Intellect

Connectionism has been gaining ground as a psychological modeling technique and shows great promise as a way to build fast, robust systems to perform intelligent tasks. Connectionism contrasts with the older tradition, that of explaining intelligent behavior in terms of the manipulation of complex symbol structures. Without expecting either style of research to be exclusively correct, it is still important to seek out the relative strengths and weaknesses of each. The partisans of the two camps have leapt to the challenge; the controversy has been fierce, as befits the prize-acceptance as the best technique for understanding people and building intelligent systems.
Language, the quintessential human activity, is often used as a touchstone for the two approaches. The traditional approach views language as composed of symbols and language use as the manipulation of symbol structures. Connectionists have largely conceded to this, accepting the idea of structure as an essential component of language.
This book calls into question this common assumption, that "structure is necessary for language modeling," by presenting a generator which produces appropriate natural language utterances without building structures along the way. It backs up this demonstration by an analysis of the generation task, which leads to the conclusion that massively parallel computation and numeric combination of evidence are in fact intrinsically necessary for generation, and that, conversely, structure building is computationally awkward.

Contents

• Preface

INTRODUCTION
• Motivations for This Work
• Preview of the Model
• Overview of the Book

DESIGN ISSUES
• Characteristics of the Generation Task
• Why Previous Research Has Missed the Point
• Design Principle
• The Principles in FIG
• On Decisions, Algorithms, and Modules

LEXICAL KNOWLEDGE AND WORD CHOICE
• Meanings of Words
• Inference
• Syntactic Properties of Words
• Other Properties of Words
• Word Choice at Run-Time
• Summary

SYNTACTIC KNOWLEDGE AND ITS USE
• Motivation
• Basics of Syntax
• Two Details
• An Example
• Synergy and Competition
• Issues and Non-issues
• What Remains to Be Done?
• History of Syntax in FIG
• Summary

PRESENTlNG AND USING RELATIONAL INFORMATION
• The Problems With Case Grammars
• Proposal
• Participatory Profiles in FIG
• Implications for Parsing
• Open Issues
• Summary

FIG'S GRAMMARS
• Some Details of Syntax in FIG
• English
• Japanese

DETAILS OF FIG
• Building the Network
• The Input
• Activation Flow
• Special Processes
• Getting the Correct Overall Behavior
• Summary of Node and Link Types
• Size and Speed

MISCELLANY REGARDING CONNECTIONISM
• Strengths and Weaknesses of Connectionism
• Why Structured Connectionism?
• Artificial Intelligence as an Experimental Science
• Past Connectionist Generation Research

HUMAN LANGUAGE PRODUCTION
• Introspection
• Pauses
• Priming Effects
• Errors
• Traditional Cognitive Models
• FIG As a Cognitive Model

A MODEL FOR NATURAL TRANSLATION
• The Need for Natural Translation
• Strategies for Machine Translation Research
• Present Technologies for Machine Translation
• Proposal
• Design Implication
• Philosophical and Software Engineering Issues
• Prospects

IN CONCLUSION
• How FIG Measures Up
• Directions for Future Work
• What Has Been Learned
• Appendix
• References
• Author Index
• Subject Index