Control of Mixed-Initiative Discourse

Through Meta-Locutionary Acts:

A Computational Model

 

David Graham Novick

 

CIS-TR-88-18

 

December 1988

 

 

This technical report is a slightly abbreviated version of the author’s dissertation presented to the Department of Computer and Information Science and the Graduate School of the University of Oregon in partial fulfillment of the requirements for the degree of Doctor of Philosophy, December 1988.

 

 

© 1988 David Graham Novick

 

 

 

 

ABSTRACT

 

CONTROL OF MIXED-INITIATIVE DISCOURSE THROUGH

META-LOCUTIONARY ACTS:  A COMPUTATIONAL MODEL

 

Human-computer interaction typically displays single-initiative interaction in which either the computer or the human controls the conversation.  The interaction is largely preplanned and depends on well-formed language.  In contrast, human-human conversations are characterized by unpredictability, ungrammatical utterances, non-verbal expression, and mixed-initiative control in which the conversants take independent actions.  Traditional natural-language systems are largely unable to handle these aspects of "feral" language.  Yet human-human interaction is coherent for the participants; the conversants take turns, make interruptions, detect and cure misunderstandings, and resolve ambiguous references.  How can these processes of control be modeled formally in a manner sufficient for use in computers?

 

Non-sentential aspects of conversation such as nods, fragmentary utterances, and correction can be seen reflecting control information for interaction.  Such actions by the conversants, based on the context of their interaction, determine the form of the conversation.  In this view ungrammaticality, for example, is not a problem but a guide to these "meta" acts.  This dissertation develops a theory of "meta-locutionary" acts that explains these control processes.  The theory extends speech-act theory to real-world conversational control and encompasses a taxonomy of meta-locutionary acts.

 

The theory of meta-locutionary acts was refined and validated by a protocol study and computational simulation.  In the protocol study, subjects were given a cošperative problem-solving task.  The conversants' interaction, both verbal and non-verbal, was transcribed as illocutionary and meta-locutionary acts.  The computational model was developed using a rule-based system written in Prolog.  The system represents the independent conversational knowledge of both conversants simultaneously, and can simulate their simultaneous action.  Simulations of the protocol conversations using the computational model showed that meta-locutionary acts are capable of providing control of mixed-initiative discourse.  The model agents can, for example, take and give turns.  A single agent can simultaneously take multiple acts of differing control.  The simulations also confirmed that conversations need not be strictly planned.  Rather, mixed-initiative interaction can be plausibly controlled by contextually determined operators.

 

This research has application to natural language processing, user interface design and multiple-agent artificial intelligence systems.  The theory of meta-locutionary acts will integrate well with existing speech-act-based natural-language systems.

 

 

 

ACKNOWLEDGMENTS

 

I wish to express appreciation to Professor Sarah A. Douglas for her guidance, inspiration, faith and perserverance.  I thank Professors Scott C. DeLancey, Stephen F. Fickas, and Kent A. Stevens for their constructive, diligent work as members of the dissertation committee.  Professor Russell S. Tomlin was very helpful with respect to discourse issues.  Barbara J. Grosz graciously reviewed my dissertation proposal and provided many detailed comments; I am grateful for her time and assistance.  I received helpful comments from the mentors and members of the ACM SIGCHI CHI'88 Doctoral Consortium.  I also wish to thank audiences at Xerox PARC, Information Sciences Institute, and the University of Oregon Cognitive and Decision Sciences Colloquium for their constructive responses to early presentations of this research.  This work was supported in part by a grant from the Foundation for the Improvement of Post-Secondary Education, U.S. Department of Education, number 84.116C, to Professors Douglas and Tomlin, and by a contract from U S West Advanced Technologies to Professor Douglas.

 

 

 

TABLE OF CONTENTS

 

Chapter

 

            I.          INTRODUCTION      1

 

                                    Speech-Act Theory      3                                  Mutuality of Knowledge            5

 

            II.         SURVEY OF RELATED WORK       9

 

                                    The Integration Problem            9

                                    Characteristics of Conversation 10

                                    Evidence for Shared Models of Conversations  11

                                    Repair-Based Conversational Interaction           15

                                    Intention, Action, and Language            21

                                    Summary          25

 

            III.       A THEORY OF META-LOCUTIONARY ACTS      27

 

                                    Approaches to the Integration Problem 27

                                    A Computational Model of Meta-Locutionary Acts       29

                                    Summary          35

 

            IV.       METHODOLOGY      37

 

                                    Methodology in Cognitive Science        37

                                    Effects to be Studied     38

                                    Protocol Analysis of Conversational Interaction 44

                                    Summary          44

 

            V.        RESULTS OF THE PROTOCOL STUDY     47

 

                                    Modeling the Conversation       47

                                    Application of the Model          52

 

            VI.       THE SIMULATION STUDY  61

 

                                    Representation  62

                                    The Rule-Based System            65

                                    Implementation of the Model as Rules   68

                                    Results of the Simulation           71

                                    Limitations and Extensions        74

 

 

            VI.       CONCLUSION          77

 

                                    Summary          77

                                    Open Issues     78

                                    Humans and Computers            82

 

APPENDIX

 

            A.        Transcript of Protocol        85

 

            B.         PREDICATE REPRESENTATIONS  93

 

            C.        CONVERSATIONAL MODELS       95

 

            D.        CONVERSATIONAL OPERATORS 99

 

            E.         SHORT TRACE OF SIMULATED CONVERSATION         103

 

            F.         FULL TRACE OF SIMULATED CONVERSATION            107

 

            G.        RULE-BASED SYSTEM:  PROGRAM          131

 

            H.        RULE-BASED SYSTEM:  META-LOCUTIONARY OPERATORS 145

 

            I.          RULE-BASED SYSTEM:  DOMAIN OPERATORS 153

 

            J.          RULE-BASED SYSTEM:  DOMAIN-SPECIFIC CLAUSES            163

 

            K.        RULE-BASED SYSTEM:  INITIAL STATE   165

 

BIBLIOGRAPHY        167

 

 

LIST OF FIGURES

 

Figure

 

            1.  Differences Between Interaction Modes       1

 

            2.  Possible Levels of Conversational Interaction           17

            3.  Taxonomy of Meta-Locutionary Acts for Turn-Taking          34

            4.  Taxonomy of Meta-Locutionary Acts for Repair of Mutual Models  35

            5.  Taxonomy of Meta-Locutionary Acts for Information           35

            6. Taxonomy of Meta-Locutionary Acts for Attention    35

            7.  Example of Random-Letter Sequences Used in Protocols     42

            8.  Example of Non-Identical Sequences Used in Third Task     43

            9.  Overhead View of the Experimental Set-Up for the Letter-Sequence Protocols        43

            10.  Illocutionary Interpretation of the Protocol  49

            11.  Operator Repeat-Act-1     51

            12.  Application of Operator Repeat-Act-1 After Act Bi6         51

            13.  Partial Transcript of Experimental Protocol 53

            14.  Form of Rule Representing Meta-Locutionary Operator     64

            15.  Meta-Locutionary Rule Do_Assert:            69

            16.  Meta‑Locutionary Rule Assertion_Received_1       69

            17.  Domain-Rule Confirmed_Next_Letter        70

            18.  Trace of Simulation of Protocol      72