Natural Language Computing: Table of Contents

Natural Language Computing
An English Generative Grammar in Prolog

ISBN: 0-8058-1525-2 cloth $69.95 || ISBN: 0-8058-1526-0 paper $36
Ordering Information: Contact Lawrence Erlbaum (order) (information)(blurb)

Prof. Ray C. Dougherty
New York University | Linguistics Department
GENERAL || RESEARCH || COURSES || WORKBOOK|| FIRST DOT

349+xlvi pages, plus a 1.44 meg floppy containing all programs in the book and Prolog Interpreters for DOS, Windows, and Macintosh.

An Introductory book that shows how to encode Chomsky's generative grammar into Prolog. It is aimed at an intelligent reader who knows little or nothing about Chomsky's grammar and less about Prolog. For linguistic tenderloins and computational greenhorns the book might bear the subtitle: Start here.

If you have any comments concerning the book, please let me know. In response to readers, I have added several pages of backup materials to help those using it in classes or on their own. If you have an interesting program, please send it to me. Perhaps we can place it on this website.

Introductory Materials from the Book

Here is the Preface and Introduction to the book in Adobe PDF format. You should install the Adobe PDF free software so you can read the newspapers of the world: The NY Times, Le Monde, Der Spiegel, and so on.

Chapters and Appendices

Natural Intelligence, Linguistics, and Prolog.
How to Read and Write in Prolog
How to Load, Run, and Edit a Prolog Program
Tables of Data as Prolog Facts and Relations
How Prolog Backtracks in Searches
The Canonical Form of a Prolog Relation
Computational Tools for Language Processing
Computational Tools for Sentence Processing

Appendices, References, Index of Names, Index

How to Use this Book: A Pictorial Essay

Figure 1. Mice yield to NLP.
Figure 2. A parser indicates the internal structure of a string of sounds, morphemes, words, or phrases.
Figure 3. Three necessary disks.
Figure 4. A natural language parser is an intelligent machine.
Figure 5. All programs in the book are available on disk or by ftp.
Figure 6. How to use the Prolog program files on disk to follow the session files in this book.
Figure 7. Three ways for a teacher to present the materials to students: screen projection, overheads, blackboard.
Figure 8. How to make overheads and presentations from Figureures and tables on disk.
Figure 9. A transparency projection setup for faculty and students doing research and development.
Figure 10. A New York University student computer workstation.
Figure 11. NLP Lotus Show disks.
Figure 12. Presenting the NLP Lotus Show files on a big monitor.

1.0. Natural Intelligence, Linguistics, and Prolog.

1.1. The Information Society

1.2. An Old Theory of Computer Intelligence

1.3. A New Theory of Computer Intelligence

1.4. A Prolog Theory of Computer Intelligence

1.5. Why Study Linguistics and Prolog?

1.6. Natural Language Computing in Prolog

1.7. Artificial Intelligence Machines

2.0. How to Read and Write in Prolog

2.1. Six Characteristics of Prolog Programs

2.2. How to Read and Write in Prolog

2.3. Four Steps to Running a Prolog Program

Figure 2.1. Three possible computer configurations.
Figure 2.2. The four steps of any Prolog session.
Figure2.3. The session (script) file is the screen contents.

2.4. Running shareware Prolog on an IBM PC

3.0. How to Load, Run, and Edit a Prolog Program

3.1. The Prolog Interpreter and Workspace

Figure 3.1. A functional view of the Prolog workspace.

3.2. How to Load and Run a Prolog Program

Figure 3.2. The form of Prolog problems, solutions, and answers.
Figure 3.3. The form of Prolog problems, solutions, and answers.
Figure 3.4. How Prolog seeks values of A,B,C...I for which sqword is true.
Figure 3.5. How Prolog seeks values of A,B,C...I for which p0310 is true.

3.3. Running Prolog on an IBM PC

3.4. Modifying the Word Squares Program

3.5. The Edit-Prolog-Edit Loop

Figure 3.6. The Edit-Prolg-Edit loop.

4.0. Tables of Data as Prolog Facts and Relations

Figure 4.1. Representing tables in Prolog.

4.1. Converting Tables into Facts and Relations

4.2. How to Extract Information from Tables

4.3. Variables in a Query

4.4. Variables in the Database

4.5. Tables with Holes: cut, fail, and an empty category

5.0. How Prolog Backtracks in Searches

Figure 5.1. Representing a bibliography in Prolog.

5.1. Information Retrieval: Bibliographic Searches

5.2. Searches with and without Backtracking

5.3. Backtracking Caused by Filters

Figure 5.2. How search1 (A,B,C) works in Session 5.1.
Figure 5.3. How search4 (A,B,C) works in Session 5.5.
Figure 5.4. How search4 (A,B,C) works in Session 5.5.

5.4. Bound and Unbound Variables

5.5. Bounded and Unbounded Variables

5.6. A Blunder Popular with Beginning Programmers

5.7. Report Generation: Formatting Output and write

6.0. The Canonical Form of a Prolog Relation

6.1. Some Irregular Verbs in English

6.2. The Six Modules of a Prolog Program

Figure 6.1. The canonical form of a Prolog relation.

6.3. The Bracketing of Logical Constituents

7.0. Computational Tools for Language Processing

7.1. The Levels of Human Language Structure

Figure 7.1. The term "grammar" is ambiguous.
Figure 7.2. Levels of linguistic structure.
Figure 7.3. A Prolog program defines each level.
Figure 7.4. A phrase marker is a linguistic data structure.
Figure 7.5. Four phrase markers for simple sentences.
Figure 7.6. The grammar is neutral between perception and production.
Figure 7.7. An overview of our interaction with the computation model of the grammar.

7.2. Morphological Parsers

Figure 7.8. Linguistic levels below the word level.
Figure 7.9. Three types of morphological problems.
Figure 7.10. In Prolog the orthographic and morpheme levels are lists of (lists of (lists)).
Figure 7.11. Prolog append is a three place predicate.
Figure 7.12. The relation among word, root, and suffix.
Figure 7.13. Two "append" statement linked by "and."
Figure 7.14. An ambiguous compound noun has two structures.

7.3. Recursion: Affixes on the Affixes

Figure 7.15. The relation "affix_all_1(word)."
Figure 7.16. "Append" can recursively remove affixes.
Figure 7.17. "Append" recursively removes prefixes and suffixes.
Figure 7.18. A "tree search" diagram on a complex word.
Figure 7.19. Independent prefix and suffix parsers in parallel.
Figure 7.20. Left-branching versus right-branching structures.

7.4. Regular and Irregular Morphology

Figure 7.21. In the workspace, Prolog uses only numbers.
Figure 7.22. The Prolog data structures of the "plural" relation.

7.5. The Minimalist Framework

Figure 7.23. The non-procedural logic minimalist theory.
Figure 7.24. An analysis of ambiguity following from combinatorial principles of morphology.
Figure 7.25. An analysis of ambiguity following from combinatorial principles of syntax.

8.0. Computational Tools for Sentence Processing

8.1. Syntax: Representations and Parsers

Figure 8.1. A coordinate conjoined compound sentence.
Figureure Figure 8.2. A subordinate compound sentence structure.
Figure 8.3. I-language assumes the grammar is a cluster of Prolog parsers in parallel.
Figure 8.4. The functional view of the organization of universal grammar.
Figure 8.5. How a Prolog parser builds structure using the lexicon and "append" to define phrases and sentences.
Figure 8.6. The sources of language samples to be fed to the parser.

8.2. Rule Governed Creativity: Derivations

Figure 8.7. The relation "np(X) :- det(A),n(B),append(A,B,X)."
Figure 8.8. A "top-down" parser searches for tokens of categories.
Figure 8.9. A "bottom-up" parser tags items lexically by category.
Figure 8.10 The application of the recursion principle.

8.3. Parsers Assign Structure to an Ordered String

Figure 8.11. How g0806 assigns structure in Session 8.5.

8.4. Top-Down and Bottom-Up Parsing

Figure 8.12. A parser defines a pair string,constituent_structure.
Figure 8.13. The parser solves a puzzle in combinatorics and geometry.
Figure 8.14. Grammar defines the geometric and combinatorial constraints of sentence construction.
Figure 8.15. A Prolog parser functions like a domino playing machine.
Figure 8.16. Prolog backtracking is equivalent to removing dominos from the board.
Figure 8.17. Parallel Prolog parsing strategies.
Figure 8.18. The domino playing parser is matching a string with a phrase marker.
Figure 8.19. The Prolog interpreter joins the pieces which are defined by the "append" relation.
Figure 8.20. Linguistic research defines the possible complexity of puzzle pieces.

8.5. Horizontal Appends: Complement Structures

Figure 8.21. Phrase markers defined by g0810 and g0811.
Figure 8.22. Phrase markers defined by g08110 and g0811.
Figure 8.23. "Horizontal appends" define the head's arguments.
Figure 8.24. "Vertical appends" define the selection properties.

8.6. Vertical Appends: Selection Restrictions

Figure 8.25. Lexical entries defined with "append" relations.
Figure 8.26. A phrase marker with some complement and adjunct constructions labeled.

Natural Language ComputingAn English Generative Grammar in Prolog

Table of Contents

ISBN: 0-8058-1525-2 cloth $69.95 || ISBN: 0-8058-1526-0 paper $36 Ordering Information: Contact Lawrence Erlbaum (order) (information)(blurb)