CLASPnet : Table of Contents

Table of Contents

Preface

Introduction

Setting the Scenery

Introducing CLASPnet

Motivations: Theoretical and Practical

1.2.1 The Nature of Thought and Language

1.2.2 The Issue of Language Acquisition

1.2.3 Bridging Connectionism and Theoretical Linguistics

1.2.4 The Cognitive Linguistics Part
1.3 Some Methodological Considerations

A Look at Related Work

2.0 Neural Networks and Natural Language
2.1 Computer Science -- Network Basics
2.2 Language Acquisition -- The Lexicon
2.3 NLP -- Making It Work
2.4 Cognitive Science -- Mixing the Bits

CLASPnet

3.0 Introduction
3.1 The Network

3.1.1 The Input Layer

3.1.1.1 The Orthographic Representation

3.1.1.2 The Semantic Representation

3.1.2 The Hidden Layers

3.1.3 The Output Layer
3.2 The Corpora

3.2.1 Introduction and Motivation

3.2.2 The Grammar and the Lexicon

3.2.3 The Parser
3.3 Training and Testing Results

3.3.1 Training

3.3.2 General Results

3.3.3 Detailed Results

3.3.3.1 Sentential Mood

3.3.3.2 ID and Status Units

3.3.3.3 Infinity, Voice, and Polarity

3.3.3.4 The Clausal Types

3.3.4 Generalization to New Patterns

3.3.4.1 An Abundance of Words

3.3.4.2 Going Down, Down, Down

3.3.4.3 Double or Quits

3.3.4.4 Parsing New Words

3.3.4.5 The Hidden Representations

3.3.4.6 Some Natural Language Samples
3.4 Experiments

3.4.1 The Importance of Being Modular

3.4.2 Increasing the Size of the Training Corpus

3.4.3 No Syntax without Semantics?

3.4.3.1 Deprivation of Orthography or Semantics

3.4.3.2 Training on Only Orthography or Semantics

3.4.4 Perceptual vs Gestalt Semantics

3.4.5 Left Justification, Right Justification, or Both?

3.4.6 The Relevance of Punctuation

3.4.6.1 Deprivation of Punctuation Marks

3.4.6.2 Training on Limited Punctuation

3.4.7 The Acquisition of Clause Spotting Capacities

3.4.8 Analyzing versus Predicting

Conclusion

4.0 Introduction
4.1 Summary of Results
4.2 Discussion
4.3 Possible Extensions
4.4 The End

List of References

Appendices

A.1 The Grammar
A.2 The Vocabulary
A.3 Generate.Input
A.4 Restore.Input
A.5 Precise Network Results

A.5.1 Chapter 3.3 Charts

A.5.1.1 Training Corpus Results

A.5.1.2 Test Corpus 1 Results

A.5.1.3 Test Corpus 2 Results

A.5.2 Chapter 3.4 Charts

A.5.2.1 The Importance of Being Modular

A.5.2.2 Increasing the Size of the Training Corpus

A.5.2.3.1 Deprivation of Orthography or Semantics

A.5.2.3.2 Training on Only Orthography or Semantics

A.5.2.4 Perceptual vs Gestalt Semantics

A.5.2.5 Left Justification, Right Justification, or Both?

A.5.2.6.1 Deprivation of Punctuation Marks

A.5.2.6.2 Training on Limited Punctuation

A.5.2.7 The Acquisition of Clause Spotting Capacities

A.5.2.8 Analyzing versus Predicting
A.6 Word by Word Analyses

A.6.1 Generated Sentences

A.6.2 Natural Language Sentences 169

Copyright 1996. All rights reserved. Ezra Van Everbroeck

Last change: 10 July 1996
http://snow.ucsd.edu/~ezra/msc/toc.html