cover human language. The rest extend these insights to other staples of computational cognitive science, such as integrating such parsers into models of human sentence processing behavior, even down to the level of brain activity.
In this regard it is important to stress that the MP is indeed a program, a particular methodological approach to generative grammar and human language, not a theory. Moreover, it is a still-developing program. In practice this means that there can be several distinct strands or choices in how to realize the basic “simplicity” tenets of the MP. The reader will see this reflected in the distinct ways the authors have approached implementation—some formal, some not, but all within the general spirit of minimalism. We hope that the resulting diversity and the success in tackling the thorny computational issues inspire a new generation of researchers to continue to explore how linguistics and computation can be made to fit together.
Robert C. Berwick
Edward P. Stabler
October, 2018
作者导语
（引自Minimalist Parsing）
本书各章节选自2015年10月10日—10月11日于麻省理工学院举办的“最简方案句法分析”研讨会。
自90年代以来，句法研究又进入了一个新的发展阶段——“最简方案”。乔姆斯基对“最简方案”的“探索”和“改进”时时代表着生成语法的最新动向。但因其还处在发展和不断的完善中，故仍存在些许问题：如怎样通过电脑程序将句子的内部结构展示出来、以及怎样将语言理论和人类认知科学关联起来。
“最简方案”的核心思想是建立“完美的句法理论”，即人类语言应该建立在一套符合逻辑、且结构简单的基础上。如句子层级是二元分支，而非三元。从经验上来讲，二元分支更符合逻辑，因为基本组合规律都为两两合并。尽管该组合规律浅显易懂，但实际操作起来并不容易，因为即使看起来结构简单的句子，也不能保证其计算机语言完全符合正确的语法规则。
本书的大部分章节探究如何将 “计算机程序”应用于“语言”，其余章节也会涉及到计算机认知科学的问题，如怎样用计算机分析语言结构、或怎样用计算机研究人脑行为等等。
毋庸置疑，“最简方案”是生成语法领域研究人类语言的重要方法。正如乔姆斯基所说，它并非“理论”，因为其还处在不断的发展和完善中。实现“最简化”的方法和途径有很多，具体细节读者可以从书中获取答案。虽途径方法众多，但其都遵循了“最简”这一原则。我们希望此书可以有效地解决“语言学”和“计算机领域”中仍然存在的一些“棘手”问题，同时也期待两个学科碰撞出新的火花，有更多有趣的发现。
Robert C. Berwick
Edward P. Stabler
2018年10月
作者简介
- Robert C. Berwick -
Professor of Computational Linguistics,
Massachusetts Institute of Technology
麻省理工学院计算机语言学教授
Research field:
Computational Cognitive Science:
Modeling the acquisition of language by children， Modeling and Computational Complexity of Human Language， Biology and the Genomics of Language， Dynamical System Theories of Language Evolution and Change.
主要研究领域：
计算机认知科学，包括儿童语言习得建模、人类语言复杂性建模研究、生物语言学和语言基因学、语言进化及动态系统理论研究。
- Edward P. Stabler -
Professor Emeritus of Linguistics, University of California Los Angeles
加州大学洛杉矶分校语言学荣誉教授
Research Field：
Language learning and variation，Human language processing，Parsing and formal language theory, Philosophy of logic and language.
主要研究领域：
语言学习和变体研究、人类语言处理、句法分析和形式语言研究、语言和逻辑哲学研究。
往期推荐：
本文版权归“理论语言学五道口站”所有，转载请联系本平台。
编辑：田英慧 王秋梅
排版：王秋梅 安镜儒
审校：王丽媛 陈旭