牛津计算语言学手册

        本书是手册性的专著。本书的38章是分别由各个领域内的50位知名专家执笔的；本书几乎涵盖了计算语言学的所有领域，反映了当前计算语言学的最新成就，能使我们对计算语言学获得全面而系统的认识；本书各章写作风格一致，内容协调，浑然一体，特别适合对计算语言学感兴趣和刚入门的读者阅读。本书文笔流畅，使用有趣的实例来介绍艰深的技术问题，尤其适合文科背景的读者阅读。

Preface
Abbreviations
Introduction
PART Ⅰ FUNDAMENTALS
  1.Phonology
  2.Morphology
  3.Lexicography
  4，Syntax
  5.Semantics
  6.Discourse
  7.Pragmatics and Dialogue
  8.Formal Grammars and Languages
  9.Complexity
PART Ⅱ PROCESSES，METHODS，AND RESOURCES
  1O.Text Segmentation
  11.Part-of-Speech Tagging
  12.Parsing
  13.Word-Sense Disambiguation
  14.Anaphora Resolution
  15.Natural Language Generation
  16.Speech Recognition
  17.Text-to-Speech Synthesis
  18.Finite-State Technology
  19.Statistical Methods
  20.Machine Learning
  21.Lexical Knowledge Acquisition
  22.Evaluation
  23.Sublanguages and Controlled Languages
  24.Corpus Linguistics
  25.Ontologies
  26.Tree-Adjoining Grammars
PART Ⅲ APPLICATIONS
  27.Machine Translation: General Overview，JOHN HUTCHINS
  28.Machine Translation: Latest Developments，HAROLD SOMERS
  29.Information Retrieval EVELYNE TZOUKERMANN，JUDITH L. KLAVANS，AND TOMEK STRZALKOWSKI
  30.Information Extraction RALPH GRISHMAN
  31.Question Answering，SANDA HARABAGIU AND DAN MOLDOVAN
  32.Text Summarization EDUARD HovY
  33.Term Extraction and Automatic Indexing，CHRISTIAN JACQUEMIN AND DIDIER BOURIGAULT
  34.Text Data Mining MARTI A. HEARST
  35.Natural Language Interaction ION ANDROUTSOPOULOS AND MARIA ARETOULAKI
  36.Natural Language in Multimodal and Multimedia Systems ELISABETH ANDR
  37.Natural Language Processing in Computer-Assisted Language Learning JOHN NERBONNE
  38.Multilingual On-Line Natural Language Processing
Notes on Contributors
Glossary
Index of Authors
Subject Index

插图：


If morphophonological processes in a language are few and local the lemma lexi-con approach can still be successful. In our example it suffices to assume two pluralendings: -s and -es. For all base forms it must be specified whether the former or thelatter of the two endings may be attached.A part from the obvious limitations with regard to the treatment of morphopho-nological rules on a more general scale the approach has some other inherent restric-tions:The algorithm is geared towards analysis. For generation purposes, one needs acompletely different algorithm and data.Interpretation algorithms are language specific because they encode both thebasic concatenation algorithm and the specific exception-handling mechanism.The approach was developed for morphosyntactic analysis. An extension tohandle more generally the segmenting of word forms into morphs is difficult toachieve.