Bridging the Gap Between Ai, Cognitive Science, and Narratology With Narrative Generation [1 ed.] 1799848647, 9781799848646 - DOKUMEN.PUB

Bridging the Gap Between Ai, Cognitive Science, and Narratology With Narrative Generation [1 ed.] 1799848647, 9781799848646

The use of cognitive science in creating stories, languages, visuals, and characters is known as narrative generation, a

323 60 23MB

English Pages 409 [435] Year 2020

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Bridging the Gap Between Ai, Cognitive Science, and Narratology With Narrative Generation [1 ed.]
 1799848647, 9781799848646

Table of contents :
Title Page
Copyright Page
Book Series
Editorial Advisory Board
Table of Contents
Detailed Table of Contents
Preface
Section 1: Cognitive Approaches to Narratology and Narrative Generation
Chapter 1: Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation
Chapter 2: Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models
Chapter 3: Extension of Clinical/Psychological Approach Using Post Narratology
Chapter 4: Defining the Verbs for “Understanding and Interpretation” of Japanese Sake
Chapter 5: 300–500 Threshold of Context Memory and Assumption Experiment
Section 2: Narrative Generation Systems and Narrative Content Generation
Chapter 6: Bridging the Gap Between Narrative Generation Systems and Narrative Contents With Kabuki-Oriented Narratology and Watakushi Monogatari
Chapter 7: Haiku Generation From Narratological Perspective
Chapter 8: On Logical Literary Work Generation and More
Chapter 9: An Analysis of Cognitive Elements and Effects of Video Commercials
Section 3: Social Perspectives of Narratological and Narrative Generation Methods
Chapter 10: Narrative Generation Using Psychological Value Variables
Chapter 11: Effect of External Activate Factors Serving as Clue
Chapter 12: Design of Narrative Creation in Innovation
Compilation of References
About the Contributors
Index
Пустая страница
Пустая страница

Citation preview

Bridging the Gap Between AI, Cognitive Science, and Narratology With Narrative Generation Takashi Ogata Iwate Prefectural University, Japan Jumpei Ono Aomori University, Japan

A volume in the Advances in Human and Social Aspects of Technology (AHSAT) Book Series

Published in the United States of America by IGI Global Information Science Reference (an imprint of IGI Global) 701 E. Chocolate Avenue Hershey PA, USA 17033 Tel: 717-533-8845 Fax: 717-533-8661 E-mail: [email protected] Web site: http://www.igi-global.com Copyright © 2021 by IGI Global. All rights reserved. No part of this publication may be reproduced, stored or distributed in any form or by any means, electronic or mechanical, including photocopying, without written permission from the publisher. Product or company names used in this set are for identification purposes only. Inclusion of the names of the products or companies does not indicate a claim of ownership by IGI Global of the trademark or registered trademark. Library of Congress Cataloging-in-Publication Data Names: Ogata, Takashi, 1958- editor. | Ono, Jumpei, 1987- editor. Title: Bridging the gap between AI, cognitive science, and narratology with narrative generation / Takashi Ogata and Jumpei Ono, editors. Description: Hershey PA : Information Science Reference, [2021] | Includes bibliographical references and index. | Summary: “This book combines traditional theories of narratology with current narrative generation practices using artificial intelligence”-- Provided by publisher. Identifiers: LCCN 2020010862 (print) | LCCN 2020010863 (ebook) | ISBN 9781799848646 (h/c) | ISBN 9781799856245 (s/c) | ISBN 9781799848653 (ebook) Subjects: LCSH: Rhetoric--Data processing. | Discourse analysis, Narrative--Data processing. | Cognitive science. | Artificial intelligence. Classification: LCC P301.5.D37 B75 2021 (print) | LCC P301.5.D37 (ebook) | DDC 808/.036--dc23 LC record available at https://lccn.loc.gov/2020010862 LC ebook record available at https://lccn.loc.gov/2020010863 This book is published in the IGI Global book series Advances in Human and Social Aspects of Technology (AHSAT) (ISSN: 2328-1316; eISSN: 2328-1324) British Cataloguing in Publication Data A Cataloguing in Publication record for this book is available from the British Library. All work contributed to this book is new, previously-unpublished material. The views expressed in this book are those of the authors, but not necessarily of the publisher. For electronic access to this publication, please contact: [email protected]

Advances in Human and Social Aspects of Technology (AHSAT) Book Series Ashish Dwivedi The University of Hull, UK

ISSN:2328-1316 EISSN:2328-1324 Mission

In recent years, the societal impact of technology has been noted as we become increasingly more connected and are presented with more digital tools and devices. With the popularity of digital devices such as cell phones and tablets, it is crucial to consider the implications of our digital dependence and the presence of technology in our everyday lives. The Advances in Human and Social Aspects of Technology (AHSAT) Book Series seeks to explore the ways in which society and human beings have been affected by technology and how the technological revolution has changed the way we conduct our lives as well as our behavior. The AHSAT book series aims to publish the most cutting-edge research on human behavior and interaction with technology and the ways in which the digital age is changing society.

Coverage • Cyber Behavior • Gender and Technology • Human-Computer Interaction • Technoself • Human Rights and Digitization • Cyber Bullying • Human Development and Technology • Technology Adoption • ICTs and social change • Information ethics

IGI Global is currently accepting manuscripts for publication within this series. To submit a proposal for a volume in this series, please contact our Acquisition Editors at [email protected] or visit: http://www.igi-global.com/publish/.

The Advances in Human and Social Aspects of Technology (AHSAT) Book Series (ISSN 2328-1316) is published by IGI Global, 701 E. Chocolate Avenue, Hershey, PA 17033-1240, USA, www.igi-global.com. This series is composed of titles available for purchase individually; each title is edited to be contextually exclusive from any other title within the series. For pricing and ordering information please visit http:// www.igi-global.com/book-series/advances-human-social-aspects-technology/37145. Postmaster: Send all address changes to above address. Copyright © 2021 IGI Global. All rights, including translation in other languages reserved by the publisher. No part of this series may be reproduced or used in any form or by any means – graphics, electronic, or mechanical, including photocopying, recording, taping, or information and retrieval systems – without written permission from the publisher, except for non commercial, educational use, including classroom teaching purposes. The views expressed in this series are those of the authors, but not necessarily of IGI Global.

Titles in this Series

For a list of additional titles in this series, please visit:

www.igi-global.com/book-series/advances-human-social-aspects-technology/37145

Dyslexia and Accessibility in the Modern Era Emerging Research and Opportunities Kamila Balharová (Pedagogical and Psychological Counseling Center, Brno, Czech Republic) Jakub Balhar (Gisat s.r.o., Czech Republic) and Věra Vojtová (Masaryk University, Czech Republic) Information Science Reference • © 2021 • 279pp • H/C (ISBN: 9781799842675) • US $165.00 Understanding the Role of Artificial Intelligence and Its Future Social Impact Salim Sheikh (Saïd Business School, University of Oxford, UK) Engineering Science Reference • © 2021 • 284pp • H/C (ISBN: 9781799846079) • US $200.00 Overcoming Barriers for Women of Color in STEM Fields Emerging Research and Opportunities Pamela M. Leggett-Robinson (PLR Consulting, USA) and Brandi Campbell Villa (Belay Consulting, USA) Information Science Reference • © 2020 • 199pp • H/C (ISBN: 9781799848585) • US $145.00 Maintaining Social Well-Being and Meaningful Work in a Highly Automated Job Market Shalin Hai-Jew (Kansas State University, USA) Business Science Reference • © 2020 • 333pp • H/C (ISBN: 9781799825098) • US $215.00 ICTs and Innovation for Didactics of Social Sciences Emilio José Delgado-Algarra (University of Huelva, Spain) Information Science Reference • © 2020 • 295pp • H/C (ISBN: 9781799828822) • US $175.00 Civic Engagement Frameworks and Strategic Leadership Practices for Organization Development Susheel Chhabra (Periyar Management and Computer College, India) and Muneesh Kumar (Maharaja Agrasen Institute of Technology (MAIT), Delhi, India) Information Science Reference • © 2020 • 314pp • H/C (ISBN: 9781799823728) • US $195.00 Civic Engagement in Social and Political Constructs Susheel Chhabra (Periyar Management and Computer College, India) Information Science Reference • © 2020 • 285pp • H/C (ISBN: 9781799823643) • US $195.00 Recent Advances in Digital Media Impacts on Identity, Sexuality, and Relationships Michelle F. Wright (Penn State University, USA) Information Science Reference • © 2020 • 347pp • H/C (ISBN: 9781799810636) • US $195.00

701 East Chocolate Avenue, Hershey, PA 17033, USA Tel: 717-533-8845 x100 • Fax: 717-533-8661 E-Mail: [email protected] • www.igi-global.com

Editorial Advisory Board Akinori Abe, Chiba University, Japan Akihito Kanai, HOSEI University, Japan Yoji Kawamura, Kindai University, Japan Kai Seino, National Rehabilitation Center for Persons With Disabilities, Japan Akira Utsumi, The University of Electro-Communications, Japan



Table of Contents

Preface..................................................................................................................................................xiii Section 1 Cognitive Approaches to Narratology and Narrative Generation Chapter 1 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation .........................1 Akihito Kanai, Hosei University, Japan Chapter 2 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models ............................17 Sheldon Schiffer, Georgia State University, USA Chapter 3 Extension of Clinical/Psychological Approach Using Post Narratology: The Proposal of a Psychological Narratology and Review for Systematization ................................................................43 Kai Seino, National Rehabilitation Center for Persons With Disabilities, Japan Shun Ishizaki, Keio University, Japan Chapter 4 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake .....................................63 Hiroki Fukushima, Kyushu Women’s University, Japan Chapter 5 300–500 Threshold of Context Memory and Assumption Experiment ................................................97 Nobuyoshi Harada, Chiba University, Japan Section 2 Narrative Generation Systems and Narrative Content Generation Chapter 6 Bridging the Gap Between Narrative Generation Systems and Narrative Contents With KabukiOriented Narratology and Watakushi Monogatari..............................................................................126 Takashi Ogata, Iwate Prefectural University, Japan  



Chapter 7 Haiku Generation From Narratological Perspective: A Circulation Between Haikus and Stories .....249 Jumpei Ono, Aomori University, Japan Takashi Ogata, Iwate Prefectural University, Japan Chapter 8 On Logical Literary Work Generation and More ...............................................................................266 Akinori Abe, Chiba University, Japan Chapter 9 An Analysis of Cognitive Elements and Effects of Video Commercials: Toward Effective Indexing and Video Production ..........................................................................................................283 Yoji Kawamura, Kindai University, Japan Section 3 Social Perspectives of Narratological and Narrative Generation Methods Chapter 10 Narrative Generation Using Psychological Value Variables: Probabilistic Model of Language Expressions and Values.......................................................................................................................311 Yasuo Tanida, Kotonoha Research Laboratory, Japan Chapter 11 Effect of External Activate Factors Serving as Clue: Creating New Products or Services Ideas With Storytelling.................................................................................................................................338 Jun Nakamura, Chuo University, Japan Chapter 12 Design of Narrative Creation in Innovation: “Signature Story” and Two Types of Pivots.................363 Akimitsu Hirota, Kindai University, Japan Compilation of References ...............................................................................................................377 About the Contributors ....................................................................................................................404 Index ...................................................................................................................................................407

Detailed Table of Contents

Preface..................................................................................................................................................xiii Section 1 Cognitive Approaches to Narratology and Narrative Generation Chapter 1 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation .........................1 Akihito Kanai, Hosei University, Japan This chapter surveys, discusses, and explores the entire concept of visual narrative structure, cognition, and generation from continuity-based to discontinuity-based perspectives. The model of visual narrative structure, including presentation and meaning, is expanded to explain the cognitive transition based on the rhetorical transition techniques and the rhetorical cutting techniques. The classification of the visual narrative structure, including rhetorical transition techniques and rhetorical cutting techniques, is useful for narrative simulation for discussing and exploring the entire visual narrative concept and generation. The rhetorical cutting techniques and the rhetorical transition techniques can reveal various cognitive effects such as reality effects and nostalgia effects, including difficulty. The determinacy-based narrative, the indeterminacy, the diversity, and the ambiguity on narrative can bridge the gaps between cognitive science, artificial intelligence, and narratology. Chapter 2 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models ............................17 Sheldon Schiffer, Georgia State University, USA Video game non-player characters (NPCs) are a type of agent that often inherits emotion models and functions from ancestor virtual agents. Few emotion models have been designed for NPCs explicitly, and therefore do not approach the expressive possibilities available to live-action performing actors nor hand-crafted animated characters. With distinct perspectives on emotion generation from multiple fields within narratology and computational cognitive psychology, the architecture of NPC emotion systems can reflect the theories and practices of performing artists. This chapter argues that the deployment of virtual agent emotion models applied to NPCs can constrain the performative aesthetic properties of NPCs. An actor-centric emotion model can accommodate creative processes for actors and may reveal what features emotion model architectures should have that are most useful for contemporary game production of photorealistic NPCs that achieve cinematic acting styles and robust narrative design.

 



Chapter 3 Extension of Clinical/Psychological Approach Using Post Narratology: The Proposal of a Psychological Narratology and Review for Systematization ................................................................43 Kai Seino, National Rehabilitation Center for Persons With Disabilities, Japan Shun Ishizaki, Keio University, Japan The purpose of this chapter is to examine the psychological approach from a narrative viewpoint. In addition, the authors discuss the possibility of collaboration with artificial intelligence (AI). They reviewed the narrative approaches in psychology. At result, they showed how the narrative approach in clinical domain was expanded. The following possibilities exist regarding a narrative generative system in a narrow sense: (1) It is the construction of a fluid system. (2) The systems generate a tale communally. (3) There are no clear beginnings and endings. (4) It is a narrative generation through a dialogue. Next, the following possibilities exist regarding a narrative generative system in broad sense: (1) It is the modeling of the stance of the narrative. (2) It is the modeling of narrative generation by the community. (3) It is the modeling of the process. (4) It is the study of the interlocutory peculiarity. Chapter 4 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake .....................................63 Hiroki Fukushima, Kyushu Women’s University, Japan In this chapter, the author attempts to define the verbs in the description of Japanese sake taste by employing 1) a usage-based approach, 2) “encyclopedic semantics” rather than a “dictionary view,” and 3) sense-making theory, drawing on data from a “sake tasting description corpus” (approximately 120,000 words). The chapter selects eight verbs of high frequency (e.g., hirogaru ‘spread’) and defines their sense(s) in a bottom-up and abductive fashion, based on a score indicating the strength of co-occurrence between terms. In this study, the authors deal with the verbs for “Understanding” or “Interpretation ”; it means, verbs that contribute to narrating the personal, individual story (contents) of the tasters. This study suggests the verbs for understanding have senses related to [Timeline] and [Space]. On the other hand, verbs do not tend to collocate with [Movement] and interestingly, the [Structure], as same as the tendency of adjectival-nouns. Chapter 5 300–500 Threshold of Context Memory and Assumption Experiment ................................................97 Nobuyoshi Harada, Chiba University, Japan This chapter investigates the threshold of context memory on numbers 300 through 500 regarding perspective with assumption experiment and theories of cognitive semantics and short-term memory. Context generation became possible with text database with lexical engineering system on artificial intelligence (AI). It was realized an infinite and continuous presentation of context. Several context presentations realized the limitation of context memory on the process of continuous context presentation on numbers 500 and 2,000, as well as the representation process with judgement for memory. The first report studied the context generating machine for Japanese pun generation. This was termed “Class B engine” or “Dajare Machine.” This study focused on the examination of the context threshold using an experiment of human brain mapping on an auditory evoked magnetic field (AEF) of a magnetic encephalogram. It also focused on the analysis of the theory of cognitive semantics and folk biology, using a theory of short-term memory of the “magical number seven.”



Section 2 Narrative Generation Systems and Narrative Content Generation Chapter 6 Bridging the Gap Between Narrative Generation Systems and Narrative Contents With KabukiOriented Narratology and Watakushi Monogatari..............................................................................126 Takashi Ogata, Iwate Prefectural University, Japan The author’s narrative generation study is based on two types of systems: the integrated narrative generation system (INGS) as a single narrative generation and reception mechanism and the Geinō information system (GIS) as a multiple narrative production and consumption mechanism. The first theme of the chapter is to introduce an idea that deals with narrative phenomena as the integration of both systems. This theme is tied to the topic of narrative content creation by presenting kabuki narrative generation or kabuki-oriented narratology and Watakushi Monogatari as a collection of narrative content to be created by the author. Hence, the second objective is to describe kabuki-oriented narrative generation and the third is to explain the ideas, thoughts, and design underlying Watakushi Monogatari in the context of internal and external narrative generation to create and distribute narrative content. Through these three themes, this chapter bridges the gap between narrative generation systems and narrative content with kabuki-oriented narratology and Watakushi Monogatari. Chapter 7 Haiku Generation From Narratological Perspective: A Circulation Between Haikus and Stories .....249 Jumpei Ono, Aomori University, Japan Takashi Ogata, Iwate Prefectural University, Japan Haiku—a form of unrhymed poetry—is popular among the Japanese. A typical haiku is composed of 17 moras and three phrases. A haiku has the possibility of scratching the surface and uncovering a hidden message through an expression of events. According to Masaoka Shiki, a haiku is a kind of literature and has high affinity with our research on generating stories. In this chapter, the authors implemented the prototype system that has two functions: first, to produce multiple haikus from a single story, and second, to engender multiple stories from haikus. The system prototyped in this chapter is based on haiku theory, which is used by the authors in their research, and is rooted in the concept’s co-occurrence information and frequency information used to generate a haiku. The method uses statistical information for selecting the words and creating the word network in the haiku. Through the aforementioned methods, the authors created a framework for a system of circulating haiku and stories and proposed a kind of narrative generation with narrative as an input. Chapter 8 On Logical Literary Work Generation and More ...............................................................................266 Akinori Abe, Chiba University, Japan In several papers, the author has discussed the possibility of the computational literary work generation. That was based on Julia Kristeva’s the concept of the intertextuality. This very simple application was achieved by random combination of the selected words/phrases/letters. It is a simple technique such as the cutup method. The author showed even such a simple method could generate literary works that can be regarded as not so bad literary works. However, for more intelligent generation, the author introduced abduction. By abduction, it is possible to guarantee the logical consistency and coherency. However, it



does not care the very cognitive aspects such as beautiful, sentimental, and valuable, which will be very important factor in the literature. In this chapter, the author will discuss to add such cognitive aspects to the logical generation of literary works. Chapter 9 An Analysis of Cognitive Elements and Effects of Video Commercials: Toward Effective Indexing and Video Production ..........................................................................................................283 Yoji Kawamura, Kindai University, Japan This chapter starts with an analysis of differences between the indexes attached to still images in the image database and cognitive elements of still images extracted from the cognitive experiment. It then analyzes the relationship between cognitive elements of each video shot generated based on still images and cognition of the video commercial produced by combining them. Lastly, it discusses how to index the video shot and production methods of video commercials based on the analyses. The image database tends to attach the records of the timing and place of shooting and features of the persons as indexes. In order to produce an attractive video commercial, it is necessary that video shots convey rich cognitive elements that are not too simple. This ensures that when combining several video shots, their cognitive elements have some consistency and consistent cognitive elements are woven into the video commercial to constitute a rich semantic network. Section 3 Social Perspectives of Narratological and Narrative Generation Methods Chapter 10 Narrative Generation Using Psychological Value Variables: Probabilistic Model of Language Expressions and Values.......................................................................................................................311 Yasuo Tanida, Kotonoha Research Laboratory, Japan In terms of story generation, the author proposes a method wherein stories can be rewritten using psychological value variables to ensure positive evaluation from readers. This method includes two phases wherein (1) a story likely to be chosen by readers from a selection of stories in the story generation system is chosen and (2) the selected story is rewritten using words that readers can relate to. It is imperative to understand the psychological values of the recipient and the relation between linguistic expressions and psychological values to complete the two phases of this study. In this chapter, the author examines the definition of psychological values and how to obtain them as well as the verification of the relation between linguistic expressions and values. This study aimed to rewrite the output of a story generation system using psychological values, and this paper is a research report on the preparation for rewriting stories. A proposal is offered for a study to construct and verify the system. Chapter 11 Effect of External Activate Factors Serving as Clue: Creating New Products or Services Ideas With Storytelling.................................................................................................................................338 Jun Nakamura, Chuo University, Japan Why is there a demand for consultants and outside directors? That is because what they are asked for objective advice and the clue—this chapter describes them as external activate factors—is precious. This chapter focuses on how the external activate factor affects human creativity, especially the process



of creating new products or services, and the author conducted an experiment to compare the effect of external activate factors at a lecture of graduate school. As a result, when the external activate factor is provided, the emergence of abduction and new rules was observed, and the effect of storytelling was confirmed. Further research is to find what kind of external activate factor is preferable. Chapter 12 Design of Narrative Creation in Innovation: “Signature Story” and Two Types of Pivots.................363 Akimitsu Hirota, Kindai University, Japan Among brand stories that affect brand equity, the importance of signature stories has been pointed out in particular. Casio Computer Co.’s G-SHOCK enjoys strong brand equity. The first model has created a new market. The development of G-SHOCK went through two years of trial and error. The trial and error that can exist only in the development of a market-creating product exert a great impact on brand equity. Stories about the trial and error of the developer and marketer during the development of a marketcreating product are called “innovation narratives” in this chapter. This study examines the process of creating an innovation narrative from the viewpoint of a dialog and identifies the existence of a chain of two types of pivots: open pivots and closed pivots. Compilation of References ...............................................................................................................377 About the Contributors ....................................................................................................................404 Index ...................................................................................................................................................407

xiii

Preface

Narrative generation studies include challenging themes and topics covering artificial intelligence (AI) and cognitive science. Recently, many studies have been conducted on various aspects of narrative generation, including story, plot, character, and language and visual representations. However, few narrative generation studies have systematically inherited and used knowledge and techniques from narratology, literary theories, and other related areas of human and social sciences. On the other hand, the editors of this book have presented the concept of “post-narratology” to approach their narrative generation study. Post-narratology refers to the narratology expanded by AI, cognitive science, and related researches, and furthered by the unique narratologies and literary theories that have been developed through each culture. One of the editors (Ogata) has presented many ideas, concepts, and systems related to the topic in “Informatics of Narratology” (Ogata & Kanai, 2010), “Computational and Cognitive Approaches to Narratology” (Ogata & Akimoto, 2016), “Content Generation Through Narrative Communication and Simulation” (Ogata & Asakawa, 2018), “Informational Narratology” (Ogata, Kawamura, & Kanai, 2018), “Post-Narratology Through Computational and Cognitive Approaches” (Ogata & Akimoto, 2019), and other books and papers. Moreover, recently, Ogata (2020a, 2020b) presents a research system that consistently collects many related areas from the viewpoint of post-narratology. These books formalize, model, and systematize previous narratology, literary theories, and relating theories, methods, and studies from the perspective of narrative generation (and other content generation) to establish bases and foundations of narratology and literary theories in narrative generation studies. In particular, the last three books (Ogata & Akimoto, 2019; Ogata, 2020a, 2020b) reveal a newly developed stage of the authors’ narrative generation study based on post-narratology. As previously described, while narrative generation is one of the challenging research themes in AI and cognitive science, many narrative generation researches do not efficiently, systematically, and essentially use the knowledge and techniques that have been developed through previous narratology and literary theories. This indicates that bridging both fields is very difficult. In such an academic situation, the objective of this book is to bridge the gap between AI, cognitive science, and narratology with narrative generation in a broad sense, including other content generation, such as a novel, poem, movie, computer game, advertisement. To present an organic, systematic, and integrated combination of both fields to develop a new research area, namely post-narratology, this book plays an important role in the creation of a new research area and has an impact on narrative generation studies including AI and cognitive science, as well as narrative studies including narratology and literary theories. The target audience of this book includes researchers and university students from fields of information (such as AI, cognitive science, natural language semantics, and content generation) and literature (such as narratology, literary theories, and related human and social sciences). Recently, the research and 

Preface

development of automated content generation has come into practice as business services of automated document generation for newspapers and various types of reports. In contrast, this book proposes the concrete description of automated generation techniques and technologies of narratives that are more difficult to formalize, model, and systematize than the aforementioned general documents. Therefore, its potential users include enterprise practitioners, engineers, and creators of diverse content generation fields, such as advertising production, computer game creation, comic and manga writing, and movie production.

TOWARD A POST-NARRATOLOGY OF NARRATIVE GENERATION THROUGH BRIDGING THE GAP BETWEEN NARRATOLOGICAL STUDIES AND AI/COGNITIVE STUDIES The editors would like to concretely supply the basic framework of this book’s approach in Figure 1. Figure 1 indicates that there is a gap between narratology (or narratological studies related to it, from a broader perspective), and AI and cognitive science (and related research fields). This book aims to reduce the gap through the medium of narrative generation. However, the narrative generation in this case Figure 1. The basic framework of the whole book’s approach

xiv

Preface

is not limited to the range of narratives in the narrow sense, such as novels and artistic works. Regarding this aspect, the “theory of narrative genre system” by Ogata (2020b) views narratives in the wide range including narratives as literary and artistic works, narratives as practical texts, narratives as social events, and psychological and mental phenomena in the narrow sense. Ogata called his approach—that aims to solve the gap between narratological studies and AI/cognitive science through narrative generation, comprehended in the broad sense—post-narratology in his recent aforementioned books (Ogata & Akimoto, 2019; Ogata, 2020a, 2020b). The editors view the future direction of post-narratology that will produce and create diverse objects or works and impact literature, arts, and societies. Hence, Figure 1 and the previous explanation provide the basic framework of the entire post-narratology research by the editors, and this edited book is located as a result to post-narratology. On the other hand, Table 1 shows a panoptic list that includes narratological studies, AI and cognitive studies, and targets, specifically, systems and models for bridging between the narratological and AI/cognitive studies by all chapters. (The terms used in this table are frequently rewritten as different terms from the words that the authors originally described.) Explaining Table 1, first, this whole book adopts theories and methods of narratological studies, including narratology, in addition to marketing and management theories. Although marketing and management theories themselves are not included in narratology in a narrow sense, narratological perspectives to those academic fields will also be able to exist in the editors’ post-narratology approach. On the other hand, the theories and methods tied to researches related to AI and cognitive sciences cover many fields such as AI-based narrative generation techniques and cognitive approaches to various areas including image recognition and social cognition. Based on these, this book’s chapters deal with narrative (generation) systems and models from literature, arts, and entertainment to management, psychological, and mental phenomena for image, virtual character with emotion, narration by disabled, gustatory sense representation, puns (dajare), novels (narratives), haiku, poems, video commercials, narration based on a sense of values, storytelling of a new product, service development, and management innovation. In this manner, all the chapters directly and indirectly treat the theme of narrative generation in a broad sense. As previously stated, this book provides an approach to the systematic post-narratology that the editors have schemed and developed. In this sense, several previous books should be referenced to obtain a better understanding of this book.

BOOK ORGANIZATION This book contains 12 chapters in all and is divided into three sections, including “Cognitive Approaches to Narratology and Narrative Generation,” “Narrative Generation Systems and Narrative Content Generation,” and “Social Perspectives of Narratological and Narrative Generation Methods.” The theme of each section and outlines of the contained chapters are described below.

SECTION 1: COGNITIVE APPROACHES TO NARRATOLOGY AND NARRATIVE GENERATION This section includes the following five chapters that explore narratological approaches to cognitive models and systems. xv

Preface

Table 1. Main concepts related to the framework of the book used in all the chapters Chapter [Author(s)]

Narratological Studies

AI and Cognitive studies

Target (System or Model)

Chapter 1 [Akihito Kanai]

Narratology/Film theory

Cognitive film theories/ Cognitive transition and cutting techniques/Narrative image simulation

Narrative film simulation system

Chapter 2 [Sheldon Schiffer]

Narratology/Film theory

Computational and cognitive psychology/AI-based agent models/Computer game technologies/Emotion

The facial expression system with emotion of NPCs in computer games

Chapter 3 [Kai Seino and Shun Ishizaki]

Narratology/Post-narratology/ Psychological narratology/Clinical psychology

Text mining/ Psychological narratology

Clinical narrative generation system accompanying linguistic analysis

Chapter 4 [Hiroki Fukushima]

Sense-making theory/Narratology

Text mining/The cognition of the sense of taste

Linguistic description of the taste of Japanese sake

Cognitive semantics/Human memory/Brain function analysis/Japanese pun generation (dajare machine)

Memory model in literary work recognition

Chapter 5 [Nobuyoshi Harada] Chapter 6 [Takashi Ogata]

Narratology/Post-narratology/ Kabuki-oriented narratology/ Creating narratives

Narrative generation system/ Geinō information system

Narrative generation systems/Watakushi Monogatari

Chapter 7 [Jumpei Ono and Takashi Ogata]

Haiku theory/Narratology/Postnarratology

Haiku generation system/ Narrative generation system

Haiku generation system/Narrative generation system

Chapter 8 [Akinori Abe]

Narratology/Intertextuality

Abduction/Literary work generation systems

Literary work generation systems, including poems and short songs

Chapter 9 [Yoji Kawamura]

Marketing theory/Advertising theory/Film theory

Image processing techniques/ Cognitive effects of image representation

Video commercial generation system by still images

Chapter 10 [Yasuo Tanida]

Human sense of values/Narratology

Quantitative stochastic model of language/Text mining/Narrative generation techniques

Rewriting system of stories

Chapter 11 [Jun Nakamura]

Business studies/Narrative studies of management fields

Abduction/Storytelling modeling

Storytelling model of new product and service creation

Chapter 12 [Akimitsu Hirota]

Marketing theory/Business studies/ Signature story in management fields

Data mining/Storytelling modeling

Innovative storytelling model of marketcreating products

Chapter 1, “Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation,” by Akihito Kanai aims to present an entire concept of visual narrative structure based on the theory of cognitive transition and cutting techniques that Kanai has been studying. Depending on Kanai’s cognitive image theory, the rhetorical cutting and transition techniques can reveal various cognitive effects, including reality effects and nostalgia effects, with the feeling of difficulty of image recognition by audiences. An important aspect of Kanai’s work is that he emphasizes audiences’ difficult rather than xvi

Preface

easy understanding of image in narrative films. His goal in narrative film rhetoric is to model, systematize, and create narrative film techniques for aiming at difficulty and delay of the understanding by the audiences. This chapter expands his previous model regarding visual narrative structures including presentations and meanings, and shows various cognitive effects, such as reality effects and nostalgia effects, including difficulty in the rhetorical cutting techniques and rhetorical transition techniques. His study is related to narratology or narratological studies in the themes of film or visual narratology and defamiliarization-related thought or theory. Moreover, although his unique cognitive approach to narrative film rhetoric directly aims to present a cognitive model, one of his future goals is to apply his presented cognitive model to actual narrative film rhetoric simulation. In Chapter 2, “Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models,” Sheldon Schiffer presents a generic emotion model in the basic components of an NPC (non-player character) in computer games as an entertaining virtual agent. The emotion-response algorithms of NPCs dealt with in this chapter need to receive relevant data from the player and game play space and elicit expressive data back into the game play environment for the player to observe and respond. In the broad perspective, as Schiffer regards the world of a computer game as a kind of narrative in which characters—including NPCs—act in narrative spaces and times, he introduces narratological studies, including Chatman who deals with linguistic and image narratives to consider the problems of camerawork of human faces. On the other hand, the author refers to several computational psychological studies that overlapped with AI and cognitive science to consider in detail a virtual agent with emotion and its action. The author considers that cinematic acting styles provide valuable methods for facial expression in video games as they connect the emotional source of actor expression to NPCs that move narratives forward and the new type of NPC model can contribute to robust narrative design for contemporary game production. This chapter describes diverse theories and techniques including through narratological studies and computational and cognitive studies to aim at realization of a facial expression model with emotion in NPCs in the narrative genre of computer game. This chapter especially focuses on the theme of facial expression of a NPC associated to emotion. Such function can have a delicate impact on the continuous narrative flow and the relationship with other game characters and human players. (For example, from a narrative field of the editors of this book, in kabuki-dance Kyōganoko Musume Dōjōji, the change of the instantaneous facial expression of the heroin, particularly, from the beautiful and gentle face of a girl to dreadful face of a ghost or departed spirit, reveals the most important depth of the narrative). Exploring the problems of facial expression with emotion in a new narrative genre of computer games is an interesting challenge because computer games are a different and unique type of narrative from many previous narrative genres such as novels, poems, dramas, and movies. In Chapter 3, “Extension of Clinical/Psychological Approach Using Post Narratology: The Proposal of a Psychological Narratology and Review for Systematization,” Kai Seino and Shun Ishizaki present the concept of psychological narratology and methods to apply it to their computational linguistic analysis that they have been studying to provide the possibility of bridging the gap between AI-based approaches and narratological studies. Another characteristic of this chapter is that they deal with the clinical domain for supporting people with disabilities such as autism spectrum disorder (ASD) and their AI-based linguistic analysis using a data mining technique is also utilized to support these people. Therefore, their study introduced in this chapter probably shows a possible direction to AI-based narrative analysis from AI-based linguistic analysis. In this regard, this chapter accurately reflects the concept of the whole book, bridging the gap between narratology, AI, and cognitive science with narrative generation. The comparison between literary narratology or previous standard narratology and psychological xvii

Preface

narratology is also interesting from the theoretical angle or the standpoint of the book’s editors. Actually, previously many narratological researches have been conducted in the contents of AI, cognitive science, and psychology. However, narratology, narratological studies, or narratives in such contexts have been regarded as a means or example for serving main academic fields. From the viewpoint of this book’s concept, narratology, narratological studies, further narratives should be focused on as primary concepts to be considered. In particular, Seino and Ishizaki explicitly focus on the psychological aspect in narratology. Chapter 4, “Defining the Verbs for ‘Understanding and Interpretation’ of Japanese Sake,” by Hiroki Fukushima intends to define the verbs for linguistically describing and representing the taste of Japanese sake. (Although Fukushima does not refer to it, Bungakuron by Natsume Soseki (1997) introduced the sense of taste as an element for forming the content in literary works, such as novels, poems, and scripts.) A series of Fukushima’s studies that aim to linguistically describe and represent human vague senses of taste of Japanese sake are closely connected to the problems of narratology. The linguistic description and representation of the senses of taste of sake will be associated with problems regarding prominently personal and individual stories and narratives, and their generation or creation. In particular, depending on “a usage-based approach,” “encyclopedic semantics,” and “a sense-making theory,” this chapter selects eight verbs that are frequently used in the taste description of Japanese sake and defines their senses in a bottom-up way and abduction method based on a score indicating the strength of cooccurrence between terms. Furthermore, this chapter clarifies the definition of the verbs dependent on the terms “timeline,” “space,” “movement,” and “structure.” In Chapter 5, “300–500 Threshold of Context Memory and Assumption Experiment,” Nobuyoshi Harada presents an attempt of the problem regarding the threshold of context memory on numbers 300 through 500. In particular, this study focuses on the examination of the context threshold through an experiment of human brain mapping on an “auditory evoked magnetic field (AEF)” of a magnetic encephalogram. Harada introduces, as a hypothesis for the future discussion, several context presentations that realize the limitation of context memory in the process of continuous context presentation on numbers 500 and 2,000. As materials for the experiment, Harada uses many puns (Japanese dajares) generated by a computer program, and this research is related to the problems of human memory processing in cases where many dajares are continuously inputted into a human brain. In this regard, this hypothetic research may contribute to the consideration of a problem concerning narratological studies and cognitive science, especially brain science.

SECTION 2: NARRATIVE GENERATION SYSTEMS AND NARRATIVE CONTENT GENERATION Four chapters included in this section relatively directly focus on narrative generation systems and narrative content generation for various narrative genres, such as narrative, haiku, poem, and advertising. Chapter 6, “Bridging the Gap Between Narrative Generation Systems and Narrative Contents With Kabuki-Oriented Narratology and Watakushi Monogatari,” by Takashi Ogata presents an idea that bridges the gap between narrative generation based on a synthetic system framework including two types of subsystems—Integrated Narrative Generation System (INGS) and Geinō Information System (GIS), and narrative contents known as Watakushi Monogatari. Kabuki-oriented narratology is also used to bridge them by introducing four kabuki theories through the classical and the contemporary. This chapter xviii

Preface

comprises of the following three major themes to complete the aforementioned idea. The first theme deals with narrative phenomena such as the integration of INGS and GIS. This theme is connected to the second theme, which is presenting the narrative generation of kabuki or kabuki-oriented narratology, and the third theme, particularly, thoughts and design underlying Watakushi Monogatari as a collection of narrative content to be created by the author in the context of internal and external narrative generation to create and distribute narrative contents. INGS and GIS in the main topics of the first theme are respectively located as a single narrative generation and reception mechanism and a multiple narrative production and consumption mechanism. The framework proposed in this chapter aims to perform the creation of actual narrative work of Watakushi Monogatari showing the aspect of internal narrative generation and social development of Watakushi Monogatari corresponding to the aspect of external narrative generation through a kind of collaborative work of INGS and GIS, and further “I” (watakushi) as the author himself. The purpose of Chapter 7, “Haiku Generation From Narratological Perspective: A Circulation Between Haikus and Stories,” by Jumpei Ono and Takashi Ogata introduces the haiku generation method into our narrative generation. There are two typical haiku definitions: the sentence is composed of 17 mora and consists of three phrases, though there are other haikus not included in the definition. This chapter presents an experimental system that can generate haikus that have the aforementioned typical definitions. However, the extreme goal of the study is not presenting a program that can make haiku generation possibility to be compared to the human haiku creation ability, but incorporating such ability into a narrative generation system, specifically, the INGS introduced in Chapter 6. Concretely, the authors discuss a kind of circulative system between haikus and narratives (stories or plots), through which one or more haiku(s) are generated in accordance with a generated narrative and one or more narratives from the generated haiku(s). The continuous execution results in the circulative generation of haiku(s) and narrative(s) and incremental narrative change through haiku generation. Hence, the proposed system framework is regarded as an alternative of narrative generation in the authors’ narrative generation study. In Chapter 8, “On Logical Literary Work Generation and More,” Akinori Abe describes the possibility and problems of computational literary work generation, such as waka, a type of Japanese classical short song with fixed form of 5, 7, 5, 7, 7 (characters), poems, and stories, largely based on the concept of intertextuality by Julia Kristeva who is a French narratologist. Although Abe’s previous approach to narrative generation was based on a relatively simple method of using a random and disconnected combination, so-called “cut-up,” of selected words, phrases, and letters, this chapter presents a more intelligent AI-based method. In particular, Abe introduces techniques of abduction into logical literary work generation to maintain the logical consistency and coherency of generated works to aim to bridge the narratological method to narrative generation systems in a broad sense. In contrast with this abduction-based logical approach, Abe indicates important factors in narrative or literary works of very cognitive aspects, such as beautiful, sentimental, and valuable representation, and discusses how to add such cognitive aspects to the logical generation of literary works. From the perspective of the editors of this book, this chapter by Abe will contribute to support actual narrative or literary works through various operations for narrative generation functions. Chapter 9, “An Analysis of Cognitive Elements and Effects of Video Commercials: Toward Effective Indexing and Video Production,” by Yoji Kawamura is a continued report of the advertising system that generates the story-based sequential collection according to still images using a prepared database of fragmentary still images of a specific product type. This chapter analyzes cognitive elements and effects in advertising commercials to produce rich and attractive advertising works. In particular, Kawamura xix

Preface

proposes a method to organically combine among video shots based on still images in the database to provide consistency or coherence to the collection using a kind of semantic network. The image database allows the attaching of records of the time and place of shooting and features of the persons as indexes. From the theme of this entire book, first, advertising works also correspond to a type of narrative genre, and this study deals with advertising narrative stories referring to management and marketing theories and image theories as an interdisciplinary narrative study or narrative generation study. Furthermore, this chapter focuses on the problem of still image indexing based on cognitive scientific effectiveness for advertising story generation.

SECTION 3: SOCIAL PERSPECTIVES OF NARRATOLOGICAL AND NARRATIVE GENERATION METHODS This section includes three chapters on the relationship between social perspectives of narratology or narratological methods and narrative generation. Chapter 10, “Narrative Generation Using Psychological Value Variables: Probabilistic Model of Language Expressions and Values,” by Yasuo Tanida designs a systematic method to rewrite a story generated by a story generation system based on “psychological value variables” for readers’ positive evaluation of the future implementation of a system for realizing the proposed design. The concept of psychological value variables is an original idea by the author, and each person comprehends and interprets the world and acts based on the comprehension and interpretation according to the difference in psychological values. The extreme goal by Tanida is probably to utilize the theory of psychological value variables to support and accelerate marketing activities of companies. The proposed study of this chapter is located in the flow to this goal and extends his previous linguistic analysis based on words of people or customers to narratological analysis, and further toward the analysis through narrative generation. Regarding AI-based methods, Tanida mainly adopts a probabilistic language expression model similar to the language analysis appearing in Chapter 3. Through Chapter 11, “Effect of External Activate Factors Serving as Clue: Creating New Products or Services Ideas With Storytelling,” Jun Nakamura is interested in the problem of the effect of storytelling for the process of creating new products or services in the context of companies. To approach to this problem, Nakamura demonstrates an experiment to compare the effect of external activate factors such as objective advice and clues by consultants and outside directors at a graduate lecture. In particular, this experiment clarifies the effect of storytelling by the external activate factor for the emergence of abduction and new rules. This chapter’s implication to the concept of this book will be stated with Chapter 12 in the next explanation. Chapter 12, “Design of Narrative Creation in Innovation: ‘Signature Story’ and Two Types of Pivots,” by Akimitsu Hirota treats the topic of narrative creation in innovation in business organizations. Hirota previously analyzed storytelling through trial and error in the development of market-creating products. However, this chapter further pursues the stories of trial and error of the developer and marketer during the development of a market-creating product, and in it, he especially labels such stories “innovation narratives.” In particular, this chapter clarifies the creating process of an innovation narrative from the angle of an organizational dialog based on two types of pivots, including open pivots and closed pivots. The importance of stories and narratives has been pointed out in the world of business administration studies, including marketing, from the perspective that they supplement and structuralize the knowledge xx

Preface

type of analysis and measuring. Chapters 11 and 12 commonly approach “organizational narratology” as a narratological genre that aims to deal with a variety of narratological phenomena seen in social organizations as narratives or stories, or to interpret diverse organizational phenomena as narratives or stories. These chapters indicate that organizational narratives and stories play important roles for the development of new products and services and enhance organizational innovation as they function as a kind of control knowledge for actually creating new rules and innovative stories. The concept of this book also discusses the connection between management knowledge and narratological knowledge in Chapters 11 and 12. Takashi Ogata Iwate Prefectural University, Japan Jumpei Ono Aomori University, Japan

REFERENCES Natsume, S. (2007). Bungaku ron (I, II) [Literary theory (I, II)]. Iwanami Shoten. (Original work published 1907) Ogata, T. (2020a). Toward an integrated approach to narrative generation: Emerging research and opportunities. IGI Global. doi:10.4018/978-1-5225-9693-6 Ogata, T. (2020b). Internal and external narrative generation based on post-narratology: Emerging research and opportunities. IGI Global. doi:10.4018/978-1-5225-9943-2 Ogata, T., & Akimoto, T. (Eds.). (2016). Computational and cognitive approaches to narratology. IGI Global. doi:10.4018/978-1-5225-0432-0 Ogata, T., & Akimoto, T. (Eds.). (2019). Post-narratology through computational and cognitive approaches. IGI Global. doi:10.4018/978-1-5225-7979-3 Ogata, T., & Asakawa, S. (Eds.). (2018). Content generation through narrative communication and simulation. IGI Global. doi:10.4018/978-1-5225-4775-4 Ogata, T., & Kanai, A. (2010). Monogatariron no jōhōgaku josetsu: Monogatari seisei no shisō to gijutsu wo megutte [An introduction to informatics of narratology: Around the thoughts and technologies of narrative generation]. Gakubunsha. Ogata, T., Kawamura, Y., & Kanai, A. (2018). Jōhō monogatariron: Jinkōchinō, ninchi, shakai-katei to monogatari seisei [Informational narratology: Artificial intelligence/cognition/social process and narrative generation]. Hakutō Shobō.

xxi

Section 1

Cognitive Approaches to Narratology and Narrative Generation

1

Chapter 1

Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation Akihito Kanai Hosei University, Japan

ABSTRACT This chapter surveys, discusses, and explores the entire concept of visual narrative structure, cognition, and generation from continuity-based to discontinuity-based perspectives. The model of visual narrative structure, including presentation and meaning, is expanded to explain the cognitive transition based on the rhetorical transition techniques and the rhetorical cutting techniques. The classification of the visual narrative structure, including rhetorical transition techniques and rhetorical cutting techniques, is useful for narrative simulation for discussing and exploring the entire visual narrative concept and generation. The rhetorical cutting techniques and the rhetorical transition techniques can reveal various cognitive effects such as reality effects and nostalgia effects, including difficulty. The determinacy-based narrative, the indeterminacy, the diversity, and the ambiguity on narrative can bridge the gaps between cognitive science, artificial intelligence, and narratology.

INTRODUCTION This chapter surveys, discusses and explores the entire concept of visual narrative structure, cognition and generation from continuity based to discontinuity based. Visual narratives such as films are analyzed in terms of the cognitive transition, the cognitive processing change of a narrative focus, in this chapter. The question of how people comprehend visual narratives such as films is difficult and complicated (Loschky et al., 2020). A viewer tends to recognize a film as a visual narrative to understand the story (Cutting, Brunick, & Canden, 2012; Cutting & Iricinschi, 2014). Although viewers’ scene perception and event comprehension based visual narrative comprehension process and generation mechanism DOI: 10.4018/978-1-7998-4864-6.ch001

Copyright © 2021, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

are important questions for the field of cognitive science or artificial intelligence, the other aspects of narrative such as the mental difficulties on cognition are also important (Corner, 2018). The mental difficulties are intentionally present, caused by a visual narrative’s core feature. For example, the mental difficulty cognition from massive indeterminacy of place and time caused by the discontinuity editing can enhance the viewpoint diversity and the cognitive transition of the viewer (Kanai, 2018). Besides the determinacy and the continuity based narrative, the indeterminacy, the diversity, the ambiguity, and the discontinuity of narratives are important questions for the field of narratology. For example, Bakhtin (1984) argued the polyphonic theory of novels. The polyphonic theory emphasized the indeterminacy, the divergence, and the ambiguity. In recent years, Hasumi (2014) argued the indeterminacy and the ambiguity of novels such as Madame Bovary. Narrative is mainly discussed in the field of narratology. Theories on narratology make a strong distinction between story (fabula) and discourse (syuzhet) (e.g., Genette, 1972; Shklovsky, 1965). Story comprehension and meaning are not as important as discourse and rhetoric themselves in theories of narratology (Chatman, 1990). For the study of film, shots, editing, and film structures themselves are equally important. A film has not only a story and meaning structure but also visual and audio structures. Story comprehension is affected by a film’s structure. Furthermore, the cognition difficulties caused by story comprehension are also affected by the structure. The cognition difficulties resulting from the structure are related to the quality and the artistry of the film. Kanai (2016, 2019) defined the rhetorical cutting technique as rhetorical techniques for creating irrational cuts in a film based on discontinuity editing. Rhetorical cutting techniques generate irrational relationships or unfamiliar situations in film and create an alienation effect (Brecht & Willett, 1964). Rhetorical cutting techniques for narrative space and time in a film can cause a viewer’s cognitive transition from being event-based and story-driven to being presentation-based and rhetoric-driven processing without cognitive effects from the story. Furthermore, this chapter discusses rhetorical transition techniques, rhetorical techniques for creating plurality and diversity in a film with some story and continuity. Rhetorical transition techniques for narrative space and time in the film can cause the viewer’s cognitive transition from event-based story-driven to presentation-based and rhetoric-driven processing with cognitive effects from the story. Rhetorical transition techniques and cutting techniques can cause a viewer’s cognitive transition. The transition and the interaction between event-based, story-driven processing and presentation-based, rhetoric-driven processing are among the most important aspects of film cognition. Only the cognitive science approach with narratology can adequately analyze the transition and the interaction. Without a cognitive science approach, narratology cannot address them. In addition, a cognitive science approach without narratology cannot address cognition difficulties in the transition and the interaction. Furthermore, narratology with cognitive science based artificial intelligence can reach the transition and the interaction, including cognition difficulties, and can do several narrative simulations based on cognitive effects for content generation. Researchers such as Ogata and Asakawa (2018) have argued for techniques such as visual narrative generation through narrative simulation using artificial intelligence techniques. Narrative simulation can be a process of composing and designing various kinds of narrative rhetoric and of testing the cognitive effects of that narrative (Kanai, 2018). Many cognitive effects from a narrative result from the effects of the determinacy and the indeterminacy on the interaction between cognitive process, story, discourse, and rhetoric themselves. For example, the indeterminacy emerging from rhetorical cutting techniques in narrative space and time in a narrative film can cause a viewer to experience cognitive transition and 2

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

intensive cognitive effects. Based on the rhetorical cutting techniques used to generate indeterminacy with regards to time, space, and characters’ actions in continuous shots, viewers can feel less constrained with regards to story comprehension and can reset their viewpoints to subtle elements, such as details of individual shots. The cognitive transition caused by the indeterminacy of narrative film rhetoric can cause cognitive effects, such as reality or nostalgia effects relating to the actual world. Those effects are discussed in a later section of this chapter. Moreover, based on the rhetorical transition techniques used to generate the plurality, the diversity, and the ambiguity with regards to time, space, and characters’ actions in continuous shots, a viewer can still feel constrained with regards to story comprehension but can reset the viewpoint to subtle elements. Based on the indeterminacy, the diversity, and the ambiguity, the narrative simulation and generation mechanism can emphasize the whole rhetorical narrative aspects of films. In this chapter, visual narrative is analyzed and surveyed, especially from the perspective of the cognitive transition relate to the narrative story and rhetorical presentation, including both continuity and discontinuity. A model of visual narrative structure, including presentation and meaning provided by Cohn & Magliano (2020), is expanded to explain the cognitive transition based on the rhetorical transition techniques and the rhetorical cutting techniques. The classification of the visual narrative structure, including rhetorical transition techniques and rhetorical cutting techniques, is useful for narrative simulation for discussing and exploring the entire visual narrative concept and generation.

BACKGROUND Visual narrative cognition involves the simultaneous processing of many factors. Generally, a story and its cognition are important factors to narrative cognition. The viewer can comprehend the totality meaning and the story from continuous events (e.g., Avrahami & Kareev, 1994). The many previous approaches primarily focused on characters’ goal-directed actions (e.g., Jhale & Young, 2010). Recent focus has turned to the visual narrative structure (e.g., Cohn, 2013, 2014; Cohn & Magliano, 2020). Although theories of story grammar inspired by Propp’s (1968) theory (e.g., Rumelhart, 1975; Thorndyke, 1977) blurred the distinction between narrative (i.e., presentation) and events (meaning), Cohn & Magliano (2020) provided a model of visual narrative structure that divides visual narrative systems across a modality of expression (i.e., graphics), organized using combinatorial systems (grammar), which express conceptual information (meaning), and it does so across both the unit and sequence levels (Figure 1). Aspects of modality involve both the unit (i.e., graphic structure) and sequence (layout) levels. The combinatorial systems organize form and meaning for both units (morphological structure) and sequences (narrative structure). Aspects of the meaning involve conceptual and event structures. Contrary to the myth of transparency based on the ecological theory of perception such as Gibson (1979), that visual narrative requires basic perceptual or event cognition alone, visual narrative systems differ in many ways and have opaque aspects, even when setting aside the modality differences of filmic and drawn narratives. For example, comics and illustrated stories vary in many conventions (such as layout, morphology, and narrative structure), despite both conveying meaning across sequential images. Figure 1 mainly features drawn visual narratives, but dynamic visual narratives appear as film, which uses actual percepts instead of drawings. In addition to the determinacy or the transparency created by the stories’ or characters’ goal-directed actions and conceptual or event meanings, the difficulties or the opaqueness from the indeterminacy or 3

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

Figure 1. A model of the structures involved in visual narratives from Cohn & Magliano (2020)

the plurality created by the visual elements of rhetoric are essential issues of film cognition (Kanai, 2017). Based on the indeterminacy or the plurality, the film can emphasize the rhetorical aspects of presentation. The aspects of rhetoric can cause cognitive effects that do not arise only through comprehending a story and an overall meaning based on continuities but which instead arise through the audiovisual situation or the cognitive transition itself. Previous studies argued for various continuities and cognition (e.g., Berliner & Cohen 2011; Magliano & Zacks, 2011; Smith 2012). The continuity means that the flow from shot to shot in the film is smooth. The continuity is an illusion created in the mind of a viewer (Smith, 2011). Thanks to spatial continuity, a viewer can understand characters’ locations. Moreover, continuity editing supports and sustains temporal manipulation as well. Film continuity is managed in the service of telling a story and thus creating a sense of continuity and discontinuity of events within the story world. In this perspective, film discontinuity is regarded as an obstacle to story comprehension. Scene perception and event comprehension theory is applied to continuity based visual narratives (Loschky et al., 2020). In the other perspective, the discontinuities or the plurality in a film themselves are important for the quality and artistry. Those aspects can cause intensive cognitive effects from the difficulties cognition. Previous studies of film cognition showed how viewers understood event-based stories, because the rhetoric of a film is generally accomplished by telling some story through explicit or implicit events. However, a film is not always created with the purpose of representing a story’s events: the rhetoric of the film may have a purpose that is not to represent an event-based story. Philosopher Deleuze (1990) said that after World War II, films had stopped being “movement-images” and instead had become “time-images.” There are films for which it is necessary to realize a film’s structure is independent of the story’s events. In such a film, a purely audiovisual situation emerges, and the point where force is

4

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

applied is placed in the description itself, in addition to the actions and goals of the characters in events. Here, a film does not always have an overall meaning; and the subtle element of each shot becomes important. With regards to the purpose of the rhetoric of the film, there is a difference between the movement-image and the time-image, as Deleuze (1990) said. Then, the parts where emphasis is placed on the elements of the film structure are different. Deleuze (1985) argued that irrational cuts in the film are related to the time-image. Furthermore, documentary film is not accomplished by the purpose of telling some story. For example, films directed by Frederick Wiseman have no main characters or events. Films such as Ex Libris: The New York Public Library (2017) directed by Wiseman consists of everything about the library. In this case, event structures and overall story are not so important. Ex Libris: The New York Public Library has a consistent narrative based on the film’s associations between the objects, the moods, and the events rather than on the continuity of the events. In the other case, art film is not accomplished by the determinacy of the event or conceptual structure. For example, films directed by Jean-Luc Godard use visual effects for unclear meanings. Films such as The Image Book (2018) directed by Godard consist of images emphasizing indeterminacy and plurality. Slow cinema or contemplative cinema using recent digital cameras such as The Woman Who Left (2016) directed by Lav Diaz and Stray Dogs (2013) directed by Tsai Ming-liang feature long continuous shots. Çağlayan (2016) discussed that slow cinema can transform boredom into an aesthetically rewarding experience through its celebration of dead time and foregrounding dramatic ambiguity. The long take performs productive functions of boredom and mind wandering by dedramatizing narrative structures. The long, continuous shot can amplify the ambiguity of the meaning of presentation. In order to discuss and explore the entire visual narrative concept such as film, not only the viewers’ scene perception and event comprehension based meaning creation mechanism, the difficulty, the indeterminacy, the diversity, and the ambiguity aspects of visual narrative caused by the cutting and the rhetorical transition techniques for the cognitive transition are also to be considered.

RHETORICAL CUTTING TECHNIQUES Based on the rhetorical cutting techniques to make the indeterminacy from the irrational relationships of time or character in continuous shots, viewers can relax constraints on story comprehension based on the difficulty of cognition and reset their viewpoints to better view subtle elements, such as individual shots details. The rhetorical cutting techniques for narrative space and time in the film can cause difficulty cognition and the viewer’s cognitive transition from event-based, story-driven determinacy to presentation-based, rhetoric-driven indeterminacy. Based on the cognitive transition, cognitive effects such as intensive reality effects can emerge. Cutting techniques can be used to create indeterminacy between (a) the rhetorical elements of two shots (Cutting 1); (b) events and images (Cutting 2); or (c) sounds and images (Cutting 3; Kanai & Ogata, 2004a). The indeterminacy created through the cutting techniques described above is used in many artistic films, television advertisements, and music videos. This chapter primarily discusses visual aspects such as Cutting 1 and Cutting 2. For many films, Cutting 1 can make the indeterminacy between two events merge in continuous shots. Cutting 2 can make indeterminacy between presentation and meaning. Figure 2 is derived from Figure 1; it illustrates the 5

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

role of the cutting techniques. Film directors such as Wiseman only use Cutting 1. On the other hand, film directors such as Godard use both Cutting 1 and Cutting 2. Additionally, the sudden appearance of the long continuous shot can involve Cutting 2. Figure 2. A model of the structures involved in visual narratives with the cutting techniques

For example, The Image Book (2018) directed by Godard used various cutting techniques. Some continuous shots of the film had no event relationship (Cutting 1). The Image Book consisted of many old films’ retouched shots, which created a discontinuity or an ambiguity regarding the events and images (Cutting 2). To cause intensive cognitive effects, the discontinuity of events must be emphasized in the rhetoric. A mix of identical and different elements in continuous shots using the rhetorical cutting techniques is important for the cognitive effects of the rhetoric of the film, such as the difficulty. Rhetorical cutting techniques resemble memories of the past (Kanai, 2019). Humans cannot remember one consistent story from the past. Especially, disaster memories may consist of many fragments like the rhetorical cutting techniques with mental difficulty. Films about disasters also may consist of many fragments with the rhetorical cutting techniques for the intentional mental difficulty. Disaster narratives not only consist of consistent stories, memories, and meanings. Conversely, in order to appropriately relate to the disaster narrative film with the rhetorical cutting techniques, viewers must relax their constraints in terms of the need to comprehend continuous events and can feel the strength and the pain of the disaster.

6

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

RHETORICAL TRANSITION TECHNIQUES For the rhetorical cutting techniques, the discontinuity of events is emphasized. On the contrary, the rhetorical transition techniques keep the continuity of events. The rhetorical transition technique is based on the continuity editing, including the plurality, the diversity, and the ambiguity other than the meaning of consistent events. Rhetorical transition techniques can change the flow of cognitive processing for narrative. The changing flow can be reduced to two categories. For many films, Transition 1 can make the plurality, the diversity, and the ambiguity between two events emerged in continuous shots. Transition 2 can make the plurality, the diversity, and the ambiguity between presentation and meaning. Figure 3 is derived from Figure 1; it illustrates the role of the transition techniques. The rhetorical transition techniques for narrative space and time in the film can cause the viewer’s cognitive transition from event-based and story-driven to presentation-based and rhetoric-driven processing with cognitive effects from the story. In this case, event-based, story-driven processing is not emphasized. Figure 3. A model of the structures involved in visual narratives with the transition techniques

Characters’ details, environment, nearby objects, elements, cognition, and rhetorical explanation can be the rhetorical transition techniques, which can generate the plurality and the diversity with regards to time, space, characters’ actions, and meaning in continuous shots. For example, the opening scene from Touch of Evil (1958) directed by Orson Wells consists of a single continuous long shot lasting three minutes and twelve seconds. Hutson et al. (2017) investigated the levels of attentional synchrony were less in the continuous long shot of Touch of Evil than in the highly edited sequence. Touch of Evil

7

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

opens on a close-up of someone setting a time bomb. The time bomb is then placed into the trunk of a car, after which a couple unknowingly gets into the car and drives off, as the camera follows them. Then, a second couple walking down the street is introduced, and the camera begins to follow them with the car always lurking around them. The relationship between the time bomb and the second couple can change the flow of cognitive processing for narrative structure (Transition 1). Touch of Evil is lacking cuts but it includes numerous objects from a Mexico–US border town. Characters’ environments and nearby objects are transition techniques to reset viewers’ viewpoints from event understanding to the plurality, the diversity, and the ambiguity of presentation (Transition 2). Generally, the heavy-handed context manipulation can change the nature of viewers’ event models of the film. Despite large differences in viewers’ understanding and event models of the film, there were small differences in attentional selection. Using the attentional selection nature, the transition techniques can reset the viewpoint to subtle elements. For cognitive science, attentional synchrony itself is an important matter. The diverse attention can cause the viewer’s cognitive transition from story-driven to rhetoric-driven processing with cognitive effects from the story, which is an important matter for narratology. For another example, Pickpocket (1959) directed by Robert Bresson frequently used close-up shots of hands picking pockets. The close-up shots of hands can be the rhetorical transition technique. First, using the attentional selection nature, the close-up shots of hands can reset the viewers’ viewpoints from the whole event structure to picking pockets events only (Transition 1). Second, the close-up shots of hands can reset the viewers’ viewpoints from pocket-picking events to presenting the characters’ hands themselves (Transition 2). Bordwell (2007) discussed two forms of editing: analytical editing and constructive editing. Analytical editing is based on the continuity of events. On the other hand, Bordwell (2012) argued for constructing editing based on Pickpocket. Bresson adhered to the idea that a film is only a sum of the elements that are necessary for visual narrative, such as the shots of the hands. Rhetorical transition techniques can be the components of constructive editing. Other than the close-up shots of hands, the close-up shots of paper money or a fake bill used in Money (1983) directed by Bresson also can be rhetorical transition techniques.

COGNITIVE EFFECTS FROM RHETORICAL CUTTING AND TRANSITION TECHNIQUES Rhetorical cutting techniques and rhetorical transition techniques can emerge various cognitive effects such as reality effects and nostalgia effects, including the difficulty, which can bridge the gaps between cognitive science, artificial intelligence, and narratology. Previous studies on film cognition primarily discussed the first type of reality effects related to the condition of being natural and less deviated with regard to movements (Visch & Tan, 2009; Rooney et al., 2012). The first type of reality effects are especially realized by Transition 2. The second type of reality effects, related to the unfamiliar situations, were argued by the Russian Formalists (Shklovsky, 1965). The indeterminacy from the rhetorical cutting techniques, especially Cutting 2, caused the second type of reality effects (Kanai, 2006). The indeterminacy, the plurality, and diversity derived from the rhetorical cutting or transition techniques creates new realities, nostalgia effects, or meanings related to unknown memories and affects that do not arise through event comprehension alone (Kanai, 2018). Particularly, the indeterminacy can emerge through the sudden appearance of an event unit (Cutting 1). In this case, a viewer sometimes 8

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

cannot apply the determinacy from the consistent events. The scene perception and event comprehension theory cannot be applied to indeterminacy based visual narratives. Therefore, attention to the audiovisual situation itself or the subtle elements is enhanced (Cutting 2). This occurs because the indeterminacy generates new cognitive effects related to unknown memories and affects that do not arise through event comprehension and meaning. For visual narratives, nostalgia effects are especially important. The connections to the past, the memories, and the present always involve the indeterminacy regarding the event’s meaning and the ambiguity regarding the place and time. The rhetorical cutting and the transition techniques and nostalgia effects are more closely related to each other. In general, nostalgia refers to a longing for the past. Past images in a film not associated with a specific event can evoke many kinds of nostalgia based on the viewer’s cognition. Nostalgia has mainly been argued by researchers like Stern (1992) as a longing for a good but lost personal and historical past. Holbrook and Schindler (1991) defined it as a preference for objects (people, places, or things) that were more common when one was younger. Nostalgia can emerge from various different types of presentation. Some nostalgia is evoked from the loss of a place’s good past. Other nostalgia is evoked from the unforgettably bad past of a place. Furthermore, nostalgia may emerge suddenly with the determinacy of the place and the time, and the indeterminacy of rhetorical elements such as irrational shots. Indeterminacy based nostalgia does not always follow comfortable cognition. To manage various kinds of nostalgia, Naito and Kanai (2012a, 2012b) redefined nostalgia as emotions arising from the rise and fall in other times in the place. Kanai (2016, 2019), classified the rhetoric of a film into four kinds of nostalgia—including non-nostalgia. Kanai focuses on three kinds of nostalgias regarding a certain place, as suggested by Davis (1979): “simple nostalgia,” “reflexive nostalgia,” and “interpreted nostalgia.” Simple nostalgia emerges from the good past with the rhetorical transition techniques. Reflexive nostalgia emerges from both the good and bad past, with the rhetorical cutting and transition techniques. Interpreted nostalgia emerges from the reality or the uncertainty of the past with the rhetorical cutting techniques. Simple nostalgia emerges from ambiguous affection for the past of a place relating to some events based on the rhetorical transition techniques (Transition 1 and Transition 2). By contrast, reflexive and interpreted nostalgia emerge from the reality and the uncertainty about the past of a place as related to unfamiliar situations, with the cognitive transition and alienation effects based on the rhetorical cutting techniques (Cutting 1 and Cutting 2). The editing, recording, and photography of a film are not always subordinate to the determinacy of the overall events and meanings. Nostalgia can emerge from every film. Nonetheless, some film directors, whose purpose is to create timeless films located in unknown places, do not want to generate nostalgia effects for viewers. Nonnostalgia based narratives can be composed using the strategy for the nostalgia effects. Moreover, every film has not only past aspects but also present aspects, as the film may be viewed in the present time. To emphasize the present aspects of films, the strategy for the non-nostalgia cognitive effects must be involved. As stated above, the focuses of the rhetorical cutting and the transition techniques are on presentation of the narrative, the mood and the object associations, or the reality related to the unfamiliar situations; rather than on the events and the meaning in the story. The indeterminacy and the ambiguity of such associations and situations cause cognitive effects that do not arise in the process of comprehending continuous events only. Kanai (2001a, 2001b) indicated that when a viewer can relax his or her constraint with regard to the need to comprehend a story, this strengthens those cognitive effects, which do not arise in the process of event comprehension. 9

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

NARRATIVE SIMULATION Narrative simulation, including the strategy for the rhetorical cutting and the transition techniques, can broaden the potential for narrative generation—especially from the perspective of the indeterminacy and the ambiguity. Although event-based plots and characters’ goal-directed actions are essential for visual narrative cognition and generation, events comprehension and meanings can also constrain this (Ogata & Kanai, 2010). The rhetorical cutting and the transition techniques, which can relax constraint with regards to the event comprehension and meaning, can be the main components of visual narrative simulation. The author had proposed a film rhetoric composition system based narrative simulation (Kanai & Ogata, 2004b; Kanai, 2018). The system can be used for narrative simulation with the rhetorical cutting techniques. The authors revised the system, which can compose nostalgia-based narrative rhetoric with the cutting and transition techniques. The user can decide which the rhetorical cutting and transition techniques to use for making discontinuities and continuities. The system then retrieves the shots from the database that fits the chosen strategy: Cutting 1, Cutting 2, Transition 1, and Transition 2. The shots are edited as a sequence of shots and outputted. By using this computational approach, we could make many different films from the same shots and use them to study the nature of the visual narrative. Computational narrative simulation can be argued as being the choice of the cutting and the transition techniques used as a strategy in order for the viewer’s cognitive processes. The generated films can reveal both the event-based story and the broad presentation aspects of the rhetoric, narrative, and cognition of the past and the film. Without using event structures and goal-oriented characters, the nostalgia-based visual narrative could be mainly simulated. The montage theory of Eisenstein (1942, 1949) can be used as a tool for more controlled and expanded narrative simulation with the transition techniques. Kanai & Kodama (2010) created 16 kinds of oneminute films using 90-second sequences of 18 shots each from The Passion of Joan of Arc (1928) directed by Carl Th. Dreyer, using the narrative simulation mechanism, and investigated the difference between the rhetorical transition aspect and the rhetorical cutting aspect in the films by use of a psychological experiment. The Passion of Joan of Arc is one of the most famous of art films containing no spoken dialogue. The shots from The Passion of Joan of Arc are famous for their usage of close-up face shots. The close-up face shots of Joan of Arc themselves can be a transition technique (Transition 1), so that quality makes them suitable for serving as the source of narrative simulation. In this case, the close-up face cuts from the sequence can make the continuity and the determinacy of the character (i.e., Joan) and the location (the cell; Transition 2). By contrast, the irrationally cut films from the sequence can make the discontinuity for the indeterminacy of the various characters or the various locations (Cutting 1). Also, some characters’ shots with a slanted angle can make the indeterminacy between presentation and meaning (Cutting 2). The Kanai & Kodama conducted the experiment based on the hypothesis “the rhetorical cutting aspect can enhance the viewpoint diversity of the viewer” and “the rhetorical transition aspect can enhance the viewer’s immersion into the story, and the viewer’s impression of character.” The results somewhat proved the hypothesis. The cutting aspect for the indeterminacy-based montage can enhance viewer viewpoint diversity. Additionally, the transition aspect for the determinacy-based montage can enhance the viewer’s immersion into continuous events.

10

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

FUTURE RESEARCH DIRECTIONS In this chapter, a model of visual narrative structure provided by Cohn & Magliano (2020) is expanded for explaining the cognitive transition caused by the rhetorical transition techniques and the rhetorical cutting techniques. An expanded visual narrative structure can be used for doing narrative simulation. Future research directions of the main themes in this chapter would be as follows. One hypothesis to be discussed more is that many of the cognitive processes that support understanding in literature, also support the processing of visual narratives such as film. Only a few studies have directly explored the extent to which processes are common across textual, filmic, or drawn narratives. Filmic narrative only is mainly discussed in this chapter. The rhetorical transition techniques can be the components of constructive editing. The details of the relationship between rhetorical transition techniques and constructive editing can be discussed more. This chapter does not discuss the difference between fiction and reality, for example, the documentary. All visual narratives such as films include both fiction and reality aspects. Film continuity for both fiction and reality are illusions created in viewers’ minds. Both fiction and reality-based visual narratives can evoke nostalgia effects. The question of the difference between fiction and reality effects is important for both cognitive science and narratology. Emotional and affective aspects other than nostalgia and difficulty are also important. Emotional and affective cognition emerge from both event-based, story-driven processing and presentation-based, rhetoric-driven processing. For example, the beauty and ugliness effects of visual narrative mainly emerged from presentation-based, rhetoric-driven processing. Also, the beauty and ugliness effects of visual narrative affect and relax event-based, story-driven processing. Moreover, the beauty and ugliness of the visual narrative can involve reverse-transition techniques, which have not been discussed in this chapter. Viewers’ scene perception and event comprehension based visual narrative comprehension process are still important matters for cognitive science and narratology. The rhetorical reverse-transition techniques for narrative space and time in the film can cause the viewer’s cognitive transition from being event-based and story-driven through being presentation-based and rhetoric-driven back to being event-based story processing again. In this cycle based on rhetorical reverse-transition techniques, viewers’ scene perception and event comprehension-based visual narrative comprehension process are enhanced. The rhetorical reverse-transition techniques can be a key concept for how people comprehend visual narrative. Finally, auditory information is vital to film cognition and making (Bordwell, 1985). Cutting and transition aspects of auditory information can be focused more.

CONCLUSION In order to discuss and explore the entire narrative concept and generation, not only the viewers’ scene perception and event comprehension based visual narrative comprehension and generation mechanism, the other aspects caused by the rhetorical cutting techniques and the rhetorical transition techniques are also to be considered. The transition and the interaction between event-based, story-driven processing and presentationbased, rhetoric-driven processing with the rhetorical cutting techniques and the rhetorical transition

11

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

techniques are one of the most important aspects of film cognition. The transition and the interaction are only analyzed by a cognitive science approach with narratology. Narratology without a cognitive science approach cannot address the transition and interaction. In addition, a cognitive science approach without narratology cannot reach the difficulties cognition in the transition and the interaction. Furthermore, narratology with a cognitive science based artificial intelligence can only address the transition and the interaction, including cognition difficulties, and can do several narrative simulations based on cognitive effects for content generation. Although the scene perception and event comprehension based visual narrative comprehension is an important question for the field of cognitive science or artificial intelligence, the other aspects of narrative are also important. The rhetorical cutting techniques and the rhetorical transition techniques can emerge various cognitive effects such as reality effects and nostalgia effects, including difficulty. Not only the determinacy based narrative, the indeterminacy, the diversity, and the ambiguity on narrative can bridge the gaps between cognitive science, artificial intelligence, and narratology.

REFERENCES Avrahami, J., & Kareev, Y. (1994). The emergence of events. Cognition, 53(3), 239–261. doi:10.1016/00100277(94)90050-7 PMID:7842635 Bakhtin, M. (1984). Problems of Dostoevsky’s poetics (C. Emerson, Trans.). University of Minnesota Press. (Original work published 1963) doi:10.5749/j.ctt22727z1 Berliner, T., & Cohen, D. J. (2011). The illusion of continuity: Active perception and the classical editing system. Journal of Film and Video, 631(1), 44–63. doi:10.5406/jfilmvideo.63.1.0044 Bordwell, D. (1985). Narration in the fiction film. University of Wisconsin Press. Bordwell, D. (2007). Poetics of cinema. Routledge. Bordwell, D. (2012). Constructive editing in Robert Bresson’s Pickpocket [Video]. Vimeo. https://vimeo. com/52312154 Brecht, B., & Willett, J. (1964). Brecht on theatre: The development of an aesthetic. Hill and Wang. Bresson, R. (1959). Pickpocket [Film]. A. Delahaie (Producer). France: Compagnie Cinématographique de France. Bresson, R. (1983). Money [Film]. J. M. Henchoz & D.T. Plantier (Producer). Çağlayan, E. (2016). The aesthetics of boredom: Slow cinema and the virtues of the long take in once upon a time in Anatolia. Projections, 10(1), 63–85. doi:10.3167/proj.2016.100108 Chatman, S. (1990). Coming to terms: The rhetoric of narrative in fiction and film. Cornell University Press. Cohn, N. (2013). Visual narrative structure. Cognitive Science, 34(3), 413–452. doi:10.1111/cogs.12016 PMID:23163777

12

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

Cohn, N. (2014). You’re a good structure, Charlie Brown: The distribution of narrative categories in comic strips. Cognitive Science, 38(7), 1317–1359. doi:10.1111/cogs.12116 PMID:24646175 Cohn, N., & Magliano, J. P. (2020). Editors’ introduction and review: visual narrative research: an emerging field in cognitive science. Topics in Cognitive Science, 12(1), 197–223. doi:10.1111/tops.12473 PMID:31865641 Corner, J. (2018). Documentary spectatorship and the navigation of “difficulty.”. In C. Brylla & M. Kramer (Eds.), Cognitive theory and documentary film (pp. 59–73). Palgrave Macmillan. doi:10.1007/978-3319-90332-3_4 Cutting, J. E., Brunick, K. L., & Canden, A. (2012). Perceiving event dynamics and parsing Hollywood films. Journal of Experimental Psychology, 38(6), 1476–1490. PMID:22449126 Cutting, J. E., & Iricinschi, C. (2014). Re-presentations of space in Hollywood movies: An event-indexing analysis. Cognitive Science, 35, 1–23. PMID:25087776 Davis, F. (1979). Yearning for yesterday: A sociology of nostalgia. Free Press. Deleuze, G. (1985). L’ Image-temps. Les Editions de Minuit. Deleuze, G. (1990). Pourparlers. Les Editions de Minuit. Diaz, L. (2016). The woman who left [Film]. L. Diaz (Producer). Dreyer, C. T. (1928). The passion of Joan of Arc [Film]. France: Société générale des films. Eisenstein, S. M. (1942). The film sense (J. Leyda, Trans.). Harcourt Brace. Eisenstein, S. M. (1949). Film form (J. Leyda, Trans.). Harcourt Brace. Genette, G. (1972). Discourse du recit: Essai de method. [Narrative discourse: An essay in method] In Figures III. Seuil. Gibson, J. J. (1979). The ecological approach to visual perception. Houghton Mifflin. Godard, J. L. (2018). The image book [Film]. Hasumi, S. (2014). Bovary fujin ron [A Theory of “Madame Bovary”]. Chikumashobō. Holbrook, M. B., & Schindler, R. M. (1991). Echoes of the dear departed past: Some work in progress on nostalgia. In R. H. Holman & M. R. Solomon (Eds.), Advances in Consumer Research, 18 (pp. 330–333). Association for Consumer Research. Hutson, J. P., Smith, T. J., Magliano, J. P., & Loschky, L. C. (2017). What is the role of the film viewer? The effects of narrative comprehension and viewing task on gaze control in film. Cognitive Research: Principles and Implications, 2(1), 46. doi:10.118641235-017-0080-5 PMID:29214207 Jhale, A., & Young, M. R. (2010). Cinematic visual discourse: Representation, generation, and evaluation. IEEE Transactions on Computational Intelligence and AI in Games, 2(2), 69–81. doi:10.1109/ TCIAIG.2010.2046486

13

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

Kanai, A. (2001a). The story as a constraint on film. Proceedings of the Third International Conference on Cognitive Science, 503–506. Kanai, A. (2001b). Cognitive process model toward the rhetoric of the film. Cognitive Studies, 8(2), 139–150. Kanai, A. (2006). Reality of unfamiliar situations on moving image. Proceedings of 19th Congress of the International Association of Empirical Aesthetics, 800–803. Kanai, A. (2016). Non-story, nostalgia, and film cognition: Nostalgia-based narrative rhetoric composition. In T. Ogata & T. Akimoto (Eds.), Computational and Cognitive Approaches to Narratology (pp. 376–390). Hershey, PA: Information Science Reference (IGI Global). Kanai, A. (2018). Narrative simulation for film rhetoric composition with or without story and nostalgia effects. In T. Ogata & S. Asakawa (Eds.), Content generation through narrative communication and simulation (pp. 148–161). Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/9781-5225-4775-4.ch003 Kanai, A. (2019). Past, non-story narrative film, and nostalgia. In T. Ogata, & T. Akimoto (Eds.), Postnarratology through computational and cognitive approaches (pp. 277–291). Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/978-1-5225-7979-3.ch006 Kanai, A., & Kodama, K. (2010). Film edit design: Story and irrational cutting. Cognitive Studies, 17(3), 444–458. Kanai, A., & Ogata, T. (2004a). Non-story processing on the film rhetoric composition system. Proceedings of 18th Congress of the International Association of Empirical Aesthetics, 433–436. Kanai, A., & Ogata, T. (2004b). Aspect of non-story processing and film rhetoric composition in the narrative generation mechanism. Proceedings of the Ninth International Symposium on Artificial Life and Robotics, 162–165. Loschky, L. C., Larson, A. M., Smith, J. T., & Magliano, J. P. (2020). The scene perception and event comprehension theory (SPECT) applied to visual narratives. Topics in Cognitive Science, 12(1), 311–351. doi:10.1111/tops.12455 PMID:31486277 Magliano, J. P., & Zacks, J. M. (2011). The impact of continuity editing in narrative film on event segmentation. Cognitive Science, 35(8), 1489–1517. doi:10.1111/j.1551-6709.2011.01202.x PMID:21972849 Naito, Y., & Kanai, A. (2012a). The narrative structure of nostalgia cognition and film. In Proceedings of the International Conference on Cognitive Science ‘12. Academic Press. Naito, Y., & Kanai, A. (2012b). Nostalgia and film creation. In Proceedings of the 26th Annual Conference of the Japanese Society for Artificial Intelligence (1N1-OS-1a-2). Yamaguchi, Japan: Academic Press. Ogata, T., & Asakawa, S. (Eds.). (2018). Content generation through narrative communication and simulation. Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/978-1-5225-4775-4

14

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

Ogata, T., & Kanai, A. (2010). Monogatariron no jyohogaku jyosetsu: Monogatariseisei no shiso to gijyutsu wo megutte [An introduction to informatics of narratology: Towards the thoughts and technologies of narrative generation]. Gakubun-sha. Propp, V. Y. (1968). Morphology of the Folktale (L. Scott, Trans.). University of Texas Press. (Original work published 1928) Rooney, R., Benson, C., & Hennessy, E. (2012). The apparent reality of movies and emotional arousal: A study using physiological and self-report measures. Poetics, 40(5), 405–422. doi:10.1016/j.poetic.2012.07.004 Rumelhart, D. E. (1975). Notes on a schema for stories. In D. G. Bobrow & A. Collins (Eds.), Representation and understanding: Studies in cognitive science (pp. 211–236). Academic Press. doi:10.1016/ B978-0-12-108550-6.50013-6 Shklovsky, V. (1965). Art as technique. In Russian formalist criticism: Four essays (L. T. Lemon & M. J. Reis, Eds. & Trans.). University of Nebraska Press. (Original work published 1917) Smith, T. J. (2012). The attentional theory of cinematic continuity. Projections, 6(1), 1–27. doi:10.3167/ proj.2012.060102 Stern, B. B. (1992). Historical and personal nostalgia in advertising text: The fin de siècle effect. Journal of Advertising, 21(4), 11–22. doi:10.1080/00913367.1992.10673382 Thorndyke, P. W. (1977). Cognitive structures in comprehension and memory of narrative discourse. Cognitive Psychology, 9(1), 77–110. doi:10.1016/0010-0285(77)90005-6 Tsai, M. (2013). Stray Dogs [Film]. V. Wang (Producer). Visch, V. T., & Tan, E. S. (2009). Categorizing moving objects into film genres: The effect of animacy attribution, emotional response, and the deviation from non-fiction. Cognition, 110(2), 265–272. doi:10.1016/j.cognition.2008.10.018 PMID:19118823 Welles, O. (1958). Touch of evil [Film]. A. Zugsmith (Producer). USA: Universal Pictures. Wiseman, F. (2017). Ex libris: The New York public library [Film].

15

 Cognitive Transition and Cutting Techniques for Narrative Film Rhetoric Simulation

KEY TERMS AND DEFINITIONS Cognitive Transition: The cognitive processing change of a narrative focus, especially the transition from story-based to non-story-based forms of narrative. Discontinuity Editing: The irrational flow from shot to shot in a film. Indeterminacy: The irrational relationships between time and character or between place and time within continuous shots. Mental Difficulty Cognition: Hard cognition caused by the indeterminacy or the divergence from the presentation. Narrative Simulation: A process of composing various kinds of narrative rhetoric and of testing the cognitive effects of a narrative. Rhetorical Cutting Technique: A technique used to irrationally cut film with the intent of creating an irrational relationship between adjacent segments. Rhetorical Transition Technique: A technique used to generate the plurality, the diversity, and the ambiguity with regards to time, space, or characters’ actions in continuous shots.

16

17

Chapter 2

Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models Sheldon Schiffer Georgia State University, USA

ABSTRACT Video game non-player characters (NPCs) are a type of agent that often inherits emotion models and functions from ancestor virtual agents. Few emotion models have been designed for NPCs explicitly, and therefore do not approach the expressive possibilities available to live-action performing actors nor hand-crafted animated characters. With distinct perspectives on emotion generation from multiple fields within narratology and computational cognitive psychology, the architecture of NPC emotion systems can reflect the theories and practices of performing artists. This chapter argues that the deployment of virtual agent emotion models applied to NPCs can constrain the performative aesthetic properties of NPCs. An actor-centric emotion model can accommodate creative processes for actors and may reveal what features emotion model architectures should have that are most useful for contemporary game production of photorealistic NPCs that achieve cinematic acting styles and robust narrative design.

INTRODUCTION “Constructing characters” is a phrase that infers distinct meanings for two participants in the creative process of computer-based media. On one hand, for the narrative architect of video games or other kinds of narrative computational media, it is a semiotic process of fiction authoring where the character designer provides a written personal history. Three-dimensional or two-dimensional models of anthropomorphic shape combined with voice can also suggest agency in a game space. A weapon-wielding muscular humanoid with big bright eyes is well equipped for video game combat and all of the emotional expression players associate with fighting. Once audio-visual elements are programmed to react to user input, these elements can signify to the observant player an imagined persona. On the other hand, for the developer of video game computer code, “constructing characters” is a process of designing a system that uses DOI: 10.4018/978-1-7998-4864-6.ch002

Copyright © 2021, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

quantitative data derived from the game program, the computer operating system or from player input data to control or trigger character animation and voicing such that the player experiences the presence of a seemingly intelligent cohesive character identity. The result of the work of the designer and the developer is a composite signified that evolves in the player’s mind over the time of game play. The manner of construction for narrative architects of video games depends on mental processes of player participation. Over the course of game play time, the player may observe actions and behaviors of Non-Player Characters (NPCs) so that a pre-game play biography and an in-game “alterbiography” (Calleja 2009) combines the NPC’s pre-game past with the evolving NPC actions the player witnesses or learns through game interaction since the start of the game. The manner of character construction for video game developers depends on computer languages whose frameworks contain data structures (primitives, classes, objects) and data behaviors (methods, functions) that can trigger and manipulate unique animations of the three-dimensional mesh model and its sound emanations (usually a voice) in ways that resemble the player’s understanding of human and animal emotional expressions. These NPC animations and sounds must be recognizable by the player as specific to the NPC’s type as a fictional narrative agent (human or non-human) and consistent with the character’s role within the world of the game. The study of character construction coincides with the related research in other disciplines. The late twentieth century coincidentally delivered strains of thought from disciplines that sought to provide taxonomies of two human endeavors – storytelling and emotion expression. Narratology addressed the former and evolved from literary theory and folklore studies to describe systematically the human perception and representation of stories in various media. The categorization of characters within stories based on a typology of roles and emotion sets afforded to those roles is one specialization within narratology. Cognitive psychology and its subdiscipline, Computational Cognitive Psychology, evolved as a reaction to Behaviorism and as an alternative explanation to the mental and emotional processes that drive human behavior. The categorization of human emotions as well as their neurological processes is one subdiscipline that Cognitive psychologist frequently consider. The two disciplines converge in computer game design because game character designers and game code developers both use models from which characters can be efficiently produced. These characters and their behaviors can be embodied as preconfigured audio-visual animation and sound synthesis systems for “static” characters or can be used in-game to spawn procedurally generated characters or behaviors. This research focuses on NPC modeling rather than Player Character modeling for the benefit of taking a simpler problem before tackling the more complex Player Character with its additional set of variables that describe each player’s unique persona and context. Daniel Vella initialized a discussion of the Player Character’s implications on game narrative that can apply to related studies of NPCs (Vella 2014). A discussion of NPC modeling benefits from the character theories of Narratology and the emotion theories of Computational Cognitive Psychology because both examine the behavior of human or human-like agency. Narratology considers how foregrounding particular NPC behaviors in the context of an audio-visual story system forms the role of a character. These behaviors are a functional necessity for story development and an elaboration in the player’s mind. Computational Cognitive Psychology considers how to represent in computational form, a simulation of the internal processing machinery of the emotive part of the human mind so that when one implements and embeds an emotion model in an agent such as an NPC, the sensory input will yield an expressive output as an appropriate behavior that simulates a coherent human-like character. To design an NPC model of emotion for video games, one must consider how emotions in characters are useful in the elaboration of narrative, and how they are generated by actors for use in animated characters. 18

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

BACKGROUND Whether one considers video games as narratives or narrative systems, the fact a player moves through the time and space of a game world provides sufficient conditions to apply the frameworks of narratology. For games with NPCs, the player experiences over time the construction of character within her mind. Each interaction with an NPC is a momentary witnessing of in-game action with all of its semiotic unwrapping over the duration of gameplay. Cognitive Psychology recognizes that expression on the body, and in particular as seen on the face, belies the mental properties and emotional states of a person (Ekman 1997). It also communicates emotional meanings from the face of one person into the eyes of another. The latter function of emotion serves narrative expression in the visual arts through the agency of character. The degree to which facial expression effects game narrative depends on the complexity of expression, and therefore the degree of immersion and believability a player may experience. Thus, an effective NPC emotion model must strive for complex expressions.

Narrative Expansion Through Facial Expression Complexity Seymour Chatman gives a comparative study of Maupassant’s short story Une Partie de campagne (A Country Excursion) (Maupassant 2017) with Jean Renoir’s rendition of the story in the film A Day in the Country (Braunberger & Renoir 1936). Chatman provides persuasive analyses of how emotions within characters as described by words in a literary narrative are interpreted through cinematography and editing to convey similar ideas in a film adaptation (Chatman 1980). Many of these emotional ideas that Chatman identifies depend on the framing of shots in the film where actors’ faces communicate their emotional states. But cinema is not video game. In Renoir’s film we are forced to see faces within a frame and in the order of a sequence of images. The dialectical process of montage allows the meanings of faces to be refined and elaborated by the image that precedes a facial closeup and by the image that follows. In video games, only cutscenes lock the player into seeing specifically framed and ordered images of faces. When the camera is released to the volition of the player, there are few if any cuts to other images. Once the camera is free to move in response to the player’s will, there is no way for the game designer to be sure that the player sees anything in a particular order. An NPCs facial expression must convey its ideas independent of other objects in the viewing screen. The need for the player to maneuver and see an NPCs face must be embedded in the mechanic of the gameplay. While lighting, music and voice can reinforce the emotional state of an NPC as well as the mood of the game space, the facial expression itself must provide the emotional clarity the game designer intends. If the goals of characters steer their action, then the emotional expression on the face of NPCs reveal how the designer wants the player to understand the character’s attitude toward their goals and the consequence of their action at a particular instance. (Later in this discussion, there will be a more robust elaboration of how Computational Cognitive Psychology informs the design of the internal workings of NPCs.) But is there enough information from the face of an NPC to “move” the narrative and cause the player to decide something significant within a game? The answer depends on the resolution of the animation. A higher resolution facial animation provides the character designer more expressive possibilities. In the past, the concept of visual resolution has described the density of pixels for a given square of screen surface. But animation resolution describes the number of vertex groups of an NPCs facial mesh that move distinctly and simultaneously for a unique emotion expression. While this description does not yet provide a quantifiable ratio of dependent and independent variables (like pixels per inch, for example), 19

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

it does communicate a problem in game animation that effects the player’s experience. Game animation must render each frame in real time (usually 60 frames each second). A character with complicated expressions simultaneously moving many parts of a mesh, can demand a lot of parallel calculations of a graphics process, thus challenging the graphics processor to complete all necessary calculations within 1/60th of a second. Cinematic acting for NPCs depends on such computational power. But with complex animated facial expressions, players can observe more detailed expression; they can read more nuanced emotional information when observing NPCs up close. Chatman’s close reading examines the polysemic phrasings of Maupassant’s story with the polysemic cinematic expression of Renoir’s filmed rendition. High resolution animation allows a similarly comparison between a character in a work of literature and its representation in a video game adaptation. Complex facial expression in game characters can allow the player to perceive NPC emotions in ways similar to human actors. NPC facial expression can then be used to effect narrative generation. Augmenting the expressive capabilities of NPCs has the side effect of augmenting narrative expressivity of video games. Narratologist Uri Margolin defines characterization as a “human or human-like individual… capable of fulfilling the argument position in the propositional form DO(X) – that is, a Narrative Agent (=NA), to whom inner states, mental properties (traits, features) or complexes of such properties (personality models) can be ascribed on the basis of textual data” (Margolin 1986). The definition is applicable especially when examining the detailed form of a game character as seen in a single frame of video game play, in a model sheet typically produced in the preproduction stages of character design, or in a written biography of a video game character. In these video game text documents, signification occurs by inference. The character’s behaviors are imagined by what its body looks like statically, and what has been reported about its past in the written form. Margolin continues by describing characterbuilding in a narrative, which is an “accumulation of a number of traits” or trait clusters “from several successive acts of the NA [Narrative Agent]…” A built character is “a unified stable constellation” of traits experienced in narrative time. While Margolin had in mind primarily literary characters, the player likewise experiences NPCs during gameplay over time and observes their actions in much the same way, gradually building an idea of a character in mind. When one thinks of actions in video game play, often the idea conjures dramatic movements that propel the body of the character or impact objects the character touches. These actions reveal internal and external states and features of NPCs through body gestures, postures and facial expressions. Facial expressions tell much about the internal mental states and actions of NPCs. Literary representations of emotion take the form either of direct predicate statements where the emotional state becomes the predicate or adverb (e.g. It is angry, charmed or It acts lovingly, scornfully) or are presented indirectly as actions or reactions in the form of a verb (e.g. It flees from fear, It flirts from arousal). Visual representations of character must use other means. Posture, gesture and facial expressions of emotion are the primary means of presenting emotion in NPCs. Early video games, much like early animation in the cinema were limited to a smaller vocabulary of emotional expressions because the processor speeds and workflow did not easily allow for much variation of expression. But more powerful graphics processors and modularized workflow allow for complex groups of simultaneous animations of different parts of an NPC face to form many more combinations and variations of emotional expression. By expanding the vocabulary of a character’s emotive expression, the possible meanings derived by the player from an NPC’s expressive system also expand. With more possible meanings entering the space of a narrative, the count of possible player experiences and interpretations grow as well. Thus as NPC emotional expressivity increases in combinatorial complexity with an expanding vocabulary of movements and gestures, 20

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

NPCs have the ability to evolve from expressing simpler cartoon-like acting styles where performative meanings are simple, discrete and the expressive variables show few dimensions, into a more robust cinematic acting style where expression is polysemic, often ambiguous and the expressive variables of the face have many more dimensions.

Agency and Acting The “construction” or “building” of Non-Player Characters by the player within a video game borrows heavily from experiences with other non-ludic art forms, such as literary fiction and films, as well as perceived experiences of simulation systems, such as software. Theatrical and cinematic narrative depend on the appearance of willful actions of characters in a story world. Agency in drama is the apparent freedom of a character to act, react, not act, or sublimate desire to achieve a goal within a fictional world. Aristotle and many modern theorists of agency attribute voluntary goal acquisition as an underlying cause of human action (Charles 2017). For NPCs and Player Characters, Janet Murray translates the concept effectively. “Agency is the satisfying power to take meaningful action and see the results of our decisions and choices” (Murray, 1997). The definition of agency has extended itself into software engineering. As NPCs in video games developed, likewise agents were run-time entities within software. The concept of agency in software engineering emerged commercially in the mid-1990s when software required run-time decisions to manage its states autonomously. Developers realized that software systems need agents to assess the state and behavior of a system’s environment to autonomously manage the system in the background so that desired measurable states are maintained. Agents in software are distinct from other software components because they do not need to wait for a user to tell them what to do. Agents can have goals that they remember pervasively, and they can be provided readable and writable access to execute methods on other software objects to accomplish those goals (Maes, 1997). Eventually, agents within software surfaced to interact with users directly using text, speech or character animation. They helped users accomplish software dependent tasks. These have been called virtual agents. The primary functional tradition of servitude for virtual agents limits their autonomy and their range of emotional expressivity. Virtual agents were not designed to entertain. But as computer games began to use facial animation for emotion elicitation of NPCs, an emotion model architecture was required. Animators provided basic iconic facial expressions of emotions and software developers integrated morph animations driven by emotion models adopted from virtual agents. Virtual agents, however, were not originally embedded in artfully crafted interactive narratives, but were relegated to software assistance for users. As emotion models of Cognitive Computational Psychology were implemented into virtual agents, validation of virtual agent efficacy affirmed their use for pedagogical applications (Rickel and Johnson 1997, Lane et al. 2013). The implementation of the same virtual agent technology for video games was an obvious temptation. But the limitations of graphics processors, central processors and memory required a low bar of aesthetic expectations for players as demonstrated in the crude character designs of the first three decades of computer games. Player aesthetic expectations of autonomously animated NPC “acting” had to be simpler to adapt to technology limitations. Following comic book artists and non-photoreal animators before them, game character artists and developers more often mimicked the expressions in the cannon of Delsartean-based typology of facial and body models to create automated animations that could express basic recognizable emotions (Nixon 2010). Virtual agent emotion models could at best validate their ground-truth assumptions based on a player’s recognition of an NPC emotion, an NPCs believability of the authenticity of expression, or the 21

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

NPCs social appropriateness. These three assumptions reflect the narrow range of expressivity a virtual agent could emote given the technological constraints on real-time 3D animation. These methods of emotion expression in agents do not consider the concept of complexity of animation expression, which can be defined as the number of simultaneous facial animations an NPC may elicit in response to stimuli as either an active or reactive expression. Simultaneous animations with higher resolution animation allow developers and animators to collaborate with actors and realize more complex emotion elicitation.

Behave Like a Human, Think Like a Machine MIT Media Lab researcher Patti Maes in 1997 wanted agents of “intelligent software” to “sense the current state of its environment and act independently to make progress toward its goal” (Maes, 1997). While describing agents, Maes uses one of the underlying concepts of artificial intelligence laid out forty years before by Alan Turing in his notable 1950 essay “Computing Machinery and Intelligence” (Turing, 1950). Turing proposed that machines could think autonomously and learn from their environment and past actions. But four decades earlier, Turing went further than Maes. He proposed that machines could respond in human-like ways to elicit enough rapport to convince a user that the machine was human. Without stating so directly, his rebuttals redrafted the idea of what “thought” was, by allowing that the production of a thought can include distinctly “mechanical” methods. The notable proposal of a continuous state machine suggests that he at least intuited some types of “thought” were not finite states that depended on a combination of true (1) or false (0) relations but were gradients that allowed for a continuous range [0,1], much as basic emotional expression states are currently given normalized values, and much as emotion expression morphs are blended between each other (e.g. blendshapes). Turing anticipated that a machine may have to perform machine-like thinking to come to human-like results. “May not machines carry out something which ought to be described as thinking but which is very different from what a man does” (Turing, 1950)? Virtual agents, as Turing described in his essay, may process “thought” like a machine, but still elicit behavior like a human. Virtual agents are programmed to elicit emotional reactions like humans while processing input data mathematically, probabilistically and algorithmically. Long before computers existed, French philosopher and mathematician, René Descartes reasoned how a machine could never think like a human (Descartes, 2006). Descartes decried in 1668 the “impossibility” of sentient AI. “...It is for all practical purposes impossible for a machine to have enough different organs to make it act in all the contingencies of life in the way in which our reason makes us act” (Descartes, 2006). The reasons he gave described the challenge of computing semantic correctness and social appropriateness with limited computing power and memory to process a wide range of meaningful expression combinations. The basic needs for semantic correctness and social appropriateness accomplish two things that humans do not always do well: teaching and entertaining. Both human activities require sensitive and dynamic awareness of the emotional state of the user. A detailed psychological model of the user or player is essential to direct the actions of an agent precisely and to measure the agent’s effectiveness. A virtual agent with eyes must know where on the user’s body of the elicitor it should look and gather input. It should also be able to interpret what that input means. Gestural and facial expression, with its potential for producing ambiguous meanings that trigger emotions, is what Descartes intuited as a combinatorial morass.

22

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

PEDAGOGIC AGENTS AND EMOTIONAL BEHAVIOR Descartes was correct. Logical processes are not effective to make decisions if available information is not complete enough to find a rational conclusion, or if premises are contradictory. Cognitive psychologists and neuroscientists contend that “emotional intelligence” allows the mind to fall back on another process that is based on probabilistic reasoning (Damasio, 1998; Johnson-Laird & Oatley, 1992). The brain contains neural networks that use pattern recognition to provide probabilistic answers to situations that rational thought cannot adequately answer. By applying emotional reasoning methods in a virtual agent, it can “predict” or recognize patterns from machine-learned (ML) instances to make behavioral decisions.

“Socially Appropriate” Behavior Rosalind Picard, a research pioneer in the field of Affective Computing, describes an emotionally intelligent agent as possessing “abilities to recognize, express, and have emotions, coupled with the ability to regulate these emotions, harness them for constructive purposes, and skillfully handles the emotions of others” (Picard, 2000). In the context of an animated video interaction, a three-dimensionally animated agent often moves and speaks in ways perceived as “socially appropriate” for the context of the useragent relationship. It should demonstrate emotionally logical behavior, follow an expected cause-andeffect sequence, that would respond to the sentience of the user. The term socially appropriate (which is sometimes called “behaviorally appropriate”) describes behaviors that fall within a normal range of variance of behaviors given a specific social situation with a known set of persons in specific types of relationships to a user (Price & Bouffard, 1974). The range of variance is affected by an expectation of satisfaction from a user and the culturally determined role the agent plays in the relationship with the user. For example, a virtual agent classmate in a learning application who assumes the role of a peer will likely behave with a broader range of appropriate behaviors than a virtual agent that is an instructor who assumes a role of pedagogic authority. Behavioral variance in virtual agents can constrain or expand user input as users adapt their decision-making variance to the variance expressed by agents (Yin & Sun, 2015). Additionally, some feature values of the human user are considered that can affect an agent’s behavior – physical location, language, culture, gender, class, education level, age, sexual orientation, country of origin and ethnicity (Gratch, Okhmatovskaia & Duncan, 2006). A virtual agent can be designed to choose for vocalization variables: diction, prosody, volume, pitch and backchannel utterances or for facial expression variables: movement direction, velocity and acceleration so that it can reflect and respond in socially appropriate ways to the user. A designer’s intention is for a virtual agent’s elicitation parameters to effectively tune to the user’s familiar behaviors so that it will maximize rapport during the interaction (Zhao, Papangelis & Cassell, 2014). Virtual agents that teach or entertain, challenge the engineer to model its algorithms to accomplish a task that has no finite outcome of “correctness”. A virtual agent with affective abilities should at least be able to fulfill each of Picard’s criteria for emotional intelligence – emotion recognition, elicitation, and regulation – to accomplish its goal. For example, a virtual piano teacher must demonstrate a nomenclature in musical theory, but also must sense (by facial or vocal expression) if the student “feels” she is able to understand more complex or simple examples, or no examples at all (Picard & Rosalind, 2000). The virtual agent must decide if its own gestural and aural expression should be firmer and more resolute in posture and tone, or gentler and more accepting. A hospital-based virtual agent, called a 23

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Hospital Buddy, provides companionship and reduces psychiatric side-effects of long-term patients. The Hospital Buddy must sense a level of trust to inquire from the patient perceptions of his hospital experience (Bickmore, Bukhari, Vardoulakis, Paasche-Orlow & Shanahan, 2012). Virtual immigration interviewers in the E.U. and the U.S. have been developed and tested for border crossings (Dormehl, 2018). They are designed to flag possible deception by visa applicants using affective biometric data, including facial micro-expressions, during inquiries for facts collected from an interview. The agent must choose questions that intend to trigger emotional responses in relation to known facts and to notice an absence of emotional response (Deahl, 2018). In all these examples of pedagogic virtual agents, they must sense the emotional state of the user and express itself in a socially appropriate way so that its tasks and manner prioritize the productive flow of information between computer to human, and back from human to computer.

Elicitation-by-Design For this discussion here forward, the term pedagogic agent shall refer to didactic, interrogative, informative and simulative virtual agents (all characteristics of a teacher). Pedagogic agents are generally designed to elicit “controlled” emotional responses so that the messages they express produce a predictable experience from the user and for the developer. For pedagogic agents, emotional expression must carry an intended message and outcome so that the user’s behavior leads to an interactive experience that enables the agent to gather the emotional information from the user as intended – surprises can defeat the software’s purpose. Pedagogic agents are not free to express the full capacity of a human personality any more than a human could doing a similar pedagogic task. To accomplish this emotion-elicitation-by-design, the author of the agent must program to specification by biasing the emotion that the agent should demonstrate during an instance of human interaction. Pedagogic agents for the most part substitute a human service that the user is presumed to need. Once the user ceases to need the service, then a pedagogic agent can become an intolerable sycophantic nuisance. The “coached” emotion-elicitation-by-design approach often fails to convey authenticity because the emotional “results” performed in a scripted interaction are intended to synchronize with a developer-permitted pedagogic task to fulfill an anticipated user need rather than respond to the spontaneous emotion elicitation of the user.

Negative Personality Traits Unpredictability, Immorality, Deceptiveness, Self-Loathing, Flippancy, Distractedness and Forgetfulness. The seven words are negative personality traits discouraged for any person assigned to a pedagogic role. Teachers, counselors, judges, doctors would not likely gain more clients if user reviews included descriptions with these negative characteristics. However, they are all attractive characteristics for fictional NPCs because they create inter-personal conflicts that test a player’s beliefs and assumptions about the game world (and perhaps the player’s world). While these characteristics may not be desired in a non-NPC emotion model (such as that of a virtual agent), each of them makes useful functional components in an NPC emotion model as it decides either what emotion its appraisal will select, or what elicitation it should animate.

24

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Neuroticism Sometimes the user appreciates hints of awareness of the negative characteristics in human behavior, or what is typically referred to in personality dimensions as “neurotic behavior” (Trull, 2013). Neurotic behavior from pedagogic agents (virtual or real) distract from the disingenuousness of a commercial-cumsocial interaction. (Consider the cynical humor of flight attendants who joke about passenger attention during pre-flight announcements.) But neuroticism tolerance varies among users and contexts; it may contradict the social “believability” of the virtual agent (Malatesta, Caridakis, Raouzaiou, & Karpouzis, 2007). While humor for the user can be the result of an agent’s neuroticism, it can be socially inappropriate for a user-agent relationship if the agent is pedagogic. Additionally, if the pedagogic agent has little or no awareness of human misdeeds or maladaptation, then a neurotic response would be implausible. Moreover, a pedagogic agent must have a narrow range of expressive variability to keep the user serviced and focused on the tasks of the software.

DISTINCT EMOTIONAL FEATURES OF NPCs An entertaining agent gathers, processes, and elicits sentient information with greater freedom of emotional expression than a pedagogic one, much as an actor has more emotionally expressive liberty than a teacher. Entertaining virtual agents are found in games and interactive media where the agent is personified to express the features of its personality that serve the game world. For clarity, this author refers to entertaining virtual agents as NPCs. Their emotion-response algorithms receive relevant data from the player and game play space and elicit expressive data back into the game play environment for the player to observe and respond. A cycle of data exchange can create rapport or dissonance between player and NPC as the game evolves and as the NPCs role requires. The narrative “facts” revealed in this exchange can be distorted or omitted by the NPCs subjective filters, much as any fictional character does in other media. Furthermore, neuroticism may be welcome insofar as it fits the role required for the world of the game. NPCs are often designed to elicit from the player a wider range of emotional responses than pedagogical agents. Similarly, live action actors and comedians or animated cartoon characters provoke emotions that are not permitted from instructors or public spokespersons. An actor, through his character, can elicit rage, fear and disgust in “socially inappropriate” ways when playing a conflicted character. Pedagogic agents such as piano teachers or public health officials usually cannot; the roles of pedagogic agents rarely allow neurotic behavior. Spectacular neuroticism as a performative aesthetic of cinematic acting is a feature of cinematic acting insofar as game players also watch movies and expect a similar degree of performative complexity, emotional authenticity and celebrity attraction (Morgans, 2018; Stuart, 2016).

Cinematic Acting Style The trend in the last half-decade has demonstrated strong interest in cinematic acting styles as games increase their ability to use higher resolution graphics, motion, and complex facial expression animation (Torres, 2014). Cinematic acting differs from theatrical acting in the production method. While performance preparation methods are very similar, cinematic acting requires the actor to deconstruct a performance into smaller units for each shot. Acting for animation of video game NPCs require actors to 25

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

decompose further – down to the gesture, posture, or facial expression. To capture motion data for high resolution animation, increased computational power is needed to recognize the face, track its inflection, and predict its movement over time. Contemporary graphics processors allow for the capture of more complex and simultaneous emotion expressions. High-resolution animations support the aesthetic of cinematic acting styles in video games. What are cinematic acting styles? Cinematic acting inherited many methods for training from theater. There are two dominant currents in cinematic acting styles. One works from a collective external vocabulary of postures, gestures and facial expressions that are loosely assigned to roles within a theatrical narrative. The task of the actor is to shape his own variation of a role as a set of expressions that convey the emotional expressions of a character for the specific narrative. This approach, often called the outside-in approach and is associated with performance theorist Vsevelod Meyorhold and Francois Delsarte, is highly adaptable between cultures and requires a studious actor to learn the emotive vocabulary of the target audience. The outside-in approach affiliates well with animation and its tradition of referencing a Delsartean taxonomy of physical expression, much like trained and practiced for silent film acting (Hart 2005). The other major approach requires a process of self-examination to identify existing emotional impulses from past lived experience so that when an actor in-character focuses on an action, the resulting emotion from obstructions or assistance are personally authentic. This inside-out approach has been affiliated with Konstantin Stanislavski and his Method approach as well as Sanford Meisner. In animated or live-action films, both currents of cinematic acting deploy complex simultaneous movements motivated by inferred psychological states that are used by actors to create character facial expression for motion capture or voice recording for video games. For this research, both cinematic acting styles provide valuable methods for facial expression in video games as they connect the emotional source of actor expression to NPCs that move narratives forward. Games that aspire to cinematic high resolutions of image and movement, such as Beyond Two Souls (Quantic Dream 2013) and The Last of Us (Naughty Dog 2013), have shown substantial consideration of the aesthetics of cinematic acting through complex animations of the face. Therefore, the methods and techniques of cinematic acting, character development and performance deserve consideration as video games embrace NPCs with cinematic acting styles.

Responsiveness to Multiple Simultaneous Stimuli The game player, attracted to cinematic acting, expects more complex animation to enrich the perceived emotional and intellectual life of the character. To develop cinematic acting for NPCs, its animation must move in complex ways that are consistent with the design of a complex screen performance; this means moving groups of vertices of an NPCs mesh in ways that represent a response from simultaneous stimuli from both external-physical and internal-mental sources. When we closely examine well-prepared actors that move in any fully developed performance style, such as naturalism or physical comedy, what makes a performance complex (and possibly “good”) is when an actor responds to multiple stimuli almost simultaneously, from sources that are external-physical and also from sources that are internal-mental. Acting Theorist John Lutterbie draws from Cognitive Science as he describes the “acting instrument” as a Dynamic System Theory model that listens inside and outside the body, filters decisions through memory, then responds with complex movement, language and gesture (Lutterbie, 2011). Complexity increases when an actor listens, acknowledges and responds with “executive control” to multiple stimuli with an intensity that seems appropriate to the scene, role, its story, and its world. 26

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Often attention to stimuli in the external-physical space appears to the player with a subjective internalmental amplification or diminishment of intensity. As game designers deploy cinematic acting styles, the Player observes stimuli generated from the NPC that seems internal-mental and occurs simultaneously with stimuli from the external-physical space. That disparity of intensity creates a competition of focus for the NPC which results in polysemic expression. The player notices this distinction and imagines, often with pleasurable uncertainty and insatiable curiosity, what the actor is thinking or feeling. Often the most pleasure-inducing experience is an ambiguous facial expression, and the audience or player invents or projects her own meanings that caused the expression. This ambiguity of focus can be built into the NPC and is fundamental to how cinematic acting styles contribute to the expansion of the narrative space. No longer is the game limited to a visual physical space. Cinematic acting styles open the narrative to the mental space of the NPC. To achieve cinematic acting, NPCs require uniquely subjective filters and algorithms to idiosyncratically appraise and elicit reactions to the emotion-evoking entities it perceives.

Cinematic Mediation of Emotion Expression Principal game characters are presented to the player base with spectacular trailers and posters, much modeled on theatrical film trailers. These advertising media reinforce character archetypes and stereotypes, and also create a set of behavioral expectations in the player before ever playing the game. Additionally, in-game NPCs are experienced by players as mediated persona through game video screens and speakers, much like film and television characters. Audio-visual moods and cues reinforce mediated emotional expressions in a game much as one observes in film and television (Smith, 2003; Wiley, 2003). The construction of mediated emotional expressions results from already familiar production methods of actor training, casting, rehearsal, recording and reperformance technologies that games use for their production. A facial expression on an NPC after a kiss or bullet to the back triggers an expected category of emotional response that can stand alone or come accompanied with sympathetic or dissonant music, sound effects or dialog. The combined facial expression and sound can affect the perceived meaning of the response and negate or affirm the expected response. These expectations do not precisely resemble real life but are a synthesis of memories of watching previous mediated experiences of similar events combined with real life memories. Actors are trained to express real emotions and to synthesize them with culturally preconceived expressions of real emotions.

Performative Mediation of Emotion Expression Mediation is not only determined by industrial or societal norms through cinema. Mediation can also be the result of an actor’s own imaginary construction of the character’s mental processes as she believes the audience will perceive them. As each elicitation of an actor or NPC is presumably the result of the flow of emotion-evoking stimuli, it is an actor’s detailed design of those mental processes for the character that provides actor-centric mediation of elicitation. Thus, what has here been defined as cinematic acting style for NPCs, with its complex mesh animations, is the result of an actor’s invention of an internal emotion model that filters sensations, identifies them, prioritizes them, associates them with memories, maps sensations to affect definitions, and selects an elicitation in response that appears as an emotional facial expression.

27

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

EMOTION MODELING OF COGNITIVE COMPUTATIONAL PYSCHOLOGY Performance theories that are foundational for cinematic acting share some concepts that describe the derivation of human emotion with those defined in the literature of Computational Cognitive Psychology. Psychologists Ortony, Clore and Collins were among the first to propose a human emotion model for simulation on the computer. While theirs was not intended for game development, it is easy to imagine that its deployment would sense emotion triggers in the game play space. The model would process the sensed data through its own character-specific filters, choose an emotion that describes the state of the NPC, and then elicit that emotion through an animatable face, body or synthetic voice. During the last thirty years, emotion models were designed to test the human affect processing system by observing them embedded in virtual agents more than NPCs. Computational Cognition Theories of emotion are collectively known as Appraisal Theories. To understand the rationale for an Actor-Centric emotion model for NPCs, a fundamental understanding of an Appraisal Theory framework for emotion models is helpful.

Appraisal Theory for NPCs Computational Cognitive Psychologists for the most part agree that emotions are generated in response to an “appraisal” of the environment and its condition to affirm one’s beliefs, provide access to fulfill desires and enable one’s intentions (BDI). A person’s environment is often laden with entities to appraise in ways that concern a person’s physical and social survival and quality of existence. Appraisal occurs initially as a person assesses the immediate situation and its immediate impact (Ortony, Clore & Collins, 1990). And for some theorists, there is a secondary appraisal (reappraisal) where one considers how to cope long-term by accepting the situation as it is and adjust one’s attitude about the situation. A person might then take action to change the situation in response (Lazarus, 1991). Entities that have emotional effects on a person are either inanimate objects, animate agents, or consequential events. A person will judge an object’s likeability, an event’s desirability (of consequences), or an agent’s (past, current or future) praiseworthiness (of actions) in terms that either favor positively or negatively one’s own BDIs. Observing and evaluating an entity (object, event or agent) and its feature values registers an emotional appraisal along one or more dimensions of measurement. The values of these measurements are stored as appraisal variables. A fundamental scale of measurement is pleasure and displeasure within a range of [-1,1]. This dimension has also been interpreted as aversion and attraction. As a two-dimensional appraisal scale, pleasure-displeasure runs on the X-axis and rises with arousal (intensity) on the Y-axis within a range of [-1,1]. Originally known as the Circumplex Model of Emotion (Russell, 1980), it has been further refined as a 12-emotion unit circle with evenly divided expression positions called the Core Affect (Yik, Russel and Steiger, 2011), as shown in Figure 1. Core affect prioritizes the “basic” emotion positions in the unit circle. A three-dimensional scale variation includes dominance on the Z-axis within a range of [-1, 1] (Mehrabian, 1996) where dominance represents the degree to which a person can control the emotion from triggering impulsive elicitations that could be anti-social, unwelcomed or inappropriate, as shown in Figure 2. Most Computational Cognitive Models based on Appraisal Theories use either two- or three-dimensional emotion mapping, as found in Embodied Conversational Agent development (Malatesta, Raouzaiou, Karpouzis & Kollias, 2009). Using two- and three-dimensional emotion maps is known to provide a system modeled on closely observed human behavior. Emotion maps allow researchers to test a virtual agent accurately in relation to a human reference (Lisetti & Hudlicka, 2015; Gratch, Marsella, Wang, Stankovic, 2009). 28

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Figure 1. Yik, Russel and Steiger (2011) represent Core Affect in a two-dimensional emotion map that evenly divides a unit circle into 12 sections

NPCs perceive entities with subjectively determined values of their feature variables based on their character-specific BDIs. For every entity that triggers an emotion, an algorithm must translate and map all the entity’s feature values in a two-dimensional Circumplex, as shown in Figure 3, or a three-dimensional volume, as shown in Figure 4, to a coordinate point positioned by its appraisal variable values -- Pleasure, Arousal and (if in three-dimensions) Dominance (PAD). These two or three dimensional values are normalized within a [-1, 1] range of a unit circle or sphere. Much of the distinction between appraisal theories is the structure of the emotion model, and therefore the path of stimuli signal through the model. Some theories emphasize memory and pattern recognition components in relation to goals and praiseworthiness, others emphasize reappraisal and changes of emotion response, while others emphasize the effect of stimuli on planning for goal acquisition. The resulting PA or PAD values may change as the appraisal process cycles through each appraisal or reappraisal over time. PA and PAD values infer a vector with a coordinate point at its tip. With each reappraisal cycle, the vector can move to a new position as the emotional state of the agent changes. A coordinate point within a unit circle or sphere falls into a region or volume that is bounded by an emotion label. The idea of the emotion label is controversial. Some Computational Cognitive Psychologist prefer to abandon labels for emotions. The trepidation to work with emotion labels stems from the subjective quality of assigning a culturally determined word to an object (an emotion) only known by

29

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Figure 2. Lance and Marsella (2010) use Mehrabian’s three-dimensional emotion map PAD to show example emotional categories

elicited effects in the body, and often repressed by the vagaries of culture (Russell & Feldman Barrett, 1999). Labeling an emotion is a linguistic problem rife with semantic contradictions of inclusion and biases. However, since NPC design depends on collaboration among creative and technical professionals, labels must be used to communicate software development processes. This author recommends the following considerations to avoid misunderstanding labeling emotion expression of the face. First, a developer must consider if in the culture of the player the face is where a specific feeling is expressed. Second, a developer must decide if the system for measuring elicitation (usually ML-based) can detect that labeled emotion on the face. Third, is the emotion word a categorical label that includes other more specific labels or is it a more detailed and descriptive word that does not contain other labels. Labels that are at the top of an inclusion hierarchy tend to draw more agreement of meaning. At the top of a hierarchy of emotion labels are “primary” or “basic” emotions. These are allegedly triggered by biological instinct and evolutionary necessity (Izard, 1992). If the label is not a primary emotion, then it is a secondary emotion. Some Cognitive Psychologist and Constructivist Neuroscientists believe that many emotions are contextually constructed, categorized and classified as expression

30

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Figure 3. Yik, Russell & Steiger (2010) use a 12-Point Affect Circumplex (12-PAC) model of Core Affect to plot the vectors of 30 mood words

concepts. Constructivists contend humans have a proclivity to express many emotions only if learned and permitted, and that some expressive characteristics may not be innate at birth (Quartz, 1999; Barrett & Satpute, 2013; Barrett, 2017). The labeling of patterns “nurtured” in context, explains the differences in emotion label meanings from one culture or social group to another. The components described in the various Appraisal Theory-based models are designed to best attain the principal objective or action of a person. Typically, object-action pairings are ascribed to an NPC. They could for example be assist player-character to attain goal, obstruct player-character to attain goal, obtain a game object or manage a game environment. If a virtual agent or NPC is intended to simulate the full psychological profile of a human, then a very complex model will be necessary to accommodate types of memory, learning, adaptation, natural and social reasoning, language acquisition and expression, motor-skill acquisition and adjustment, and other influences. Additionally, since reappraisal of an entity can be triggered by internal stimuli from memory, a scheduling algorithm must control the bussing of data through the components, thus allowing for more than one path through the system from sensation to elicitation, and allowing for interruption by cycles of signal between components.

31

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Figure 4. Zhang, S., Wu, Z., Meng, H. M., and Cai, L. (2010) use Mehrabian’s three-dimensional emotion map to plot facial expressions of a talking avatar from an expression database

SOLUTIONS AND RECOMMENDATIONS For an NPC emotion model to accommodate the complexity of perceiving and processing multiple emotional entity categories of stimuli simultaneously, components that sense and process emotion-evoking entities must be built into the emotion model for sensing, evaluating, prioritizing and selecting affect. Ultimately, the path of either a sensation signal or a processed affect selection is controlled by algorithms that schedule the flow of data from sensation to appraisal to elicitation so that behavior is synchronized to receive new sensations and to reconsider past sensations (reappraisal) held in memory. Present sensations and memories compete for attention. Simultaneous appraisal and elicitation with reappraisal require parallel processing and scheduling to manage the complexity of computational processes. To achieve cinematic action and therefore high-resolution animated emotion expression, complexity in an emotion model must increase with the growth of more specific components and more non-linear scheduling of affect selection and elicitation.

An Actor-Centric Emotion Model Extracting the principal points identified thus far, an actor-centric emotion model should have components that allow for the NPC performative characteristics. Arrays, structs or vectors serve to store

32

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

immutable ordered data that describe the implicit biases of an agent. A Feed-Forward Neural Network (FFNN) provides the structure to algorithmically process input from multiple handlers-listeners or sensors into a gradient output that will result into an animated elicitation. These structures are used in Table 1 to describe how an emotional model may deploy the critical performative characteristics of an Actor-Centric emotion model. Table 1. Data Structures and Algorithms for Performative Characteristics Performative Characteristic

Data Structure(s)

Algorithmic Method(s)

Expression of Negative Traits

Arrays or vectors that store coefficients to alter values of post-appraisal dimensional vector

Receives post-appraisal signal as dimensional vector. Each dimensional value is multiplied by a coefficient C Î ¡

Multiple Sensations Simultaneous Sensations

Multiple FFNNs or structs with object handlers-listeners and feature recognition array

FFNNs or structs receive signals from environment during same frame (appraisal instance).

Internal Stimuli

Array 1 stores feature values of significant objects from previous appraisals (memories). Array 2 stores relevant structures indexed by object type whose methods can trigger internal sensations.

If external stimuli are recognized as similar to significant objects in array 1, then method in structure of matching object type in array 2 triggers internal sensation.

External Stimuli

Objects in game space: Struct with vector representing feature values. Objects in player space: FFNN recognizes object and converts to vector representing feature values.

External stimuli are recognized as similar to objects in array 1, then method in structure of matching object type in array 2 triggers internal sensation.

Self-Oriented Goals

Data types that permit gradient values, such as 3D vectors. Or, any variable if the goal is not an emotional state or change in emotional state.

Method must check goal value and current state value. Difference may trigger new appraisal. Method might check player variables to decide to forgo action to serve self and instead engage player.

Modularity

Affect Derivation Component object or struct contains arrays or links to FFNNs. More modules increase ability to recognize more emotional entities.

Method switch can enable or disable modules as needed to minimize computational expense as agent changes state or space.

Scalability

Within Affect Derivation Component, increasing Affect Components allows for more granular sensitivity to variations of entity feature variables.

Method switch can enable or disable components as needed to minimize computational expense as agent requires greater sensitivity to entity feature variables.

Affective Loop

The entire emotion model runs as a process. It contains two cycles. Cycle 1: entirely internal and allows reappraisals. Cycle 2: partially external to the agent, allows the agent to send out elicitation and receive response from other agents or the player.

Scheduling of signal vector must allow for each component to complete its task. Must store in memory significant appraisals for subsequent recognition or reappraisal. Must regulate and prioritize competing signals to determine which should occupy appraisal bus: new signal or reappraisal.

A generic emotion model as shown below in Figure 5, presents the basic components of an NPC. This generic model was synthesized from the MAMID model developed by Hudlicka (2003), from the FeelMe model developed by Broekens and DeGroot (2004), and from the FAtiMA model developed by

33

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Dias, Mascarenhas and Paiva (2014). The resulting model satisfies Picard’s componential requirements. Table 2 elaborates on the function of each component. Inside of the Affect Derivation component in Figure 5 is what is depicted in Figure 6. The generic emotion model uses FAtiMA’s modularity feature. The Appraisal Components provide the Appraisal Frame the current values of each Appraisal Component’s predictive result. This author proposes that each component consist of an FFNN that recognizes the entity and its emotion relevant feature variables and values to predict appraisal variable value for each affect from [0, 1]. For a facial emotion expression, that would be the gradient value of an emotion expression of which there is usually at least six expressions plus neutral. Figure 5. A generic emotion model that combines features from MAMID (Hudlicka, 2003), FeelMe (Broekens and DeGroot, 2004) and FAtiMA (Dias, Mascarenhas and Paiva, 2014)

34

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Each Affect Derivation Component consists of a data structure that receives input from an Appraisal Filter. The Appraisal Filter receives an appraisal signal that recognizes each emotion expression for each respective Appraisal Component. The filter will likely implement an object recognition neural network or an array of feature value ranges – both will process a vector whose elements and their values are within range to process and predict a meaning. For a facial emotion expression system, the Appraisal Filter will look for facial features in a video stream that fit the parameter values of an emotional expression on a human face. Then the signal must be sent to the appropriate Appraisal Component within the Affect Derivation unit. An Appraisal Component will evaluate a feature of the face that best indicates a value of a specific emotion expression type (happiness, anger, etc.). Then the Appraisal Component must use an emotion recognition FFNN to predict the expressive meaning of the signal. The resulting gradient value will reside in the Appraisal Frame. All Appraisal Frames will be evaluated to choose a final Affect Derivation for the appraisal instance. The resulting derivation will be the Affective State, which will make a Behavior Selection, and in turn trigger a Behavior Elicitation.

Applying the Affective Loop for Emotion Model Dynamics A consistent theme in the discussion of emotion models is the idea that affectively relevant sense data runs a cycle between an agent and its environment. Sundström calls this cycle the Affective Loop (2005). The Affective Loop has three stages familiar to actors from improvisation and rehearsal exercises: 1) a human user elicits an emotional expression, 2) a system responds with an appropriate emotional expression, and 3) the human user internalizes and interprets the affective meaning of the eliciting system, and 4) the loop repeats. A similar concept is found in the Repetition Exercise of renowned theater director and acting teacher Sanford Meisner (Meisner, 1987). The three stages of the Affective Loop have parallel stages in the process of developing an actor’s skill at finding authentic emotional responses to interactions with other characters, objects and events observed in a scene. The Affective Loop expands human-to-human interaction, as practiced in the dramatic arts, to a “design principle” of the Affective Computing domain that implements motion-capture-performer-to-computer and computer-to-player principles. The Affective Loop design principles are interpreted for player to NPC interaction as follows from Sundström (2005). Embodiment is enhanced as the player’s body expresses its affect through the game controller or other sensing interface. Some part of the player’s elicitation data becomes input that the NPC’s emotion system can sense. Likewise, the player can sense the NPC sensing the player, and sensing the NPC eliciting its organismic idiosyncrasies through its own body. Flow is another attribute where, as the player interacts, there is a sense that there is always potentially a new sensation coming from which to repeat the Affective Loop. Or, an NPC can react to internal stimuli from memory. The emotion model not only repeats as loops, but it is always open for new stimuli from internal or external sources that allow each loop iteration to elicit distinct values. Another Affective Loop attribute occurs when NPCs are designed with elicitation ambiguity. When there is more than one meaning and possibly more than one stimulus attributed to an NPC elicitation, the uncertainty can create interest in subsequent sense information with the belief that clarity of meaning will be revealed over time. Ambiguity is accomplished with facial expressions by always allowing multiple simultaneous stimuli sensations and appraisals causing distinct affect selections that can blend in the facial elicitation. Lastly, Natural-But-Designed Expressions are characteristics that value elicitation forms found in nature, such as shapes of the face (eyes, mouth, ears) and their movements that resemble shapes and 35

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Table 2. Component descriptions of the generic emotion model in Figure 5 Component Name

Description

Sensible event-laden environment

Container of all events, objects and agents that can be sensed by emotional agent.

Sensorial reception

Events that are accepted by sensors.

Appraisal filtration

Filters perception of objects, agents, events in terms of beliefs, desires (goals) and intentions (BDI).

Affective memory association

Storage of perceptions, affects and elicitations.

Affect Derivation

Pleasure-Arousal-Dominance mapped emotion vector algorithm.

Behavior selection

Resulting animation of face indicating current emotional state.

Affective memory association

Storage of perceptions, affects and elicitations change the “rules” or “nets” (and don’t maintain a memory of events.

Behavioral elicitation

The physical movement of the mesh as gesture or facial expression.

Figure 6. Detail of Affect Derivation Component from FAtiMA component (Dias, Mascarenhas, & Paiva, 2014)

movements of recognizable natural objects. For example, when eyebrows rise in surprise or rage, they take the shape of volcanos or other swollen vessels filled with energy. With its flexible deployment, the Affective Loop as applied within a system provides a guide that reinforces an NPCs resemblance to intelligent behavior and appearance of living entities. The Affective Loop intends continuous engagement by providing the player with familiar organismic processes.

Subjectively Filtered Imperfect Perception and Memory NPC emotion models should also allow for occluded, skewed or false perceptions of sensed data by deploying subjectively tuned feature value distorting filters consistent with the NPCs idiosyncratic personality. Subjectivity in virtual agents is a known topic in software engineering, and it is used to localize agent behaviors to restricted domains, even when the agents are NPCs (Omicini & Ossowski, 2003). Designing subjectivity in an NPC then allows for building “imperfect”, non-machine-like or

36

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

organismic features that simulate human constraints on intelligence. Thus, rationality can be limited as the NPCs character design requires. Similarly, an NPCs memory can also be designed as “imperfect”. Memories can be subject to disappearance, erosion and distortion over time. For an NPC to show signs of organismic entropy is for its memory to simulate decay and self-doubt as the game play proceeds. Furthering organismic limitations, an NPC may not always respond too quickly, nor should it always respond without the delaying force of anticipatory doubt, remorse or regret. Similarly, NPCs can reflect before choosing a course of action. They can likewise elicit a choice with temerity or tentativeness. People often mumble and stumble because they doubt their choices, forget their memories, become distracted by competing but pertinent stimuli, or fail to figure out a course of action in due time. Occasionally unclear elicitation is the result of an overloaded overwhelmed system. Such “imperfections” built into an emotion model make for NPCs that cause curiosity in players and reinforce the lifelikeness of the world that spawned them.

Modularity and Scaling of Emotion Resolution Less an organismic and more a mechanical concern, the most adaptable emotion models are those that allow for modular components that increase the variety of sensible entities and the resolution of sensing input data. Modularized emotion models can allow an increase or decrease in the number of feature variables an NPC can sense, and an increase in the complexity of elicitation. Modularity gives a designer the ability to use the same model with fewer components for simple mass-produced NPCs, while for more complex supporting characters, and for very complex antagonists or “buddy” characters, more modules can expand the character’s emotion complexity. Modularity also allows for real time scaling up or down, where if a complex character has a brief and simple appearance, its model can retain less data with fewer components, or vice-versa. Thus, when designing NPC emotion models, scale of entity detection and emotion elicitation may vary enormously, depending on the importance of a character in a game. Like the limitations of scale, games provide circumscribed interactions with NPCs that reveal an incomplete view of a character. We get to know an NPC over time through a mediated context that serves the whole game. There is no need to model more than what the game design will allow the player to sense during anticipated game play time. A model needs to show only the emotions that the game mechanic allows. Like many forms of humanistic expression that comprise an art form, players must infer what they do not sense. And they cannot sense everything at once.

FUTURE RESEARCH DIRECTIONS With an emotion model that is modular, it becomes conceivable to implement a procedurally generated emotion as part of a procedurally generated NPC and thus within a procedurally generated narrative system. Procedurally generated game components can become principal features of adaptive games that reshape themselves to the variables of the player. Adaptivity would require some mechanism that can categorically predict the features of the player so that generated game components would spawn in variations that best suit the designer’s prediction of what the player needs to experience. A predictive system would require a robust machine learning algorithm that enables adaptivity in ways productive to the aesthetic of the game and player-designer relationship. From the perspective of the actor who must perform the character motion, it remains to be seen if the workflow proposed in this research is more 37

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

or less consistent with the performance preparation process. This question needs further research and experimentation. Future work described is viable, reproduceable and could provide a new paradigm for production of photorealistic NPCs and cinematic acting in video games.

CONCLUSION Enabled by processor innovations and the techniques of machine learning, photorealistic NPC design and development can adopt the performative aesthetics of motion picture acting with its complex facial animations. The challenge that lies ahead for developers is to design an emotion model and working process that adapts to existing entertainment methods and techniques. Past practices that relied on handcrafted morph animations have provided remarkable accomplishments at creating life-like characters for computer games. However, animator-centric NPC emotion modeling based on morph animations is a cost prohibitive method for most game developers and designers. This research promotes a synthesis of the structures of emotion developed by Computational Cognitive Psychology with the most influential Performance Theories and acting training methods and practice toward the service of procedurally generated character and narrative. Future work will reveal if putting a human actor’s creative process at the center of emotion modeling and the workflow of character design will make NPCs more enjoyable and closer to cinematic style acting for computer-based games.

REFERENCES Bickmore, T., Bukhari, L., Vardoulakis, L., Paasche-Orlow, M. & Shanahan, C. (2012). Hospital buddy: a persistent emotional support companion agent for hospital patients. Academic Press. Braunberger, P. (Producer), & Renoir, J. (Director). (1936). A Day in the Country [Motion picture]. France: Panthéon Productions. Broekens, J., & DeGroot, D. (2004). Scalable and flexible appraisal models for virtual agents. In Proceedings of the Fifth Game-on International Conference (pp. 208-215). Academic Press. Calleja, G. (2009). Experiential narrative in game environments in breaking new ground: innovation in games, play, practice and theory. Proceedings of DiGRA 2009. Chatman, S. (1980). What novels can do that films can’t (and vice versa). Critical Inquiry, 7(1), 121–140. doi:10.1086/448091 Damasio, A. R. (1998). Emotion in the perspective of an integrated nervous system. Brain Research. Brain Research Reviews, 26(2-3), 83–86. doi:10.1016/S0165-0173(97)00064-7 PMID:9651488 de Maupassant, G. (2007). The Short Stories of Guy de Maupassant. Black’s Readers Service. Deahl, D. (2018), The EU plans to test an AI lie detector at border points. The Verge. https://www. theverge.com/2018/10/31/18049906/eu-artificial-intelligence-ai-lie-detector-border-points-immigration Descartes, R. (2006), A discourse on the method of correctly conducting one’s reason and seeking truth in the sciences (I. MacLean, Trans.). Oxford University Press.

38

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Dias, J., Mascarenhas, S., & Paiva, A. (2014). Fatima modular: Towards an agent architecture with a generic appraisal framework. In Emotion Modeling (pp. 44–56). Springer. doi:10.1007/978-3-319-12973-0_3 Dog, N. (2013). The Last of Us [Computer video game]. Sony Interactive Entertainment. Dormehl, L. (2018), A.I. border agents could use machine smarts to tell if travelers are lying. Digital Trends. https://www.digitaltrends.com/cool-tech/ai-border-airport-virtual-agent/ Dream, Q. (2013). Beyond Two Souls [Computer video game]. Sony Interactive Entertainment. Ekman, P. (1997). What the face reveals: Basic and applied studies of spontaneous expression using the Facial Action Coding System (FACS). Oxford University Press. Gratch, J., Marsella, S., Wang, N., & Stankovic, B. (2009). Assessing the validity of appraisal-based models of emotion. In 3rd International Conference on Affective Computing and Intelligent Interaction and Workshops (pp. 1-8). IEEE. 10.1109/ACII.2009.5349443 Gratch, J., Okhmatovskaia, A., & Duncan, S. (2006, November). Virtual humans for the study of rapport in cross cultural settings. 25th Army Science Conference, 27-30. Hart, H. (2005). Do You See What I See? The Impact of Delsarte on Silent Film Acting. Mime Journal, 23(1), 184–199. doi:10.5642/mimejournal.20052301.11 Hudlicka, E. (2003). Modeling effects of behavior moderators on performance: Evaluation of the MAMID methodology and architecture. Proceedings of BRIMS, 12. Lane, H. C., Cahill, C., Foutz, S., Auerbach, D., Noren, D., Lussenhop, C., & Swartout, W. (2013). The Effects of a Pedagogical Agent for Informal Science Education on Learner Behaviors and Self-efficacy. In Artificial Intelligence in Education. AIED 2013. Lecture Notes in Computer Science, (vol. 7926). Springer. Lazarus, R. S. (1991). Cognition and motivation in emotion. The American Psychologist, 46(4), 352–367. doi:10.1037/0003-066X.46.4.352 PMID:2048794 Lisetti, C., & Hudlicka, E. (2015). Why and how to build emotion-based agent architectures. In R. A. Calvo, S. K. D’Mello, J. Gratch, & A. Kappas (Eds.), Oxford Library of Psychology. The Oxford Handbook of Affective Computing (pp. 94–109). Oxford University Press. Lutterbie, J. (2011). Towards a general theory of acting: Cognitive science and performance. Palgrave Macmillan. doi:10.1057/9780230119468 Maes, P. (1997). Intelligent Software. IUI ‘97 Proceedings of the 2nd International Conference on Intelligent User Interfaces, 41-43. Malatesta, L., Caridakis, G., Raouzaiou, A., & Karpouzis, K. (2007). Agent personality traits in virtual environments based on appraisal theory predictions. Artificial and ambient intelligence, language, speech and gesture for expressive characters, 7. Malatesta, L., Raouzaiou, A., Karpouzis, K., & Kollias, S. (2009). Towards modeling embodied conversational agent character profiles using appraisal theory predictions in expression synthesis. Applied Intelligence, 30(1), 58–64. doi:10.100710489-007-0076-9

39

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Margolin, U. (1986). The doer and the deed: Action as a basis for characterization in narrative. Poetics Today, 7(2), 205–225. doi:10.2307/1772759 Mehrabian, A. (1996). Pleasure-arousal-dominance: A general framework for describing and measuring individual differences in temperament. Current Psychology (New Brunswick, N.J.), 14(4), 261–292. doi:10.1007/BF02686918 Meisner, S. (1987). Sanford Meisner on acting. Vintage Random House. Morgans, J. (2018). How video games cast actors just like movies do. Vice. https://www.vice.com/en_uk/ article/gymde7/how-video-games-cast-actors-just-like-movies-do Murray, J. H. (1997). Hamlet on the Holodeck: The future of narrative in cyberspace. MIT Press. Nixon, M., Pasquier, P., & El-Nasr, M. S. (2010). DelsArtMap: Applying delsarte’s aesthetic system to virtual agents. In International Conference on Intelligent Virtual Agents (pp. 139-145). Springer. 10.1007/978-3-642-15892-6_15 Omicini, A., & Ossowski, S. (2003). Objective versus subjective coordination in the engineering of agent systems. In Intelligent Information Agents (pp. 179–202). Springer. doi:10.1007/3-540-36561-3_9 Ortony, A., Clore, G. L., & Collins, A. (1990). The cognitive structure of emotions. Cambridge University Press. Picard, R. W., & Rosalind, W. (2000). Toward agents that recognize emotion. Vivek-Bombay, 13(1), 3–13. Price, R. H., & Bouffard, D. L. (1974). Behavioral appropriateness and situational constraint as dimensions of social behavior. Journal of Personality and Social Psychology, 30(4), 579–586. doi:10.1037/h0037037 Rickel, J., & Johnson, W. L. (1997). Integrating pedagogical capabilities in a virtual environment agent. In Proceedings of the First International Conference on Autonomous Agents (pp. 30-38). 10.1145/267658.267664 Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39(6), 1161–1178. doi:10.1037/h0077714 Smith, G. M. (2003). Film structure and the emotion system. Cambridge University Press. doi:10.1017/ CBO9780511497759 Stuart, K. (2016). Video games where people matter? The strange future of emotional AI. The Guardian. https://www.theguardian.com/technology/2016/oct/12/video-game-characters-emotional-ai-developers Sundström, P. (2005). Exploring the affective. University of Sweden. http://www.diva-portal.org/smash/ record.jsf?pid=diva2:1041047 Torres, S. (2014), Video game characters modeled after real people. Venture Beat. https://venturebeat. com/2014/06/28/video-game-characters-modeled-after-real-people/ Trull, T. J., & Widiger, T. A. (2013). Dimensional models of personality: The five-factor model and the DSM-5. Dialogues in Clinical Neuroscience, 15(2), 135. PMID:24174888

40

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

Turing, A. (1950, October). Computing machinery and intelligence. Mind, 59(236), 433–460. doi:10.1093/ mind/LIX.236.433 Vella, D. (2014). Modeling the semiotic structure of game characters. Proceedings of DiGRA 2014. Wiley, N. (2003). Emotion and film theory. Studies in Symbolic Interaction, 26, 169–187. doi:10.1016/ S0163-2396(02)26012-3 Yik, M., Russell, J. A., & Steiger, J. H. (2011). A 12-point circumplex structure of core affect. Emotion (Washington, D.C.), 11(4), 705–731. doi:10.1037/a0023980 PMID:21707162 Yin, M., & Sun, Y. (2015). Human behavior models for virtual agents in repeated decision making under uncertainty. Proceedings of the 14th International Conference on Autonomous Agents and Multiagent Systems. Zhao, R., Papangelis, A., & Cassell, J. (2014). Towards a dyadic computational model of rapport management for human-virtual agent interaction. In Intelligent Virtual Agents. IVA 2014. Lecture Notes in Computer Science, (vol. 8637). Springer. doi:10.1007/978-3-319-09767-1_62

ADDITIONAL READING Feldman Barrett, L. (2017). How emotions are made: The secret life of the brain. Houghton Mifflin Harcourt. Hudlicka, E. (2008, August). Affective computing for game design. In Proceedings of the 4th Intl. North American Conference on Intelligent Games and Simulation (pp. 5-12). McGill University Montreal, Canada. Hudlicka, E., & Broekens, J. (2009, September). Foundations for modelling emotions in game characters: Modelling emotion effects on cognition. In 2009 3rd International Conference on Affective Computing and Intelligent Interaction and Workshops (pp. 1-6). IEEE. Marsella, S. C., Carnicke, S. M., Gratch, J., Okhmatovskaia, A., & Rizzo, A. (2006, August). An exploration of delsarte’s structural acting system. In International Workshop on Intelligent Virtual Agents (pp. 80-92). Springer, Berlin, Heidelberg. 10.1007/11821830_7 Popescu, A., Broekens, J., & Van Someren, M. (2013). Gamygdala: An emotion engine for games. IEEE Transactions on Affective Computing, 5(1), 32–44. doi:10.1109/T-AFFC.2013.24 Scherer, K. R., Bänziger, T., & Roesch, E. (Eds.). (2010). A Blueprint for Affective Computing: A sourcebook and manual. Oxford University Press. Vick, E. (2009). Emotion notions: Modeling personality in game character AI. Cengage Learning. Warpefelt, H., & Verhagen, H. (2015). Towards an updated typology of non-player character roles. In Proceedings of the International Conference on Game and Entertainment Technologies.

41

 Multi-Disciplinary Paths to Actor-Centric Non-Player Character Emotion Models

KEY TERMS AND DEFINITIONS Affective Loop: A cycle of sensing and reacting between an agent and its environment. It models the process of human interaction with things external to the body. Agency: The apparent freedom of a character to act, react, not act, or sublimate desire to achieve a goal within a fictional world. Agent: (In software development) An artificially intelligent software object that can monitor a program’s state and execute commands autonomously to achieve those goals. (In literary and performance theory) an object or being that appears to behave with an intention to achieve a goal. Actions taken toward goal attainment foreground the agency of the object or being. Appraisal Variables: A set of values categorically assigned to objects that an agent senses that can affect the emotional state of the agent. Computational Cognitive Psychology: A subdiscipline of Cognitive Psychology and Cognitive Science that draws from Neuroscience and Communication Theories to develop models of cognitive experience that can be programmed as computer-based application and tested with data collected from human subjects. Emotion Model: An abstract representation of the components that comprise the process for an agent of sensing appraisal variables in objects, identifying objects, processing objects in relation to beliefs, desires and intentions, and eliciting emotions. High Resolution Animation: An animation method, often assisted by motion capture technology, that uses very detailed movement to animate a photorealistic game character. Non-Player Character (NPC): Characters in a video game, excluding the one performed by the player during gameplay. Performative Mediation: The process of integrating information outside the domain of the game design for the construction of a character. An actor’s personal experiences, an awareness of the distorting features of camera optics. These are external information that often is considered in an actor’s performance preparation for a character and its role. Player Character: The character in a game performed by the player. Virtual Agents: An artificially intelligent software object programmed with an emotion model with the intent of teaching or assisting a user complete a task or learn a process.

42

43

Chapter 3

Extension of Clinical/ Psychological Approach Using Post Narratology: The Proposal of a Psychological Narratology and Review for Systematization Kai Seino National Rehabilitation Center for Persons With Disabilities, Japan Shun Ishizaki Keio University, Japan

ABSTRACT The purpose of this chapter is to examine the psychological approach from a narrative viewpoint. In addition, the authors discuss the possibility of collaboration with artificial intelligence (AI). They reviewed the narrative approaches in psychology. At result, they showed how the narrative approach in clinical domain was expanded. The following possibilities exist regarding a narrative generative system in a narrow sense: (1) It is the construction of a fluid system. (2) The systems generate a tale communally. (3) There are no clear beginnings and endings. (4) It is a narrative generation through a dialogue. Next, the following possibilities exist regarding a narrative generative system in broad sense: (1) It is the modeling of the stance of the narrative. (2) It is the modeling of narrative generation by the community. (3) It is the modeling of the process. (4) It is the study of the interlocutory peculiarity.

DOI: 10.4018/978-1-7998-4864-6.ch003

Copyright © 2021, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

 Extension of Clinical/Psychological Approach Using Post Narratology

INTRODUCTION In this chapter, we examine the potential of combining the narrative as a problem-solving method with computational techniques such as artificial intelligence (A.I.), and the applicability of the integrated approach in a clinical environment. As a general term, “narrative” is understood as the “act of discussing and being discussed using language (Spoken Language and written language, etc)” (Yamada, 2007b). “Narrative” can also be defined more narrowly as in the term “narratology,” which refers to the reproduction of real or fictional events and is always accompanied by time constraints (Prince, 2003). Bruner (1987) suggested that the narrative provides individuals with a method for understanding and thinking about their everyday lives and interactions with others. In recent years, the narrative approach has been utilized in clinical domains such as medicine, nursing, psychology, and welfare. The narrative may be thought of as an evidence base containing conventional, objective data. It may also be conceptualized as a narrative base, wherein individual subjectivity and originality are considered important, and this conceptualization of narrative leads to narrative-based medicine and narrative-based practice. Furthermore, an evidence base is considered objective science and is based on modernism, whereas a narrative base is relative, post-modernist, and post-structuralist. In the clinical domain, the approach is key to any narrative, and is called the narrative approach. For example, representation is a narrative therapy in psychotherapy. However, the conceptualization of a narrative in recent years has not been limited to the narrative approach and is considered to have a wider extent and possibility. One key to understanding the wider possibility of a narrative is the term “narrative generative system,” which is advocated by Ogata, the author and editor of this book. The narrative generative system conceived by Ogata not only considers the structure or form of a story, but also incorporates the process by which the story is constituted and received. In addition, “computational narratology,” one of the underlying theories of the narrative generative system, is a concept resulting from the fusion of computational technology and story. Authors have divided narratology into literary narratology and psychological narratology from the viewpoint of a narrative generation system (Seino 2016). A literary narratology considers the contents and form of a story, whereas psychological narratology is called a story treatment in the clinical domain and is a narrative approach. Against the backdrop of Ogata’s theory, Seino et al (2016, 2018, 2019) proposed a new technique for data mining designed to provide support for persons with disabilities. The technique provides a method for the analysis and generation of a story, through a combination of text-mining technology and the viewpoint of literary narratology. The results demonstrated that a theory could be established based on a narrative generative system. However, to investigate the practical implications of the system for the theory and method of a psychological narratology, a discussion of technology (e.g., computer, software, coding) is omitted. Therefore, the psychological approach is presented in this chapter from the viewpoint of a narrative, and aims at clarifying the possibility of a psychological narratology. As aforementioned, the purpose of this chapter is to explore the theory and the method of using narrative in the psychological approach. Medical treatment and support for any illness that exists in the clinical domain is defined as a psychological approach. In this chapter, the possibility of extending the psychological approach through collaboration with AI or cognitive science has been examined. Specifically, we review the research and practice of psychology, psychiatry, and a proximity domain in recent years and their use of narrative. Our discussion is broadly focused on the approach and practice relevant not only to narrative therapy but also to a narrative, or to narratology.

44

 Extension of Clinical/Psychological Approach Using Post Narratology

A DEFINITION AND THEORY OF A NARRATIVE Definition of “Narrative” A narrative has been defined in multiple ways, and there are also many theoretical positions on the topic (Riessman, 1991). For example, one definition of narrative is that it is an explanation of the experience of a continuous story or people (Bloor et all, 2006). A narrative may also be communicated using several methods (Clandinin, 2006). For example, it is possible for a narrative to originate from a document, such as an interview, a diary, a biography, or an ethnography (Liebich, Tuval-Mashiach et all, 1998). In this study, three types of narrative are defined: first, a literary narrative is a story found in literature, including mythology and folk tales; second, a personal narrative is a story of an individual and family; third, a social narrative is a social story about a group of people. In addition, narratology, which is the theory of a narrative, is defined in the narrowest sense in the literary theory of “literary narratology.” Moreover, a narrative may be used outside of the field of literature, as in a clinical domain such as psychology, medicine, nursing, and social work that provides services to support persons. The combination of narratology and the use of a narrative in a clinical domain is called “psychological narratology.” Both literary narratology and psychological narratology are explained in detail later in this chapter.

Literary Narratology as a Basic Theory The narrowest definition of narratology is regarded as a literary theory, which is the study of the nature, form, and function of narrative (Prince, 2003). This section gives an overview of the theories of narratology proposed by Propp (1968) and Genette (1980). Ogata (2010) organized narratology into five categories that establish the theoretical underpinnings of narratology: (1) “The Poetics” of Aristotle (1997); (2) structuralism, which developed based on the ideas of Saussure; (3) the literary art movement of Russian Formalism (Kuwano & Oishi, 1988); (4) British and American literary theory (Yamoto, 1974); and (5) the literary theory of structuralism (Tsuchida, Aoyagi, & Ito, 1996). Narrative comprises “story” and “discourse.” According to narratology (Prince, 2003), the story is the content of the narrative while the discourse concerns the expressive side. Narratology employs an analysis of both story and discourse, and the relation between the story and discourse has become the main theme of narratology. To illustrate, Russian Formalism, which is one of the theoretical underpinnings for narratology, regarded the components of narrative as being fabula (corresponding to the story) and sjuzet (corresponding to the discourse). Forster (1927) defined the components of narrative as “story” and “plot” (i.e., discourse). He observed that “story talks about a case and an event according to progress of time,” whereas the plot focuses on the causal relation of incidents and events. Propp (1968) described the fundamental components of any Russian folktale. In particular, he defined the narratological notions of “function” and characters’ “roles.” “Function” is a character’s act defined from the viewpoint of its significance in the story, and Propp described 31 types of “functions” that compose Russian folktales. Characters’ “roles” are classified based on their range of performable “functions” and, as a result, he described the following seven roles”: (1) villain, (2) dispatcher, (3) helper, (4) princess or prize and her father, (5) donor, (6) hero, and (7) false hero. Genette (1980) classified narrative into three components: story, discourse, and narrating. Furthermore, Genette (1980) divided discourse into three categories: (1) “tense,” which is the temporal relation between discourse and story; (2) “mood,” which is the reproduction of the story through the discourse; and (3) “voice,” which is the relationship between the narrating 45

 Extension of Clinical/Psychological Approach Using Post Narratology

and the narrated and the relationship between the narrating and text. Tense, mood, and voice are further divided into several sub-items. In applied studies of narratology, some researchers analyze and generate stories based on the theories constructed by Propp and Genette (Akimoto & Ogata, 2013; Ogata, 2011; Klein et al, 1974). For example, in the study of the narrative generation system, Ogata (2015) devised a conceptual dictionary that can provide the basic components of narrative; specifically, character, objects, and place. Regarding these applied studies, Ogata (1999) suggested a study framework called “computational narratology.” This research offered a literary study frame that fused an understanding of narratology with the generation of narrative through the use of AI and cognitive science, thereby broadening the concept of computational narratology.

Interdisciplinary Narrative Generative System In this section, “interdisciplinary narrative generative system,” which is derived from Ogata’s narrative generation system, is outlined. Considerable research on story generation has been conducted in the domains of cognitive science and AI. Ogata and Kanai (2010) have completed advanced research on the interdisciplinary approach for the narrative generative system, which introduced narratology and literary theory. In addition, Ogata (2011) proposed the concept for the narrative generative system, which continues to be developed at two levels: at a broad level, the narrative generative system constitutes a method for symbolically understanding a human being and society in aggregate; and at a narrow level, a narrative generative system consists of a computer program. Recently, research in the clinical domain, which is based on Ogata’s concepts, has been progressing. Specifically, the research is on disease and disability, and the results will be used for clinical support (Aoki et al, 2017; Seino et al, 2016, 2018). This research realizes the concept of Ogata’s narrative generative system at a narrow and broad level. Two underlying concepts in Ogata’s narrative generative system are extended literary theory and multiplex tale structure (Ogata, 2001). Extended literary theories unite various disciplines and include the dynamic state analysis of a phenomenon, the experimental creation of a new literary genre, and a concept that aims at deployment (Ogata, 2001). The descriptive method of an integrative meta-level is theorized as a model of a literary phenomenon. A multiplex story structure encompasses the idea of integrating the understanding and explanation of a literary phenomenon (Ogata 2000), and it is performed by multiplexing the process of generation-acceptance. The underlying idea of Ogata’s narrative generative system is the method of the meta-level which unifies a discipline. It is a model which aims at analysis and a certain creation of a phenomenon. In addition, with multiplex story structure, a phenomenon can be realized as a model that is understood and explained by the multiplexing of a generation-acceptance process. Authors argue about the kind of suggestion this idea may provide to practice and research in the clinical domain in the “Discussion” section.

The Research Framework for Narrative Several studies have suggested how to treat a narrative. The relationship between the concept of the narrative and the usage of the narrative varies (Clandinin, 2006). The content of the narrative fluctuates as well (Mishler, 1995). In a macroscopic position, the framework of a narrative includes all life activities. In a micro position, a narrative (story) expresses the causalities of an event (Riessman, 1993). In addition, the narrative approach is defined as “the method of making the concept of a narrative key and 46

 Extension of Clinical/Psychological Approach Using Post Narratology

Table 1. A humanities and social science classification system, and the contained discipline Classification

The contained discipline

1

Humanities / social science research that treats narrative directly

History, the philosophy of history, literary criticism, psychology

2

Humanities social science research relevant to narrative

European philosophy, hermeneutics

3

Practice research that utilizes narrative

Clinical psychology and sociology, risk psychology, economy and business administration, folklore, engineering works and city planning

presenting a certain phenomenon” (Noguchi, 2009). Conversely, researchers have criticized research on the narrative. Nelson (2011) noted that qualitative research based on the narrative was not fully discerned and the meaning of the research was not fully explained to the reader. Numerous disciplines study narratives. Fujii et al (2001) classified the research in connection with the story of the discipline of humanities and social science into three categories (Table 1). In addition, with regard to the entire discipline, Ogata (2018) classified the division between a scientific domain and a story into three more categories (Table 2). Table 2. The division between a discipline and narrative Classification

The contained discipline

1

Symbol level [Literature, art, and an entertainments domain]

Narratology, poetics, rhetoric, literary criticism, linguistics, folklore, folktale study, and an old tale, entertainment research, Noh play theory, kabuki theory etc., cultural anthropology and folklore, philosophy and thought, movie theory and cinematic review

2

A brain, a nerve, and a psychological level [Psychiatry and a psychological domain]

Psychology, psychiatry and psychopathology, psychoanalysis, tale treatment, neuro-psychoanalysis, brain science (The author added a postscript.)

3

Social level [Social domain]

Sociology, history, business administration, economics, jurisprudence, marketing theory, and advertising theory

Moreover, Mishler (1995) classified research into three categories as Models of Narrative Analysis (Table 3).

THE NARRATIVE IN CLINICAL / PSYCHOLOGICAL APPROACH A Clinical Narrative Turn and Narrative-Based Medicine (NBM) The purpose of this research is to contribute to problem-solving for clinical issues. Therefore, we examined psychological narratology from the viewpoint of a narrative, and discussed its extension and the possibility of applying it to other areas of research. The conventional clinical domain attaches importance to evidence and the basis for demonstrating that the evidence is objective. The concept of an evidence-based approach originated from evidence-based

47

 Extension of Clinical/Psychological Approach Using Post Narratology

Table 3. Models of narrative analysis Category

Subcategory

1

Reference and temporal order: The “telling” and the “told”

1. Recapitulating the told in the telling 2. Reconstructing the told from the telling 3. Imposing a told on the telling 4. Making a telling from the told

2

Textual coherence and structure: Narrative strategies

1. Textual poetics: Figuration, tropes, and style 2. Discourse linguistics: Oral narratives

3

Narrative functions: Contexts and consequences

1. Narrativization of experience: Cognition, memory, self 2. Narrative and culture: Myths, rituals, performance 3. Storytelling in interactional and institutional contexts 4. The politics of narrative: Power, conflict, and resistance

Source: Mishler (1995)

medicine (EBM) in the 1990s (Guyatt, 1991; Sackett, Richardson, Straus, Rosenberg, & Haynes, 1997), and there is objective evidence in this regard. Sackett et al. (1997) define EBM as the “the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of the individual patient.” After the introduction of EBM, the concept was extended to other fields such as nursing, welfare, psychology, and education. Conversely, the idea of “evidence-based” has been reasonably criticized and the following criticisms have been made: (1) it downplays qualitative studies on the hierarchy of evidence (Concato, 2004; Manuel, Fang, Bellamy, & Bledsoe, 2008); (2) in the framework of objective fact, it is all but impossible to acquire the opinions of socially-oppressed minorities, and therefore, this method fails to capture a pluralistic reality (Gergen, 1994a); and (3) it is also impossible to imitate the medical EBM model in a practice domain based on social sciences because the EBM model assumes only natural sciences as a research foundation (Yamabe, 2012). Based on the limits of objective facts, social constructivism, which provides the relativist conception of reality, has come to be recognized (Gergen, 1994b, 1999). Against the backdrop of social constructivism, a trend that allows a reconsideration of EBM from the perspective of the narrative is Narrative Based Medicine (NBM) (Greenhalgh & Hurwits, 1998). Saito (2003) identified the following five points as elements of the NBM approach: (1) illness as a story; (2) the patient as a storyteller; (3) the allowance of ambiguity of phenomena; (4) the insignificance of linear cause and effect (it is not considered important that a specific cause and effect should be joined together linearly); and (5) dialogue as a treatment. The narrative approach has paid close attention to the narrative understanding of NBM. Since the 1990s, the narrative approach has attracted greater attention in interpersonal help domains, such as medical care, nursing, and psychology. The development of epistemology and the methodology that attached importance to narrative is known as the “Narrative Turn.” (Denzin & Lincoln, 2000). Noguchi defined this narrative approach as “going closer to reality with the clues in the forms of narratives” (Noguchi, 2005). From this background, we may be able to develop an effective method if evidence and a narrative can be unified by the integration of quantitative and qualitative analysis in a clinical context. In this chapter, it may become a help of integration of evidence and a narrative to examine integration of psychological approach and computational technology by making a narrative into a key.

48

 Extension of Clinical/Psychological Approach Using Post Narratology

The Ideological Background of Narrative Approach The narrative approach is set against the context of the philosophy of post-modern thought and poststructuralism, and the theory of social constructionism. Post-modern philosophy originates in semiology or literary criticism from the first. Therefore, post-modern ideology has a tale and a deep relation. In addition, social constructionism was greatly impacted by the study of the concept of the text theory by Derrida (1978), and the power of Foucault (1975). The social constructionism way of thinking can be summarized in the following five points. (1) Reality is socially constituted by language. (2) It is social practice to begin to develop a meaningful language, and existence of the individual mind is not regarded as evident. (3) Meaningful discourse does not exist without a “relation.” If meaningful discourse does not exist, the “object“ and the “act” cannot occur. (4) Feelings such as “It is right” and “this is good” are produced out of a community. However, there are many instances in which “right” and “good” were limited to the community. (5) Any enforcements must be open to evaluation and criticism (Gergen, 1994). In other words, social constructionism brought about the paradigm shift that converted the method of actual recognition into relativism from conventional absolutism. These theories are used to analyze personal narratives and social stories in the context of society and culture. Lyotard (1979) described a category of classification of “grand narratives” that can be used to justify various supporting stories that are seen as “little narratives.” Foucault (1980) accounted for a classification of “a dominant story,” which influences a certain situation, and “an alternative story,” which appears when the dominant story is doubted. Yoshimoto’s (1968) community illusion may also be connected to dominant and alternative story. Therefore, it was claimed that a tale is an indispensable concept for the interpretation of man’s world or its historical recognition (Danto 1965).

REVIEW FOR SYSTEMATIZATION OF A PSYCHOLOGICAL NARRATOLOGY The system of a psychological narratology is proposed in this section, which shows that the clinical / psychological approach is extensible through post-narratology. Post-narratology is a narrative generative system that includes story generation by A.I. In order to actuate story generation, a literature review is performed with a clinical/ mental approach considering that the extension to post-narratology is possible. The system then constructs the story through AI and authors argue about the ability of a bridge to be carried out and whether by re-extending the clinical/mental approaches through story generation, a clinical contribution is possible. These approaches are concretely aimed at the approach (research and practice) in clinical domains, such as psychology. Moreover, what can be called a narrative approach is dealt with from the viewpoint of post-narratology, in which the possibility of extension of the present psychological narratology is proposed. A psychological narratology is the theory of the clinical/ psychological approach that is relevant to the narrative in the broadest sense that the author has proposed. Therefore, the following psychological narratologies are outlined and considered. ・Narrative therapy: A type of psychotherapy and family therapy/ The client pays to talk to the clinician. ・Self-help group: A type of group treatment in which one receives support from peers (companions or the parties concerned) ・Open Dialogue: A therapeutic intervention for an acute term mental disease, in which a dialogue is considered important. 49

 Extension of Clinical/Psychological Approach Using Post Narratology

・Tohjisha-Kenkyu (In Japanese) (Self-directed studies): An empowerment approach produced out of the life movement of the affected party.

Psychotherapy Outline In clinical psychology or the counseling field, a “narrative” has come to be associated with the narrative therapy of psychoanalysis, analytic psychology, a gestalt therapy, and family therapy since the 1980s. In a narrow sense, psychological narratology is narrative therapy in psychotherapy and family therapy. Narrative therapy was advocated by White and Epston (1990) and it developed through the context of family therapy in the 1980s. The core belief was an awareness of the importance of how narration and the tale of the client are respected. White and Epston (1990) applied a classification of Foucault to narrative therapy, aiming to excavate the new “alternative story” from the “dominant story“ that restricts life. This excavation is conducted by “externalizing” and seeking a “unique resolution” for the problem. “Formation of outside,” “discovery of unique outcome,” “re-writing,” a “remembering,” a “definition festival,” “outsider witness,” and a “therapeutic letter” are among the techniques used in this process (White & Epston 1990). In addition, Spence (1983) described the occurrence in which a narrator is in treatment, and a hearer hears the talk of a narrator as a tale. In analytic psychology, as with traditional clinic psychology, the role of a “tale,” like the myth, is touched upon by Kawai (2001), who presupposes that the process of psychotherapy is a tale. In addition, in Gestalt therapy, Polster (1987) presupposed that the recovery of harmony of the heart requires a dialog, and proposed the method of making achieving this harmony is by having a real person sit on an “empty chair” that is assumed to be there and engage in a dialog.

Consideration of a proposal Narrative therapy creates a narrative that provides the client with a tool for problem solving. A typical narrative approach heads toward the solution to a question as the client completes a tale. We propose the following possibility from the viewpoint of post narratology: what is treated is the discovery or description of a “dominant story” and an “alternative story,” and the technique of externalization. There are two implications of this suggestion. First, the “dominant story” and “alternative story” may be identified in order to construct the theory of a narrative generative system. Second, computational technology may be able to realize two stories, and if so, a narrative generative system may be expanded.

Other Psychological Narratologies There are multiple other psychological narratologies that have been advanced in the last 30 years. Elliott (2005) divided narratives into two classifications: a “first order narrative” and “second order narrative.” The former focuses on individual’s discussion of their own experience, whereas the latter centers on researchers’ discussion about a personal quest to understand the social world. McAdams and Aubin (1998) analyzed a recited life story and, through the classifications of “a story of recovery” and “a story of the turning point,” identified fixed forms, such as “sacred stories.”

50

 Extension of Clinical/Psychological Approach Using Post Narratology

Moreover, there is a theory known as “literary narratology about story and discourse,” that includes psychological narratology. Bruner (1986) classified forms of narrative into the (1) “narrative mode,” which includes the time-linking information about an event, and (2) “paradigmatic mode,” which includes information about the necessary relations between multiple events as well as causal relations. Morgan (2000) views a story (corresponding to the narrative) as “an event connected on temporal axes in succession which becomes a plot (corresponding to the discourse).” Adam (1999) contended that a narrative is composed of time, characters, predicates, and events that can change the predicates. Yamada (2007a) distinguished four levels of narrative: (1) real, (2) mutual action, (3) text, and (4) model. In line with Yamada’s classification, Shojima (2008) classified narrative into four acts: (1) the side of mutual acts, (2) the side of the social act, or social historic linkup, (3) something that cannot be discussed because it exists outside of the narrative, and (4) multiple lives or voices surrounding the life of the narrator. Some studies have examined the narrative approaches of persons with disabilities. For example, Tagaki’s (2008) study examined the speech of persons with disabilities, whereas others have focused on the psychological details of schizophrenia (Komori & Yamada, 2001), narrative approaches to mental disorders (Buchanan-Barker & Barker, 2008; Vuokila-Oikkonen, Janhonen, & Vaisanen, 2004), the speech of aphasic persons (Nouchi, 2008), and the life of a small child undergoing treatment for cancer (Saiki, 2008). Studies of these narrative approaches have thus far used psychological narratology. However, these earlier studies, except for the current authors’ research, are yet to closely examine the term “narrative” or to adopt a literary narratology method.

Psychological Narratologies Other Than Narrative Therapy Additional clinical approaches are discussed below that are not narrative therapy, but nevertheless contain what the authors view as psychological narratology. These approaches are related to a tale and are considered useful for the extension of a narratology and a study of a narrative generative system.

Self-help Group Outline A self-help group is practiced among those who suffer from the same illness and offers support through peer relations. For example, self-help groups may be formed for individuals addicted to alcohol, drugs/ psychotropic substances, or gambling, or who are dealing with the same life obstacle or chronic malady. In recent years, self-help group have come to be viewed as a form of a narrative approach (Noguchi 2002). “Alcoholics Anonymous (AA)” was the first self-help group, which was developed in the United States in 1935, and is for recovery from alcohol addiction. AA was effective for treating alcohol addiction, and over time, many other groups adopted the method used by AA to deal not only with addiction or dependence but also for other problems. The typical definition of a self-help group is that of Kats &E. I. Bender (1976), who defined a self-help group as a small group of companions (peers) who provide mutual aid for the achievement of a specific purpose. AA meetings are regularly scheduled and the members who gather at each meeting talk in turn about various topics. The talk may be of experiences from one’s past or the utterance of a single word may constitute a speaking turn (Noguchi 1996). The self-group does not have a technical definition (Ito, 2000).

51

 Extension of Clinical/Psychological Approach Using Post Narratology

Little consideration The acts that a self-help group perform are the primary methods of a group. With the aim of aiding recovery, the talk among members is considered a narrative approach. Sakiyama (2018) indicated that self-help group research was a domain where research was advanced from the viewpoint of narratology. In addition, Sakiyama (2018) arranged the self group as it is what constructs “alternative story” ; it compared with “dominant story” which a professional makes. The first-line specialty of support of a self-help group is “experiential knowledge.” Experiential knowledge is the knowledge based on a member’s experience, and serves as a source of support for a self-help group. The knowledge of members is more practical and more comprehensive than the knowledge of professionals (Borkman 1976). Each member’s “experiential knowledge” is discussed and the collective knowledge if the group is accumulated and utilized as one tale knowledge from a group. We can interpret sharing of knowledge as the process of making one big narrative from two or more members’ narratives. The process may be adopted into theory as a narrative generative system and may be realized as a system. For example, studies have proposed collect many texts that were spoken and generating the tale based on some typical elements. In addition, Ito (2000) examined a self-help group from the perspective of an individual tale. The results of the study indicated that the tale should have been regarded as an object with the effect of showing not the collective knowledge system of a group, but of a participant’s self. In other words, participants produced or consumed “the good talk,” through which each participant realized a sense of togetherness with self-excellence and the group. To determine the degree of change for each participant, it was presupposed that it was important to determine the standard by which a participant made a tale “the good talk.” In other words, it was important for a tale or suggestion to achieve a set basis for valuation. An evaluation of the tale could be conducted based on how the group absorbed the tale and how it affected the production and consumption of future tales.

Open Dialogue Outline In recent years, Open Dialogue (OD) is another psychological approach that has attracted attention in Japanese psychiatry. OD started in Finland in the 1980s and is a method of therapeutic intervention for an acute-term mental disease. However, an open dialog is not a mere technique or system. Seikkula, a central figure who developed the open dialog emphasizes that OD is a “technique,” “philosophy,” and a “way of thinking,” but not a “medical treatment process.” In OD, two or more staff conduct a crisis intervention within 24 hours of a request. Patients and families practice OD every day by sitting in a circle and performing “OD.” OD seldom uses a drug but makes a patient’s critical situation cease (Saio 2014). All matters are discussed and determined as a characteristic portion of OD at the members’ meeting place. In addition, an expert team discusses a patient and a family (exchange) in their presence, as they reflect on what is being shared with them. Olson, Seikkula, and Ziedonis (2014) summarized the following key OD items: The Twelve Key Elements of Fidelity to Dialogic Practice in Open Dialogue (Olson, Seikkula, and Ziedonis, 2014) 1. Two (or More) Therapists in the Team Meeting 52

 Extension of Clinical/Psychological Approach Using Post Narratology

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

Participation of Family and Network Using Open-Ended Questions Responding to Clients’ Utterances Emphasizing the Present Moment Eliciting Multiple Viewpoints Use of a Relational Focus in the Dialogue Responding to Problem Discourse or Behavior in a Matter-of-Fact Style That is Attentive to Meanings Emphasizing the Clients’ Own Words and Stories, Not Symptoms Conversation Amongst Professionals (Reflections) in the Treatment Meetings Being Transparent Tolerating Uncertainty

Little Consideration OD is considered to be a narrative approach given that a conversation called a dialog becomes a key medical treatment. OD contains narrative therapy and a common feature of OD is that it has an ideological background. According to Saito (2014), OD has various philosophical bases. It is a basedof Bateson’s double bind theory, Bakhtin’s polyphony theory of a dialog principle, and Vygotsky’s theory of psychology. In addition, according to Saito (2014), OD is perceived of as a cure within the context of critical psychiatry or post psychiatry (Thomas & Braken, 2004). Post psychiatry considers the technique of scientific psychiatry and seeks new moral medical treatment, given that the present psychiatry is criticized. In addition, it is also a common feature with narrative therapy to use reflecting. The 12 elements of OD consider tale generation and its effect, and the process is thought provoking. A dialogue is performed by two or more members of various positions. In addition, open-ended questions are used as an interlocutory trigger. OD may be able to be realized as an automatic dialogue system or a narrative system by speech recognition. By pulling out two or more viewpoints and observing relationships, an automatic dialogue system may be able to return utilize an extraction method to return an element as the narrative is being generated.

Tohjisha-Kenkyu (Self-directed studies) Outline In recent years, the “Tohjisha-Kenkyu” approach that originated in Japan has attracted significant international attention. Tohjisha-Kenkyu may be introduced as an open dialog in Japan (Saito 2016) and is an approach commonly used with patients with schizophrenia. In recent years, it has also been developed as a treatment for persons with developmental disabilities. Tohjisha-Kenkyu is an empowerment approach, and it happened among the activity and liveing of the person with disabilities who held schizophrenia etc (Mukaiyachi, 2013). There are seven main characteristics of Tohjisha-Kenkyu (Mukaiyachi, 2013). (1) Tohjisha-Kenkyu is a self-help program for the parties concerned, such as individuals with schizophrenia who live in the local region. The party concerned becomes “a hero of difficulties.” (2) The wisdom that dissolves difficulty is in one’s experiences of the experiences of one’s companions. (3) The areas of exploration are 53

 Extension of Clinical/Psychological Approach Using Post Narratology

the difficulties of life faced by the concerned party. (4) The party concerned with a disability investigates his symptoms with companions. They cooperate with all involved parties and talk with one another to explore difficulties in an open-minded manner without judgment. Occasionally they develop a figure (picture) and action plan. An important discovery is the identification of a pattern and the mechanisms by which a difficult even occurs and the meaning and possibility that exist in a difficult context. (5) While repeating positive trial and error, a unique understanding and idea that are produced improvisatorially may lead to an important discovery of “how to help oneself.” The means for helping oneself are found there, in a unique understanding of oneself. The participant values sharing and implementing their discoveries in life and sharing with a friend. (6) It is a method of mere “problem solving.” It changes one’s attitude toward facing the occurrence considered to be a problem, and determines how to resolve it. It examines and cherishes the “language“ that related the experience in which a person has lived, and lives now. It can be called a “program of language” that transforms “being connected” with detailed “acting” into an actual life scene. (7) The Tohjisha-Kenkyu program is necessary and effective for a supporter.

Little consideration Tohjisha-Kenkyu provides suggestions for examining a narrative generative system and how to connect narrative generation for problem solving. In Tohjisha-Kenkyu, it is assumed that the party concerned will themselves become a hero of difficulties. In addition, Tohjisha-Kenkyu allows the narrative to pass and has suggested that it is a narrative approach tied to the problem solving of the party concerned. In addition, Noguchi (2002) reported that Tohjisha-Kenkyu is “a community of narration” that makes it “the community of the tale.” Many audiences are in a “narrative community,” and new narration becomes more certain with an existing audience. A narrative community is the space wherein new narration is shared and established. In addition, the aspects of changing attitudes, and how to identify and position attitudes Tohjisha-Kenkyuare are also common in narrative therapy. Moreover, the conceptualization of one’s own experience and that of a friend is similar to the notion of “experiential knowledge” in a selfhelp group. In addition, when considering tale generation, an improvisatorial, unique understanding or an idea is important and shows how to utilize the result, which creates a “relation“ as a “behavior“ in a scene received by the narrative generative system.

Approaches relevant to other narratives The approaches relevant to other tales are enumerated as a reference. Arai (2014) identified narrative therapy, collaborative therapy, and reflection as narrative approaches from the perspective of social work. Arai (2014) also reported that narration of the illness, the self-help group Tohjisha-Kenkyu, the human library, and the narrative interview are approaches related to the narrative approach. Other than this, authors show “the social story for an autistic child (Gray 2000) “, and “the Wellness Recovery Action Plan for a person with disabilities (Copeland 2002)” as approach relevant to a narrative. In addition, Kato (2010) reported that an item on narrative was considered in preparation of ICD-11 (the International Statistical Classification of Diseases and Related Health) by the WHO (the World Health Organization). However, the item on narrative was not adopted into ICD-11, which was released in 2018.

54

 Extension of Clinical/Psychological Approach Using Post Narratology

PSYCHOLOGICAL APPROACH AND NARRATIVE GENERATION BY A.I. In this section, two studies of the clinical/ psychological approach and the possibility of narrative generation by A.I. are introduced.

Narrative Generation For Employment Support of Persons With Disabilities In this section, the authors’ research is introduced, in which we extracted text elements and employed a system of story generation using A.I. Specifically, we analyzed a disabled person’s free descriptive answer to an interview question using the method of “informatics tale analysis” to extract the elements of the story. This method combines the technology of text mining, and some theories of a literary narratology. Text mining involves the quantitative analysis of text data based on natural language processing or data mining. After we extracted the element of the story from the individual’s text response, the system proposed creating a collectivestory. An example of the narrative is shown below. Please note that to protect respondents’ privacy, some answers are altered, but the basic meanings remain unchanged. In the authors’ study, both the wider and narrow senses of narrative were used depending on the particular contexts of the analysis. [Concerning work description] In response to the question: “Describe your job in detail.” Example 1: Since my legs were injured, I have been assigned paperwork. I am now engaged in computer work. (Physical disability) [Concerning care and support at the workplace] In response to the question: “Specify the care and support provided in your workplace as a means of assisting your career.” Example 2: At my workplace, employees often get acquainted with coworkers at informal meetings, such as drinks parties. I did not attend these parties because of my disease, so I had no opportunity to get to know them. The other day, I took the opportunity of participating in one, and it worked much more effectively than I had expected. (Mental disorder) Example 3: I first wanted to work without mentioning my disease. The other day, a relative of a person who had supported me asked me to work in their office. When I began working there, I soon found that this relative did not care about my condition, so it was a disastrous year. From this experience, I realized that I should conceal my disease because people consider me to be inferior to healthy people. (Intractable disease) The authors’ study analyzed the descriptive texts that reflected the personal stories of persons with disabilities, and clarified social stories of disability based on the results of multivariate analysis involving text mining. Note that the descriptive texts analyzed in this study were narrative in the narrow sense while others were narratives in the wider sense. This is an example from our research entitled “Employment Support and Story Generation for Students with Developmental Disabilities.” The results indicated that story generation tends to materialize through the combining of existing elements with those from previous research (Seino, Enomoto, & Miyazawa, 2018). The figure below shows a story generative system. The analysis of job descriptions analysis results were used to create the image. For cases in which this system was created, it was important to consider how elements were combined. The probability of a story being generated is attained by calculating the probability that it may happen and the strength of the relationship between the text elements. In addition, in order to utilize the result of the support generation simulation, it is important to understand what kind of elements have been collected. For example, 55

 Extension of Clinical/Psychological Approach Using Post Narratology

Figure 1. The image of the job description story generative system

the elements of support needs may be collected based on responses to a specific problem, and it may be possible to examine what kind of problem and support needs occur at what time, thus enabling the preparation of appropriate support.

Narrative Generation for Paper Creation Support of a Student With a Developmental Disease Tendency Joint research by Ogata and Aoki et al. (Aoki, et al. 2018) proposed students’ paper creation support and narrative generative system based on the cognitive tendencies and action patterns of individuals with autism spectrum disorders. Aoki et al (2018) suggested that a paper creation might address the problem of a student with autism spectrum disorder. they explained why the problem occurs based on the cognitive tendencies and the action patterns of people with autism spectrum disorder, who may be sensitive to “a surprise and a gap” in the “discontinuous nature” of a story. Therefore, a story (paper creation), after making it continuous, stops the reaction of a student with autism spectrum disorder. The evaluation mechanism was set up and support was proposed for advancing a story by issuing directions on a macro and a micro level.

56

 Extension of Clinical/Psychological Approach Using Post Narratology

DISCUSSION This section has following discussions. First, literary narratology as basic theory is compared to psychological narratology as a practice or application theory. Second, the extension and possibility of a psychological narratology are considered.

Comparison of Two Narratologies Literary narratology focuses on a literary narrative while a psychological story centers on social and personal stories. Both literary and psychological narratologies share some characteristics. For example, Russian Formalism, Forster and Bruner, and Morgan all reached the consensus that narrative can be classified into two elements: (1) those aspects of the narrative that align along the event in a time lapse (corresponding to the story) and (2) those aspects of the narrative that align along the event in a causal relation (corresponding to the discourse). Additionally, an interest in the components of the story is seen commonly in Propp (1968), Greimas (1966), Ogata (2015), and Adam (1999). The story consists of characters (including their roles and actions), functions, time, objects, places, predicates, and events that can affect the predicates. Additionally, Genette’s (1980) classification of “voice” is divided into the time for narrating, a standard, and the persona of the narrator. The words and narrative have similar elements as the classification by Elliot (2005), Yamada (2007a), and Shojima (2008). In other words, literary narratology and psychological narratology may have a common viewpoint. Conversely, there is a distinct difference between literary and psychological narratologies. Unlike literary narratology, psychological narratology analyzes social and personal narratives, allowing the relationship between individuals, their culture, and society to be clarified.

Extension and Possibility of a Psychological Narratology In this chapter, the psychological narratology was divided into some items and reviewed. The authors showed that the narrative approach in clinical domain, defined as psychological narratology, went beyond the narrative therapy in psychotherapy and was expanded to various experiments and research. Moreover, is thought that four common features are among the psychological narratologies reviewed in this chapter. As the first, many practice is not simple technology and method but stance, and a way of thinking. It is thought that this is related to narrative approach of the background of philosophy or an ideology. The approach will be fixed supposing if it is technology and a method. However, in fact, since it has narrative approach against the background of social constructionism, fixation is avoided and relativized. As the second, the narrative can point out being made communally. . Moreover, if it is a specialist and the concerned party, the concerned party will be the hero of the tale, and will become a narrator. In narrative therapy, the alternative story of a client is reconstructed in the separation between a therapist and a client. In addition, in a self-help group, an open dialog, and in Tohjisha-Kenkyu, a community is a group completes a tale. The third point is that it is considered to be an effective form of treatment that provides an account of its support effect. In narrative therapy, there is temporary directivity of an alternative story of being strong and carrying on. However, in all practice, the contents or forms which are told are not uniquely absolute and are relative. It is thought that the process of the practice has produced an effect. The practice is free dialogue among members of the community and the concerned party. The

57

 Extension of Clinical/Psychological Approach Using Post Narratology

fourth point may be obvious if it seen from the method of psychological approach, but all the tales are built through a dialog. They were not monologues and were constructed through colloquial expressions. Based on these points, the narrative generative system is considered in both a narrow sense and a broad sense. The following possibilities exist regarding the automatic tale generation machine realized as a narrative generative system in a narrow sense, i.e., a computer program etc. (1) It is the construction of a fluid system. For example, a random element is taken in accepted. This system carries out narrative generation by combining elements. (2)Two or more persons are the systems that generate a tale communally. (3) There are no clear beginnings and endings. For example, it increases the flexibility of an item or an element. Since the generation of a narrative, it can be conceived so that the system aims at the process of generation itself. (4) It is a system of the tale generation through simple dialogue but not an input. In addition, the following is mentioned as a suggestion for the practice of this research. A psychological narratology is not merely a methodology but is also the mechanism for converting of a relative sense of values. It is given significance with conversion (or improvement) of the worth of the tale of the party concerned and the party concerned. The following specific suggestions are based on such a system. (1) Consider the community itself which generates a tale and utilizes it to be a system and as well as its construction and practical use. (2) The idea of a system and the realization that this idea places more value on the process rather than the contents of tale generation. (3) Get participation and collaboration of the concerned parties towards realization and practical use of a system.

FUTURE RESEARCH DIRECTIONS This chapter reviewed research on the narrative approach of a clinical domain, and aimed to utilize the actual extension of a narrative generative system. Finally, it did not show a clear objective conclusion like the target narrative approach. The following suggestions should be used to guide future work in this area: 1. Examine the ideology and the philosophical background of a psychological narratology. 2. It is treating overall an ideology and philosophy, and a literary narratology and a psychological narratology. 3. Consider research of tales from the humanities and social science system as a reference for psychological narratology. 4. Consider the narrative approach of a clinical domain more broadly as a psychological narratology. In addition, the following are suggested as future research possibilities: a. Modeling and implementation of a system as a narrative generative system in a narrow sense (system which produces the detailed contents of language) from suggestions in this chapter. b. Carry out modeling or elaboration of the existing system based on suggestions from this research as a narrative generative system (system which recognizes an actual phenomenon and computes a meaning) in a broad sense.

58

 Extension of Clinical/Psychological Approach Using Post Narratology

CONCLUSION This research defines the narrative approach of a clinical domain as a psychological narratology, and examines its extension and possibility. As a result of the review, psychological narratology was confirmed to have widely expanded beyond the so-called narrative therapy of psychotherapy and family therapy. In particular, the self-help group, the open dialog which attracted attention in recent years, and Tohjisha-Kenkyu also have many common features with narrative therapy, and are seen to be using the same narrative approach. In addition, suggestions from these practices are thought-provoking and also utilize expansion, realization, and the system of the narrative generative system in a narrow sense and a broad sense for actual problem solving. It is meaningful to consider the practice of a clinical domain from the viewpoint of a narrative, as well as in respect to clinical reduction. I expect that expansion of new discovery and practice will be performed because the party concerned, a supporter, the clinical man of deeds, and a researcher in connection with a theme, and the researcher of related domains, such as A.I., computing science, cognitive science, literary theory, and philosophy, will argue and collaborate in the future.

REFERENCES Adam, J. M. (1999). Le récit [The narrative]. Presses Universitaires de France. Akimoto, T., & Ogata, T. (2013). Towards a discourse mechanism in narrative generation system: Proposal of a system introducing narrative discourse theory and reception theory. Cognitive Studies, 20(4), 396–420. Aoki, S., Ogata, T., & Ono, J. (2018). Narrative generation related to cognitive patterns seen in ASD -From the perspective of “Surprise”. In Proceedings of the 35th Annual Meeting of the Japanese Cognitive Science Society (pp. 652-660). Academic Press. Aristotle. (1997). Poetics (N. Matsumoto & M. Oka, Trans.). Iwanami Shoten. (Original work published 335 B.C.) Borkman, T. (1976). Experiential knowledge: A new concept for the analysis of self-help groups. The Social Service Review, 50(3), 445–456. doi:10.1086/643401 Bruner, J. (1986). Actual minds, possible worlds. Harvard University Press. Buchanan-Barker, P., & Barker, P. J. (2008). The Tidal Commitments: Extending the value base of mental health recovery. Journal of Psychiatric and Mental Health Nursing, 15(2), 93–100. doi:10.1111/j.13652850.2007.01209.x PMID:18211556 Clandinin, D. J. (Ed.). (2007). Handbook of narrative inquiry: Mapping a methodology. Sage Publications Inc. doi:10.4135/9781452226552 Copeland, M. E. (2002). Wellness recovery action plan (2nd ed.). Peach Press. doi:10.1300/J004v17n03_09 Danto, A. C. (1965). Analytical philosophy of history. Cambridge University Press. Denzin, N. K., & Lincoln, Y. S. (Eds.). (2000). Handbook of qualitative research. Sage Publications. 59

 Extension of Clinical/Psychological Approach Using Post Narratology

Derrida, J. (1978). Writing and difference. London: Routledge & Kegan. (Original work published 1967) Elliott, J. (2005). Using narrative in social research: Qualitative and quantitative approaches. Sage Publications. doi:10.4135/9780857020246 Forster, E. M. (1927). Aspects of the novel. Edward Arnold. Foucault, M. (1980). Power/knowledge (C. Gordon, L. Marshall, J. Mepham, & K. Soper, Trans.). Pantheon Books. (Original work published 1972, 1975, 1976, 1977) Genette, G. (1980). Narrative discourse: An essay in method (J. E. Lewin, Trans.). Cornell University Press. (Original work published 1972) Gergen, K. J. (1994a). Toward transformation in social knowledge. Sage Publications. Gergen, K. J. (1994b). Realities and relationships. Harvard University Press. Gergen, K. J. (1999). An invitation to social construction. Sage Publications. Gray, C. A. (2000). The new social story book: Illustrated edition. Future Horizons Inc. Greenhalgh, T., & Hurwitz, B. (Eds.). (1998). Narrative based medicine. BMJ Books. Guyatt, G. H. (1991). Evidence-based medicine. American College of Physicians Journal Club, 114, A-16. Ito, T. (2000). Serufu herupu gurupu to kojin no monogatari [Self-help groups and personal stories]. Japanese Sociological Review, 51(1), 88–103. Kato, S. (2010). Gendai seishinigaku niokeru rejiriansu no gainen no igi [Significance of the concept of the resilience in present psychiatry]. In Resilience: The new paradigm of present (pp. 1-23). Kanehara & Co., ltd. Katz, A. H., & Bender, E. I. (1976). Self-help groups in western society: History and prospects. Journal of Behavioural Science, 12(3), 265–282. doi:10.1177/002188637601200302 Kawai, H. (2001). Shinriryoho niokeru Monogatari no Igi [Significance of the “narrative” in psychotherapy]. Japanese Journal of Psychotherapy, 27, 3–7. Komori, Y., & Yamada, M. (2001). Seishinbunretsubyou no kazokushinrikyouiku niokeru naratibu apurochi [Narrative approach in family psychology education of schizophrenia]. Family Therapy Research, 18(2), 143–150. Lyotard, J. F. (1979). La condition postmoderne [The postmodern condition]. Minuit. Manuel, J., Fang, L., Bellamy, J. L., & Bledsoe, S. E. (2008). Evaluating evidence. In R. M. Grinnell & Y. A. Unrau (Eds.), Social work research and evaluation: Foundation of evidence-based practice (8th ed., pp. 482–495). Oxford University Press. McAdams, D. P., & De St. Aubin, E. (Eds.) (1998). Generativity and adult development: How and why we care for the next generation. American Psychological Association. Mishler, E. G. (1995). Models of narrative analysis: A typology. Journal of Narrative and Life History, 5(2), 87–123. doi:10.1075/jnlh.5.2.01mod

60

 Extension of Clinical/Psychological Approach Using Post Narratology

Morgan, A. (2000). What is narrative therapy? An easy-to-read introduction. Dulwich Center Publication. Mukaiyachi, I. (2013) Tojisha-kenkyu toha: Tojisha-kenkyu no rinen to kousei [What is a TojishaKenkyu? The concept and structure of a Tojisha-Kenkyu]. Tojisya-Kenkyu network (https://toukennet. jp/?page_id=56) Nelson, C. D. (2011). Narratives of classroom life: Changing conceptions of knowledge. TESOL Quarterly, 45(3), 463–485. doi:10.5054/tq.2011.256799 Noguchi, Y. (Ed.). (2005). Naratibu no rinshoshakaigaku [Clinical sociology of narrative]. Keiso shobo. Ogata, T. (1999). Proposal of computational narratology. In Proceedings of IPSJ SIG Computers and the Humanities (pp. 53-60). Academic Press. Ogata, T. (2010). “Monogatari seisei sisutemu” no haikei oyobi monogatari to bungaku no aida [Backgrounds of the “narrative generation system” and “relations between narrative and literature”]. In T. Ogata & A. Kanai (Eds.), Monogatariron no jouhougaku josetsu: monogatari seisei no shisou to gijutsu wo megutte [An introduction to the informatics of narratology: about the thought and technologies of narrative generation] (pp. 186–258). Gakubunsha. Ogata, T. (2011). Narrative generation system as the practice of “informatics of narratology.”. Journal of Japan Society for Fuzzy Theory and Intelligent Informatics, 23(5), 676–685. doi:10.3156/jsoft.23.5_676 Olson, M. E., Seikkula, J., & Ziedonis, D. (2014). The key elements of dialogic practice in open dialogue: Fidelity criteria. https://www.umassmed.edu/globalassets/psychiatry/open-dialogue/keyelementsv1.109022014.pdf Polster, E. (1987). Every person’s life is worth a novel. W. E. Norton & Company Inc. Prince, G. (2003). A dictionary of narratology (Revised ed.). University of Nebraska Press. Propp, V. Y. (1968). Morphology of the folktale (L. Scott, Trans.). University of Texas Press. (Original work published 1928) Riessman, C. K. (1991). Beyond reductionism: Narrative genres in divorce accounts. Journal of Narrative and Life History, 1(1), 41–68. doi:10.1075/jnlh.1.1.04bey Riessman, C. K. (1993). Narrative analysis. Sage. Saiki, S. (2008). Shounigan no kodomo no toubyoutaiken [Narrative based medicine and clinical intellect? Narrative of a pediatric cancer patient]. In Y. Yamada (Ed.), Jinsei to yamai no katari [Narrative of life and illness] (pp. 103–132). University of Tokyo Press. Saio, T. (2014). Open dialogue approach to acute psychosis: Is its effectiveness well-established? Clinical Evaluation, 42(2), 531–537. Saito, S. (2003). Naratibubeisudomedisun to rinshochi [Narrative based medicine and clinical intellect]. In Y. Yamada (Ed.), Jinsei to yamai no katari [Narrative of the life and the illness] (pp. 133–163). University of Tokyo Press.

61

 Extension of Clinical/Psychological Approach Using Post Narratology

Seino, K., Enomoto, Y., & Miyazawa, S. (2018). Narrative analysis of employment support for students with developmental disabilities: Using an objective analysis of free-expression answers. In T. Ogata & S. Asakawa (Eds.), Content generation through narrative communication and simulation (pp. 341-357). Information Science Reference (IGI Global). Seino, K., Haruna, Y., & Ishizaki, S. (2016). Employment status and support needs of persons with disabilities in Japan: An analysis of narrative using narratology and text mining on a national survey. In T. Ogata, & T. Akimoto (Eds.), Computational and cognitive approaches to narratology (pp. 245-275). Information Science Reference (IGI Global). Shojima, S. (2008). Multilayered contexts in which narrative is generated and interpreted: Toward the next narrative based inquiry. Kyoto University Research Studies in Education, 54, 138–151. Spence, C. D. (1983). Narrative persuasion. Psychoanalysis and Contemporary Thought, 6, 457–468. Tagaki, M. (2008). Experiences of people with acquired long-term physical disability: Life stories of persons with spinal cord injuries in Japan. Qualitative Psychology Nexus, 7, 51–79. Thomas, P., & Braken, P. (2004). Critical psychiatry in practice. Advances in Psychiatric Treatment, 10(5), 361–370. doi:10.1192/apt.10.5.361 Vuokila-Oikkonen, P., Janhonen, S., & Vaisanen, L. (2004). “Shared-rhythm cooperation” in cooperative team meetings in acute psychiatric inpatient care. Journal of Psychiatric and Mental Health Nursing, 11(2), 129–140. doi:10.1111/j.1365-2850.2003.00679.x PMID:15009487 Yamabe, S. (2012). Sosharuwaku niokeru ebidensu tohananika: Purorogu [What is the evidence in the social work?: Prologue]. In Japanese Society for the Study of Social Welfare (Ed.), Taironshakaihukushigaku [Discussion social welfare studies] (pp. 85-87). Chuohoki Publishing. Yamada, Y. (2007a). Shitsutekikenkyuu niokeru taiwatekimoderukouseihou [Interactive model construction method in qualitative research]. Japanese Journal of Qualitative Psychology, 6, 174–194. Yamada, Y. (2007b). Shitsutekishinrigaku no houhou: Katari o kiku [Method of qualitative psychology: Listening to narrative]. Shinyousha. Yoshimoto, T. (1968). Kyodo Gensoron [The theory of communal fantasy]. Kawade Shobo Shinsha.

KEY TERMS AND DEFINITIONS Literary Narrative: A narrative found in literature, including mythology and folktales. Literary Narratology: The theory about story and discourse in the narrow sense. Narrative in the Narrow Sense: A reproduction of a real or fictional event that is always accompanied by time restraints. Narrative in the Wider Sense: The individual’s act of narrating using his or her own language. Psychological Narratology: The theory of the clinical/ psychological approach that is relevant to the narrative in the broadest sense.

62

63

Chapter 4

Defining the Verbs for “Understanding and Interpretation” of Japanese Sake Hiroki Fukushima Kyushu Women’s University, Japan

ABSTRACT In this chapter, the author attempts to define the verbs in the description of Japanese sake taste by employing 1) a usage-based approach, 2) “encyclopedic semantics” rather than a “dictionary view,” and 3) sense-making theory, drawing on data from a “sake tasting description corpus” (approximately 120,000 words). The chapter selects eight verbs of high frequency (e.g., hirogaru ‘spread’) and defines their sense(s) in a bottom-up and abductive fashion, based on a score indicating the strength of co-occurrence between terms. In this study, the authors deal with the verbs for “Understanding” or “Interpretation ”; it means, verbs that contribute to narrating the personal, individual story (contents) of the tasters. This study suggests the verbs for understanding have senses related to [Timeline] and [Space]. On the other hand, verbs do not tend to collocate with [Movement] and interestingly, the [Structure], as same as the tendency of adjectival-nouns.

INTRODUCTION In this study, the author attempts to give encyclopaedic definitions to the verbs in a Sake corpus based on Sake tasting related terminology. This study is in a series of Sake terminology studies. In the previous study (Fukushima 2020b), the author tried to give encycropaedic definitions to adjectives and adjectivalnouns. In this terminology series, the author emphasizes the importance of predicates (verbs, adjectives, and adjectival nouns). They are the starting point of our cognition of taste or flavor as an “event”. As is often the case with wine sommeliers, the expression or description in Sake tasting tends to be regarded as a reductional ‘flavor-finding’ process. However, in this study series, the author respect Sake tasting expressions as the fruits of the dynamic event construction and sense-making process. DOI: 10.4018/978-1-7998-4864-6.ch004

Copyright © 2021, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

There are some Japanese linguistic studies on Sake taste terms (e.g., (Matsuura, 1992; Otsuka, Suwa, & Yamaguchi, 2015; Utsunomiya, Isogai, Iwata, & Nakano, 2006), but the investigation is in its infancy, largely because of the historical context of the Sake brewing industry. As the flavors of Sake have expanded, more expressions have been required. However, the study of the development of verbal expressions for Sake has been neglected. Technical terms for brewing and descriptive terms used to indicate some of the off-flavors of a Sake are emphasized, leaving terms to describe appealing flavors unstudied. In response, Fukushima (2014) compiled a small encyclopedia listing Sake taste words. Considered epoch-making in the Sake industry,1 this work has inspired other work, leading to the development of a method of defining Sake taste terms.

BACKGROUND Theoretical Background The author regards cognitive content generation of taste as a kind of aesthetic appreciation process. Sake, wine, coffee, cigarettes or other items have communities of lovers, from professional tasters to amateur enthusiasts. They express differences in taste or flavor by brands and vintage years, and many comment-supporting tools have been proposed. One typical example is “flavor wheels”. The process of aesthetic appreciation of these items is as follows: Aesthetic experience ® reflection ® expression (primarily by language) For this process, the phenomenological background theory, from the aesthetic experience to the reflection and explanation (“emergence-motivated event construction”), is explained in Fukushima (2020a, 2020b).

Sense-making theory To define the aesthetic Sake taste terms, I draw on “sense-making theory”(Fukaya & Tanaka, 1996; Tanaka & Fukaya, 1998), arguing that the meaning of a word is determined through various levels of interaction. The interaction includes the relationships among words (i.e., co-occurrence), between words and sentence or context, and even between people (i.e., communication level). Fukaya and Tanaka claim the sense of a single word cannot be determined a priori; rather, the sense is cooperatively “made” during the communication process. This chapter concurs with sense-making theory on this dynamic aspect of word sense.

Emergence-motivated event construction The author proposes to define Sake taste by focusing on adjectivals and verbs, as an alternative to a dominantly used method focusing on nouns. The author calls this latter way of verbalization an “objectmotivated event construction” where the experiencer primarily uses nouns to describe the event of tasting. This is commonly found in English tasting comments by wine sommeliers, as in “I feel a note of black cherry, cassis, and the rich flavor of the oak,” where the sommelier detects the elements of the flavor and verbalizes them, perhaps selecting the terms from his or her list of tasting words.

64

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

This is analogous to an “audio” or a “visual event construction,” where an event is reported objectively. As an alternative event construction, Fukushima (2020a, 2020b) proposed the theory of “emergencemotivated event construction.” The way of constructing the event of tasting (or smelling) differs from the process of vision. When we taste something—that is, when we have an event construction of tasting or when we conceptualize what we taste in our mouths—what we feel first is not the element of taste, such as sweetness, acidity, apple flavor, or other flavors in Sake (as expressed by nouns), but the emergence of the tasting event itself. Supporting emergence-motivated event construction means that adjectives, adjectival nouns, and verbs (but not nouns) take the leading role in the taste description. The recognition of the emergence of an event is primarily expressed by predicates (verbs, adjectives and adjectival nouns).

Linguistic background This chapter’s proposed method of defining taste terms has the following characteristics, in harmony with the themes of cognitive linguistics: • • •

the dynamic aspect of word senses usage-based “sencyclopaedic semantics” rather than a “dictionary view”

The meaning of a taste term is often different from the word’s general definition. For example, when tōmei ‘clear’ is used to express the taste of Japanese Sake, it could represent (or modify) the lightness of the body, or clean sweetness, or a quick fading of the aftertaste. However, ‘light body’, ‘sweetness’, or ‘aftertaste’ would never be listed in the definition of tōmei ‘clear’ in general dictionaries. As illustrated in this example, I am emphasizing the dynamic aspect of word senses. The “sense of a word” is not fixed and static as in a dictionary. It varies, depending on the context; it is made dynamically in communication. This is the dynamic aspect of a word’s meaning.

Narratological Background: Narrative Generation and Narrative Analysis The approaches in narratology are roughly classified into two kinds: narrative generation and narrative analysis. (Ogata, 2020) The primary approach in content generation study is to develop a computational (or computer-assisted) content generation system. In this book, the chapters in Part 2 are illustrative examples. As an applicational domain of content generation, the author has carried out some studies focusing on “cognitive content” as a story, which is called “Cognitive Content Generation” (Fukushima 2020a). The cognitive content is, in short, the symbolic expression (including non-verbal representations) of a mental image of the five senses. As a case of the study of the cognitive content, the author has been studying the domain of taste. Over the past few years, the author has conducted studies on cognitive content generation of taste. In (Fukushima, 2020a), the author presented a methodology for bridging the gap between the mental representation of taste and the external representation (i.e., words, drawing) as content generation. In this study, the author attempts to analyze the cognitive content of taste, with a tasting comments corpus. There are some text types expressing the sense of taste: recipes, essays, advertisements and food 65

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

packaging, online restaurant reviews, and so on. At this stage, the author does not define how much of these are in the scope of narratology. Among these genres, this study deals with tasting comments concerning Sake. Japanese Sake, which is also called rice wine, has recently become more characterful in its flavor and taste.

MAIN FOCUS OF THE CHAPTER Verbs for Describing the Taste of Sake This study aims to define the sense of verbs in Sake tasting description, asking the question, what kind of verbs are used in the Sake corpus? Appendix 4 shows the top 100 verbs in the Sake corpus. Note that the table does not include a number of functional verbs (e.g., naru ‘be’, suru ‘do’). Figure 1 shows the hierarchical clustering chart of the top 29 words (word frequency of over 100). From the hierarchy branching, Cluster 01 appears to be independent of other clusters (Clusters 02-05). As a general trend, verbs in Cluster 01 and Clusters 02-05 seem to differ in usage tendency. Cluster 01 consists of the following verbs: tsukuru ‘(Sake) making,’ kamosu ‘brew,’ hiyasu ‘chill,’ motsu ‘have,’ and tanoshimeru ‘enjoy(able).’ These words seem not to be the expressions for the taste or flavor, from the viewpoint of describing the quality of the experience. Here, in order to clarify the role of the verbs, let us take a look at a standard example of Sake reviewing in books (Sentence 1) and co-occurrence networks with nouns (Figure 2). The “Azuma-ichi” which aims at the optimum quality of Sake during meals uses Yamada-nishiki, cultivated by the brewer himself. [Ingredients] The privately-incubated Kumamoto yeast is used. Azuma-ichi constantly produces the fine junmaiginjo (pure rice premium Sake) by fermenting Kumamoto yeast for a long time at low temperature where the yeast is difficult to be active. [Brewing Techniques] The harmony of calm fragrance and umami, refreshing sourness and light taste is splendid. [Flavor and Taste] The sharpness of the aftertaste and the faint feeling of gas are impressive. [Flavor and Taste] You can enjoy the taste for a long time even after opening. It is a Japanese Sake that goes well with foods. [How to enjoy] For Azuma-ichi, Yamada-Nishiki Junmai-ginjo, (Sake Competition, 2019); translated by the author, [Caption] added by the author) Figure 2 shows the MDS (Multi-Dimensional Scaling) chart (drawing nouns and verbs, word frequency of over 100) Cluster 01 contains words for the general taste and flavor of Sake. Sake, taste, aroma and other words in Cluster 01 are common words in Sake tasting comments. Some words like umami, balance, ginjo and 66

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Figure 1. Hierarchical Clustering Chart of verbs in the Sake description Note, the bar graph on the left of the figure shows the frequency of word usage.

pure-rice (cf. Appendix 1) describe some quality of the taste, but these words can be used in almost any context in Sake descriptions. Clusters 02 and 03 seem to be clusters for the domain of taste and flavor. Cluster 04 consists of words for the [Space] or [Parts of the mouth] tag, including some words for the [Timeline] tag (e.g., last, after flavor). Clusters 01 to 04 can be roughly divided into the “taste / flavor” category, which describes the sensory impression in the tasting comments. On the other hand, Clusters 05 and 06 seem to have another role in the tasting comments (Cluster 07 is for ‘other words’ or ‘un-clustered words.’). Words in Cluster 05 mainly concern Sake brewing (Yamada-Nishiki is a brand of special rice for Sake). And Cluster 06 concerns words of classification (designation) of Sake (‘dai-ginjo’), drinking situation (‘chill’), food pairing (‘meal’). Clusters 05 and 06 can be divided into the “brewing” category. 67

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Figure 2. Multi-Dimensional Scaling chart (nouns and verbs, word frequency 100+ times)

Note. Verbs are framed, and the bold frames are the verbs for “understanding” the taste of Sake.

The Verbs to be Defined in this Study If divided in terms of narratology, verbs in Clusters 05-06 are regarded as verbs for “explaining (Erklaren)”, and verbs in Clusters 01-04 can be considered as verbs for “understanding” or “interpretation” (Dilthey, 1900). In this study, the author focuses on the verbs that contribute to describing the understanding of the tasters’ experience. The verbs for the understanding are, in other words, verbs for e xpressing some quality of the taste or flavor. The quality of the taste of flavor is personal and is a private sensation felt in the mouth. It cannot be objectively (or scientifically) explained with a ‘taste-sensor machine’. More concretely, the verbs to be defined in this study are the following (see also Figure 2.): hirogaru ‘spread’, nokoru ‘remain’, tatsu ‘arise’, hiku ‘fade’, arawareru ‘appear’, and kieru ‘vanish’. In addition to these verbs from Figure 2, the following frequently used verbs are investigated: tsuduku ‘continue’ and fukuramu ‘swell’.

68

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

METHOD Corpus and Text Coding (Mining) Tool CORPUS For this analysis, the author uses a Sake corpus, a corpus of Japanese Sake tasting expressions primarily taken from Sake-reviewing books and magazines written in Japanese. The Sake corpus also includes tasting comments and expressions provided by six tasters, myself included. Table 2 summarizes the details. Note that the paragraphs refer to the different Sake brand descriptions. In total, the Sake corpus consists of 120,789 words. Table 1.  Details of the Sake corpus Details Tokens

120,789

Types

6,018

Type Token Ratio

20.07

Sentences

5,582

Paragraphs (brands of Sake)

2,388

Average Frequency Standard Deviation

10.50 64.55

The Sake corpus draws on data from 14 books and magazines, as well as data from a tasting experiment. At first glance, the number of consulted books may seem limited. However, the books or magazines for Sake tend to introduce the brewing method of Sake brands, not the taste or flavor. Thus, eliciting the “tasting comment data” from published material is a difficult task. Sentence 1 below is a typical example of a flavor description in a Sake magazine. Figure 3 shows an image where such descriptions appear (basic translations are given in the bottom part of the image). 2. a. kajitsukei no hanayaka-na kaori to, ‘Fruity, elegant flavor and’ b. pukkuri-to-shita umami ga kōchū de fukurami, ‘pulump-umami swells in the mouth,’ c. taoyaka ni nagare-te iku. hōrei na Sake o jikkan. ‘then it calmly flows. (I) realized the rich, splendid taste.

69

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Figure 3. A standard style of tasting comment description in Japanese Sake books and magazines. Captured from Sake Catalog 755 (Umehara, 1999)

Coding tool As a text-mining tool, the author used KH Coder (Higuchi, 2004). For the entire process of corpus analysis (i.e., word extraction with ChaSen, PoS tagging, listing word frequency, making KWIC concordance, and drawing word networks), KH Coder was used. KH Coder is free software for quantitative content analysis or text mining. It is also used for computational linguistics and can be used to analyze Catalan, Chinese (simplified), Dutch, English, French, German, Italian, Japanese, Korean, Portuguese, Russian, Slovenian, and Spanish text (“KH Coder Index Page,” 22-09-2019).

How to Define Verbs Calculation of the co-occurrence score with KWIC Using KWIC, we can calculate the “score” for determining which words most frequently co-occur with the target term. The calculated “score” is used to indicate the strength of co-occurrence relationships. For a detailed explanation of KWIC, see Appendix 2; the protocol for calculating a score is shown in Appendix 3. In this analysis, I considered the words with a score of over 1.000 in defining the target term (i.e., approximately the top 30 words in the concordance list).

Defining the target terms The terms to be used to define the target term are selected from the list of the top words in the concordance. This task relies on the knowledge and skill of the individual producing the definition. If the person

70

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

knows very little about Sake or lacks tasting skill or does not understand the meaning of the words in the list, he or she will clearly be unable to define the target term properly. Moreover, people are likely to define the same term differently.2 Although the method is not fixed or rigidly defined, I set the following tags (in square brackets and bold text) as the points to be defined. [Flavor] or [Taste]: [Flavor] or [Taste] terms tend to modify taste or flavor (e.g., Words for [Taste] include: taste, sweetness, bitterness, acidity, etc.; Words for [Flavor] include: flavor, scent, floral, fruity flavor, etc.). [Dominance]: [Dominance] describes which tastes among the basic tastes (i.e., sweet, acidity, umami, bitter, astringency, and dry3) are more likely to co-occur with the target term. [Structure]: [Structure] terms include words for the structure or physical texture of the Sake (e.g., bodī ‘body’, waku / wakugumi / kokkaku ‘frame’, rinkaku ‘contour’, katamari ‘lump’, futoi ‘bold’). Examples of the usage of [Structure] terms include instantiations of a “conceptual metaphor” (Lakoff & Johnson, 1980), “taste is building” or “taste is body.” Sentence (3) Shows an example of the two conceptual metaphors. 3. kokkaku ga shikkari-shita, futoi umami ‘(I feel) the firm framed, bold umami’ In (3), the first part of the phrase kokkaku ga shikari-shita ‘firm framed’ involves the equation of “building” with “taste”, where the elements of “taste” are conceptualized as corresponding to (or “mapped onto”) those of “building.” Stated differently, the taste is conceptualized as having a frame, just like a building has a frame, and the stability of the “frame” of the taste is expressed by shikari-shita ‘firm’. Similarly, the latter part of (3) involves the equation of “human body” with “taste”, where the elements of “taste” are conceptualized as corresponding to those of the “human body”. Because of this correspondence, the taste term umami can accompany a modifier futoi ‘(lit.) fat’. [Organoleptic Feelings]: [Organoleptic Feelings] terms include terms on [Texture]. The majority of [Organoleptic Feelings] are words for [Texture] or mouthfeels (e.g., kanshoku ‘feeling’, tacchi ‘touch’, sofuto ‘soft’, kurīmī ‘creamy’). Words like marui ‘round’, tsutsumu ‘wrapping’, naderu ‘stroking’, and other words for stimulus are not tagged as [Texture], but as [Organoleptic Feelings]. For [Organoleptic Feelings], onomatopoeic terms are often used (e.g., zara-zara ‘rough texture’, suru-suru ‘smooth’). [Space] or [Parts of the mouth]: [Parts of the mouth] terms refer to the words for the parts of mouth used to indicate the point or the place where the taste or flavor is felt (e.g., kuchi ‘mouth’,, hanasaki ‘nose tip’, shita ‘tongue’). These words can be found in expressions like the one in (4), an instance of shitasaki ‘tip of the tongue’. [Space] is a tag for the words that indicate the place (but not the part of the mouth), like oku ‘back (of the mouth)’, sayu ‘right and left’, and so on. 4. Shitasaki de amami o kanjiru ‘(I feel) sweetness on the tip of the tongue.’ [Timeline]: [Timeline] terms describe the timeline of the tasting experience, from beginning to end (i.e., in my “emergence-motivated event construction” model, from the emergence of the taste to its disappearance). In a tasting comment, describing the way the taste or flavor appears and disappears is very important. Words for appearance are, for example, saisho ‘beginning’, arawareru ‘appear’, tatsu 71

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

‘stand’ (cf. Section 4.2.6), and hanasaki ‘tip of the nose’. Words for disappearance include kieru ‘fade’, kireru ‘finish’, nokoru ‘remain’, and atokuchi ‘aftertaste’. It is notable that some words for the part of the mouth often (indirectly) refer to the point of the timeline (e.g., ‘nose tip’ = ‘the starting point’, or ‘back of the mouth’ = ‘the last point’). [Movement]: [Movement] terms include verbs modified by adjectivals expressing the movement of the taste. Representative examples include: fukuramu ‘swell’, hirogaru ‘spread’, nagare ‘flow’, tadayō ‘drift’, and osamaru ‘subside’. [Characteristic Words]: [Characteristic Words] terms include words that supplement the information of the target terms but are important to describe the characteristics of the taste, such as sukoshi ‘a little’. [Related Words]: If the PoS of the word is the same as the PoS of the target term (i.e., adjectives or adjectival nouns in the concordance list), and if the word has an intralinguistic relationship with the target term, such as synonymy, antonymy, hyponymy, or gradable antonymy (cf.(Lehrer, 2009)), then the word should be included. For example, if the description includes yawarakaku marui ‘soft and round’, and the target term is yawarakai ‘soft’, marui ‘round’ should be listed as a related word, as it expresses a synonymous meaning as ‘soft’ in the Sake tasting context.

RESULTS Hirogaru ‘Spread’ As Table 2 shows, hirogaru ‘spread’ co-occurs with the following words;

[Taste] Among the five basic tastes, the term hirogaru ‘spread’ often co-occurs with amami ‘sweetness’ [4, 9, 19], umami ‘umami’ [10, 13], and sanmi ‘acidity’[14]. As stated above, these three tastes are all the primary tastes for Sake. Thus, hirogaru ‘spread’ describes the essential and desirable taste of Sake.

[Flavor] In the concordance list, only two words for flavor are listed: kōmi and kaori ‘flavor’ [7, 8]. The words kōmi and kaori are rather abstract words, and no specific flavor words (e.g., apple, melon, etc.) are seen in the concordance list.

[Dominance] The term hirogaru ‘spread’ is used more frequently to describe “taste” than “flavor,” as indicated by the higher token frequency of the taste terms for ‘sweetness’ [4, 9, 19], ‘umami’ [10, 13], and ‘acidity’ [14] versus the flavor terms [7, 8].

72

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Table 2. Concordance list for hirogaru ‘spread’ N

Word

Pron.

Eng. Trans.

PoS

Total

LT

RT

Score

1



aji

taste

noun

48

29

19

22.57

2



kuchi

mouth

noun

52

42

10

16.18

3

口中

kōchū

mouth

noun

37

32

5

14.97

4

甘み

amami

sweetness

noun

33

24

9

12.25

5



shita

tongue

noun

34

16

18

10.13

6

全体

zentai

entire

adv.

23

20

3

9.53

7

香味

kōmi

flavor and taste

noun

17

14

3

9.00

8

香り

kaori

flavor

noun

21

15

6

8.88

9

甘味

amami

sweetness

noun

22

15

7

8.47

10

旨み

umami

umami

noun

15

11

4

6.13

11

優しい

yasashii

tender

adj.

12

9

3

5.73

12

左右

sayū

right and left

noun

15

11

4

5.50

13

旨味

umami

umami

noun

13

8

5

5.48

14

酸味

sanmi

acidity

noun

14

6

8

5.43

15

最後

saigo

last

noun

11

1

10

4.42

16

丸い

marui

round

adj.

6

5

1

3.78

17

上顎

uwaago

maxilla

noun

11

7

4

3.70

18



yoko

side

noun

9

9

0

3.57

19

甘い

amai

sweet

adj.

14

11

3

3.57

20

薄い

usui

thin

adj.

8

4

4

3.57

21

感じ

kanjiru

feeling

noun

4

1

3

3.50

22

飲む

nomu

drink

verb

7

0

7

3.33

23

その後

sonoato

later

adv.

7

0

7

3.25

24



oku

back

noun

12

5

7

3.20

25



kome

rice

noun

10

4

6

3.20

26



Sake

Sake

noun

9

3

6

3.08

27

きれい

kirei

clean

AN

5

4

1

2.75

28

印象

inshō

impression

noun

8

3

5

2.73

29

味わい

ajiwai

taste

noun

7

5

2

2.65

30

香気

kōki

aroma

noun

6

5

1

2.53

Note. From left to right, the columns in the table give the ranking (1 through 21) of the score (N) of the co-occurring words (Word), the English translation of the co-occurring words (English Trans.), pronunciation of the co-occurring words (pronunciation), the part of speech (PoS) of the words, the total number of co-occurrences (Total), the number of occurrences to the left of the target term (LT), the number of occurrences to the right side of the target term (RT), and the scores.

[Space and Body-part] Hirogaru ‘spread’ is originally a word for spatial nuance. Naturally, hirogaru ‘spread’ frequently collocates with words ofthe [Space] and {Parts of the Mouth] tag, as seen in kuchi ‘mouth’ [2], kōchū ‘mouth’ [3]

73

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

and zentai ‘entire’ [6], sayū, ‘right and left’ [12], uwaago, ‘maxilla’ [17], yoko, ‘side’ [18], oku, ‘back’ [24]. These collocating words include almost all areas of the mouth cavity. Note that hirogaru ‘spread’ can be used for describing aroma diffusing in the nose cavity, though words about ‘nose’ are not seen in the concordance list.

Table 3. Concordance list for nokoru ‘remain’ N

Word

Pron.

Eng. Trans.

PoS

Total

LT

RT

Score

1



aji

taste

noun

46

31

15

18.65

2

最後

saigo

last

adv.

27

19

8

11.30

3



shita

tongue

noun

28

15

13

10.15

4

少し

sukoshi

a little

adv.

19

8

11

9.12

5

余韻

yoin

aftertaste

noun

16

11

5

6.33

6

長い

nagai

long

adj.

9

7

2

6.03

7

苦味

nigami

bitterness

noun

12

10

2

5.45

8

印象

inshō

impression

noun

14

12

2

5.23

9



kuchi

mouth

noun

15

10

5

5.03

10

薄い

usui

thin

adj.

12

8

4

4.93

11



oku

back

noun

17

12

5

4.70

12

雑味

zatsumi

odd taste

noun

8

6

2

4.50

13

酸味

sanmi

acidity

noun

11

6

5

4.37

14

刺激

shigeki

stimulus

noun

11

11

0

4.27

15

口中

kōchū

mouth

noun

8

3

5

3.83

16

香り

kaori

flavor

noun

11

3

8

3.82

17

甘み

amami

sweetness

noun

12

7

5

3.73

18



Sake

Sake

noun

8

0

8

3.67

19

甘い

amai

sweet

adj.

10

4

6

3.45

20



saki

tip

noun

10

8

2

3.35

21

丸い

marui

round

adj.

5

3

2

3.20

22

感じ

kanjiru

feeling

noun

9

5

4

3.15

23

上顎

uwaago

maxilla

noun

9

1

8

3.07

24

甘味

amami

sweetness

noun

6

4

2

3.03

25

その後

sonoato

later

adv.

5

1

4

3.00

26

舌先

shitasaki

tip of tongue

noun

8

3

5

2.83

27

渋味

shibumi

astringency

noun

8

6

2

2.78

28

渋み

shibumi

astringency

noun

6

6

0

2.58

29

味わい

ajiwai

taste

noun

4

1

3

2.45

30

消える

kieru

disappear

verb

7

3

4

2.37

74

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Nokoru ‘REMAIN’ [Taste] In the taste domain, nigami ‘bitterness’ [7], zatsumi ‘odd taste’ [12], and sanmi ‘acidity’ [13] and even shibumi ‘astringency’ [27] are characteristic collocational words for nokoru ‘remain.’ These tastes are not the primary tastes for Sake. Too much astringency, bitterness, and acidity spoil the balance of Sake, but the proper quantity of secondary tastes can give richness to the taste of Sake.

[Dominance] Though both flavor and taste terms can co-occur with nokoru ‘remain’, from the concordance list, the frequency dominance is clearly given to the domain of taste.

[Timeline] Nokoru ‘remain’ is originally a word for describing the last part of the taste. Therefore, some words for indicating the latter period can be seen in the concordance list: saigo ‘last’ [2], yoin ‘aftertaste’ [5, 34], sonoato ‘later’ [25], nagai ‘long’ [6].

[Space] Nokoru ‘remain’ seems to often co-occur with the parts of the mouth: shita ‘tongue’ [3], oku ‘back of the mouth’ [11] kōchū ‘mouth’ [15], uwaago ‘maxilla’ [23], shitasaki ‘tip of the tongue’ [26]. These words have a role in pointing out where the taste “remains.” The remaining tastes (i.e., bitterness and odd taste) tend to be felt at the back [11] of the tongue.

[Organoleptic Feelings] For organoleptic feelings, shigeki ‘stimulus’ [17] is seen in the list. The word ‘stimulus’ generally means (undesirable) dryness of the alcohol.

[Aesthetic Terms] Yoin ‘aftertaste’ [5], usui ‘thin (weak)’ [10], and marui ‘round’ [21]

[Verbs] For the co-occurring verbs, kieru ‘vanish’ [30] and hirogaru ‘spread’ [31] are listed.

75

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Tatsu ‘Arise’ The literal English translation of tatsu is ‘stand.’ However, if the verb tatsu is used for describing some quality of taste, it means ‘arise’ or ‘appear.’ Table 4. Concordance list for tatsu ‘arise’ N

Word

Pron.

Eng. Trans.

PoS

Total

LT

RT

Score

1

香り

kaori

flavor

noun

61

48

13

31.70

2

昇る

noboru

rise

verb

22

0

22

22.00

3

香気

kōki

aroma

noun

26

24

2

12.58

4

華やか

hanayaka

gorgeous

AN

16

9

7

8.53

5

湧く

waku

well up

verb

7

7

0

7.00

6

口中

kōchū

mouth

noun

15

7

8

6.92

7



san

acid

noun

12

5

7

5.40

8



aji

taste

noun

17

4

13

5.38

9

甘い

amai

sweet

adj.

15

9

6

5.23

10

果実香

kamika

fruit fragrance

noun

11

10

1

4.83

11

リンゴ

ringo

apple

noun

8

3

5

3.77

12

酸味

sanmi

acidity

noun

5

1

4

3.20

13

甘味

amami

sweetness

noun

8

5

3

3.03

14

上げる

ageru

raise

verb

3

0

3

3.00

15

クリーミー

kurīmī

creamy

AN

5

1

4

2.95

16

全体

zentai

entire

AN

5

0

5

2.87

17

鼻先

hanasaki

nose tip

noun

5

5

0

2.50

18

香味

kōmi

flavor and taste

noun

6

4

2

2.28

19

爽やか

sawayaka

freshness

AN

8

6

2

2.08

20

ふくよか

fukuyoka

plump

AN

9

3

6

2.05

21

苦味

nigami

bitterness

noun

4

2

2

2.00

22



saki

tip

noun

4

4

0

2.00

23

シャープ

shāpu

sharp

adj.

5

2

3

1.83

24

軽妙

keimyō

light and easy

AN

3

0

3

1.83

25

芳香

hōkō

fragrance

noun

4

4

0

1.83

26

軽い

karui

light

adj.

5

2

3

1.75

27

イチゴ

ichigo

strawberry

noun

4

1

3

1.73

28

アルコール

arukōru

alcohol

noun

4

2

2

1.67

29

含む

fukumu

contain

verb

7

2

5

1.60

30

スムーズ

sumūzu

smooth

AN

2

0

2

1.50

31

辛味

karami

dryness

noun

3

2

1

1.50

32

舌先

shitasaki

tip of tongue

noun

3

2

1

1.50

33



momo

peach

noun

2

0

2

1.50

76

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Typical usage of tatsu ‘arise’ is seen in example (5). 5. A citrus type (lime) flavor gently arises. Raimu kei no kaori ga hikaeme ni tatsu For Wakaebisu Junmai-Daiginjo, (JS, 2001)

[Flavor] The verb tatsu ‘arise’ significantly co-occurs with words for flavor. In addition to ‘flavor’ itself [1, 3, 18], phrases for fruity flavor should be specially mentioned. In the concordance list, many names of fruits [10] are listed; ringo ‘apple’ [11], ichigo ‘strawberry’[27, banana ‘banana’ [34] and meron ‘melon’ [39].

[Taste] As well as the flavor words, tastes relating to fruits collocate with tatsu ‘arise’; sanmi ‘acidity’ [7, 12], amami ‘sweetness’ [13]. Among the primary tastes for Sake (i.e., sweetness, acidity, and umami), umami is not listed in the collocation of tatsu ‘arise.’ This wo_uld be because of the types of Sake. There are mainly four or more types of Sake, including the fruity type and the umami type (others are the clear ‘tasteless’ type, matured type, and so on). From the concordance list, tatsu tends to be used for the fruity type, rather than the umami type of Sake. For the secondary taste, nigami ‘bitterness’ [21] and karami ‘driness’ [31] are listed, but their rank is rather low.

[Space] The verb tatsu ‘arise’ often describes the first part of one experience of tasting Sake, but interestingly, there are no words for the [Timeline] tag in the concordance list. Instead, there are some words for the front part of the mouth and nose; hanasaki ‘nose tip’ [17], shitasaki ‘the tip of the tongue’ [32] and simply saki ‘tip’ [22]. Note that some words for the entire space of the mouth are also seen; kōchū ’mouth’ [6] and zentai ‘whole’ [16].

[Aesthetic Concepts] Tatsu ‘arise’ significantly co-occurs with aesthetic concepts: hanayaka ‘gorgeous’ [4], kurīmī ‘creamy’ [15], sawayaka ‘freshness’ [19], fukuyoka ‘plump’ [20], shāpu ‘sharp’ [23], keimyō ‘light and easy’ [24], hōkō ‘fragrance’ [25], karui ‘light’ [26], sumūzu ‘smooth’ [30], katai ‘firm’ [37], wakawakashii ‘youthful’ [41] These aesthetic concepts listed above tend to modify the flavor, rather than the taste. Let’s take a look at an example (6). 6. An elegant ginjo-flavor (fruity flavor) arises. Yuuga na ginjō ka ga kaori tatsu 77

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

For Shiki-zakura, (SSI, 2019), underlined words indicate the aesthetic concepts.

Tsuduku ‘Continue’ [Taste] In the taste [3] domain, the typical collocation is found in words for the secondary taste; nigami ‘bitterness’ [6, 25], and san ‘acid’ [9]. ‘Sweetness’ [17, 22] is listed as always, but the rank is relatively low.

[Timeline] As shown in Table 5, various words for describing the timeline are listed; saigo ‘last’ [7], sonoato ‘later’ [12], saisho ‘first’ [19], atokuchi ‘aftertaste’ [23], atoni ‘after’ [27]. Other words than saisho ‘first’ are words for indicating “later” or “after in the timeline.

[Space] As well as the timeline domain, words for the space domain also co-occur with the verb tuzuku ‘continue.’ The listed words are: kuchi ‘mouth’ [13, shitasaki ‘the tip of the tongue’ [14], uwaago ‘maxilla’ [16], and shita ‘tongue’ [24].

[Organoleptic Feelings] The word shibire ‘numbness’ [21] is a unique word for Tsuzuku ‘continue.’ It describes the drying sensation or rough ‘mouthfeel’[4]. Numbness has a relation to astringency, as well as the feelings associated with tannin in wines. As for the taste of Sake, the shibire ‘numbness’ can represent the sensation of the astringency, acidity, maybe bitterness, and dryness of the alcohol.

[Characteristic words] The words junmai ‘pure rice,’ meigara ‘brand’ [8], kambam ‘signboard’ [5] are characteristic. However, these words are not for describing some quality of taste, but for mentioning the background of the Sake being reviewed.

[Aesthetic Terms] Tsuyoi ‘strong’ [26] and yowai ‘weak’ [29] are listed. Futoi ‘bold’ and nōmitsu ‘dense’ represent the way the thick taste ‘continues.’

78

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Table 5. Concordance list for tsuduku ‘continue’ N

Word

Pron.

Eng. Trans.

PoS

Total

LT

RT

Score

1

長い

nagai

long

2

余韻

yoin

aftertaste

noun

20

16

4

9.58

3



aji

taste

noun

6

1

5

3.70

4

感触

kanshoku

feel

noun

5

4

1

2.45

5

看板

kamban

signboard

noun

2

0

2

2.00

6

苦味

nigami

bitterness

noun

3

2

1

2.00

7

最後

saigo

last

noun

4

1

3

2.00

8

銘柄

meigara

brand

noun

3

0

3

2.00

9



san

acid

noun

3

1

2

1.75

10

純米

junmai

pure rice

noun

5

2

3

1.75

11

印象

inshō

impression

noun

5

4

1

1.53

12

その後

sonoato

later

adv.

3

3

0

1.50

13



kuchi

mouth

noun

3

1

2

1.50

14

舌先

shitasaki

tip of tongue

noun

3

0

3

1.50

15

流れ

nagare

flow

noun

4

1

3

1.50

16

上顎

uwaago

maxilla

noun

4

2

2

1.40

17

甘み

amami

sweetness

noun

4

3

1

1.33

18

感じる

kanjiru

feel

verb

6

3

3

1.30

19

最初

saisho

first

noun

3

1

2

1.25

20

香り

kaori

flavor

noun

4

2

2

1.20

21

痺れ

shibire

numbness

noun

3

3

0

1.17

22

甘味

amami

sweetness

noun

3

1

2

1.08

23

後口

atokuchi

aftertaste

noun

3

2

1

1.08

24



shita

tongue

noun

3

0

3

1.08

25

苦み

nigami

bitterness

noun

3

1

2

1.03

26

強い

tsuyoi

strong

adj.

1

1

0

1.00

27

後に

atoni

after

adv.

1

1

0

1.00

28

山田錦

Yamada-Nishiki

Yamada-Nishiki

noun

1

0

1

1.00

29

弱い

yowai

weak

adj.

1

1

0

1.00

30

精悍

seikan

virile

AN

1

0

1

1.00

31

創業

sōgyō

establish

noun

3

3

0

1.00

32

太い

futoi

bold

adj.

1

0

1

1.00

33

濃密

nōmitsu

dense

AN

1

0

1

1.00

adj.

17

17

0

16.50

79

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Fukuramu ‘Swell’ Table 6. Concordance list for fukuramu ‘swell’ N

Word

Pron.

Eng. Trans.

PoS

Total

LT

RT

Score

1



aji

taste

noun

29

23

6

13.08

2

厚い

atsui

thick

adj.

15

7

8

10.83

3

感じる

kanjiru

feel

verb

11

2

9

4.53

4

香り

kaori

flavor

noun

12

9

3

3.72

5

丸い

marui

round

adj.

5

4

1

3.53

6

口中

kōchū

mouth

noun

8

7

1

3.37

7

味わい

ajiwai

taste

noun

8

1

7

3.00

8

吟醸香

ginjō-ka

ginjo-flavor

noun

7

7

0

2.95

9

香味

kōmi

flavor and taste

noun

8

4

4

2.75

10

旨み

umami

umami

noun

6

5

1

2.70

11

優しい

yasashii

tender

adj.

4

4

0

2.53

12

香気

kōki

aroma

noun

6

5

1

2.45

13

広がる

hirogaru

spread

verb

6

3

3

2.33

14

良い

yoi

good

adj.

5

1

4

2.03

15

引き出す

hikidasu

take out

verb

4

1

3

2.00

16

厚み

atsumi

thickness

noun

4

3

1

2.00

17

太い

futoi

thick

adj.

3

1

2

2.00

18

有す

yūsu

possess

verb

4

0

4

2.00

19

きれい

kirei

clean

AN

6

3

3

1.95

20



haba

width

noun

4

2

2

1.83

21

後口

atokuchi

aftertaste

noun

7

1

6

1.77

22

丸み

marumi

roundness

noun

4

3

1

1.70

23

持つ

motsu

have

verb

4

1

3

1.70

24

甘み

amami

sweetness

noun

5

3

2

1.67

25

穏やか

odayaka

calm

AN

3

2

1

1.50

26

保つ

tamotsu

keep

verb

3

0

3

1.50

27

良好

ryōkō

good

AN

3

0

3

1.50

28

立つ

tatsu

stand

verb

5

3

2

1.40

29

純米

junmai

pure rice

noun

6

3

3

1.38

30

張り

hari

tension

noun

3

1

2

1.33

[Taste] The words for the taste domain are umami [10], sweetness [24], and acidity [33, 41]. These are the primary tastes of Sake and are often listed in the concordance list. Judging from Table 6, the rank of these

80

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

primary taste words is relatively low.

[Flavor] The unique words for the flavor domain are ginjōka ‘ginjō-flavor’ [8]. Ginjōka is the aroma that is produced by a special brewing process (ginjō-brewing, see Appendix 1). Ginjōka generally contain fruity flavors, such as melon [36], apple, green apple, pear, or banana.

[Dominance] Fukuramu ‘swell’ can co-occur both with the domains of taste and flavor. In comparison to other verbs, the concordance list (Table 6) tells us that the flavor domain [4, 8, 9, 12, 36] relatively dominates over the taste domain [1, 7, 10, 24, 33, 41].

[Timeline] Words for the timeline domain: atokuchi ‘aftertaste’ [21]

[Space] Words for the space domain: kōchū ’in the mouth’ [6]

[Aesthetic Terms] Atsui/atsumi ‘thick(ness)’ [2, 16] is a unique term for fukuramu ‘swell’. Thickness is the experience of Sake with a rich, full-bodied, or umami taste (or flavor). Futoi ‘bold’ [17] or haba ‘width’ [20] have a similar nuance to atsui. Other aesthetic terms in the concordance list are: marui/marumi ‘round(ness)’ [5, 22], yasashii ‘tender’ [11], kirei ‘clean’ [19], odayaka ‘calm’ [25].

Hiku ‘Fade’ The verb hiku ‘fade’ is literally translated as ‘pull.’ However, in a tasting description, ‘fade’ or ‘go down’ would be the better translations.

[Taste] Words for sweetness [3], umami [4], and acidity [19, 25, 26] are listed. These are the primary tastes for Sake, therefore, hiku ‘fade’ illustrates the manner of fading or the going down of a comfortable and desirable taste.

81

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Table 7. Concordance list for hiku ‘fade’ N

Word

Pron.

1



ato

2



3

甘味

4

Eng. Trans.

PoS

Total

LT

RT

Score

after

noun

36

36

0

17.75

o

tail

noun

24

24

0

12

amami

sweetness

noun

23

11

12

10.933

旨味

umami

umami

noun

16

5

11

7.367

5

軽い

karui

light

adj.

9

0

9

5.917

6

後口

atokuchi

aftertaste

noun

7

3

4

3.067

7

切れ

kire

clearing

noun

5

1

4

3.033

8



aji

taste

noun

8

5

3

3

9

感触

kanshoku

feel

noun

7

2

5

2.7

10

優しい

yasashii

tender

adj.

4

3

1

2.667

11

穏やか

odayaka

calm

AN

5

2

3

2.583

12

きれい

kirei

clean

AN

6

5

1

2.333

13

後に

atoni

after

adv.

5

5

0

2.333

14

印象

inshō

impression

noun

5

1

4

2.233

15

香り

kaori

flavor

noun

8

5

3

2.15

17

浅い

asai

shallow

adj.

2

2

0

2

18

丸み

marumi

roundness

noun

4

1

3

1.833

19



san

acid

noun

6

4

2

1.833

20



oku

back

noun

5

2

3

1.7

21

感じる

kanjiru

feel

verb

7

3

4

1.633

22

香味

kōmi

flavor and taste

noun

3

0

3

1.583

23

なめらか

nameraka

smooth

AN

2

1

1

1.5

24

余韻

yoin

aftertaste

noun

4

2

2

1.5

25

甘酸っぱい

amazuppai

sweet and sour

adj.

3

2

1

1.4

26

酸味

sanmi

acidity

noun

5

2

3

1.35

27

タッチ

tacchi

touch

noun

4

4

0

1.333

28

深い

fukai

deep

adj.

2

0

2

1.333

29

味わい

ajiwai

taste

noun

4

2

2

1.333

30

太い

futoi

thick

adj.

2

0

2

1.25

[Dominance] We can accept the dominance of [Taste]. As for the [Flavor] category, kaori ‘flavor’ [15] and kōmi ‘flavor’ [24] are listed, but the ranking is low, and no specific flavor names are seen in the list.

82

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

[Timeline] Hiku ‘fade’ describes a decreasing or vanishing. Consequently, the [Timeline] words, especially words for the latter timing tend to collocate: atokuchi ‘aftertaste’ [6], kire ‘clearing’ [7], atoni ‘after’ [16], yoin ‘after flavor’ [24].

Kieru ‘Vanish’ [Taste] We can recognize [Dominance] in [Taste], rather than [Flavor]. Then, what kind of tastes vanish during Sake tasting? In the concordance list, we can find “stimulating tastes” for Sake; nigami ‘bitterness’ [13], and shibumi ‘astringency’ [25]. These words sometimes appear in other concordance lists, but the frequency is characteristically high in the list for Kieru ’vanish’. Note that too much of these stimulating tastes is generally thought to create “odd tastes” in Sake. However, a moderate amount of bitterness or astringency often gives the feeling of outline or structure to a Sake.

[Space (Parts of the mouth)] Bitterness and astringency tend to be felt around the root of the tongue or the peripheral area of the mouth cavity (i.e., hard palate, soft palate, or gums). In Table 11, there are some words of the parts of the mouth; oku ‘back’ [7], shita ‘tongue’ [8], uwaago ‘maxilla’ [9], kōchū ‘mouth’ [10], nodo ‘throat’ [12], shitasaki ‘tip of the tongue’ [19], saki ‘tip’ [26]. As shown above, there are a large number of words for the root of the tongue, or the peripheral area of the mouth cavity. This implies that the bitter taste or astringency determines the final impression of the Sake taste.

[Timeline] Kieru ‘vanish’ is an event in the last part of one tasting attempt. In the concordance list, we can find some words for describing the last part of the timeline; yoin ‘aftertaste’[3], nokoru ‘remain’ [14], atoaji ‘aftertaste’ [16] nokosu ‘leave’ [18]. Collocating with the words for [Space] (especially the back area of the mouth), these words are applied for describing the bitter taste and astringency in the Sake. On the one hand, some words for the first part of the Timeline; arawareru ‘appear’ [23], saisho ‘first’ [24]. Arawareru or saisho are often collocated with words of the front parts of the mouth (i.e., shitasaki ‘tip of the tongue’ [19], saki ‘tip’ [26]) in the [Space] domain. These words can be regarded as antonyms for kieru ‘vanish’, but they are often collocated together. Take a look at example (7). 7. Chiizu no aji ga kieru koro, umami no yoin ga arawareru When the taste of cheese disappears, the rich aftertaste appears. 83

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Table 8. Concordance list for kieru ‘vanish’ N 1

Word 味

Pron. aji

Eng. Trans. taste

PoS noun

Total 13

LT 12

RT 1

Score 6.25

2

印象

inshō

impression

noun

11

11

0

4.867

3

余韻

yoin

aftertaste

noun

12

6

6

4.633

4



kuchi

mouth

noun

10

2

8

3.35

5

甘み

amami

sweetness

noun

8

5

3

3.2

6

感じる

kanjiru

feel

verb

9

8

1

3.117

7



oku

back

noun

10

5

5

2.967

8



shita

tongue

noun

9

0

9

2.8

9

上顎

uwaago

maxilla

noun

7

1

6

2.75

10

口中

kōchū

mouth

noun

8

3

5

2.7

11

スッ

su’

quickly

onoma

5

5

0

2.5

12



nodo

throat

noun

9

2

7

2.433

13

苦味

nigami

bitterness

noun

7

3

4

2.4

14

残る

nokoru

remain

verb

7

4

3

2.367

15

軽い

karui

light

adj.

6

3

3

1.817

16

後味

atoaji

aftertaste

noun

3

0

3

1.667

17

甘味

amami

sweetness

noun

4

3

1

1.583

18

残す

nokosu

leave

verb

4

4

0

1.583

19

舌先

shitasaki

tip of tongue

noun

5

1

4

1.533

20

すーっ

sū’

smoothly

adv.

3

3

0

1.5

21

透明

tōmei

transparent

AN

4

0

4

1.4

22

香り

kaori

flavor

noun

3

2

1

1.333

23

現れる

arawareru

appear

verb

3

2

1

1.25

24

最初

saisho

first

noun

3

0

3

1.25

25

渋み

shibumi

astringency

noun

2

1

1

1.25

26



saki

tip

noun

3

1

2

1.25

27

溶ける

tokeru

dissolve

verb

2

2

0

1.2

28

引っかかる

hikkakaru

stuck

verb

4

3

1

1.117

29

感じ

kanjiru

feeling

noun

3

3

0

1.083

This example illustrates the vanishing of one taste, and the contrasting appearance of another taste. In this way, antonyms for kieru ‘vanish (disappear)’ are often collocated in order to give some contrast between the vanishing taste and the appearing or remaining taste.

84

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

[Onomatopoeias] Onomatopoeic expressions are the most distinctive modifiers for vanish. In general, making a taste description can be rephrased to point out the existing taste on the tongue. The way of ‘vanishing’ or finishing of a taste is hard to recognize and cannot be seen visually. In a previous study (Fukushima, 2017), the author raised the theory that onomatopoeic expressions in taste descriptions are employed to describe breaking points, turning points, and changing processes of taste, rather than stable states. Kieru ‘vanish’ is a typical example of the “breaking points” or “turning points” of a taste.

Tadayou ‘Flow’ [Dominance] Tadayou ‘flow’ has an exceptional feature of dominance. While other terms tend to indicate their dominance for Taste, Tadayou ‘flow’ shows its dominance in the Flavor domain. In the concordance list, words for the flavor domain are ranked higher: ‘flavor’ (kaori [1], kōmi [6], fūmi [10]), kōki ‘aroma’ [3],

[Space] Except for the [Taste] or [Flavor] words, the major part is in the [Space] domain, especially body parts: bikū ‘nasal cavity’[9], oku ‘back’[13], hanasaki ‘nose tip’ [14], shita ‘tongue’ [15], hana ‘nose’ [19]. Within the co-occurrence of tadayou ‘flow’, these body parts terms are used for indicating the place the flavor or taste is felt. The feeling of tadayou ‘flowing and retaining’ is a sensation of gas (not liquid). Thus, ‘the nose’ or ‘the nasal cavity’ is seen in the list.

[Aesthetic Terms] In the concordance list, the following terms are thought of as aesthetic terms for Sake; kirei ‘clean’, kihin ‘elegance,’ sugasugasii ‘refreshing.’ These terms have some clean and clear qualities in common. Kihin is one of the best complimentary words for the taste of Sake. It refers to the elegant, polished, and refined quality of Sake.

CONCLUSION AND FUTURE RESEARCH DIRECTIONS In this study, the author attempted to give encyclopaedic definitions to the verbs in a Sake corpus based on Sake tasting related terminology. In this terminology, the author emphasized the importance of the predicates (verbs, adjectives, and adjectival nouns). They are the starting point of our cognition of taste or flavor as an “event”. Through the terminological studies, the author proposed a defining method using collocation tags (Taste, Flavor, Timeline, and so on). With the collocation tags, we can analyze what types of words tend to co-occur with the target terms. 85

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Table 9. Concordance list for tadayou ‘flow’ N

Word

Pron.

Eng. Trans.

PoS

Total

LT

RT

Score

1

香り

kaori

flavor

noun

19

16

3

7.317

2

かすか

kasuka

faint

AN

6

5

1

3

3

香気

kōki

aroma

noun

8

4

4

2.95

4



Sake

Sake

noun

10

6

4

2.933

5

酸味

sanmi

acidity

noun

7

5

2

2.65

6

香味

kōmi

flavor and taste

noun

6

6

0

2.5

7

リンゴ

ringo

apple

noun

4

2

2

2.4

8

きれい

kirei

clean

AN

6

1

5

2.283

9

鼻腔

bikū

nasal cavity

noun

4

3

1

2

10

風味

fūmi

flavor

noun

4

4

0

2

11

気品

kihin

dignity

noun

3

2

1

1.833

12

甘味

amami

sweetness

noun

4

2

2

1.667

13



oku

back

noun

4

3

1

1.5

14

鼻先

hanasaki

nose tip

noun

4

4

0

1.5

15



shita

tongue

noun

5

3

2

1.483

16



aji

taste

noun

5

2

3

1.417

17

甘み

amami

sweetness

noun

4

2

2

1.4

18

清々しい

sugasugasii

refreshing

adj.

2

1

1

1.333

19



hana

nose

noun

4

2

2

1.25

20

ブドウ

budō

grape

noun

2

1

1

1.2

21



san

acid

noun

3

0

3

1.033

22

スマート

sumāto

smart

AN

1

0

1

1

23

プレミアム

puremiamu

premium

AN

1

0

1

1

24

果実

kajitsuka

fruit

noun

1

0

1

1

25

果実香

kamika

fruit fragrance

noun

1

0

1

1

26

感じ

kanjiru

feeling

noun

1

0

1

1

27

甘辛い

amakarai

salty-sweet

adj.

1

0

1

1

28

厚い

atsui

thick

adj.

1

1

0

1

29

香木

kōboku

fragrant wood

noun

1

0

1

1

30



momo

peach

noun

1

0

1

1

31

雰囲気

fun’iki

atmosphere

noun

2

2

0

1

32

余韻

yoin

aftertaste

noun

1

0

1

1

The author examined the meanings of 16 adjectives and adjectival nouns (Fukushima, 2020b)), and eight verbs (this study). In conclusion, this terminology is summarized in the following table. As a brief summarization of this Sake terminology study, let us compare the overall tendency of each PoS (part of speech, i.e., adjectives, adjective-nouns, and verbs) based on Table 10. Table 10 shows two

86

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Table 10. Tag-collocation with the Target Terms

Adjectives

AdjectiveNouns

Target Terms (Defined Terms)

Taste

Flavor

Dominance

Yawarakai ‘soft’



¡

T

Karui ‘light’



T

Katai ‘firm’



T

Marui ‘round’



T



Futoi ‘bold’



T



Kobashii ‘roasted aromatic’



Yasashii ‘tender’



Chikaraduyoi ‘powerful’



Kirei ‘clean’





F/T

Odayaka ‘calm’





F/T

Sawayaka ‘fresh’





F/T

Tomei ‘clear’



Fukuyoka ‘plump’

¡

Hanayaka ‘gorgeous’ Maroyaka ‘mellow’



Fukuzatsu ‘complex’



Hirogaru ‘spread’



Nokoru ‘remain’



Tatsu ‘arise’ Verbs



Movement

Timeline

Structure

Space (Parts of the mouth)

Organoleptic Feelings



● ●

● ●



● ●

F/T



T



T

● ●







T ¡





F



F/T





T T

● ●





F ●





F

¡

¡





Tsuduku ‘continue’



Fukuramu ‘plump’



Hiku ‘fade’



T



Kieru ‘vanish’



T





Arawareru ‘appear’



¡

T





Tadayou ‘flow’

¡



F

● ●



Note. ●: strong collocation, ¡: weak collocation, T: Taste dominant, F: Flavor dominant, F/T: Both Flavor and Taste collocate with the target term. Adjectives and Adjectival-Nouns are studied in (Fukushima, 2020b).

tendencies: first, each PoS has its unique role, as well as each word. Secondly, among one PoS, there can be seen some functional differentiation. As a general tendency, adjectives can collocate with broad categories. Concerning adjectival-nouns, kirei ‘clean’ collocates with many kinds of categories, but the others’ primary function seems to modify the taste or flavor itself. Adjectival nouns seem to hardly collocate with the [Movement] or [Structure] domain. The collocations of the verbs seem to concentrate on the [Timeline] and [Space] domain. On the other hand, verbs do not tend to collocate with the [Movement] or [Structure] domains, the same as the tendency of adjectival nouns.

87

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Adjectives Common features for adjectives While they commonly occur with Taste, the co-occurrence frequency with fragrance is low (cf. adjective verbs). On the whole, adjectives seem to co-occur with various domains (i.e., [Tags]), and there is no specific tag specific modification function.

Differentiation among adjectives There are two groups among adjectives: first, terms that tend to co-occur with the [Space] tag and [Organoleptic Feelings] tag (i.e., yawarakai ‘soft’, Karui ‘light’, Katai ‘firm’). Second, terms that tend to co-occur with the [Movement], [Timeline], and [Structure] tags (i.e., marui ‘round’, futoi ‘bold’, kobashii ‘roasted aromatic’).

Adjective-Nouns Common features for adjective-nouns Adjective-nouns tend to collocate not only with the [Taste] tag but also with [Flavor] tag. Concerning adjectives, the [Movement] tag and the [Timeline] tag often share the target terms. On the one hand, in the case of adjectival-nouns, few terms collocate with Movement but some words collocate with Timeline.

Differentiation among adjectival-nouns For adjectival-nouns, grouping the terms by tags seems difficult, because of the less frequent collocating between the target terms and tags. But if roughly divided, we can see a group of ‘tag-collocating terms’ and ‘less tag-collocating terms.’ Tag-collocating terms include; kirei ‘clean’, odayaka ‘calm’, fukuyoka ‘plump’, and hanayaka ‘gorgeous’. As with adjectives, these terms work as modifiers of taste or flavor in the collocation with the tags. ‘Less tag-collocating terms’ (i.e., sawayaka ‘fresh’, tōmei ‘clear’, maroyaka ‘mellow’, and fukuzatsu ‘complex’) are terms that are less likely to collocate with other tags. The primary function of these terms is not modifying tastes. Rather, these terms express some aesthetic quality by itself. Thus, ‘less tag-collocating terms’ work as predicates (not modifying words) in taste description sentences.

Verbs Common features for verbs The co-occurrence frequency between the listed verbs and the [Movement] tag or the [Structure] tag is relatively low. On the one hand, the [Timeline] and [Space] tags seem to co-occur frequently. While adjectives and adjectival-nouns indicate a property (or some quality) of the taste, verbs describe the motion of the liquid (or flavor gas), appearance/disappearance, and the changing process of the taste. Thus, verbs tend to collocate with [Timeline] and [Space] tag words, that are the markers of chronological time flow or the place movement, appearance, or disappearance occurs. 88

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Differentiation among verbs Similar to adjectival-nouns, there is differentiation in the dominance of Taste and Flavor. Taste-dominant terms are: nokoru ‘remain’, tsuduku ‘continue’, hiku ‘fade’, kieru ‘vanish’, and arawareru ‘appear’. Flavor-dominant terms are: tatsu ‘arise’ and tadayou ‘flow’. Another differentiation among verbs are seen in the collocation with [Organoleptic Feelings] tag. Nokoru ‘remain’, tsuduku ‘continue’, hiku ‘fade’ tend to collocate with [Organoleptic Feelings] tag.

FUTURE STUDIES In this study series, the author has defined the co-occurrence unit size as one sentence (‘sentence cooccurrence’). In future studies, collocational tendencies should be more microscopically investigated from the viewpoint of more close word collocation, such as ‘dependency’ or ‘modification’ relationships. As well the analyzing unit size, the genre or domain of the corpus is important. Since the meanings of the aesthetic terms depend on the domain, the comparison among various domains such as wines, whiskeys, perfumes, or others would be worth investigating. As well as the diversity among domains, the variety among the speaker groups would be impressive. There would be significant term usage variation between professional tasters and novices. Likewise, variations between languages (e.g., comparing the wine descriptions in English and in French), book genre (e.g., comparing the commercial magazines and critique essays), generation, or gender would reveal the novel features of the aesthetic terms. In any case, success or failure largely depends on the quantity and the quality of the corpus. Another future research direction is on grammar theory. In this study, we analyzed the co-occurrence relationship between [Tags] and each part of speech (PoS), as summarized in figure 10. We found some differences in co-occurrence characteristics among PoS and within PoS, as stated above. However, it should be noted that this analysis employs a particular grammatical theory or PoS classification method. There is an influential grammatical theory called “four major grammar” in Japanese grammar. The classification and the way of the perspective of the PoS are different, respectively. In this study, we could not give a sufficient explanation as to why adjectives are difficult to co-occur with [Flavor] tag, whereas adjective verbs are easy to co-occur with [Flavor] tag. In order to give a clear answer to this question, it may be necessary to review the grammatical theory that we stand on.

REFERENCES Caballero, R. (2007). Manner-of-motion verbs in wine description. Journal of Pragmatics, 39(12), 2095–2114. doi:10.1016/j.pragma.2007.07.005 Caballero, R. (2017). From the glass through the nose and the mouth: Motion in the description of sensory data about wine in English and Spanish. Terminology, 23(1), 66–88. doi:10.1075/term.23.1.03cab Dilthey, W. (1900). Die Entstehung der Hermeneutik. Retrieved from https://archive.org/details/bub_ gb_icZZAAAAcAAJ

89

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

Fukaya, M., & Tanaka, S. (1996). The sense-making theory of ( no Imizuke ron). Kinokuniya. Fukushima, H. (2014). Encyclopedia of Sake Terms. Keio University. Fukushima, H. (2020a in press). Cognitive Contents Generation: A Basic Introduction (Ninchiteki monogatari seisei heno shotai). In T. Ogata (Ed.), Various Aspect of Post Narratology (Post Narratology no shoso). Shinyosya. Fukushima, H. (2020b in press). Defining the aesthetic sake taste terms: A usage-based approach. In K. Toratani (Ed.), The Language of Japanese Food: Cognitive Perspectives and Beyond. John Benjamins. Gauntner, J. (2011). Sake Handbook. North Clarendon: Tuttle Publishing. Lehrer, A. C. N.-P. L. (2009). Wine & conversation (2nd ed). Oxford: Oxford University Press. López-Arroyo, B., & Roberts, R. P. (2014). English and Spanish descriptors in wine tasting terminology. Terminology, 20(1), 25–49. doi:10.1075/term.20.1.02lop Matsuura, T. (1992). Terms for tasting Sake (Sake wo ajiwau kotoba). In I. Tajima & K. Niwa (Eds.), Gendai Nihongo no kenkyū. Izumi Shoin. Otsuka H. (2004). Analysis for Evaluated Expressions of Tasting Japanese Sake. doi:10.11517/pjsai. JSAI04.0.147.0 Otsuka, H., Suwa, M., & Yamaguchi, K. (2015). Studies of Expressions to Taste Japanese Sake by Creating Onomatopoeia. Japanese Society for Artificial Intelligence. doi:10.11517/pjsai.JSAI2015.0_2N5OS16b5 Paradis, C., & Eeg-Olofsson, M. (2013). Describing Sensory Experience: The Genre of Wine Reviews. Metaphor and Symbol, 28(1), 22–40. doi:10.1080/10926488.2013.742838 Sake Competition. (2019). The Official Guidbook for Sake Competition 2019. Pia. Tanaka, S., & Fukaya, M. (1998). The Evolvement of the ( no Tenkai). Kinokuniya. Utsunomiya, H., Isogai, A., Iwata, H., & Nakano, S. (2006). Flavor Terminology and Reference Standards for Sensory Analysis of Sake. Journal of the Blewing Society of Japan, 101(10), 730–739.

KEY TERMS AND DEFINITIONS Sake: Sake (Japanese Sake, also spelled saké) is a fermented alcoholic beverage made from rice, commonly referred to as Japanese rice wine. For more information, see Appendix 1. Sense-Making Theory: A semantic theory advocated by Fukaya and Tanaka (Fukaya & Tanaka 1996; Tanaka & Fukaya, 1998). Fukaya and Tanaka claim that the sense of a single word cannot be determined a priori; rather, the sense is cooperatively “made” during the ongoing communication process.

90

 Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

ENDNOTES 1 2

3





In 2014, this work received the Good Design Award in Japan. This type of subjective approach may be avoided in some research fields such as experimental psychology and cognitive science. However, the sense of taste itself is firmly in the subjective domain. It cannot be shared directly with another person, nor can it be represented by sensor information from a “tasting machine,” as the sense of taste cannot be reduced to the sum of its component elements. Astringency and dryness are not basic tastes in an anatomical context (they are algesthesia), but in this chapter, I regard them as “basic tastes.” Note that a salty taste will almost never be noted in the taste of Sake.

91

Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

APPENDIX 1 Sake (Japanese Sake, also spelled saké) is a fermented alcoholic beverage made from rice, commonly referred to as Japanese rice wine. Sake is a fermented food product like soy sauce, miso, and katsuobushi ‘dried bonito’, standard seasonings for Japanese foods. Premium Sake uses special rice called Sake rice, suited for brewing premium Sake (Gauntner, 2011). There are various types of Sake, just as there are various types of wine. Red wine can be roughly broken into three categories: full bodied, medium bodied, and light bodied. For white wine, the gradation from sweet to dry is generally used. In the case of Japanese Sake, sweet types (amakuchi) and dry types (karakuchi) are the most common categories. Light, pale types (tanrei), and full, rich types (nōjun) also appear. In the 1980s and 1990s, pale and dry (tanrei-karakuchi) types of Sake were popular (and remain so). However, around 2000, there was a boom in ginjō-shu (premium Sake made from highly polished rice using a special technique). Ginjō-shu has a floral and fruity flavor.

Figure 4. Window showing KWIC concordance results for the word “say” (Higuchi 2017)

92

Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

APPENDIX 2 In order to clarify which words co-occur with the target words (the adjectives and adjectival nouns listed above), KWIC (Key Words in Context), or simply “concordance,” is used. With KWIC, we can analyze how a target word is used in a corpus. In this appendix, an example of KWIC using KH Coder, a text mining application, is shown. Figure AP-2 illustrates the KWIC for the word “say” in a sample corpus. KH Coder is a very useful text mining tool. It provides statistics, multi-dimensional scale (MDS), word networks, and other helpful visuals. Using the Collocation Stats window in KH Coder (Figure AP-2), it is easy to determine which words frequently appear before and after the target word (or node word). In Figure AP-2, the statistics show that the word “hear” appears thrice in a position two words before (L2) and twice just before (L1) the node word “say.” In addition, we see clearly that words like “Red,” “Shirt,” and “Porcupine” are often used in association with “say” in this corpus. (KH Coder Reference Manual: Higuchi, 2017, Revised)

APPENDIX 3 Concordance score is calculated using the function f(w) shown below, where l1 is the frequency of a certain word w that appears just before the node word; l2 is its frequency, two words before the node word; r1 is its frequency just after the node word; and r2 is its frequency, two words after the node word. In general, the greater the frequency that a certain word w appears before or after the node word (li+ ri), the larger the value of f(w). In calculating the value of f(w), frequencies (li + ri) are divided by “i,” Figure 5. Collocation statistics window (Higuchi, 2017)

93

Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

which weighs the frequencies according to their distance from the node word. Thus, words that appear nearer to the node word (i.e., with a smaller “i”) have greater weight than those that occur five words before or after the node word. In this formula, the frequencies of words that appear just before and after are simply added, since they are divided by unity. (“KH Coder Reference Manual”: Higuchi, 2017)

94

Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

APPENDIX 4

Table 11. Frequency of the verbs in Sake Corpus N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

感じる

word

飲む 広がる 残る 思う 立つ 醸す 含む ふくらむ 引く 入る 消える 現れる 持つ 冷やす 帯びる 落ち着く 続く 出る 出す 漂う 残す 見せる 抜ける 造る 熟す まとまる 変わる 楽しめる 備える 合う できる 引き出す 合わせる 進む まとう 練れる もつ 使う 磨く たたえる 楽しむ 増す 流れる 仕込む 保つ 立ち上がる もたらす 行う 伴う

pron

Eng. Trans

freq. 890

51

kanjiru

feel

52

nomu

drink

416

53

hirogaru

spread

352

54

nokoru

remain

264

55

omou

think

199

56

tatsu

stand

186

57

kamosu

brew

178

58

fukumu

include

155

59

fukuramu

swell

150

60

hiku

draw

145

61

hairu

enter

145

62

kieru

vanish

141

63

arawareru

appear

120

64

motsu

have

113

65

hiyasu

chill

102

66

obiru

take on

101

67

ochituku

settle down

94

68

tsuduku

continue

92

69

deru

come out

91

70

dasu

put out

85

71

tadayou

drift

82

72

nokosu

leave

77

73

miseru

show

76

74

nukeru

slip out

75

75

tsukuru

make

73

76

jukusu

age

72

77

matomaru

come together

70

78

kawaru

change

68

79

tanosimeru

enjoyable

65

80

sonaeru

equip

62

81

au

match up

60

82

dekiru

I can.

60

83

hikidsasu

pull out

59

84

awaseru

match

59

85

susumu

proceed

59

86

matou

take on

59

87

nereru

knead

58

88

motsu

have

57

89

tsukau

use

56

90

migaku

polish

53

91

tataeru

praise

53

92

tanoshimu

enjoy

52

93

masu

increase

51

94

nagareru

flow

50

95

shikomu

train

49

96

tamotsu

keep

47

97

tachiagaru

stand up

47

98

motarasu

bring

45

99

okonau

do

42

100

tomonau

come with

42

101

上げる 目指す 戻る 用いる 覆う でる 押す 取る 取れる 言う 入れる 引き締める 味わう 与える 向く 混じる 似る 示す 膨らむ こだわる まとめる 飲み込む 際立つ 知る 伝わる 比べる 枯れる 引き締まる 搾る 開く 誇る 思える 置く 撫でる つける 押し上げる 伸びる 包む ざらつく 引っかかる 仕上げる 焦がす 食べる 抑える すべる 効く 仕上がる 上がる 醸し出す 成す 来る

ageru

raise

41

mezasu

aim for

41

modoru

Back

41

mochiiru

use

41

oou

cover

40

deru

come out

39

osu

push

38

toru

take

38

toreru

come off

37

iu

say

36

ireru

put in

36

hikishimeru

tighten

35

ajiwau

taste

34

ataeru

give

34

muku

face

32

majiru

mingle

32

niru

resemble

32

shimesu

denote

32

fukuramu

swell

32

kodawaru

stick to

32

matomeru

put together

32

nomikomu

swallow

31

kiwadatsu

stand out

31

shiru

know

31

tsutawaru

pass

31

kuraberu

compare

31

kareru

wither

30

hikishimaru

tighten up

29

shiboru

squeeze

29

hiraku

open

28

hokoru

boast

28

omoeru

feel

28

oku

place

28

naderu

stroke

28

tsukeru

attach

28

oshiageru

push up

27

nobiru

grow

27

tsutsumu

wrap

27

zaratsuku

gritty

27

hikkakaru

get stuck

26

shiageru

finish off

26

kogasu

burn

26

taberu

eat

26

osaeru

curb

26

suberu

slip

26

kiku

do good

25

shiagaru

finish

25

agaru

rise

25

kamoshidasu

brew

25

nasu

make

25

kuru

come

25

95

Defining the Verbs for “Understanding and Interpretation” of Japanese Sake

APPENDIX 5

Multi-dimensional Scaling Performing multi-dimensional scaling with many (50 to 100) words results in a plot that may be difficult to interpret. To make interpretation easier, KH Coder can perform a cluster analysis of words and present these clusters using different colors. Given that only 12 colors can be used at a time to identify clusters on the screen, a maximum of 12 clusters can be specified. This function is not available for three-dimensional plots. If [Adjacent clusters] is checked under the “Multi-Dimensional Scaling Options” area on the right, cluster analysis is performed based on the score obtained in multi-dimensional scaling. This cluster analysis uses the Ward method based on Euclidian distance. Thus, words plotting close to each other are classified in the same cluster. Clearly, clustering can assist interpretation of the results of multi- dimensional scaling. This type of cluster analysis is also applied in “Concept Mapping” proposed by W. M. K. Trochim (1989), and is thought to be an effective method for exploring data with an unknown group or category structure (Afifi & Clark 1996).

96

97

Chapter 5

300–500 Threshold of Context Memory and Assumption Experiment Nobuyoshi Harada Chiba University, Japan

ABSTRACT This chapter investigates the threshold of context memory on numbers 300 through 500 regarding perspective with assumption experiment and theories of cognitive semantics and short-term memory. Context generation became possible with text database with lexical engineering system on artificial intelligence (AI). It was realized an infinite and continuous presentation of context. Several context presentations realized the limitation of context memory on the process of continuous context presentation on numbers 500 and 2,000, as well as the representation process with judgement for memory. The first report studied the context generating machine for Japanese pun generation. This was termed “Class B engine” or “Dajare Machine.” This study focused on the examination of the context threshold using an experiment of human brain mapping on an auditory evoked magnetic field (AEF) of a magnetic encephalogram. It also focused on the analysis of the theory of cognitive semantics and folk biology, using a theory of short-term memory of the “magical number seven.”

INTRODUCTION There was no discussion on the limitations in the memory of context in the “Poetics” of Aristotle (1949) because it is extremely difficult to generate and expose (present) the context on the over number of limitations of humans. The context-generating system of artificial intelligence (AI) was possible to generate the unlimited number of contexts. One example was the “dajare machine” (Japanese pun-generating machine), which could generate an unlimited number of contexts, with the over number of limitations of the homogeneous context. A qualitative difference was observed between over- and underlimitation on the effect of memory construction (Appendix 1). DOI: 10.4018/978-1-7998-4864-6.ch005

Copyright © 2021, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

 300–500 Threshold of Context Memory and Assumption Experiment

This chapter describes the quantitative difference in the number of contexts exposed by the assumed experiment of the auditory evoked magnetic field (AEF). The establishment of context memory was investigated with two conditions of the over and under numbers of thresholds of context memory. This chapter also presents the concept of the context memory threshold from the perspective of working memory of magical number seven and the semantic classification of taxonomy on hierarchical structure. The part of the function of the context threshold of taxonomy on hierarchical structure was reflected in the function of the long-term working memory. The narrative and pun had a functional base of long-term declarative memory. The human memory function was classified into short-term memory and long-term memory on the property of time course of memory preservation. Long-term memory was divided into two types of memory, namely, declarative memory and nondeclarative memory, with the property of recollection for consciousness (Squire & Zola-Morgan, 1985). Tulving (1972) reported the definition of the two types of long-term memory for semantic memory and episodic memory. Semantic memory is the memory of word and word semantics and concepts and the semantics and relationships of perceptual objects, as organized memory. Episodic memory is the memory of events that individuals experience. According to Tulving, semantic memory is constructed by repeated experience, while the memory of the situation of acquisition disappears. Episodic memory is constructed by a single trial experience, while the memory of the situation of acquisition is maintained with the memory of the episode. The memory of the narrative plot and pun is constructed by a single trial experience, while the memory of the situation of acquisition disappears. The memory of the narrative plot and pun is midway between semantic and episodic memory. The pun was possible to adopt for one narrative division with the definition of the narrative genre (Ogata, 2018), which was included and dominated by narrative factors, as the broadly defined narrative, which did not have independence of artwork and had some property of accessory and method of other artworks. Adam (1984) described the predictability of narrative and narrative genre (fantastic story, journalism story, and laughter) of discourse. The expectations of narrative and individual narrative genre of discourse were established by the predictability of the organizational form of narrative and the individual discourse of narrative genre, which was supported by the shared knowledge between the narrator and listener and general reasoning. In this chapter, the authors of this study discuss the threshold of context memory. A difference in the memory context was found in the process of recalling context memory with different contexts of numbers of 500 and 2,000. The process of recalling the sound file of 500 contexts provoked the sense of having already heard these sounds. This was not the same for the process of 2,000 contexts. This result indicated a limitation and threshold of context memory with numbers between 500 and 2,000. The threshold may control the status of the memory and novelty contexts, as it controls the presentation of the number of contexts. This chapter first discusses the case study and background of the context threshold of 300–500. The case study of memory construction with “dajare machine” is the first case of overexposure (and presentation) of the context memory threshold using an AI system. This chapter presents a case study of 300 thresholds in “Shippaigaku,” the 316 cases of counseling in “A Common Casebook,” and number of limitations of 300 of Rakugo Stories. This chapter also discusses the main focus of the assumption experiment. Study 1 shows the background of human brain mapping of the relationship between anticipation and variation of average waveform of an AEF. Study 2 shows the actual case of variation of average waveform with the anticipation of the 98

 300–500 Threshold of Context Memory and Assumption Experiment

sound of mora with the status of memory-reflected familiarity. Study 3 shows the effect of memory status as familiarity on stimulus anticipation of four-character Chinese idiom with an auditory evoked magnetic field. The preprocessing of the first sound changed the response of the second sound with the elicitation anticipation for arrival of the second sound. Study 4 shows the design of the assumption experiment. The difference in memory status is discussed from the perspective of an AEF between 500 and 2000 Japanese pun presentation. The difference in memory status made it possible to change the average waveform of the second sound with the preprocessing of the first sound of the elicitation of the anticipation of the second sound. This chapter finally discusses the concept of the context memory threshold. This discussion is constructed with two perspectives: the hierarchical structure of semantics as cognitive semantics and magical number seven as the function of working memory. The first discussion is constructed with the peats of the hierarchical structure of semantics, folk biology, and layers of semantics. This first discussion presents the perspective of the function of three layers of semantics. The second discussion presents the perspective of magical number seven as a function of working memory. The perspective of the first and second discussions is combined and investigated in the last discussion of the hypothesized concept with the semantic layer and magical number seven. This perspective is supported by the hierarchical structure of the long-term working memory.

BACKGROUND AND CASE STUDY Dajare Machine The dajare machine, also known as a “class B engine,” generates Japanese puns, both continuously and infinitely, using a lexical engineering system of AI. This system was developed to present the concept of the language engineering study group, Kotoba Kougaku (linking art and engineering), and research results. The dajare machine was named as “slightly generating infinite puns” because it generates an infinite number of Japanese puns (Abe, 2011). The dajare machine creates Japanese puns through either a printed or voice output. There are four types of content. The first three patterns are combinations of two similar words, including an operation to insert a sound word, such as the end of a sentence. The fall of the sentence reflects the contemporary social situation.

500 to 2,000 Sound Files This was the first case reflecting the threshold of context memory using a dajare machine. The threshold was found via a process of presenting Japanese puns of sound files generated in a number of contexts. The 38th Language Engineering Workshop was held on August 26 and 27, 2011, at Kanagawa University Yokohama Campus, Building Number 23. The dajare machine was exhibited by Professor Kazumitsu Matsuzawa’s laboratory. To simplify the mechanism, the system was displayed to play the sound files. This was generated and recorded using a dajare machine. The author had an opportunity to obtain a comment from Professor Matsumoto regarding the number of contexts of Japanese puns and subjects’ responses related to the memory of context. 99

 300–500 Threshold of Context Memory and Assumption Experiment

Figure 1. Class B engine system developed to present the concept of the language engineering study group Kotoba Kougaku and research results

In trials where Japanese puns were regenerated in 500 contexts, participants could recognize regenerated puns the second time they heard the sound files. However, when puns were regenerated in 2,000 contexts, the participants were unable to recognize the regenerated puns upon hearing them the second time. This result reflects the presence of a memory threshold between 500 and 2,000 contexts. The data indicated a qualitative difference in the context of memory processes when exposed to 500 and 2,000 contexts related to recall. Thus, the result reflects the limitation property of context memory.

100

 300–500 Threshold of Context Memory and Assumption Experiment

Shippaigaku Shippaigaku (i.e., the Japanese word for “failure analysis”) concerns the sampling and analysis of failures to create failure measures and manual operation of failure status (Hatanaka, 2000). Chapter 8 of Shippaigaku provides a discussion on the limits of manipulation number failure objects for the construction of measures and manual operations. The number of limitations of manipulation should not exceed 300. If manual chapter numbers exceed 300, an omission of the discovery of the chapter could be provoked over 300 chapters (e.g., 20,000). A knowledge unit of failures and their cause and effect could have limitations structured in groups and hierarchical arrangements. The construction of the structure of manual chapters was required for grouping and hierarchical arrangement to understand the background of failures. Grouping and hierarchical arrangement of chapters were reflected with manual chapter threshold of 300.

An Uncommon Casebook An Uncommon Casebook is a collection of 316 case studies published by Dr. Milton H. Erickson (O’Hanlon & Hexum, 1990). Erickson grouped these cases into 45 units with nine classified sets: (1) habits and impulses; (2) physical problems and distress; (3) sexual problems; (4) sleep problems; (5) fear and emotional problems; (6) cognitive and communication problems; (7) marriage, family, and human relationship issues; (8) serious disabilities in behavior or cognition; and (9) other issues. Units were grouped as finger sucking and claw biting issues. There were three layers within each group and unit. The number of members in each group and/or unit deviated between 1 and 12 and from 2 to 9. These averages did not deviate significantly from 7. Seven was known as the limitation of short-term memory (Miller, 1956). The group set structure and unit of cases showed a limitation to the number related to recognition, classification, and systematization of counseling cases.

Limitation of Memory Related to Rakugo One chapter of the book Shippaigaku described limitations related to human memory. This section presents limitations related to number memory in the Japanese storytelling termed “rakugo” (Hatanaka, 2000). Enshou was a famous master of rakugo storytelling. A rakugo storyteller has approximately 100 stories in his/her life. However, Enshou’s 300 stories were relayed with accurate memory. The book Shippaigaku noted that 300 was the limitation of the memory of numbers.

MAIN FOCUS 1: ASSUMPTION EXPERIMENT Observation: 300 to 500 Context Memory The threshold of context memory assumed an experiment of 300 to 500 threshold of context memory with a fall on a brain mapping experiment using magnetoencephalogram and electroencephalogram based on the difference of memory status reflected with familiarity of context between under and over number of threshold on auditory evoked magnetoencephalographic and electroencephalographic fields. The threshold of context memory that exists between 500 and 2,000 on context number was reflected with a recognition process for the regeneration of 500 to 2,000 contexts of Japanese puns. The difference 101

 300–500 Threshold of Context Memory and Assumption Experiment

in recognition allowed us to reflect the states of degree of memory for words of Japanese puns between 500 and 2,000 contexts. Moreover, different states of the degree of memory of words allowed expression with the degree of familiarity of words. The effect of word familiarity allowed the investigation of the variation in AEF and auditory evoked electrical potential (AP) in human subjects. It is possible to show the experimental design of the examination of the effect of the 300–500 threshold with the difference in provocation of word familiarity between 500 and 2,000 Japanese puns.

Study 1: Anticipation and AEF Figure 2. Response anticipation of the averaged waveform on different inter-stimulus intervals and word familiarity

Engstrom, Kelso, and Holroyd (1996) observed that the reaction time for tapping to a metronome sound was a Gaussian distributed relative to the metronome timing for short inter stimulus intervals (ISIs), such as 1 or 2 s. On the contrary, the reaction time for long ISIs, such as 4 or 8 s, was distributed as a large, sharp, and narrow peak almost 200 ms following the metronome. The difference in the response between the short and long ISI conditions was thought to be derived from anticipating the timing of the metronome. The subject could not anticipate the timing of the metronome with a long ISI but could anticipate it with a short ISI. The averaged waveform of the N100m component with changes in the amplitude and latency were also observed with changes in the length of the ISI in the fixed ISI form, in the short ISI (i.e., 1–2 s) and long ISI (i.e., 4–8 s) conditions. Näätänen (1986) reported a large N100m component in long ISI

102

 300–500 Threshold of Context Memory and Assumption Experiment

conditions, such as 4 or 8 s. The amplitude and latency of the N100m component increased with the increment of the ISI interval from 1 to 2 s to 4 to 8 s (Appendix 2). In a previous study (Amano, 1996), subjects were asked to press a button as soon as they recognized the phoneme /k/ in Japanese spoken words or nonsense words that consisted of four morae. The phoneme was either in the first, second, third, or fourth mora, and its position was always similar in words and nonsense words. The response time was measured under three conditions of different speeds with which the words were played (i.e., word duration): time-compressed (× 0.5), normal (× 1.0), and time-expanded (× 2.0). The response time for phoneme /k/ was faster in words than in nonwords for all three duration conditions but only if the phoneme was in the third or fourth mora. When the phoneme was in the first or second mora, response times for the phoneme were similar in words and nonwords, regardless of the duration condition. Nonsense words are, by definition, completely unfamiliar; therefore, subjects could not anticipate the mora sounds in this condition. Thus, in this study, the familiarity of a word facilitated the detection of a phoneme in the third or fourth mora of 4-mora-long words. Amano (1996) suggested that the brain processes word sounds (phoneme recognition) independently of word duration and that the facilitatory effect of word familiarity on word recognition is independent of word duration. If Amano’s (1996) finding that the reaction time for a particular phoneme does not depend on word duration is a reflection of the general word recognition function of the human brain, it implies that the brain processes word sounds similarly for different word durations. Furthermore, it implies that the effect of word familiarity on word recognition will be similar for different word durations and between continuous word sound streams and clearly separated phoneme sounds. The reaction time was different in the third and fourth morae of words between words and nonwords; the difference in anticipation function appeared on the onset time of the third and fourth morae. The average waveform of the word sound (speech sound) showed the N100m component at 100 ms of sound onset time and broad N100m complex after the N100m component. The broad N100m complex was considered the sustained field, and the sustained dipolar field pattern was estimated as one single ECD with location similar for N100m (Eulitz, Diesch, Pantev, Hampson, & Elbert, 1995; Hari, 1991). The average wave of the N100m complex for word sounds could be adopted as word-related complex (WRC). The effect of familiarity was observed on the amplitude and latency of the WRC. The decrease in familiarity was expressed as the increment of amplitude and prolongation of latency of WRC (Harada, Iwaki et al., 2005a).

Study 2: Word Familiarity with AEF The effect of word familiarity of auditory stimuli on the word-recognizing function of the human brain was investigated (Harada, N., Iwaki, S., Nakagawa, S., Amano, S., Yamaguchi, M., & Tonoike, M. 2005a,b). Word familiarity, characterized by facilitation and accuracy of word recognition, is an index of the relative ease of word perception (Amano, Kondo, & Kakehi, 1995; Connine, Mullennix, Shernoff, & Yele, 1990). The author studied the effect of word familiarity using four morae of Japanese auditory words as auditory stimuli on the elicitation of AEFs with a word-naming task. The author selected words from a database of the lexical properties of Japanese words based on auditory familiarity (Amano & Kondo, 1999). The author grouped and presented the four morae of Japanese auditory words in four classes of degree of familiarity. The researcher used three sound conditions as the auditory stimuli for recording the AEFs. The first condition was a normal length of a word sound (word 1/1 condition). The second was 103

 300–500 Threshold of Context Memory and Assumption Experiment

shortened by 50% for the length of a word sound (word 1/2 condition). The third was a simple pure tone at 1 kHz with 100 ms duration (1 kHz, 100 ms condition). This was the control condition to record the simplest AEF waveform. The researcher observed the waveforms of AEF on channels of the maximum amplitude of N100m components in the left and right hemispheres. Figure 3. Average waves of auditory evoked magnetic field of the magnetoencephalogram of the four Japanese mora sounds used in four classes related to the degree of familiarity (The thickness and pattern of line indicates familiarity [thick line, 7.0–5.5; thick dotted line, 5.5–4.0; fine line, 4.0–2.5; thin dotted line, 2.5–1.0]) Source: Harada, Iwaki, Nakagawa, Amano, Yamaguchi, & Tonoike, 2005a

The author observed WRCs after the N100m component as large broad components in the word 1/1 and 1/2 conditions in the left hemisphere channel, rather than the 1 kHz 100 ms condition (Figures 1 and 2). As shown in Figure 3, the peak amplitude of WRCs in the word 1/2 condition significantly decreased (F(3/27) = 9.44, p = 0.000195). The 50% amplitude crossing latency of WRCs of the maximum amplitude in the four conditions of familiarities significantly shortened (F(3/27) = 11.71, p = 0.0000427) with an increment in the value of familiarity on the decreasing curve of WRCs (Figure 2). Some scholars reported that the increment in the regularity of interstimulus intervals (with the increment of exponent n of fluctuation 1/fn) provoked a decrement in the N100m amplitude and latency shortening with the construction of memory traces of stimuli or short-term memory traces (Harada et al., 2001; Harada, Masuda et al., 2005). The amplitude decrement and 50% amplitude crossing latency shortening of WRCs were provoked by an increment in the value of familiarity. This may be reflected in

104

 300–500 Threshold of Context Memory and Assumption Experiment

the internal construction of memory traces for word sounds (long-term memory traces) constructed with experience from exposure to and generation of word sounds in a lifetime (Harada, Iwaki et al., 2005a). The researcher observed the effect of word familiarity of auditory stimulus of four morae of Japanese auditory words on the amplitude decrement and 50% amplitude crossing latency shortening of WRC in the word 1/2 condition. Long-term memory traces constructed with the experience of word sounds resulted in the WRC waveform in the left hemisphere.

Study 3: Effect of Memory (Familiarity) on Stimulus Anticipation of Four-Chinese-Character Idiom with AEF The author investigated the effect of word familiarity and incongruity of auditory stimuli with fourChinese-character idioms (four kaiji idiom) with a difference in familiarity of four-Chinese-character idiom and congruity or incongruity of idiom with exchange of forward and backward idioms with the same familiarity group (Harada, Iwaki, & Tonoike, 2006c). A previous study reported that they expressed the effect of word familiarity on AEF on increment in amplitude and shortening of latency of the N100m component of WRC on AEF (Harada, Iwaki et al., 2005a). The expression of congruity and incongruity of auditory stimuli of four-character Japanese idiom were investigated at different levels of familiarity (i.e., high and low familiarity of Japanese) with or without exchange of the forward and backward parts of the idiom (Appendix 2). Figure 4. Average waveforms of auditory evoked magnetic fields of magnetoencephalographs elicited by word sounds of congruity and incongruity in a Japanese sentence of four Chinese characters with high (7.0–5.0) and low (2.0–1.0) degrees of familiarity (recorded in channels with maximum amplitude of N100m components of the left hemisphere channel located on the supratemporal auditory cortex) Source: Harada, Iwaki, Tonoike, 2006c

105

 300–500 Threshold of Context Memory and Assumption Experiment

The author grouped and presented the four-character idiom (four kaiji idiom) of Japanese auditory words into two classes (i.e., high, 7.0–5.5, and low, 2.0–1.0) of degree of familiarity. The researcher separated the four-character idiom using the forward and backward parts of the idiom. The author investigated the congruity and incongruity conditions of idioms with and without exchange of forward and backward parts of idiom with the group of idioms with the same familiarity. Time interval of auditory stimuli of the forward and backward parts of the idiom was located at 1000 ms. The researcher observed waveforms of AEF of the forward and backward parts of idiom sounds on channels of the maximum amplitude of N100m components in the left and right hemispheres. The author observed WRCs after the N100m component in the left hemisphere as large broad components in four conditions of high and low familiarity and with congruity and incongruity on the forward and backward parts of the idiom. The peak amplitude and latency of the WRCs of the forward part had no difference in the four conditions. The peak amplitude and latency of WRCs of the backward part decreased only with high familiarity and no exchange (congruity) condition. There was no difference in the amplitude of WRCs in the backward part of the three other conditions (i.e., high familiarity and exchange, low familiarity and no exchange, and high familiarity and exchange). Auditory stimuli of the backward part of the idiom with high familiarity and no exchange condition allowed anticipation of the stimulus presentation. Instead, the stimulus presentation of the backward idiom part exchanged with high familiarity condition was not anticipated, even in low familiarity conditions with exchange and no exchange conditions. The researcher observed that the sound set of idioms expressed WRC amplitude decrement and latency shortening of the backward idiom on the condition of anticipating the backward idiom. The anticipation process may reflect the internal construction of memory traces for word sounds (i.e., long-term memory traces) constructed with the experience of exposure to and generation of word sounds. The amplitude decrement and latency shortening of N100m of WRC allowed the investigation of the provocation of anticipation of auditory stimuli with construction of memory of word sounds on constructed familiarity.

Study 4: Design of Assumption Experiment to Investigate for Context Memory Threshold of 300 to 500 Numbers The sections above illustrate the general structure of the experiment. To examine the difference between the construction of context memory and presentation of Japanese puns with a number between 500 and 2,000, it is necessary to study the difference in amplitude and latency of AEF and AP of WRC of Japanese puns (Harada, 2019). Construction of context memory was different in the presentation of the number of Japanese puns between 500 and 2,000. Memory construction was reflected in anticipation of the sound of context and allowed the examination the provocation auditory response of AEF and AP of WRC. The author investigated memory construction of 500 and 2,000 Japanese puns by provocation of AEF and AP of WRC. Japanese puns can be prepared using the dajare machine for numbers 500 to 2,000, with 400 for the measurement of AEF and AP of WRC. The researcher was also able to sample the Japanese pun from the Web data available on its website. The author sampled the Japanese pun with two criteria of selection: 1) easiness to understand for individuals without knowledge or expertise and 2) clearly distinguishable structures of the forward and backward parts. The experimental design consisted of two processes: presentation process and measurement process. The author presented Japanese pun groups in the range of 500–2,000. After the presentation of the Japa106

 300–500 Threshold of Context Memory and Assumption Experiment

nese pun sound, the listener has a 15-min break; then, the measurement of the auditory response starts. The author analyzed the measurement of AEF and AP of WRC with 100 pun sounds sampled randomly from the range 500–2000 of Japanese pun groups. The author represented the sampled 100 puns with 200 new Japanese puns with random mixed conditions. The researcher asked the listener to assess new or already heard sounds for the presented Japanese pun with two selection buttons. The author separated part of the Japanese pun sounds into the forward and backward parts, with a time interval of 1000 ms. The researcher averaged the auditory responses 100 times for the measurement of AEF and AE of WRC. The design of experiment was constructed with two processes of pre- and postpresentation of word sound under two conditions of 500 and 2,000 numbers.The 500 and 2,000 sounds of Japanese puns were in the prepresentation process and then the 100 responses of selected word were averaged for AEF. The selected 100 word sounds were presented a postpresentation with 200 new Japanese pun word sounds. The auditory response of 100-word sounds were averaged for AEF. Subsequently, the amplitude and latency of the AEF of WRC were compared with the backward part of the response of WRC between pre- and postpresentation of word sound. The 500-word sound presentation condition decreased the amplitude and shortened the latency on WRC with construction anticipation with the structure of context memory. Figure 5. Designed experiment of results of auditory evoked magnetic field of 500 or 2,000 word sound presentation of Japanese puns

107

 300–500 Threshold of Context Memory and Assumption Experiment

The author studied the difference in amplitude and latency of AEF and AP of WRCs with a backward sound of Japanese puns from a selection of 100 puns from the range of 500–2,000 groups. It was impossible to anticipate the forward part sound of the Japanese pun when the sound of the stimuli arrives. Instead, the researcher could influence the anticipation of the backward part sound with the construction of memory of the context of the Japanese pun. If context memory of the Japanese pun was constructed with the first presentation, the sound memory of the Japanese pun was easily provoked with the forward part of the sound presentation. This arrived with the backward sound, as it easily anticipated it via sound memory. It was possible to produce anticipation of the backward part sound with decrement of amplitude and shortening of latency of the N100m and N1 component of AEF and AP of WRCs. If context memory was not constructed, there was no amplitude decrement or latency shortening with AEF and AP of WRCs. It was possible to investigate the construction of context memory by the variation of AEF and AP of WRCs. The author was able to observe the amplitude decrement and latency shortening with 500 Japanese pun groups, but not with 2,000 Japanese puns. This resulted in the construction of context memory in the 500 Japanese pun groups, but not in the 2,000 Japanese pun groups. Figure 6. The detection of context memory threshold 300–500 is shown by the difference in the establishment of memory of the pun presented in the range of 500–2000

The design of the experimental steps is shown in Figure 6 for the detection of the threshold of context memory with the difference in the status of memory of the pun, which was presented under or over the threshold.

108

 300–500 Threshold of Context Memory and Assumption Experiment

Step (1) consisted of the 500 and 2000 pun presentation. On the adoption of the case study of the dajare machine, the 500 pun presentation established the memory, while the 2000 pun presentation did not. In the first and second phrases, one pun constructed. This is the memory establishment process of a pun. Step (2) consisted of a memory checking process for two conditions. The 100 puns were corrected from the groups of 500 and 2000 puns. The corrected 100 puns were presented after the 500 and 2000 pun presentations with short break. An AEF of 100 puns was recorded and averaged 100 times for the detection of the average waveform. In the second pun presentation of 90 times (Step 2-1), the presentation of the first phrase recalled the second phrase and anticipated the second pun. The condition of 500 puns recalled and anticipated the second pun. The condition of 2000 puns did not recall or anticipate the second pun. At the time before the second phrase presentation, the sound arrival of the second pun was anticipated. In Step (2-2), after the second phrase presentation, the facilitation of auditory informational processing in the auditory cortex occurred. The condition of the 500 pun presentation maintained the anticipation, while the condition of 2000 pun presentation did not. The result of Step (2-2) appeared on the variation of the average waveform of AEF. The condition of 500 puns allowed the facilitation of informational processing, while the condition of 2000 puns did not. The condition of 500 puns of AEF decreased the peak amplitude and shortened the latency, while the condition of 2000 puns did not. The difference in AEF between the conditions of 500 and 2000 puns indicated the difference in the establishment of memory for the presented pun and showed the existence of the threshold of context memory.

MAIN FOCUS 2: CONCEPT OF CONTEXT MEMORY THRESHOLD Hypothesis of Semantic Layer and Magical Number Seven The author examined the 300–500 threshold with concepts of short-term memory limitation of the magical number seven and semantic layer development of words in early childhood. The number of objects held in short-term memory was 7 ± 2. The researcher constructed a semantic layer of words with the basic level on a fast process. The superordinate level was on the second process (4 years). The subordinate level was in the third process (5 or 6 years). The researcher expressed the 300–500 threshold functional base with the function of short-term memory of the magical number seven on three semantic layers of word for context memory. The 300–500 threshold can be shown as 7 × 7 × 7 = 343 (memory number of short-term memory process was presented as 7 ± 2) and with the taxonomic function of long-term working memory (Ericsson and Kintsch, 1995).

Structural Model of Context Memory Threshold Shippaigaku presented a manipulation threshold of 300 in the chapter number of operations manual of machine (Hatanaka, 2000). It was necessary to construct the structure of chapters for their grouping and hierarchical arrangement to understand the background of failure reflected with 300 chapters. The hypothesis of context memory threshold was influenced by memory property of grouping and hierarchical arrangement of memory target of context. The author discussed the memory property of hierar109

 300–500 Threshold of Context Memory and Assumption Experiment

chical arrangement by the theory of the semantic layer; it was rerating for the property of the layer on the memory target. Further, the author illustrated the memory property of grouping by describing the short-term memory concept of the magical number seven (Miller, 1956); it was rerating for property of limitation and operability on memory target. The author discussed the hypothesis of context memory threshold using the semantic layer theory and the theory of short-term memory concept of the magical number seven.

Hierarchical Arrangement of Memory Hierarchical Structure of Semantics The author constructed a category from a hierarchical structure in relation to inclusion. The semantic feature theory of structural semantics shows hierarchical structures of semantic relations as hyponyms (i.e., low-layered words). These include the semantic features of hyperonyms (i.e., high-layered words). These belong to the idea of the discrete and binary oppositional feature of the category (Katz & Fodor, 1963; Katz & Postal, 1964). All category levels of the hierarchy have equal importance in structural semantics. Instead, cognitive semantics attach special importance to a single category level in the hierarchical category. The specific category, or basic level, plays a central role as the structural base for the superordinate level (i.e., higher layer of the basic level) and subordinate level (i.e., lower layer of the basic level). The basic level depends on the semantic theory of the prototype. Cognitive semantics describes the function of a prototype. The semantic category of words, namely, the prototype and nonprototype, are central members (prototype = robin or sparrow of the bird member). The peripheral members (nonprototype = ostrich or penguin of the bird member) in a semantic category have a subjective judgment of prototypicality (Barsalou, 1985). Members of this category do not have equal importance (Rosch, 1973, 1977). The role of the prototype includes (1) a short response time to discriminate belonging to the members of the category; (2) the facility to image a typical example of a category member; and (3) ease of learning. The basic level is the prototype of the taxonomic layer in a hierarchical structure. Its functional roles include (1) concentration of number information compared with other layers; (2) first layer of a category for children to learn words; (3) maximum layer of frequency of use; (4) expression with one word; and (5) less difference in interferences between languages.

Folk Biology All humans have the word taxonomy of animals and plants. Folk biology, a research area in the field of anthropology, is a study on the structure of the taxonomy of words regarding the natural world of animals and plants. Berlin (1972, 1992) and Berlin, Breedlove, and Raven (1973) reported that all human words have the hierarchical taxonomic structure of the natural world. In addition, layers of hierarchy were limited to the fifth or sixth. Limitations on the number of layers come from limitations of a human cognitive function. The first layer, termed “unique beginner,” corresponds to the words “plant” or “animal.” The second layer, termed “life form category,” corresponds to “tree, bush, vine, grass or herb” or “bird, fish or mammal.” The third layer, termed “generic category,” corresponds to “oak or maple” or “cat or sheep.” This

110

 300–500 Threshold of Context Memory and Assumption Experiment

layer, which includes the maximum number of words, plays the role of a central layer in hierarchical taxonomy. The fourth layer is termed “specific category.” The fifth layer is termed “varietals category.” Berlin (1972, 1992) reported the existence of an intermediate layer between the “life form category” and “generic category.” He called this the “intermediate category.” Rosch, Carolyn, Wayne, David, and Penny (1976) reported that the “generic category” functions as the basic level for the learning process of language.

Semantic Layer The basic category is the first layer for children to learn words. The high taxonomic layer in the hierarchical structure is for children to learn words at the age of 4 years. Low-high taxonomic layer is for children to learn words at the age of 5 or 6 years. The three layers of the generic category (or basic level), plus the low-high taxonomic layer, are fundamental players in the construction of the semantic concept of words in a hierarchical structure in the developmental process. The maximum number of taxonomic layers in folk biology was five (or low frequency six) from the limitation of human recognition. High-frequency layers include “life form category,” “generic category,” and “specific category.” “Generic category” has been used for the first time in context, while the highest layer of “unique beginner” and lowest layer of “varietal category” have been used with low frequency. Many languages do not have a taxonomy layer of “unique beginner.” The “varietal category has been used for objects of particular interest within a culture. Three semantic layers have played a role in the facilitation of categories in the recognition process. The role of “basic level” was explained as a category with a maximum number of common factors and the same category members. This category has a minimal number of common factors with different category members. A high layer of category has some similar factors with other category members. A low layer of category has some similar factors with other category members. The “basic level” layer balances with the sameness of the belonging factor with internal and external category members. This perspective is applicable for three layers. One or two layers are insufficient for the category group. When the category is able to include factors, it should be classified in another category. Four or five layers are too many. When the different categories include the same factor, they should be classified in another category. The three-layer structure was enhanced as an efficient categorization of the object.

Working Memory Magical Number Seven Miller (1956) discussed the limitations of short-term memory using the experimental results of the limits of one-dimensional absolute judgment. The experiment of a one-dimensional absolute judgment task was proposed to a subject having a corresponding response to a stimulus (previously learned). Stimuli were varied in one dimension (e.g., 10 different tones varying in pitch). The response of the subject was kept for accuracy of judgment, with the number up to five or six for different stimuli. The accuracy of judgment decreased with an increase in the number of stimuli over five or six. Miller (1956) discussed limitations surrounding short-term memory based on differences in judgment accuracy with a number of stimuli (more or less than five or six). He termed the limitations of short-term memory using the results of the experiment as the magical number seven. According to Miller (1956), 111

 300–500 Threshold of Context Memory and Assumption Experiment

humans are limited to the number of objects they can hold in their short-term memory (7 ± 2). Memory limitation of the number or word is termed “chunk” (i.e., it does not exceed 7 ± 2). Chunk is different to objects (i.e., the number was 7, the letter was 6, and the word was 5, approximately). The short-term memory limitation of 7 ± 2 may play a role in the process of operability in memory. Various studies on short-term memory report a different number of limitations. It is clear that there is a limit to the number (nearly seven) related to the function of short-term memory. Ericsson and Kintsch (1995) reported the taxonomic structure of chunks of the magical number seven. Chase and Ericsson (1981) found the expert performance memorization of a list of 30 digits. They were able to assess a hierarchical retrieval organization that experts used to memorize lists of 30 digits. The function of long-term working memory was found in memorization of text. The memorization process of text uses retrieval cues to the hierarchical organization of encoded text to provide access to text information (Fletcher 1981) (Appendix 3)

Hypothesized Concepts: Semantic Layer and Magical Number Seven Construction of the context memory threshold of 300–500 was hypothesized using the concepts of the semantic layer and magical number seven. The threshold number was explained with a three-layer structure of folk biology and limitation of the number seven for short-term memory. If the grouping role of short-term memory limitation seven was plated on all three layers, the number of manipulated objects could show 7 × 7 × 7 = 343 on the three layers. The number 343 did not deviate from 300. The number of manipulated objects for variation was 8 × 8 × 8 = 512 and 9 × 9 × 9 = 729, which indicates that the magical number deviates according to the object. Numbers 512 and 729 exceeded 500. However, these numbers were clearly lower than 2,000. The number seven could facilitate the threshold for the manipulation of memory of the recall function in the process of grouping and layer forming. Grouping and layer forming with the number seven may result in smooth memory recall without extreme effort. A three-layer structure and the number seven could play a factor in quantitative constraints for expression of the threshold of context memory on the background of the theories of folk biology, cognitive semantics, and short-term memory as the magical number seven. The threshold in the range 300–500 was expressed with a grouping function in short-term memory on the magical number 7 ± 2 in three layers of the hierarchical structure of semantics. Three layers corresponded, as the first layer was the life form category and superordinate level, the second layer was generic category and basic level, and the third layer was a specific category and subordinate level, based on the theories of folk biology and cognitive semantics. Two concepts of the magical number seven and three major layers of the theories of folk biology and cognitive semantics allowed us to show the threshold of context memory of the numbers in the range 300–500. The concept of long-term working memory was possible to help develop the research threshold of 300–500 (Appendix 3).

112

 300–500 Threshold of Context Memory and Assumption Experiment

Figure 7. Concept of expression of threshold contest memory constructed for the property of grouping and hierarchical arrangement by the theories of short-term memory as the magical number 7 ± 2, semantic layers of cognitive semantics, and folk biology

DISCUSSION Perspective of Limitation of the Chunk Layer The significance of the context memory threshold was discussed from the perspective of long-term working memory. A study on long-term working memory presented the perspective of layer structure between chunk and memory key, and the component of chunk and memory key can also create a layer structure. The limitation of the chunk component layer expressed the limitation memory, which had a relationship with chunks. The concept of context memory threshold proposed the layer limitation of working memory as three layers from the perspective of cognitive semantics. The limitation of the layer brought the limitation of the number of memories as context. The concept of context memory threshold was possible to present the limitation of the chunk layer in the study of long-term working memory, as it was possible to play the limitation of memory in the number of components in the layer structure.

Overcoming the Limitation of Context Memory The significance of context memory was discussed from the perspective of context generation with the AI system. It was difficult to generate an over-number context of threshold with human activity. In con-

113

 300–500 Threshold of Context Memory and Assumption Experiment

trast, the context generation with AI system was almost infinite, as in the case of the “dajare machine.” The AI system of infinite context generation led to the first investigation of the limitation and threshold of context memory. The AI system presents qualitative differences in the generated context of over or under the threshold.

Framework for Describing World The significance of context memory threshold was discussed from the perspective of the utilization of AI for the external environment of humans. The perspective of context memory provides a framework for describing the world. Shippaigaku indicated directions for understanding its background. An object number of 300 was described as a limitation number for recognition. This object number was a framework for others in describing the background for the world’s failure. If a report of the outside world was produced with AI, a framework with limitation 300 was helpful in its understanding. Stories generated about the outside world through narratology also utilized this framework. In Shippaigaku, a normal rakugo storyteller has approximately 100 stories. Several objects are possible within the framework, decreasing it from 300 to 100. The context threshold is a tool in technology maintenance related to the novelty of context. It provides a framework for others when describing the world (Appendix 4).

Finiteness of Context and Music Social significance and social function of context memory threshold were discussed from the perspective of the lack of construction of boring and monotonous feelings for the limitation of context and music. The finiteness of music was discussed based on the law of conceptual design of music patents (Mizuno, 2017). Music (as melody), which was expressed in notes on musical score, was finite. The property of finite music was brought into Western music. Contemporary musicians have the assumption of finiteness of music, reuse of music, and a combination of musical components. They also attempted to produce new grooves, feelings, and ambience. They have constructed their music with frequency variation over time; the number of patterns they have been able to express has been limited. Moreover, the chance to express new music patterns has been limited and has decreased. The depletion of the new music patterns was discussed in the phenomenon of the appearance of the same music on various irrelevant places of music composition. The reproduction environment of music has been greatly developed as a result of Internet technology, while the depletion of new music patterns has significantly progressed. Conversely, music listeners have not had the pressure of feeling bored by music; rather, they have had fun listening to music and experienced mental well-being. It was possible to maintain the novelty of music through the function of the memory threshold. In the modern world, highly developed technologies for the visual, auditory, and/or somatosensory presentation of information have led to a reduction in the novelty of information. The degree of loss of informational novelty has rapidly progressed through the establishment of photography technologies in media. Maintenance of the novelty of context is an important issue in the information management of various media. This perspective of context memory limitation may help technology in maintaining the novelty of context while controlling the number of presenting contexts. 114

 300–500 Threshold of Context Memory and Assumption Experiment

Walter Benjamin The quality of duplication technology has evolved. The development of technology has been ravaged with the presentation of context. The technology of visual, auditory, and somatosensory information has been greatly developed. The novelty of information in the inexperienced world has been seriously damaged. The degree of this loss of “aura,” or informational novelty (Benjamin, 1936), has progressed from the establishment of phototechnology to an informational presentation of media. Benjamin (1936) discussed the loss of novelty of the context of aura presentation with duplication technology. He described this loss with the technology of photographs, which coincided with the publication of his book. According to Benjamin (1936), Around 1900, technical reproduction had reached a standard that not only permitted it to reproduce all transmitted works of art and thus to cause the most profound change in their impact upon the public; it also had captured a place of its own among the artistic processes. (p. 12) He also discussed the loss of the novelty element (Benjamin, 1936, p. 15) as follows: Even the most perfect reproduction of a work of art is lacking in one element: Its presence in time and space, its unique existence at the place where it happens to be […] One might subsume the eliminated element in the term “aura” and go on to say: That which withers in the age of mechanical reproduction is the aura of the work of art. Adorno (1971) presented the risk of duplication technology to enhance the attitude of loss motivation without resistance. During the same period, there was an argument regarding the monotonous labor of the “Ford model.” This provoked a sense of helplessness and violation of human mental health. Actually, mental health of humans was not infringed upon with the loss of novelty of context presentation. Humans did not perceive the loss of novelty as boring. Quality mental health for the informational environment of human life was not a serious issue . People enjoyed music, videos, and movies. Context memory with a threshold of 300–500 increases the sense of novelty for information. This remains true when information and novelty are damaged or provoke a sense of boredom.

CONCLUSION The chapter discussed the threshold of context memory on numbers 300–500 from the perspective of the assumption experiment and concept for the threshold with semantic layer and working memory. The property of the AEF was able to express the difference in memory states over (2000) or under (500) the threshold of context memory. This chapter discussed the concept of threshold with a semantic layer and working memory. The perspective of the semantic layer presented from the investigation of the hierarchical structure of cognitive semantics, folk biology, and semantic layers of basic categories. The perspective of working memory presented the study on “magical number seven” as the function of short-term memory. The study on long-term working memory presented the bridge between the three semantic layers and magical number seven. The three semantic layer limitations expressed the limitation

115

 300–500 Threshold of Context Memory and Assumption Experiment

of the chunk layers of long-term working memory and the limitation of the number of memory, as the number of context memory thresholds.

REFERENCES Abe, A. (2011). What is “Kotoba-kougaku” (language engineering) Intelligence and Information, (Journal of the Japan Society for Fuzzy Theory and Intelligent Informatics), 23(5), 666-675. Adam, J. M. (1984). Le récit. Presses Universitaires de Paris. Adorno, T. W. (1971). Fukyo waon: Kanri syakai ni okeru ongaku (N. Sanko & T. Takatsuji, Trans.). Ongakunotomosha. (Original work published 1938) Amano, S. (1996). Lexical facilitation of phoneme perception in time-modified speech. The Journal of the Acoustical Society of Japan (E), 17(1), 9–14. doi:10.1250/ast.17.9 Amano, S., & Kondo, T. (1999). NTT database series, vol. 1, word familiarity. Tokyo, Japan: Sanseido. Amano, S., Kondo, T., & Kakehi, K. (1995). Modality dependency of familiarity ratings of Japanese words. Perception & Psychophysics, 57(5), 598–603. doi:10.3758/BF03213265 PMID:7644320 Aristotelēs. (1949). Poetics. Iwanami Shoten. Barsalou, L. W. (1985). Ideals, central tendency, and frequency of instantiation as determinants of graded structure in categories. Journal of Experimental Psychology. Learning, Memory, and Cognition, 11(4), 629–654. doi:10.1037/0278-7393.11.1-4.629 PMID:2932520 Benjamin, W. (1936). Das Kunstwerk im Zeitalter seiner technischen Reproduzierbarkeit [The work of art in the age of mechanical reproduction]. Shoubunsha. Berlin, B. (1972). Speculations on the growth of ethno botanical nomenclature. Language in Society, 1(1), 51–86. doi:10.1017/S0047404500006540 Berlin, B. (1992). Ethno biological classification: Principle of categorization of plants and animals in traditional society. Princeton University Press. doi:10.1515/9781400862597 Berlin, B., Breedlove, D. E., & Raven, P. H. (1973). General principles of classification and nomenclature in folk biology. American Anthropologist, 75(1), 214–242. doi:10.1525/aa.1973.75.1.02a00140 Bernard, H. R. (2006, September 28). Honoring Peter Killworth’s contribution to social network theory. Paper presented at the University of Southampton. Chase, W. G., & Ericsson, K. A. (1981). Skilled memory. In J. R. Anderson (Ed.), Cognitive skills and their acquisition (pp. 141–189). Erlbaum. Connine, C. M., Mullennix, J., Shernoff, E., & Yele, J. (1990). Word familiarity and frequency in visual and auditory word recognition. Journal of Experimental Psychology, 16, 1084–1096. PMID:2148581 Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of mental storage capacity. Behavioral and Brain Sciences, 24(1), 87–185. doi:10.1017/S0140525X01003922 PMID:11515286

116

 300–500 Threshold of Context Memory and Assumption Experiment

Engstrom, D. A., Kelso, J. A. S., & Holroyd, T. (1996). Reaction-anticipation transitions in human perception-action patterns. Human Movement Science, 15(6), 809–832. doi:10.1016/S0167-9457(96)00031-0 Ericsson, K. A., & Kintsch, W. (1995). Long-term working memory. Psychological Review, 102(2), 211–245. doi:10.1037/0033-295X.102.2.211 PMID:7740089 Eulitz, C., Diesch, E., Pantev, C., Hampson, S., & Elbert, T. (1995). Magnetic and electric brain activity evoked by the processing of tone and vowel stimuli. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 15(4), 2748–2755. doi:10.1523/JNEUROSCI.15-04-02748.1995 PMID:7722626 Fletcher, C. R. (1981). Short-term memory processes in text comprehension. Journal of Verbal Learning and Verbal Behavior, 20(5), 264–274. doi:10.1016/S0022-5371(81)90183-3 Freedman, D. J., Riesenhuber, M., Poggio, T., & Miller, E. K. (2002). Visual categorization and the primate prefrontal cortex: Neurophysiology and behavior. Journal of Neurophysiology, 88(2), 929–941. doi:10.1152/jn.2002.88.2.929 PMID:12163542 Gladwell, M. (2000). The tipping point: How little things can make a big difference. Asuka Shin Sha. Harada, N. (2019). Question of 300-500 context threshold and assumption experiment. Proceedings of the Society for Language Engineering “Kotoba-Kougaku”, 62, 1-24. Harada, N., Iwaki, S., Nakagawa, S., Amano, S., Yamaguchi, M., & Tonoike, M. (2005a). Effect of word familiarity on auditory evoked magnetic fields. In K. Zilles, J.-B. Poline, & C. Grady (Eds.), Neuro Image: Proceeding of 11th Annual Meeting of the Organization for Human Brain Mapping (OHBM) (26, Supplement (1), 1115). Amsterdam: Elsevier. Harada, N., Iwaki, S., Nakagawa, S., Amano, S., Yamaguchi, M., & Tonoike, M. (2005b). Effect of word familiarity on auditory evoked magnetic fields. In Proceeding of the 20th Annual Meeting of Japan Biomagnetism and Bioelectromagnetics Society (pp.36-37). Academic Press. Harada, N., Iwaki, S., & Tonoike, M. (2006a). Effect of the incongruity controlled by semantic distance on visually evoked magnetic fields. In K. Zilles, J.-B. Poline & C. Grady (Eds.), Neuro Image: Proceeding of 12th Annual Meeting of the Organization for Human Brain Mapping (OHBM) (31, 440). Amsterdam: Elsevier. Harada, N., Iwaki, S., & Tonoike, M. (2006b). Effect of the incongruity controlled by semantic distance on visually evoked magnetic fields. In Proceeding of the 21th Annual Meeting of Japan Biomagnetism and Bioelectromagnetics Society, 19(1) (pp.56-57). Academic Press. Harada, N., Iwaki, S., & Tonoike, M. (2006c). Effect of word familiarity for expression of incongruity. Proceedings of the Society for Language Engineering “Kotoba-Kougaku”, 24, 36-54. Harada, N., Masuda, T., Endo, H., Nakamura, Y., & Takeda, T. (2001). Evoked magnetic fields caused by auditory stimuli with 1/fn fluctuation in inter-stimulus intervals. The Journal of Japan Biomagnetism and Bioelectromagnetics Society, 13(2), 1–12.

117

 300–500 Threshold of Context Memory and Assumption Experiment

Harada, N., Masuda, T., Endo, H., Yukihiro, N., Takeda, T., & Tonoike, M. (2005). Effect of 1/f fluctuation in inter-stimulus intervals on auditory evoked magnetic fields. Neuroscience Letters, 13(379), 223–228. doi:10.1016/j.neulet.2005.01.012 PMID:15843068 Hari, R. (1991). Activation of the human auditory cortex by speech sounds. Acta Oto-Laryngologica, 111(491), 132–137. doi:10.3109/00016489109136790 PMID:1814144 Hatanaka, T. (2000). Recommendation of “Shippaigaku” (study of failure case analysis). Koudansha. Katz, J., & Fodor, J. (1963). Structure of semantic theory. Language, 39(2), 170–210. doi:10.2307/411200 Katz, J., & Postal, P. (1964). An integrated theory of linguistic description. MIT Press. Mäntysalo, S., & Näätänen, R. (1987). The duration of a neuronal trace of an auditory stimulus as indicated by event-related potentials. Biological Psychology, 24(3), 183–195. doi:10.1016/0301-0511(87)90001-9 PMID:3663794 Martin, A. (2007). The representation of object concepts in the brain. Annual Review of Psychology, 58(1), 25–45. doi:10.1146/annurev.psych.57.102904.190143 PMID:16968210 McCarty, C., Killworth, P. D., Bernard, H. R., Johnsen, E., & Shelley, G. (2000). Comparing two methods for estimating network size. Human Organization, 60(1), 28–39. doi:10.17730/humo.60.1.efx5t9gjtgmga73y McNab, F., Varrone, A., Farde, L., Jucaite, A., Bystritsky, P., Forssberg, H., & Klingberg, T. (2009). Changes in cortical dopamine D1 receptor binding associated with cognitive training. Science, 323(5915), 800–802. doi:10.1126cience.1166102 PMID:19197069 Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63(2), 81–97. doi:10.1037/h0043158 PMID:13310704 Mizuno, T. (2017). Legal design, accelerating creativity and innovation through law. Film Art Inc. Näätänen, R. (1986). N2 wave of the evoked potential: Scalp reflection of neuronal mismach of orienting theory? In J. Strelau, F. H. Earley, & A. Gale (Eds.), The biological bases of personality and behavior. Psychophysiology, performance, and applications. 2 (pp. 59–75). Hemisphere. Näätänen, R., & Gaillard, A. W. K. (1983). The orienting refrex and the N2 deflection of the event-related potential (ERP). In A. W. K. Gaillard & W. Ritter (Eds.), Tutorials in ERP: Endogenous components (pp. 119–141). North-Holland. doi:10.1016/S0166-4115(08)62036-1 Näätänen, R., Sams, M., & Alho, K. (1986). The mismatch negativity: The ERP sign of cerebral mismatch process. Electroencephalography and Clinical Neurophysiology. Supplement, 38, 172–178. Näätänen, R., Simpson, M., & Loveless, N. E. (1982). Stimulus deviance and evoked potentials. Biological Psychology, 14(1-2), 53–98. doi:10.1016/0301-0511(82)90017-5 PMID:7104425 O’Hanlon, W. H., & Hexum, A. L. (1990). An uncommon casebook: The complete clinical work of Milton H. Erickson, M.D. Norton. Ogata, T. (2018). Monogatari-seisei no monogatari-ronaruiha posuto-naratoroji. Cognitive Studies, 25(2), 200–217.

118

 300–500 Threshold of Context Memory and Assumption Experiment

Purves, D. (2008). Principles of cognitive neuroscience. Sinauer Associates Inc. Rosch, E. (1973). Natural categories. Cognitive Psychology, 4(3), 328–350. doi:10.1016/00100285(73)90017-0 Rosch, E. (1977). Human categorization. In N. Warren (Ed.), Advances in cross-cultural psychology (pp. 1–72). Academic Press. Rosch, E., Carolyn, M., Wayne, G., David, J., & Penny, B. B. (1976). Basic objects in natural categories. Cognitive Psychology, 8(3), 382–439. doi:10.1016/0010-0285(76)90013-X Sams, M., Hari, R., Rif, J., & Knuutila, J. (1993). The human auditory sensory memory trace persists about 10 sec: Neuromagnetic evidence. Journal of Cognitive Neuroscience, 5(3), 363–370. doi:10.1162/ jocn.1993.5.3.363 PMID:23972223 Squire, L. R., & Zola-Morgan, S. (1985). The neuropsychology of memory: New links between humans and experimental animals. Annals of the New York Academy of Sciences, 444(1), 137–149. doi:10.1111/j.1749-6632.1985.tb37585.x PMID:3925849 Takakura, K., & Okubo, A. (Eds.). (1994). MEG Noujizu no kiso to Rinshou. Asakura Publishing Co. Ltd. Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.), Organization of memory (pp. 381–403). Academic Press.

KEY TERMS AND DEFINITIONS Auditory Evoked Magnetic Field: The auditory evoked magnetic field was the component of the average magnetic response induced by the electrical current of the primary auditory nerve system. The largest and most widely studies component was the N100m component, reflecting the response activity of the primary auditory cortex on the right and left temporal lobes for tone presentation. Class B Engine: The “class B engine” was an “almost infinite” Japanese pun-generating system, using a lexical engineering system of artificial intelligence. This system was developed to present the concept of the language engineering study group, Kotoba Kougaku (linking art and engineering), and research results. Cognitive Semantics: Cognitive semantics was the research field of the semantics investigated with the perspective of human perception and recognition. This research field was developed by George Lakoff, Leonard Talmy, Ronald Langacker, and Gilles Fauconnier. Context Memory: The context memory, which reflected the memory status of the narrative and pun, had a functional base of the long-term memory declarative memory. Long-term memory is divided into two types: semantic memory (constructed by repeated experience, situation of acquisition disappeared) and episodic memory (constructed by single trial experience, situation of acquisition was maintained with the memory of episode). Context memory, which is similar to the memory of the narrative plot and pun, is constructed by a single trial experience, while the memory of the situation of acquisition disappears. The context memory is a midway between semantic memory and episodic memory. Human Brain Mapping: Human brain mapping is a noninvasive technology used to investigate human brain function for visual, auditory, and somatosensory stimulation. Representative technologies

119

 300–500 Threshold of Context Memory and Assumption Experiment

were electroencephalography, magnetoencephalography, functional magnetic resonance imaging, and near-infrared spectroscopy (NIRS). Magical Number Seven: The magical number seven was the functional limitation of memorizing numbers and characters. The function reflected the activity of working memory, which is a system of short-term memory, and the functional activity of the frontal lobe. Short-Term Memory: The human memory function was classified into short-term memory and long-term memory on the property of time course of the preservation of memory. This definition reflected the property of the time of status of memory preservation. Working memory is one component of the short-term memory.

120

300–500 Threshold of Context Memory and Assumption Experiment

APPENDIX 1

Human Brain Mapping of Context The detection of the effect of narrative plot with provocation of emotional response was demanded for 90–100 times in the stimulus presentation of the plot. The response to the plot presentation with emotional reaction differed from that of the first time presentation to the second time or after trials. The response to the plot presentation significantly decreased from first to second or after. It was difficult to detect the response to the plot with emotional response with the repetition of the same plot presentation. The 90–100 numbers of different plots with emotional provocation were necessary for the detection of the effect of the narrative plot on the average waveform of magnetoencephalogram with auditory and visual stimulus presentation. The response for the pun had a similar property to the response to the presentation of the narrative plot. The response to pun presentation differed and decreased from first to second or after the presentation. The response decrement for the narrative plot and pun was a problem even in functional magnetic resonance imaging with repetitive stimulus presentation, which was in accordance with the experimental design. The response of the sensory nervous system of the N100m component was gradually changed with the repetition of the stimulus presentation. The peak amplitude of N100m component was decreased exponentially with repetition of stimulus presentation, which showed the habituation of AEF (Harada, Masuda, Endo, Nakamura, & Takeda, 2001). In contrast, the response of the stimulus of the narrative plot and pun significantly decreased (as digitally) from first to second or after presentation.

APPENDIX 2

Anticipation and Decrement of Amplitude and Latency of Auditory Evoked Magnetic Field The variation in decrement of amplitude and latency of average waveform reflected the function of the elicitation of the mismatch field (MMF) component with the anticipation for stimulus. The MMF component disappeared in the long ISI conditions. Instead, a reduced N100m component and increased MMF were observed with a shortened ISI, such as 1 or 2 s (Mäntysalo & Näätänen, 1987; Näätänen & Gaillard, 1983; Näätänen, Sams, & Alho, 1986; Näätänen, Simpson, & Loveless, 1982; Sams, Hari, Rif, & Knuutila, 1993). Mäntysalo and Näätänen (1897) reported that the inflexion point in the ISI range for change from elicitation to disappearance of the MMF was 2–4 s, while Sams et al. (1993) reported that it was 3–6 s. Variations in the length and randomness of the ISI produced opposite changes in the MMF and N100m component.

121

300–500 Threshold of Context Memory and Assumption Experiment

The range of ISI produced a Gaussian distribution that resulted in the range of times for the elicitation of the MMF overlapping with a short ISI and a sharp and narrow peak overlapping with the disappearance of the MMF with a long ISI. The disappearance of the MMF might have occurred because of the loss of anticipation of the stimulus with a long ISI. The author suggests that the opposing changes in the MMF and N100m component are related to the temporal correlations of the input stimulus sequence and that this process is responsible for both the establishment and breakdown of anticipation behavior. The timing of the tone with a long ISI could not be anticipated by the subject; thus, a memory trace was not constructed for long ISIs, and the absence of a memory trace resulted in the disappearance of the MMF. Subjects could also not anticipate the timing of the stimulus with random changes in the 1/f0 ISI condition, which inhibited the formation of the memory trace, and the MMF was not elicited. Instead, the 1/f∞ ISI condition resulted in anticipation and memory trace formation, and the MMF was elicited. The small change in ISI regularity from 1/f0 to 1/f1 enabled anticipation of the timing of the stimulus arrival. A low-level stochastic correlation in the stimulus train in 1/f1 changed the informational processing of the stimulus in the primary auditory area, which was reflected in the elicitation of the MMF. The temporal sensitivity of the auditory cortex has been determined from posttetanic potentiation in a rat’s auditory cortex slice; posttetanic potentiation was maximally inhibited with an ISI time ranging from 1 to 2 s. The author considered that functional verification of the timing of stimulus arrival and internal monitoring clock for the stimulus timing may be performed in the primary auditory area.

APPENDIX 3

Function of the Frontal Lobe on Working Memory and Function of Taxonomical Expression The categorization function of the human brain was discussed with a relationship between visual temporal lobe and frontal lobe in a functional magnetic resonance imaging study (Martin, 2007). The visual image was separated in the temporal lobe, and categorical judgment was executed in the frontal lobe. Short-term memory (e.g., working memory) was conducted in the frontal lobe. The frontal lobe was responsible for the function “magical number 7 ± 2” as the short-term memory. The function “magical number 7 ± 2” was expressed with the function of the short-term memory, which was related to the “working memory” function located in the frontal lobe investigated in the functional brain mapping study (McNab et al., 2009). The expression of the threshold function in the range of numbers 300–500 in context memory was described with the perspectives of the combination of theories of magical number 7 ± 2 and semantic layer structure. Ericsson and Kintsch (1995) reported the taxonomic structure of chunks of the magical number seven, discussed long-term working memory, and discussed the over seven chunks used in human life.

122

300–500 Threshold of Context Memory and Assumption Experiment

The results of the study on the monkey’s nerve cell and visual evoked magnetic field (VEF) of the human frontal and temporal lobe provided the semantic taxonomic classification. The semantic discrimination of brain function was reported with the investigation of nerve cells of the frontal lobe and visual cortex of the monkey (Freedman, Riesenhuber, Poggio, & Miller, 2002). The semantic judgment of “dog” or “cat” was requested with the visual images constructed with the image mixing technology of “morphing” with the variation of mixing ratio from 0% to 100% on “dog” and “cat” images with semantic judgment. In the judgment of “dog,” the activity of the nerve cell was gradually increased with the increment of mixing ratio of the “dog” image in the visual cortex. The activity of the nerve cell of the frontal lobe was digitally increased to > 50% mixing ratio of the “dog” image. The study described the results of the experiment, the increment of the activity of the nerve cell of the visual cortex reflected the form recognition, and the increment of the activity of the nerve cell of the frontal lobe reflected the semantic judgement. The semantic provocation effect of animal body parts was investigated with the experiment of VEF (Harada, Iwaki, & Tonoike, 2006a, b). The visual image of the experiment was constructed with the “line” drawing of animals with changing the “body” and “head.” The combination of parts of “body” and “head” was controlled with the semantic distance on the structure of the word constitution of Five Thesaurus of Japanese. The first condition was the combination of the original and normal “body” and “head” conditions. The second condition was the combination with changing in the group of taxonomy class relating to generic category on folk biology study. The third condition was the combination with changing intergroups of the taxonomy class. The semantic distance of the combination of the parts of body and head was increased in the condition from the first and second to third with the taxonomic joint number intergroups on the semantic layer structure of the thesaurus. The VEF was recorded for the projection of the image of the three types of the “line” drawing animal images with the conditions of the combination of “body” and “head” on the semantic distance of the first (0 joint) and second (2 joints) to third (4 joints) condition. The average waveforms of the root mean square (RMS) value were calculated with the left frontal, right frontal, left occipital, and right occipital area channels. The average waveform of the VEF of the RMS in the left-occipital area was investigated with the peak amplitude of 170 ms and 240 ms. The peak amplitude of 170 ms of the third condition was significantly decreased with the amplitude of the first and second conditions. The peak amplitudes of 240 ms of the second and the third conditions were significantly decreased with the amplitude of the first condition. The peak amplitude of the average waveform of the left frontal area on 380 ms was significantly increased from the first, second, and third conditions. The peak amplitude of 170 ms differed between the first, second, and third conditions and reflected the extraction process of the large difference of the visual image. The peak amplitude of 240 ms differed between the first, second, and third conditions and reflected the extraction process of the small differences in the visual image. The steps of informational processing of the visual image were indicated for two steps of processing of the large and small difference extractions. The two steps of extraction of difference of images allowed the classification of three levels of the taxonomic difference with the variation in semantic distance with the number of joints. The increment of semantic distance was used to enhance the activity of the left frontal area observed with a peak amplitude of 380 ms. The variation of the peak amplitude of 170 ms and 240 ms of the visual cortex of the left occipital area showed the semantic classification of three layers of taxonomy. The activity of the left frontal lobe

123

300–500 Threshold of Context Memory and Assumption Experiment

of 380 ms was enhanced with discrimination of the semantic definition of the visual cortex. This result showed that the visual cortex allowed the classification of the visual image with three layers of taxonomy. The combination of the function of the short-term memory of the magical number 7 ± 2 on the frontal lobe, and the semantic classification function of three layers on the visual cortex enabled the expression of the function of the threshold of context memory of the range of numbers 300–500.

APPENDIX 4

Dunbar’s Number Dunbar reported the relationship (correlation) between the volume of the cortex (brain) and group size of the number of members in monkeys. The estimated value in humans was 150. This amount implied the hypothesis that 150 was the threshold of the cognitive function for the group size, which allowed to durably maintain the social relationship. A higher amount of group members would request a restriction rule to maintain stable social relationships. Bernard (2006) and McCarty, Killworth, Bernard, Johnsen, and Shelley (2000) reported the number of 290 (almost twice of 150) for the stable social group, which they estimated by field research on several social groups in America. Gladwell (2000) analyzed the case study of the company of WL Goa, which had < 150 users of the office building and company parking lot. The limitation to < 150 users contributed to identifying and inhibiting social issues. The number 290 was extremely close to the number 300 of the threshold of the context. The deviation of the magical number seven was reported plus and minus two. The number 7×7×7=343 could deviate to 6×6×6=216 and 5×5×5=125. The number 216 and 125 was closer to Dunbar’s number between 150 and 290. Cowan (2001) reported the magical number was estimated as four with the strictly controlled experiment. In the perspective of the case study of the story maker, Japanese novelist of Kenzo Kitakata made some comment of the relationship of memory between the person and the story in the FUJI TV program of in “Typewriters” at 2020/7/18 on 10.28-11.50. He made a comment that, “The first process story making was the provocation of person in the image field, the second process was the starting up the flue of story”. This comment showed the linkage of memory function between person and story. The common cognitive base between personal identification and context recognition could be used. The system of personal identification can be adopted for the system of context recognition.

124

Section 2

Narrative Generation Systems and Narrative Content Generation

126

Chapter 6

Bridging the Gap Between Narrative Generation Systems and Narrative Contents With Kabuki-Oriented Narratology and Watakushi Monogatari Takashi Ogata https://orcid.org/0000-0001-9181-8876 Iwate Prefectural University, Japan

ABSTRACT The author’s narrative generation study is based on two types of systems: the integrated narrative generation system (INGS) as a single narrative generation and reception mechanism and the Geinō information system (GIS) as a multiple narrative production and consumption mechanism. The first theme of the chapter is to introduce an idea that deals with narrative phenomena as the integration of both systems. This theme is tied to the topic of narrative content creation by presenting kabuki narrative generation or kabuki-oriented narratology and Watakushi Monogatari as a collection of narrative content to be created by the author. Hence, the second objective is to describe kabuki-oriented narrative generation and the third is to explain the ideas, thoughts, and design underlying Watakushi Monogatari in the context of internal and external narrative generation to create and distribute narrative content. Through these three themes, this chapter bridges the gap between narrative generation systems and narrative content with kabuki-oriented narratology and Watakushi Monogatari.

INTRODUCTION The author published a series of books (Ogata, 2020a, 2020e) that systematically Shōwan integrated approach to studying narrative generation. These books assert that research on content generation and narrative generation forms an important, challenging field that has a fundamentally interdisciplinary DOI: 10.4018/978-1-7998-4864-6.ch006

Copyright © 2021, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

characteristic. This characteristic is based on artificial intelligence (AI) (including cognitive science) and narratology (including literary theories), grounded in an integrated, interdisciplinary narrative generation system of study through a model of multiple narrative structures, as well as other philosophical considerations. In the first part of this chapter, the author describes the meanings of the concept of an integrated approach to narrative generation. “Integration,” in this case, connotes a broad range of senses beyond the integration conveyed in the Integrated Narrative Generation System (INGS) that the author has been designing and developing. The term “integration” includes the following: 1. Integration Among Modules in INGS: This is the main meaning of “integration” in INGS. At the most macro-level of the module division, INGS has three modules: story generation, narrative discourse generation, and narrative representation. These modules have a mutual relationship based on the simple process of order. At the same time, they are associated with each other through various connections. 2. Integration Between INGS and the Geinō Information System (GIS): Although the author’s narrative generation study can be seized as a grand whole that includes personal and individual narrative generation-reception process and social narrative production-consumption process, INGS functions under organic relationships within another system framework: GIS. In the author’s research system, GIS corresponds to a system that deals with the abovementioned social narrative production-consumption aspect. Furthermore, when the author elaborates on the stage of Watakushi Monogatari, which will be described in this chapter, the mutual cooperation of INGS and GIS will be necessary to proceed in a comprehensive manner. 3. Integration as Multiple Narrative Structures: This is a manifold kind of integration in the sense that narrative generation phenomena are composed of three macro levels: the level of the brain (super personal level A), the symbolic/linguistic level, and the social/collective/collaborative/institutional/communal level (super personal level B). In terms of Takaaki Yoshimoto’s (1924–2012) philosophical system or his main books that Ogata (2020e) frequently refers to, these three levels correspond to the levels of mental phenomena (Yoshimoto, 1971, 2008), linguistic phenomena (Yoshimoto 1965), and social phenomena or communal illusion (Yoshimoto, 1968). 4. Integration Among Research, System Development, Social Development, and Creation: The author’s narrative generation study does not necessarily fall under academia alone; it covers the level of social development or distribution, and the private/personal levels of the production and creation of actual narrative works, via the level of system design and development (although basic, prototypical designs and development are academic, the level will be tied to a more pragmatic stage) from the angle of research. In a wide sense, the “research” is integrated because it is carried out in a broad, successive area. 5. Integration Between Information and Literature (or Expanded Literary Theory): The author’s expanded literary theory (Ogata, 2002a, 2002b, 2003a, 2003b, 2014; Ogata, Imabuchi, & Akimoto, 2014) was the earliest approach to the theme of “integration” in the author’s narrative generation study. It aims to build a new narrative study by synthesizing informational science, engineering (including AI and cognitive science), and literary theories (such as narratology and narrative theory). 6. Integration Among Diverse Academic Areas: Since developing the expanded literary theory described earlier, the author’s integrated/synthetic narrative generation study has been extended to an integrated approach that covers a wide range of fields beyond the fusion of information technol-

127

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

ogy (centering on AI and cognitive science) and narratology (including literary theories). Chapter 3 (Ogata, 2020c) in Ogata (2020a) refers to the primary relevant areas. 7. Integration Among Diverse Narrative Genres (or “Narrative Genre System”): This study does not limit narratives to the field of literature, and understands the genre of narratives to exist in broader areas. Chapter 2 (Ogata, 2020b) in Ogata (2020a) outlines the narrative genre system, which is composed of five categories, and provides Japanese narrative genres for the central part, presenting concrete examples of the narrative genre system. This is partially related to the fact that the extreme goal of this integrated approach to narrative generation is to produce personal and private narrative content called Watakushi Monogatari. In particular, “I” (watakushi) is most closely linked to the production of Watakushi Monogatari and many parts of knowledge, forming the narrative content necessary for a narrative, and directly relating to the nation or history of Japan, which deeply connects with “I” (watakushi). Hence, the narrative genre system is constructed based on diverse Japanese narrative genres. Although this chapter more or less pertains to all of the above topics, it has strong ties with both 2 (Integration between INGS and the Geinō Information System (GIS)) and 4 (Integration among Research, System Development, Social Development, and Creation). Regarding the latter, this chapter aims to describe the fundamental strategy that bridges the research and development (R&D) of INGS and GIS for the production/creation and social development/distribution of R&D through GIS. The author intends to achieve this goal by considering the following three themes. As demonstrated by the title, this chapter attempts to bridge the gap between narrative generation systems and narrative content with kabuki-oriented narratology and Watakushi Monogatari in the author’s research system. • • •

The Study of Major Narrative Generation Systems, INGS and GIS: In the following section, the study is called “(A) INGS and GIS.” The Study of Survey, Analysis, and the Systematization of Kabuki: Similarly, this study is called “(B) Kabuki.” The Study of External Narrative Generation and Internal Narrative Generation for Watakushi Monogatari: This study is called “(C) Internal and External.”

Ogata (2020e) includes these three themes, and Ogata (2020a), whose work corresponds to the first section in the sense of topics, examines these themes at a fundamental level. Through these two books, the author presents many (or almost all) concepts concerning the author’s integrated approach to narrative generation and basic considerations of them. However, these books have not reached the scale of organic integration among systems (INGS and GIS), kabuki studies, or Watakushi Monogatari in the context of internal and external narrative generation. Stated in the previous part, the central strategy of this chapter is to aggregate the three kinds of themes to form a consistent research story. In the following section of this INTRODUCTION, concerning these three themes, the author summarizes the respective descriptions of future directions based on Ogata (2020a, 2020e) to establish the key topics and problems in this chapter. First, concerning (A) INGS and GIS, Chapter 1 (Ogata, 2020f) in Ogata (2020e) remarks on INGS architecture, its philosophical ideas, and mechanisms, such as generation, as well as content mechanisms and application systems. INGS is currently being carried out at the experimental level. INGS develops the aspects of “expanded literary theory” by incorporating the literary theories of Propp (1968), Genette 128

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

(1972), and Jauss (1970) into the system and covers the implementation of a story, narrative discourse, and narrative representation. Through the developmental process of INGS, several components and mechanisms were also applied, including conceptual dictionaries, a state-event transformation knowledge base, and story generation mechanisms using these dictionaries and knowledge base The implementation enabled effective constructive research on the complex mechanisms in which an entire performance becomes possible through the interactions of diverse narrative techniques and expertise. INGS architecture has been put into effect as a modularized, flexible program for easy extension, transformation, and revision; this characteristic is tied to a kind of development that enables the gradual incorporation of various ideas and modules into the system. The author aims to address the issue of development based on a circulative, repetitive process by organically consolidating partial revisions into the entire architecture using its plastic feature. In addition to continuing the progressive development of INGS, in FUTURE RESEARCH DIRECTIONS in Chapter 1 (Ogata, 2020f) in Ogata (2020e), the author strives to bridge the system to the social and personal “practices” of narrative generation. The following topics are covered in the ongoing system development of INGS: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

Basic Generation Methods Content Knowledge for Generation Expanding Content Knowledge Knowledge Acquisition and Learning Mechanisms Natural Language Generation GIS (the Geinō Information System) or the Social Narrative Production and Distribution System Expanded Literary Theory Japanese Literary Theories and Narratology Philosophical Concepts Related to Expanded Theory Multiple Narrative Structures Social Narrative Distribution (in Plans for Social and Individual Development) “My [the author’s]” Narrative Creation (in Plans for Social and Individual Development) Evaluating Narrative Generation Systems

In these future works, 6, 8, 11, and 12 are closely linked to this paper. In 6, GIS is not only an application system of INGS, but is also an extension of INGS that allows it to be tied to a social/business framework for the production and distribution of generated narratives. In particular, GIS is designed to contain one or more INGS. In 8, an especially important direction of the expanded literary theory is to analyze traditional or modern/contemporary narratological research and literary theories in Japan, and to connect them with INGS (and GIS). In 11, this topic implies that narratives produced by a narrative generation system (which include descriptions created in collaboration with humans) and the narrative generation system itself are associated with society, namely the outside of the system. There are diverse possibilities, such as business. Forms like “virtual publishing” and “virtual entertainment production” by GIS may make clear images for introducing the aspect of business into narrative generation systems. In 12, this concept refers to narratives pursuant to the author’s private themes and materials. In particular, INGS introduces private knowledge content into the story content knowledge base and other narrative content knowledge bases.

129

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Next, the author describes (B) Kabuki. The following is the FUTURE RESEARCH DIRECTIONS in Chapter 4 (Ogata, 2020h) in Ogata (2020e). The main short-term plans concerning kabuki’s survey and analysis are presented: 1. The Modeling and Design of GIS with INGS, Based on the Introduction of Elements and Knowledge of Kabuki 2. Deepening the Survey and Analysis of Kabuki Itself as an Information System 3. An Experimental Implementation and the Verification of GIS with INGS 4. An Applicable System Design and Implementation, Including Social Distribution Experiments By the term “GIS with INGS,” the author means the following. In the sense that GIS is a broad framework in which INGS (correctly, one or more INGS[s]) is included, the author has used the term GIS with INGS or “GIS including INGS.” This chapter employs the terms, especially GIS with INGS. In the above list, 1, 2, and 3 are particularly relevant to this chapter. In 1, the first objective is to design a more detailed GIS with the INGS model, extended through a survey of kabuki. This also draws on the next theme, specifically, deepening the investigation of kabuki itself. Regarding this direction, a literature survey by the author on kabuki describes, in depth, Kabuki as Documents in BACKGROUND in Chapter 2 (Ogata, 2020g) in Ogata (2020e). Furthermore, with a fresh perspective, the author delves into the concepts of “person,” “story,” tsukushi and naimaze in kabuki in the section INTRODUCING KABUKI KNOWLEDGE INTO GIS WITH INGS: FROM ANALYSIS TO GENERATION. A future study would involve producing deeper, broader examinations of numerous kabuki elements. Extending the study’s range to other elements in kabuki is also regarded as a vital path for future research. In 2, the kabuki analysis, along with the system design and implementation, forms a mutual process in that the exploration of kabuki shifts toward the systematization of GIS with INGS. In this vein, the system design and implementation require more detailed, precise, and systematic scrutiny of kabuki. Depending on the analysis, Ogata (2020e) describes experimental programs using the concepts of “person,” “story,” naimaze and tsukushi. One future possibility would be to conduct an experiment to incorporate the proposed program on persons and other elements into INGS and GIS. Although the topic of 3 accompanies the expansion and precision of INGS, it is possible to use the current, tentative version in the GIS experimental system, because it could act as an entire system. Another trajectory for future research could be to develop a meta-level mechanism to control the mutual relationship between the two mechanisms, in order to produce a sequential story of narrative work from the narrative content. In addition, multiple narrative structures could be seen in each constituent element in the narrative generation-reception and production-consumption mechanisms. INGS may be able to tackle this issue by enhancing knowledge bases for elements, including persons and other features. Last, concerning (C) Internal and External, the following describes FUTURE RESEARCH DIRECTIONS in Chapter 4 (Ogata, 2020h) in Ogata (2020e): 1. GIS 2. Developing Watakushi Monogataris 3. Philosophical Considerations of Institutional Narratives in the Broadest Sense In the above list, 1 and 2 are closely related to this chapter. 1 covers designing and developing a system with INGS. An attempt to grasp its design and development is considered, incorporating, with more 130

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

awareness, the experimental system based on the analysis of kabuki as a mechanism of GIS with INGS. Further, the author strives to design the virtual construction of a troupe or a geinō production company (entertainment agency)—an actual form of the GIS—and to prepare geinō characters in various components in GIS. Concerning 2, the author plans to contemplate and concretize a myriad of information and generation methods in relation to the conceptual dictionaries in INGS and GIS. Stated above, although Ogata (2020a, 2020e) describes the direction of future research for the three topics, this chapter covers challenges whereby the previous narratology cannot be extended to the pertinent approaches to solve them. In other words, the author tackles the gap between previous narratology and information modeling using AI and cognitive science. As concretely as possible, this chapter reveals the strategy needed to synthesize the above topics to finally actualize them as Watakushi Monogatari. In regards to (A) INGS and GIS, based on the first chapter (Ogata, 2020f) in Ogata (2020e), a narrative generation system called INGS corresponds to a system that creates simple pieces of narrative content individually, while another system, called GIS, produces many pieces of narrative content in a circulative manner. By studying the narrative generation of kabuki, the author presents mechanisms and methods for how INGS and GIS can integrate and collaborate. In particular, this chapter aims to design, and partially develop, a multiple narrative generation mechanism in which INGS and GIS are organically and mutually executed. In (B) Kabuki, for the author, kabuki is an essential research theme that offers a comprehensive, interdisciplinary approach to narrative generation. The narrative multiplicity in kabuki is typical of a multiple narrative structures model, which serves as a critical analysis of the author’s narrative generation strategy. The author believes that the dramaturgy of kabuki provides interesting, crucial examples in which new narrative content is produced by combining and transforming previous narrative content, along with the fragments. Hence, the author scrutinizes the broad spectrum of analyses of kabuki’s elements, seen from the angle of producing a kabuki work on stage. After reviewing narratological studies of kabuki by the author, this section explores several narratological references or books, including Sakusha Shikihō Kezairoku [Authors’ Ceremonial Methods Kezairoku] (1972) (a theoretical essay written by a kabuki writer during the Edo period) and Sekai Kōmoku (1916) (a database of sekais [worlds] used to create kabuki works from the Edo period). Moreover, the author investigates two contemporary kabuki researchers and their books. The basic methods to approach the two references (i.e., books) include analyzing them from the angle of narrative generation, and utilizing the acquired results in the narrative generation system design and implementation. This chapter delves into the possibilities of their new usage and application through partial system design and program development. Specifically, this section on kabuki discusses the problems that previous kabuki narratology or narratological studies of kabuki could not handle, as well as new techniques to solve them. The author refers to the gap between previous narratology in literary theories and information modeling using AI and cognitive science. The author also systematically examines kabuki elements from the viewpoint of producing a kabuki work on stage. Finally, in terms of (C) Internal and External, through narrative and literary histories, the author describes diverse methods to create and distribute narrative content. For instance, some narrative content in a specific genre is produced through collaborative efforts by many people who play different roles and distribute content at a physical site, such as the stage of a theater. In another genre such as novels, narrative content is written using papers and characters and is spread through printed books. Today, computational technology and narrative generation systems are regarded as tools or media that create new methods and possibilities for narrative creation and distribution.

131

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

The author presents two related concepts, namely “internal” and “external” narrative generation, which are associated with the issue of producing and circulating narratives. Internal narrative generation indicates that narrative content is created using a narrative generation system, especially GIS with INGS. Further, it relates back to the author himself, and includes a sub-genre called Watakushi Monogatari, which, in its current form, is a narrative vision to be written using the author’s own materials. On the other hand, external narrative generation corresponds to the social distribution of narrative content, using a mechanism in GIS in particular. This section covers the basic designs of internal and external narrative generation, respectively. In his description of internal narrative generation, the author deals with the topic of narrative content creation supported by GIS with INGS, and narrative content that includes a concrete approach to Watakushi Monogatari. This chapter also presents the design of external narrative generation, which is the mechanism for organically allocating created narrative content. Moreover, both internal and external narrative generation are explained according to an organic relationship. This part discusses the problem whereby previous narratology should be bridged to informational and computational methods through the author’s narrative generation study, regarding narrative creation and distribution. In other words, the author recognizes the gap between previous narratology and information modeling (which uses AI and cognitive science), and utilizes a research direction to connect them. This chapter consists of the following sections. The BACKGROUND deals with Post-Narratology and the Post-Narratology of Narrative Generation, which examines an integrated narrative generation approach based on the multiple narrative structures model and other philosophical considerations by the author. Next, Narrative Generation Mechanisms Based on Post-Narratology focuses on the following: (1) INGS and (2) GIS as the most important elements in the above strategy and (3) previous attempts to combine INGS and GIS. The section on kabuki—The Narrative Generation of Kabuki and Systematic Surveys and Analyses of Kabuki Elements, Seen from the Viewpoint of Producing a Kabuki Work on Stage—summarizes preceding approaches to the narrative generation of kabuki. The former part reviews the author’s narrative generation study of kabuki or kabuki as a narrative generation study. In the latter part, kabuki’s multiple elements (perceived based on the author’s narrative generation system and multiple narrative structures model) form the core outcomes of the analysis of kabuki, or the starting point of the author’s narrative generation study on kabuki. Finally, the subsequent section—Internal and External Narrative Generation—introduces former approaches to internal and external narrative generation. The following three sections are located in the main parts of this chapter. The first is A MULTIPLE NARRATIVE GENERATION MECHANISM THROUGH TWO TYPES OF NARRATIVE GENERATION SYSTEMS: GIS WITH INGS, which includes Basic Elements and the Process of GIS, Incorporating the Watakushi Character, and The Overall Behavior of GIS, Including a Watakushi Character with INGS. One of the three key parts of this section covers the conceptual design of an integrated mechanism for INGS and GIS or GIS with (i.e., including) INGS. The foremost section is NARRATOLOGY OF KABUKI: ORIENTED TOWARD NARRATIVE GENERATION. The author has used the term of “kabuki as language.” By kabuki as language, the author means that kabuki has produced many types of linguistic documents (e.g., essays, reports, critiques) and existed alongside them, although an individual performance on stage is regarded as the most critical component of the art form. This section briefly reviews diverse types (or genres) of kabuki documents through examples. First, the paper centers on contemporary kabuki researchers and critics Yukio Hattori and Tamotsu Watanabe, as well as their books; perceived as theory-oriented writers, their thoughts are useful for the author’s narrative generation approach to kabuki. Next, as older documents, the author examines Sakusha Shikihō 132

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Kezairoku and Sekai Kōmoku in depth from the narrative generation system’s perspective, which are to be applied to narrative generation mechanisms based on GIS with INGS. In the next section, the author explores ways to solve the challenges of the gap between former narratological studies related to kabuki, and narrative generation modeling based on AI and cognitive science; the author refers to several paths for future research. As for the final theme, THE CONCEPT OF WATAKUSHI MONOGATARI AND DESIGNING INTERNAL AND EXTERNAL NARRATIVE GENERATION METHODS TO CREATE AND DISTRIBUTE NARRATIVE CONTENT, the first part describes internal narrative generation—particularly its methods and practices supported by GIS with INGS—to produce narrative content such as Watakushi Monogatari. The next part details the basic procedures of external narrative generation, which organically allocate narrative content created by the proposed internal narrative generation mechanism. The third section outlines the integrated relationship between internal and external narrative generation to make this study count as an actual attempt. Finally, the chapter states how the author bridged the gap that was mentioned in the first section, and refers to several possibilities for future research. The following portion, SUPPLEMENTAL DISCUSSION AND DERECTIONS FOR FUTURE RESEARCH, addresses four topics not dealt with previously, and additional themes related to GIS with INGS, kabuki, and Watakushi Monogatari. In addition, the author examines possibilities for future research, including problems faced when designing and executing GIS with INGS, systematizing Sakusha Shikihō Kezairoku and Sekai Kōmoku, and developing Watakushi Monogatari.

BACKGROUND Based on the above, Ogata (2020a, 2020e) comprehensively and systematically describes his narrative generation study, and summarizes prior investigations. This background reviews the chief themes that emerge in these two books. The following topics are examined: 1. Post-Narratology and the Post-Narratology of Narrative Generation: This entails an integrated narrative generation approach based on the multiple narrative structures model and other philosophical considerations by the author. 2. Narrative Generation Mechanisms Based on Post-Narratology: This includes INGS and GIS as the most critical elements in the above approach, as well as previous attempts to combine INGS and GIS. 3. The Narrative Generation of Kabuki: This is a new, interdisciplinary approach to kabuki from the angle of a constructive view of narrative generation. 4. Systematic Surveys and Analyses of Kabuki Elements, Seen from the Viewpoint of Producing a Kabuki Work on Stage: This part summarizes the systematic analyses of kabuki elements, seen from the perspective of producing a kabuki work on stage. 5. Internal and External Narrative Generation: The final part of the background gives examples that describe different methods to create and distribute narrative content through narrative and literary histories.

133

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Post-Narratology and the Post-Narratology of Narrative Generation The author summarizes an integrated narrative generation approach based on the multiple narrative structures model and other philosophical considerations. First, grounded in Ogata (2020a) (especially, chapters 3 (Ogata, 2020c) and 4 (Ogata, 2020d)), the author reviews the post-narratology and narrative generation or post-narratology of narrative generation and focuses on important issues. (Regarding the concept of post-narratology, the author has been discussing through Ogata (2019b) and other papers.) Narrative generation systems comprise a new narratology, and the evolution of their conceptual/technical construction is the process of building narratology itself. The author discusses narrative generation systems based on essential ideas of narratology. First, in Narrative as Multiple Communication, Fujii (2004) observes the origin of narrative in the discourse on dialogue. He sees the opportunity for dialogue to be transformed into a story via the author’s differentiation and multiplicity. Fujii presents the example of a conversation between a mother and her child, in which the mother takes on the child’s role in speaking to the child using the first person. This real-life instance illustrates the narrator’s multiplicity. In particular, a single narrator is differentiated into multiple, virtual narrators (in both role and function), revealing the narrator’s manifold nature. In a narrative, the narrator and narratee are not substantive, realistic entities, but can rather be called “agents” in the sense that their roles/functions, and the information surrounding them, determine their existence. By referring to interactive discourse as communication, as well as to the functional differentiation of the narrator’s/narratee’s multiplicity/multiplexing, it is possible to seek a single property of a story in multiple communication. Multiple communication differs from the singular communication of daily conversation, in which there is a direct confrontation between sender and receiver, in that the actual narrator and narratee are functionally differentiated into multiple roles. In Narrative as Simulation, the author considers several issues from the standpoint of narrative as a type of simulation, which occurs when a certain subject is expressed with a limited number of characteristics. Through the interactions with subjects, the object’s behavior is artificially simulated, and we can take into account ways of simulating the human in the social state of a narrative (perhaps including nature as well). The temporal development and spatial expansion of the interaction between the characters and the environment (society and nature) is a means of carrying out a thick simulation. This thickness does not refer simply to detail but is linked to the issue of multiple communication, which surely permeates through the opportunity that the narrator offers. In a narrative, in addition to the story—or rather beyond it—the dimension of discourse is vital. The narrator harnesses the diverse functions of those who carry the discourse, or of the mediator between the story and the discourse. The narrative embodies the discourse, which explicitly or implicitly strongly involves the narrator. Accordingly, the narrative, as a simulation, juxtaposes the thickness of not only the content of events but also the mechanisms of language. There are various types of simulations, such as scenario/management simulations and simulations in the natural world. When the narrator’s opportunity comes to the foreground (-should, -must be) in this way, the narrative that simulates the author is original and unique. Moreover, if we think about the matter in terms of multiple communication’s special features, the concept of the narrator within the narrative is differentiated into the actual narrator (i.e., the author) and the fictional narrator that appears in the narrative (a kind of character). The discourse possesses a complex nature, which is realized through this convergence. In some cases, the narrator’s unique qualities (e.g., uncertainty, ambiguity, multiplicity) can wholly obscure the concrete nature of the simulation of events.

134

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

In “Transposition (1): Narratology as a Production and Narrative Generation System, the author’s goal is to actualize narratology as a narrative generation system, so it goes without saying that this project necessitates the transposition of traditional narratology. The first point entails a transposition from an analytical, hermeneutical direction to a production-oriented path. The achievements of narratology’s great founders formed a very strong model for organization and orientation; it probably once contained the potential to develop the processes of generation and production, but in its successive phases, the practical application of textual analysis and interpretation became mainstream. Along with the trends of acceptance theory in literary investigations and the flourishing of cultural studies (as reading comprehension and the interpretation of texts became mainstream in literary research), narratological research came to coexist with other schools of thought and is now viewed as a tool for textual interpretation. Of course, logic and practice became established whereby interpretation and acceptance simultaneously became acts of projection. For example, reception theory implies that active interpretation on the receiver’s side is somewhat a productive act (i.e., acceptance is brought into the category of production). According to this logic, in the end, in doing the opposite of what its name suggests, acceptance theory has contributed to the expansion of authorial nature. What the author wants to say here is different: that it is literally an act of production. Looking at the origin of this idea, narratology is tied to the genealogy of rhetoric, or was envisioned with an aim toward the present revival of rhetoric. For Genette (1972), the root concept of narratology explicitly involves reviving this kind of literary tradition. Propp (1968) holds that the field of dependent morphology is centered on the search for the morphogenesis of plants and animals, a generative way of thinking. From this perspective, early narratology focused on production, and hence became production oriented. In this sense, it is not necessary to deliberately stress transposition; later on, narratology came to be treated as a single method of literary comprehension and acceptance. Concerning Transposition (2): From Narratology as Informatics to Expanded Literary Theory, the production-oriented transposition outlined above is realized more directly as a computer program that serves as a narrative generation system. For that reason, this narratology is based on informatics, information theory, and technology (especially AI and cognitive science), as well as the fusion of literary and narratology techniques. This section also denotes the transposition of narratology as information. From the author’s perspective regarding narratology as a whole, however, he aims to bring a greater, more complex meaning to this field, including not only including information but also various other ideas and intentions; he has referred to this project as expanded literary theory. In speaking of “expanded,” we should note that narratology originated as a dramatic expansion beyond the realm of text in a narrow sense as the object of literary research. Regarding methodology, narratology has been extended in terms of a deepening literary analysis by using methods of microscopic internal examination. Hence, narratology is already a broadened literary theory. Thus, the author’s approach can be called “expanded literary theory.” In Institutional Perspective, thinking of literature and narrative in conjunction with the concept of the institution is more common. For narratives and literature, the author takes an “institutional perspective.” The phrase has become a household word among members of the institutional population; it carries an academic meaning as well as an everyday sense, but the author does not necessarily (want to) use it in an academic way. The novelist Kenji Nakagami (1946–1992) frequently uses “institution” in his literary discourse. In particular, he uses “narrative institution.” In general, this phrase refers to the structure of the spell of the narrative, which is unconsciously expressed by the writer and cannot be deliberately avoided. The author of this book was greatly influenced by Yukio Mishima (1925–1970) as a boy. Later on, the author read the latest works by Yukio Mishima (2000–2006) in their entirety, 135

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

but given the concept of anti-psychologism and the pragmatic nature of words, the author realized once again that Mishima’s literature had given birth to his affinity for the institutional perspective, as well as his attachment to the idea of familiarity. Mishima is an incredible writer with a direct interest in political and social institutions. Regarding the section Narrative Generation System, for the author, the construction of narratology is synchronized with building a narrative generation system. To put it in extreme terms, in this research, the narrative generation system corresponds to the theory of narratology itself. A narrative is a cross-border concept; in response to it, narratology has evolved as movement to bring about a unified perspective. However, a narrative generation system goes further since it seeks to actualize narratology into a cohesive, convergent core that synthesizes analysis and production. A narrative generation system serves as a framework for incorporating the cross-border technique of narrative into this cohesive, convergent method. Chapter 1 (Ogata, 2020f) in Ogata (2020e) provides a concrete description of a narrative generation system (INGS), but, as mentioned above, the entire book is organized around a stance that expresses various aspects of the narrative generation system. The author touches upon the concepts of literature versus narrative. Like a narrative, literature is a cross-border notion. Within the framework of a narrative generation system, narrative and literature overlap, and contrast with each other at certain points like the points on a fan. Narratology expands outward toward the scope of the target text (a literary text in a narrow sense); the author’s research is based on this foundation. At the same time, the opposite tendency arises: to go against the current and return to the literary text. From this angle, the narrative generation system’s specific output is literary text. Simultaneously, the narrative generation system is oriented toward diverse fields of study, actions, and practices that lie at the center of literary thought. “Narratology”—which actualizes the narrative generation system—has expanded since “literary theory” originates in the trend of the movement from narrative to literature. Next, the author reviews the concept of circular narrative generation based on Chapter 4 (Ogata, 2020d) in Ogata, (2020a). This study deals with the phenomena of narrative generation, involving ever evolving and generated things at the social and personal levels. The three types of philosophical ideas in the same group are “circular narrative control,” “norm-deviation,” and “fluidity-fixation.” At the macro level, they dynamically control the abovementioned, ongoing generating narrative generation. “Circular narrative control” implies that a narrative generation procedure is performed through a sequential process without a beginning or end, and it continues across different media and genres. This concept can be used to control INGS, as described in Chapter 2 (Ogata, 2020g) in Ogata (2020e). Many narrative generation systems produce a specific type of output information from a determined type of input information. One basic idea is to be able to create diverse types of output information from various pieces of input information grounded in circular narrative control. The following description explains this notion using two examples. This concept is related to developed mechanisms. An attempt is made in the narrative discourse phase of INGS (a more detailed explanation will be provided in a sub-section of the next section). The control mechanism—which applies the reception theory by Jauss (1970)—iteratively repeats the narrative discourse generation according to the interactions between the narrator and narratee involved in the system. The control mechanism embodies the idea of circular narrative control. Although this mechanism is meant for a particular component of generation, it was originally intended to perform a free, flexible narrative type of ordering, for which the sequence/order of the operating modules/phases is not preliminarily decided upon. INGS, in the current stage, partially encompasses the concept in the 136

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

abovementioned sense by introducing a music generation and transformation mechanism, which the author has been developing in piecemeal fashion. The idea is not only that music is the accompanied expressive medium of semiotic narrative representation, but also that its input-output relationship can be quite flexible and diverse. This idea has been generalized to the connection between music and narrative as an essential notion for narrative generation, via the appearance of circular narrative control. Circular narrative control entails the design of a narrative generation system envisioned by the author, in particular the current form of INGS. The first meaning is iterative narrative generation, as shown in the experimental, implemented mechanism for the narrative discourse phase, explained earlier. This concept of the first meaning will be tied to a method through which a narrative generation cycle is perpetuated to obtain numerous generated outcomes; it will be further synthesized for an edited, individual narrative work. In contrast, the second meaning refers to the free, flexible control of the applicable order of narrative generation phases. For example, INGS should not fix the generation order among all modules or generation units, including conceptual structure generation and surface expression generation. INGS should be designed to enable various generation courses and repletion. Furthermore, GIS is a medium for circular narrative control in that the narrative production and consumption mechanism causes the myriad levels of narratives to repeat (such as individual narrative works, an actor’s authentic narratives, and the entire narrative of a sequence of each productive narrative).

Narrative Generation Mechanisms Based on Post-Narratology This section introduces the following topics: overviews of (1) INGS and (2) GIS; and (3) previous attempts to combine INGS and GIS.

An Overview of the Integrated Narrative Generation System This section deals with INGS as the most important element in the above approach, or rather, a concretization of the author’s earlier described thoughts on narrative generation. The following is according to the flow of Chapter 1 (Ogata, 2020f) in Ogata (2020e). THE ARCHITECTURE OF INTEGRATED NARRATIVE GENERATION SYSTEMS explains the architecture underlying INGS in terms of the mechanism (The Generation Mechanism of INGS) and content (Content Knowledge in INGS). The main generation mechanisms contain the entire process: the story generation, the narrative discourse, and the narrative representation mechanisms. The generation controls are also included in the explanation. The central components of content knowledge are conceptual dictionaries and narrative content knowledge bases. Two large sections, The Generation Mechanism of INGS and Content Knowledge in INGS, describe the above topics. Figure 1 is shown. The primary generation modules are for the story, narrative discourse, and expression (including language, music, and picture images). The story and narrative discourse are embodied by their respective conceptual representations. All modules are controlled by each local control mechanism, which are collectively under global dominance. Story and discourse generation are performed using story and discourse techniques, respectively. Furthermore, INGS has dictionaries (e.g., for concepts and language) and various knowledge bases (e.g., for story content and state-event transformation). In addition, as stated in previously, several literary theories are incorporated into INGS for the practice of expanded literary theory. Although the figure below of INGS architecture has been used in prior papers, and several alterations have been made to the architecture, the author uses this figure in this version 137

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Figure 1. The architecture of INGS

because the foundation will not change in this first draft (however, the next extended draft will use a developed version of the figure). Regarding The Generation Mechanism of INGS, as mentioned earlier, the author considers aspects of INGS architecture through the division of content and mechanism (or form). Firstly, this section explores the formal attribute of mechanism. For Entire Generation Process, INGS proposes mechanisms to control the overall generation process, as well as several smaller points. The control mechanism’s chief function at each stage of generation is to operate the characteristics of a narrative to direct the use of narrative techniques. Although many of these mechanisms are simple, the Jauss-based one (which, comparatively speaking, is precisely designed) is only employed in the narrative discourse mechanism. One characteristic of narrative generation control in INGS is circulative control, which is grounded in the use of a fundamental concept of the author’s narrative generation. INGS aims to implement a form of circular narrative control. The INGS mechanisms are partially developed through the diversity of generation routes. Ideally, this control mechanism means that all the generation mechanisms can receive any type of input information. The user can select the generation route at the starting time in a session in any of the following ways: 1. The user selects a possible route: The user can choose a free combination of “story, narrative discourse, sentence, visual image, original music, and variation music.” 2. The system automatically determines a generation route. 3. The system automatically determines the next module at the ending time of a generation module. 4. If the user does not select any of the previous options, the system determines the order: “story → narrative discourse → sentence → visual image → variation music.”

138

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

In general, a generation module harnesses the previous output data as the input data. However, for example, for the order “story → sentence → discourse,” the discourse module takes the output data from the story module as the input data. Such details are explained in the respective sub-sections. The information in this table will be shown in the sections for each module. For Story Generation Process, INGS provides mechanisms that control both the overall narrative generation process and each point of the narrative generation. During the phase of story generation, the system produces a story using a tree structure with three basic elements: event, relation, and state, organized by time. Each event is constructed using an event concept and semantically related noun concepts. The conceptual representation is used as a frame structure to semantically categorize each noun concept, using eight basic cases (“agent,” “counter-agent,” “object,” “location,” “time,” “instrument,” “from,” and “to”) and nine special cases (“adverb,” “possessive,” “situation,” “purpose,” “experiencer,” “source,” “idiom,” “information,” and “as”). Each event accompanies the preposition state and the resulting state at the beginning and end of the event. At the same time, each relationship links an event or aspect of the tree structure (which includes events) to another event or aspect of the tree to create a new structure, whereby events and states have a top-level relationship. Many relationships (such as “cause-effect,” “sequence,” and “script”) are created. States and events correspond to static and dynamic informational elements, respectively. A state’s central functions are to manage knowledge about the story’s world and to handle the flow from event to event. Although a story is a temporary organization of events, its semantic consistency is governed by relationships and states. For Narrative Discourse Mechanism, like the story generation mechanism, the narrative discourse mechanism is composed of narrative discourse techniques applied to its developing structure. The similarity between the story generation and narrative discourse mechanisms is grounded in the following characteristic of INGS. Namely, both mechanisms play the role of conceptual processing in narrative generation in INGS. Here, conceptual processing refers to deep, semantic narrative generation in INGS, in contrast to surface narrative processing, which uses various representative media in the narrative expression mechanisms in INGS. Although the concrete processing of the narrative discourse mechanism differs from story generation processing, they perform similar procedures, formally speaking. The narrative discourse module executes generation and circular that employ the narrative discourse theory by Genette (1972), and the reception theory by Jauss (1970), respectively. The use of these theories symbolize a concretization in the narrative discourse mechanism of expanded literary theory, along with the introduction of Propp’s hypothesis on the story generation mechanism. The latter mechanism (grounded in Jauss’s reception theory) reveals a possible implementation of narrative generation control based on numerous possibilities. A narrative discourse (generated or modified by the Genette-based mechanism) corresponds to the structural aspect of narrating a story tree structure, including a sequence or series of events. The discourse structure to be narrated and its required surface media (such as language and images) are points distinguished within the system, and are described in terms of conceptual representation, as stated above. The narrative discourse should correspond to a transformed story structure. From the angle of expanded literary theory (which simply uses narratological and literary theories), in the key part of the process, Genette’s philosophy develops a set of discourse methods as the root of narrative techniques. The Genette-based and Jauss-based control mechanisms are linked to this phase. However, INGS, a comprehensive system, incorporates existing narratological and literary theories/approaches into the entire system framework. Hence, the author examines the generalization of theory-based mechanisms by looking at the design and development of INGS. 139

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Regarding the concrete process of narrative discourse, a generated story structure is transformed into a narrative discourse (as the narration’s structure) according to a temporal sequence of events and states. A narrative discourse structure can reorganize (or further ignore) the temporal order of events in the story. Additionally, in INGS, the narrative discourse structure to be narrated is differentiated from the narrative representation by the surface representation media (such as language, images, and music). Narrative discourse structures are described using corresponding conceptual representations, like in story structures. Thus, a narrative discourse coincides with a transformed story structure. Although it is a tree structure in which events are the main elements, the different relations from story structures are used. A narrative discourse structure is described by the tree form, transformed from a story tree; the events and states in the narrative discourse structure are same as in the story structure. The Narrative Representation Mechanism engenders surface representations and contains the following sub-mechanisms: 1. The natural language generation mechanism creates a natural language representation or expression. 2. The image generation mechanism forms an image presentation or expression, namely visual images of moving and still pictures. 3. The music generation mechanism produces musical representations or expressions. Previous studies on INGS have focused on creating conceptual representations, while the above mechanisms for producing surface representations (especially language and images) remain at the invention/ framework stage. The following sub-section discusses natural language (1) and music generation (3). In addition, an experimental function of the mechanism for generating narrative images (2) yields pictures that correspond to elements in INGS’ conceptual dictionaries, using image management rules to synthesize them. In response to a story or narrative discourse structure, the mechanism generates a text command file, which delivers a series of processing commands to the image representation interface. In the future, the author aims to construct an image generation mechanism in INGS based on the similar concept of the music generation mechanism. In particular, the plan treats the generated tree structure of narrative conceptual representation as the tree structure of image representation; it inserts various rhetorical elements of imagery, corresponding to rhythm and tempo in the case of music, for example. The fundamental grammar of image representation—which corresponds to language-based case structures and sentence patterns—may effectively control concrete image representation through a camera work function or other means. Characters can be employed in the generated narrative visual representation. (The above part is the direct reference.)

An Overview of the Geinō Information System GIS is the most vital element in the above approach, or a concretization of the philosophy underlying the author’s narrative generation Chapter 3 (Ogata, 2020c) in Ogata (2020a). The following introduction is according to the reference. In The Sequential and Multiple Narrative Generation System and the Geinō Information System, there are phenomena called narrative genres and mixed media. In the previous section, The Development and Distribution Mechanism, the author tried to capture such phenomena in terms of content, but they have occurred naturally since ancient times. For example, in the case of kabuki, in 140

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

addition to texts that are performed, during the Edo period, there was also a type of critical (and even advertising) narrative, such as written commentary on each actor. Scripts have also been reading through books. At present, it is not uncommon for films, novels, comics, games, and other forms to be linked. In the case of music, various conversions are also performed, such as recording and editing live musical or theatrical performances at venues (e.g., records and CDs) and distributing them via TV or another kind of network. In many cases, the task of textual conversion between media is not done purely at the individual level but by an organization (such as a company). This category includes publishers, film and video production companies, TV and radio stations, advertising agencies and production businesses, game production enterprises, and entertainment production. These are not the only entities that engage in such profit-generating activities; some artistic groups promote a certain art form based on the artistic lineage or houses (the iemoto system), such as floral design (ikebana), the tea ceremony (sadō) (both popular rituals in Japan), and literature (novels, poetry, tanka, haiku, and other forms of literature). Coterie associations focusing on manga and other interests, as well as (depending on one’s point of view) religious groups, can be included too. It is normal for organizations, companies, and other groups to be involved in a “narrative generation system as genre production.” However, up to the previous stage of simple narrative generation, in contrast to the organizational involvement in which the creation of a text remains internal, and at the stage of sequential or composite narrative generation, the center of gravity of organizational engagement shifts from a relationship with individual works to wielding control over them. In other words, the difference is that the production and control of multiple works is carried out at this time, whereas in the past, this stage entailed creating a single work. In this sense, there has been a qualitative change between the two. If we consider the problem in this way, the subject does not necessarily need to be a group or an organization. From this angle, we can also grasp the mechanism of the “author” as an individual, such as a novelist. This point is connected to the GIS described below. Yet broadly speaking, we discuss this issue in terms of the general trend of its superimposition on the organizational mechanism. Geinō Information System: Overall Architecture and Thought introduces GIS and addresses the basic concept of transforming INGS into GIS. The system architecture is currently fluid and will be expanded and revised in future concrete design/implementation. In addition, “why geinō” is an important point. For the author, narratives are not literature and arts but rather geinō. In the narrative genre system (Chapter 2 (Ogata, 2020b) in Ogata (2020a), narrative as literature and arts is only one part. Further, this is also related to the fact that the author’s narrative generation study (or what the author ultimately desires to achieve with the narrative generation study) is not what is advanced and avant-garde but instead “commonplace” and “non-unique.” According to the author, a narrative is the farthest from anything represented by words such as “advanced,” “avant-garde,” and “innovative.” Figure 2 portrays a conceptual composition of the GIS. The INGS mechanism is included as a component. The architecture will be changed in future versions. Regarding previous attempts, refer to papers such as Amino, Kawamura, and Ogata (2002). GIS corresponds to a narrative production-consumption mechanism that encompasses the multiple use of INGS as a narrative generation-reception mechanism. The system is, at the highest level, composed of a sender mechanism, a receiver mechanism, and a geinō history. Simply speaking, GIS continuously produces geinō content through the interactions between the sender and receiver mechanisms. In the sender mechanism, a variety of methods coexist and, at the highest level, it branches out into geinō works and geinōjins (each a kind of sender mechanism). Thus, the GIS also creates content related to the subjects of geinō works, called geinōjins, in addition to engendering content 141

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Figure 2. The architecture of GIS

as geinō works. Further, as mentioned in the above review of informational narratology, receivers of a narrative frequently realize the level at which they themselves are the senders, beyond their positioning as pure reception subjects. In this sense, receivers can also be viewed as subjects, who create narratives. GIS produces diverse content through the above mutuality between the sender and receiver(s). However, the production-consumption process leads to a sequence of pieces of content, in addition to content as a simple object. A sequence of pieces of content is referred to as “geinō history.” Namely, from a more macro angle, GIS produces geinō histories according to the mutuality between the sender and receiver(s).

Previous Approaches to Combining INGS and GIS In TOWARD INTRODUCING KABUKI KNOWLEDGE INTO GIS WITH INGS: FROM ANALYSIS TO GENERATION in Chapter 2 (Ogata, 2020g) in Ogata (2020e), based on the above framework or components, the author surveyed several kabuki elements, including “person,” “story,” tsukushi

142

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

or zukushi, and naimaze. Although the person and story aspects are included in the fifteen elements, tsukushi and naimaze are not. They belong to another group in the author’s framework of the narrative generation study of kabuki. The latter elements are a group of universal narrative techniques that can be applied to various components. In particular, the fifteen kabuki elements are not concrete narrative techniques of kabuki, but instead are meant to construct a kabuki phenomenon. They can contain various narrative methods employed in kabuki. However, tsukushi and naimaze are respectively a narrative approach or set of narrative strategies that can be used in numerous elements of kabuki. For example, the narrative technique(s) of naimaze can be employed in terms of the person (character), story, and place. Although the examined narrative approaches for the person, story, tsukushi, and naimaze are not organically incorporated into INGS and GIS or GIS with INGS, they are positioned as important knowledge components of the unified system. Therefore, the following section reveals the macro-level framework of GIS with INGS. Regarding A Macro Framework of GIS with INGS, a kabuki play and sequence are produced through a comprehensive narrative generation and production mechanism (the GIS with INGS in the author’s model), using various types of kabuki knowledge, as stated in the previous section. These knowledge elements correspond to each knowledge base, storing historical, cultural, and other information on kabuki. The kabuki-related, domain specific knowledge bases are intended to be included in GIS, while more general, domain-dependent knowledge bases are intended to be developed in INGS. A single generation/ reception of a kabuki play repeatedly extends to production and consumption at the social level. The author considers a macro-level design plan for GIS with INGS. As described repeatedly in this chapter, INGS is a simple narrative generation system that corresponds to the first mechanism in modeling the author’s comprehensive narrative generation research. However, the current version of INGS cannot create synthetic narratives using knowledge acquired through the analysis of kabuki from the angle of a narrative generation system. For the second mechanism, it is necessary to focus on the metalevel mechanism, which controls the use of one or more INGS(s), namely GIS. This is the meaning of the term “GIS with INGS.” The above findings will contribute to any part of INGS and GIS. Yet it is too difficult to simply divide them, for instance, by simply thinking that the “character” and the “actor” in the “person” can, respectively, be associated with INGS and GIS. Notwithstanding, a character is partially a narrative element to be stored in a portion of GIS in its connection to the narrative world (sekai). Regarding the framework of the relationship between INGS and GIS, as shown in Figure 3, the INGS treats narrative generation as a simple subject. By contrast, the GIS produces sequential narratives on a broad level, which includes each piece of narrative work and narratives of geinōjins (entertainment performers) such as ordinary people, actors, and characters.

The Narrative Generation of Kabuki This section covers the author’s narrative generation study of kabuki, or kabuki as a narrative generation study. This paper examines kabuki according to the constructive (computational) method or the cognitive scientific approach. The author describes the meaning (although this has been the technique of cognitive science, many researchers do not know). This is also a pre-conditional concept of the author’s expanded literary theory or post-narratology. The above is the theoretical direction. (As an engineering or constructive/productive method, this section uses the kabuki analysis based on the constructive method to produce kabuki narratives grounded in any direction. The following part refers to chapter 2 (Ogata, 2020g)

143

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Figure 3. Expanded narrative generation by INGS and GIS

(TOWARD INTRODUCING KABUKI KNOWLEDGE INTO GIS WITH INGS: FROM ANALYSIS TO GENERATION) in Ogata (2020e), delving into previous concrete attempts or the current findings. Based on the above framework or components, the author surveyed several kabuki elements including “person,” “story,” tsukushi or zukushi, and naimaze. Although the person and story are included in the fifteen elements, tsukushi and naimaze are not included. They belong to another group in the author’s framework of the narrative generation study of kabuki. The latter elements are a group of universal narrative techniques that can be applied to various components. In particular, the fifteen kabuki elements are not firm narrative techniques of kabuki, but instead are intended to build a kabuki phenomenon. They contain diverse narrative strategies employed in kabuki. However, tsukushi and naimaze are, respectively, a narrative approach or set of narrative techniques that can be used in numerous aspects of kabuki. For example, the narrative strategy or methods of naimaze can be used for the person (character), story, or place. Although the investigated narrative approaches for the person, story, tsukushi, and naimaze are not organically incorporated into INGS and GIS, or GIS with INGS, they are positioned as critical knowl-

144

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

edge components of a unified system. Hence, the following section reveals the macro-level framework of GIS with INGS. Regarding A Macro Framework of GIS with INGS, a kabuki play and sequence are produced through a comprehensive narrative generation and production mechanism (the GIS with INGS in the author’s model), using various types of kabuki knowledge, as stated in the previous section. These knowledge elements correspond to each knowledge base, storing historical, cultural, and other information on kabuki. The kabuki-related, domain-specific knowledge bases are intended to be included in GIS, while more general, domain-dependent knowledge bases are intended to be developed in INGS. A single generation/ reception of a kabuki play repeatedly extends to production and consumption at the social level. The author considers a macro-level design plan for GIS with INGS. As described repeatedly in this chapter, INGS is a simple narrative generation system that corresponds to the first mechanism in modeling the author’s comprehensive narrative generation research. However, the current version of INGS cannot create synthetic narratives using knowledge acquired through the analysis of kabuki from the angle of a narrative generation system. For the second mechanism, it is necessary to focus on the metalevel mechanism, which controls the use of one or more INGS(s), namely GIS. This is the meaning of the term “GIS with INGS.” The above findings will contribute to any part of INGS and GIS. Yet it is too difficult to simply divide them, for instance, by simply thinking that the “character” and the “actor” in the “person” can, respectively, be associated with INGS and GIS. Notwithstanding, a character is partially a narrative element to be stored in a portion of GIS in its connection to the narrative world (sekai). Regarding the framework of the relationship between INGS and GIS, as shown in Figure 3, the INGS treats narrative generation as a simple subject. By contrast, the GIS produces sequential narratives on a broad level, which includes each piece of narrative work and narratives of geinōjins (entertainment performers) such as ordinary people, actors, and characters. As for Two Multiple Elements and Two Technical Concepts: Person, Story, Naimaze, and Tsukushi, based on prior studies (Ogata, 2016a, 2018e, 2019c, 2019d, 2020g; Ogata, Fukuda, Ono, & Ito, 2020), the author supplements two types of multiple elements—person and story—and two kinds of technical concepts—naimaze and tsukushi (zukushi). These investigations reveal the multiple structures of the persons or people of kabuki. In kabuki, there are numerous “persons,” such as “actors” (performers), (dramatic) “characters” and “real persons.” These personas temporally and spatially form (and possess) multiple existences and relationships. In terms of Preparatory System Development Based on the Analysis of Kabuki, this section discusses preparatory system development using methods that the author acquired through prior studies on kabuki. One of the goals is to contemplate, in detail, narrative functions, mechanisms and techniques regarding kabuki in order to implement GIS and INGS in the future. Thus, the proposed systems are relatively simple, although they are grounded in a consistent approach: “generation based on editing.” In particular, the systems employ actual kabuki texts and edit them using narrative techniques accumulated through the author’s analysis of kabuki. As Ogata (2019c, 2019d, 2020g) outlines the fundamental technique, this section revises and expands the description based on the summary. In A Sequential Generation Example, the author traces a narrative generation process through the connections of several generations using the methods of person, story, naimaze, and tsukushi. For Analysis and Generation of a Person’s Information, the author scrutinizeShōwacquired, described, and structuralized property information of kabuki actors is linked to the GIS and INGS. Figure 3 includes the conceptual figure of the GIS. Kabuki actors’ geinōjin and life information is stored in the geinō work resource (geinō resource) and life work resource (life resource). Figure 3 also indicates that 145

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

the GIS with INGS starts out as a narrative and becomes a sequence of narratives, receives them, and consumes them. Moreover, this figure displays the circulation of these processes. In Selecting and Editing Stories, Ogata chooses two stories on kabuki works, Sukeroku and Narukami, that Ichikawa Ebizō XI frequently produced. Kanjinchō is added as another example. The three kabuki works have relatively simple, clear scenarios. The author shows these scenarios’ synopses (from Wikipedia in Japanese), their property data, and generated sentences. For Using Naimaze, the dramaturgy of kabuki is constructed based on kata and shukō. Kata refers to the typical patterns of diverse elements in kabuki, including characters, places/scenes, and stories/plots. The tradition of kabuki contains different kata. Kata is also related to the concept of sekai (world) in kabuki. The author provides explanations of sekai in a post-narratology chapter (Ogata, 2019b) and the INGS chapter (Ogata, 2019a) in a previous study (Ogata & Akimoto, 2019). Regarding the history of kabuki, the document Sekai Kōmoku (1916), from the Edo period, gives a systematic description (as a kind of database/knowledge base) of sekai. Sekai Kōmoku provides actual works in many narrative genres including nō, ningyō jōruri, novels, famous characters, and typical stories/plots. In the production of a kabuki work, one more sekai are chosen to determine the essential kata to carry out a work efficiently. However, early kabuki writers did not simply apply the kata to the work. The second concept, shukō, is equal to special, individual narrative techniques for transforming kata to make a work more interesting. Naimaze, which fuses different stories (as kata) to form a new story, is also a type of shukō. Applying a naimaze method to a story/plot is the most general usage. Naimaze is also related to the element of a “character” as a “person.” In particular, although it is possible for naimaze to mix the “stories” or “plots” of two or more narratives, the use of the naimaze of “stories” or “plots” is tied to the naimaze of “characters” in different works. Similarly, introducing “characters” from different works into a new one engenders an associative relationship between new and old works. Ogata (2019c, 2019d, 2020g) presents an approach for designing the narrative techniques of naimaze for INGS (and GIS) by applying the method for combining “moves,” as explained in Propp’s (1968) narratological study. However, the following example is simpler. An example synthesizes the synopses of Sukeroku, Narukami, and Kanjinchō into a new scenario using naimaze techniques. The author employs the conceptual structures of Sukeroku, Narukami, and Kanjinchō. This form of a conceptual structure is the story representation that is used in INGS. Hence, the narrative data described here can be harnessed in the synthesis with INGS. First, the author transforms the sentences in each story into a collection of simple sentences using a semi-automatic script generation program developed by Arai, Ono, and Ogata (2016). Second, each modified sentence is altered manually using an input format. Using these input data, stories of Sukeroku and Kanjinchō are inserted into the story of Narukami as the foundation. Sukeroku and Kanjinchō are connected to Narukami via the points in which large, special changes appear. These connecting stories cut through the natural flow of Narukami’s story. Ogata and Kanai (2010) study this type of story-cutting rhetoric. Sukeroku, the hero in Sukeroku, reaches new places after the events of “Narukami is angry” or “Yoritomo is angry.” Kumonotaemahime, the heroine in Narukami, is changed to Agemaki, who is the heroine in Sukeroku, based on the common trait of being female. For Using Tsukushi, tsukushi is comprehended from the angle of techniques through which objects respectively relate, meaning that objects are inserted (or sprinkled) into a narrative work/part. Methods of tsukushi may correspond to lateral techniques for expanding objects. In the basic approaches of tsukushi, strategies for connecting objects to one another are included and divided into methods by semantic and non-semantic relations. The former, for instance, includes the geographical relationships of places and the conceptual relationships of objects. The latter contains relationships through dajare (puns) and 146

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

linguistic sounds. Many spatial objects can be inserted into a narrative using geographical movement and temporal change. The typical structure is michiyuki. For A Final Example and the Positioning in GIS with INGS, according to the story construction, the final story structures are produced. These include the generated explanation sentences of Ichikawa Ebizō XI, the generated explanations of the three kabuki works (i.e., Sukeroku, Narukami, and Kanjinchō), the works’ generated synopses, and the synthesized (and expanded) kabuki synopses that apply the techniques of naimaze and tsukushi.

Systematic Surveys and Analyses of Kabuki Elements, Seen from the Viewpoint of Producing a Kabuki Work on Stage The multiple elements of kabuki—perceived based on the author’s narrative generation study and multiple narrative structures model (which the author addresses in this section)—are the central outcomes of the author’s kabuki investigation. Alternatively, the consideration of these kabuki elements serves as a point of departure for a more full-fledged narrative generation study of kabuki. Regarding these multiple components, after uncovering initial findings, the author refines and extends the elements by providing abundant examples. However, these efforts are conducted using the same framework, which itself has not been altered or expanded. The narrative generation study of kabuki does not aim to merely perform an analysis. Since its goal is to be connected to the design and development of narrative generation systems based on the knowledge and techniques of kabuki, the author makes several attempts related to system modeling and system implementation, which depend on the same framework of multiple elements of kabuki. In contrast, currently, grounded in the previous accumulation of data, the author is rethinking the framework of the multiple elements of kabuki in order to change and expand the framework. The author will contemplate these issues in a future study. The author analyzes the multiplicity of kabuki with regard to the fifteen elements, displayed in Figure 4. The term, “multiple,” in the caption of this figure indicates that each element has multiplex aspects. The author regards the multiplicity as one of the important features in kabuki. For instance, the element of a “person” in kabuki can be divided into the following three aspects: the real “person” who lives in the actual world; the “person” as an actor who has a history of performing roles; and the “person” of the kabuki character that appears on stage. Each person has his/her own history or story. In this section, the author explains the main elements of kabuki, or narrates them as a story, from the perspective of an audience that is viewing kabuki plays in a kabuki theater. In the following, the author describes these multiple elements of kabuki according to a flow or story of an audience in a kabuki theater. Let us start this story with a scene where an audience enters a kabuki theater; for example, Kabukiza in Tokyo. What is the audience of kabuki ([a] Audience)? Kabuki is primarily meant to be seen in a theater by an audience. Of course, this does not mean that viewing kabuki in the physical space of a theater is an absolute experience for the audience. However, kabuki is closely, essentially, and mutually bound to the theater’s existence. The audience also plays a role in the multiple editing of kabuki. The difference between seeing a kabuki play as a movie and in the theater is greater than the difference between seeing a movie on TV and at the cinema. The perspective (camera work) in a movie has already been prepared based on the creator’s intention and technique. The camera work is often regarded as the most creative narrative technique in a movie. Rather, when an audience views a movie at the cinema, the stance in the work does not change according to the place in which the audience witnesses it. For kabuki, there are

147

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Figure 4. Multiple narrative structures of kabuki

different outlooks according to the place in the theater. An audience’s positioning in a theater is decided upon either intentionally or accidentally. First, the audience experiences [b] Theater, Stage, and Seating. Traditional stage dramas in Japan incorporate theaters into necessary components in respective genres. In kabuki and other Japanese geinō genres, a theater is not a universal space where all (or diverse) theatrical genres are performed. In particular, no, ningyō jōruri, and kabuki take place in exclusive theaters and are performed on stages specifically suited for them. The hanamichi is a mechanism on the kabuki stage that has multiple uses as a unique device, and is effectively utilized in real kabuki theaters in Japan. The audience members view a [c] Stage Performance. Kabuki during the Edo period was a contemporary art form of the time. Kabuki plays were performed in a style called tōshi-jōen, in which all parts are played continuously over one or two days. However, the actual form was not necessarily simple, as there were complex rules or performance systems. Two types of newly created works—jidai-mono and sewa-mono—were mutually performed over a series of kabuki performances. In contrast, after the Meiji era, the midori-jōen, a then-new performance style in which the classical works of Edo kabuki were repeatedly performed, gradually spread and became popular. Numerous new works (scenarios)—including the sub-genre of shin-kabuki (new/modern kabuki)—were published by many writers, including famous ones such as Kidō Okamoto (1872–1939) and Seika Mayama (1878-1948). The performances and dances in the theater have respective [d] Style(s) (Form(s) or Pattern(s). The term “style” here means “performance style.” Different styles are adapted for the stage in kabuki plays such as dance, music, narration, and action forms or patterns by actors. A variety of multiplicities and possibilities exist based on the constraints of prepared elements and combinations. The music uta-mono (played as actors carry out actions) includes myriad genres such as nagauta, kiyomoto, tokiwazu, and ō-zatsuma. The narration by a takemoto, a type of katari-mono (a narrative genre)—corresponding to 148

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

a tayū in ningyō jōruri—was transformed so that human actors could play the roles instead of artificial puppets. In ningyō jōruri, all parts of a work’s scenario are narrated by several tayū; a kabuki work is divided into the katari (narration) by a takemoto and the serifu (speech), given by human actors. The artistic blend of kabuki and ningyō jōruri gave birth to another aesthetic technique called ningyō-buri, a special performance style in which a human actor imitates (or simulates) the action of a puppet in ningyō jōruri. An interesting, complex structure in kabuki is also represented in the example. Although any method could be called a “direction” for a stage performance, [e] Direction in kabuki has unique features. The attribute of a kabuki play—whereby different historical periods and places in a narrative are mixed (rather than representing a realistic standard)—influences the production of each actual stage. Various styles of blending, or the overlapping of costumes, languages, speech, and music, sometimes exceed ordinary situations. The characters’ make-up and costumes are designed according to the form or type of traditional norm to be used for each typical character. Such symbolic and formal qualities frequently ignore the realistic consistency and naturalness of the entire staging. The author considers that the above characteristics show the “expressionism” of kabuki. Kabuki frequently emphasizes the beauty and effectiveness of expression more than reality. One reason may be the disordered confusion of forms through historical accumulation. Some actors, such as Ichikawa Danjūrō IX (1838–1903) during the Meiji era, proposed innovative ways of restoring kabuki to a realistic style. However, since overly realistic and symbolic styles form a tense tradition and a unique beauty involving the audience, kabuki grew into a more heterogeneous, synthesized dramatic genre than one of pure realism. In addition to the dramatic characters in dramas and dances on stage, players or actors appear. [f] Person in kabuki has multiple senses. A “person” can be an “actor” who performs on stage, or a (dramatic) “character” in the narrative of a work. These two kinds of individuals form numerous relationships on the temporal and spatial levels. First, a person plays a character in a kabuki play; at the same time, the existence accompanying the real name also has an essential meaning. The aspect that “an actor plays” in kabuki frequently has a more important significance than representing a universal drama or story. A kabuki scenario from the Edo era was not written with the names of the characters in the story but with the names of the actors who performed the drama. In particular, a scenario was ordinarily written according to the characteristics of the actors in classical kabuki works. Hence, the actors were not selected based on any artistic viewpoint grounded in a preliminary scenario. Although there are many critics of the weak points of such a star system and, after the Meiji era, scenario-centered drama production became a popular method, the tradition of a star system that emphasizes the actors has persisted in contemporary kabuki. This means that many audiences go to a kabuki theater to see their favorite actors. Dramas and dances are called [g] Works. Through the historical development of kabuki, a work was often adapted to other ones using various types of editing and revision methods. Although influential relationships among works are ordinary phenomena in the current artistic environment, very quick and active transformational phenomena, especially during the Edo period, would have partially relied on the lack of awareness of copywriting. A more practical reason for such an outstanding production style was that a revised work was easy to adapt to the unique traits of the author, theater, actor, or troupe. This is a very interesting problem to be considered in association with the theoretical and methodological narrative production techniques of kabuki, using the concepts of sekai (“world”), naimaze (“blending”), and shukō (“ingenuity” or “device” as ideas and contrivances). In the previous section, the author uses words such as “drama” and “dance.” These kinds of [h] Genres exist in kabuki. Creating a new work based on one or more past ones in the same or another genre is a common approach in Japanese dramatic or geinō genres, such as nō and ningyō jōruri. This 149

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

method of creation reached a unique, extreme level in kabuki. Kabuki is a synthetic collection of diverse geinō genres, including nō, kyōgen, and ningyō jōruri as dramatic geinō genres. Generally speaking, the transformation among diverse genres in narratives is a universal phenomenon in Japanese literature, geinō, and the arts. In this case, narratives are a type of versatile, general software. In contrast, genres are media for narratives in the form of software, and a narrative can be embodied in many different media. Kabuki has a number of genres and works. They are narratives in the broad sense and commonly have a [i] Story and Plot. In kabuki narratives, we can see the multiplicity (overlapping features) of stories or plots. This facet is also tied to sekai, naimaze, and shukō mentioned in [g] Work. The history of kabuki resulted in the accumulation of stories and plots. Many narrative stories/plots are blended into fresh, more complex versions. Thus, audiences of kabuki can associate the fragments of stories/plots that they already know, and imagine another story/plot from the work performed on stage. A kabuki work is usually performed by several actors, and takes place on the narrative stage in terms of place ([j] Actor and Place). According to Tamotsu Watanabe (1989), a limited number of actors in a theater troupe can play multiple repertories of kabuki by playing various dramatic characters. A theater troupe is needed to make preliminary preparations and the necessary arrangements to construct a myriad of places on stage: settings for fictional locales and costumes for the actors’ roles. Since both characters and costumes are categorized in a patterned system, the combinations can be limited. This combinatorial mechanism enables actors to play different roles using a restricted number of costumes and closely related make-up. Likewise, places on the stage can be set up by using a certain number of settings and combinations. Such features of kabuki engender the artistic and narratological aspect in which the realism of performing is not respected. Similarly, a kabuki narrative includes the unique temporal scope in which the narrative is developed; for instance, a specific [k] Time (Era or Age). Different historical times are frequently mixed in a kabuki play. Such temporal multiplicity has been linked to the censorship system of the Edo government, in addition to artistic requirements. Direct representation of contemporary events was forbidden under the political circumstances of the time. A work based on a modern, real incident repeatedly changed the temporal setting to a past age, and was expressed in a different temporal setting, using the sekai of another era. Changes of time were simultaneously tied to changes in spatial elements, the components for space and place. We can see in kabuki an interesting phenomenon regarding temporal and spatial transformation in which the spaces and characters in different historical periods are overlapped. Seeing a kabuki narrative from a different angle, each narrative has [l] Material or Topic. Materials or topics commonly used in kabuki works already exist and have been categorized under numerous groups. For instance, when a kabuki work is produced using certain existing material, and subsequent work is created based on editing a previous work, the latter work becomes a variation of the former while using the same material/topic as the original. This ongoing flow has resulted in multiple groups of kabuki works according to the same respective materials. Although this may be similar to the theme of a “real incident,” the material or topic referred to here is not necessarily a genuine event. Fictional occurrences are also utilized. Further, “real” and “fictional” can be changed through a kabuki generation sequence or circulative evolution. A kabuki work produced from a true event can transform into a fictional narrative. Kabuki works/narratives frequently use a [m] Real Incident, either from modern times or the past. Stories/plots in kabuki are often constructed based on actual incidents; for instance, historically famous events and facts for jidai-mono, namely “historical story type.” In contrast, sewa-mono (“contemporary, realistic story type”) frequently employs present-day events from the daily lives of ordinary people. Typically used historical sets of events have been repeatedly introduced and arranged in many kabuki 150

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

works. The links between a real incident and its fictionalization are complex. The fictional parts of a true occurrence, such as a character’s formation, sometimes have a large influence on people’s imaginings. Such blending is similar to the cases of the above story/plot, time/era, and material/topic, which reveal kabuki dramaturgy based on a complicated mix (naimaze in the broad sense) of the original ingredients. Audiences who experience kabuki during a certain period may be aware of various naimaze techniques. Audiences see a fixed number of works involving the above diverse elements played out on stage. Although audiences only view stage performances, we can imagine that these real plays have respective scenarios (which in the book of a scenario is called daichō, a kind of hidden existence before a scene is actually played on stage). This is the aspect of [n] Production of Scenarios (“Daichō”). The production of authentic performance scenarios had a very simple format in the early days of kabuki, which only contained a single structure with a few scenes. However, kabuki works gradually came to have large-scale, complicated structures. The production mechanisms also became more systematic, thanks to excellent authors, especially from ningyō jōruri, such as Chikamatsu Monzaemon (1653–1725) and Ki-no Kaion (1663–1742). As a result, a new type of scenario production style was invented, through which several authors created a scenario by working jointly. Moreover, as the above naimaze techniques came to be utilized, kabuki daichōs became long and intricate. Currently, one can buy scenario books for staging kabuki plays at the shop in Kokutitsu Gekijō (the National Theater in Tokyo). This audience in this story bought a pamphlet called sujigaki or banzuke at a shop. The book is not unlike the above scenario books, and includes explanations of things such as today’s kabuki works, performances, actors, places, historical periods, and in-depth critical articles. Through the tradition of kabuki, [o] Text for stage performances became abundant. Many people buy sujigaki instead of pure scenario books in the theater. Unlike reading a novel, witnessing a drama on stage implies, in a primary sense, perceiving humans’ physical actions in various settings and decors. A drama on stage is not a representation solely based on the enumeration of words, but is also a collection of bodily actions, and sometimes the stage itself accompanying these actions. At the same time, kabuki comes from a tradition of essays originating in the yakusha hyōbanki of the Edo period, a series of books that comment on kabuki actors. The style’s existence has relied on the fact that an audience views a kabuki play on stage in a theater. This indicates that the texts of kabuki consist of individual stage performances as a collection of bodily actions by actors and stage settings. This custom has not yet been corrupted by the present-day kabuki scene. The kabuki tradition of text production has developed a different style of criticism, such as theoretical and methodological approaches that go beyond each actor toward kabuki narratives/stories. Through scenarios/stories that audiences view at a theater as several kabuki works, the author reveals diverse elements for kabuki narratives. A vital concept in this story is “multiple narrative structure” in the sense that an element has relationships with invisible existences on stage. For example, a narrative character that the audience sees on stage synthesizes several elements in multiple methods such as actors, or a person in the real world. Moreover, a kabuki work that the audience is currently and truly observing has diverse relationships with the preceding works, as well as works of the future. In particular, a kabuki work appears, so to speak, as a bundle in the present time of myriad multiple relationships.

Internal and External Narrative Generation The first topic for discussing internal and external narrative generation is the narrative generation-reception process and the narrative production-consumption process in a multiple narrative structures model. The narrative generation-reception process has individual humans as its units, whereby a novelist, using 151

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

thought, narrows down a plan and expresses it in words and a reader interprets it. However, observing the process more realistically, in the case of a novel (a book), various tasks are involved and performed by individuals other than the author, such as the publishing company and specific people (e.g., editors) who structure the book; they insinuate themselves into the actual process. When a certain narrative goes into social circulation, diverse elements penetrate it so that it cannot be fully expressed by the generationreception model at the individual level. Many people besides the author intervene and are engaged in this generation process; in totality, they make up what could be called the author/collective author. The aim of the multiple narrative structures model is to develop a unified understanding of the narrative generation-reception process as an individual one and the narrative production-consumption process as a social one. The functions that connect the two, and the subject responsible, comprise the work content for narrative generation, aiming only for abstract effects, regardless of whether the objective of narrative generation is an individual thing, a group, or an organizational thing; the functions are techniques and technology for narrative generation. The above multiple narrative structures model bridges narrative generation as a simple body to the problem of social development, specifically regarding how to develop and distribute socially generated narratives; the precondition is the element of a “generated narrative.” In the author’s framework of narrative generation study, the generated narrative simply means “a narrative generated by a narrative generation system” and the multiple narrative structures model is a synthesized framework for connecting it to the direction (problem) of how this narrative should be socially developed and spread. At the same time, the author extends the concept of the “generated narrative” to narrative content, which strengthens the characteristics of a work. This extended concept can include application systems associated with a narrative generation system, such as computer gaming systems. The author expands this flow to the direction of more literary/artistic narratives by the author himself. In the above discussion, the author calls the social development of narrative generation “external narrative generation” and the issue of the narrative as the work/content (by the author) “internal narrative generation.” The framework of GIS with INGS corresponds to the concrete architecture for realizing external narrative generation, while internal narrative generation will be incarnated through Watakushi Monogatari. These are the author’s future plans. The following section describes the generation process of Watakushi Monogagari and GIS with INGS based on TOWARD EXTERNAL OR SOCIAL PRACTICES OF NARRATIVE GENERATION (Chapter 4 (Ogata, 2020h) in (Ogata, 2020e). This section covers the developmental plan for both the external (social) and internal (personal and individual) directions. These directions are related to the bridge, from research to content creation. This bridging may influence the growth of narrative generation systems, including INGS. The entire narrative generation system (as technology) is a broad framework containing “the narrative generation system as a single system,” which is equivalent to INGS. It refers to a comprehensive, narrative production-consumption model or system, in contrast to the narrative generation-reception model or system of INGS. Their overlapping is a concrete form of the multiple narrative structures model. The external development (i.e., social expansion) of INGS corresponds to the narrative content produced by INGS, ranging from collaborative production forms with humans based on supporting methods, to societies as an external area of systems, which more firmly extend to the social distribution and consumption mechanisms. The area involved is positioned in the outermost zone of the multiple narrative structures model. A typical form is business; this can include new possibilities, such as a virtual geinō production company. In addition, trials of pure artistic and literary experiments also become possible.

152

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Originally, the author’s narrative generation system study was an extremely individual, personal attempt; hence, the development approaches to the internal, personal, and individual fields contrast with the above external directions. Simultaneously, internal development is conducted using INGS and the narrative production-consumption mechanism, including INGS. External and social development require more technological, mechanical, and formal discussion. In contrast, since internal and individual development require the author’s “personal” knowledge as content knowledge, we need to discuss themes related to narrative contents. In line with the above, the author introduces GIS, which contains INGS, to the larger system framework. It is possible to accomplish both the external (social) and internal (personal and individual) facets of INGS through GIS. Throughout the author’s study, GIS was originally planned as a system for realizing the social development of narrative generation via the multiple, complex uses of the narrative generation system as a single system, namely INGS. Simultaneously, GIS aims to incorporate the author’s individual and personal narrative content into the mechanism based on various geinō genres, especially kabuki and ningyō jōruri. Thus, the GIS system is a framework for expanding both externally and internally. Although it is different from INGS, GIS is still in the stage of fragmental implementation, and is explained according to the vision for execution. In the GIS architecture, first, the sender mechanism receives the external information, namely traces of information output by the receiver mechanism. This occurs via the “interpretation strategy” mechanism inside the sender mechanism. Next, the receiver makes an interpretation according to the interpretation framework to produce a narrative. However, the narrative does not mean that something is being created according to the receiver’s requirements and expectations. INGS is positioned in the center of the narrative generation by the sender mechanism in the GIS architecture. The sequence and reception of the narrative generation by the sender mechanism form the next geinō history. For processing, more special knowledge bases not included in INGS (i.e. the “work resource,” the “life resource,” and the “geinōjin resource”) are used. Although these resources are similar to the narrative content knowledge bases in INGS, these ones store GIS-specialized content knowledge. The geinōjin resource stores character or persona information on geinōjins. Through the life resource, the life histories of geinōjins are extracted from the geinōjin resource. The work resource refers to the processed knowledge of geinō works; compared to the content knowledge stored in the narrative content knowledge bases in INGS, this information is grounded in more specific narratives. In particular, the work resource aims to store narrative content knowledge that centers on kabuki and ningyō jōruri. Regarding the division between the narrative content knowledge bases in INGS and the work resource in GIS, geinō-related content knowledge is stored in the work resource. The receiver mechanism includes any mechanical reception subjects, in addition to the human receiver. These output any reception results based on the reception experiences of an individual work, a sequence of works, and the life history of a geinōjin. A geinō history is sequentially formed through the circular and repetitive interactions between the sender and receiver mechanisms. The content of a geinō history is classified into both a geinō event and the sequence, as well as a life event and the sequence. The former means each generated work while the sequence and the latter correspond to the actor’s flow of life in a geinōjin. Works here include scandals focused on an actor’s life. Regarding the division of the content knowledge of INGS and GIS, the knowledge of INGS is more general. Further, each dictionary corresponds to the most general knowledge group. GIS’s knowledge, as more special expertise, is divided into large two groups along the axis of generality/specialty. The most

153

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

unique group in GIS is the author’s private, individual knowledge, and the various knowledge resources are prepared as more general knowledge. Although the kabuki knowledge mechanisms are linked to INGS, in terms of relations with GIS, many elements are currently included in the work resource; these encompass “fact (event),” “work,” “genre,” “material,” “story or plot,” “place (stage),” “time (era),” “geinō style,” “text,” “production,” “direction,” and “staging.” These were stated in the previous part of this chapter. However, works in GIS do not necessarily indicate geinō or artistic literary works in the narrow sense. They include scandals about the life stories of geinōjins. Instead, they are tied to the geinōjin and life resources. On the other hand, only a minor element is associated with the geinō history and reception mechanism. Precisely speaking, the geinō history is related to comparatively many elements, such as the historical features of works, characters, actors, and staging. In the future, analyzing kabuki by focusing on the dynamic aspects of narrative production and consumption will be required to extend the knowledge resources of GIS. Thus, GIS modeling will help to design the study of kabuki more organically. The Watakushi Monogatari is created by using GIS or INGS in a special way. The systems consist of an independent, general organization, based on which it imports personal and private knowledge content. Potentially, the “I” (watakushi) that exists outside the systems can perceive the narratives created there as “fluid” and can express a narrative on another level: a Watakushi Monogatari that is fixed. Alternatively, the cycle or repetition of narrative generation by the GIS, equipped with INGS, may aggregate all fluid narratives formed at some point to produce a “Watakushi Monogatari that has been fixed.” GIS is a mechanism for setting up the “external I” (watakushi) of INGS within the system. It is possible to imagine other possibilities. In the end, the aim to build an organization that brings together multiple concrete mechanisms. Figure 5 shows, in abstract form, a narrative concretization mechanism based on the concepts of “fluid” and “fixed.” Documents on the personal and private levels are correlated with narrative content knowledge bases grounded in diverse levels of INGS, as well as GIS locations. What the author wants to focus on in this chapter is not the aspect of narrative generation mechanisms, but rather the personal and private knowledge content itself, on the foundation of which the mechanism for creating the Watakushi Monogatari is built. This knowledge content makes up the most personal and private knowledge content and knowledge bases used by the GIS with INGS. The GIS as a whole, including the INGS, continuously and repetitively generates/gives birth to fluid narratives through its interface. The fluid narratives are cyclically, repetitively, and continuously generated using the GIS’s troupe (including actors) and stage equipment. In this fluid narrative generation, the process can be identified as a work, where the receiver can observe myriad aspects of the process at any time using the interface. This expression can emerge in different forms. At present, this generation image is not especially distinguished from the general GIS narrative generation image. The work to reach a conclusion on this point remains ahead of us. The descriptive text that forms the knowledge content needed for the Watakushi Monogatari, as discussed above, is also an independent unit by itself, or several partitioned units of writing. This is the knowledge used to create narrative content through the narrative generation system, yet at the same time, it makes up text that appears like narrative content in various senses. As such, it is not necessary for the narrative content to be used exclusively inside the system, but it is possible to try to publish it in books and magazines, on paper and online, as multiple writings appropriately partitioned and ordered in some form or content. We can imagine that a series has the words The Narrative Generation of XXX in all of its titles (in this case, “narrative generation” simultaneously includes the meaning “anti-/nonnarrative generation” (Ogata & Kanai, 2010)). Although this is a connection to the “categorization of 154

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Figure 5. Image of Watakushi Monogatari generation/production based on “fluid” and “fixed” (Source: Figure 3, Chapter 4 (Ogata, 2020h) in Ogata (2020e))

personal texts” discussed in the next section, its titles make up a unit belonging to the category of criticism (essays), contrasting with the academic papers that have been at the center of the author’s writing so far. Moreover, the fluid form of the Watakushi Monogatari (operated in computers or online) and its fixed state (expressed in the forms of books, paper-based or electronic) make up another group called narrative works and content. To prepare for producing or creating Watakushi Monogatari in the future, Figure 6 summarizes the knowledge content necessary for Watakushi Monogatari, in particular a kind of knowledge resource forming the main materials of Watakushi Monogatari. Figure 6 presents the overall private and personal knowledge content for producing Watakushi Monogatari by using GIS with INGS. The “temporal divisions” signify the time needed to cover the materials prepared, while the “text” implies the types of text to be concretely collected or created and edited within that timeframe. The author considers different kinds of knowledge to create the Watakushi Monogatari. The first “temporal divisions” form the basic structure of private and personal knowledge content, i.e., texts correlated with four types of temporal division: • •



Present Continuous: The first temporal division began March 11, 2011 and will undoubtedly continue for the time being. Past: The next temporal division comprises the memories that demarcate “my/myself,” which the author defines as the period from November 25, 1970 to March 11, 2011. The first day is the day Yukio Mishima died; subjectively, that was when the author’s conscious and self-aware days of life began. Past Perfect: This is the third temporal division.

155

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Figure 6. Overview of knowledge content needed for Watakushi Monogatari (Source: Figure 4, Chapter 4 (Ogata, 2020h) in Ogata (2020e))



Historical: This is the fourth temporal division. This refers to a temporal division where “I/myself” do not exist, but the existence of those connected to “me” is indeterminate. There is a vast historical space that transcends the private, where time intermingles without distinction.

Next, in accordance with the abovementioned temporal divisions, the author constructs a chronological account of events deemed facts, as represented by the following list. The content of “personal texts” and “personal readings” should not be written anew but rather edited and organized: •

• •

156

Event Descriptions: A temporal, chronological list of past matters thought to have happened to the author (or which “I/myself” think have happened), as well as a somewhat detailed explanation thereof. Outside of that, there are descriptions of issues directly related to (but not directly experienced by) the author, forming the outer edge of the author’s experiences. Personal Chronologies: Seen from a bottom-up perspective, this refers to the chronological information of the individual-level temporal divisions (past perfect, past, present continuous), made up of the abovementioned aggregate of event descriptions as “personal chronologies.” Social and Historical Chronologies: These are general chronological descriptions that do not overlap with any account of “my own life” as the author, and can broadly be divided into three

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

areas: a chronology of Japan during and after the Meiji era, a chronology of Japan before that, and a chronology of other relevant regions that are not Japan, meaning the world. There are additional two categories: “personal texts” and “personal readings.” The “personal texts” is a chain of many statements written by the author in the past and present continuous. The list is a kind of discography but does not necessarily contain only presented or published statements. Instead, it is a more comprehensive, inclusive personal/private list that also encompasses unpublished content. Ideally, it is an aggregation of all texts written by the author to date, including texts that the author simply did not publish anywhere and texts that were not approved by the system (e.g., documents rejected, papers, research plans). It would be an enormous undertaking if the author wanted to include remnants of manuscript revisions, but it goes without saying that the author does not think to save them because they are excellent. Rather, the author saves them because they exist, doing nothing but saving them as a record of a man without qualities. In that sense, it might be more meaningful to save things that are not excellent. As such, these are private things, but they are also other things that are normally dissolved in statistics, and the I/other narrative without qualities includes such things as well. The author categorizes the author’s personal texts from 1970 to 2011 into the following four types: • • • • •

Published Academic Content: These includes papers and essays in academic journals (paperbased and electronic), as well as books. Unpublished Content Related to Academic Activities: Unpublished texts related to academic papers and other research activities. Materials from Conferences and Classes: Projection materials and handouts (e.g., resumes) related to talks and lessons. Content That Was Not Published: Content stored privately, without it having been presented or published anywhere. Lesson Records and Large Volumes of Notes from Self-study, Research Gatherings, and Other Occasions Written for Record or Contemplative Reasoning: These encompass unarranged records and texts.

The “personal readings” cover all or main content types that the author has read up until this point. However, this is figurative, because it actually includes everything that the author has heard and seen, yet if the author were to summarize it all and call it appreciated, that would give it a strange nuance, so the author says read instead. This is likely an expression of the author’s disposition to think of centering on reading. Specifically, this includes novels, music, pictures, movies, TV series, radio, newspapers, magazines, and plays. It also pertains to things the author is reading in present continuous. The author considers the following literary texts as the principal subjects of reading, investigation, analysis, and dissected use, as they are deeply tied to the content aspect of narrative generation systems, as well as the existential facet, since it is a list not formed by research necessity, but rather personal interest and curiosity. The categories mix; genres and authors are identified with other ones where that is not the case. Categories like kabuki, ningyō jōruri, and Yukio Mishima carry the meanings of subjects that should be treated or that the author wants to treat with exceptional care: •

Japanese Literature from Antiquity to the Present: Works: narratives, novels, poems, wakas (a genre of Japanese short songs), and haikus. 157

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

• • • • • • •

World Literature from Antiquity to the Present: Works in the world of narratives, novels, poems, and plays. Japanese Literature from Antiquity to the Present: Commentaries, critiques, and miscellaneous writings in Japan. Music: Western classical music and jazz were vital, yet, perhaps the truly important ones are the other genres. Video Works: Movies, TV dramas, and others. Kabuki and Ningyō Jōruri and Texts: Research and commentaries, in addition to the works. Works by Yukio Mishima: Related research and commentaries, in addition to his literary works. Other Miscellaneous Narratives: All narratives, such as manga (comics) and TV commercials.

In Chapter 4 (Ogata, 2010) in Ogata and Kanai (2010), the author presents a comprehensive list of books the author had read until a given point. Figure 7 is a conceptual image of a mechanism for generating Watakushi Monogatari that connects with other kinds of knowledge. A part of the above book list (1973–1974) is also described in this figure. In terms of private and personal stories for the author, the R&D of INGS and GIS is conducted to realize this kind of image. Figure 7. Conceptual image of Watakushi Monogatari generation

A MULTIPLE NARRATIVE GENERATION MECHANISM THROUGH TWO TYPES OF NARRATIVE GENERATION SYSTEMS: GIS WITH INGS This section describes GIS with INGS based on the post-narratology of narrative generation or a macro design of an integrated mechanism of GIS and INGS. The author presents the conceptual design of an

158

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

integrated mechanism of INGS and GIS, or GIS with (including) INGS. In particular, this section covers the macro-level design of the system, which the author has been calling GIS with INGS or GIS including INGS. Previously, although the consideration of the macro-level design has been addressed, the goal here is to link the GIS with INGS in order to connect to the level of design regarding actual program functions, structures, and data from simple conceptual designs.

Basic Elements and the Process of GIS GIS functions as a whole and includes INGS. Ogata and Amino (2002) examine partial implementation via a computer program of GIS architecture that they conceptually provide. This attempt is an example of GIS’s execution in the first section. The study sets a virtual geinōjin character, a woman called Kumiko, and simulates her geinō activities and life using Propp’s literary theory (developed by Ogata and Amino), which defines the typical structure of folktales based on the relationships of 31 basic event concepts called “functions.” These functions relate to the generated stories of geinō activities and life trajectories in the form of another level’s narratives and narrative generation, beyond each narrative content and life event. Narratives/stories to be created by this system form a group of individual narratives/stories. In this sense, the proposed system is a collective system of narratives/stories. Hence, this study implements the concept of GIS, realizing it whereby a narrative/story at the collective level is produced through the sequence of much smaller narratives/stories. This study also presents the idea that each small narrative/ story is generated continuously and sequentially under the generation strategy of grand narratives/stories in the above sense. Both the generation of each narrative/story and narrative/story as the sequence are equally included in narrative generation. In the author’s current system framework, each small narrative/story is created using INGS, and large narratives/stories, as the sequence, are engendered through GIS. Further, GIS involves comprehensive positioning, and GIS encompasses INGS. Although Ogata and Amino focus on the generation of sequential and large narratives/stories, and do not deal with each narrative generation, in this chapter, GIS with INGS (which in parallel and in a duplicate way generates both aspects) corresponds to an extended model that complements the above study. In this section, the author scrutinizes the development of the above study’s primitive presentation, moving toward an efficient, thorough approach. However, the author reveals the macro design level’s consideration before the system is implemented. Based on Figure 2, the main macro-level components in the GIS architecture are the geinōjin resource, the geinō work resource, and the life work resource. When a geinōjin character stored in the geinōjin resource is focused on, this geinōjin character produces individual geinō work, and the geinō work’s line forms the sequence of each geinō work using the geinō work resource. At the same time, the geinōjin character produces each life work, and the life work’s line functions according to each life work using the life work resource. •

The Geinōjin Resource and the Geinōjin Character: A GIS is like a geinō production agency or organization. Although the translation is close to “entertainment agency” in English, the author uses “geinō production” because the author’s theory is founded on the Japanese form of geinō, literature/narrative, and is inspired by its classical or traditional forms such as kabuki, ningyōjōruri, and no. GIS, as a geinō production agency, contains many geinōjin characters. GIS stores a geinōjin resource as an important component. As shown in the author’s analysis of kabuki, each geinōjin character is a multiple existence and includes the following sub-components:

159

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

The Aspect of the Geinōjin Character as an Actress/Actor: Each geinōjin character has unique property information. Each activity forms a geinō work, and the chain/sequence corresponds to the discography as a geinō work chain/sequence for the actress or actor. If the spatial information is temporal information, the sequence is temporal as well. The property information, as spatial information, also changes according to the temporal progression. ◦◦ The Aspect of the Geinōjin Character as an Ordinary Person: This aspect also has unique property information (spatial), and each life event and chronological (temporal) sequence is formed. ◦◦ The Geinōjin Character as a Character in a Geinō Work: Another multiple element in the geinōjin character is the facet as the dramatic character in a geinō work, which is a little different from the above two elements. Also, the property information, regarded as spatial information in the above elements, certainly gives a consistent feature whereby each geinōjin character has the potential to become a dramatic character. However, the geinōjin character, who plays diverse dramatic roles, needs to have new property information to play a different role each time. This type of property information can be interpreted as the element that forms part of the model of each dramatic character. Hence, we can divide the model of a dramatic character into such property information, and the part related to dramatic events is induced according to the property information. The Geinō Work Resource: A geinōjin character plays many geinō works/narratives through his/ her discography, particularly a geinō history. The information related to the works that the geinōjin character appears in (or which is meant for production) is built into the geinō work resource. The geinō work resource also contains information relevant to the dramatic characters. The Life Work Resource: Likewise, the life work resource stores various types of information related to the lives of geinōjin characters, which helps to build their life narratives/stories.

◦◦





In the following sub-sections, the author considers prospective concrete content.

The Geinōjin Resource and the Geinōjin Character As described above, the most basic components of GIS architecture are the geinōjin character, the geinō work resource, and the life work resource. They are the internal elements, a sequential narrative production mechanism. At the same time, they are related to INGS, which is an individual narrative generation mechanism and is part of GIS. Although the basic role of a geinōjin character in the primary elements of GIS is to appear in works as an actress/actor in produced narratives, the same geinōjin characters perform in produced life narratives/stories as other existences or ordinary persons. In the framework of GIS with INGS, “narrative authors” are simply positioned as INGS and GIS. As frequently stated, regarding the distinction, INGS is a mechanism for generating each narrative as a single body, and GIS is a mechanism for producing sequential narratives based on single narratives. In terms of a broader perspective, narratives in this case include narratives as life works, more than narratives as geinō works. At this point, although the Japanese term geinō is frequently translated as “entertainment” in English, there is a large gap between these two words. By geinō, in this chapter, the author does not mean mere content meant for entertainment. Instead, the author means the narrative, which encompasses a representation of the work through a one-time performance. INGS can create life events for each geinōjin character as a real person, and GIS incorporates the function to produce a corresponding life 160

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

narrative/story as a line of life events. Yet although the “authors” in INGS and GIS may need to have corresponding and individual characters, at present, both INGS and GIS have not personified figures as narrative authors. Regarding this problem, the author believes that a geinōjin character can also create a work as an author, and he/she can perform in the work created. In early kabuki, the leader (zagashira) of a troupe (ichiza) wrote kabuki works to be performed by the ichiza. This was ordinary. In such a case, the geinōjin character, as a kabuki writer, overlaps with INGS and GIS. A question to be asked is, “Who is the author or writer of a life narrative work?” An adequate answer is to consider that a life narrative work is written by a geinōjin character, who proceeds along a lifeline. Hence, in creating a life narrative work, the corresponding geinōjin character overlaps with INGS and GIS. In the above framework, each geinōjin character, in parallel, executes and repeats two types of practices and creations of geinō works and life works to make the prospective geinō history/story and life history/story. By focusing on a geinōjin character, he/she continues to participate in practices such as performances, and appears on stage; at the same time, he/she performs his/her life events. His/her life constitutes both a geinō history/story and a life history/story. These histories/stories are performed with a holistic viewpoint in mind. Furthermore, when we perceive the entire geinō production agency or a geinō group/troupe (which synthesizes and fuses the geinō history and life history of each geinōjin character included in these agencies and troupes), we also establish a grand narrative/story from a broader standpoint. Now, what information is needed for the property information of a geinōjin character? As mentioned above, since a geinōjin character contains three different aspects (including characters as an actress/actor, as a real existence, and as a dramatic character), it is necessary to define the corresponding types of property information for the three elements. Because the property information of a dramatic character is the information to be incorporated into the next geinō work resource, the author will describe it in the next section. As for the remaining two elements, as an actress/actor, the geinōjin character needs pertinent property information; this includes his/her name and the basic strategies/concepts of his/her career. The geinōjin character, as a real person, has a name, sexuality, birthplace, career, interests, hobbies, and a unique personality. In many cases, these two aspects of the geinōjin character cannot be seen as completely divided. Of course, there may be a type of geinōjin character who intentionally, consciously, and clearly differentiates the character as a real person from the character as a geinōjin. However, two kinds of property information for these characters are mutually mixed in many situations. In addition, characters, as real people, have a vagueness about them, and it is challenging to classify them under the above life work resource. One idea is to define characters as real persons in the life work resource. As described above, three types of property information (for the three categories of a geinōjin character) leave room for consideration to develop a clearer definition. Figure 8 depicts the respective types of property information and problems that arise when categorizing the three multiple elements of a geinōjin character.

The Geinō Work Resource By geinō work resource, the author means the resource necessary to create concrete geinō works/content in which geinōjin characters, as actresses/actors, perform using various styles. The geinō work resource is an essential information resource for GIS in the sense that it is most directly linked to GIS and INGS as a narrative generation system. 161

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Figure 8. Categorization of a geinōjin character and related problems

If we consider the geinō work resource in a broader sense, it is pertinent to the narrative content knowledge base in INGS. The narrative content knowledge base contains diverse techniques for utilizing the narrative generation of the narrative generation mechanism in INGS and mostly stores methods for story generation. The story content knowledge base covers know-how and strategies for the entire structure of a story for the partial structure of a story. Given the current version of the narrative content knowledge base (including the story content knowledge base), INGS is tentative systematization in progress. The author refers to different narratological studies and literary theories and analyzes narrative/story works, encompassing numerous genres (e.g., novels, folktales, theater, ads, manga [comics]). In this section, the author discusses the differences and relationship between the narrative content knowledge base in INGS and the geinō work resource in GIS. The information stored in the geinō work resource is concrete, comprehensive information on narratives, compared to the information in the narrative content knowledge base. By “concrete” information, the author indicates the following: Narrative knowledge in the narrative content knowledge base is abstract, universal knowledge, effective for all narratives. In contrast, narrative knowledge for the geinō work resource is more specific, meant for particular narrative genres and works. By “comprehensive,” the author implies a more delicate meaning. The narrative knowledge and techniques in the narrative content knowledge base can be used generally, regardless of specific narrative genres. In this sense, the narrative content knowledge base has a comprehensive characteristic. The comprehensiveness is the result. In contrast, the comprehensiveness of the geinō work resource is intended to apply to diverse genres of narratives. The narrative content knowledge base is comprehensive in a bottom-up way, while the geinō work resource is comprehensive in a top-down manner.

162

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

The concreteness and comprehensiveness in the above sense of the geinō work resource rely on the characteristic of intending to create discographies of geinōjin characters. They are related to the point of how a geinōjin character, as an actress/actor, performs in each geinō work in any style. For example, the discography of an actual geinōjin character includes different records regarding performances and other practices. He/she may perform as a dramatic character in a TV drama, as a talented performer on a TV variety Shōwas a singer in a popular song program, as a character in a TV ad, as a narrator in a documentary, as a radio performer in an animated TV program, as a dramatic character in a movie, and as a dramatic character in a play. Since it is difficult for the current GIS with INGS system(s) to concretize all of these narrative representations, they should be realized in simple and alternative forms. For instance, it is possible to substitute these narrative representations by generating narrative skeletons such as scenarios, plots, or stories of works in which geinōjin characters perform. In any case, the geinō work resource must be able to deal with myriad narrative genres. The following types of scenarios, plots, and stories are included: • • • • • • • • •

Folktale-like narrative scenarios, plots, or stories Ad-like narrative scenarios, plots, or stories Game-like narrative scenarios, plots, or stories Drama-like narrative scenarios, plots, or stories Kabuki-like narrative scenarios, plots, or stories Manga-like narrative scenarios, plots, or stories Movie-like narrative scenarios, plots, or stories TV drama-like narrative scenarios, plots, or stories Conte-like narrative scenarios, plots, or stories

These narratives can be applied to other genres beyond the ones mentioned here. In this case, names according to narrative content (instead of names of narrative genres) are also needed; for instance, narrative genres of family trouble and double suicide, which are popular and frequently appear in kabuki and ningyō-jōoruri. This kind of narrative is closely associated with narrative content, and is included in the narrative resource for “narrative types commonly used in many narrative genres.” Let us return to the discussion of narrative scenarios according to specific genres. If an ad-like narrative scenario is an original narrative work, its derivative narrative works may also exist. The derivative narrative works are distributed through their transformation into different media from the original medium. Geinōjin characters can also perform in derivative narratives. • • • • •

Deviated narrative works from folktale-like narrative scenarios: Ex.) A folktale is narrated on a radio or TV drama. Deviated narrative works from ad-like narrative scenarios: Ex.) A magazine ad is adapted to a novel or a book. Deviated narrative works from game-like narrative scenarios: Ex.) A game is adapted to a novel or a book. Deviated narrative works from kabuki-like narrative scenarios: Ex.) A kabuki work is adapted to a ningyō jōruri work. Deviated narrative works from manga-like narrative scenarios: Ex.) A manga work is adapted to a game and a kabuki work. 163

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

• • •

Deviated narrative works from movie-like narrative scenarios: Ex.) A movie work is adapted to a TV drama and a game. Deviated narrative works from TV drama-like narrative scenarios: Ex.) A TV drama is adapted to a movie and manga works. Deviated narrative works from conte-like narrative scenarios: Ex.) A conte (short story) is adapted to a TV drama and film.

For example, a geinōjin character performs in TV dramas and movies with a “folktale-like scenario,” TV and Internet ads with an “ad-like scenario,” computer games and movies with a “game-like scenario,” staged plays and their cinematized works with a “kabuki-like scenario,” and adapted movies, TV dramas, and kabuki. An important term is “narrative media,” which enables us to realize various narrative genres. These media need to be contained in the knowledge base of the geinō work resource. At this point, let us organize the main types of narrative information to be encompassed in the geinō work resource. First, narrative genres exist. Chapter 2 (Ogata, 2020b) in Ogata (2020a) introduces the author’s “narrative genre system,” which divides narrative genres into five broad categories (listed below) and shows their sub-categories and examples, primarily based on Japanese literary genres: 1. Narratives as Works in the Narrow Sense or Works in the Narrow Sense in Which Narratives Appear (or Narratives Are Included) 2. Narratives as Works in the Broad Sense or Works in the Broad Sense in Which Narratives Appear (or Narratives Are Included) 3. Narratives as Social and Emergent Phenomena or Social and Emergent Phenomena in Which Narratives Appear (or Narratives Are Included) 4. Narratives That Invade Real Phenomena or Real Phenomena in Which Narratives Appear (or Narratives Are Included) 5. Narratives as Human Physiological and Psychological Natural Phenomena or Human Physiological and Psychological Natural Phenomena in Which Narratives Appear (or Narratives Are Included) This narrative genre system is organized as a collection; it contains narrative genres in the broad sense, such as mental narratives (e.g., dreams) and social narratives (e.g., festivals, real events), in addition to ordinary narratives (e.g., literary and narrative works). Thus, for example, geinōjin characters can also perform the narration of a real event (news story). Next, the geinō work resource covers the concept of “narrative media,” which is divided into several levels: media as devices or machines (e.g., TV, radio); media as more direct representation (e.g., sound, characters), and media as social organizations or agencies (e.g., TV stations). The geinō work resource includes the concept of “narrative type.” Compared to the above genres and media, the narrative type is the most substantial kind of content information for narrative works. The information is most directly tied to narrative plots/stories. For example, there are events and their sequences (the author employs a temporal event sequence story and an event sequence for a narrative development plot), dramatic characters, objects, places (spaces), and times. In order to analyze this type of narrative information, the author begins each topic in the first phase to categorize the results into several groups, in which methods in a group can be used for a common goal from the angle of crossgenres or trans-genres. For instance, although a narrative type of kabuki, “family troubles” (oie sodo

164

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

mono), emerged through the examination of kabuki, generalization beyond narrative genres enables us to apply the method to diverse narrative genres other than kabuki. Regarding a concrete narrative work/content performed by a geinōjin character, first, there is a story/plot in which specific characters, objects, spaces or places, and times appear. The story/plot and its elements are defined using the narrative content knowledge bases and related narrative techniques. Furthermore, to realize a narrative in a specific geinō genre, the above abstract elements of narrative are concretized by combining certain media to be performed by geinōjin characters. By blending diverse narrative elements (e.g., different genres’ works and patterns, various dramatic characters in different works), myriad places and times (ages) become connected to present the concept of naimaze, which is a traditional method for creating a new work in kabuki. Strictly speaking, this concept means a specialized dramaturgy, in particular methods for creating new works by blending different sekai (worlds), produced by several excellent kabuki writers or creators in an age (from the middle to the latter half of the Edo period), especially Sakurada Jisuke I and Tsuruya Nanboku IV. This methodological concept of narrative generation will be generalized into the creative direction of a principle of narrative generation, which contains numerous concrete techniques for producing new works from the amalgamation of two or more works and their many components. The author aims to examine different problems of narrative generation, including narrative deviation and transformation among many genres and media, from the perspective of naimaze as a generalized methodological concept. The author introduces naimaze in The Narrative Generation of Kabuki in the BACKGROUND section of this chapter.

The Life Work Resource and Geinōjin Characters Each geinōjin character lives in an actual world, and the continuous path of events shapes the geinōjin character’s “life course” (also called “life story” or “life history”). Linking with INGS, GIS can produce the life stories of geinōjin characters, in addition to narratives as geinō works. Hence, life stories are part of the narratives that are generated regarding each geinōjin character. GIS stores the life work resource with the geinō work resource. The life work resource contains the material information for creating the life stories of geinōjin characters, mainly the information for constructing their life events. People commonly have typical life events, shown in the following list: • • • • • • • •

Life events around birth: Father, mother, ancestor, place, time or era, etc. Life events around school: Entrance, graduation, registered absences, leaving school, etc. Life events around love: First love, lost love, marriage, immorality, etc. Life events around friendship: Friendship among members of the same sex, the opposite sex, etc. Life events around jobs: Part-time jobs, employment, job changes, social failures, social successes, etc. Life events around family: Relationships with/among mothers and fathers, brothers and sisters, husbands and wives (marriage, divorce), children, etc. Life events around hobbies: Sports, reading, music, etc. Life events around geinō: Debut, work, job, failures, successes, etc. (This is the boundary area with the geinō work resource.)

These categories and examples cover only some of the life events and are not sufficiently systematized, but this list provides tentative content according to particular themes such as birth, schooling, and 165

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

love. Seen from a spatial angle, life events form a hierarchical structure that includes several high-level event categories, as well as low-level event categories belonging to the higher categories. The low-level categories can be subdivided into smaller units. Smaller categories encompass many events. At any hierarchical level, events move into the range of the narrative content knowledge base in INGS from the range of the life work resource, because events after the level of an elaboration can be interpreted as event development, to be used generally, and are not contained in life events. Since the information to be stored in the life work resource in GIS corresponds to the comparatively large categories for forming people’s lives, knowledge for detailed, concrete development is stored in the narrative content knowledge base, especially the story content knowledge base. However, the above description is still a general direction, and the isolation of both areas will be considered in the future. Regarding the actual development of each geinōjin character’s life events, a portion of the lower event categories in the life event structure is activated in each timing to construct a geinō sequence of events in an efficient form. Each event sequence is concretized through a specific temporal range, including eras and seasons, a certain space (place), participating people, and the relationships among them. A geinōjin character’s life events indicate the relationships among geinō events, which are performed according to his/her name as an actress/actor in order to produce his/her life history/story, which is a relatively individual event sequence. A geinōjin character’s entire life story is a narrative, just as the entire geinō discography is a narrative/story. Additionally, since each narrative in a geinō discography is respectively a narrative, each event in a life story (e.g., birth, lost love, job success or failure) is also a narrative. As mentioned above, these events should be included in a narrative as a single body to be generated by INGS. Thus, INGS contains the mechanism for generating life events/narratives from the viewpoint of the linked model with GIS. There may be various design methods and ideas about the implementation of the life narrative generation program. The typical directions are the top-down and bottom-up strategies. For the former, a pre-determined, basic, broad scenario of a life narrative is created at the first starting point of the simulation of an entire life narrative, or the large part, by a geinōjin character. Next, small events are produced according to the broad scenario. The bottom-up strategy did not previously have such a broad scenario. The strategy engenders the life events of a geinōjin character in his/her relationships with other geinō narratives’ generation, and the entire life event sequence is continuously and sequentially created through the above repetition. The most useful approach is the hybrid method of the top-down and bottom-up strategies. For example, in the first stage, a basic and broad scenario for the extension of a life story, many of the parts remain in a non-detailed state. In the next step, the system completes in detail through the simulation of concretization and elaboration in engagement with context and situations. In this process, a planned or predicted broad life story sometimes shifts to a different life story. Both the top-down and bottom-up strategies for generating life narratives can be applied to create geinō work sequences using the geinō work resource. Lastly, Watakushi Monogatari, described in THE CONCEPT OF WATAKUSHI MONOGATARI AND DESIGNING INTERNAL AND EXTERNAL NARRATIVE GENERATION METHODS TO CREATE AND DISTRIBUTE NARRATIVE CONTENT, is a vision of the narrative(s) to be created based on the Watakushi Character, which was discussed in Incorporating the Watakushi Character. Regarding the material information for creating Watakushi Monogatari, the author partly systematizes the life information based on “I” as the author. This was stated in Internal and External Narrative Generation in the BACKGROUND of this chapter. The life work resource in this section is an information unit for storing the common knowledge resource for geinōjin characters. The systematization 166

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

of life narratives, as the material information for Watakushi Monogatari, is the information for more a concrete and individual “I” (Watakushi Character). In the future, the systematization of life narratives for Watakushi Monogatari and the Watakushi Character will indicate how to construct life narratives as more general knowledge.

GIS and INGS in the Integrated Mechanism First, the author summarizes the above description. The construction of GIS requires establishing geinōjin characters and providing the corresponding property information for each geinōjin character. The property information is divided into data for an actress/actor and for an ordinary human. The former and latter pieces of information are respectively related to the discography of the geinō work performance (geinō history) and the life narrative. Moreover, the above discographies (a discography is a type of narrative) are tied to the geinō work resource, and the life narratives are linked to the life (work) resource. These narratives are collective in the sense that many individual narratives are gathered. GIS can produce an entire narrative, which is a collection of many individual narratives, while INGS generates each narrative included in the narrative as a collection. At the same time, the division of roles is relevant to the characteristics of the knowledge bases and resources for producing narratives. In particular, the geinō work and life work resources of GIS deal with a more macro-level narrative, such as entire narrative structures in a specific narrative genre, and the framework for the development of important events in lives. The narrative content knowledge bases (especially the story content knowledge base) deal with knowledge for developing more detailed, general events. The author discusses the division of roles of GIS and INGS in the productions of geinō and life narratives. As described above, although the knowledge contained in INGS is only general knowledge for narrative generation, GIS’s knowledge is more concrete, individual knowledge based on each narrative work. For example, the story content knowledge base in INGS only contains fragmented narratives or story knowledge. The author asserts that the image of knowledge resources to be contained in GIS can become clear by applying Sekai Kōmoku in kabuki to it. This means that Sekai Kōmoku, or the knowledge structures seen in Sekai Kōmoku, can be beneficially applied to the geinō work resource in GIS. A detailed description of Sekai Kōmoku is provided in the following section, The Survey and Analysis of Sekai Kōmoku; it is a professional knowledge system that was used for writing kabuki scenarios during part of the Edo period. Sekai Kōmoku categorizes the typical patterns of kabuki narratives, and, for all sekai (worlds), describes or lists main dramatic characters; the original books, including novels and histories, of the sekai; and ninghyo-joruri works written based on the sekai. Kabuki writers from that time period referred to diverse sekai mentioned in Sekai Kōmoku, and added new shukō to a sekai or several sekai(s) to create a new kabuki scenario. Although Sekai Kōmoku does not directly address narrative plots and places, as the kabuki writers in that age completely knew the details, they selected one or more sekai(s) and determined basic characters, plots, spaces, and times to add shukō or twists to the sekai(s) for differentiating them from previous and existing narratives. Thus, they produced fresh narrative scenarios. The author believes that GIS, especially the geinō work resource, can be defined and stored in a similar form with the narrative knowledge structures of Sekai Kōmoku. For instance, regarding a narrative sekai as a kind of retrieval item, related narrative works, characters, and others are organically represented. In the real Sekai Kōmoku, the readers arrange and edit the content as the names of titles and characters. Given the possibility of a new version of Sekai Kōmoku in GIS, as the related information stored in the Sekai Kōmoku is computationally reorganized, the users decide on narrative 167

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

macro-level structures, forms, and content linked to GIS and INGS, complete with narrative scenarios using more detailed information. As stated above, in an actual narrative generation process based on the macro-micro incorporation of INGS and GIS, the geinō work and life work resources are provided by GIS to produce relative macro narratives. Therefore, GIS can create comparatively macro narrative structures and content, in addition to sequential narrative generation, including each narrative generation.

Incorporating the Watakushi Character Earlier, the author deals with the general structure of GIS with INGS. This section states the GIS with INGS from the viewpoint of relationships with the author’s Watakushi Monogatari. Watakushi Monogatari is also created through GIS with INGS. As mentioned above, geinōjin characters in GIS basically perform in produced narratives. For example, they perform various geinō narrative works. Furthermore, they can create narrative works as authors or writers. Although the concept of “author” is not clearly articulated in GIS with INGS, the geinōjin characters can be narrative authors or writers who create narrative works by themselves. Moreover, GIS with INGS, which focuses on Watakushi Monogatari, is necessary to prepare one or more geinōjin character(s) who can be author(s) or writer(s). By Watakushi Monogatari, the author intends to produce a unique narrative work. In the plan of Watakushi Monogatari, the author designs the form in which this real “I” (watakushi) exists with one or more geinōjin character(s) as virtual characters, or enters into the geinōjin characters. Although GIS is basically composed of a collection of geinōjin characters as virtual characters, this real “I” (watakushi) invades the geinōjin characters. In particular, the geinōjin character, who has half a virtual existence and half a real “I” (watakushi) existence, produces narrative works tentatively called Watakushi Monogatari, and socially distributes the narrative works through GIS. Hence, the process of the geinōjin character, including “I” (watakushi), encompasses the act of creating Watakushi Monogatari, in addition to the act of distribution. Simultaneously, regarding the production, creation, and distribution of Watakushi Monogatari, the production and distribution of other narratives by GIS are continuously facilitated through other authors, narrators, and actresses/actors. The acts of creation and distribution of Watakushi Monogatari by “I” (watakushi), as a part of geinōjin characters, are embedded in diverse narrative acts in GIS. In addition, this “I” (watakushi) and Watakushi Monogatari are not completely independent things separated from narrative generation acts by other subjects. They are existences that are influenced from the external elements and situations and can influence external things. Figure 9 shows an overview of the creation process of GIS with INGS and Watakushi Monogatari. It is vital to discuss the concrete, detailed form of the concept in this figure. For example, when the geinōjin character mixed with “I” (watakushi) creates Watakushi Monogatari, the geinōjin character creates the Watakushi Monogatari by using or editing various types of knowledge, which are analyzed in Toward Internal Narrative Generation: The Creation of Watakushi Monogatari in this chapter. At the same time, the knowledge and resources stored in GIS and INGS are also used in the creation. They can be employed from the angle of automatic narrative generation. Furthermore, Watakushi Monogatari and the part created once can be restructured using the knowledge and resources in GIS and INGS. The following tentative list contains several methods for the creation (production) of Watakushi Monogatari based on “I” (watakushi), a geinōjin character including an “I” (watakushi) or a geinōjin character as an “I” (watakushi), and GIS with INGS: 168

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Figure 9. The concept of the creation process of GIS with INGS and Watakushi Monogatari

• •



The geinōjin character as an “I” (watakushi) thinks up the plots and stories of a Watakushi Monogatari and the parts through GIS and INGS, in addition to only the living “I” (watakushi). “I” (watakushi) can develop the plots or parts for Watakushi Monogatari. Simultaneously, this “I” (watakushi) refers to the plots or parts that the GIS with INGS system generates. Further, “I” (watakushi) can expand a provisionally made plot through GIS with INGS and translate a onetime narrative created by the system. Like these, “I” (watakushi) can create the parts of Watakushi Monogatari through various interactions between the system of GIS with INGS. More micro parts for narrative development and expansion, such as event development, explanations, and descriptions of constituent elements of a story and event, and narrative discourse techniques at the level of the transformation of event development, can use GIS with INGS. This generates the possibilities of narrative micro development, other than the possibilities thought by “I” (watakushi) or beyond them, in addition to the elements of macro narrative expansion.

“I” (watakushi) can create Watakushi Monogatari considering the above possibilities. The above description covers formal methods regarding the creation of Watakushi Monogatari by “I” (watakushi). Regarding formal methods of the next distribution process of Watakushi Monogatari, the author considers the following possibilities: •

“I” (watakushi) makes it possible to send Watakushi Monogatari to the distribution process via the created parts. These parts can be like initial “seeds” for the final forms, which do not comprise a fixed narrative, and can be transformed in continuous phases.

169

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents



Narratives, as initial “seeds,” can be modified after the distribution process. In this case, the distribution process is part of a production process. (This type of creation is compared to the popular creation and distribution of novels and stories in modern Japanese literature, though it is doubtful whether many writers clearly understand that this kind of creation-distribution is a unified creation process.) In any case, first, a part of a novel is published in a literary magazine and, continuously, the novel is published in the form of a book. During transformation, a novel, which is serially published in the medium of a magazine, is altered through corrections, revisions, and further reorganization. Using this comparison, part of Watakushi Monogatari—which was first published in a “magazine”—changes the structures and content of an entire Watakushi Monogatari and its parts before the Watakushi Monogatari is finished. A difference is whether the objects to be modified comprise the entirety or parts of a novel/narrative. However, if we regard a part in the case of a whole, the same structure is recursively repeated. The difference in phenomenological form (in the case of Watakushi Monogatari) strengthens the characteristic whereby the narrative creation process enters the narrative distribution process.

The Overall Behavior of GIS, Including a Watakushi Character with INGS Since the Watakushi Character here is a geinōjin character in GIS, it functions like other geinōjin characters or as one of the many geinōjin characters. The operation means that a Watakushi Character cyclically creates and performs narrative works. However, the genres or narrative kinds it creates are limited. In particular, only Watakushi Monogatari itself or the content for preparing Watakushi Monogatari are the narrative genres/types produced by the Watakushi Character. Furthermore, the Watakushi Character is not a pure or completed virtual AI character; half of it is occupied by the “I” (watakushi) in the real world. Hence, Watakushi Monogatari is completed through the collaborative act of the Watakushi Character and the “I” (watakushi) in reality. Although an overview of the concrete repetition and cyclical process is mentioned in the following list, a detailed discussion will be given in THE CONCEPT OF WATAKUSHI MONOGATARI AND DESIGNING INTERNAL AND EXTERNAL NARRATIVE GENERATION METHODS TO CREATE AND DISTRIBUTE NARRATIVE CONTENT: • •



170

First, there is a situation in which many geinōjin characters cyclically continue to produce and perform diverse narratives. In other words, it is a fluid narrative situation. In this fluid situation of narrative generation, the Watakushi Character intermittently creates fixed, comparatively short narrative works through concerted action with the real “I” (watakushi) included in the Watakushi Character. The author tentatively calls the narrative collection “fragmented narratives.” Moreover, based on the “fragmental narratives,” the Watakushi Character and the real “I” (watakushi), who is involved in the Watakushi Character, continue the creative act toward Watakushi Monogatari, which is a more systematic, fixed narrative. The author calls the systematic, fixed narrative Watakushi Monogatari. Alternatively, it could be called “the narrative of fluidity and fixation.” However, the Watakushi Monogatari in the above meaning may be the Watakushi Monogatari in the narrow sense. In this case, the total narrative, combined the “fragmented narratives” and “the narrative of fluidity and fixation,” may be called Watakushi Monogatari.

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Abstractly speaking, the fixed narratives, which are sometimes produced in the forever fluid situation, evolve into fragmented narratives. Further, the narrative of fluidity and fixation—as the fixed narrative that integrates, edits, and adapts fragmented narratives—is produced. At the theoretical level, the narrative of fluidity and fixation is also fluidized toward other circulative, repetitive narrative situations.

NARRATOLOGY OF KABUKI: ORIENTED TOWARD NARRATIVE GENERATION The author has frequently argued that in general, kabuki is discussed with a focus on the stage, and moreover, on so-called “patterns,” while kabuki retains an extremely large volume of “language” since the Edo period, and this diversity of language in kabuki forms one factor that has supported the resilience and continuity of kabuki. Therefore, kabuki should be studied through the stage and also through language. By kabuki as language, the author means that kabuki has produced many types of linguistic documents, including essays, reports, and critiques, and existed with them, though an individual performance on the stage is regarded as the most important element in kabuki. Chapter 2 (Ogata, 2020g) in Ogata (2020e) organizes and classifies the types of “language” found in kabuki as follows: •

• •



Research ◦◦ General and introductory studies ◦◦ Studies focusing on actors ◦◦ Studies focusing on authors ◦◦ Studies focusing on works or plays ◦◦ Studies focusing on music ◦◦ Studies focusing on theaters and environments ◦◦ Folkloric and geinō studies ◦◦ Historical studies ◦◦ Sociological studies ◦◦ Literary, artistic, aesthetic, and critical studies ◦◦ Fringe studies ◦◦ Informational and cognitive studies ◦◦ Dictionaries and encyclopedias Dramatic Critics ◦◦ Dramatic critics (from early modern to modern) [synchronic] ◦◦ Dramatic critics (contemporary) [diachronic] Reports ◦◦ Reports (a) (pamphlets in theaters) ◦◦ Reports (b) (other documents including picture collections) ◦◦ Magazines ◦◦ Actor’s talks Works ◦◦ Scenarios (daichō) in kabuki and related genres including no and ningyō jōruri

This section introduces and discusses topics relating to “Literary, artistic, aesthetic, and critical studies” from previous Research, and “Dramatic critics (from early modern to modern) [synchronic]” 171

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

and “Dramatic critics (contemporary) [diachronic]” from Dramatic Critics. In the aforementioned classification, “Literary, artistic, aesthetic, and critical studies” from Research and Dramatic Critics (“Dramatic critics (from early modern to modern) [synchronic]” and “Dramatic critics (contemporary) [diachronic]”) are similar; however, this indicates that in the case of kabuki, some critiques may be more like comparative studies, while others may be similar to criticism. First, the author mentions Yukio Hattori and Tamotsu Watanabe, two present-day kabuki researchers who have discussed numerous topics of thought and philosophy concerning kabuki, and present their statements from perspectives of thought and philosophy. Then, the author focuses on a work of kabuki theory (Sakusha Shikihō Kezairoku, which means the rules and principles that kabuki writers and directors create and produce kabuki’s scenarios and stage performances) and a type of instructional work or knowledge base (the aforementioned Sekai Kōmoku, which means the outline and details of the sekais (worlds) that are used for writing kabuki scnerios) written in the Edo period, and considers how each is linked to the author’s studies of narrative generation in kabuki in an essential sense. It should be noted that Sekai Kōmoku is neither a research work nor a critical work, but a type of instructional text or guide for creating kabuki. The aforementioned classification does not have an item suited to this type of document and therefore requires expansion in the future.

The Thought and Philosophy of Two Present-Day Kabuki Researchers While a large volume of kabuki-related “language” accumulated through the Edo period, academic study of kabuki flourished particularly in the Meiji era. Kabuki has had a rich tradition of critical and commentary texts since the Edo period, and thus, the modern kabuki world has many critical researchers or research-based critics who practice texts of a type that combine scholarly methods and critical methods. Yukio Hattori and Tamotsu Watanabe, discussed here, are a new type of scholars we will call researchers-cum-critics. If pressed, the author would say that Hattori is a writer who leans towards being a researcher and Watanabe is a writer who leans towards being a critic, and that both authors are exceptionally skilled researchers and critics. The author of this chapter is not prepared to systematically discuss an overview of their research/criticism, so the author will merely consider one book by each and present their outlines.

Yukio Hattori’s The Structure of Kabuki Yukio Hattori is a noted kabuki researcher who worked at Kokuritsu Gekijō (the National Theater of Japan) and produced many works. His early work, The Structure of Kabuki (Hattori, 1970), is an important text in the literature on kabuki research. It was published on November 25, 1970—the very date on which author Yukio Mishima died by suicide at the Self-Defense Force Base in Ichigaya, Tokyo after inciting force members to rise to action. Around this period, the climate concerning speech was heavily influenced by the wave of structuralism sweeping across the world of Japanese thought and philosophy. Hattori, who was still comparatively young, probably wrote this work while conscious of riding this tide of modern thought and modern philosophy. As previously described, structuralism also formed the broad tides of thought and philosophy in Japan, and many thinkers and researchers in the humanities and social fields were influenced by it to an extent. Examples include Masao Yamaguchi (1931–2013) and Chizuko Ueno (1948–). However, few researchers seriously pursued formal or structural analyses of narratives, as observed in structuralist narratology. Kazuhiko Komatsu (1947–), a folklorist and folktale researcher, 172

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

was a representative of this kind of researchers. Moreover, very few researchers went beyond this and assimilated their attempts and outcomes to incorporate them into systems of thought and philosophy. In this sense, structuralism and narratology were generally inclined to remain a passing trend in thought and philosophy and did not go beyond the level of scholarship for the sake of scholarship in Japan. To supplement the above, Masao Yamaguchi’s structuralism swept through the journalistic world of thought and philosophy in Japan at one time (around the 1970s and 1980s), directly influencing the novel-writing techniques of authors such as Kenzaburō Ōe (1935–) (for example, Dōjidai Game [The Game of Contemporaneity] (Ōe, 1979)). Although his structuralist theory of cultural anthropology, scattered with concepts such as “center vs. periphery” and “the trickster,” was of interest to the author of this chapter, it was not attractive enough to make the author comprehensively explore it. Around that time, Japan seemed to have developed a formula closely linking structuralism with Yamaguchi, however, this structuralism differs from what the author regards as structuralism. From the author’s perspective of structuralism, merely in the sense symbolized by formal, structured narratology (the author believed that Yamaguchi and others lacked this perspective of detailed formal, structural object analysis), Japanese structuralism as represented by Yamaguchi seemed extremely journalistic, particularly emphasizing only the obvious parts or the remarkable parts in the structure of the object. In other words, in the aurhor’s opinion, their analysis and examination seemed to be from an extremely “elevated” vantage point. They had seemed to suddenly leap to semantics before conducting a formal and structural analysis of the object at a “lower” level. However, this is merely the author’s impression, and a careful reexamination may produce countervailing evidence. This trend also extended to economics and other fields, leading to the appearance of “the trickster theory of business administration” and the like, which applied folklorist and narratological concepts mechanically or arbitrarily to business and organizational phenomena (for example, Teramoto and Nakanishi (1993)). If such a thing were called dynamic structuralism, what the author sought would be—to put it ironically and in extreme terms—static structuralism, and could even be deemed as something desirable or something that had already ended by the 1980s and 1990s. Hattori’s structuralist approach to kabuki can probably be considered to function on the foundation of Hattori’s own approach to his later works and research. However, it was possibly linked to the aforementioned trends concerning structuralism in general, and approaches such as this were rarely developed further within kabuki research. (The author has not yet investigated works and papers on structuralist kabuki theory, but may find related works.) For example, Toshio Kawatake’s (1924–2013) kabuki studies (Kawatake, 1989, 1992, 2003) are widely read, and while they are individually highly attractive, they focus on the “style” of kabuki (such as the beautiful and peculiar style); although they are organized, they are not systematic in an essential or structural sense). However, the author intends to focus on Hattori’s structuralist kabuki theory and draws attention to it again. Hattori’s structural approach to kabuki focuses on structure rather than style. In extreme terms, it is a study centered on extracting the elements of kabuki and identifying the relationships between them. In practicing this attempt, Hattori first noted the people involved in kabuki and then examined them beginning with those towards the outer circles—the financiers of the troupes—and moving in order through the actors to the creators. Hattori pays particular attention to the existence of “actors” (“yakushas” or “haiyūs” in Japanese) in kabuki. In other words, he places actors at the core of the structure of kabuki and clarifies that the concept is more comprehensive, overreaching, and fundamental than the concept of actors as we usually imagine them. In this way, Hattori considers actors to be a central element in the structure of kabuki. In fact, passages concerning actors account for the largest portion of this book. To wit, Hattori discusses 173

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

the creation and development of kabuki actors in historical terms, explains the structure of actors’ “art” and its making by invoking many concepts, such as “styles,” “roles,” “forms,” and “spirit,” and finally explains that the authorial function known as “kyōgen tsukuri” (lit. play-maker) was originally performed by the actors. This attention to actors does not mean that Hattori simply regarded actors as more important among the many elements; rather, it means that the actors were the most important in actual kabuki. However, Hattori also notes the “creators,” who separated and became independent from actors over time, and specifically investigates and analyzes structural characteristics of kabuki “works.” At the same time, the reason creators were traditionally positioned below actors in kabuki is revealed with statements about the roles of actors. Specifically, in early kabuki, the duties of creators could not be considered to occupy a clear position among the various structural elements of kabuki. It is possible that creators existed only as a function, and there were no actual creators in a pure sense. In fact, in early kabuki, a certain work was written by the stronger actors of a troupe (company) called “ichiza” in Japanese, who then performed it themselves. The task of creating and performing works was the responsibility of the actors. At last, creators came to exist separately from actors, but this characteristic in early kabuki also influenced the later development of kabuki and formed the tradition that still continues, wherein actors occupy a central, core position in kabuki. A mysterious phenomenon that exists even today is that kabuki has a designation called “playwright” (kyōgen sakusha). Inferring from this title, most people would normally think it refers to a person who creates kabuki plays. However, their actual job is not to write plays, but to perform general duties around the stage. Naturally, this does not mean that jobs like this are not important, and it is a highly important role in the process and setting of constructing an actual stage performance. However, this is not typically the task we would associate with a creator, nor is it the job of a producer. In modern terms, it could provisionally be considered similar to the role of an assistant director in film or television. However, in the world of kabuki in the Edo period, the aforementioned designation in which actors also functioned as creators and the duties of a creator were subsumed within those of actors was gradually followed by the formation and independence of the creator as a separate position. Moreover, this position came to be borne by a group of playwrights or kyōgen sakushas belonging within a so-called department known as the “play office” (kyōgen-beya) that was included in a kabuki troupe or ichiza. In these circumstances, it is true that the creator arose as an occupation that was independent of the actor in kabuki. Nevertheless, they were still positioned below the actor in charge of the troupe (ichiza), especially below one or more key actors in charge of other actors in the ichiza, produced individual plays, and oversaw all the work needed in the ichiza, such as formulating strategies for the flow of several plays. Later, after many kabuki works came to be written by external creators not belonging to a specific troupe in the Meiji era, playwrights (kyōgen tsukuris) and play offices (kyōgen-beyas) lost power and eventually became mere shadows of their former selves, as discussed. This work by Hattori views “actors” as the most important people involved in kabuki, but in the initial section, places the “financiers,” who have the capital capacity to actually produce kabuki, further outside them, thus revealing the true state of kabuki against the background of the economic infrastructure within society. Based on this, statements concerning actors account for the largest portion of this book, and “the structure of [their] ‘art’ and its making” is clarified. However, this book does not simply discuss actors. It also focuses on the play-maker (kyōgen tsukuri), which was originally positioned as an actor’s ability, and discusses the internal structure of kabuki works they created. In this regard, references are also made to the kabuki theoretical work Sakusha Shikihō Kezairoku and the instructional text Sekai 174

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Kōmoku—both written in the Edo period (discussed in detail in the section Surveys and Analyses of Two Documents from the Narrative Generation System’s Point of View in this chapter), and theories of related narratives and literature such as nō and haiku as theories on the “shigumi” of kabuki. Shigumi approximately means methods for creating the structures and developments of kabuki works and plays. Hattori’s structural theory of kabuki can be considered in comparison with the multipl narrative structures model the author has proposed. Alternatively, as it closely resembles the author’s multipl narrative structures model, it can be viewed by incorporating the structural model of kabuki into the multipl narrative structures model. Another possibility is to understand the model of the structure of kabuki as an example of the multipl narrative structures model. However, the multipl narrative structures model is a model for narrative generation, and naturally, it places the concept of the “creator” at its center. Nevertheless, as described, Hattori convincingly states in this book that kabuki has traditionally (and presently) regarded the “actor” as more important than the “creator.” How should this point be considered?—In this regard, however it would be useful to consider Hattori’s concept of the “actor” to be different from—or greatly surpass or expand upon—the concept of actors as we imagine them today. According to Hattori, the “actor” in kabuki was not someone originally defined as a person who only acts; it was nearly the only position that made up a troupe. In other words, kabuki was born from troupes (ichizas) as a gathering of actors. Watanabe (1989) also emphasizes in his writing that kabuki is a form of geinō born out of troups or ichizas. Actors, in this case, therefore did not merely perform plays on a stage, but comprehensively undertook a range of jobs that accompanied this. We can then say that these duties also included a job that was equivalent to that of a “creator.” In other words, kabuki did not originally have a “creator” as an independent person, and the so-called “actor” was not an actor who (merely) performed in the narrow sense in which actors are presently understood, but in an expanded sense, who also performed the roles of the creator, the director, and the producer in a troupe (ichiza). Accordingly, although they have the designation of being an “actor,” they are not actors in the narrow sense that we imagine today. This situation is not particularly mysterious from the perspective of the aurhor’s multipl narrative structures model. In other words, in the multipl narrative structures model, the title of an agent as a person or organization involved in narrative generation is no more than a symbol used for convenience, and what is important is the actuality. Therefore, in this case, the agent known as the “actor” contains the “actor,” the “creator,” the “director,” the “producer,” or a swathe of other functions in the narrative. The entirety of this is fulfilled by several people known as actors. In this sense, the actor is not merely an agent operating just as an “actor.” In more definite terms, the author has said that some actors are key actors while others are not, and so on and so forth, but will omit that here. In the preceding section, the author examined the structure of GIS with INGS, and the author noted there that GIS with INGS does not explicitly deconstruct the function of “creator,” and also that in the narrow sense, a geinōjin character (entertainer, performer, or player character) equivalent to the “actor” may also simultaneously function as a creator. The reason this structure arose is probably that GIS with INGS is not a narrative generation system in the narrow sense, but no more than an “geinō information system.” If “geinō” is placed at the center, the existence and concept of the “creator” in a narrow sense shifts to the background or is rendered dormant. It should be noted that another reason the concept of the creator is not brought forward in geinō is that geinō works are often not new works that exhibit creativity in the way that we think of it. Many geinō works are repeated performances or revisions of existing works. In nō, a narrative geinō predating kabuki, the job of the creator existed more explicitly broken down than in kabuki, although there is considerable overlap with the actor, and many works that have been passed down clearly show the name of the 175

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

creator. However, Zeami (c. 1363–c. 1443), the most important nō creator and its greatest theoretician, states that nō works should always use characters appearing in existing works or well known historical or fictional characters and add new innovations to them (Zeami, 1983). Ningyō jōruri and kabuki also follow this kind of tradition (although at the same time, ningyō jōruri and kabuki create narratives that use topics contemporary to the period more often than nō). However, notwithstanding these circumstances, Hattori also writes about more specialized creators as those opposed to creators in the sense of actors who are also creators. This is due to the process observed as we pass through the early developmental stages of kabuki, in which the substance of the job the actor must do to also be a creator gradually becomes more complicated, and more importantly, kabuki narratives gradually become more complicated, thereby requiring the position of the creator in a narrow sense, the creator as a specialist. Amid these statements, Hattori refers to Sakusha Shikihō Kezairoku and Sekai Kōmoku, a kabuki theoretical work (the former) and an instructional text and knowledge base (the latter), both written in the Edo period, and also mentions the creator’s mechanism and the mechanism within the creator’s process for creating a narrative. The final chapter of The Structure of Kabuki refers to something called “kabuki works.” Like many kabuki texts, Hattori does not provide style-based commentary or explanations of “katas,” which mean a kind of typical and settled patterns or forms for various aspects in kabuki, on individual works performed on stage or individual genres, but rather pursues the core topic of what the fundamental principles of kabuki narratives are. Here, Hattori primarily considers “jo-ha-kyū”, which has underlined Japanese geinō from ancient times and was positioned by Zeami as the most fundamental principle of nō. (Although it is diffilcut to explain the concept of jo-ha-kyū because it has deiverse meanings, oen of the imterpretations is “slow introduction, normal-speed development, and rapid finale.” However, it is not simpley an idea according to speed. For example, “ha” means a kind of breakdown or interruption of the first state, i.e. “jo.”) However, more interestingly, Hattori locates the fundamental principles of kabuki narratives in haikai as a literary genre of short poems including haiku, and moreover in the Danrin-ha (Danrin school) of haikai. Particular attention is paid to renga-like and rengu-like methods in which several individual waka and haiku are connected, rather than the structure and form of the individual haiku, and Hattori considers this method to display phenomenon coupling principles in narratives unique to kabuki. In addition, renga and rengu are two types of linked-verse forms, and waka is a genre of Japanese classical short songs. Incidentally, Hattori describes, as a type of generative grammar, the process by which early kabuki’s structure as a Shōwattained more complexity than a simple structure containing only a single phenomenon and then one part became fixed as a kata. Descriptions like this have probably been influenced by the structuralism and narratology from when this book was written, but this is a characteristic no longer seen in later kabuki study, or in humanities study that includes it. However, such an idea is useful in identifying phases of complication in kabuki, in addition to narrative and geinō works in general, and merits reexamination from the perspective of the author’s narrative generation system research in particular.

Tamotsu Wanatabe’s Philosophical or Narratological Thought of Kabuki Tamotsu Watanabe is a present-day kabuki researcher and kabuki critic, but unlike other purely intellectual researchers, critics, and commentators who provide impressionistic criticism, he is supported by a foundation of thought and philosophy, and moreover, is an author of timely kabuki commentary, making him a rarity. Watanabe has written sporadically on thought and philosophy (Watanabe, 2004, 176

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

2012, 2018), but here, the author presents and discusses the unique, particular thought and philosophy written in Musume Dōjōji (Watanabe, 1986), a research text and critique concerning the kabuki known as “Dōjōji-mono”, which the author and other researchers are currently studying, and in particular, the work Kyōganoko Musume Dōjōji, which is considered a masterpiece of kabuki dance. (Kabuki script: Kyōganoko Musume Dōjōji, 1970; Video: Bandō, 2003) The kabuki dance Kyōganoko Musume Dōjōji was not an original kabuki work in the beginning. At the same time as the ancient legendary tale called Dōjōji Densetsu (“The Legend of Dōjōji”) passed through Buddhist fable, nō and other writing and illustrated scrolls came to be incorporated into kabuki and made into various works as Dōjōji-mono, and were crystallized as Kyōganoko Musume Dōjōji in kabuki dance. In this way, narratives concerning Dōjōji have a long history of transition stretching over more than a thousand years. In addition, Dōjōji is a temple in Kishū or Kii province in the Edo period (modern-day Wakayama prefecture) and Dōjōji-mono includes many narrative works in diverse genres based on Dōjōji Densetsu. Specifically, Kyōganoko Musume Dōjōji forms a type of sequel, originating in the Legend of Dōjōji, an ancient legend relating to the temple of Dōjōji. Once the Legend of Dōjōji was remade as a kabuki dance, it developed into a work with its own contents beyond the original legend, even though many parts were omitted or simplified in the plot of the narrative. According to Yasuda (1989), the Legend of Dōjōji underwent the following transitions to become Kyōganoko Musume Dōjōji in kabuki: The narrative of Dōjōji has long historical transitions over one thousand years. The oldest records of the Legend of Dōjōji can be seen in the scene 129, Ki no Kuni Muro no Kōri no Ashiki On’na [The Evil Woman from Muro District, Kii Province] (No. 129) (1974) in Dainipponkoku Hokkekyō Genki (or Honchō Hokkekyō Genki) (1974). The first half is a tale of female passion, in which a woman who fell in love with a young monk but was betrayed burned the young monk together with a bell, while the second half is a tale of salvation, in which the woman and young monk were saved by the power of Buddhist law. Next, act 14, scene 3 of Konjaku Monogatari Shū (1993, 1994, 1996, 1999a, 1999b) also describes this as Ki no Kuni no Dōjōji no Sō Hōke wo Utsushi Hemi wo Sukueru Koto [The Tale of a Monk of Dōjōji in Kii Province Who Transcribes the Lotus Sutra and Saves a Snake] (Vol. 14, No. 3) (1993). The story is mostly the same as in Dainipponkoku Hokkekyō Genki, although there is more interest in the young man’s difficulties in the first half. Genkyō Shakusho (1965) describes the man as “Anchin” from Kuramadera (Kuramadera temple), but refers to the woman only as “a widow” and emphasizes the man’s perspective. In the illustrated scroll Dōjōji Engi (1982), held by Dōjōji in Wakayama and said to date to the Ōei era (1394–1428) in the Muromachi period, the woman transforms while chasing the man and takes the form of a snake to cross a river and leaps into the river. This woman is also described in specific detail as “the wife of the master, a person known as Kiyotsugi Shōji,” of the inn in “Masago, Muro District, Kii Province,” and the perspective from the woman’s side is growing. The name “Kiyohime” first appears in Dōjōji Genzai Uroko (Asada & Namiki, 1995), a ningyō jōruri dated around the 18 th century. The first volume of Dōjōji Monogatari (1982), which is a story composed of three volumes that was published in Manji 3 (1661), depicts a travel scene in which the monk makes a pilgrimage to Kumano Province (modern-day Wakayama Prefecture). The second volume reveals the story of the women transforming into a serpent out of passion and burning the monk to death, and the third volume presents the story of the woman who comes to perform a memorial service for the bell, transforms into a serpent and hides in the bell, and a tale of salvation in which the monk and the woman perform a memorial service by transcribing the Lotus Sutra. The scene in the tale of salvation where the bell is restored and the memorial service is performed was incorporated into the nō Kanemaki by 177

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Kanze Kojirō Nobumitsu (?1435 or 1450–1516) (at present, this work is rarely played), and the later nō Dōjōji (1998) which is of unknown provenance, and simplifies the dialog and narration in Kanemaki and emphasizes the dance section. In Dōjōji, the serpent transforms into a courtesan and appears where the memorial service for the bell is being performed, then causes the bell to fall from hatred and hides within it. This serpent does not pray to enter Nirvana, but vents her hatred on the bell. Kanemaki focused on the male perspective, which views the woman’s acts as evil, but Dōjōji gives rise to a perspective that examines the woman’s sadness. To this point, the Legend of Dōjōji has transmuted from a tale in which the man has a spiritual awakening and attains Buddhahood to an expression of the woman’s sadness and hatred, and later, the Legend of Dōjōji comes to be reread as a story from the woman’s perspective. In the early modern period, the Legend of Dōjōji comes to be incorporated into kabuki and ningyō jōruri. The kabuki dance plays with the bell of Dōjōji begin with Katari Dōjōji or Ko-Dōjōji (also known as Furu-Dōjōji). “Dōjōji Kane no Dan (Act of Dōjōji’s Bell)” in Kokon Hauta Taizen (Kurihara, 1886) provides the major verse and prose for the nō plays Dōjōji and Miidera [Miidera Temple] (1998) and is considered a pioneering work in Dōjōji dance. The first Dōjōji-mono in kabuki is Wakanoura Kataonami (premiered in Genroku 2 (1689) in the Edo period). Kabuki and ningyō jōruri Dōjōji-mono developed under each other’s influence. The first half of the tale of the Dōjōji has been reconstituted in performances to better integrate it with early-modern realistic developments, and today, only the part where Kiyohime turns into a snake and crosses the Hidaka-gawa (Hidake River) remains in dance plays in both kabuki and ningyō jōruri. However, the main interest in the tale of Dōjōji was instead focused on the sequel. In other words, interest in the Legend of Dōjōji moved away from the story as a whole and towards connections between the bell, the serpent, and the woman. In the third month of Hōreki 3 (1753), Musume Dōjōji saw one type of completion as a kabuki dance in Kyōganoko Musume Dōjōji, danced by Nakamura Tomijūrō I (1719–1786), in the third scene of the kabuki, Otokodate Hatsugai Soga (1985) at the Edo Nakamura-za. This work inherited the form of the nō Dōjōji without change, but the verse and prose did not concretely show the Legend of Dōjōji itself and did not tell the tale of the past of the woman, the main character. This work focused, in the end, on showing the dance itself with its many changes, rather than the development of the story. In other words, the Legend of Dōjōji is not expressed in the verse and prose and has been relegated to information that provides the background for the work. In this way, Kyōganoko Musume Dōjōji wandered far from the original legend. Moreover, the tone of the dance is still bright and flashy on the surface. However, a terrible sense of fear leaks from it, and ultimately, the climax of the story explodes. This extreme contrast between the flashy gaudiness and terrible darkness forms the attraction of this work. This important work that showcases kabuki dance is known as a work played by the representative on’nagata (male performers who play female roles) of the time, such as Nakamura Utaemon VI (1917–2001) in the Shōwa era (1926–1989) and Bandō Tamasaburō V (1950–) in the Shōwa era, Heisei era (1989–2019), and current Reiwa era (2019–) in recent years, and is a masterpiece that exhibits to the greatest extent the properties of kabuki dance in its delicateness, fineness, dramatic characteristics, and subtle changes, among other features. Tamotsu Watanabe’s Musume Dōjōji is a significant work that comprehensively describes the historical origin of Kyōganoko Musume Dōjōji and its development, combined with the paths of genius on’nagata actors such as Segawa Kikunojō I (1693–1749) and Nakamura Tomijūrō I in the Edo period. The statements in Musume Dōjōji mostly follow the process of the origin and development of Kyōganoko Musume Dōjōji, closely relating these to actors (on’nagata). The main on’nagata who appear here include Segawa Kikunojō I, who first developed Musume Dōjōji (Momochidori Musume Dōjōji (1960; Nagauta Genpon 178

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Shūsei Kankōkai, 1927)), Nakamura Tomijūrō I, who completed Kyōganoko Musume Dōjōji, Segawa Kikunojō II (1741–1773) and III (1751–1810), Arashi Hinasuke I (1741–1796), Bandō Mitsugorō I (1745–1782), Nakamura Nakazō I (1736–1790), Ichikawa Uzaemon IX (1791–1819), Onoe Matsusuke I (1744–1815), Iwai Hanshirō IV (1747–1800) and V (1776–1847), Yoshizawa Iroha I (1755–1810), Sawamura Tanosuke II (1845–1878), and Ichikawa Dan’nosuke I (1684–1740). The final chapter is the first to provide the reader with statements concerning the structure of the narrative of Kyōganoko Musume Dōjōji itself. Specifically, Watanabe deconstructs the acting (dancing) of the female main character of Kyōganoko Musume Dōjōji into the three elements of mind, gestures, and lyrics, and has created a table containing the concrete contents (values) for 11 parts of the modern performance form consisting of 14 parts in total. Here, “mind” is a term that expresses the true nature of the character (at the time), such as “girl” or “ghost,” “gestures” means the type of physical movement, such as the dances shirabyōshi no odori, hanagasa odori, or nō-gakari, and “lyrics” refers to the song or music with lyrics, such as hauta, taue-uta, or wakite-bushi. As seen in Table 1 (equivalent in the structure to Musume Dōjōji, p. 479, table, upper 4 rows), the contents of the mind, gestures, and lyrics continually change for each part. Table 1. Mind, gestures, and lyrics of the female main character in Kyōganoko Musume Dōjōji and the phases of changes therein 心 (Kokoro: Mind)

Scenes

振り (Furi: Gestures)

歌詞 (Kashi: Lyrics)

1. 道行 [Michiyuki]

亡霊(嫉妬—芝居) [Bōrei (Shitto—Shibai): Departed spirit (Jealousy—Play)]

生娘(流行唄) [Kimusume (Ryūkō-uta): Innocent girl (Popular song]

白拍子(能の道行) [Shirabyōshi (The michiyuki of Nō]

2. 乱拍子 [Ran-byōshi]

娘 [Musume: Girl]

白拍子 [Shirabyōshi]

能 [Nō]

3. 中啓の舞 [Chūkei no Mai]



白拍子

能(鐘づくし) [Nō (Kanezukushi)]

4. 手踊り [Te-odori]

真奈児の娘の嫉妬 [Manago no musume no shitto: The jealousy of Manago’s daughter]

生娘

端唄(娼婦のエロティシズム) [Ha-uta (Shōfu no eroticism)]

5. 鞠唄 [Kemari-uta]



鞠つき(踊り) [Maritsuki (Odori)]

廓づくし(遊女) [Kuruwazukushi (Yūjo)]

6. 花笠 [Hanagasa]



花笠踊 [Hanagasa-odori]

わきて節 [Wakite-bushi]

7. くどき [Kudoki]

恋する女(裏切られた女) [Koisuru on’na (Uragirareta on’na): Woman in love (Betrayed woman)]



恋と嫉妬の小唄 [Koi to shitto no kouta (A song of love and jealousy)]

8. 山づくし [Yama-zukushi]



鞨鼓の踊り [Kakko no Odori]

山づくし [Yamazukushi]

9. 手踊り [Te-odori]

真奈児の娘の嫉妬

生娘

端唄(娼婦のエロティシズム)

10. 鈴太鼓 [Suzu-daiko: Bell and drum]



鈴太鼓 [Suzu-daiko]

田植唄(早乙女踊) [Taue-uta (Saotome-odori)]

11. 鐘入り [Kaneiri]

本性 [Honshō: True character]

能がかり [Nō-gakari: Nō-like gestures]



12. 祈り [Inori]







13. 蛇体 [Jatai]

亡霊

能がかり



14. 押戻し [Oshimodoshi]

亡霊

能がかり

芝居とのつながり [Shibai to no tsunagari (The connection with the Dōjōji story)]

179

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Watanabe proposes an extended table that adds the elements “costume,” “props,” and “physical points” to the structure of the dance in Kyōganoko Musume Dōjōji based on these three elements, and completes it by inserting values showing the concrete contents for these elements of “mind,” “gestures,” “lyrics,” “costume,” “props,” and “physical points” for each of the 14 parts of Kyōganoko Musume Dōjōji. Watanabe’s work contains many keen observations, and the author has frequently cited it in his previous research (Ogata, 2018b, 2019c, 2020g; Kawai, & Ogata, 2019, 2020). With regard to Musume Dōjōji, much of the final chapter in particular is spent on considerations of thought and philosophy. Watanabe’s analysis of Kyōganoko Musume Dōjōji is based on his thoughts described below, which the author deeply identifys with. The major aspects of his thought are demonstrated through the following quotations. First, the combination of the elements of mind, gestures, and lyrics in the development of Kyōganoko Musume Dōjōji does not have a consistent method (Watanabe’s “law”). Watanabe states the following in regard to this: “The six elements, including not only mind, gestures, and lyrics, but also the color of the costumes, props, and physical points, are in opposition in each scene, and moreover, they are independent between scenes. The contrast between what appears to be a ghost (mind) but is a girl (gestures), or between what appears to be a girl but is a courtesan (lyrics) becomes in other scenes what appears to be a ghost (mind) but is a girl (gestures), what appears to be a girl but is a courtesan (gestures), and what appears to be a theatrical travel scene depicting travel in nō then becomes a ranbyōshi dance while still remaining nō. This completely lacks logic, consistency, and unity, and has no law. If there were a law, it would be the law that there is no law. The law of “absence.” Completely contradictory things and unrelated things are combined together to establish a relationship.” [p. 473] However, it is pointed out that the law of no laws has the following type of method: “Each of the elements in this table are like the colored glass in a kaleidoscope. The elements themselves, like the pieces of colored glass, have no relation to each other. They are scraped together and form a pattern, and if you give them a shake (after each scene), they create a different pattern. They create mostly accidental relationships from moment to moment. These relationships are not created by each part linking to the others organically, but merely by differences in shape. One scene is formed by these relationships. This kind of thing is repeated infinitely, and can be inserted or substituted freely anywhere.” [pp. 473–476] The lack of law here really means the lack of the following: “Here, things like an overall, coherent structure, topic, meaning, which modern artworks are certain to have, are completely absent. Usually, each of the elements operates together to express the topic or meaning of a work, and each of the elements are a part that forms the whole. No matter how many parts are gathered, they cannot make the whole on their own. There is a center that gathers the parts, and the whole takes form only through this center. This center is the topic and the meaning of the work.” [p. 476] However, Watanabe holds that Kyōganoko Musume Dōjōji lacks a center or topic or meaning in the sense set out above. (“However, this table lacks what is known as a center.” [p. 476]) In addition, “What is here are independent parts, and just the arbitrary relationships created by the parts, and these meaningless relationships alone continue infinitely. Odd relationships, in which the parts are themselves the whole, and the whole is in other words the parts, continues.” [p. 476] This so-called methodless method, however, is not completely without target, but actually forms a strategy constituted with a view to a target. To explain, this is because Nakamura Tomijūrō I, who first created this work, was able to implement the aforementioned methodless method and thereby “absorb all manner of entertainment genres. Nō, kabuki, dance (mai and odori), folk entertainment (minzoku geinō), party songs (zashiki-uta), popular songs. It extends from the nō favored by the upper classes to the popular songs of the cities, and from the performances of geinin (entertainers) in the pleasure quarters 180

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

to the local entertainment (geinō) in the tall-grassed countryside. These things were all absorbed.” [p. 476] What is more, “What is important here is that there is one center and these things are not things that have been inserted towards the center. They have not been inserted, but synthesized. In other words, there was not a single center initially, and the thing itself was formed by synthesis. If there had been one center, the things that are all independent from one another would certainly not have formed a harmonious whole to this degree. They formed a harmonious whole because in the first place, they were nothing but synthesis.” [pp. 476–477] According to Watanabe, this is also a narrative structure he calls “absence—the structure of the void.” “Everything has a structure that it can be placed into, and once it enters this structure, each of its parts become independent and assert their own existence. For example, the spoken passage in the seventh scene (dan or act) seems like a confession of love by a betrayed woman, but the woman seems like a girl and also like a courtesan, and each of these personalities assert themselves, and from this discord arises a multifaceted, multiplex vision of the woman. This vision is precisely what is composed of the relationships, and the individual relationships themselves form a single whole (the vision). The whole exists only within these relations. This is a property of the structure of the void.” [p. 477] Thus, this structure of the void is “a mishmash, a patchwork,” [p. 477], but “the fact that this mishmash, patchwork is neither merely a mishmash nor a patchwork is due to the presence of the relationships that make the whole by themselves, and what establishes these relationships is the structure of the void, which rejects any topic or meaning. The mishmash, patchwork itself is the source of life that makes Kyōganoko Musume Dōjōji.” [p. 477] The aforementioned may be summarized thus: Kyōganoko Musume Dōjōji is not a narrative of a type where central elements (important semantic elements) and other elements exist in orderly organization within a story with a climax, such as in Western narratives. It is simply a narrative in which diverse narrative elements bridging various genres are just synthesized, and that rejects its own structure in which these converge into the meaning or topic as a whole. This creates an organized, semantic whole, or rather a synthetic whole. This explains the structure of Kyōganoko Musume Dōjōji as a work in which various parts covered by “all manner of entertainment genres” are arranged haphazardly within the broader narrative framework (which largely consists of the undepicted scenes of the narrative). The aforementioned is a discussion of the uniqueness of so-called narrative structure (in a broad sense) in Kyōganoko Musume Dōjōji, but Watanabe further continues to attempt a discussion of the characteristics of the dances in this work. Specifically, he examines the dance in Musume Dōjōji under the Shigayama school (a school of dance), and states that the dance is not a symbol that points to something outside of itself. (“The choreography in the descriptive explanation is overwhelmingly small, and almost all of it consists of gestures unrelated to the meaning of the lyrics or the dramatic setting.” [p. 480]) This also underlies the aforementioned fact that narratives are merely a synthesis of various elements. Narratives and dances are not organized as tributes to something, but exist as independent entities, and neither exists as some sort of meaning or something that can be reconstituted as a “narrative” in the form of a certain aggregation of meaning. Speaking boldly, Kyōganoko Musume Dōjōji as discussed by Watanabe is a pure narrative that does not serve as a tribute to anything other than a narrative being a narrative, and is entirely different from narratives like advertisements, for instance, whose existence is evidenced by serving something other than the narrative. In the case of advertising, we are forced to be conscious of the effect in correlation with the purpose. However, pure narratives do not have a certain purpose, and “assessment” itself is not possible. For example, “The idea here is that dance is the language of the body itself, nothing more, nothing less. This language is uttered only from within the very movements of the 181

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

body, and is not controlled by anything else. Of course, it cannot be translated into everyday language at all. Moreover, it cannot be concealed or diluted by psychological descriptions or emotions. If it rejects the ‘meaning’ of the lyrics on one hand and excludes ‘psychological descriptions’ and ‘emotions’ on the other hand, all that remains is just the very movements of the body.” [p. 481] In this chapter’s author’s experience of examining kabuki, he has frequently felt uneasy upon seeing modern kabuki from the Meiji era and later (which is often called Shin Kabuki (New Kabuki)). For example, in Kidō Okamoto’s (1969) Banchō Sarayashiki [The Dish Mansion at Banchō], the master, who would not have punished the chambermaid he secretly loved for merely breaking one of the family’s heirloom plates, could not forgive her for doubting him for even a moment, and in the end, brutally murdered her. This work retains the start and end of the original legend of the dish mansion, but replaces the history linking the two with a mentalist narrative. This psychological story could be construed as a simple explanation to draw out the morals of the brutal murder in accordance with kabuki aesthetics, but in present-day kabuki performances, it is usual for the psychological drama section to occupy a central position and the final brutal murder to be treated as a mere addition. This direction is the “progress and development” in modern kabuki or Shin Kabuki and was regarded as an achievement in Kyōganoko Musume Dōjōji and stands far from the meaninglessness or “nature as amusement” of kabuki; however, in truth, the progress and development is no more than a change biased towards the meaning of individual psychology and things. The author constantly has this impression. (Yet, if Banchō Sarayashiki, which the author has presented here as an example, is seen as a rare work that grotesquely materializes complicated psychological reasons, its narrative could be interpreted anew as something with a range exceeding a mere psychological drama. In the author’s experience, he had a similar feeling when he saw Yūzō Yamamoto’s (1887–1974) Shin Kabuki, Sakazaki Dewanokami (1977). Since the Meiji era, Japanese literature and narratives incorporated and applied “mentalist realism” that rushed in from the West, but this climate notwithstanding, kabuki plays have their own biases—perhaps we should think that the so-called “expressionist” bias exceeding realism from the Edo period and later still remained, without going into decline.) So, Tamotsu Watanabe’s views on kabuki, as presented above, offer an important suggestion regarding the author’s narrative generation research for kabuki, or his narrative generation research in general, in terms of proposing narratives as a broad framework capable of arbitrarily synthesizing narratives from various genres.

Surveys and Analyses of Two Documents from the Narrative Generation System’s Point of View The author surveys and analyzes in detail Sakusha Shikihō Kezairoku and Sekai Komōku from the narrative generation system’s point of view to be actually applied to narrative generation mechanisms based on INGS and GIS.

The Survey and Analysis of Sakusha Shikihō Kezairoku In this section, based on the Ogata (2019e) and Hattoris’s (1970), the author shows the structure and content of Kezairoku. Namiki Shōza II (?–1807), an Edo-period kabuki writer (and not Namiki Shōza I (1730–1773), who was famous for Yadonashi Danshichi (Yadonashi Danshichi Shigure no Karakasa) [The first performance 1768] and other kabuki works), penned the renowned work Sakusha Shikihō 182

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Kezairoku (1972) under the name Nyūgatei Ganyū in Kyōwa 1 (1801). Namiki Shōza II is not regarded as an exceptional kabuki writer, but his theoretical abilities and writing abilities were regarded extremely highly (Iura, 1990; Kominami, 1998), and although this particular work is not especially long, it is known as a systematic, theoretical description on the creation of kabuki, which was rare in the Edo period and even more so today. This book exists in several printed versions, each differing from the others (Iura, 1990, 1991), but here, the author mainly refers to the revised edition by Masakatsu Gunji (1913–1998) (Sakusha Shikihō Kezairoku, 1972), which is easier to obtain today, and its revisions, annotations, and commentary. However, Iura points out several issues with Gunji’s revisions and annotations, so at times, the author also adds Iura’s opinions on amendments. Moreover, the aforementioned Yukio Hattori’s The Structure of Kabuki also mentions Sakusha Shikihō Kezairoku, so the author has also referred to it. The author (Ogata, & Ono, 2019) has evaluated this book and the following statements are also an expansion thereof. The overall composition of Sakusha Shikihō Kezairoku is: • • •

• • • • • • • • • • • • • • • • •

[Introductory part] 作者差別之事 [Sakusha Sabetsu no Koto] 浄瑠璃作者連名 [Jōruri Sakusha Renmei]: Nyūgatei Ganyū regards Chikamatsu Monzaemon, Ki no Kaion, Bunkōdō (Matsuda Wakichi) (the mid Edo period), and Namiki Senryū (Namiki Sōsuke) (1695–1751) as Kyōgen Sakusha Shiten’no (“The Greatest Four Jōruri Writers”) and shows other 56 jōruri writers. 歌舞伎作者之部 [Kabuki Sakusha no Koto]: This section shows 82 kabuki writers are listed and also includes the figures of Goka Jūyō no Den and Nidō Shibetsu no Den (which are introduced in the following part by the author of this chapter). 看板之心得 [Kanban no Kokoroe] 三都狂言替り有事 [Santo Kyōgen Kawari Aru Koto] 竪筋横筋之事 [Tatesuji Yokosuji no Koto] 四季人情差別之事 [Shiki Ninjō Sabetsu no Koto] 狂言之場行工合之事 [Kyōgen no Ba Yuki Guai no Koto] 作者金言之事 [Sakusha Kingen no Koto] 作者役場心得之事 [Sakusha Yakuba Kokoroe no Koto] 役者役場之事 [Yakusha Yakuba no Koto] 役場甲乙之事 [Yakuba Kōotsu no Koto] 役割番附之事 [Yakuwari Banzuke no Koto]: 36 kabuki actors are treated. 表八枚看板の事 [Omote Hachimai Kanban no Koto] 作者支配之事 [Sakusha Shihai no Koto] 江戸紋番附之事 [Edo-mon Banzuke no Koto] 大阪絵看板之事 [Osaka Ekanban no Koto] 一夜附の事 [Ichiyazuke no Koto]/役納る之事 [Yaku Osamuru no Koto]/作者出勤之事 [Sakusha Shukkin no Koto] 幕明の景様之事 [Makuake no Keiyō no Koto]/引上て下に置ぬ [Hikiagete Shita ni Okanu]/一 件男女にて引しめる事 [Ikken Danjo nite Hikishimeru Koto]

The aforementioned includes several interesting items, but the one that has attracted attention for many years is the figure titled Sakusha Kokoroe no Koto: Goka Jūyō no Den (The Writer’s Knowledge: 183

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

The Commentary of Five Flowers and Ten Leaves),” listed in “Kabuki Sakusha no Bu (Kabuki writer’s section).” The figure is reproduced as Figure 10 by adding the English alphabet. The main text does not directly explain this figure, but Iura (1990) argues that the text following the figure offers clues to its meaning. Furthermore, the Sakusha Shikihō Kezairoku book was not originally published for and distributed to the public, but a secret book passed down through generations of kabuki writers and the associated people. However, given the central position occupied by this figure and the book’s secret charFigure 10. Sakusha kokoroe no koto: Goka jūyō no den

acter, it can be imagined that verbal explanation may have been deliberately omitted (Iura, 1990, 1991). The Goka Jūyō no Den forms a type of method for creating kabuki plays. It is explained below with reference to Gunji’s revisions and annotations and the aforementioned Iura paper. From this figure, the narrative world develops within the following narrative structure. This is equivalent to the five-stage composition for plays that is popular in kabuki and ningyō jōruri. •



184

景様 (Keiyō): Gunji writes this using different characters that mean “appearance” and holds that it means showing the form of things in the opening act. Conversely, Iura believes that it rather means visual appeal, as is suggested by the similar terms keigoto (dance plays in ningyō jōruri) and keibutsu (seasonal features), among others. In other words, it is not the internal features, but the beauty of form, and mainly a good outward appearance. 頂上 (Yama): Gunji considers this to correspond to “gradation.” Conversely, Iura argues that it means “the peak in the latter half within the first half of the entire drama.” This is equivalent to the

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents



• •

peak in the “world” in kabuki works considered period pieces. Two major classifications of kabuki narratives are jidaimono (period pieces) and sewamono (pieces about everyday life). Jidaimono are narratives developed with a focus on the samurai class and nobles, rather than commoners, in periods earlier than when they were created, while sewamono are narratives developed based on the actual lives of commoners in the current period when they were created. Many kabuki works had a structure that develops as jidaimono-like narratives in the first half and continues with development as sewamono-like narratives in the second half. The yama referred to here, according to Iura, refers to the peak in the jidaimono section in the first half of the work. 揺 (Yusuri): According to Gunji, this means the thing that fluctuates or is the most active. Conversely, Iura states that it refers to a dispute that again overturns the situation that had settled down in the yama, and that this portion occupies a mini-scene-like position that is not equal to the other four parts and has a range of functions and contents that draw out the next scene. 大曲 (Ōkuruwa): According to Gunji, this is the scene where the narrative makes a single large development. Conversely, Iura argues that it contrasts with the yama, and means a major narrative development in the contemporary world. 鎌入 (Kamaireru): This means reaping, or in other words, the denouement (Gunji), or the scene with the resolution of all the complications and changes (Iura).

Kominami (1998) compares the narrative structure depicted in this figure with Freytag’s structural diagram for Western classical plays. As a comparative theory, it has not been described in detail, but can be thought of as an awareness of a problem that views the five-stage structural diagram in Sakusha Shikihō Kezairoku in a more generalized form. In broad terms, this direction aligns with the perspectives and targets of my research, which understands Sakusha Shikihō Kezairoku as a narrative generation model to be a form of “narratology,” or positions it as a collection of groups of techniques for a narrative generation system. In this sense, comparing the diagram in Goka Jūyō no Den with the jo–ha–kyū model, the ki–shō–ten–gō (“introduction–development–turn–conclusion”) model, Propp’s morphology model, Prince’s narrative grammar model, and the cognitive science-based story grammar model, among others, and assessing it are both important tasks. Returning to the contents of Sakusha Shikihō Kezairoku, what is considered important after the aforementioned portions in relation to the narrative structure and techniques in kabuki is the statement portion in Tatesuji Yokosuji no Koto (On Vertical and Horizontal Stripes). It states that kabuki narratives develop through a certain shigumi (plot) and shukō. Here, shigumi means an ordinary method of development along the lines of narrative pattern, and shukō means the creator’s own unique innovations and ideas for that. Goka Jūyō no Den also uses the terms shigumi and shukō, and below that, describes the two representative patterns for methods of narrative development, jo–ha–kyū and below that, ki–shō– ten–gō. The shigumi means to basically follow these, while the shukō means inserting unique mutations and differences into it. Note that Zeami viewed jo–ha–kyū systematically as a comprehensive basic principle that stretched from single nō narratives and their performance compositions that would last from one day to several days, to finely detailed compositions of a single sound or word. Sakusha Shikihō Kezairoku invokes jo–ha–kyū as a concept that shows (one face of) the basic structure of a single work. By contrast, ki– shō–ten–gō is a concept that indicates a typical (normal) composition of a four-line Chinese poem (a poem of four lines with five characters in each). The relationship between jo–ha–kyū and ki–shō–ten–gō in Sakusha Shikihō Kezairoku is such that jo is linked to ki and shō, ha is linked to shō and ten, and kyū 185

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

is linked to ten and gō. The interpretation that the transitions between elements in jo–ha–kyū uses and is supplemented by ki–shō–ten–gō is also possible. The Tatesuji Yokosuji no Koto section of Sakusha Shikihō Kezairoku also contains the following statement that appears to relate to the sekai (“world”) seen in the Goka Jūyō no Den section of the same work, and the naimaze (blending) (Miura, 2000) relating to that: “As the world also serves to establish the main stripes, an ordinary world does not work on a play. Because the stripes are assembled, they are called vertical and horizontal stripes. Let us insert the horizontal stripes of Ishikawa Gozaemon into the vertical stripes of Taikōki [The Life of Toyotomi Hideyoshi] (1996). In addition, Kashiwade, Kinnari, Sakurako, Katsurako, Keyamura Rokusuke, etc., are all horizontal stripes. The vertical stripes are the sekai (world), and the horizontal stripes are the shukō. The vertical is jo. Even if the stripes are mixed until the end, it does not work. If the horizontal stripes are brought out towards the middle, they begin to work and make the play look new. This is an important aspect.” However, according to Urayama (1974), the method described here cannot necessarily be considered the blending of sekai in the original sense. The method called the shukō here means inserting into the sekai specific people in the narrative before the sekai is formed. In contrast, “blending” means a method of fusing two or more sekai in one narrative. (However, Iura notes that if the shukō part is expanded or systematized by the sekai, Tatesuji Yokosuji no Koto will approach blending in its original sense.) Conversely, the above-cited Kominami mentions the ideas of Tatesuji Yokosuji no Koto. First, he adopts a more generalized view of plot as a method of structuring the world of the narrative with multiple layers. In addition, he understands this as an idea broadly seen in renga, the use of legendary backgrounds in painting, waka, haiku, and other genres of Japanese poetry and art. On the other hand, shukō is understood as a comprehensive, method-based concept relating to all elements involved in stage production, including production, acting, techniques, and costumes, which include a variety of techniques to create changes in the narrative. Next, another figure Nidō Shibetsu no Den (The Commentary of Two Paths and Four Differences) which Gunji annotated with “Precepts for living on the creator’s path,” is depicted as Figure 11. Even here, the figure is not explained in the main text and Gunji’s commentary is only brief, and it is therefore difficult to interpret accurately. The author would like to think that it can be reinterpreted as a type classification for characters, slightly modifying the understanding that it is a description concerning creators. A more detailed explanation than Gunji’s commentary is provided by Iura (1991), but that also ultimately expands this Nidō Shibetsu no Den beyond its narrow meaning of the path of the creator and believes, “The creator must write these diverse people and the world they form,” and “The world of people is a huge stage complicated and constructed by the fame and wealth, and the living, as well as the jōkon (ability to follow the Buddhist path), jitsujō (true feelings), gekon (lack of ability to follow the Buddhist path), and hakujō (unfeelingness), as shown in the ‘two paths and four differences,’ and various things.” In addition, the Santo Kyōgen Kawari Aru Koto (That Plays Change in Three Cities) section raises the kawa (“skin”), “niku (“flesh”), and hone (“bones”) theory,” which represents the character of kabuki performed in the three cities where kabuki was held with state approval in the Edo period as Kyō (kawa), Edo (hone), and Ōsaka (niku), but Sakusha Shikihō Kezairoku develops a generalized theory related to the characterization of kabuki works. It argues, “The stripes are the bones, the plot (shigumi) is the flesh, and the script (serifu) is the skin,” and “If the bones, flesh, and skin are all continuous, they create an excellent work.” According to Iura, hone refers to a strong, filled out character, niku refers to a richly filled out character, and kawa refers to a smooth, beautiful character. In addition, the points that should be considered important in jidaimono, oiemono (plays about upheavals in feudal clans), 186

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Figure 11. Nidō shibetsu no den

and sewamono (which according to Iura includes the everyday life parts in the entirety of a narrative, in which case they become everyday life scenes) are represented through the metaphor of bones, flesh, and skin, as “Jidaimono should have solid bones, oiemono should have abundant flesh, and sewamono should have fine skin.” (Note that this theory of kawa, niku, and hone is thought to be originally based on discussion by Zeami (1983).)

The Survey and Analysis of Sekai Kōmoku The author shows and discusses the content and structure of Sekai Kōmoku based on Ogata (2019e), Hattori’s The Structure of Kabuki (1972), and other references. The sekai (world) in Sekai Komōku means the typical or normal systematized knowledge for each individual narrative that has been used in creating kabuki. One sekai is a system consisting of the plot of the narrative, the main characters, the main settings (stage or spaces), the main temporal scope (period, season, etc.), and other elements. A set of sekais can be thought of as a type of knowledge base shared between people involved in creating kabuki and the recipients. Sekai Kōmoku (1916) has been regarded as an Edo-period document concerning “sekais” It shows in concrete detail how sekais were understood kabuki at the time. Sekai Kōmoku was not a document providing abstract explanations of what sekais are in kabuki, but a type of practical handbook or manual—albeit a secret one permitted for use only by top-class creators accepted by their fellows, and not for use by the general public. Sekai Kōmoku lists many practical examples of works that have been or could be used as raw material for kabuki at the time, using a consistent description method. Katō (1997a, 1997b) states that Sekai Kōmoku began when the famous kabuki writer Kanai Sanshō (1732–1798)—viewed as a rival of the equally famous writer Sakurada Jisuke I (1734–1806)—gathered and documented the sekais of many renowned kabuki of the time. Sekai Kōmoku is a type of practical secret text for sekai sadame (world-fixing)—a regular function held at the start of autumn every year in the mid-Edo period to determine the basic direction for kabuki plays that would be performed in each theater in the coming year—and was handed down mainly among kabuki creators who followed Sakurada Jisuke’s line, and once the custom of sekai sadame faded towards the end of the Edo period, it

187

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

came to be distributed more widely. In any case, it is not the kind of book that was in general distribution. Furthermore, Sekai Kōmoku exists in three editions—the National Diet Library (Kokuritsu Kokkai Toshokan) edition, the University of Tokyo College of Arts and Sciences edition, and the Tsukuba University edition—and although there are significant differences between two of these editions in particular, they are all considered to spring from the same root. The reader is referred to Katō’s aforementioned paper. Incidentally, the University of Tokyo College of Arts and Sciences edition is a transcription of the National Diet Library edition by kabuki and ningyō jōruri researcher Kuroki Kanzō (1882–1930). For reference, the aforementioned sekai sadame is discussed briefly in terms of its relation to the present research. Kominami (1998) describes kabuki performances particularly in the period when the sekai sadame function was still actually performed as a magnificent dramatic world over a span of one year. This idea connects with my GIS narrative generation = narrative production–consumption. GIS is positioned as a mechanism to cyclically create a narrative at a broader level than as a chain of individual narrative works, but the formulation of an annual performance strategy based on specific world under sekai sadame in the Edo period can be viewed as a concrete, exceptional example of a manner of operation for the GIS narrative production–consumption system. In Japan, this kind of idea of identifying a set of individual narratives as a narrative on another level has existed for some time. For example, Zeami viewed the concept of jo–ha–kyū in terms of the structure of a single nō work, or in terms of the structure of a fragment of acting by an actor, zooming in to the closest level, and conversely in terms of the overall structure of one or several days’ performance, zooming out to the furthest level. For Zeami, jo–ha–kyū was not a theory or method fixed or tied to a specific level, but occupied a position as a universal formal and structural theory, and a method that went beyond identification with or affixing to any level. The author’s concept of narrative generation is similar, and narrative generation is equally valid on various levels in the multipl narrative structures, from the micro to the macro. GIS with INGS is a term that indicates narrative generation in general, from the micro to the macro, in which INGS bears the generation of narratives as single units and GIS bears the production of compound or linked narratives. The aforementioned, sekai sadame to formulate a strategy for a chain of performances in a year-long time framework means, in the author’s words, a strategy to ascertain a representative segment for compound or linked narrative production in the kabuki GIS. Thus, according to the National Diet Library edition, Sekai Kōmoku is divided into the following four sections, each of which contains many or few world descriptions: 1. 歌舞伎時代狂言世界之部 (Kabuki Jidai Kyōgen Sekai no Bu) [58 sekai]: (1) 日本武尊 (Yamato Takeru), (2) 神功皇后 (Jingū Kōgō), (3) 仁徳天皇 (Nintoku Ten’nō), (4) 衣通姫 (Sotoorihime), (5) 浦島 (Urashima), (6) 松浦佐用姫 (Matsurasayohime), (7) 聖徳太子 (Shōtokutaishi), (8) 大 職冠 (Taishōkkan), (9) 天智天皇 (Tenchi Ten’nō), (10) 大友皇子 (Ōtomo no Ōji), (11) 大友眞鳥 (Ōtomo no Matori), (12) 百合若 (Yuriwaka), (13) 安部仲麿 (Abe no Nakamaro), (14) 弓削道鏡 (Yuge Dōkyō), (15) 中将姫 (Chuujōhime), (16) 田村丸 (Tamuramaru), (17) 融大臣 (Tōru Daijin), (18) 小野篁 (Ono no Takamura), (19) 小町 (Komachi), (20) 松風村雨 (MatsukazeMurasame), (21) 業平 (Narihira), (22) 道風 (Tōfū), (23) 北野御記 (Kitanogyoki), (24) 蝉丸 (Semimaru), (25) 将門純友 (Shōmon Sumitomo), (26) 惟茂 (Koremochi), (27) 源氏六十帖 (Genji Rokujujō), (28) 四天王 (Shiten’nō), (29) 奥州攻 (Ōshūzeme), (30) 殺生石 (Sesshōseki), (31) 保元物語 (Hōgen Monogatari), (32) 平治物語 (Heiji Monogatari), (33) 平家物語 (Heike Monogatari), (34) 頼政 (Yorimasa), (35) 伊豆日記 (Izu Nikki), (36) 木曽 (Kiso), (37) 義経記 (Gikeiki), (38) 源平軍並に頼 朝治世 (Geipei Ikusa Narabini Yoritomo Chisei), (39) 曾我 (Soga), (40) 頼家治世 (Yoriie Chisei), 188

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

(41) 實朝治世 (Sanetomo Chisei), (42) 和田合戦 (Wada Kassen), (43) 鉢木 (Hachinoki), (44) 太 平記 (Taiheiki), (45) 東山 (Higashiyama), (46) 甲陽軍 (Kōyōgun), (47) 出世奴 (Shusseyakko), (48) 小栗 (Woguri), (49) 苅萱 (Karukaya), (50) 三庄太夫 (Sanshō Dayū), (51) 愛護 (Aigo), (52) 角田川 (Sumidagawa), (53) 信田 (Shinoda), (54) 俊徳丸 (Shuntokumaru), (55) 望月 (Mochizuki), (56) 放下僧 (Hōgesō), (57) 大内之介 (Ōuchinosuke), (58) 甲賀三郎 (Kōga Saburō) 2. 御家狂言之内敵討之部ならびに類 (Oie Kyōgen no Uchi Tekiuchi no Bu Narabini Rui) [7 sekai]: (1) 義士傳 (Gishiden), (2) 伊賀上野 (Iga Ueno), (3) 浄瑠璃坂 (Jōrurizaka), (4) 亀山 (Kameyama), (5) 御堂前 (Midōmae), (6) 巌柳島 (Ganryūjima), (7) 非人敵討 (Hinintekiuchi) 3. 歌舞妓世話狂言世界之部 (Kabuki Sewa Kyōgen Sekai no Bu) [75 sekai]: (1) 一代男 (Ichidai Otoko), (2) 與之助 (Yonosuke), (3) お七 (Oshichi), (4) 曽根崎 (Sonezaki), (5) 重井筒(Shigei no Tsutsu), (6) お千代半兵衛 (Ochiyo Hanbei), (7) お花半七 (Ohana Hanshichi), (8) 三勝半七 (Sankatsu Hanshichi), (9) お染久松 (Osome Hisamatsu), (10) お菊幸助 (Okiku Kōsuke), (11) お 梅粂之助 (Oume Kumenosuke), (12) 丹波與作 (Tanba Yosaku), (13) 山崎與治兵衛 (Yamazaki Yojibei), (14) お夏清十郎 (Onatsu Seijūrō), (15) 夕霧伊左衛門 (Yūgiri Izaemon), (16) 玉屋新 兵衛 (Tamaya Shinbei), (17) 梅川忠兵衛 (Umagawa Chūbei), (18) 小金彦惣 (Okane Hikosō), (19) お三茂兵衛 (Osan Mohei), (20) 小春治兵衛 (Koharu Jihei), (21) 金村屋 (Kanamuraya), (22) 助六総角 (Sukeroku Agemaki), (23) 椀久松山 (Wankyū Matsuyama), (24) 淀屋 (Yodoya), (25) 本町二丁目 (Honchō Nichōme), (26) お亀與兵衛 (Okame Yohei), (27) お萬源兵衛 (Oman Genbei), (28) おきさ治郎兵衛 (Okisa Jirobei), (29) おしゅん傳兵衛 (Oshun Denbei), (30) 梅 川新七 (Umegawa Shinshichi), (31) 小いな半兵衛 (Koina Hanbei), (32) お勘半九郎 (Okan Hankurō), (33) 樽屋おせん (Taruya Osen), (34) おさい道徳 (Osai Dōtoku), (35) 神田與吉萬 字屋高崎 (Kanda Yokichi Manjiya Takasaki), (36) 金屋金五郎額の小さん (Kanaya Kingorō Hitai no Osan), (37) おはん長右衛門 (Ohan Chōemon), (38) 信田妻 (Shinodazuma), (39) 道成 寺 (Dōjōji), (40) 富士淺間 (Fuji Asama), (41) 薄雪 (Usuyuki), (42) 清玄 (Seigen), (43) 鳴神上 人 (Narukami Shōnin), (44) 七草四郎 (Shichikusa Shirō), (45) 高尾 (Takao), (46) 粂豊勝 (Kume Toyokatsu), (47) 天竺徳兵衛 (Tenjiku Tokubei), (48) あこぎ平治 (Akogi Heiji), (49) 累 (Kasane), (50) 東金茂衛門 (Tōgane Moemon), (51) 鑓の権三 (Yari no Gonza), (52) 五人男 (Gonin Otoko), (53) 團七 (Danshichi), (54) 黒船 (Kurofune), (55) 濡髪 (Nuregami), (56) 梅の由兵衛 (Ume no Yoshibei), (57) 源五兵衛 (Gengobei), (58) 湯の勘右衛門 (Yu no Kan’emon), (59) 帯刀男達, (60) 神祇組 (Jingigumi), (61) 銘職人 (Mei Shokunin), (62) 佛教大師 (Bukkyō Daishi), (63) 役 行者 (Eki no Gyōja), (64) 弘法大師 (Kōbō Daishi), (65) 柿本記僧正 (Kakinomoto no Kisōjō), (66) 空也上人 (Kūya Shōnin), (67) 法然上人 (Hōnen Shōnin), (68) 新泉馬上人, (69) 日蓮上人 (Nichiren Shōnin), (70) 遊行上人 (Yugyō Shōnin), (71) 西行 (Saigyō), (72) 兼好 (Kenkō), (73) 一 休 (Ikkyū), (74) 自然居士 (Jinenkoji), (75) 釋氏名目 (Shakushi Meimoku) 4. 神祇之部 (Jingi no Bu) [3 sekai]: 淡島 (Awashima),常陸帯 (Hitachiobi),高砂 (Takasago) Each world consists of the following items: ① 役名 [Yakumei: Role names]: A list of major characters who appear in the sekai. ② 引書 [Hikigaki: References]: A list of ancient (authoritative) literature in which the sekai appears. ③ 義太夫 [Gidayū: Gidayū jōruri]: Lists the names of gidayū jōruri (jōruri works with dramatic recitation accompanied on the shamisen) works that use the sekai.

189

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

For example, the Woguri world, which has comparatively few items, in the Kabuki Jidai Kyōgen Sekai no Bu (Kabuki Period Play World Section), includes the following contents: •

• •

Role names: 小栗判官兼氏實は孫五郎 (Woguri Hangan Kaneuji, Jitsu ha Magogorō), 横山治 郎 (Yokoyama Jirō), 栗橋太郎 (Kurihashi Tarō), 戌亥局 (Inui no Tsubone), 青蟇の長 (Aohaka no Osa), 鬼次新左衛門 (Oniji Shinzaemon), 常陸小萩 (Hitachi no Kohagi), 遊行上人 (Yugyō Shōnin), 池庄司時門 (Ike no Shōji Tokikado), 照手の姫 (Terute no Hime), 鬼王源太 (Oniō Genta), 後藤新左衛門 (Gotō Shinzaemon), 後藤三郎 (Gotō Saburō), 横山郡司(式部太郎) (Yokoyama Gunji (Shikibu Tarō)). References: 『小栗實記』 (Woguri Jikki), 『新編鎌倉志』 (Shinpen Kamakurashi). Gidayū jōruri: 『小栗判官』 (Weguri Hangan), 『小栗判官軍街道』 (Woguri Hangan Ikusa Kaidō), 『今様小栗判官』 (Imayō Woguri Hangan), 『忠臣金短冊』 (Chūshingura Kin no Tanzaku), 『鬼鹿毛武蔵鐙』 (Onikage Musashi Abumi).

To explain, the aforementioned example means that the Woguri, which consists of 14 characters, was originally recognized as an independent world based on Wguri Jikki (The Authentic Account of Woguri) and Shinpen Kamakura Shi (New Topography of Kamakura), and Woguri Hangan (Judge Woguri) and the following works were formed based on this world. Many of us presently do not have detailed knowledge about Woguri Jikki or Woguri Hangan, but the writers or creators and associated people at the time when they were used in sekai sadame would have been entirely familiar with their contents. In fact, the contents of various narratives listed here combine to provide concrete form to the attributes and characteristics of the characters, the period (time), and place or stage (space) where the narrative is developed, and outlines of the narrative’s story and plot in the world of Woguri. According to Urayama (1974), sekais did not exist as a continually fixed thing, but changed and interchanged over time. Using the discussion of vertical and horizontal stripes, that is, the shigumi and shukō, in the previously mentioned Sakusha Shikihō Kezairoku as an example, much of what was considered a shukō in this case were elements such as characters in sewamono. However, in Sekai Kōmoku, sewamono also form many of the worlds as previously described. From this, the very method of combining shigumi and shukō that was mentioned in Sakusha Shikihō Kezairoku may not be able to be termed blending as discussed, but when a sewamono-like narrative that forms the foundation for a shukō, as seen in Sekai Kōmoku, is established as a sekai, the aforementioned discussion of horizontal stripes–vertical stripes or shigumi–shikō approaches the discussion of blending methods or for strategy. As a discussion relating to sekai in Kabuki, the author wishs to present research by Shigeo Moriyama (1914–2000), who analyzed sekais and blending based on many actual works and subsequently presented a more generalized theory of sekai. Moriyama’s (1993) Tsuruya Nanboku: Naimaze no Sekai [Tsuruya Nanboku: The World of Blending] gives hardly any conceptual explanation in the body, and relates in extremely concrete detail (or “in a surge”) across more than 370 pages the state of blending in the various works of Tsuruya Nanboku IV (1755–1829), a kabuki writer from the Edo period, and to fully comprehend this, the reader is forced to read the group of works set as source materials and then make careful notes while reading through, but in the two pages of the “Afterword,” a general explanation of blending finally appears. According to this explanation:

190

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

• •

• •

Naimaze (blending) is destroying existing sekais and reorganizing them as different worlds. Doing so not only demolishes the consistent order, patterns, and narratives of the traditional worldview, but creates a new, larger-scale, compound sekai. Naimaze historically began with mixed performances of the highlights of a few works, and gradually built up to the doubling or compounding of different sekais from different works. However, this historical development or complication was influenced by the actors’ desire for a diversity of performance, especially their desire to unveil transformation performances, in addition to the growing complication of audiences’ wishes for dramatic performances. The evolution of creators’ dramaturgic skills was linked to the discovery of a new drama in the conflict between the orders of multiple sekais. Works that had been the norm were parodied, and works that appeared to be parties came to have meaning as new norms. Tsuruya Nanboku IV freely used naimaze as a conscious dramaturgic method. Nanboku used sekais from various narratives and inherited, invaded, or transmuted them to give birth to one new sekai after another. Thus, the strategic operation of sekai leads to the birth of novel sekais.

Moriyama’s description of naimaze is concrete, using Nanboku as its raw material, but it is treated as a highly comprehensive narrative technique or strategy. The aurhot also thinks that naimaze is not a mere means for creating kabuki narratives, but the most important composition principle. Here, naimaze may be thought of as the narrative creation principle, by which the “sekais”—characters, events, historical background, plot-lines, etc.—of existing narratives—myths, legends, folk tales, stories, history, etc.—are used as the base for creating a new narrative work through the synthesis and correlation of other narrative sekais by a range of methods. This is an overarching term that contains diverse methods and techniques. It should also be noted that Uchida, Kondō, and Funato (1987) propose the creation of a computer system from Sekai Kōmoku. The proposed system has functions to attach furigana (reading information) to all items that appear in Sekai Kōmoku and can search for the names of relevant works from the name of a character, among others. Although it is rather simple, the author presents it as a pioneering achievement. Inspired by Sekai Kōmoku in the Edo period, the author is drawn to the idea that it could be possible to create a modern-day Sekai Kōmoku. In this chapter, the author discusses the scheme for Watakushi Monogatari, but from the perspective of using various knowledge sources as raw materials for it. They include a variety of events from ancient times to the early modern period and the present day, and certain types among them could be considered to already exist as worlds. In particular, early modern and modern history has several event units that can be recognized as worlds. Thus, as seen in the Edo-period Sekai Kōmoku, sekais are often given concrete form through the writing as works of literature and narratives, rather than history texts. Based on this idea, the construction and establishment of sekais through works of literature and narratives is needed for the early modern and modern periods. In any case, the author aims to compose a certain something among the knowledge sources for Watakushi Monogatari, either by borrowing the concept of a sekai as discovered by kabuki, or by referring to the composition of Sekai Kōmoku. In the previous sections, the author attempted to consider Sakusha Shikihō Kezairoku and Sekai Kōmoku from the perspective of narrative generation system research, and as demonstrated in several places, these works have parts that overlap. In particular, the Goka Jūyō no Den and the theory of horizontal and vertical stripes (shigumi–shukō), which form the heart of Sakusha Shikihō Kezairoku, are intimately related to the concept of worlds. In the future, I aim to use my considerations mixing the two as a bridge to reach the construction of one or more methods within a narrative generation organization 191

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

that utilizes kabuki’s narrative methods or GIS with INGS, incorporating sekais, naimaze, shigumi–shukō, horizontal and vertical stripes, and jo–ha–kyū.

THE CONCEPT OF WATAKUSHI MONOGATARI AND DESIGING INTERNAL AND EXTERNAL NARRATIVE GENERATION METHODS TO CREATE AND DISTRIBUTE NARRATIVE CONTENT In the two sections that have thus far formed the bulk of this chapter, A MULTIPLE NARRATIVE GENERATION MECHANISM THROUGH TWO TYPES OF NARRATIVE GENERATION SYSTEMS: GIS WITH INGS and NARRATOLOGY OF KABUKI: ORIENTED TOWARD NARRATIVE GENERATION, the author first outlined the structure of GIS with INGS and in so doing established the Watakushi Character as a type of geinōjin resource with the aim and purpose of creating Watakushi Monogatari. In the subsequent section on kabuki, the author introduced several theoretical and practical studies and documents surrounding kabuki in the modern and Edo periods. On the basis of this, in this section the author will discuss Watakushi Monogatari as concretely as possible. First, the Figure 12. The structure of this section

author will discuss how private knowledgea for Watakushi Monogatari fundamentally constitutes the content of the Watakushi Character, and then the author will describe the internal and external plans for Watakushi Monogatari. Here, the internal plan is a plan or direction in relation to aspects concerning the contents of narrative content (which is further separated into content and formal aspects). At the same time, the external plan directly signifies how the contents of the narrative content unfold externally

192

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

and socially; at the same time, this also has a meaning in relation to the method by which it creates the narrative contents of Watakushi Monogatari. In other words, in Watakushi Monogatari the process of narrative creation and the process of societal distribution are not cut and dried, but rather they are more closely related in an essential sense, and the internal and external plans or directions of Watakushi Monogatari here are discussed as an intimately intertwined and integrative form. Based on the above, the structure of this section can be illustrated as shown in Figure 12.

The Basic Structure of GIS for Watakushi Monogatari The GIS was originally envisioned as neutral and generic. In other words, it was a type of tool that was positioned as an applied system for narrative generation research. Thus, the traditional design excluded consideration of the specific “author” or “geinōjin character” elements that create and distribute narrative content through the GIS. During the design and development of a system that is intended to function as a tool or an application, the system designer or developer is usually the provider of that tool or application and, as such, he/she is sharply distinguished from those who use the tool to produce content. However, one of the most important goals of this study on narrative generation systems is to examine the author’s production of personal and private works in a format that is more than a simple autobiography or a firstperson novel. Moreover, in order to go beyond the level on which there is a tool, and that tool has a use, the author is also aiming for a tool form that is simultaneously a work of art. In this section’s title, the term “internal narrative generation” refers to an author’s generation of narratives on the personal and private levels, and signifies a narrative work or narrative content that is produced by processing a private narrative and Watakushi Monogatari. The author considers this narrative and its generation on the personal and private levels from the perspective of Watakushi Monogatari’s conceptual goal. Having had many opportunities to discuss Watakushi Monogatari, the author has long had the idea for a personal study as well as the intention to bring that idea to fruition (Ogata, 2018c, 2018d, 2020d, 2020h). As set out earlier in this chapter, Watakushi Monogatari refers to a narrative based on the author’s personal subject matter, objects, and topics, and is further generated using GIS with INGS. To confirm Watakushi Monogatari’s basic concepts, the author would first like to describe it in a somewhat abstract sense, based on the various abovementioned papers. Watakushi Monogatari is created using GIS with INGS in a special way. The “I” (the Watakushi Character) that exists half within the GIS and half outside of it is able to perceive the fluid narrative created and expressed through GIS with INGS. Furthermore, the character is able to generate a “narrative as a fixed thing” on a separate level. Through these processes, the Watakushi Character, in conjunction with GIS with INGS, can collect all the fluid narratives generated at a specific point in a cycle or iteration of narrative generation using GIS with INGS to generate a “fixed Watakushi Monogatari” on a higher level. In short, GIS with INGS and the Watakushi Character, half of which is positioned internally and half of which is positioned externally, ultimately arrive at a fixed Watakushi Monogatari through a cycle of fluid and fixed narrative generation on a number of levels (although this provisionally final fixed narrative, when viewed from a higher level still, can be one phase of a fluid narrative). Both the concept of narrative generation in Watakushi Monogatari through the flow and fixation described above and the personal and private knowledge that should be used to create Watakushi Monogatari are presented in the section titled Concepts of Internal and External Narrative Generation in

193

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

the BACKGROUND in this chapter, and the author will explore them in more detail in the next section. Chapter 4 (Ogata, 2020h) in Ogata (2020e) includes a detailed explanation of the relevant framework. Moreover, although the conceptual image of Watakushi Monogatari generation is shown in Figure 9, the author presents a more specific and newly extended view below. The geinōjin resource within the GIS includes a Watakushi Character as one of the geinōjin characters. This Watakushi Character is half artificial, by virtue of belonging to the GIS, while the other half corresponds to an “I” (watakushi) of reality, signifying that there are aspects of the Watakushi Character that automatically generate narrative as well as aspects of narrative that are directly generated by this “I” (watakushi) of reality. The Watakushi Character contains both private and personal knowledge related to this “I” (watakushi), and a discussion of its specific contents, namely what kind of content comprises each respective category, will follow in the upcoming section titled Toward Internal Narrative Generation: The Creation of Watakushi Monogatari. As described above, the actual generation of narrative takes place through GIS with INGS and the Watakushi Character, and “fluidity and fixation” are concepts that are involved in this generation. Chapter 4 (Ogata, 2020d) in Ogata (2020a) offers a foundational view of this concept. At present, the author considers real narrative generation through fluidity and fixation to be represented by the following image. First, to the author, Watakushi Monogatari as the final narrative content is a fixed narrative. Indeed, this emerges from the interaction between fluid and fixed narrative generation. The specific generation process can be understood as follows (details of which may be subject to change in the future). That is to say, the Watakushi Character is constantly monitoring (as an artificial entity, this is primarily the Watakushi Character’s performative role) the narrative (automatically) being constantly generated by GIS with INGS (this has so far been viewed as fluid narrative). Based on these narratives (though, of course, not only these), the Watakushi Character, as a real entity, creates a fixed narrative. Even during this generation process, it may be possible to use an automatic generation mechanism in some form. In this sense, fixed narratives are assumed to be fragmentary, and none are very long. Their short duration allows those fragmentary fixed narratives to be subsumed again into the GIS with INGS and used for fluid narrative generation. Through this recursive repetition, the Watakushi Character creates the final Watakushi Monogatari as a larger fixed narrative.

Toward Internal Narrative Generation: The Creation of Watakushi Monogatari In the sense of narrative content, the term “content” refers, in short, to what is written in that narrative. By contrast, the narrative form––for example, where a narrative is written in text, as in a novel––refers to narrative structure and the type of composition used (i.e., sentence length, word choice, whether action is emphasized over psychological description or vice versa, etc.). It is easy to imagine that every narrative has different narrative content. It is also true that not all narratives are written in the same form. However, there are a great many narratives that take very similar forms but have different narrative content. Perhaps changing or even innovating narrative forms is a task that requires a more conscious strategy than changes to or innovation in narrative content. Furthermore, while the formal aspect is a crucial element for the narrative, many authors are not explicitly aware of the narrative’s form. Although various arguments can be made, narrative consists of both content and formal aspects. In the following pages, the author will discuss the narrative created by Watakushi Monogatari by separating it into two parts, namely the formal aspect and the content aspect, to clarify what the author means by internal narrative generation.

194

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

The Conception of Content in Internal Narrative Generation The author will now expand on Ogata’s (2020e) description of narrative generation research’s development toward the internal, specifically the development and creation of Watakushi Monogatari. This section describes according to the following order, Fundamental Knowledge Sources for the Narrative Content of Watakushi Monogatari, Two Directions of the Narrative Content of Watakushi Monogatari and Their Fusion: From Fluidity/Fixation To In Motion/Static. (1) Fundamental Knowledge Sources for the Narrative Content of Watakushi Monogatari Although this does not constitute the narrative content of Watakushi Monogatari itself, the author will first consider the knowledge group that should be the source of its development materials. This knowledge group was first organized by Ogata (2018c, 2018d), and Chapter 4 (Ogata, 2020h) in Ogata’s (2020e) provides an extended description. These are summarized in the section titled Internal and External Narrative Generation in the BACKGROUND section in this chapter. Here, as an internal concept of narrative, the author will provide what can be considered “material” for the as-is realization of Watakushi Monogatari’s narrative content; however, a worthwhile task for the future will be to provide such actual, specific material information in a more organized form and structure the narrative content material using fuller descriptions. The following list shows the knowledge that is needed for Watakushi Monogatari: 【A. Temporal Division】 1. The temporal division of “Present Continuous” 2. The temporal division of “Past” 3. The temporal division of “Past Perfect” 4. The temporal division of “Historical” 【B. Text Categorization】 1. Event descriptions 2. Personal chronologies 3. Social and historical chronologies 4. Personal texts a. Published academic contents b. Unpublished contents related to academic activities c. Materials from conferences and classes d. Contents that was not published (contents stored privately without it having been presented or published anywhere) e. Lesson records and large volumes of notes from self-study, research gatherings, and other occasions written for record or contemplative reasons 【C. Personal readings】 1. Japanese literature from antiquity to the present (works: narratives, novels, poems, waka, haiku, etc.) 2. World literature from antiquity to the present (works: narratives, novels, poems, plays, etc.) 3. Japanese literature from antiquity to the present (commentaries, critiques, miscellaneous writings, etc.) 195

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

4. World literature from antiquity to the present (commentaries, critiques, miscellaneous writings, etc.) 5. Music (especially Western classical music and jazz, but perhaps the truly important are the other genres) 6. Video works like movies and TV 7. Kabuki and ningyō jōruri and texts (research and commentaries) 8. Works by Yukio Mishima (and related research and commentaries) 9. Other miscellaneous narratives Of the above, B-4 “Personal discourse” consists of the existing personal documents themselves, but the question exists as to how specifically these should be linked to Watakushi Monogatari’s narrative content. The same holds for “personal ‘reading’” in C. Closely examining these is a task for the future; however, broadly speaking, there is both a direct as well as a more indirect method for addressing the issue of bridging this gap. The direct method makes direct use of individual texts, while the indirect method processes and uses the text in some way. A close examination of the latter, i.e., “personal ‘reading,’” is also a task for the future, but it may be possible to use this in various forms, including based on direct quotation and processing. Under “personal ‘reading,’” the author has listed only the novels the author himself has read from junior high school up to university (the majority of which are modern classics, although there are also a number of Japanese classics) with the intention of later using them as narrative content for what the author calls Watakushi Monogatari. In addition to this, in the future, specific consideration must be given to the question of how information should be added, processed, and used, and in Appendix 1 to this chapter, the author provides an expanded list containing detailed information. However, this list is a tentative version to aim at future completion for introducing it as narrative networking knowledge into GIS and the creation of Watakushi Monogatari. Next, the knowledge that is to be examined, considered, and, more specifically, created as the substance of Watakushi Monogatari’s narrative content corresponds to items 1-3 in B, namely “Event description,” “Personal chronicle,” and “Social and historical chronicles” for each of the “time categories” that are recorded in A above. Here, B-1 “event description” refers to the description or explanation of a relevant past event that the author has directly or indirectly experienced. The event description groups are arranged chronologically, corresponding to “Personal chronicle” and “Social and historical chronicle.” Here, “Personal chronicle” refers to a chronicle that includes, at its core, a group of events that are directly related to the author, while also including a group of events that are not directly related to the author, but are closely, albeit indirectly, linked to the former set of events. On the other hand, the phrase “Social and historical chronicle” refers to a general chronicle that does not overlap with the author’s life and can be very roughly separated into the following three fields: the first is the social and historical chronicle of Japan since the Meiji period; the second is the social and historical chronicle of an historically earlier Japan; and the third is the social and historical chronicle of relevant regions outside of Japan, that is, the world. These “Event descriptions,” “Personal chronicles,” and “Social and historical chronicles” have the following relationship. When one clicks, so to speak, on a particular description under a “Personal chronicle” or a “Social and historical chronicle,” a relatively detailed description of the corresponding event description appears. There are two types of event descriptions, namely personal/private (“Personal/ private events”) and social (“Social/historical events”), which, respectively, correspond to 1. “Present continuous time segments,” 2. “Past time segments,” 3. “Things spoken of in the past perfect tense,” and 4. “Historical periods and epochs” in A, namely “Time segments.” 196

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Unlike B-4, writing these anew is time consuming. It is likely that these descriptions alone would be sufficient to produce one or more narratives. However, these are not the final Watakushi Monogatari, but rather information and texts that form the materials for it. In the section titled The Overall Behavior of GIS, Including a Watakushi Character with INGS, the author describes how the fluid narratives constantly created by GIS with INGS constitute a collection of fragmentary narratives from which one or more fixed narratives are created through the Watakushi Character, but conversely, these event descriptions and personal as well as social and historical chronicles may be fragmentary narrative groups created by the Watakushi Character’s “I (watakushi)” and then forming a bridge to Watakushi Monogatari. A deeper, more substantial description remains a task for the future, but below the author sets a general policy for each of these items with respect to the information that should be described in the future. Thus, as indicated above, the author would like to stress that the specific events in each of the time categories are only described tentatively for the purpose of formulating a general policy; thus, not only may they be expanded and detail or greater specificity added, they may also be changed, removed, or restructured. [The temporal division of “Present Continuous”] This is the present form that has been in existence since March 11, 2011 (Heisei 23 in Japanese era name), and it must continue until a certain point in time in the future. It is demarcated as a time segment, containing personal and private events as well as social and historical events. However, while the author uses the term “historical” as a shared classification term, social events, in this case, are either prior to or in the process of being historicized, and so they are an intermediate step. In addition, private and personal events, and social and historical events have overlapping boundaries that cannot be clearly separated from one another. This situation is common to other time segments as well; however, especially in the present, which is an ongoing time segment, personal and private events are strongly shaped by social and historical events, and so the area of ambiguity is also broader. The lists below provide examples of each event: A. Private and personal events 1. Personal and private situations during and immediately following the Great East Japan Earthquake. 2. Personal and private situations at universities. 3. Personal and private situations in the “countryside” in Japan (rural life). 4. Private and personal “pair illusion” situations (Yoshimoto, 1968; Ogata, 2020h). 5. Private and personal life situations amidst the novel coronavirus pandemic that began around the end of 2019 and spread globally (i.e., a further present ongoing situation in the present ongoing time segment). B. Social and historical events 1. Social (as well as political, economic, and cultural) conditions in Japan during and immediately following the Great East Japan Earthquake. 2. Social (as well as political, economic, and cultural) conditions in Japan since the Great East Japan Earthquake and continuing up to the present. 3. Global social (as well as political, economic, and cultural) conditions during the same period. 4. The situation in Japan and around the world surrounding the novel coronavirus pandemic that began globally around the end of 2019 (i.e., a further present ongoing situation in the present ongoing time segment).

197

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

[The temporal division of “Past”] This is a time segment that was demarcated based on the author’s memories, which span the period beginning on November 25, 1970 (Shōwa 45), and ending on March 11, 2011 (Heisei 23). As the author has mentioned several times, November 25, 1970, is the date on which the writer Yukio Mishima died; to the author, this is not only a date of social and historical significance, it also holds significance as the boundary beyond which the author gained a real awareness of his interest in literature and narrative, ergo, this is a date that marks an artistic period for the author on both a personal and a private level. Many important social and historical events occurred between that day and the day of the Great East Japan Earthquake; however, the precision or sharpness with which the author recalls them differs depending upon the strength of their connection to his personal and private events. As such, the author shall only cite those social and historical events with which he has a strong connection. The following are examples of personal and private as well as social and historical events that can be categorized as such: A. Private and personal events 1. Memories of personal and private events on the date of Yukio Mishima’s suicide at the SelfDefense Force Base in Ichigaya, Tokyo, i.e., the Mishima Incident. 2. Private and personal situations in “school” society. 3. Private and personal situations in “family” society. 4. Private and personal situations as individual illusions. 5. Private and personal situations during the experience of “work” at a “company.” 6. Private and personal situations during the experience of “work” (i.e., research, teaching, etc.) at universities. B. Social and historical events 1. Yukio Mishima commits suicide at the Self-Defense Force Ichigaya Camp. 2. Serial lynchings and murders are carried out by the Coalition Red Army and the Asama-Sanso Incident. 3. The Japan Airlines Osutaka-yama (Mt. Osutaka) crash. 4. The chain of collapse of socialist states and regimes that began with the overthrow of the Soviet Union. 5. The Gulf War. 6. Chaos and the sudden rise of emerging Islamic states. 7. China’s move from recovery to establishing economic hegemony. 8. The collapse of Japan’s “bubble” society. 9. Emperor Shōwa’s (Hirohito’s) death. 10. The Great Hanshin earthquake. 11. The Aum Shinrikyo incidents. 12. The acceleration of the general weakening of Japanese society under the Jun’ichirō Koizumi (1942–) and Heizo Takenaka (1951–) regimes. 13. Economic and social stagnation and decline in Japan. 14. The Lehman Brothers’ fall. 15. Emperor Heisei’s (Akihito’s) retirement and the Crown Prince’s (Hironomiya’s) ascension as Emperor Reiwa (Naruhito). 16. The continuation of Shinzō Abe’s (1954–) cabinet.

198

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

[The temporal division of “Past Perfect”] This is the period of time in which the author’s “I (watakushi)” becomes indeterminate or where his “myself” objectively did not exist; however, people directly connected to the author, such as his parents, did exist, covering the period from around 1925 (Shōwa 1) to November 25, 1970 (Shōwa 45). Therefore, for many periods of time during which the author was not yet in the world, his personal and private events also did not exist. Even after the author came into the world, he, for a while, had no memory of private or personal events (although there are some such events that precede his own memory, like the date of his birth); memory-based descriptions of the author’s personal or private events became possible from around the early 1960s, and they become clearer after 1964 (Shōwa 39). A. Private and personal events 1. Memories originate mostly from hearsay, the author’s parent’s event groups, and those of the people around them. 2. Dim, vague memories of event groups (several episodes) from the author’s early childhood (and the people surrounding them). 3. A group of events from the years between kindergarten and elementary school, when the author’s memory becomes clearer, and the events group (of people) related to it (numerous episodes). B. Social and historical events 1 The eve of World War II and the Pacific War. 2 World War II and the Pacific War. 3 The period of chaos and confusion following Japan’s defeat. 4 A period of strong economic growth. [The temporal division of “Historical”] This denotes a temporal segment during which the “I (watakushi)” in the author’s mind does not exist within it, the existences of the various people who are connected with the “I (watakushi)” itself become indeterminate, and moreover, time indistinctly mixes with the vast historical space that goes beyond the private. A criterion linked to the author’s “I (watakushi)” may make it possible to jump to a certain place in history. This criterion is basically considered to be private and personal interest as well as curiosity, since there is no direct personal relationship of any kind between the events that are categorized here and the “I (watakushi).” However, if this relationship is not always direct, and spiritual elements can also be involved, it can be said that personal and private interest and curiosity may, for some reason, in the context of the author’s own “I (watakushi),” also become private and personal necessities. For example, the fact that the land where the author’s ancestors lived for a long time was (indirectly) on the side of the losers in the Battle of Sekigahara (1600), which is of great historical significance, and also (directly) on the losing side in the Boshin War (1868–1869), which was of vital significance over 150 years ago, both serve as strong psychological impetuses to remember these conflicts and their accompanying social conditions. Of course, the author neither directly experienced the battle or the war nor did the author has any kind of exchange with anyone who observed them; however, such a statement can be made to the extent that it is not impossible for events of that kind to stand out as leaving a somewhat prominent impression among the many groups of events that are, for example, arranged chronologically in textbooks. There are many other issues that must be examined and compiled with respect to these social and historical events (and the same can also be said for all other types of events). A systematic discussion 199

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

on this topic is a task for the future; however, it is important to mention the difference in the character of the so-called universality of social and historical events. In other words, this is the extent to which a social or historical event is global or, conversely, the extent to which it is local. Supposing there was an incident or event that had a fairly equal impact on all the regions of the world, this event would have a highly global character. The novel coronavirus that is currently sweeping the globe is certainly likely to be such a highly global phenomenon. Conversely, there are social and historical events that have only a localized impact on a particular region. For example, many of the events that occurred in Edo-period Japan were very localized events when viewed on a global scale. However, the social and historical events here are to be taken up and organized neither from the perspective of universality nor from the perspective of global to local ranking; rather, they are things to be extracted and organized from among a large number of events, from the perspective of this performative “I (watakushi)” character that exists within a specific, limited time and space. Thus, if we say they have a hierarchy of importance (even if this is no more than a simple reflection of the amount of description), it would mean that these events are organized from the personal and private perspective, which is distinct from the global–local or universal– specific perspective. In the author’s case, these social and historical events tend to be organized around narrative and literature, and below is a mixture of general historical nomenclature and period-specific characteristics of narrative, literature, and theater and the performing arts (it is particularly odd that the descriptions of each period include both parts, incorporating narrative and literature as well as theater and performing arts and their historical attributes). This may need to be reexamined and improved in the future. Furthermore, since there is a no definitive picture of a method for organizing social and historical events in the world, the author records only topics that are of interest to him below, in no particular order. A. Social and historical events (Japan) 1. The society and history of the Japanese islands and their surroundings prior to the genesis of written language—The Jōmon and Yayoi periods. 2. The birth of Chinese civilization and culture, particularly the period during which the Japanese language was born and explored as a system of written language. 3. The age of ancient ballads and myths. 4. The Heian period, which witnessed the creation and explosive development of Japanese poetry and “monogatari (narratives).” 5. Society in the age of the samurai, from the Kamakura period to the Muromachi period, which witnessed a multistage transition from monogatari to “setsuwa (legends)” and then to “katarimono (narrating arts).” 6. From the Warring States period to the era of the three revolutionary figures, namely Oda Nobunaga (1534–1582), Toyotomi Hideyoshi (1537–1598), and Tokugawa Ieyasu (1543–1616). 7. The Edo period, in which there was a sphere of stability and circulation. 8. The independent world of Japanese puppet theater (ningyō jōruri) and kabuki in the Edo period. 9. [Overview of history] The history and status of theater and the performing arts in Japan. 10. [Overview of history] The history and status of literature and narratives in Japan. B. Social and historical events (world) 1. The East Asian sphere centered on ancient China. 2. The South Asian sphere centered on ancient India. 3. The ancient Middle East. 200

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

4. 5. 6. 7. 8. 9. 10. 11. 12.

The European world from ancient Greece to ancient Rome. The golden age of China (i.e., the Tang and the Song dynasties). The Western world and the Mongolian Empire. India before colonization, and Southeast Asia and China. Islamic society. Modern Europe and colonialism. The existence of the United States of America. The existence of Africa. Social ideologies and their conflicts, i.e., democracy and socialism, capitalism and socialism, and liberalism and imperialism. (2) Two Directions of the Narrative Content of Watakushi Monogatari and Their Fusion: From Fluidity and Fixation to In Motion/Static

The concepts of fluidity and fixation are not related to narrative content; instead, they comprise an idea that is related to a narrative generation strategy. However, the author also applies these contrasting concepts to narrative content aspects as they relate to Watakushi Monogatari. For that reason, the author will modify the terms a bit, terming the narrative content concept corresponding to fluidity “in motion” and the term corresponding to fixation “static.” Motion and static are two polar directions relating to narrative content in Watakushi Monogatari, with the text’s final narrative content being created through their fusion. In this way, the terms “in motion” and “static” are both concepts that are involved in creating Watakushi Monogatari as well as polar concepts for controlling Watakushi Monogatari, so to speak. However, unlike fluidity/fixation, motion and static are involved in ascribing direction on a level that is close to the narrative’s content. What then are the respective characteristics of “in motion” and “static” in narrative content? In what ways are “movement” and “stillness” respectively realized in relation to Watakushi Monogatari? This is not the sort of question for which there is an objective answer, and so the author’s answer must be woven together in a form that is either in line with or constrained by his preferences. In this way, the closer one gets to a narrative’s content aspect, the stronger the need to weave together or create solutions based on one’s own personal and private interests and preferences. Although this is no more than an initial consideration, the author proposes the (personal and private) idea of linking “in motion” and “static” to “a picaresque novel-type narrative” and “a speculative novel-type narrative,” respectively. First, the author will explain how in motion = a picaresque novel narrative. When the author graduated from the various stories “aimed at children” (or arranged for children), such as fairy tales, folk tales, and mystery and science fiction novels, among others, the works that held the most appeal around the time the author began reading “adult” novels were those by Ernest Hemingway (1899–1961), André Malraux (1901–1976), and Takeo Arishima (1878–1923). It seems to the author that a kind of “adventure” (including in the symbolic sense) resides in each of their novels. The author feels that their novels depict characters who take an active role in terms of seizing control of their own life directions and taking action that reflects their own ideas, concepts, and intuitions. For example, in Arishima’s (1964) relatively short novel Umareizuru Nayami [Troubled Birth/The Agony of Coming into the World], we witness the emergence of a type of character who, rather than assuming other people’s affairs through, for instance, psychological confrontation and personal relations, resignedly tries to take life actions according to his fate, although his life changes fatalistically (in Oedipus (1972) by Sophocles, which Aristotle (384–322 BC) references in Poetics (Aristotle, 1997), the protagonist’s adversary is the same “fate”) as a result of 201

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

conflicts arising from his ideas, concepts, and intentions. This novel does not belong to the adventure genre, but the author considers it to be an extremely adventurous novel that tells an adventurous story, both at the time of its publication and still today. Aru On’na [A Certain Woman] (Arishima, 1995), which is by the same author, is a similarly adventurous novel. The author also perceived For Whom the Bell Tolls (1994), The Human Condition (Malraux, 1990), Gone with the Wind (Mitchell, 2014), and Wuthering Heights (Bronte, 2002) as adventurous novels with adventurous narratives. When adventure or adventurous novels and stories are considered in this broad sense, just about any novel or story can be viewed as one or the other. There is a particularly strong trend of saying that novels and stories that evince great strength, so to speak, were so from the outset. Although the author thinks that this idea is partially correct, the characteristics of the author’s concept of this type of adventurous novel or story can ultimately also be seen in kabuki and puppet theater (particularly in the Edo period) as well as in various Yukio Mishima’s novels (e.g., Ao no Jidai [The Age of Blue] (Mishima, 1950), Kinjiki [Forbidden Colors] (Mishima, 1951, 1953), Kinkakuji [The Temple of the Golden Pavilion] (Mishima, 1956), Honba [Runaway Horses] (1969b), etc.). From the rather broad view that all novels and stories are adventurous ones, the author would like to briefly consider a slightly more limited view, that is, just what (in his estimation) are adventure novels or adventure stories? When considering this, the term “picaresque novel” comes to my mind. In a narrow sense, this refers to a type of novel that was popular in Spain in the 16th and 17th centuries, and the author shall expand on the concept a little here, treating it as a “picaresque novel or story.” Under this broad definition, there are many picaresque novels and stories among the literature from other periods and regions. For example, as stated above, Mishima’s Ao no Jidai, Kinjiki, Kinkakuji, and Honba all seem to the author to be picaresque novels and stories. The protagonists in these novels include some who are not so-called “bad guys” and who all try to act and go about their lives according to their own beliefs, ideas, and outlooks. These beliefs, ideas, and outlooks are, in many cases, not the kind that other people and society at large admire; on the contrary, they are the kind that are reviled and hated. Many of Seichō Matsumoto’s (1909–1992) novels (he wrote several Japanese mystery novels, in the broad sense of the genre, that many people still read, such as Yakō no Kaidan [Stairs That Shine at Night] (Matsumoto, 1961), Linkage [Renkan] (Matsumoto, 1962), Warui Yatsura [Bad Sorts] (Matsumoto, 1980), Kurokawa no Techō [Pocketbook of Black Leather] (Matsumoto, 1981), etc.) are more directly picaresque. These Matsumoto novels’ protagonists are criminals, but they each have a strong will to act and live their lives on the basis of their beliefs and ideas. Moreover, in many cases, the detective role, which the typical mystery novel’s protagonist would fill, is relegated to minor characters in these novels, with the reader’s strongest emotions being reserved for the criminal protagonist instead. When the author considers the characteristics of those novels and stories that he deems picaresque, the author devises character categories such as “the unrepentant character” and the “character who rushes in.” These traits make for appealing characters. The unrepentant character, in particular, is heavily influenced by kabuki. Kabuki characters are significantly stylized; heroes are heroes and bad guys (villains) are bad guys (villains). Moreover, in kabuki, the bad guy (villain) is central to the play, with the best actors often playing those roles. Indeed, in kabuki, unrepentance is a villainous characteristic; hence, the bad guys’ actions and lives are villainous to the end. In many cases in kabuki, the bad guys (villains) are ultimately punished with tragic deaths, yet audiences cheer and applaud or show enthusiasm for them for their consistent behavior as bad guys (villains). As character traits, unrepentance and rushing in are more important to the author than classifying other character traits, such as whether a character is active and socially appealing or, conversely, whether a character has no social appeal whatsoever and is 202

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

repellent. Since the protagonists in the original picaresque novels are so-called villains, naturally, they are shunned by society, and they bear a black mark, but the author is desirous of seeing unrepentance and the headstrong character trait paired with the so-called “B-grade” or “averageness” and “(though contradictory) the characteristics of a character that does not stand out in one’s memory.” A protagonist who is a wholehearted “bad guy” is a reversal of the typical protagonist characterization as a perfectly sensible, respectable person; that is, the two are alike after all. The author would like to see a protagonist character who is simultaneously a “bad guy” and an ordinary, average person. The picaresque novel (or story) the author is seeking is, as described above, one that balances the “unrepentant, headstrong” characteristics with those of the “ordinary, average” person. The author will now discuss another characteristic, namely “static,” that the author is seeking in Watakushi Monogatari’s narrative content. Static is symbolized by speculative novel narratives. As used here, the term “speculative novel” is the author’s own; that is, it is distinct from the science fiction genre, referring instead to novels and stories that contain a substantial amount of or focus on philosophical thought or critical discourse. Static = speculative novel narratives. Broadly speaking, by “speculative fiction,” the author means novels and stories that feature a character who is opposite to those in the picaresque novels that were discussed above. That is, if in a picaresque novel, the protagonist’s fundamental strategy involves taking action, rushing in, making breakthroughs, and maintaining a non-reflective psychological state, the speculative novel narrative is without action and is therefore reflective and static. Speculative fiction stories’ protagonists are motionless, static, and stagnant, constantly reflecting and indulging in internal thoughts. If picaresque novels’ protagonists “act without thinking,” then those in speculative fiction stories “think without acting.” However, if we consider the behaviors that are depicted in picaresque novels to be external, while the actions in speculative fiction are internal, the difference between the two can be attributed to a difference in behavioral direction; taking the above picaresque and speculative fiction narratives to be “external picaresque novel narratives” and “internal picaresque novel narratives,” respectively, these distinctions may constitute a broad internal classification of picaresque novel storylines. Of course, this is the author’s idea. We can conclude here that speculative fiction stories, which the author has called external picaresque novel narratives, are narratives in which the protagonists’ internal actions, that is, their thoughts, are extremely active and unreflective. Which novels or stories, then, does the author feel correspond to speculative fiction or internal picaresque novel stories? Japanese author Yutaka Haniya (1909–1997) wrote only a few novels; his most voluminous works were essays, in which the most important topics pertained to the relationship between politics and humanity. Shirei [Death Spirits] (Haniya, 1998) is among Haniya’s novels; he wrote it intermittently, starting in his youth and continuing into his later years, but he never finished it. Shirei is set in pre-Pacific War Japan, and several characters are developed through ideological dialogue and philosophical considerations about humanity, the universe, and politics. In his novels and stories, Haniya proposes philosophical concepts such as jidōritsu no fukai (the discomfort of the law of identity) and kyotai (the void). Shirei is contemporary Japan’s most important work of speculative fiction. Other Japanese novels, especially those from the contemporary era, including Hiroshi Noma’s (1915–1991) (1983-1984) Seinen no Wa [The Ring of Young People] and Yukio Mishima’s Kinkakuji and Utsukushii Hoshi [A Beautiful Star] (Mishima, 1962), seem to fall into the speculative fiction or internal picaresque novel stories category. Furthermore, Yukio Mishima’s Taiyō to Tetsu [Sun and Steel] (Mishima, 1968) represents a new speculative narrative style, seemingly fusing the novel, essay, and poem forms. Seinen no Wa is an extremely long novel that is set in Osaka, Japan, before the Pacific War. The plot develops extremely slowly and somewhat redundantly, with multiple protagonists engaging in a great many philosophical 203

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

discussions about humanity and political issues. In Mishima’s Kinkakuji, the introverted, unsociable protagonist creates his own unique, dogmatic notion, and a philosophy of “beauty” based on his own complexes and sense of inferiority; on the basis of this, he then sets fire to the real-life Kinkaku (Golden Pavilion), which is a symbol of corrupted beauty, and in the final scene, he feels a sense of bliss watching the Golden Pavilion burn. Mishima’s Utsukushii Hoshi has not been translated into English, but a magnificent cosmology of UFOs unfolds through several characters’ speech, damaging the novelic structure. Among modern and contemporary literary works outside of Japan, Victor Hugo’s (1802–1885) (1982) Les Misérables, Herman Melville’s (1819–1991) (2009) Moby Dick, Fyodor Dostoevsky’s (1821–1881) Crime and Punishment (Dostoevsky, 2014), The Brothers Karamazov (Dostoevsky, 2003), etc., Thomas Mann’s (1875–1955) (2005) The Magic Mountain, and Hermann Hesse’s (1877–1962) Demian (2013), Steppenwolf (2015), etc., all seem to fall into the category of stories the author calls speculative fiction or internal picaresque novel narratives. In Les Misérables, the narrator pauses the plot (for quite some time) to observe the development of external events, often providing lengthy commentaries on particular topics (especially the monastery and the sewers). This novel is closer to “explanatory” than “speculative” (since it offers an overwhelming amount of knowledge or is concerned with presenting details or special knowledge about specific topics), and the author would like to cite this as a special form of speculative fiction or internal picaresque narrative. Hesse’s Demian discusses the possibility of a dualistic ideology and philosophy similar to ancient Persia’s Zoroastrianism through dialogue between the young boy Demian and the narrator, Sinclair, which helps Sinclair overcome the spiritual crisis of his youth. To summarize the above, what the author seeks from the narrative content of Watakushi Monogatari, as well as the policy or strategy at the macro-level of its production, is, on the one hand, a picaresque novel narrative or an external picaresque novel narrative, and, on the other, a speculative fiction story or an internal picaresque novel narrative. Moreover, the author seeks a story in which both these characteristics are present. When considering such a coexistence, the author is reminded, for example, of Yukio Mishima’s Kinkakuji, which was mentioned above. This is a type of picaresque novel in which the protagonist sets fire to the Golden Pavilion in a grand, kabuki-like event. More broadly speaking, as a whole, it is also a kind of adventure novel (in which as a player in an adventure game the protagonist achieves the goal of setting fire to the Golden Pavilion in the final scene after a series of trials and errors). At the same time, the protagonist is a brooding, introverted boy/young man who is tormented by various complexes, including one relating to his appearance and a sense of inferiority. He never receives others’ approval, and since he is the protagonist, his thoughts and ideas, which could be called philosophical, are interspersed in various episodes throughout the story. When read, this novel assumes the form of memories emanating from the protagonist’s “I,” and, as readers, we wonder how this gloomy, introverted, uneducated young man is able to conceptualize and express such fine philosophical thought in writing. To the author of this chapter, Kinkakuji is among a small number of novels (narratives) that fuse the characteristics of a picaresque or external picaresque novel narrative and a speculative fiction or an internal picaresque novel narrative. We can also point to Seinen no Wa as another example of this fusion in Japanese novels. Besides this novel’s protagonist, Taguchi, who is a young man with a crippled leg and an unbalanced personality, plays a major role. In his booklet about the zentai shōsetsu (total novel), Inoue (2015) discusses Mishima’s Hōjō no Umi [The Sea of Fertility] (1969c-1971) (Haru no Yuki [Spring Snow] (1969a), Honba [Rusnaway Hoses] (1969), Akatsuki no Tera [The Temple of Dawn] (1970), and Ten’nin Gosui [The Decay of the Angel] (1971)) and Hiroshi Noma’s Seinen no Wa as two representative examples of contemporary Japanese novels that are total novels or that have a strong orientation toward being complete novels; however, here, the author has unintentionally employed these 204

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

two novelists’ works as representative examples of “picaresque novel stories or external picaresque novel narratives and speculative fiction stories or internal picaresque novel narratives” in contemporary Japan. While it can be said that the two authors contrast starkly in terms of their respective novels’ styles and ideas, as Inoue pointed out, they can be considered to share an unexpected commonality with respect to novelic methodology. Of course, what the author has described above is ultimately only a policy or a strategy for creating the narrative content in Watakushi Monogatari, and it does not mean that the narrative content uses or borrows from the stories that are told in Kinkakuji, Hōjō no Umi, or Seinen no Wa. In that sense, the kind of policy or strategy for narrative content that the author describes here should be classified as a relatively formal one among the elements relating to Watakushi Monogatari’s narrative content. In practice, Watakushi Monogatari’s specific narrative content is to be invented independently, by the author, using the abovementioned materials, subject matter, and automatic narrative generation function.

The Conception of Form in Internal Narrative Generation Form in stories such as novels is a highly intriguing subject. When the author first read the Japanese translation of James Joyce’s (1882–1941) (2010) Ulysses, what struck the author most was the form in which it presented adventure and the variety of that form. One of the translators was the novelist Saiichi Maruya (1925–2012). Maruya was able to translate one chapter, in which Joyce delivers narrative content while tracing the history of the English language from ancient English up to contemporary colloquial English, into Japanese by tracing the history of Japanese expressions from ancient Japanese to medieval and early modern (Edo-period) Japanese, the Meiji period, and finally, contemporary colloquial Japanese. What lingered in the author’s memory was not so much the content of the story as the story’s expression and the intensity of the awareness of form that made such an adventure possible. The final chapter is written using a formal technique known as stream of consciousness. The stream of consciousness technique was attempted prior to Ulysses in Natsume Sōseki’s (1867–1916) (2014) Kōfu [The Miner]. Writers whose work the author read around that time, including William Faulkner (1897–1962), Yasunari Kawabata (1899–1972), Kenji Nakagami (1946–1992), etc., and Riichi Yokomitsu (1898–1947), whom the author read later, were all highly aware of novelic and story form. At the same time, Yukio Mishima’s classical form is also a kind of formal adventure, going in the opposite direction. For example, Mishima’s novel Shiosai [The Sound of Waves] (Mishima, 1954a) deliberately imitates the Greek novel Daphnis and Chloe (Longus, 2019), and several of his kabuki works also imitated the classical kabuki style of the Edo period (the current custom in kabuki performances is to use jōshiki-maku (drawing curtain of three colors) for so-called classical Edo-period kabuki and to use donchō (drop curtain) for dance and modern or New kabuki since the Meiji era; however, performances of Mishima’s Iwashi Uri Koi no Hikiami [Sardine Vender Love’s Trawl] (1954b) written in a pseudo-classical style, use the same jōshiki-maku as in classical kabuki). Mishima also wrote many other works, incorporating formal techniques such as those used in French psychologism novels and classical drama. Narratology was originally influenced by Russian Formalism, which is one of the sources from which it arose. It emphasized aspects of form in stories and narratives, and argued that innovation in form is equivalent to the innovation of a narrative. Natsume Sōseki’s (2007) Bungakuron [Literary Theory] deals with narrative content in literature (mainly novels, poems, and plays), but its handling is very formal. That is to say, Sōseki does not directly criticize literary content (i.e., characters’ personalities and ideas, temporal and spatial characteristics, etc.) as is usually the case with literary criticism; rather, he formally categorizes the literary content, 205

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

so to speak. In that sense, he takes a very formal approach. Takaaki Yoshimoto’s (1965) Gengo ni totte Bi to ha Nani ka [What is Beauty in Language?] also engages in a historical discussion of the question of the evolution of literary works based on the formal characteristics of the language in which they are written. Various other examples can also be mentioned, but it is clear that form is an extremely important topic in literature. However, when the author speaks of formal concepts here, he does not suggest that “form” only includes those literary works that are oriented toward the form of adventure. It goes without saying that formality exists equally in works that are avant-garde and those that are not. Otherwise, neither novels nor stories could possess substance. In the broadest sense, form in narratives encompasses all the methods by which narrative content is actually expressed and can be seen equally in any masterpiece as in any slipshod work, and in any artistic as in any popular one. In literature, that is oriented toward the venturous form or the diversity of form mentioned above, special attention is paid to the question of mediating, so to speak, the value assessment of narratives or literary forms. Preceding this type of value assessment however, there is the question of form as a more universal subject. Naturally, the form the author is referring to here, and in a more substantive sense toward Watakushi Monogatari, is form in the sense of being in line with the specific realities of the work. Form in the sense of what form the narrative work to be generated takes is form at the level of the work itself. In light of the above, the author will separate the discussion into “higher-level” and “lower-level” form, based on this issue or awareness of it in narratives and literature. Here, lower-level form is form in the sense of encompassing all forms that must be present in all literature. Higher-level form, by contrast, encompasses issues of the conscientious use and application of and experimentation with form, such as superior, adventurous, avant-garde, and (intentionally used) classical forms in narrative and literature. Note that the terms higher and lower do not imply value or significance. Lower-level suggests the sense of the foundational formal foundation, so to speak, while higher-level suggests the sense of conscientious use and application based on it. As preparation for full-scale systematization in the future, the author will now discuss lower-level form, meaning all the methods and techniques that are used in narratives and literature to actually express and convey narrative content. The (lower-level) form in narratives can be further separated into two levels. The author will use the example of a novel, since narratives are expressed entirely in language (note that various issues and difficulties arise when we make such a consideration with regard to other types of narratives that, unlike the novel, are not expressed entirely using language, such as theater and the performing arts). One is form on the level preceding individual sentences, and another is form on the level of the set of sentences. Using the INGS that the author has developed, the generative process is separated into three parts: the story generation process, the narrative discourse (generation) process, and the narrative representation (generation) process (particularly the linguistic representation (generation) process in relation to this discussion). The narrative discourse generation and the narrative representation generation processes are relevant to form as discussed here, and it follows that “form at the level preceding individual sentences” above corresponds to form for narrative representation (“form as narrative representation”), while “form at the set of sentences level” corresponds to form for narrative discourse (“form as narrative discourse”). The first, form as narrative representation, deals primarily with form in relation to sentences, words, letters, and sounds. What rules and methods exist within these and, moreover, what manner of information is present depends, of course, on the type of language. The main language the author uses in Watakushi 206

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Monogatari is Japanese. In Japanese, words and sounds are, of course, different from other languages such as English, but there are many other differences. For example, it is safe to say that Japanese lacks English’s strict syntax rules. In Japanese sentences, it is the “particles” or “te-ni-o (wo)-ha” that often play a major role in establishing meaning, while the order in which the words in a sentence are arranged is irrelevant to establishing the sentence’s core meaning. Furthermore, while in English a vast number of words can be created from a small number of characters, there is an extremely large number of characters that are used in Japanese, particularly kanji, as well as other types of characters, such as the hiragana and katakana alphabets, and Roman characters, which are also present. Japanese sentences are mixtures of these different character types. In considering the form that is used to express narrative content in Watakushi Monogatari, we must examine these characteristic features of the Japanese language and devise a formal system based on it. However, English and many other languages can be considered to share the use of sentences, words, letters, and sounds as the main elements of narrative expression. The author would now like to consider two vectors of form, “the form of narrative expression unrelated to the semantic content of the narrative” and, conversely, “the form of narrative expression related to the semantic content of the narrative.” For example, “sentence length” is a form of narrative expression that is extremely important, with some writers, such as Hemingway and Yasunari Kawabata, frequently using short sentences, while others, such as Marcel Proust (1871–1922) and Yutaka Haniya, often compose long ones. Riichi Yokomitsu’s short novel Kikai [Machine] (Yokomitsu, 1931) is also written primarily in extremely long sentences, although Yokomitsu is famous for changing his style of narrative expression forms in each new work. This demonstrates that sentence length is an essential element that creates stylistic differences between writers. On the premise that stories’ styles are included in their meanings, the form individual sentence length is also involved in meaning; however, this differs from narrative expression forms in the sense of, for example, “what words one uses to describe roughly the same things at the story or event level.” In the latter, every word retains its own particular meaning, and word choice corresponds to the form of narrative content on a more semantic, content-related level. In relation to this, the different forms relating to character types in Japanese, such as whether a word is expressed in kanji, hiragana, katakana, or using the Roman alphabet, create a large, essential characteristic in written narratives, but this kind of character selection is a form of narrative expression on a level that is further removed from semantic content than word choice. Thus, “involved or uninvolved in the semantic content of the narrative” is perhaps a comparatively binary concept, and while a strict dichotomy is impossible, this kind of distinction can certainly be made. The next category, that is, the narrative discourse form, is, in the simplest terms, a level of form that focuses on the composition as a set of sentences, as described above. However, it is extremely difficult to clearly and comprehensively consider and organize its true state. Nevertheless, attempts to accomplish this, including the task of integrating with the form at the level of narrative expression above, are essential in aiming for a true system of narrative discourse theory (in the sense of being integrated with narrative expression) that extends well beyond the scope of Genette’s (1972) narrative discourse theory. This can be demonstrated with an example. Suppose that, in a given story, we are handling “A scene from the day of the Great East Japan Earthquake” and “Events on the day of Mishima’s ritual suicide” (both of which are episodes). Form on the narrative discourse level discussed here can be seen in the following examples: “The events of the day of the Great East Japan Earthquake are narrated through the narrator’s evocation of memory in the present time and space, through the order in which the narrator recounts their memories,” and “The events of the day of Mishima’s ritual suicide are narrated based on a stance of objective coverage of the event following the passage of time and with the aid of newspaper articles 207

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

and other articles from the time.” It becomes clear that there are differences in the form of narrative discourse, such as certain events in the story being narrated as organized by memory, and narration using a style of incident reporting that follows a chronological order. Moreover, it is possible to go on to list numerous examples of this form of narrative discourse one after the other, such as the number of times the incidents or events in the abovementioned materials are expressed in the narrative overall, whether through brief summary descriptions or using lengthy descriptions, etc. A more definitive level of the form of narrative discourse appears next. For example, these include themes relating to the expressive style of narrative discourse, such as description, conversations, mental representations, etc., and themes relating to depicting actions (description), explaining external and internal character attributes, and providing exposition through descriptions, topics, and objects. In determining the various elements for the form of narrative discourse, each of its parts will be superficially realized on the narrative expression form level. For example, characters’ actions in descriptive passages are expressed in short sentences using complex words, and conversations are expressed in long sentences using only hiragana. Conversely, in a more macroscopic direction, there is also form in the sense of the “composition” of each relatively large block that constitutes the whole narrative or a part thereof, as in a novel. Ulysses is, on the one hand, a novel that takes an avant-garde form, but in many respects it is a novel that adopts a classical form that was created based on the structure of Homer’s (2006) epic Odysseus; that is, it was constructed using a form that fuses the two. Moreover, this novel was also created using a variety of formal structural principles that vary almost from chapter to chapter. However, if we view the individual chapters that constitute it as a whole, each is carried through by a consistent formal principle. In this sense, there are innumerable possibilities for composition when viewed from a macro perspective, but the extent of the temporal scope in novels of equal length, for instance, will be reflected in the narrative’s characteristics. Ulysses, The Brothers Karamazov, Shirei (Death Spirits), and Seinen no Wa are novels that unfold within a relatively short span of time and are composed of a condensed set of tightly interconnected episodes. On the other hand, Genji Monogatari [The Tale of Genji] (1993, 1994, 1995, 1996, 1997), War and Peace (Tolstoy, 2010), Jean-Christophe (Rolland, 2011, 2018, 2007), and Hōjō no Umi (The Sea of Fertility) are narratives that unfold over an extremely long span of time, depicting a person’s life, or at least half of their lifetime, and even generations of people, through a loosely connected series of incidents or episodes. Find below a list of the major categories above as well as the main formal elements that might be included within each. It should be noted, however, that this list is only tentative, containing a few examples that attempt to encompass the narrative form: A. Forms of narrative representation ◦◦ Sentence length ◦◦ The amount of conversation and description in a sentence, and whether these are many or few sentences as well as how they are controlled ◦◦ Conversational response duration (long or short) ◦◦ Paragraph length ◦◦ The frequency at which kanji are used in the text ◦◦ Similarly, the frequency of hiragana’s use ◦◦ Similarly, whether (or not) a lot of katakana is used ◦◦ Similarly, the frequency at which Roman alphabet characters are used ◦◦ The distribution of character types, i.e., kanji, hiragana, katakana, and Roman alphabet characters, in the text 208

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

◦◦ Whether complex words are used (frequently or infrequently) in the text ◦◦ How heavily (or lightly) a sentence is punctuated ◦◦ The types of word endings used in a sentence ◦◦ Is the text written in literary or spoken language? B. Forms of narrative discourse ◦◦ How is the timeframe of the story handled as the narrative unfolds? Does the narrative develop through the story or is the narrative structured in an order that is different from the story’s? ◦◦ To what extent is the “narrator” actualized in the narrative and to what extent do they have power? Is the narrator visible or invisible? Is the narrator strong or weak? ◦◦ Is the development of the narrative scene- or summary-centric? ◦◦ How long (or short) are the scenes in the context of narrative development? ◦◦ As a method for segmenting the entire narrative, divide it into several “sections” or, alternatively, “chapters,” divide it using numbers, etc., divide it without using symbols (a new paragraph, etc.), undivided, etc. ◦◦ How many parts of speech, such as nouns, verbs, adjectives, adjectival nouns, etc., are included, and how are they balanced? ◦◦ Is there a lot of (or a little) description? ◦◦ Is there a lot of (or a little) explanation? ◦◦ How detailed is each description? Is it relatively detailed or rough? ◦◦ What is the length of each description? Is it relatively long or short? ◦◦ How detailed is each explanation? Is it relatively detailed or rough? ◦◦ How long is each explanation? Is it comparatively long or short? ◦◦ How does personhood, i.e., first-, second-, and third-person, and others (e.g., Riichi Yokomitsu introduced the concept of the fourth-person in his novel Kikai (Machine)). ◦◦ How many characters appear, and how many of them are major characters? ◦◦ Is there a protagonist? ◦◦ Is there a single protagonist or are there several? ◦◦ Does the character (the protagonist) have a broad (or a narrow) scope of action? ◦◦ For how long does a character (e.g., the protagonist) appear in the narrative? ◦◦ What personality does the character (the protagonist) have? ◦◦ To what extent are material elements featured? ◦◦ To what extent are spatial elements (places or stages) featured? ◦◦ Is there much movement between locations? ◦◦ To what extent are abstract elements featured? ◦◦ To what extent are emotional components featured? ◦◦ From what point of view, i.e., the single (internal), omniscient, or dynamic (non-fixed) perspective, etc., does the narrative develop? ◦◦ Does the narrative have an overall comedic or tragic effect? ◦◦ Is the narrative as a whole mundane or extraordinary (fantastical)? ◦◦ Are animals or inanimate objects anthropomorphized in the narrative? ◦◦ Are humans animalized or objectified (reverse anthropomorphism) in the narrative?

209

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

In contrast to the lower-rank narrative forms described above, there are high-order narrative forms. This is related to the so-called “adventurous form” in Watakushi Monogatari, and it can be said that this adventurous form was one of the major objectives of Watakushi Monogatari’s creation; although higherlevel narrative form is an important topic, this is something the author wishs to discuss in a separate paper.

Toward External Narrative Generation: The Sociological Distribution of Watakushi Monogatari In the previous section titled “Internal narrative generation,” the author addressed the issue of the Watakushi Monogatari content that should be written in the future or, more specifically, the materials that should be used to prepare for this content’s creation. In this sense, the discussion has yet to address the full narrative content, but in Watakushi Monogatari, the generation and distribution processes are separate topics. The generation process is naturally performed through GIS with INGS (as described Figure 13. The fluidity and fixation in Watakushi Monogatari or the narrative of possibility and the narrative of fixation

above) and the Watakushi Character that overlaps this “I (watakushi).” Meanwhile, distribution is accomplished through GIS and other distributive media. However, in Watakushi Monogatari, the generation and distribution processes are not completely distinct; rather, the two are interrelated. The author now withes to explain this interrelatedness in Watakushi Monogatari. As a precondition for clarifying the interrelated generation and distribution processes in Watakushi Monogatari, Figure 13 demonstrates fixation and fluidity. In Figure 13, the “narrative possibility (fluid state)” is in constant motion. From there, the “fragmentary fixation of narrative” occurs, at times, in various forms. This process is called “primary fixation.” Moreover, from there, or possibly directly from the fluid narrative, a “systemic fixation of narrative” arises. This process is called “secondary fixation,” but this can be further separated into “secondary fixation A” and “secondary fixation B.” Furthermore, fragmentary and systemic fixed narratives forward

210

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Figure 14. The creation and distribution of Watakushi Monogatari and their cyclical process

information back to “possible narratives (fluid state)” on various levels. In other words, this process gives rise to a great cycle. Here, part of the fixation of narrative, in any form, escapes the generation process and enters the distribution process. We can try tweaking this slightly to accommodate a diagram that incorporates the three elements: the generation and distribution processes, and the abovementioned circulation (Figure 14). In Figure14, the process of narrative generation (fluid state) produces a fixed work. Some never enter the actual distribution process, while others will. However, these processes are not unidirectional. In other words, once a narrative is fixed or once a narrative has entered the distribution process, it can return to the generation process as a type of material. Fixed narratives that return play the same role in the generation process as other materials and resources that the process utilizes. In this way, the generation process and fixed narratives and the generation and distribution processes form a reciprocal, cyclical process. The final stage of this process is the complete, systematic Watakushi Monogatari. The external directions of narrative generation or external narrative generation can be simply interpreted the distribution from the production or creation through an internal process. However, as shown in the two

211

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

figures above, Figure 13 and Figure 14, the Watakushi Monogatari method includes a cyclical process comprised of generation and distribution, and the narrative generation process can also be included in external narrative generation. Alternatively, external narrative generation should be understood in terms of the cyclical relationship between the narrative generation and the narrative distribution processes. Incidentally, the author mentioned above that “The final stage of this [cyclical] process is the complete, systematic Watakushi Monogatari.” Since Figure 13 focuses only on the cyclical process of narrative production and circulation, the distinction between the work as a work-in-progress and the final work is not made explicit. In Figure 14, however, a distinction is made between a “fragmentary fixed narrative” and a “systematic fixed narrative,” and a distinction can be made between the former as a narrative-in-progress and the latter as a final narrative. This shows that, on the basis of the two reciprocal foundational concepts of fluidity and fixation, the distinction between a work-in-progress and the final work becomes less apparent; however, at the same time, the two must, of course, be distinguished from each other. There is, however, likely no essential difference between the two with regard to the internal logic of the narrative. At a certain point in the Watakushi Monogatari generation process—this is probably the point at which the idea of “completion” carries a great deal of weight in the author’s mind—the final form of Watakushi Monogatari is transformed into a book (in this case, “book” is not limited to printed works but is more symbolic in nature). In this sense, in the final “book” form, at the point of conservative fixation, the narrative of the purely fluid process (perhaps Watakushi Monogatari in “the fluid process as a generative process”) is ultimately fixed to form Watakushi Monogatari. An important consideration in relation to personal strategy is the degree to which we bring that to a point in time or, to put it another way, the place to which we bring the “end of the narrative.” The question also arises as to how the narrative, as a final, fixed work, should be distinguished from the narrative as a fixed work in the course of narrative generation. In this regard, the author proposes that there are two types of fixation: fixation that is performed through the author own intent and fixation that is done for other reasons. For example, there is a form in which narrative is fixed, where the recipient takes some manner of action within the vortex of fluid narrative generation and, for Watakushi Monogatari, fixation is performed deliberately by a part of the Watakushi Character other than the real “I (watakushi).” Alternatively, we can also consider that fixation during the process and fixation in the final work are realized on different scales. When considering this issue, while there is the idea of calling the final fixed narrative Watakushi Monogatari and referring to the in-progress narrative by a different title, there is also the idea of considering the scope that encompasses both to be Watakushi Monogatari and the in-progress narrative to be a special part of the same Watakushi Monogatari. Adopting the latter view, the final fixation refers to one large work in its entirety, the Watakushi Monogatari, whereas the fixation of the work-in-progress refers to pieces, fragments, and parts of the narrative related to Watakushi Monogatari on a smaller scale. In other words, in the course of generating the Watakushi Monogatari, independent, fixed narratives that are smaller than the final Watakushi Monogatari arise, and these are differentiated from Watakushi Monogatari’s main body in the form of, for example, “short narratives” or “narratives of medium-length (novellas).” Of course, this or part of the source of a type of narrative from which that fixed narrative originated can in some ways influence parts that should again be part of the Watakushi Monogatari as a whole. Incidentally, fixed narratives realized through the medium of (paper) books where they go beyond the fluid process are an important form to the author, and the possibility of the circulation and development of narratives related to Watakushi Monogatari in the form of “paper books” cannot be excluded. In this case, the fixed narrative that enters the distribution process will be published as various books. At the 212

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

same time, the author believes there is also a need to extend this concept of the book to include books in a sense that goes beyond the antiquated form of paper books to include “superbooks.” A superbook is the form that a fluid narrative takes when it becomes a fixed narrative, but additional fluid elements are incorporated in its fixation, and this form is the idea the author means to describe with the term “superbook.” A fixed narrative can be fixed not only in the traditional form of a paper book, but it can also be realized as digital books, which are fluid in themselves. In such cases, the boundary between fixed and fluid narratives becomes extremely blurred or (possibly) extremely creative, fluid, and ambiguous. The author would like to further consider this notion of a superbook. In its simplest form this would be a “hypertext schema” in such stories where the text of the final work is linked to other parts of the work, annotations, explanatory text, etc., through the use of hyperlinks. The simplest form of this hypertext narrative scheme––with links to annotations and explanations––is also achieved in paper books in the form of annotations (and annotations of annotations). Some novels are heavily annotated. For example, the Japanese novelist and politician Yasuo Tanaka (1956–) included numerous annotations on various things that were popular at the time at the end of his book Nantonaku Kurisutaru [Somehow Crystal] (Tanaka, 1981). A more complex “hypertext schema” can also be considered. For example, this could be a form in which one part of a narrative is connected with another part of the same narrative through links. For example, there may be intra-narrative relationships, such as a particular word being connected to the same word or an associated word in another section or a certain event may be connected to the same or an associated (single) event in another part, or a particular structural unit may be connected to the same or an associated structural unit in another section, etc. There may also be various other relationships. If we can say that various readings of a single narrative become possible through these relationships, we will have a complex version of a novel, like the type represented by Cortázar’s (1914–1984) (1991) Hopscotch. The above examples evince methods based on internal textual relationships within a single fixed narrative work. By contrast, we can envisage a situation in which behind a fixed narrative work another type of information is associated with it. This is possible when there is a mass of connected information behind a fixed narrative work. For example, if the final output of the narrative generated from INGS is (in the case of novel-type narrative generation) language, where a conceptual information group is present behind the narrative’s superficial language, conceptual information that is, in some sense, related to that superficial language can be extracted. In other words, this means a multi-layered narrative. However, this is not multiplicity in a narrative that is directly expressed in language (as is the case for multiplicity in traditional narratives); rather, this is a form of narrative multiplicity in which there is another layer or multiple layers of narrative hidden behind the narrative that is expressed via language which, while, of course, related, exists on a different level. For hypertext mechanisms in ordinary narratives, links point to another part of the narrative that is expressed directly using language; however, in the sense of the superbook that is described here, links point to specific locations behind the narrative, so to speak, which exist on a different level than the point at which the narrative begins. Links that point to a level other than the narrative’s include, among their types, annotations in and of themselves, but in the sense that we are discussing here with respect to a different level of narrative, these are not the narratives behind the narrative. This means that as a superbook Watakushi Monogatari incorporates multiple relationships between the multiple background narratives that may be present. (Taking the example of a narrative and the annotation of its individual parts and supposing that the individual annotations are based on another large narrative chunk that organically systematizes them, we come close to the idea of a superbook, as the term is used here.) The entire work, including background narratives, is Watakushi Monogatari as a whole, but the background narratives cannot be directly viewed from the surface. However, it is also 213

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

true that one or more underlying narratives can be produced as similar to and in conjunction with the narrative that is visible from the surface. Perhaps these underlying narratives could be a collection of fragments or a mechanical chronicle that lacks a normal narrative’s structure, and thus cannot be called individual works in their own right. As a superbook, then, Watakushi Monogatari not only allows for a hypertext scheme between superficial linguistic expressions, it also enables a type of relationship between those superficial linguistic expressions and a deeper narrative structure, i.e., underlying narratives. The latter signifies one aspect of the multiple narrative structure in Watakushi Monogatari. This also hints at “fixed narrative = the issue of a fluid narrative appearing in the final work.” In this sense, in relation to Watakushi Monogatari, there is the opportunity for “fluidity and fixation” in all stages of narrative generation (including the final stage). It is also present at the final stage. Since such a form is unlikely to be fully expressed in the conventional “book” form, a superbook is needed, and it is essential to explore the possibilities of such a form. (With regard to multiplicity in Watakushi Monogatari, one can imagine that many superbooks are constantly being written in a fluid manner emanating from a single, extremely condensed superbook built into GIS with INGS.)

An Integrated Method Between Internal and External Narrative Generations The next section designs a basic mechanism of the external narrative generation that organically distributes narrative contents created by the proposed internal narrative generation mechanism. In particular, the author focuses on the integrated relationships between the internal narrative generation and external narrative generation to progress this research to an actual attempt. As is clear from the above, internal narrative generation, that is, the act of generating (producing) a real narrative work, and the act of external narrative generation or the external presentation (the distribution and circulation) of that generated (produced) narrative are not two separate, distinct processes within the Watakushi Monogatari framework of narrative generation research; rather, the two processes are very closely related to one another. Moreover, it is important to note that the two are quite intrinsically—that is, in a way that would not be possible without the use of such a framework—linked. In the course of Watakushi Monogatari’s creation, there can be various forms of fixed narrative prior to the final fixed state. However, while remaining fixed, this state is simultaneously involved in various ways in the creation of Watakushi Monogatari and the final work. For example, where, in the course of Watakushi Monogatari’s creation, a part or the current form is fixed in the form of a “book,” although this certainly represents the fixation of a narrative at that point in time, at the same time Watakushi Monogatari and that “book” may have some relationship as something that continues to change. For example, the fixed narrative (as pieces or fragments of a larger whole) can rejoin the Watakushi Monogatari generation process as a fluid subprocess, changing its fixed form, and influencing various parts of the fluid narrative. Considered in this way, the concept of fixation here is not necessarily associated with the concept of an “end,” but is rather established in very close connection with the concept of a fluid process. In other words, the narrative generation process surrounding Watakushi Monogatari is not simply a fixed process in the sense that the overall Watakushi Monogatari process ends when fluidity becomes fixed. If we consider this more pragmatically, that is, if we consider the question of the specific process by which the author’s “I (watakushi)” becomes the Watakushi Character and creates a narrative, we can publish fixed fragmentary or fragmented narratives from time to time during this process as fixed media, for example, as books. However, this is not simply a case of combining many such books to construct or 214

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Figure 15. From GIS with INGS and kabuki methods to Watakushi Monogatari

restore the whole. Those pieces, fragments, and fragmentary narratives are not later incorporated as-is into the overall Watakushi Monogatari; fixation promptly moves to a fluid phase, and the fixed narrative enters a phase in which it is again changed and used in the fluid process. The phenomena of internal narrative generation (creation and production) and external narrative generation (production–consumption) in Watakushi Monogatari can be understood as this representation of the narrative generation process. To simplify, from a highly pragmatic perspective, and as shown in Figure 15, the Watakushi Character and “I (watakushi),” using GIS with INGS, generate a number of intermediate Watakushi Monogatari as a group of fragmentary narratives, utilizing knowledge and methods related to kabuki as the narrative technique that this paper introduced and connecting them in the final Watakushi Monogatari as a kind of compilation. The fragmentary group of Watakushi Monogatari in this case (Watakushi Monogatari in a broad sense) and the final Watakushi Monogatari (Watakushi Monogatari in a narrow sense) are all fixed narratives; however, the process that generates Watakushi Monogatari in the broad sense and the one that produces the final Watakushi Monogatari in the narrow sense are not solely fluid states of narrative generation and, particularly in the final Watakushi Monogatari, as described above, they are understood as fluid narratives through multiplicitous structures of underlying narratives or through a superbook’s special function.

215

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

SUPPLEMENTAL DISCUSSION AND DIRECTIONS FOR FUTURE RESEARCH In this section, the author provides a supplemental consideration based on the following four topics, as well as directions for future research regarding the entirety of this chapter.

Supplemental Discussion of Several Topics This section addresses four themes that the author does not refer to in previous parts or additional topics for this chapter.

Utilizing and Extending Sekai Kōmoku The first topic is tied to both GIS and kabuki. One of the new ideas in this chapter is that the author aims to construct the basic structure of the geinō work resource in the resources in GIS based on Sekai Kōmoku in kabuki. This part attempts to provide a supplemental discussion. The author continues to analyze Sekai Kōmoku in detail in the context of the narrative generation study of kabuki. The essential goal of the investigation is not necessarily to uncover new possibilities for the study of kabuki, but to utilize these documents for the author’s development of narrative generation systems. From a broader viewpoint, dealing with prior narratives and literature from the perspective of Sekai Kōmoku, and expanding and changing the range of their uses, are themes relevant to the expanded literary theory and post-narratology that the author considers. The author’s expanded literary theory and post-narratology (which is the recent term) are substantially developing, introducing analyses of kabuki studies, as well as further “narratologies in Japanese literature” (which is not the narratological interpretation of Japanese literature) beyond the previous usage of the narratology of Propp, Junette, and Jause. The author affirms that Sekai Kōmoku can be utilized as a general, universal method/framework for addressing narratives both systematically and structurally. The author’s idea, in a more pragmatic sense, is to develop a universal, general knowledge source based on Sekai Kōmoku or its essential ideas. Sekai Kōmoku, collected in a document or organized as text, is simply the enumeration of many narrative titles, characters, and so on. Readers or users during the Edo era employed personal and organizational knowledge obtained by professional people; they expanded, embodied, and concretized such knowledge to produce kabuki works. Consequently, to introduce Sekai Kōmoku and its essential ideas into the author’s narrative generation study, it is necessary to analyze the invisible knowledge (which is frequently “tacit knowledge”) to formalize and give structure to the underlying principles. In turn, the representation can be transformed into a conceptual, structured description for use in narrative generation systems, especially GIS with INGS. Actually positioning and utilizing Sekai Kōmoku as a major method in GIS requires the difficult restructuring of Sekai Kōmoku, including the above tasks. The author introduces a previous attempt related to Sekai Kōmoku or its future use and expansion. The author describes the systematic collection of Japanese folktales as the conceptual representation structure by Common Lisp. he collection can be used as a narrative example in the GIS with INGS version of Sekai Kōmoku. In the next stage, the author considers the intertextual synthesis of several narrative works using the technique of naimaze in kabuki. In the processing order by GIS with INGS, GIS is in charge of defining a macro-level narrative structure, and INGS elaborates and transforms its structure to generate the final narrative scenario.

216

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Life Narratives in GIS GIS with INGS treats life narratives as a narrative type. This chapter concretely discusses life narratives, focusing on positioning and usage in GIS. Events in the lives of geinōjin characters need to be stored in the system. For example, scripts presented by Schank and Abelson (1977), which are a typical knowledge representation in semiotic AI, are related to life narratives. Existing scripts are organized based on each event’s unit in people’s daily lives. In contrast, the subject in the context of the author’s life narrative is how the life events, as broader units, are organically defined. The specific consideration of life narratives requires collecting the author’s life and other people’s lives as materials, and organizing them for use in GIS with INGS. The information to be contained in GIS is the broad (hierarchical) framework of lives. INGS elaborates the framework according to more detailed information. As with life narratives, this detailed information can be commonly used for ordinary narratives. These kinds of fragmented forms in the narrative knowledge of INGS can be generally used for both ordinary and life narratives. As shown in Internal and External Narrative Generation, in the BACKGROUND, the consideration about the author’s “life” occurs in association with Watakushi Monogatari. This analysis classifies the “time” regarding the author’s life into several types, and categorizes related and other information. The primitive idea for the respective content is described in The Conception of Content in Internal Narrative Generation. This is studied as an approach bound in the author’s “I” (watakushi). The content is unique in the author’s “I” (watakushi). It is possible to divert and adapt the framework for the design and development of life narratives in GIS.

Developing Watakushi Monogatari Watakushi Monogatari provides the idea for a narrative (or narratives) to be created by the author. Generally, studies of computer systems assume use by external people other than the systems’ designers or developers. Many computer systems are examined in terms of the development of tools, no matter whether they are support systems or alternative systems. Different computational content studies develop AI computers and narrative generation systems that select existing or past artists as materials and learn and imitate their methods of content generation and representation to produce similar content. In contrast, Watakushi Monogatari, presented in this chapter, is the conception of narrative content creation to be written by the author. Therefore, it is very private and personal. As a similar example, Hori (2007) writes his essays using the creative activity aid program that he himself implemented. Although he originally designed his creative activity aid system as a general one for supporting users’ creative activities, he added himself as a user and supported his writing of papers beyond simply verifying the usage. He employs the system that he researched and developed for his own creative activities. The author intends to use a developed system for the narrative(s) he will write, Watakushi Monogatari, instead of practical documents, such as presentations and papers. Watakushi Monogatari does not simply support (through imitation) objects written based on existing content methods without altering the framework of previous content methods. This study is different from similar research, and attempts to provide a new approach to narrative creation by developing the method of producing Watakushi Monogatari using the author’s narrative generation system study, integrated in GIS with INGS. One of the most unique points of Watakushi Monogatari, under the framework of the narrative generation study, is that it presents a narrative generation method or model in which both the narrative generation process

217

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

(a kind of fluid situation) and its representation process (similarly, a fixed situation) are essentially liked and interact with each other. The author considers the above topic from the angle of generality and concreteness. For example, several studies (including the author’s) have been conducted that formalize Propp’s narrative theory (“the morphology of the folktale”) and utilize it in narrative generation systems in cognitive science and AI, but they have been frequently criticized. One opinion (Tanaka, 2010) asserts that although Propp’s narrative research (in addition to similar narratological studies) can be used as a theory of narrative generation (as a general theory), when we concretize research to apply it to actual narrative generation, it is difficult to assign sufficient narrative generation abilities to the theoretical framework, owing to problems related to the data’s details and common sense narrative processing. As for criticisms of Propp’s theory from the perspective of AI and cognitive science, like Gervás (2013) in Spain, the author believes that Propp’s narrative research includes very rich, concrete narratological knowledge and insights. The surface analysis of Propp’s theory by researches in AI and cognitive science has ignored the rich characteristics that Propp’s theory has. The author has sometimes mentioned this fact (for instance, Ogata and Kanai (2010)). In the first place, research that comprehensively systematizes Propp’s fertile narratology does not exist. Scholars like Takana maintain that sufficient systematization, as a narrative generation mechanism of Propp’s theory, is difficult, despite the theoretical possibility, perhaps because these scholars do not comprehend the material content. In contrast, the above R&D—based on the imitation of specific artists and content—forms an approach in which general narrative generation, as well as other genres’ content generation, shift in the direction of concrete content production according to each objective. Regarding this point, the focus moves toward the concreteness of content, based on generality as a tool. In many cases, such research has a general tendency that positions a specific artist’s imitation as a means for aiming toward performance improvement. The author’s conception of Watakushi Monogatari aims for the creation of the author’s narrative(s), though they are, at first, certainly experimental and trial work(s). In this sense, the author strives for very concrete, special, and personal results. The final evaluation criteria include pragmatic, actual, and concrete achievements of purposes; for example, with narrative generation systems, whether: good works are created (though the evaluation of good works is not single), works that the author himself can satisfy are made, and narratives can be socially distributed. Of course, at several stages of the development process, the feedback from concrete works and content is addressed in the system as a general tool. This cycle is extremely turned around, contending for actual works and content, and finishes with the completion of Watakushi Monogatari. Although this book’s title is Bridging the Gap between Narratology, Artificial Intelligence, and Cognitive Science Through Narrative Generation, from the above description, the trial of Watakushi Monogatari in this chapter goes beyond narratology and literary theories. Rather, the author endeavors to bridge the gap between the narrative itself, and AI and cognitive science, through narrative generation.

The Creation and Distribution of Novels and Kabuki in Japan In this chapter, the author asserts that the creation and distribution of Watakushi Monogatari are not two divided things, but rather a mutually related process. In other words, the internal narrative generation in the author’s proposed concept (i.e., the process of creating narrative content) and the external narrative generation (i.e., the social distribution and development of created narrative content) are mutually complex. In line with this explanation, the author believes that the internal creation and the external 218

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

development of previous and actual narrative production in various genres may have similar mutual relationships; hence, the author will pursue such analysis in future studies. In this section, the author only reveals a tentative consideration of the pertinence between creation and social development (e.g., distribution), especially in modern novels and kabuki. First, the author examines a typical method that has frequently been adopted in modern Japanese novels. In the following phenomena, creation and distribution occur based through a mutual (i.e., an obvious awareness regarding narrative creation and distribution), blended, whole process. A novel is partially published via the medium of a magazine. Next, the novel of the magazine is changed to the form of a published book. In the process, the novel, which was continuously published in the magazine, receives small primary corrections and revisions. In comparing Watakushi Monogatari with the above process, the sections for Watakushi Monogatari may correspond to the “fragmented narratives” shown in Figure 13. At this point, the stated process and the case of Watakushi Monogatari are similar. However, in the continuous change from magazine to book in the previous example, most parts of a novel published in a magazine (or several magazines) are synthesized into a novel as a book, notwithstanding the number of corrections and revisions. In contrast, in Watakushi Monogatari, only the collection and synthesis of fragmented narratives do not directly construct a larger Watakushi Monogatari to a substantial extent. An essential change takes place in the case of Watakushi Monogatari, beyond the number of corrections and revisions in the narrative transformation, from the fragmented narratives as the Watakushi Monogatari in a broad sense, to the Watakushi Monogatari in a narrow sense. Watakushi Monogatari forms more closed, fundamentally inseparable mutual relationships between narrative creation (i.e., internal narrative generation) and narrative distribution and social development (i.e., external narrative generation). The author deals with kabuki as another topic. The author scrutinizes the creation method of a scenario (the scenario book was traditionally called daichō in the world of kabuki) in kabuki narratives. Watanabe (1989) respectively calls the levels of an actual performance and the scenario (daichō) “first text” and “second text.” This indicates that scenarios do not dominate performances on stage in kabuki. In the history of kabuki, the text’s level of scenarios (daichōs) existed, except during the period of early kabuki. Notwithstanding, kabuki stresses the role of the first text, and many kabuki critiques have traditionally been carried out based on materials of the first text instead of the second one. This fact is connected to the tradition of the strong power of actors. Hattori (1970) states that, when we compare actors to writers, the existence of actors is primary, and the existence of writers is secondary. The author is interested in discovering what kinds of processes and method scenarios (daichōs) in kabuki, at the level of the second text, have been created. In other words, what type of mutual relationship has emerged regarding scenario (daichō in the second text) creation (i.e., internal narrative generation in this study) and actors’ stage performances (i.e., external narrative generation)? The author divides the scenario (daichō) creation process into three types: • • •

A situation where there is no scenario (daichō). Actors produce a drama through their mutual practical acts. The scenario (daichō) is sometimes written as the play’s report after the actual performance. Actors begin to perform a play according to the narrative’s outline. Through this process, various additions and alterations take place, and the scenario (daichō) is changed to reflect them. A kabuki playwright composes a steady scenario (daichō), and the actors perform the drama based on it. However, during the performance, including the rehearsal, diverse corrections are

219

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

incorporated into the scenario, such as detailed and structural revisions. A final first text, a stage performance, is differentiated and transformed from the original second text, a scenario. Kabuki scenarios (daichōs), as the second texts, are created in accordance with numerous ways. Actors, as subjects in a stage performance, play an essential role in scenario creation. This tradition has been passed on to contemporary kabuki, in which both narrative social distribution and delivery as a performance on stage, and narrative creation, as a scenario (daichō), are more or less merged. The author may be straying from the central topic here, but an influential method of narrative generation system studies in AI is “narrative generation as the simulation of actions and the mentality of characters.” Actors’ role is great in kabuki and is often confused with the AI method for narrative generation. However, these two things are clearly not the same. The narrative creation method in kabuki is not “narrative generation by character(s) in a story,” but “narrative generation by actor(s)” or “narrative generation by both actor(s) and writer(s).” Of course, when one or more actors create the narrative of a stage performance in kabuki, an actor can simulate a specific character’s action and mentality. However, an actor in kabuki does not completely become one with a character. In many kabuki plays, an actor can play several roles. Kabuki actors do not completely emphasize specific characters. Rather, actors in a kabuki performance play many parts overlapping with writers. During the first period of kabuki history, the positions of “writer” and “actor” in kabuki were not separate. A certain type of actor played characters on the kabuki stage and was also a writer. Many kabuki works in the early period were created by actors. This means that stage performances by actors included the function of narrative works’ creation. Both of the above two things are inseparable. The mutual connectiveness between the internal and external narrative generation in kabuki will influence the creation of Watakushi Monogatari.

Future Research Directions Regarding This Whole Chapter There are diverse paths for future research regarding this chapter. A main theme is the system development and content creation in terms of various aspects, such as the following: •

220

Design and implementation of GIS with INGS: A key future research topic is the program development and implementation based on the policy and macro design described in this chapter. This program tentatively requires the following modules: (1) A program module for generating and changing the geinō information (discography) and life information of a geinōjin character at a certain moment. (2) A program module for a narrative genre and work that the geinōjin character appears and plays. (3) A program module for creating a narrative scenario using INGS based on the geinō work resource, which is similar to the knowledge system presented in Sekai Kōmoku in GIS. The geinōjin character is also formed as a dramatic character, in addition to an actress/actor. (4) A program module connected with the above program module (3) to generate a life story of the geinōjin character at a certain point based on the life work resource in GIS. (5) A geinō narrative work and life narrative work, at a certain point for the geinōjin character, are reflected in the extension of the geinōjin character’s discography and life story. Narratives produced using GIS with INGS, which roughly has the above program modules, are approximately divided into the following parts: (1) The entire discography of a geinōjin character; (2) The entire life story of a geinōjin character; (3) Geinō narratives (scenarios) created at a certain point in time; (4) The part of the life story created at a certain point in time; and (5) The geinōjin character expanded at a certain

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents





point of time. The basic criteria for the investigation of the first stage system are the following: The first criterion is whether various features of the narrative, generated based on the thought of the macro design of GIS with INGS, stated through this chapter, are realized according to the thoughts of macro design. The second is whether the program, developed in accordance with the macro design’s thoughts, can produce knowledge and insight that transcend the first prediction in some way. The results of the first verification will be linked with the next programming actions, such as the revision, expansion, and correction of the program’s configuration, algorithms, and data structures. The second verification results will cause more radical actions toward reconsidering the program’s basic macro design thought itself. Systematization of Sakusha Shikihō Kezairoku and Sekai Kōmoku: Further analysis and modeling of Sekai Kōmoku from the angle of narrative generation will be continued according to the policy described in The Survey and Analysis of Sekai Kōmoku. In future studies of Sakusha Shikihō Kezairoku, the author will determine whether the schema of a multiple grammatical structure, shown in Figure 10, can be formed as a set of computer programing descriptions. Moreover, the author will advance other elements’ systematization. The author predicts that the first Sekai Kōmoku will be able to be positioned in GIS with INGS and, further, that Watakushi Monogatari, as a useful document, will provide a generalized method beyond an object for the analysis of kabuki. The latter Sakusha Shikihō Kezairoku will be reconsidered from a similar standpoint. The next sub-section complements these topics. Creating and Developing Watakushi Monogatari: One of the most important paths for future research is to proceed with the actual creation of Watakushi Monogatari to move toward the pragmatic phase from the theoretical and conceptual stages. In the future, detailed, individual topics will be discovered based on a comprehensive theme. As made clear through the discussion in this chapter, the fixed narrative(s) in Watakushi Monogatari not only indicate the final fixed narrative. Through all phases of the generation, creation, and production of Watakushi Monogatari, fixed narratives are made in flexible, free, and diverse forms. The degree to which fixed narratives were limited and regulated in terms of influencing later narrative generation is small; their narratives can also be used effectively, freely, and flexibly for later processes of fluid narrative generation. The creation of Watakushi Monogatari will be experimentally moved ahead by tentatively making fixed narratives from fluid narrative generation, namely performing narrative creation as a kind of trial and error. As stated above, the author will enter the pragmatic phase for the creation of Watakushi Monogatari in the future.

Kezairoku and Sekai Kōmoku as Narrative Generation What do we observe in Japanese narratologies and literary theories when viewing them from the angle of narrative generation study? This means that the knowledge acquired through the analysis of narratives is introduced into narrative generation models and systems as narrative generation techniques. Even if we could survey many narratological and literary studies and criticisms in Japan, we would probably see diverse research and criticisms with different characteristics. Even if we could model and further systematize research or critiques partially and formally, how can they be stored in an integrated manner into one or more narrative generation systems? Simply speaking, there are methods or directions between two extremes. The first approach is the direction that individually examines numerous narratological and literary theoretical studies and cri221

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

tiques in Japan to develop one or more broad narratological and literary theoretical systems, based on the idea of a higher-level, more synthesized narratological system. Another path stresses individually using surveyed methods without an entire systematization like the above example. The author provisionally calls the former direction a systematic method and the latter direction an arbitrary one. They both have pros and cons. For example, although the former is clear and valuable theoretically or from an academic viewpoint, it requires enormous efforts and is inconvenient when we directly try to apply it to narrative generation systems in a pragmatic sense. In contrast, the pros and cons of the arbitrary approach are the inverse of the above case. It is easy to directly and practically use a narrative technique and the technical groups in narrative generation systems. However, such an approach may be interpreted as too individual and arbitrary in theoretical and academic senses. In any case, the author affirms that an eclectic method, including both directions or approaches, is realistically appropriate for the practical goal of the author’s narrative generation study. The aim is, based on the process considering each object of Japanese narratological and literary theoretical research and critiques, to model and systematize them from the perspective of narrative generation research, and to build an entire theoretical system. Like this, a progression that blends both directions is in demand. Meanwhile, narratologies and literary theories in Japan, like those addressed in this chapter, can be arranged assuming reflection on the narrative content knowledge bases in INGS and the narrative work resources in GIS. Specifically, a kind of editing method of narrative content seen in Sekai Kōmoku will give an indication as to the systematic construction method in GIS. Figures of Goka Jūyō no Den” and Nidō Shibetsu no Den in Sakusha Shikihō Kezairoku, as well as several theories related to sekais (worlds) and naimaze, such as shigumi–shuko and yokosuji–tatesuji, are parts that need a detailed description and conceptualization. In addition, algorithms reference contemporary narratology and narrative theories. Goka Jūyō no Den, which includes the concepts of jo–ha–kyū by Zeami and ki–shō–ten–gō by Chinese poets, indicates a framework of synthesized narrative grammar. Other narratologies and literary theories in Japan also have diverse types of knowledge and insights that may appear through productive features in the connection with narrative generation. For instance, the idea of kyojitsu hiniku by Chikamatsu Monzaemon, is linked to the discussion about the adjustments of fiction and fact in narrative. There are books that can contribute to shedding light on the techniques of narrative generation, including Bungakuron by Natsume Sōseki, Gengo ni totte Bi toha Nani ka by Takaaki Yoshimoto, and Nihon Shōsetsu Gijutsushi by Naomi Watanabe (2012). The books by Yoshimoto and Watanabe are especially related to the aspect of narrative discourse. The literary theory by Sōseki covers both narrative content/stories and narrative discourse. For example, Nihon Mukashibanashi Taisei (Seki, 1978a, 1978b, 1978c, 1978d, 1978e, 1979a, 1979b, 1979c, 1979d, 1980; Seki, Nomura, & Ōshima, 1979, 1980) collects Japanese folktales and categorizes them into many kinds. Such studies will contribute to the facet of narrative content/ stories. At the same time, this study accounts for the topic of narrative variation and transformation. It may be useful to examine the concrete methods of narrative discourse. As stated above, kabuki and other approaches to narratives and literature will make a substantial contribution to the author’s narrative generation study in various senses. The following works will be scrutinized in parallel in the future: • •

222

Analysis of narratologies and literary theories in Japan The organization from the angle of narrative generation: ◦◦ Modeling and systematization toward/as INGS ◦◦ Modeling and systematization toward/as GIS

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

This chapter covers two books by two contemporary researchers or critics and two classical documents written in the Edo period for the sake of introducing the author’s narrative generation study in any form. In this chapter, the author could not present the computer programs for these materials; notwithstanding, the author describes the elementary plan for modeling and systematizing the above two classical results as mechanisms of narrative generation. In recent kabuki, imaging techniques such as CG are often used on stage. In examples the author recently observed, Konjaku Kyōen Senbonzakura by Nakamura Shidō II (1972–), Sawamura Kuniya I(1978–), and Hatsune Miku (as a virtual character), etc. (Minamiza, Kyoto, August, 2019) and Oguri by Ichikawa En’nosuke IV (1975–) and Nakamura Hayato I (1993–), etc. (Shinbashi Enbujō, Ttokyo, Oct., 2019) are used frequently. Yukinojō Henge by Bandō Tamasaburō V, Ichikawa Chūsha IX (1965–), Nakamura Shichinosuke II (1983–), etc. (Kabukiza, Tokyo, August, 2019) are often used as visual images as a movie. However, there are no studies or practices that generate narratives such as scenarios, stories, and plots. As Ogata (2018a, 2018b) describes in detail, kabuki means a theatrical art that is performed on stage and witnessed by an audience. At the same time, kabuki involves enormous documents through language and images. They include documents called narratology or literary theories. Sakusha Shikihō Kezairoku (1972) and Sekai Kōmoku (1916) are documents of that type. Regarding this point, narratives such as scenarios, stories, and plots are sufficiently addressed in the case of kabuki. In the author’s examination of kabuki (Ogata, 2016b, 2018a, 2018b, 2019c, 2020g; Ogata & Ono, 2019), the author analyzes and categorizes the multiple structural elements composing a kabuki work to be performed on stage, as well as related technical and rhetorical components. Next, the author proceeds to inspect each element and its relationships, and studies several topics. Furthermore, the study is extended to the level of sequential, repetitive, and circulative work generation. These surveys are done to introduce the results into GIS with INGS. Since a goal of GIS with INGS is develop the narrative generation mechanisms of kabuki in various senses, they include generating scenarios like kabuki narratives, and producing stage performance structures like kabuki from a story or plot, etc. Surveying Sakusha Shikihō Kezairoku and Sekai Kōmoku will also be a future meaningful topic in terms of modeling and systematization as concrete mechanisms in GIS with INGS.

Sakusha Shikihō Kezairoku as Narrative Generation In various theories written in Sakusha Shikihō Kezairoku, the figure of Goka Jūyō no Den (Figure 10) provides a kind of formalized grammatical description. However, this book’s author, Nyūgatei Ganyū, does not explain the figure. Other researchers do not state detailed explanations. The author would like to interpret this grammatical figure as a narrative development method based on a multiple structure. In particular, the author interprets this figure based on the division of “sequence (or chain),” “concept,” and “method.” • • •

[Sequence a] ◦◦ 景様 (Keiyō)‐頂上 (Yama)‐揺 (Yusuri)‐大曲 (Ōkuruwa)‐鎌入 (Kamaireru) [Concept a] ◦◦ Sekai (World) [Concept b] ◦◦ Shigumi ◦◦ Shukō 223

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

• • •

[Sequence b] ◦◦ Jo–Ha–Kyū [Sequence c] ◦◦ Ki–Shō–Ten–Gō [Complicated Method] ◦◦ Jo is connected with Ki–Shō, Ha is connected with Shō–Ten, and Kyū is connected with Ten–Gō. The moving among elements in Jo–Ha–Kyū are expanded using Ki–Shō–Ten–Gō.

The author interprets this hierarchical, multiple story grammar of kabuki. First, the above “Sequence a” is a global structure of a kabuki scenario to be finally composed. “Sequence b” is the most abstract structure and “Sequence c” is the intermediate structure. In contrast to these, “Sekai” in “Concept a” corresponds to the most primary principle of kabuki scenarios, and both “Shigumi” and “Shukō” (in “Concept b”) guide the sekai toward a specific scenario structure. Moreover, the concept of Tate-suji (vertical line) and Yoko-suji (lateral line) may be related to the above narrative development. Based on the above ideas, the author advances more concrete consideration of multiple grammar for kabuki scenarios.

Sekai Kōmoku as Narrative Generation Next, the author examines how Sekai Kōmoku can be described as a computer program. First, the author explores a method that faithfully follows the description shown in Sekai Kōmoku. For the preparation, elements in each unit in Sekai Kōmoku need to be conceptually structured, represented, and described in some manner. The key elements include names of characters (including fictional characters and real ones), prior works (including various narrative genres), and ningyō jōruri works. As the first “names of characters” also appear in prior narrative works, the decomposition and anatomy of related narrative works in some way are essential tasks for program design and implementation. The decomposition and anatomy can be executed for the names of characters, plots and stories, dramatic places, (important) objects, etc. The first subject for programming is decomposing and dissecting necessary and possible elements for narrative generation until a necessary and possible level is reached. If the knowledge structure of Sekai Kōmoku corresponds to the data in the computer program, the use and operation should be performed using other mechanisms. For example, the narrative grammar and narrative techniques written in Sakusha Shikihō Kezairoku may be able to be implemented for use and operation. Utilizing examples of characters and narrative works stored in the description of Sekai Kōmoku, the mechanism can be composed of relatively global narrative frameworks, which are included in the narrative production part of GIS in the author’s study framework. Thereafter, a type of narrative skeleton, as a result of the above process, is passed on to the module of INGS to place various concreteness and elaboration within it.

How the Narrative Generation Study of Kabuki is Tied to GIS with INGS and Watakushi Monogatari In this section, the author considers a way to introduce theories and methods in Sakusha Shikihō Kezairoku and Sekai Kōmoku into GIS with INGS. Generally speaking, the computer program that models and implements the theories of Sakusha Shikihō Kezairoku and the knowledge structure of Sekai Kōmoku is positioned as parts of GIS. It, namely GIS with these elements, produces the most global structure of a narrative scenario, and INGS generates a detailed description, namely conceptual representation 224

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

structures of a narrative. Although this process reveals narrative generation as a single substance, its consecutive processing forms a narrative sequence for geinō. Narratives produced through these processes correspond to narratives as works, except for life narratives. If we, so to speak, overestimate the value in the theories and methods of kabuki discussed here, the geinō in GIS means that kabuki and GIS form a system to be concretely developed, reflecting and introducing the results of the author’s narrative generation study of kabuki. This is an important problem for this study. Sekai Kōmoku indicates the systematic composition method of the geinō work resource. One future topic based on the extension of Sekai Kōmoku is to construct a new type of narrative knowledge base in which personally defined and collected narrative knowledge is stored. Works of Watakushi Monogatari, through GIS with INGS, are created by both the systems themselves and a Watakushi Character as the author’s “I” (wataskushi), who utilizes the knowledge base systematized like Sekai Kōmoku, and the development methods.

CONCLUSION This chapter dealt with the following themes in three central sections except for the INTRODUCTION, BACKGROUND, SUPPLEMENTAL DISCUSSION AND DIRECTIONS FOR FUTURE RESEARCH, and this CONCLUSION: 1. A MULTIPLE NARRATIVE GENERATION MECHANISM THROUGH TWO TYPES OF NARRATIVE GENERATION SYSTEMS: GIS WITH INGS addressed topics regarding the design of GIS with INGS. 2. NARRATOLOGY OF KABUKI: ORIENTED TOWARD NARRATIVE GENERATION focused on two contemporary kabuki researchers and two classical documents of kabuki. 3. THE CONCEPT OF WATAKUSHI MONOGATARI AND DESIGNING INTERNAL AND EXTERNAL NARRATIVE GENERATION METHODS TO CREATE AND DISTRIBUTE NARRATIVE CONTENT discussed the concepts of internal and external narrative generation, as well as their mutual relationships. In 1, the author described major elements and processes of GIS. The elements include the geinōjin resource, the geinō work resource, the life work resource, and geinōjin characters. Furthermore, this broader section discusses how the function of Watakushi Monogatari’s creation is incorporated into GIS with INGS, and explains the overall behavior of GIS with INGS in the context of associations with Watakushi Monogatari. In 2, the author surveyed the philosophies of two contemporary kabuki researchers, namely Yukio Hattori’s The Structure of Kabuki and Tamotsu Wanatabe’s philosophical and narratological thought of kabuki through a typical kabuki-dance, Kyōganoko Musume Dōjōji. Next, as classical theories of kabuki, the author scrutinized two documents, Sakusha Shikihō Kezairoku and Sekai Kōmoku from the narrative generation system’s point of view. Furthermore, the following sections considered the design in terms of the narrative generation of these documents or related parts. In 3, after this section shows the basic configuration of GIS with INGS for Watakushi Monogatari, the author discussed pertinent themes around internal and external narrative generation. Internal narrative generation means the creation of Watakushi Monogatari; part of the policy of content by internal 225

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

narrative generation includes fundamental knowledge sources for the narrative content of Watakushi Monogatari, and two directions of narrative content for Watakushi Monogatari based on the concepts of fluidity and fixation or movement and stillness. Although the external narrative generation is regarded as the sociological distribution of Watakushi Monogatari, it is essentially associated with the aspect of internal narrative generation. Hence, the final part addressed an integrated method between internal and external narrative generation. Finally, the macro-level goal of this chapter is for the author to concretely examine how to bridge the gap between narrative generation systems and narrative content with kabuki-oriented narratology and Watakushi Monogatari. This goal is also partially related to the title, a theme of the entirety of this book. Through this chapter, the author achieves this goal using the following methods. First, the narrative generation systems correspond to GIS with INGS, which the author has been proposing and developing, and the target of narrative content is set, including Watakushi Monogatari and related narrative content. Moreover, the gap between GIS with INGS and Watakushi Monogatari is bridged based on narrative knowledge, including kabuki-oriented narratological knowledge and diverse knowledge resources, such as knowledge related to the author’s “I” (watakushi). Based on the above established framework, which depends on GIS with INGS, Watakushi Monogatari, and other knowledge resources, the author would like to develop necessary systems and knowledge, and experimentally create fragmented content for Watakushi Monogatari.

ACKNOWLEDGMENT The research for this chapter was supported by the Japan Society for the Promotion of Science (JSPS KAKENHI), Grant No. 18K18509.

REFERENCES Amino, T., Kawamura, Y., & Ogata, T. (2002). Hierarchical generation of geinō-idol stories—Toward geinō information system and narrative marketing. In Proceedings of 17th Congress of the International Association of Empirical Aesthetics (pp. 549–552). Rome, Italy: University of Rome Tre. Arai, T., Ono, J., & Ogata, T. (2016). Semi-automatic generation of events sequence knowledge for narrative generation: The use in an integrated narrative generation system. In Proceedings of the 30th Annual Conference of the Japanese Society for Artificial Intelligence (3P1-1in2). Tokyo, Japan: Japanese Society for Artificial Intelligence. Arishima, T. (1964). Umareizuru nayami. In Arishima Takeo shū (pp. 350–393). Tokyo, Japan: Chikuma Shobō. (Original work published 1918) Arishima, T. (1995). Aru on’na. Tokyo, Japan: Shinchōsha. (Original work published 1919) Aristotle. (1997). Poetics (M. Heath, Trans.). London, UK: Penguin Classics. (Original work published 4th century BC)

226

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Asada, I., & Namiki, S. (1995). Dōjōji genzai uroko. In Toyotakeza jōruri syū, 3 (pp. 7-100). Tokyo, Japan: Kokusho Kankōkai. (First performed in 1742) Bandō Tamasaburō, V. (2003). Bandō Tamasaburō buyōsyū [Motion Picture DVD]. Syōchiku Home Video. Bronte, E. (2002). Wuthering heights. London, UK: Penguin Books. (Original work publishued 1847) Cortázar, J. (1991). Hopscotch: A novel (M. Heath, Trans.). Pantheon. (Original work published 1963) Dainipponkoku hokkekyō genki. (1974). In Ōjōden Hokkegenki (pp. 43-219). Tokyo, Japan: Iwanami Shoten. (Original work published 1040–1043) Dōjōji. (1998). In Yōkyoku hyakuban, Shin nihon koten bungaku taikei, 57 [Nō 100 songs, New Japanese classic literature collection, 57] (pp. 52–57). Tokyo, Japan: Iwanami Shoten. (Original work published the Muromachi period) Dōjōji engi. (1982). In Muromachi jidai monogatari taisei, daiju, ten-hamo [The Muromachi period narrative collection, No. 10, ten-hamo] (pp. 97–103). Tokyo, Japan: Kadokawa Shoten. (Original work published the Muromachi period) Dōjōji monogatari. (1982). In Muromachi jidai monogatari taisei, daiju, ten-hamo [The Muromachi period narrative collection, No. 10, ten-hamo] (pp. 104–124). Tokyo, Japan: Kadokawa Shoten. (Original work published 1660, the Edo period) Dostoevsky, F. (2003). The brothers Karamazov: A novel in four parts and an epilogue (D. McDuff, Trans.). Penguin Classics. (Original work published 1880) Dostoevsky, F. (2014). Crime and punishment (O. Ready, Trans.). Penguin Classics. (Original work published 1866) Fujii, S. (2004). Monogatari riron kōgi [Lectures on the theory of narrative]. Tokyo Daigaku Shuppankai. Genette, G. (1972). Discours du récit, Essai de méthode, Figures III. Paris: Seuil. Genji monogatari. (1993). Shin nihon koten bungaku taikei, 19 [New Japanese classic literature collection, 19]. Tokyo, Japan: Iwanami Shoten. (Original work published early 11th century, the Heian period) Genji monogatari. (1994). Shin nihon koten bungaku taikei, 20 [New Japanese classic literature collection, 20]. Tokyo, Japan: Iwanami Shoten. (Original work published early 11th century, the Heian period) Genji monogatari. (1995). Shin nihon koten bungaku taikei, 21 [New Japanese classic literature collection, 21]. Tokyo, Japan: Iwanami Shoten. (Original work published early 11th century, the Heian period) Genji monogatari. (1996). Shin nihon koten bungaku taikei, 22 [New Japanese classic literature collection, 22]. Tokyo, Japan: Iwanami Shoten. (Original work published early 11th century, the Heian period) Genji monogatari. (1997). Shin nihon koten bungaku taikei, 23 [New Japanese classic literature collection, 23]. Tokyo, Japan: Iwanami Shoten. (Original work published early 11th century, the Heian period) Genkyō shakusho. (1965). In Shintei zōho kokushi taikei, 31 [New revision, supplement, national histories collection, 31]. Tokyo, Japan: Toshikawa Kobunkan. (Original work published 1322 (the Kamakura period)

227

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Gervás, P. (2013). Propp’s morphology of the folk tale as a grammar for generation. OpenAccess Series in Informatics., 32, 106–122. doi:10.4230/OASIcs.CMN.2013.106 Haniya, Y. (1998). Shirei [Death Spirits]. Tokyo, Japan: Kōdansha (Original work publishued 1948-1995) Hattori, Y. (1970). Kabuki no kōzō [The structure of kabuki]. Tokyo, Japan: Iwanami Shoten. Hemingway, E. (1994). For whom the bell tolls. Arrow. (Original work published 1940) Hesse, H. (2013). Demian: The Story of Emil Sinclair’s Youth (D. Searls, Trans.). Penguin Random House. (Original work published 1919) Hesse, H. (2015). Steppenwolf (D. Horrocks, Trans.). Picador USA. (Original work published 1927) Homer. (2006). The Iliad & The Odyssey (B. Williams, Trans.). TX: El Paso Norte Press. (Original work published the late eighth or early seventh century BC) Hori, K. (2007). Sōzō katsudō shien system no riron to ōyō [Theories and applications of creativive activity aid]. Tokyo, Japan: Ohmsha. Hugo, V. (1982). Les miserable (N. Denny, Trans.). Penguin Classics. (Original work published 1862) Inoue, T. (2015). Mishima Yukio “Hōjō no umi” vs Noma Hiroshi “Seinen no wa”—Sengo bungaku to zentai shōsetsu [Mishima Yukio The sea of fertility vs. Noma Hirosi The ring of young people: Post-war literature and total novels]. Shintensha. Iura, Y. (1990). “Sakusha shikihō kezairoku” kō (zen)—Mondaiten no shiteki to kaimei [Considering “Sakusha shikihō kezairoku” (the first part): Indication and elucidation of problems]. Gakuen, 610, 2–13. Iura, Y. (1991). “Sakusha shikihō kezairoku” kō (go)—Mondaiten no shiteki to kaimei [Considerating “Sakusha shikihō kezairoku” (the second part): Indication and elucidation of problems). Gakuen, 615, 1–13. Jauss, H. R. (1970). Literaturgeschichte als provokation. Frankfurt am Main: Suhrkamp Verlag. Joyce, J. (2010). Ulysses. Wordsworth Editions. (Original work published 1922) Katō, T. (1997a). “Sekai kōmoku” no shohon—Sono ichizuke to tensha no keii [Various books of “sekai kōmoku”: Their positioning and the process of transcription]. Gakushūin Daigaku Kokugo Kokubungaku Kaishi, 40, 79–94. Katō, T. (1997b). “Sekai kōmoku” no seiritsu [The establishment of “Sekai kōmoku”]. Gakshūuin Diagaku Jinbun Kagaku Ronshū, 6, 77–100. Kawai, M., & Ogata, T. (2019). An analysis of “Kyōganoko musume Dōjōji” for the generation of the staging structures of kabuki from narrative plots. In Proceedings of the 63th Special Interest Group on Language Sense Processing Engineering (pp. 25–61). Tokyo, Japan: The Japanese Society for Artificial Intelligence. Kawai, M., Ono, J., & Ogata, T. (2020). Analyzing the stage performance structure of kabuki-dance “Kyoganoko Musume Dojoji.” In Proceedings of the 34th Annual Conference of the Japanese Society of Artificial Intelligence (3D1-OS-22a-04). Tokyo, Japan: Japanese Society of Artificial Intelligence.

228

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Kawatake, T. (1989). Kabuki-bi ron [On kabuki’s beauty]. Tokyo Daigaku Shuppankai. Kawatake, T. (1992). Kabuki: Its beauty and tradition (C. Holmes, Trans.). Japan Arts Council. Kawatake, T. (2003). Kabuki: Baroque fusion of the arts (Connel Hoff, F., & J., Trans.). Tokyo, Japan: The International House of Japan. Ki no kuni muro no kōri no ashiki on’na. (1974). In Ōjōden Hokkegenki (pp. 217–219). Tokyo, Japan: Iwanami Shoten. (Original work published 1040–1043) Ki no kuni no Dōjōji no sō hōke wo utsushi hemi wo sukueru koto. (1993). In Konjaku monogatari shū, 3 (pp. 286–290). Tokyo, Japan: Iwanami Shoten. (Original work published the end of the Heian period) Kominami, T. (1998). “Kezairoku” no kōsatsu—Engeki ni okeru kankyaku ninshiki no shomondai [Consideration of “Kezairoku”: Problems of audience recognition in a play]. Otaru Joshi Tanki Daigaku Kenkyū Kiyō, 27, 1–42. Konjaku monogatari shū, 1. (1999a). In Shin nihon koten bungaku taikei, 33 [New Japanese classic literature collection, 33]. Tokyo, Japan: Iwanami Shoten. (Original work published the end of the Heian period) Konjaku monogatari shū, 2. (1999b). In Shin nihon koten bungaku taikei, 34 [New Japanese classic literature collection, 34]. Tokyo, Japan: Iwanami Shoten. (Original work published the end of the Heian period) Konjaku monogatari shū, 3. (1993). In Shin nihon koten bungaku taikei, 35 [New Japanese classic literature collection, 35]. Tokyo, Japan: Iwanami Shoten. (Original work published the end of the Heian period) Konjaku monogatari shū, 4. (1994). In Shin nihon koten bungaku taikei, 36 [New Japanese classic literature collection, 36]. Tokyo, Japan: Iwanami Shoten. (Original work published the end of the Heian period) Konjaku monogatari shū, 5. (1996). In Shin nihon koten bungaku taikei, 37 [New Japanese classic literature collection, 37]. Tokyo, Japan: Iwanami Shoten. (Original work published the end of the Heian period) Kurihara, K. (Ed.). (1886). Kokon hauta taizen (Kaichū keikobon). Eirakudō. Kyōganoko musume Dōjōji. (1970). In Meisaku kabuki zenshū, 19, Buyōgeki shū [Masterpeace kabuki collection, 19, Kabuki-dance works] (pp. 15–30). Tokyo, Japan: Tokyo Sōgen Shinsha. (First performed in 1753. Written by Fujimoto Tobun and composed by Kineya Yasaburō) Longus. (2019). Daphnis and Chloe (E. Bowie, Trans.). Cambridge, UK: Cambridge University Press. (Original work published the second century) Malraux, A. (1990). Man’s fate (H. M. Chevalier, Trans.). Vintage. (Original work published 1933) Mann, T. (2005). The magic mountain (J. E. Woods, Trans.). Everyman. (Original work published 1924) Matsumoto, S. (1961). Warui yatsura [Bad Sorts]. Tokyo, Japan: Shinchōsha. Matsumoto, S. (1962). Renkan [Linkage]. Tokyo, Japan: Kōdansha. Matsumoto, S. (1980). Kurokawa no techō [Pocketbook of black leather]. Tokyo, Japan: Shinchōsha. Matsumoto, S. (1981). Yakō no Kaidan [Stairs That Shine at Night]. Tokyo, Japan: Shinchōsha.

229

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Melville, H. ’s (2009). Moby-Dick: Or, the whale. London, UK: Penguin Classics. (Original work published 1851) Miidera (1998). In Yōkyoku hyakuban, Shin nihon koten bungaku taikei, 57 [Nō 100 songs, New Japanese classic literature collection, 57] (pp. 52–57). Tokyo, Japan: Iwanami Shoten. (Original work published the Muromachi period) Mishima, Y. (1950). Ao no Jidai. Tokyo, Japan: Shinchōsha. Mishima, Y. (1954a). Shiosai. Tokyo, Japan: Shinchōsha. Mishima, Y. (1954b). Iwashiuri koi no hikiami [The sardine seller’s net of love]. Engekikai, 11. Mishima, Y. (1956). Kinkakuji. Tokyo, Japan: Shinchōsha. Mishima, Y. (1962). Utsukushii hoshi. Tokyo, Japan: Shinchōsha. Mishima, Y. (1968). Taiyō to tetsu. Tokyo, Japan: Kōdansha. Mishima, Y. (1969a). Haru no yuki. Tokyo, Japan: Shinchōsha. Mishima, Y. (1969b). Honba. Tokyo, Japan: Shinchōsha. Mishima, Y. (1970). Akatsuki no tera. Tokyo, Japan: Shinchōsha. Mishima, Y. (1971). Ten’nin gosui. Tokyo, Japan: Shinchōsha. Mishima, Y. (1951, 1953). Kinjiki. Tokyo, Japan: Shinchōsha. Mishima, Y. (1969c-1971). Hōjō no umi (4 books). Tokyo, Japan: Shinchōsha. Mishima, Y. (2000–2006). Ketteiban Mishima Yukio zenshū, 1–42 [Complete works of Mishima Yukio (definitive edition), Vols. 1–42]. Tokyo, Japan: Shinchōsha. Mitchell, M. (2014). Gone with the wind. Pan Books. (Original work published 1936) Miura, H. (2000). Naimaze. In Y. Hattori, T. Tomita, & T. Hirosue, (Eds.). Shinpan kabuki jiten [New edition kabuki dictionry] (p. 310). Tokyo, Japan: Heibonsha. Momochidori musume Dōjōji. (1960). In T. Takano (Ed.), Nihon kayō shūsei, 9, Kinseihen (p. 86). Tokyo, Japan: Tokyodō Shuppan. Moriyama, S. (1993). Tsuruya Nanboku—Naimaze no sekai [Tsuruya Nanboku: Naimaze’s world]. Tokyo, Japan: San’ichi Shobō. Nagauta genpon shūsei kankōkai. (Ed.). (1927). Momochidori musume Dōjōji (Nagauta genpon shūsei, Vol. 1). Tokyo, Japan: Nagauta Genpon Shūsei Kankōkai. Natsume, S. (2007). Bungaku ron (I, II) [Literary theory (I, II)]. Tokyo, Japan: Iwanami Shoten. (Original work published 1907) Natsume, S. (2014). Kōfu. Tokyo, Japan: Iwanami Shoten. (Original work published 1908)

230

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Noma, H. (1983-84). Seinen no wa, 1–5. Tokyo, Japan: Iwanami Shoten. (Original work published 1949–1968) Ōe, K. (1979). Dōjidai game. Tokyo, Japan: Shinchōsha. Ogata, T. (2002a). Expanded literary theory: Cognitive/computational expansion of literary theories and narratology. In Proceedings of the 17th Congress of the International Association of Empirical Aesthetics (pp. 163–166). Rome, Italy: University of Rome Tre. Ogata, T. (2002b). The concept of system narratology: From the viewpoint of expanded literary theory. In Proceedings of the 3rd International Workshop of Literature in Cognition and Computer in held in PRICAI2002 (W7-1, pp. 1–10). Tokyo, Japan: Japanese Cognitive Science Society. Ogata, T. (2003a). Monogatari no tajūsei to kakuchō bungakuriron no gainen—System narratology ni mukete, I [Narrative multiplicity and the concept of expanded literary theory: Toward a system narratology, I]. In M. Yoshida (Ed.), Fukuzatsu-kei shakai riron no shin chihei [New paradigm of complex social system theory] (pp. 127–181). Tokyo, Japan: Senshu Daigaku Shuppankyoku. Ogata, T. (2003b). Kakuchō bungakuriron no kokoromi—System narratology ni mukete, II [Attempts of expanded literary theory: Toward a system narratology II]. In M. Yoshida (Ed.), Fukuzatsu-kei shakai riron no shin chihei [New paradigm of complex social system theory] (pp. 309–356). Tokyo, Japan: Senshu Daigaku Shuppankyoku. Ogata, T. (2010). Shōsetsu—Ryūdō to kotei, Sakuhin no hō he [Novels: Fluidity and fixation, toward works]. In T. Ogata & A. Kanai (Eds.), Monogatariron no jōhōgaku josetsu—Monogatari seisei no shisō to gijutsu wo megutte [An introduction to informatics of narratology: Around the thoughts and technologies of narrative generation] (pp. 130–169). Tokyo, Japan: Gakubunsha. Ogata, T. (2014). Expanded literary theory for automatic narrative generation. In Proceedings of Joint 7th International Conference on Soft Computing and Intelligent Systems and 15th International Symposium on Advanced Intelligent Systems (pp. 1558–1563). Tokyo, Japan: Japan Society of Fuzzy Theory and Intelligent Informatics. Ogata, T. (2016a). Narrative generation of zukushi—Towards its systematization through surveying kabuki and developing a proper noun conceptual dictionary in INGS. In Proceedings of the 30th Annual Conference of the Japanese Society of Artificial Intelligence (1K5-OS-06b-2). Tokyo, Japan: The Japanese Society for Artificial Intelligence. Ogata, T. (2016b). Kabuki as multiple narrative structures. In T. Ogata & T. Akimoto (Eds.), Computational and cognitive approaches to narratology (pp. 391–422). Hershey, PA: Information Science Reference (IGI Global). Ogata, T. (2018a). Kabuki ni mukete (1)—Shiisei to henshūsei no monogatari kara tajūsei to jitsuzon no monogatari he [Toward kabuki (1): From narratives of arbitrariness and editing to narratives of multiplicity and real existence]. In T. Ogata, Y. Kawamura, & A. Kanai (Eds.), Jōhō monogatariron—Jinkōchinō, ninchi, shakai katei to monogatari seisei [Informational narratology: Artificial intelligence/cognition/ social process and narrative generation] (pp. 187–208). Tokyo, Japan: Hakutō Shobō.

231

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Ogata, T. (2018b). Kabuki ni mukete (2)—Tajū monogatari kōzō no shosō [Toward kabuki (2): Aspects of multiple narrative structures]. In T. Ogata, Y. Kawamura, & A. Kanai (Eds.), Jōhō monogatariron— Jinkōchinō, ninchi, shakai katei to monogatari seisei [Informational narratology: Artificial intelligence/ cognition/social process and narrative generation] (pp. 209–244). Tokyo, Japan: Hakutō Shobō. Ogata, T. (2018c). Gaibu he no monogatari seisei mata ha geinō jōhō system ni mukete [Narrative generation to the outside or toward a geinō information system]. In T. Ogata, Y. Kawamura, & A. Kanai (Eds.), Jōhō monogatariron—Jinkōchinō, ninchi, shakai katei to monogatari seisei [Informational narratology: Artificial intelligence/cognition/social process and narrative generation] (pp. 327–353). Tokyo, Japan: Hakutō Shobō. Ogata, T. (2018d). Naibu he no monogatari seisei mata ha watakushi monogatari ni mukete [Narrative generation to the inside or toward I narratives]. In T. Ogata, Y. Kawamura, & A. Kanai (Eds.), Jōhō monogatariron—Jinkōchinō, ninchi, shakai katei to monogatari seisei [Informational narratology: Artificial intelligence/cognition/social process and narrative generation] (pp. 355–370). Tokyo, Japan: Hakutō Shobō. Ogata, T. (2018e). An integrated approach to narrative generation: From Mishima and kabuki to narrative generation systems. In T. Ogata & S. Asakawa (Eds.), Content generation through narrative communication and simulation (pp. 49–147). Hershey, PA: Information Science Reference (IGI Global) Ogata, T. (2019a). A computational, cognitive, and narratological approach to narrative generation. In T. Ogata, & T. Akimoto (Eds.), Post-narratology through computational and cognitive approaches (pp.184). Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/978-1-5225-7979-3.ch001 Ogata, T. (2019b). Toward a post-narratology or the narratology of narrative Generation. In T. Ogata, & T. Akimoto (Eds.), Post-narratology through computational and cognitive approaches (pp. 85–142). Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/978-1-5225-7979-3.ch002 Ogata, T. (2019c). Kabuki as Multiple Narrative Structures and Narrative Generation. In T. Ogata & T. Akimoto (Eds.), Post-narratology through computational and cognitive approaches (pp. 192–275). Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/978-1-5225-7979-3.ch005 Ogata, T. (2019d). A method of naimaze of narratives based on kabuki analyses and Propp’s move techniques for an automated narrative generation system. Journal of Robotics. Networking and Artificial Life, 6(2), 71–78. doi:10.2991/jrnal.k.190829.001 Ogata, T. (2019e). Toward the incorporation of Japan’s narratologies and literary theories into the narrative generation systems by the author. In Proceedings of the 36th Annual Conference of the Japan Cognitive Science Society (pp. 442–451). Kanagawa, Japan: Cognitive Science Society. Ogata, T. (2020a). Toward an integrated approach to narrative generation: Emerging eesearch and opportunities. Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/978-1-5225-9693-6 Ogata, T. (2020b). Areas of narratives or narrative genres. In Toward an integrated approach to narrative generation: Emerging research and opportunities (pp. 59–161). Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/978-1-5225-9693-6.ch002

232

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Ogata, T. (2020c). Narratology and post-narratology. In Toward an integrated approach to narrative generation: Emerging research and opportunities (pp. 162–314). Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/978-1-5225-9693-6.ch003 Ogata, T. (2020d). Theoretical or philosophical considerations for an integrated narrative generation approach. In Toward an integrated approach to narrative generation: Emerging research and opportunities (pp. 315–403). Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/978-1-52259693-6.ch004 Ogata, T. (2020e). Internal and external narrative generation based on post-narratology: Emerging research and opportunities. Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/9781-5225-9943-2 Ogata, T. (2020f). An integrated narrative generation system: Synthesis and expansion. In Internal and external narrative generation based on post-narratology: Emerging research and opportunities (pp. 1–108). Hershey, PA: Information Science Reference (IGI Global). Ogata, T. (2020g). Kabuki as a synthetic narrative: Synthesis and expansion. In Internal and external narrative generation based on post-narratology: Emerging research and opportunities (pp. 109–254). Hershey, PA: Information Science Reference (IGI Global). Ogata, T. (2020h). Toward external and internal practices of narrative generation. In Internal and external narrative generation based on post-narratology: Emerging research and opportunities (pp. 328–427). Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/978-1-5225-9943-2.ch004 Ogata, T., & Akimoto, T. (Eds.). (2019). Post-narratology through computational and cognitive approaches. Hershey, PA: Information Science Reference (IGI Global). doi:10.4018/978-1-5225-7979-3 Ogata, T., & Amino, T. (2002). Tajū monogatari kōzō to geinō character no monogatari seisei system [A narrative generation system of multiple narrative structures and a geinō character]. In Proceedings of the 19th Annual Meeting of the Japanese Cognitive Science Society (pp. 130–131). Tokyo, Japan: Japanese Cognitive Science Society. Ogata, T., Fukuda, K., Ono, J., & Ito, K. (2020). Considering naimaze in kabuki narratives and a method. In Proceedings of the 34th Annual Conference of the Japanese Society of Artificial Intelligence (3P5OS-16b-02). Tokyo, Japan: Japanese Society of Artificial Intelligence. Ogata, T., Imabuchi, S., & Akimoto, T. (2014). Narratology and narrative generation: Expanded literary theory and the Integration as a narrative generation system (2). In Proceedings of the 5th Augmented Human International Conference (P-10). ACM Digital Library. 10.1145/2582051.2582077 Ogata, T., & Kanai, A. (2010). Monogatariron no jōhōgaku josetsu—Monogatari seisei no shisō to gijutsu wo megutte [An introduction to informatics of narratology: Around the thoughts and technologies of narrative generation]. Tokyo, Japan: Gakubunsha. Ogata, T., & Ono, J. (2019). Two elements and two techniques for the narrative generation of kabuki. In Proceedings of the 33th Annual Conference of the Japanese Society of Artificial Intelligence (1F2NFC-1-03). Tokyo, Japan: Japanese Society of Artificial Intelligence.

233

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Okamoto, K. (1969). Banchō sarayashiki. In Meisaku kabuki zenshū, 20, Shinkabuki shū [Masterpeace kabuki collection, 20, New kabuki works] (pp. 189–244). Tokyo, Japan: Tokyo Sōgen Shinsha. (First performed in 1916) Otokodate hatsugai soga. (1985). In Kabuki daichō shūsei, 8 [Kabuki script collection, 8] (pp. 75–240). Tokyo, Japan: Benseisha. (First performed in 1759. Written by Fujimoto Tobun) Propp, V. Y. (1968). Morphology of the folktale (L. Scott, Trans.). University of Texas Press. (Original work published 1928) Rolland, R. (2007). Jean-Christophe: Journey’s end (G. Cannan, Trans.). Echo Library. Rolland, R. (2011). Jean-Christophe (Vol. I; G. Cannan, Trans.). Aeterna. Rolland, R. (2018). Jean-Christophe in Paris: The market-place, Antoinette, the house (G. Cannan, Trans.). Sagwan Press. Sakusha shikihō kezairoku. (1972). In M. Gunji (Ed.), Kinsei geidōron (Nihon shisō taikei, 61) [Theories of early modern times’ arts (Japanese classic phirosophical thoughts collection, Vol. 61)] (pp. 493–532). Tokyo, Japan: Iwanami Shoten. (Original work published 1801) Schank, R. C., & Abelson, R. P. (1977). Scripts, plans, goals, and understanding: An inquiry into human knowledge structures. Lawrence Erlbaum. Sekai kōmoku. (1916). In Sekai kōmoku, shibai nenjū gyōji, gekikai chinwa (Chinsho kankōkai sōsyo, 9) [Compendium of narrative worlds, annual events related to kabuki plays, rare tales in theatrical circles (Rare books publication society series, Vol. 9)] (pp. 6–46). Tokyo, Japan: Chinsyo kankōkai (Written in the late Edo period (before 1791)) Seki, K. (1978a). Nihon mukashibanashi taisei, 2—Honkaku mukashibanashi, 1 [Compilation of Japanese folktales, Vol. 2: Ordinary folktales, No. 1]. Tokyo, Japan: Kadokawa Shoten. Seki, K. (1978b). Nihon mukashibanashi taisei, 3—Honkaku mukashibanashi, 2 [Compilation of Japanese folktales, Vol. 3: Ordinary folktales, No. 2]. Tokyo, Japan: Kadokawa Shoten. Seki, K. (1978c). Nihon mukashibanashi taisei, 4—Honkaku mukashibanashi, 3 [Compilation of Japanese folktales, Vol. 4: Ordinary folktales, No. 3]. Tokyo, Japan: Kadokawa Shoten. Seki, K. (1978d). Nihon mukashibanashi taisei, 5—Honkaku mukashibanashi, 4 [Compilation of Japanese folktales, Vol. 5: Ordinary folktales, No. 4]. Tokyo, Japan: Kadokawa Shoten. Seki, K. (1978e). Nihon mukashibanashi taisei, 6—Honkaku mukashibanashi, 5 [Compilation of Japanese folktales, Vol. 6: Ordinary folktales, No. 5]. Tokyo, Japan: Kadokawa Shoten. Seki, K. (1979a). Nihon mukashibanashi taisei, 1—Dōbutsu monogatari [Compilation of Japanese folktales, Vol. 1: Animal tales]. Tokyo, Japan: Kadokawa Shoten. Seki, K. (1979b). Nihon mukashibanashi taisei, 7—Honkaku mukashibanashi, 6 [Compilation of Japanese folktales, Vol. 7: Ordinary folktales, No. 6]. Tokyo, Japan: Kadokawa Shoten.

234

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Seki, K. (1979c). Nihon mukashibanashi taisei, 8—Warai banashi, 1 [Compilation of Japanese folktales, Vol. 8: Jokes and anecdotes, No. 1]. Tokyo, Japan: Kadokawa Shoten. Seki, K. (1979d). Nihon mukashibanashi taisei, 9—Warai-banashi (Shōwa), 2 [Compilation of Japanese folktales, Vol. 9: Jokes and anecdotes, No. 2]. Tokyo, Japan: Kadokawa Shoten. Seki, K. (1980). Nihon mukashibanashi taisei, 10—Warai-banashi (Shōwa), 3 [Compilation of Japanese folktales, Vol. 10: Jokes and anecdotes, No. 3]. Tokyo, Japan: Kadokawa Shoten. Seki, K., Nomura, J., & Ōshima, H. (Eds.). (1979). Nihon mukashibanashi taisei, 12—Kenkyū hen [Compilation of Japanese folktales, 12: Researches]. Tokyo, Japan: Kadokawa Shoten. Seki, K., Nomura, J., & Ōshima, H. (Eds.). (1980). Nihon mukashibanashi taisei, 11—Shiryō hen [Compilation of Japanese folktales, 11: Data]. Tokyo, Japan: Kadokawa Shoten. Sophocles. (1972). Oedipus the king (A. Burgess, Trans.). Minneapolis, MN: University of Minnesota Press. (Original work first performed around 429 BC) Taikōki. (1996). In Shin nihon koten bungaku taikei, 60 [New Japanese classic literature collection, Vol. 60]. Tokyo, Japan: Iwanami Shoten (Original work published 1626) Tanaka, K. (2010). Kigō to saiki—Kigōron no keishiki, Program no hitsuzen [Symbol and recursion: The forms of semiotics, the inevitability of programs]. Tokyo, Japan: Tokyo Daigaku Shuppankai (The University of Tokyo Press) Tanaka, Y. (1981). Nantonaku kurisutaru. Tokyo, Japan: Chikuma Shobō Shinsha. Teramoto, Y., & Nakanishi, A. (1993). Chi no tenkansha tachi—Business trickstar [Transformers of knowledge: Business trickstars]. Tokyo, Japan: NTT Shuppan. Tolstoy, L. (2010). War and peace (A. Mandelker, Ed., MaudeL.MaudeA., Trans.). Oxford University Press. (Original work published 1969) Uchida, Y., Kondō, H., & Funato, M. (1987). “Sekai kōmoku” sakuin—1985 nendo kyōdō kenkyu hōkoku [“Sekai kōmoku” index: 1985 joint study report]. Kyōritsu Joshi Daigaku Bungei Gakubu Kiyō, 33, 173–286. Urayama, M. (1974). Naimaze to sekai. Geinō no kagaku, 5, 103–120. Watanabe, N. (2012). Nihon shōsetsu gijutsushi [A history of technologies of Japanese novels]. Tokyo, Japan: Shinchōsha. Watanabe, T. (1986). Musume Dōjōji. Tokyo, Japan: Shinshindo Shuppan. Watanabe, T. (1989). Kabuki—Kajō naru kigō no mori [Kabuki: The forest of excessive signs]. Tokyo, Japan: Shin’yōsha. Watanabe, T. (2004). Kabuki—Kata no miryoku [Kabuki: Charm of forms]. Tokyo, Japan: Kadokawa Shoten. Watanabe, T. (2012). Watashi no kabuki henreki—Aru gekihyō-ka no kokuhaku [My kabuki wandering: Confession of a drama critic]. Tokyo, Japan: Engeki Shuppansha.

235

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Watanabe, T. (2018). Kudaime Danjūrō [Danjūrō IX]. Tokyo, Japan: Engeki Shuppansha. Yamamoto, Y. (1977). Sakazaki Dewanokami. In Yamamoto Yūzō zenshū, 1 [The complete works of Yamamoto Yūzō, 1]. (pp. 287–351). Tokyo, Japan: (Original work published 1921) Yasuda, N. (1989). Dōjōji densetssu kara “Kyōganoko Musume Dōjōji” he [From Dōjōji legnds to Kyōganoko Musume Dōjōji]. Shōtoku Gakuen Gifu Kyōiku Daigaku Kokugo Kokubungaku, 8, 1–12. Yokomitsu, R. (1931). Kikai. Tokyo, Japan: Hakusuisha. Yoshimoto, T. (1965). Gengo ni totte bi toha nanika [What is beauty for language?]. Tokyo, Japan: Keisō Shobō. Yoshimoto, T. (1968). Kyōdō gensō ron [The theory of communal Illusion]. Tokyo, Japan: Kawade Shobō Shinsha. Yoshimoto, T. (1971). Shinteki genshōron josetsu [A theory of mental phenomena: Introduction]. Tokyo, Japan: Hokuyōsha. Yoshimoto, T. (2008). Shinteki genshōron honron [A theory of mental phenomena: Main issues]. Tokyo, Japan: Bunka Kagaku Kōtō Kenkyūin Shuppankyoku. Zeami, M. (1983). Sandō. In M. Yamazaki (Ed.), Zeami (pp. 253–268). Tokyo, Japan: Chūōkōronsha. (Original work published 1423)

KEY TERMS AND DEFINITIONS External Narrative Generation: External narrative generation refers to the social development and distribution of the narrative generation systems, including the Integrated Narrative Generation System (INGS) and the Geinō Information System (GIS), as well as content produced by the narrative generation systems, such as computer games and automatic generation stories and novels. The first goal of external narrative generation is to develop a computational environment for using the above systems. Geinō Information System (GIS): This indicates a system model of geinō production and consumption. GIS has been considered a framework in which various levels of narrative production processes are driven by writers, receivers, characters, and actors. Furthermore, GIS is a model for circulative narrative generation, which repeats the narrative generation process. In contrast, in the author’s narrative generation study, the INGS corresponds to a single-level narrative generation system. GIS With INGS: Although GIS and INGS are relatively different narrative generation systems, they function as a linked mechanism. INGS plays a role in creating individual narratives, and GIS produces each sequential narrative by connecting the narratives produced by INGS into a broader story line. GIS is a more comprehensive framework of narrative generation that contains INGS. Integrated Narrative Generation System (INGS): INGS is a synthetic narrative generation system architecture that integrates previous studies by the author. From the broadest angle, INGS is divided into knowledge and procedures. INGS aims to be a narrative synthesizer that integrates a variety of narrative techniques, methods, rhetoric, and knowledge into an organic, dynamic generation framework.

236

 Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Internal Narrative Generation: This term forms the counterpart to the above external narrative generation. The internal narrative generation corresponds to the aspect of creating narrative works by the author, using developed narrative generation systems, mainly INGS and GIS. The author integrates such narrative works according to the concept of Watakushi Monogatari. Kabuki-Oriented Narratology: The author’s goal, based on kabuki-oriented narratology, is to scrutinize narratological research and critiques in the world of kabuki and apply the results to the modeling, design, and development of the narrative generation mechanisms of kabuki. In this chapter, the author focuses on two contemporary kabuki scholars, Yukio Hattori (1932–2007) and Tamotsu Watanabe (1936–), and two classical kabuki documents, Sakusha Shikihō Kezairoku (1972) and Sekai Kōmoku (1916). Narrative Generation of Kabuki: The narrative generation study of kabuki is a type of kabuki research according to constructive, productive, engineering, computational, and cognitive scientific methods. It aims to analyze diverse components of kabuki based on one or more narrative generation systems. Further, the narrative generation of kabuki attempts to utilize the acquired results for various narrative generation mechanisms as narrative techniques and knowledge. Post-Narratology: A concept for a field in which narratology and literary theories are organically introduced into computational and cognitive approaches. The author calls this type of narratology postnarratology, in contrast to classical narratology, which only focuses on only human and social scientific fields. However, the author’s concept of post-narratology deals with classical narrative knowledge, such as kabuki, and positively introduces the Japanese cultural worlds to which the author himself belongs. In various senses, the conception of post-narratology exceeds an interdisciplinary narratology between literary and informational areas. Sakusha Shikihō Kezairoku: This is a comprehensive, theoretical document written during the Edo era (1801) by a real kabuki writer, Namiki Shōza II. The author pays attention to the parts in which unique theories and ways of how kabuki scenarios should be written and created are described, including a type of grammatical statement showing a narrative development structure and a technique for inserting narrative contrivances (shukō in Japanese) into a determined or stereotyped flow (shigumi) according to a sekai (world) as a typical type of narrative. Sekai Kōmoku: Sekai means a narrative system or collection (a kind of knowledge base) that has been used to produce kabuki scenarios. Sekai Kōmoku is a document written in the middle of the Edo era that systematically collects many sekais as a knowledge base. In this chapter, the author analyzes the book’s entire structure and the mechanism of each sekai’s unit. In the future, an interesting and practical theme would be to pursue the author’s unique, new Sekai Kōmoku that perhaps treats modern and contemporary periods and topics. Watakushi Monogatari: Watakushi Monogatari (“My Story” or “My Narrative,” or “I-Story” or “INarrative”) means a new narrative genre that should be practiced using narrative generation systems. Although it is a type of story or narrative genre in which narrative worlds and narrations are pursued and represented through “I” (watakushi), this does not necessarily mean that the narrative world to be generated should be narrow and the field related only to “I.” Rather, it means that the first-person world gradually opens to the second-person and third-person worlds to construct an entire narrative dependent on their organic relationships.

237

Bridging the Gap Between Narrative Generation Systems and Narrative Contents

APPENDIX Table 1.­ Author 有島武郎[Arishima Takeo]

Ernest Hemingway

Franz Kafka

Title

生れ出る悩み [Umareizuru Nayami: The Agony of Coming into the World]

Publication 1918

Period(s) within the work The Meiji era

Location(s) within the work Iwanai, Sapporo (Hokkaido, Japan)

カインの末裔 [Cain no Matsuei: The Descendants of Cain]

1918

The Meiji era

Matsukawa farm (Hokkaido, Japan)

小さき者へ [Chiisaki Mono He]

1918

Taishō 6 (1917)

Tokyo

或る女 [Aru On’na: A Certain Woman]

1919

The Meiji era

Yokohama, Japan, A ship bound for USA, Seattle, USA

A Farewell to Arms

1929

During World War I

Italy, Switzerland

The Sun Also Rises

1926

During World War I

USA, Paris, Pamplona (Spain)

For Whom the Bell Tolls

1940

During Spanish Civil War

Spain

The Old Man and the Sea

1952

20th century

Cuba, Gulf of Mexico, Africa

The Killers

1927

20th century

New Jersey, Brentwood

The Snows of Kilimanjaro

1936

20th century

Africa,Mt. Kilimanjaro, Europe

The Metamorphosis

1915

Gregor’s house

The Judgment

1916

St. Pederburg

The Castle

1926

Vest Vest Count’s castle property, Court office, The junk room in K’s workplace

The Trial

1925

America

1927

USA, Germany, New York port, Hotel Occidental

André Malraux

The Human Condition

1958

1927

Shanghai

John Steinbeck

The Grapes of Wrath

1939

1930’s

Oklahoma, Route 66, Arizona Desert, Rocky Mountains, California

William Faulkner

The Sound and the Fury

1929

1910, 1928

Harvard University, Cambridge Street, Mottson Town

Light in August

1932

20th century

Alabama, Mississippi

Henry Miller

Tropic of Capricorn

1939

20th century

New York, Brooklyn

Tropic of Cancer

1934

1930’s

France

芥川龍之介[Akutagawa Ryūnosuke]

羅生門 [Rashōmon]

1917

The Heian period

Kyoto

樋口一葉[Higuchi Ichiyō]

志賀直哉 [Shiga Naoya]

William Shakespeare

Albert Camus

山本有三[Yamamoto Yuzō]

鼻 [Hana: The Nose]

1917

The Heian period

Ikeo, Uji, Kyoto

河童 [Kappa]

1927

20th century

Hotakadake [Mt. Hotaka]

たけくらべ [Takekurabe]

1986

The Meiji era

Yoshiwara, Senzoku Shrine, Otori Shrine (Tokyo)

にごりえ [Nigorie]

1895

The Meiji era

Fukuyamacho, Matuyama

十三夜 [Jūsan’ya: The Thirteenth Night]

1895

The Meiji era

城の崎にて [Kinosaki Nite: At KinosakiI]

1917

The Taishō era

Kinosaki Onsen (Hyōgo)

小僧の神様 [Kozō no Kamisama]

1967

The Taishō era

Kanda (Tokyo)

暗夜行路 [An’ya Kōro: A Dark Night’s Passing]

1921

From the Meiji era to Taishō era Onomichi, Kyoto

Hamlet

1603

Macbeth

1623

Othello

1622

King Lear

1606

Denmark 1040–1057

Scotland, Glamis Venice

The Stranger

1942

20th century

Algiers, Algeria

The Plague

1947

20th century

Oran city, Algeria Amsterdam

The Fall

1956

20th century

Exile and the Kingdom

1957

20th century

路傍の石 [Robō no Ishi: A Stone by the Roadside]

1940

The mid Meiji era

真実一路 [Shinjitsu Ichiro: Towards the Truth]

1936

The Meiji era

石坂洋次郎[Ishizaka Yojirō]

若い人 [Wakai Hito]

1937

The Shōwa era

石川達三[Ishikawa Tatsuzō]

私ひとりの私 [Watakushi Hitori no Watakushi]

1964

The Shōwa era

Nezu, Hongō (Tokyo)

A port town in a northern area (Japan)

Margaret Mitchell

Gone with the Wind

1936

1860’s

Southern Georgia, Atlanta

Erich Remark

Arch of Triumph

1945

1938–1939

Paris

continued on following page

238

Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Table 1. Continued Author Emily Brontë Lev Tolstoy

Fyodor Dostoevsky

Ivan Turgenev

Alexander Pushkin

Maxim Gorky

Title Wuthering Heights

武者小路実篤[Mushanokōji Saneatsu]

1847

Period(s) within the work 19th century

Location(s) within the work Yorkshire (England), Haworth (Yorkshire, England)

Resurrection

1899

19th century

Russia, Siberia

Anna Karenina

1877

1870’s

Russia, Moscow, St. Peterburg

War and Peace

1867

19th century

Russia, Austria

Childhood

1852

19th century

Moscow, Russia

Boyhood

1854

19th century

Moscow, Petrovskoy, Russia

Youth

1856

19th century

Petrovskoy, Moscow, Russia

Poor Folk

1846

19th century

St. Petersburg

Crime and Punishment

1866

19th century

St. Petersburg, Siberia

The Idiot

1874

19th century

Russia, Switzerland

Demon

1873

1869

Russia

The Brothers Karamazov

1880

19th century

Russia

The Raw Youth

1875

19th century

Russia

The House of the Dead

1862

19th century

Siberia, Russia

Notes from Underground

1864

19th century

St. Petersburg

Humiliated and Insulted

1861

19th century

St. Petersburg

The Village of Stepanchikovo

1859

19th century

Stepanchikoovo Village

First Love

1860

1833

Moscow, Lake Neskunui

Fathers and Sons

1862

May, 1859

Heidelberg

Rudin

1857

19th century

Russia

On the Eve

1860

19th century

Venice, Bulgaria, Balkans

A Nest of the Gentry

1859

19th century

Baden-Baden, Germany

Smoke

1867

19th century

Russia

Virginity

1877

19th century

Russia

The Captain’s Daughter

1836

1773–1775

Russia, Don River Basin, Bella Gorx fortress

Eugene Onegin

1833

1819–1825

The Queen of Spades

1834

St. Petersburg Russia

My Childhood

1914

19th century

Russia

In the World

1916

19th century

Russia

My Universities

1923

19th century

Russia

Mother

1907

19th century

Russia

The Lower Depths Nikolai Ostrovskii

Publication

How the Steel Was Tempered

1901-1902

19th century

Russia

1932

20th century

Russia

友情 [Yujō: Friendship]

1920

The Meiji era

Japan, Europe

愛と死 [Ai to Shi: Love and Death]

1939

The Taishō era

Japan, Paris

伊藤左千夫 [Itō Sachio]

野菊の墓 [Nogiku no Haka: The Wild Daisy]

1906

The Meiji era

Matsudo (Chiba)

Hans Theodor Storm

Immensee

1851

19th century

Schleswig-Holstein

Viola Tricolor

1874

19the century

Germany

Johann Wolfgang von Goethe

The Sorrows of Young Werther

1774

18th century

Germany

Jacques-Henri Bernardin de SaintPierre

Paul and Virginia

Faust

1808, 1833 1787

Abbé Prévost

Manon Lescaut

1731

Madame de La Fayette

La Princesse de Cleves

1678

15th century–16th century

Germany

18th century

Mauritius Island

16th century

Louvre Palace

France, USA

George Sand

Little Fadette

1849

19th century

Franch countryside

Colette

Green Wheat

1923

20th century

Paris

20th century

Françoise Sagan

Hello Sadness

1954

Alain-Fournier

The Great Meaulnes

1913

James Hilton

Goodbye, Mr. Chips]

1934

From the end of 19th century to the beginning of 20th century) 19th century

Cote d’Azur France, Sologne England

Hans Christian Andersen

Billedbog uden Billeder

1839

Saint-Exupéry

The Little Prince

1839

Denmark

Night Flight

1931

20th century

Lisbon, Patagonia

Charles Lindbergh

The Spirit of St. Louis

1953

1923

Over Atrantic Ocean, Le Bourget airport (Paris)

Sahara desert

Charles Dickens

Christmas Carol

1843

19th century

London

Annelies Frank

The Diary of a Young Girl

1947

Dering World War II

Amsterdam, Netherlands

continued on following page

239

Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Table 1. Continued Author

Title

西川一三[Nishikawa Kazumi]

秘境西域八年の潜行 [Hikyō Saiiki Hachinen no Senkō]

岸上大作[Kishigami Taisaku]

もうひとつの意志表示 [Mouhitotsu no Ishihyoji]

Publication

Period(s) within the work

Location(s) within the work

1967

1945, 1950

Japan, Mongolia, Gobi Desert, Tibet

1973

Shōwa 30’s

Tokyo

永山則夫[Nagayama Norio]

無知の涙 [Muchi no Namida]

1971

The Shōwa era

Prison

高野悦子 [Takano Etsuko]

二十歳の原点 [Hatachi no Genten]

1971

Jan/2–Jun/22, 1969

Kyoto

Hermann Hesse

Thomas Mann

Beneath the Wheel

1905

19th century

Germany

Demian

1920

20th century

Latin school in a small town in Jarmany

Narcissus and Goldmund

1957

Steppenwolf

1927

20th century

Germany

Germany, Europe

Tonio Kroger

1913

Early 20th century

Lübeck, Germany, Gymnasium, Munich, Denmark)

Death in Venice

1912

Early 20th century

Munich, Adriatic Sea, Venice

Buddenbrooks

1901

19th century

Lübeck, Germany

The Magic Mountain

1924

World War I

Royal Highness

1909

Swiss Alps, Davos Grand Grimburg

Doctor Faustus

1947

Germany

The Holy Sinner

1951

Flanders, Artois

Lotte in Weimar: The Beloved Returns

1939

Germany

Confessions of Felix Krull

1954

River Rhine, Romania

Joseph and His Brothers

1933

Ancient times

Egypt, Babylon

Little Herr Friedemann

1897

19th century

Germany

20th century

Italy

After 1878

Bavarian Kingdom

Mario and the Magician

1930

Friedrich Nietzsche

Thus Spoke Zarathustra

1885

Rainer Maria Rilke

The Notebooks of Malte Laurids Brigge

1910

Hans Carossa

A Childhood

1922

Das Jahr der schonen Tauschungen

1941

1897–1898

Munich

Jean-Paul Sartre

Nausea

1938

1932

France Ancient Greece

André Gide

Jean Genet

The Wall

1937

The Flies

1943

Ancient Greece

Strait is the Gate

1909

19th century

The Pastoral Symphony

1919

Gustave Flaubert

Stendhal

Spain France France

The Fruits of the Earth

1897

19th century

Africa

Atarashiki Kate

1935

20th century

Africa

If It Die, an Autobiography

1921

19th century–20th century

France

The Vatican Cellars

1914

20the century

Rome, Bucharest, Paris

The Counterfeiters

1925

20th century

Paris, France

Our Lady of the Flowers

1944

20th century

Paris, Prison

Miracle of the Rose

1946

20th century

Metley

The Thief’s Journal

1949

20th century

Spain, Italy, Austria, Czechoslovakia, Poland, Germany, Belgium)

The Maids Guy de Maupassant

Paris, Denmark

1947

20th century

Pierre and Jean

1887-1888

19th century

Le Havre

A Woman’s Life

1883

19th century

Les Pouples

Ball of Fat

1880

19 th century

Rouen

Madame Bovary

1857

19th century

Rouen, Tost

Sentimental Education

1869

1840–1867

Paris

Bouvard et Pecuchet

1881

19th century

Paris, Proudhon Street

The Charterhouse of Parma

1839

Napoleonic’s era

Parma, Lake Como, Farnese Tower, Italy

1820’s

Italy

The Red and the Black

1830

Vanina Vanini

1829

Besancon, France

de Laclos

Dangerous Liaisons

1782

18th century

France

David Herbert Lawrence

Sons and Lovers

1913

19th century

Wales

Women in Love

1920

Wales

The Rainbow

1912

Wales

Lady Chatterley’s Lover

1928

The White Peacock

1911

St. Mawr

1925

During World War I

Venice

After World War I

England, New Mexico

continued on following page

240

Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Table 1. Continued Author Mikhail Sholokho

Title

Publication

Period(s) within the work

Location(s) within the work

The Man Who Died

1929

Ancient times

Gorgoda Hill

Deshabarikko

1925

1917

Russia

And Quiet Flows the Don

1928-1940

During World War I

Tatar Ski Village near Byoshchenskaya in Don area

Virgin Soil Upturned

1932, 1959

Jan., 1930

Gremyachi Lok Village in the River Don Basin

Romain Rolland

Jean-Christophe

1904

19th century

Germany, River Rhine, Berlin, Paris, USA, Switzerland, Rome

梶井基次郎 [Kajii Motojirō]

檸檬 [Lemon]

1931

Early 20th century

Kyōgoku, Maruzen, Teramachi Street, Kyoto

佐藤春夫 [Satō Haruo]

田園の憂鬱 [Den’en no Yūutsu: Melancholy in the Country]

1919

Early 20th century

Nakazato-mura, Tsuzuki-gun, Kanagawa Prefecture, Japan

中島敦 [Nakajima Atsushi]

都会の憂鬱 [Tokai no Yūutsu: Melancholy in the City]

1922

Early 20th century

Musashino, Tokyo

山月記 [Sangetsuki]

1942

The Tō period (China)

Rousei, China

坂口安吾[Sakaguchi Ango]

白痴 [Hakuchi: The Idiot]

1947

During World War II

Kamata, Tokyo

古井由吉 [Furui Yoshikitichi]

行き隠れ [Yukigakure]

1972

The Shōwa era

Japan

川端康成[Kawabata Yasunari]

伊豆の踊り子 [Izuno Odoriko: The Dancing Girl of Izu]

1927

The Taishō era

Izu, Yugasima, Amagi, Shimoda

三島由紀夫[Mishima Yukio]

花のワルツ [Hana no Warutsu: Waltzu of the Flowers]

1936

The Shōwa era

雪国 [Yukiguni: Snow Country]

1937

The Taishō era-The Shōwa era

Yuzawa Onsen

古都 [Koto: The Old Capital]

1962

The Shōwa era

Kyoto, Nakagyō, Yasaka Shrine

千羽鶴 [Senbazuru: Thousand Cranes]

1952

The Shōwa era

Engaku temple, Kamakura, Takeda, Kyushu

掌の小説 [Tenohira (Tanagokoro) no Shōsetsu: Palm-of-the-Hand Stories]

1971

禽獣 [Kinju]

1935

The Shōwa era

Hibiya, Zen temple

抒情歌 [Jojōka]

1934

The Shōwa era

Hanatake Peach Blossoms on the Coast

愛する人達 [Aisuru Hitotachi]

1941

The Shōwa era

日も月も [Hi mo Tsuki mo]

1953

Shōwa 20’s

Kamakura

名人 [Meijin: The Master of Go]

1954

Jan/18–Dec/4, 1940

Atami

川のある下町の話 [Kawa no Aru Shitamachi no Hanashi]

1954

Shōwa 20’s

Tokyo

美しさと哀しみと [Utsukushisa to Kanashimi to: Beauty and Sadness]

1965

The Shōwa era

Kyoto, Kita Kamakura, Saihoji temple

眠れる美女 [Nemureru Bijo: The House of the Sleeping Beauties]

1961

The Shōwa era

Japanese-style hotel

みづうみ [Miduumi: The Lake]

1955

The Shōwa era

Tokyo, Shinsyu (Nagano), Karuizawa

片腕 [Takaude: One Arm]

1965

The Shōwa era

たんぽぽ [Tanpopo: Dandelions]

1972

The Shōwa era

東京のひと [Tokyo no Hito: Women in Tokyo]

1955

1950’s

Tokyo

女であること [Onna de Aru Koto]

1956-1957

The Shōwa era

Tokyo

美しい星 [Utsukushii Hoshi: A Beautiful Star]

1962

The Shōwa era

Han’no (Saitama), Mount. Rakan, Kanazawa Uchinada, Sendai

盗賊 [Tōzoku: Thieves]

1948

1930’s

San’nomiya (Kōbe), Murasakino (Kyoto)

Ikutachō, Jokoji temple, Miidera temple, Nishiizu

仮面の告白 [Kamen no Kokuhaku: Confessions of a Mask]

1949

Before/During World War II

Tokyo, Karuizawa

純白の夜 [Jumpaku no Yoru]

1950

1948

Azabu, Tokyo, Kamakura

愛の渇き [Ai no Kawaki: Thirst for Love]

1950

The Shōwa era

Umeda, Toyonaka, Osaka

青の時代 [Ao no Jidai]

1950

1923

Chiba, Tokyo, Toriizaki

禁色 [Kinjiki: Forbidden Colors]

1950–1951

Tokyo, Izu, Kyoto

潮騒 [Shiosai: The Sound of Waves]

1951-1953 1954

The Shōwa era

Utashima

沈める滝 [Sizumeru Taki]

1955

The Shōwa era

Tokyo, Nigata Kinkakuji temple, Naryu misaki, Nanzenji temple, Maizuru, Ryuhoji temple

金閣寺 [Kinkakuji: The Temple of the Golden Pavilion]

1956

After World War II

美徳のよろめき [Bitoku no Yoromeki]

1957

The Shōwa era

Tokyo

The Shōwa era

Yotsuya Higasi-Shinanocho, New York, Mexico

鏡子の家 [Kyōko no Ie: Kyoko’s House]

1959

continued on following page

241

Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Table 1. Continued Author

Raymond Radiguet Oscar Wilde 森鷗外 [Mori Ōgai]

泉鏡花 [Izumi Kyōka]

Title

宴のあと [Utage no Ato: After the Banquet]

Publication

Period(s) within the work

Location(s) within the work

1960

The Shōwa era

Tokyo

獣の戯れ [Kemono no Tawamure: The Frolic of the Beasts]

1961

The Shōwa era

Nishi-Izu, Ginza, Tokyo, Hamamatsu

午後の曳航 [Gogo no Eikō: The Sailor Who Fell from Grace with the Sea]

1963

The Shōwa era

Yamate, Tomioka, Yokohama, Yokohama port

音楽 [Ongaku: Music]

1965

The Shōwa era

Hibiya, Kōfu, Izu), Namida-bashi

絹と明察 [Kinu to Meisatsu: Silk and Insight]

1964

The Shōwa era

Ōmi (Shiga), Konkai Kōmyōji temple

遠乗会 [Tōnorikai]

1958

The Shōwa era

真夏の死 [Manatsu no Shi: Death in Midsummer]

1965

The Shōwa era

Izu

卵 [Tamago: Eggs]

1966

The Shōwa era

Tokyo

憂国 [Yūkoku: Patriotism]

1961

1936

Yotsuya, Tokyo

三熊野詣 [Mikumano Mōdo: Acts of Worship]

1953

The Shōwa era

Kumano, Kumano Nachi Taisya Shrine, Kumano Hayatama Taisya Shrine, Kumano Hongu Shurine

橋づくし [Hashi Dukushi]

1956

The Shōwa era

Tokyo, Shinbashi, Miyoshibashi, Tsukijibashi, Irifunebashi, Akatsukibashi, Sakaibashi, Bizenbashi

癩王のテラス [Raiō no Terasu]

1969

Late 12th century

Cambodia

鹿鳴館 [Rokumeikan]

1957

1886

Rokumeikan, Hibiya

豊饒の海(春の雪) [Hojo no Umi (Haru no Yuki): The Sea of Fertility (Spring Snow)]

1969

The Meiji era–The Shōwa era

Tokyo, Yamato (Nara)

豊饒の海(奔馬) [Hojo no Umi (Honba): The Sea of Fertility (Runaway Horses)]

1969

May, 1932–Dec, 1933

Osaka, Ōgami Jinja Shrine, Mt. Miwa, Isagawa Jinja Shrine, Yanagawa, Yamanashi, Manchuria, Ise Jingū Shrine, Izu

豊饒の海(暁の寺) [Hojo no Umi (Akatsuki no Tera):The Sea of Fertility (The Temple of Dawn)]

1970

1941–1945, 1952–1967

Bangkok, India, Gotemba

豊饒の海(天人五衰) [Hojo no Umi (Ten’nin Gosui):The Sea of Fertility:The Decay of the Angel]

1971

1970–1975

Miho no Matsubara (Shizuoka), Tokyo, Gesshūji (Yamato)

Devil in the Flesh

1923

1917

France

Count d’Orgel’s Ball

1924

1920

Robson, Basque Country London

The Picture of Dorian Gray

1891

19th century

Salome

1893

Ancient times

舞姫 [Maihime: The Dancing Girl ]

1907

Late 19th century

Germany, Saigon, Klosterstrasse, Berlin, Victoria Theater, Russia

ヰタ・セクスアリス [Vita Sexualis]

1909

The Meiji eara

Tokyo

青年 [Seinen: Young Men]

1913

The Meiji era

Tokyo

雁 [Gan: The Wild Geese]

1913

1880

Muenzaka, Tokyo

渋江抽斎 [Shibue Chūsai]

1916

The Edo period

Edo (Tokyo) River Takase, Kyoto

高瀬舟 [Takasebune]

1916

The Edo period

歌行燈 [Uta-andon: A Song by Lantern Light]

1912

The Meiji era

高野聖 [Kōya-hijiri: The Holy Man of Mount Koōya]

1908

The Meiji era

Wakasa, Echizen, Shinshu, Amou Tōge

婦系図 [On’na Keizu]

1908

The Meiji era

Yushima, Tokyo

日本橋 [Nihonbashi]

1914

The Meiji era

Nihonbashi, Tokyo

尾崎紅葉 [Ozaki Kōyō]

金色夜叉 [Konjijki Yasha: The Golden Demon]

1898-1903

The Meiji era

Atami, Shiobara, Tokyo

太宰治 [Dazai Osamu]

斜陽 [Shayō: The Setting Sun]

1947

1945

Nishikata, Tokyo, Izu

The Shōwa era

晩年 [Ban’nen: The Late Years]

1936

新釈諸国噺 [Shinshaku Shokokubanashi: A New Cersion of Countries’ Tales]

1945

津軽 [Tsugaru]

1944

The Shōwa era

Kanagimachi, Kodomari, Tsugaru (Aomori)

人間失格 [Ningen Shikkaku: No Longer Human]

1948

The Shōwa era

Tokyo

ヴィヨンの妻 [Viyon no Tsuma: Villon’s Wife]

1947

The Shōwa era

Asakusa park, Koganei, Nakano Station, Tokyo

Japan

continued on following page

242

Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Table 1. Continued Author René Descartes

道元 [Dōgen]

Title The World

Publication

Principles of Philosophy

1644

Discourse on the Method

1637

The Passions of the Soul

1649

正法眼蔵随聞記 [Shōhō Genzō Zuimonki: The Treasury of the True Dharma Eye: Record of Things Heard]

世阿弥 [Zeami]

風姿花伝 [Fūshikaden]

親鸞[Shinran]

歎異抄 [Tan’nishō]

Period(s) within the work

Location(s) within the work

1633

1235-1238 Early 15th century The end of the Kamakura era

Old Testament New Testament 古代歌謡集 [Kodai Kayō Shū]

Ancient times

Ancient Yamato area

中世・近世歌謡集 [Chūsei Kindai Kayō Shū]

The Kamukura period–The Edo period

Many areas in Japan

万葉集 [Man’yōshū] 古今和歌集 [Kokin Wakashū] 新古今和歌集 [Shin Kokin Wakashū] 金塊和歌集 [Kinkaiwakashū] 和泉式部歌集 [Izumi Shikibu Kashū] 義経記 [Gikeiki] 狭衣物語 [Sagoromo Monogatari]

紫式部 [Murasaki Shikibu]

上田秋成 [Ueda Akinari] 式亭三馬 [Shikitei Sanba]

井原西鶴 [Ihara Saikaku]

Ancient times

Many areas in Japan

c. 912

Kyoto, many areas in Japan

c. 1210

Kyoto, many areas in Japan

The Kamuraku period The Heian period From the Northern and Southern Dynasties to the The Heiji rebellion early Muromachi period

Kyoto, Ishiyama-dera Temple, Echizen, Rokujogawara, Kyoto, Mino

The Heian period

The Heian period

Tsukushi, Kyoto

夜の寝覚 [Yoru no Nezame]

The end of the Heian period

The Heian period

Kyoto

平家物語 [Heike Monogatari: The Tale of the Heike]

The Kamakura Period

The Kamakura period

源氏物語 [Genji Monogatarri: The Tale of Genji] 紫式部日記 [Murasakishikibu Nikki: The Diary of Lady Murasaki]

与謝野晶子[Yosano Akiko]

The end of the Nara period

源氏物語 [Genji Monogatari: The Tale of Genji]

The Heian Period The early 13th century 1911-1940

Kyoto, Kamakura, Setouchi, Dan’noura Kyoto, Suma, Akashi, Tsukushi, Many places in Japan

1008–1010

Kyoto

The Heian period

Kyoto, Suma, Akashi, Tsukushi, Many places in Japan

雨月物語 [Ugetsu Monogatari]

1776

The Edo period

Sanuki, Kōyasan, Dōjōji

春雨物語 [Harusame Monogatari]

1808

The Edo period

Kyoto, Sagami, Ōmi, Settsu

浮世風呂 [Ukiyoburo]

1809-1813

The Edo period

Edo (Tokyo)

浮世床 [Ukiyodoko]

1809

The Edo period

Edo (Tokyo)

好色一代男 [Kōshoku Ichidai Otoko: The Life of an Amorous Man]

1682

The Edo period

Kyoto, Osaka, Kyūshū, Ohara, Sado, Shimabara, Sano, Murotsu, Shinmachi, Nyogogashima

好色五人女 [Kōshoku Gonin On’na: Five Women Who Loved Love]

1686

The Edo period

Himeji, Osaka, Komagome Kisshōji teimple, Satsuma

日本永代蔵 [Nippon Eitaigura: The Eternal Storehouse of Japan]

1688

The Edo period

Edo, Kyoto, Izumi, Nanto, Ōtsu, Kinokuni, Sakata, Bungo, Fushimi, Suruga, Chikuzen, Echizen, Nagasaki, Yamazaki, Yamato, Hitachi, Sakushū, Bushū, Senshū, Yamashiro

世間胸算用 [Seken Munezan’yo: Reckonings that Carry Men Through the World or This Scheming World]

1692

The Edo period

Nagamachi, Sakai, Edo

西鶴織留 [Saikaku Oridome]

1694

The Edo period

藤原道綱母[Fujiwara no Michitsuna no Haha]

蜻蛉日記 [Kagerō Nikki]

974

957–974

Kyoto

和泉式部 [Izumi Shikibu]

和泉式部日記 [Izumi Shikibu Nikki]

1008

1003–1004

Kyoto

菅原孝標女[Sugawara no Takasue no Musume]

更級日記 [Sarashina Nikki]

藤原長子[Fujiwara no Nagako]

讃岐典侍日記 [Sanuki no Suke Nikki]

清少納言 [Sei Shōnagon]

枕草子 [Makura no Sohi]

The Middle of the Heian period

1020–1059

Kyoto

The End of the Heian period

The time of Emperor Horikawa and Emperor Toba)

Kyoto

The Middle of the Heian period

The Middle of the Heian period

Kyoto

continued on following page

243

Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Table 1. Continued Author 吉田兼好[Yoshida Kenkō]

Title 徒然草 [Tsurezuregusa]

鴨長明 [Kamo no Chōmei]

方丈記 [Hōjōki]

為永春水[Tamenaga Shunsui]

梅暦 [Ume Goyomi]

源信 [Genshin]

往生要集 [Ōjōyōshū]

夏目漱石[Natsume Sōseki]

吾輩は猫である [Wagahai ha Neko de aru: I Am a Cat]

永井荷風 [Nagai Kafū]

Ninanji temple, Kyoto

The End of the Edo period

Edo

The Meiji era

Tokyo

985 1905-1907

Jigoku [Hell]

1907

The Meiji era

Tokyo, Matsuyama

The Meiji era

Tokyo, kyoto, Ueno Park

坑夫 [Kōfu: The Miner]

1908

The Meiji era

A cole mine

三四郎 [Sanshirō]

1909

The Meiji era

Tarin, Tokyo

門 [Mon: The Gate]

1911

The Meiji era

Tokyo, Kamakura

それから [Sorekara: And Then]

1910

The Meiji era

Tokyo

こころ [Kokoro]

1914

The Meiji period

Kamakura, Yuigahama, Tokyo, Zoōhigaya

行人 [Kōjin:The Wayfarer]

1914

The Meiji era

Tokyo, Osaka

道草 [Michikusa]

1915

The Meiji era

Hongō, Tokyo

彼岸過迄 [Higansugimade: To the Spring Equinox and Beyond]

1912

The Meiji era

Ogawachō, Tokyo, Kansai

明暗 [Meian: Light and Darkness]

1917

The Meiji era

Tokyo, Izu

あめりか物語 [Amerika Monogatari: American Stories]

1908

Meiji 36–40

USA

1909

Meiji 40

France

1962-1965 1911

1917–1959

Ushigome, Tokyo

The Meiji era

Asakusa, Tokyo

腕くらべ [Udekurabe: Geisha in Rivalry]

1918

The Meiji era

Tokyo, Tōhoku

日和下駄 [Hiyori Geta]

1915

The Meiji era

Tokyo

おかめ笹 [Okame-zasa]

1920

The Meiji era

Tokyo

つゆのあとさき [Tsuyu no Atosaki: During the Rains]

1931

The Meiji era

Ginza, Tokyo

ひかげの花 [Hikage no Hana: Flowers in the Shade]

1946

The Meiji era

Tokyo

墨東綺譚 [Bokutō Kitan: A Strange Tale from East of the River]

1937

The Meiji era

Tamanoi, Mukōjima, Tokyo

踊り子 [Odoriko]

1948

Tokyo

雨瀟々 [Ame Shōshō]

1922

Ginza, Tokyo

浮雲 [Ukigumo: The Drifting Cloud]

1887-1889

The Meiji era

Tokyo

其面影 [Sono Omokage: An Adopted Husband]

1907

The Meiji era

Tokyo, Tenshin

平凡 [Heibon]

1908

The Meiji era

Tokyo

新世帯 [Shinsetai]

1909

The Meiji era

Tokyo

足迹 [Ashiato]

1910

The Meiji era

Tokyo, Asakusa, Yushima

黴 [Kabi]

1912

The Meiji era

Kyushū, Osaka

爛 [Tadare]

1913

The Meiji era

Tokyo

あらくれ [Arakure: Rough Living]

1915

The Meiji era

Tokyo, Shitaya, Hongō

仮装人物 [Kasō Jinbutsu]

1938

The Meiji era

Tokyo

縮図 [Shukuzu: Epitome]

1944

The Meiji era

Tokyo, Hakusan, Fujimi

破戒 [Hakai: The Broken]

1906

The End of the Meiji perido

Shinshū (Nagano)

春 [Haru: Spring]

1908

The Meiji era

Tokyo

家 [Ie: The Family]

1910

The Meiji era

Kiso (Nagano), Nagoya

新生 [Shinsei]

1919

The Meiji era

Tokyo, Paris

1853–1886

Kisodani (Shinshū), Tokyo

夜明け前 [Yoake Mae: Before the Dawn] 谷崎潤一郎[Tanizaki Jun’ichirō]

1170’s–1180’s

1908

すみだ川 [Sumidagawa: River Sumida]

島崎藤村[Shimazaki Tōson]

1212 1833-1835

Location(s) within the work Ninanji temple, Kyoto

坊ちゃん [Botchan]

断腸亭日乗 [Danchōtei Nichijō]

徳田秋声 [Tokuda Shūsei]

Period(s) within the work

虞美人草 [Gubijinsō: The Poppy]

ふらんす物語 [Furansu Monogatari: Frence Stories]

二葉亭四迷[Futabatei Shimei]

Publication

The End of the Kamuraku The Kamura period period

恐怖時代 [Kyōfu Jizai]

1932, 1935 1920

The Edo period

饒太郎 [Jōtarō]

1914

お艶殺し [Otsuya Goroshi]

1915

The Edo era

お国と五平 [Okuni to Gohei]

1922

The Edo era

Nasunogahara

痴人の愛 [Chijin no Ai: Naomi]

1925

1917-1922

Asakusa, Ōmori, Yokohama

細雪 [Sasameyuki: The Makioka Sisters]

1944–1948

1936–1941

Osaka, Ashiya, Ōgaki, Tokyo

Surugaya, Fukagawa, Mukōjima

continued on following page

244

Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Table 1. Continued Author 舟橋聖一[Funahashi Seiichi]

Title

雪夫人絵図 [Yuki Fujin Ezu: Portrait of Madame Yuki] ある女の遠景 [Aru On’na no Enkei ]

Publication

Period(s) within the work

1948

The Shōwa era

Location(s) within the work

1963

The Shōwa era

Tokyo

好きな女の胸飾り [Sukina On’na no Munakazari]

1967

The Shōwa era

Tokyo

木石 [Bokuseki]

1938

裾野 [Susono]

1952

The Shōwa era

とりかえばや秘文 [Torikaebaya Hibun]

1966

The Taishō era

Tokyo

瀬戸内晴美[Setouchi Harumi]

かの子繚乱 [Kanoko Ryōran]

1979

The Meiji era–The Shōwa era

Tokyo, France

岡本かの子[Okamoto Kanoko]

金魚繚乱 [Kingyo Ryōran: A Riot of Goldfish]

1937

The Taishō era

Tokyo

堀辰雄 [Hori Tatsuo]

福永武彦[Fukunaga Takehiko]

伊藤整 [Itō Sei]

安部公房 [Abe Kōbō]

大江健三郎[Ōe Kenzaburō]

高橋和巳[Takahashi Kazumi]

燃ゆる頬 [Moyuru Hoho]

1939

Karuizawa

聖家族 [Sei Kazoku: The Holy Family]

1932

Karuizawa

風立ちぬ [Kazetachinu: The Wind Has Risen]

1938

風土 [Fūdo]

1952

Japan

草の花 [Kusa no Hana: Flowers of Grass]

1954

Around 1945

Japan

忘却の河 [Bōkyaku no Kawa]

1964

Shōwa 30’s

Japan

死の島 [Shi no Shima]

1971

1953–1954

Hiroshima, Tokyo

廃市 [Haishi]

1960

The Shōwa era

A quiet, rural town with an abandoned loneliness

海市 [Kaishi]

1968

The Shōwa era

Izu, Tokyo

典子の生きかた [Noriko no Ikikata]

1953

The Shōwa era

Japan

若い詩人の肖像 [Wakai Hito no Syōzō]

1956

1928

Otaru, Tokyo

鳴海仙吉 [Narumi Senkichi]

1950

Around 1945

Hokkaido

氾濫 [Hanran]

1958

1940’s

Japan

砂の女 [Suna no On’na: The Woman in the Dunes]

1962

Dunes

燃えつきた地図 [Moetsukita Chizu: The Ruined Map]

1967

City

箱男 [Hakootoko: The Box Man]

1973

Ueno, Shinjuku

方舟さくら丸 [Hakobune Sakuramaru: The Ark Sakura]

1984

Underground quarry cave

万延元年のフットボール [Man’en Gan’nen no Football: The Silent Cry]

1967

みずから我が涙をぬぐいたまう日 [Mizukara Waga Namida wo Nugui Tamau Hi: The Day He Himself Shall Wipe My Tears Away]

1972

洪水はわが魂に及び [Kōzui ha Waga Tamashii ni Oyobi: The Flood Invades My Spirit]

1973

1960’s

Shikoku

The Shōwa era

Tokyo

悲の器 [Hi no Utsuwa: Vessel of Sorrow]

1962

Around World War II

Japan

捨子物語 [Sutego Monogatari]

1968

The End of the Pacific War

Osaka

我が心は石にあらず [Waga Kokoro ha Ishi ni Arazu]

1967

1950’s–1960’s

Japan

憂鬱なる党派 [Yūutsu naru Toha: A Melancholy Faction]

1965

The Shōwa era

Japan

邪宗門 [Jashūmon: Heretical Faith]

1965

Early Shōwa era

Japan

日本の悪霊 [Nihon no Akuryō]

1980

The Shōwa era

Japan

堕落 [Daraku]

1969

Around World War II

Japan, Manchuria

散華 [Sange]

1963

Around World War II

Japan

白く塗りたる墓 [Shiroku Nuritaru Haka]

1971

The Shōwa era

Japan

1968-未完

The Shōwa era

Japan

The Shōwa era

Japan

黄昏の橋 [Tasogare no Hashi] わが解体 [Waga Kaitai] 真継伸彦[Matsugi Nobuhiko]

Karuizawa 1923–1939

1971

文学の責任 [Bungaku no Sekinin]

1963

光る声 [Hikaru Koe]

1966

continued on following page

245

Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Table 1. Continued Author

Title

Publication

Period(s) within the work

Location(s) within the work

柴田翔 [Shibata Shō]

されどわれらが日々 [Saredo Wareraga Hibi]

倉橋由美子[Kurahashi Yumiko]

スミヤキストQの冒険 [Sumiyakisuto Q no Bōken: Adventures of Sumiyakisto Q]

1964

Sensitivity center on an isolated island

アマノン国探訪記 [Amanon-koku Tanbōki: Journey to Amanon ]

1986

Phantom Land Amanon

愛の生活 [Ai no Seikatsu: Love Life]

1968

夢の時間 [Yume no Jikan: Time of Dreams]

1970

岸辺のない海 [Kishibe no Nai Umi: The Shoreless Sea]

1974

夜になっても遊び続けろ [Yoru ni Nattemo Asobitsuzukero]

1974

金井美恵子[Kanai Mieko]

1964

1955

Tokyo

贈る言葉 [Okuru Kotoba]

1966

1960’s

Tokyo

わたしの中の彼へ [Watashi no Naka no Kare he]

1970

小田実 [Oda Makoto]

アメリカ [America]

1962

The Shōwa era

USA

丸山健二[Maruyama Kenji]

夏の流れ [Natsu no Nagare: Summer’s Passage]

1967

The Shōwa era

Japan

正午なり [Shōgonari]

1968

The Shōwa era

Japan

石川淳 [Ishikawa Jun]

白頭吟 [Hakutōgin]

1957

The Shōwa era

Japan

焼け跡のイエス [Yakeato no Yes]

1946

After World War II

Ueno

村上龍[Murakami Ryū]

限りなく透明に近いブルー [Kagirinaku Tomei ni Chikai Būu: Almost Transparent Blue]

1976

The Shōwa era

Fussa, Tokyo

外岡秀俊[Sotooka Hidetoshi]

北帰行 [Hokkikō]

1976

The Shōwa era

Train, Morioka

稲垣足穂 [Inagaki Taruho]

一千一秒物語 [Issen Ichibyō Monogatari: One Thousand and One Second Stories]

1923

小栗蟲太郎[Oguri Chutarō]

黒死館殺人事件 [Kokushikan Satsujinjiken]

1935

野間宏 [Noma Hiroshi]

暗い絵 [Kurai E: Dark Pictures]

1947

The End of the Pacific War

Osaka, Kyoto

崩壊感覚 [Hōkai Kankaku]

1948

The End of the Pacific War

真空地帯 [Shinkū Chitai: Zone of Emptiness]

1952

1944

島尾敏雄 [Shimao Toshio]

中上健次[Nakagami Kenji]

James Joyce

Japanese army

わが塔はそこに立つ [Waga Tō ha Soko ni Tastu]

1962

The Shōwa era

Kyoto

さいころの空 [Saikoro no Sora]

1959

The Shōwa era

Kabutochō, Tokyo

青年の環 [Seinen no Wa] 武田泰淳 [Takeda Taijun]

Kōza, Kanagawa

風媒花 [Fūbaika]

1949, 1950, 1966, 1970 1952

Jul–Sep, 1939

Osaka

After the Pacific War

Japan, China

貴族の階段 [Kizoku no Kaidan]

1959

The age of the 2.26 incident

森と湖の祭り [Mori to Mizuumi no Matsuri]

1958

The Shōwa era

Hokkaiso, Lake Akan

富士 [Fuji]

1971

The Shōwa era

A Mental hospital at Mt. Fuji

快楽 [Keraku]

1972

Shōwa 10’s

Tokyo

司馬遷 [Shibasen]

1943

Ancient China

China

ひかりごけ [Hikarigoke: Luminous Moss]

1953

The Shōwa era

Rausu, Hokkaido

出発はついに訪れず [Shuppatsu ha Tsuini Otozurezu]

1964

Durging Pacific War

Kakeromajima

死の棘 [Shi no Toge: The Sting of Death]

1977

The Shōwa era

枯木灘 [Karekinada: The Sea of Withered Trees]

1977

The Shōwa era

Kishū (Wakayama)

地の果て 至上の時 [Chi no Hate Shijō no Toki: Supreme Time at the End of the Earth]

1983

The Shōwa era

Singu, Kishū (Wakayama)

千年の愉楽 [Sen’nen no Yuraku: A Thousand Years of Pleasure]

1982

The Shōwa era

Singu, Kishū (Wakayama)

熊野集 [Kumanoshū]

1984

The Shōwa era

Kishū (Wakayama)

岬 [Misaki: The Cape]

1976

The Shōwa era

Kishū (Wakayama)

Ulysses

1922

Jun/16, 1904

Dublin, Ireland

Dubliners

1914

Dublin, Ireland

continued on following page

246

Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Table 1. Continued Author

Title A Portrait of the Artist as a Young Man

Virginia Woolf Henry James

Émile Zola

Honoré de Balzac

Publication

Period(s) within the work

1916

Finnegans Wake

1939

To the Lighthouse

1927

Mrs Dalloway

1925

Location(s) within the work Dublin, Ireland Ramsey, Isle of Skye

After World War I

London

The Bostonians

1886

19th century

Mississippi, Boston, Harvard University

The American

1877

19th century

Europe, Paris

The Wings of the Dove

1902

19th century

Europe

Roderick Hudson

1875

19th century

Europe

The Turn of the Screw

1898

19th century

England

Therese Raquin

1867

19th century

Oreluan, Algeria

L’Assommoir

1877

19th century

Paris

Nana

1880

Second Empire

Paris

The Joy of Living

1884

1863–1873

Bonneville Village, France

Germinal

1885

1860’s

Coal mine town monsue

La Terre

1887

19th century

The Lily of the Valley

1836

Eugenie Grandet The Quest of the Absolute

1833 1834-1835

France France Country town Saumur, Paris, India

French royal period

France

Ivan Goncharov

Oblomov

1859

19th century

Russia

Marquis de Sade

美徳の不幸 [Bitoku no Fukō: The Misfortunes of Virtue]

1791

18th century

France

悪徳の栄え [Akutoku no Sakae: Juliette] The 120 Days of Sodom

Georges Bataille

1797–1801

18th century

France

1785

18th century

Schwarzwald old castle, Schilling castle

Philosophy in the Bedroom

1795

18th century

France

悲惨物語 [Hisan Monogatari]

1800

18th century

France

Madame Edwarda

1937

20th century

Paris

Story of the Eye

1928

20th century

Spain

The abbot C

1971

During World Wor II

France

The Accursed Share: The Notion of Expenditure

1951

A future age

Future Empire EHS

Erotism

1961

Guillaume Apollinaire

The Eleven Thousand Rods

1907

沼正三 [Numa Shōzō]

Kachikujin Yapū

1970

John Cleland

Fanny Hill

Jean-Jacques Rousseau

Emile

Michel de Montaigne

1748-1749

18th century

London

1762

18th century

France

The New Heloise

1761

18th century

France

Confessions

1,781

18th century

France

Reveries of a Solitary Walker

1778

18th century

France

Essays

1580

Niccolò Machiavelli

The Prince

1532

Georg Wilhelm Friedrich Hegel

The Phenomenology of Spirit

1807

Vladimir Nabokov

Paris

Lolita

1955

1947–1949

Europe, USA

The Real Life of Sebastian Knight

1941

1899–1936

Paris

小島信夫 [Kojima Nobuo]

抱擁家族 [Hōyō Kazoku: Embracing Family]

1965

The Shōwa era

Setagaya, Tokyo

Homer

Iliad

Ancient Greece

Troy

François Rabelais

Gargantua and Pantagruel]

島 [Shima: Island]

Victor Hugo

The Hunchback of Notre-Dame

Gottfried Keller

Green Henry

Aldous Huxley

Point Counter Point

1956 8th century BC

An island Poitiers, Oreluan, Paris, Utopia, Theleme Abbey

1532-1564 1831 1879-1880

1482

Paris

19th century

Munich

During World War I

Europe

During World War I

Paris

1928

Roger Martin du Gard

The Thibaults

1922

Samuel Beckett

Malone Dies

1951

Andrei Bely

The Silver Dove

1910

Louis-Ferdinand Céline

Journey to the End of the Night

1932

Russia

Cesare Pavese

The Fine Summer

1949

20th century

Italy

Günter Grass

The Tin Drum

1959

1954(1899–1954)

Danzig Freedom City

Saul Bellow

Dangling Man

1944

20th century

Chicago

Jerome David Salinger

Nine Stories

1953

20th century

USA

continued on following page

247

Bridging the Gap Between Narrative Generation Systems and Narrative Contents

Table 1. Continued Author

Title

Publication 1967

Period(s) within the work 20th century

Location(s) within the work

Jean-Marie Le Clézio

Terra Amata

Nice

The Interrogation

1963

20the century

France

Philippe Sollers

The Park

1961

20th century

France

Drieu La Rochelle

Gilles

1939

20th century

France

Jorge Vargas Llosa

Conversation in the Cathedral

1969

20th century

Bar La Cathedral

José Donoso

The Obscene Bird of Night

1970

20th century

Julio Cortázar

Hopscotch

1963

20th century

Paris

Manuel Puig

Kiss of the Spider Woman

1976

20th century

Prison in Buenosaires

Gabriel García Márquez

The Autumn of the Patriarch

1975

石原慎太郎[Ishihara Shintarō]

化石の森 [Kaseki no Mori: Petrified Forest]

1970

The Shōwa era

Tokyo

高橋源一郎[Takahashi Genichirō]

ジョン・レノン対火星人 [John Lennon Tai Kaseijin: John Lennon vs the Martians]

1985

The Shōwa era

Tokyo Detention Center

埴谷雄高 [Haniya Yutaka]

死霊 [Shirei: Death Spirits]

Shōwa 10s

Tokyo

闇のなかの黒い馬 [Yami no Naka no Kuroi Uma: Black Horses in the Darkness and Other Stories] Aleksandr Solzhenitsyn

1970

The Gulag Archipelago

1973-1975

Former Soviet Union

Concentration Camp Glag, USSR

Cancer Ward

1968-1969

1955

USSR

1950’s

USSR

One Day in the Life of Ivan Denisovich Red Wheel

248

1945-95

fictional country facing the Caribbean Sea

1962 1993-1997

Eastern Prussia, Tannenberg

249

Chapter 7

Haiku Generation From Narratological Perspective:

A Circulation Between Haikus and Stories Jumpei Ono Aomori University, Japan Takashi Ogata https://orcid.org/0000-0001-9181-8876 Iwate Prefectural University, Japan

ABSTRACT Haiku—a form of unrhymed poetry—is popular among the Japanese. A typical haiku is composed of 17 moras and three phrases. A haiku has the possibility of scratching the surface and uncovering a hidden message through an expression of events. According to Masaoka Shiki, a haiku is a kind of literature and has high affinity with our research on generating stories. In this chapter, the authors implemented the prototype system that has two functions: first, to produce multiple haikus from a single story, and second, to engender multiple stories from haikus. The system prototyped in this chapter is based on haiku theory, which is used by the authors in their research, and is rooted in the concept’s co-occurrence information and frequency information used to generate a haiku. The method uses statistical information for selecting the words and creating the word network in the haiku. Through the aforementioned methods, the authors created a framework for a system of circulating haiku and stories and proposed a kind of narrative generation with narrative as an input.

INTRODUCTION A haiku is a type of poem. This art form emerged in Japan and has been familiar to Japanese people for many years. A haiku poem is composed of 17 moras and 3 lines: the first line has 5 moras, the second 7, and the third 5. However, this standard definition is only a reference and does not define all types of haiku. Furthermore, many poets have attempted to describe haiku, with Matsuo Basho and Masaoka DOI: 10.4018/978-1-7998-4864-6.ch007

Copyright © 2021, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

 Haiku Generation From Narratological Perspective

Shiki being some of the famous haiku poets. In Haikai Taiyo, Masaoka Shiki (1966) mentions that haiku is a literary form, pointing out that what makes it different from other types of literature is the number of moras, although haiku does not necessarily consist of 17 moras. In Haiku no Tukuriyo (“How to Make a Haiku”), Takahama Kyoshi (2009), an apprentice of Masaoka Shiki, comments on the need for seasonal words and cuts based on 17 moras. However, there is a view that haiku does not have a strict formal style. Ishihara (2012) thus asserts that a haiku style is defined on the basis of a poetic mindset (emotions, scenery, etc.). As maintained by Masaoka (1966), haiku is a literary form, and it shows a high affinity for our research on generating stories. In this chapter, we propose a system for the following two functions and implement a prototype. The first function is to produce multiple haikus from a single story. The second function is to generate multiple stories from haiku. The authors have approached the haiku production using two methods. One of them is grounded in the structural aspects of haiku, while the other are based on scrutinizing the transitions of the parts of speech in haiku. Ito and Ogata (2019) consider that a haiku consists of parts of speech, and they created haiku using the transition patterns of parts of speech in each line. Specifically, this approach utilizes the transition patterns of nouns and verbs only (Ito, Igarashi, & Ogata, 2018). The target of analysis to acquire the transition pattern is 63 phrases published in Oku no Hosomichi (Matsuo, 1702; Imoto, Hisatomi, Hori, Yamashita, & Maruyama,2008). First, each phrase is subjected to a morphological analysis, and only the nouns and verbs of the obtained parts of speech are examined. The haiku is a fragmentary set of words and is an experimental effort based on the expectation that nouns and verbs make up the majority of it. In this paper, we focus only on the number of occurrences of nouns and verbs. If other parts of speech (e.g., adjectives) appear first, we begin counting from the first noun or verb that appears in the line. The second strategy for constructing a haiku uses a time series based on deep learning and long short-term memory (LSTM). The third strategy uses umeji, which harnesses the method of devising a story. The authors concentrate on a strategy of dividing the haiku into word and character units, grounded in the approach of deep learning employed by Ono and Ogata (2018, 2019). In each type of learning, the 63 phrases in Oku no Hosomichi were used to study the data. Igarashi harnessed a “chainer” for the framework of deep learning and produced a haiku by learning to construct the LSTM. The learned pattern is the appearance pattern of the division unit. The target haiku was given a start symbol and end symbol. The mechanism was taught 10,000 times under the above conditions. The authors tried deep learning for the appearance pattern of words in a haiku; the appearance pattern of characters in the haiku for the learning method is described. With these learning data, a haiku was formed by concatenating words or characters from the start symbol to the end symbol. The basic procedures are as follows: 1. 2. 3. 4.

Begin using the start symbol. Select the element based on the learning data for the element that appears next to the start symbol. If the end symbol is selected, the haiku is complete. Otherwise, go to Step 4. Choose elements based on learning data for elements that appear next to the selected element. Return to Step 3.

With the above technique, if the number of times needed to learn is small, both very short and very long haikus are generated. When the number of times needed to learn has reached 1,000, the results that satisfy the form of the haiku are generated. At 2,000 times, the same haiku was created as input data. Most 250

 Haiku Generation From Narratological Perspective

haikus constructed after 5,000 times became the same as the input data. First, this subsection explains the experimental outcomes. Second, we discuss the results. Through this experiment, we found that the comprehensibility of the haiku and the degree to which it is interesting tend to be inversely proportional. Although there is no clear difference, this finding will serve as a foothold for experiments on a larger scale. In this research, by assessing the generated haiku based on two types of evaluation criteria, we discovered that comprehensibility and interest in the learning process tend to be inversely proportional. This chapter aims to develop a system that establishes a story summary and new stories by producing haikus.

BACKGROUND Haiku Generation Based on Umeji Another approach is to model how poets describe the process of writing haiku. The method of generating one statistically has a limit in exceeding a certain level. Therefore, it may be significant to derive a computer science model from the poet’s approach to writing haiku. We studied a computational science technique to constructing a haiku based on umeji, as described by Takahama (2009). He demonstrated in Haiku no Tukuriyo the first approach to haiku creation by “arranging 17 moras randomly.” Thus, statistically random creation is not an incorrect way of forming haiku. However, deep learning for haikus is not a suitable strategy because it is based on existing haikus. Here, we explain the architecture of the proposed system. There are two large modules. One produces a haiku using a story. The other generates a story from a haiku. For the module that produces a haiku from a story, we used the module that we mentioned previously. This module focuses on a sentence in the story, collects and edits related sentences to generate three lines, and combines them into one haiku. The module extracts elements such as people, places, and verbs from the observed sentences and gathers sentences with the same components. In this paper, we extend the method and comprehensively establish a haiku using the module. Instead of devising a single haiku, we create one from every angle. With this approach, the elements that constitute the story are thoroughly extracted. The discourse concept can be used as an exhaustive extraction technique. In particular, the notion of the viewpoint corresponds to the method. Haikus shaped by this method can be perceived as a network of story components. When producing a story from a haiku, all or a part of the generated haiku can be input. The components of the story that have been broken down into weak connections are once again given a strong connection by writing the story. This process uses an integrated narrative generation system. The technique suggested in this paper includes the above-mentioned work by Takahama. However, we propose not only a haiku but also a cyclical creation strategy, which includes generating a story from a haiku. This cyclical procedure is used in the process of forming a story described by Ogata and is based on the haiku theory outlined by Nitta (2018). Nitta stated that the recipient can transform the haiku context (narrative) and refer to the transformation as a strolling ontology. Recipients have their own context and accept the haiku context (narrative) in that context. The cyclical generation set forth in this paper represents the transformation. Generating a haiku by using a story as an input means creating a haiku, and generating a story using a haiku as an input embodies acceptance.

251

 Haiku Generation From Narratological Perspective

We approached the haiku generation from two aspects. The first one involves focusing on haiku’s structural features. One entail analyzing the part-of-speech transitions in the haiku; the other requires deep learning to produce the haiku. These attempts yielded particular results. Another technique is to model the haiku generation method described by the poet. We consider the first approach of forming a haiku statistically, to have a limit in exceeding a certain level. Therefore, the second method can be beneficial in creating a computer science model from the haiku creation approach commented on by the poet. We studied a haiku generation method based on computational science grounded in umeji, referred to by Takahama. In Haiku no Tukuriyo (2009), Takahama showed that randomly arranging 17 mora is the first haiku creation method. Hence, it is not wrong to produce a statistically random haiku. However, the deep learning of a haiku is not suitable because it uses an existing haiku. As mentioned above, this paper uses the approach by Takahama. However, we propose not only haiku generation but also cyclical formation, including devising a story from a haiku. This cyclical generation is used in the story process described by Ogata, and it refers to the haiku theory outlined by Nitta (2018). Nitta stated that the recipient transforms the haiku context (narrative) and expresses the transformation as a strolling ontology. Recipients have their own context through which they accept the haiku context (narrative). The cyclical generation set forth in this paper represents the transformation. Shaping a haiku by using a story as an input is the process of creating a haiku, and generating a story by using a haiku as an input is a process of acceptance. From the perspective of narratology, creating a haiku has a discourse aspect. Discourse is the concept presented by Genette (1972). Thus, haiku plays the role of a discourse technique. For example, it is possible to make a haiku by cutting a story from a particular viewpoint. Alternatively, a specific event is missing. A haiku can be established by employing various discourse-technique-like operations, such as reversing the time series of events. The following paragraph exemplifies the architecture of the proposed system, which uses two large modules. One generates a haiku from the story, while the other produces stories from the haiku. The module that creates a haiku from the story uses the module that we have previously published. It considers the sentences in the story, collects and edits sentences related to it, generates three lines, and combines these lines into one haiku. The module extracts elements such as people, places, or verbs from the observed sentences and collects sentences with the same elements. In this chapter, the technique is expanded to comprehensively form haikus by employing modules. Rather than creating a single haiku, it generates haiku from every perspective. With this type of haiku generation, the components that make up the story are thoroughly extracted. The concept of discourse can be used as a comprehensive extraction approach. In particular, the concept of the viewpoint corresponds to the method. The haiku generated using this strategy can be regarded as a network of story components. When generating a story from a haiku, all or a part of the haiku consists of input. A story component that has been broken down into weak links can once again be given strong links as a story through story production. This story creation uses an integrated narrative generation system. We suggest a system for generating stories and haikus using the above technique. When the haiku is defined on the basis of Masaoka’s standards, the usual works using ordinary words are not good haikus. Based on this, using the deep-learning method only can lead to a type of haiku generation that looks like natural language. Moreover, this method has the possibility of approaching an expression of events via the surface, as well as a hidden message by expressing those events. In another study, we conducted the commercial message generation, and we believe in using the knowledge of this paper in the generation. 252

 Haiku Generation From Narratological Perspective

We carried out experiments to produce sentences (like those in a haiku) through deep learning, with haikus used as learning data. Even in this generation, as the number of times needed to learn increased, the comprehensibility of the sentences tended to comprise the output. However, when considering the fun of writing a haiku, it is hard to say that its comprehensibility was not necessarily interesting. The hypothesis was that haikus that seemed rather difficult to understand might be more interesting. Therefore, by evaluating the comprehensibility and fun at each constant learning number, we investigated the relationship between the two values.

Haiku Generation Based on the gap Technique Ito, Ono and Ogata (2019) proposed haiku generation based on the gap technique. Regarding this method of production, we consider the following two techniques: (1) selecting two sentences that have a character as a designated agent, from a narrative, and (2) generating a haiku using the elements from these sentences. We expect to form a haiku to express an overview of the character’s behavior in the narrative. What follows is a generation example using the following narrative: Abstract sentences have a designated agent, and two sentences are chosen from this abstraction: When “wolf” is selected, we subtract “The wolf attacks sheep in a stock farm” and “The wolf escapes.” Figure 1. Input story

253

 Haiku Generation From Narratological Perspective

An abstract verb case and a purpose for the sentences are chosen in (1) Extract “attacks” and “sheep” from the first sentence. Abstract “escapes” from the second sentence. “The wolf” is S, and “attacks” is V1. “Sheep” is P1, and “escapes” is V2. Then, we make a format and arrange the words in a sequence for each selected sentence: Here, we use “S V1+P1 V2,” setting elements according to this format: “狼 (the wolf) 襲う(attacks) 羊を (sheep) 逃げる (escape).” Adjustments are made to form a proper haiku: “the wolf” is added as a mora Kirezi, because “the wolf” consists of four moras. Similarly, “escapes” is three moras, so two moras are added to Kirezi. Finally, we end up with the following haiku: “狼 (the wolf) や (ya) 襲う(attacks) 羊 (sheep) を 逃げ る (escape)かな (kana).”

GENERATION SYSTEM THROUGH CIRCULATION BETWEEN HAIKUS AND STORIES Our main goal here is to create a haiku and a story cyclically. Cyclic in this case implies a mechanism whereby mutual inputs can be inserted into a production system, and one input can output the other. The simplest form of this cycle is when a haiku is formed from a story, and the original story is produced from the generated haiku. This is an example of haiku generation as a narrative creation technique. In addition, we consider a method to shape a story, unlike the original one, based on the generated haiku. This is an attempt to form a new story through the mutual generation of the haiku and the story. A haiku is generated from a certain story, and we assess whether the original story can be produced on the basis of the generated haiku. We propose a haiku/story flow, as shown in Fig. 2. The haiku and the Figure 2. Mutual generation by translating the haiku and the story

narrative are connected cyclically by two processes: (1) narrative generation from the haiku and (2) the creation of the haiku from the narrative. When starting a story, first, a story is given as input to the system. Such a story is transformed into a haiku that focuses on some point in the process of producing a haiku from the story. The haiku obtained through this process is developed into a story once again through story generation from the haiku. The story obtained at this time is elaborated around the elements that the input haiku is paying attention to, so we expect that the center of the original story and the topic will change.

254

 Haiku Generation From Narratological Perspective

To start a haiku, first, we input a haiku into the system. At this time, the input haiku considers both the existing haiku and the haiku produced using the haiku generation method proposed in this system. Next, we move on to creating stories from a haiku. One generation method is selected to form a story. Next, we move on to create a haiku from a story. The generated narrative is converted back into the haiku format using the chosen method. At this time, when a sentence is converted from a story to a haiku, the output changes depending on the technique used, and a resemblance to the first input haiku may be generated. Interconversion uses not only superficial expressions but also structural aspects. Figure 3 depicts an example of one of the readers’ interpretations of the haiku shown in Ito et al. (2017), which was written and analyzed. To create a story from a haiku, we asked the reader to read and interpret it. We performed an experiment in which the reader produced a story about the haiku. We examined the relationship between the haiku and the narrative regarding the development of the expression from the narrative obtained from the experiment. From the results, we devised a haiku and a semantic network of stories, as presented in Fig. 3. A story generation approach using a semantic network and an integrated story Figure 3. Semantic network of the haiku and the story (Source: Ito, Arai, & Ogata, 2017)

system was also developed. Specifically, we tried to create a new story by replacing words in the haiku in the interpretation of the haiku itself, with different words using a concept dictionary. Here, the haiku’s structure is generated at the level of the case structure, without analyzing the meaning up to the level shown in Fig. 3. Attempts were made to create a haiku using all four elements of the story (in this case, people, things, places, and events).

255

 Haiku Generation From Narratological Perspective

IMPLEMENTATION OF THE GENERATION SYSTEM This section covers the methods of generation for haikus and stories. The first subsection portrays a haiku generation flow. The next subsection describes the story generation flow. Moreover, the last subsection contains a discussion. The system was made using Common Lisp.

METHOD FOR GENERATING A HAIKU The section explains the following steps to produce a haiku: 1. Select one or more events in the input story. 2. Construct haiku structures. 3. Make an empty story structure.

Step 1. This step includes selecting a sentence that has a character as a designated agent and choosing a sentence that appears just before the selected sentence or immediately afterward. A haiku is produced using the elements from these sentences. We expect that it will be created so as to describe one scene from the narrative. We demonstrate a generation example using a narrative. An abstract sentence has a designated agent, and a sentence is selected that appears just before the chosen sentence or immediately afterward: When “the wolf” is chosen and abstracted, the sentence “the wolf escapes” appears, followed by “Alain hits the wolf” and “Alain misses the wolf.” In this example, “Alain hits the wolf” was selected. Figure 4. Example of input story

256

 Haiku Generation From Narratological Perspective

Step 2. Abstract sentences contain a word representing a designated location. Two sentences from this abstraction are selected: When “his house” is chosen, abstract sentences contain “his house”: “A father is embarrassed by a wolf in his house,” “His son, Alain, encourages his father in his house,” and “Alain washes off his clothes in his house.” In this example, we selected “Alain washes off his clothes in his house.” A designated agent, a verb case, and a purpose from the sentences were selected. “Alain,” “washes off,” and “his clothes” come from “Alain washes off his clothes in his house.” “Alain” is S, “washes off” is V, and “his clothes” is P, while “his house” is L. The format is made, and the words are arranged in the sequence for each of the selected sentences. Here, the format “L S+V2P” is used, and the elements are arranged accordingly: “家屋 [his house] アラン [Alain] 洗う[washes off] 衣服 [his clothes].” The elements are adjusted to properly form a haiku: “アラン [Alain] 洗う [washes off]” is added to a mora kireji (Table 1), because it comprises six moras. Similarly, “家屋 [his house]” and “衣服 [his clothes]” are three moras, so two moras are added to kireji. Finally, we obtain the haiku: “家屋かな アラン洗うや 衣服なり.” In particular, we make a rule using two moras kireji. “けり(keri),” “たり (tari),” and “なり(nari)” are auxiliary verbs in classical Japanese grammar. “けり” indicates the past and mourning, and it is connected to the conjunctive form of the conjugated word by utilizing the La variant. “なり” denotes assertion/existence, and it is used in the adjective verb Nari inflectional type to be attached to nouns, adnominal forms, adverbs, and particles. “たり” means completion and survival, and it can be used as a continuous form of an inflection word by utilizing the La variant. “たり” has a more conclusive meaning, depending on the context, and can be put into a physical form by employing the adjective verb tari inflection type. Here, the nouns are nari and tari, and the verbs are keri and tari. Table 1. Kireji table Mora

Kireji

One mora

やら し ぞ か よ せ や つ れ ぬ ず に へ け じ

Two moras

かな けり たり

Three moras

もがな

Figure 5. Example of generated haiku structure and sentences

Step 3 For the events, the input story structure is removed. The system obtains an empty story structure, and outputs generate haikus and an empty story structure. Figure 6 shows an example of the output data.

257

 Haiku Generation From Narratological Perspective

Figure 6. Haiku structure and a story structure

METHOD FOR STORY GENERATION USING A HAIKU This section explains the following generation steps. 1. Select one verb (a noun) from the input haiku. 2. Obtain a verb with a similar frequency of information from the frequency information of the selected verb (noun). 3. Insert the generated events into an empty tree.

Step 1 A verb or noun is chosen using frequency information (Appendix 1). The c verb (noun) is used for the first generation of the event.

Step 2 The system chooses verbs for event production. The verb is selected on the basis of the number of empty events in the empty story structure. Figure 6 outlines examples of the generation functions. Figure 7 depicts the verb frequency data and verb corresponding data. (Appendix 2 contains more examples.)

Step 3 The events generated in Step 2 are inserted into an empty tree. Figure 9 shows an example of a generated story.

258

 Haiku Generation From Narratological Perspective

Figure 7. Examples of functions

Generation Experiments This section covers two experiments, such as those in Figure 10. The first experiment demonstrates a cycle between a story and a haiku. The second experiment depicts continuous transformation between stories and haikus. Two experiments used a common story (Figure 4) as input data. Input stories, generated haikus, and generated stories are calculated as corresponding parameters. The parameter is used to compare the stories and haikus. If the two parameters are separated, the content in the story translates to haiku generation.

Regression Experiment Regression experiments generate a haiku from a story and use the generated haiku to check whether a story that is close to the original can be generated. Figure 11 shows examples of the generated story. The co-occurrence information score of the haiku generated from the story and the co-occurrence information score of the story generated from the haiku were compared, and a story with a difference within 0.2 was judged to be a regressed story.

259

 Haiku Generation From Narratological Perspective

Figure 8. Example of corresponding information

Transformation Experiment Figure 9. Example of a generated story

260

 Haiku Generation From Narratological Perspective

Figure 10. Example of a generated story

Regression experiments generate a haiku from a story and use the generated haiku to examine whether a story that is not close to the original can be generated. Figure 12 shows examples of the generated story. The co-occurrence information score of the haiku generated from the story and the co-occurrence information score of the story generated from the haiku were compared, and a story with a difference within 0.2 was judged to be a regressed story.

RELATED WORKS Haiku generation research has been conducted since 2016, and researchers at Microsoft Development Ltd. and Hokkaido University have proposed a method using deep learning. In both cases, learning was attempted by employing the LSTM method with Matsuo Basho’s or Kobayashi Issa’s haikus as input. Furthermore, researchers at Hokkaido University have attempted to generate haiku from input images by matching images with haikus. Both methods are based on statistical data. Furthermore, researchers at the Kyoto Institute of Technology have proposed haiku generation using generative adversarial networks, which are widely used in the field of image generation. In this study, the standard used for the evaluation was whether the generated haiku looked like a human-created haiku. Researchers at Ben Gurion University of the Negev approached creativity in haiku generation, with the word associative norm as the axis of generation. When broadening their horizons and looking at the study of poetry generation, researchers at Tsinghua University tried to create Chinese poetry. Their study controls sentiment. Therefore, a sentiment controllable poetry generation model was defined, and the researchers evaluated the sentiment of the generated haiku. The core idea of AI Issa-kun (Chibi), which the research team of Hokkaido University first worked on, was learning, and learning with hiragana haiku data that could measure the number of moras to reliably form the five–seven–five structure. Thus, about 20,000 Kobayashi Issa’s Hiragana Haiku character

261

 Haiku Generation From Narratological Perspective

Figure 11. Examples of a generated story in the regression experiment

strings were divided into morphemes by the morphological analysis engine MeCab. The haiku divided into morphemes was input into the Recurrent Neural Network (RNN), and Learn, which Yoneda, Yokoyama, Yamashita, and Kawamura (2018) proposed as the methods that learn relations between an image and a haiku and generate haikus showing relationships with a motif in an image. Wu, Klyen, Ito, and Chen (2017) proposed generating haiku-like sentences with elements that were uttered during the talk with Chatbot. Hirota, Oka, Araki, and Tanaka (2018) proposed methods that learn a large quantity of haiku data and character data strings containing five or seven syllables, and generate haikus using those data. In these previous studies, haiku was generated using elements that were strongly interrelated. As a result, the generated haiku is a common haiku that can be expected. The authors think that it is difficult to reach “surprising” results compared to previous studies. In addition, the authors think that the previous haiku generation methods are only a combination of elements. morphemes appear.

262

 Haiku Generation From Narratological Perspective

Figure 12. Example of a generated story in the transformation experiment

CONCLUSION In this paper, the authors propose a haiku generation system based on mutual transformation between a narrative and haiku-like sentences. In particular, this paper presents a method for generating haiku-like sentences from a narrative fragment. In addition, the authors have been developing an integrated narrative generation system that automatically generates narratives. Thus, this paper aims to combine this mechanism with the above-mentioned system in the future. Basic elements for generating a narrative and a haiku-like sentence are commonly the selection, arrangement, and modification of the elements. To generate haiku-like sentences from a narrative, the proposed method selects particular elements from a narrative and arranges them in the form of a haiku-like sentence. Further, it provides variations to the haiku-like sentence. This paper shows several forms of haiku-like sentence generation to experimentally generate haiku-like sentences from a narrative. The authors discovered that by evaluating the generated haiku based on two types of evaluation criteria, the understandability and interest in the learning process tend to be inversely proportional. In addition, although the number of learning times that make

263

 Haiku Generation From Narratological Perspective

understandability and interestingness increase in a well-balanced manner was not found in this experiment scale, the authors would like to discover the number of learning that harmonizes interestingness and understandability in the future. To realize the framework described, the following issues should be considered: Morphological analysis of haiku: Since haiku is a set of fragmentary words, ordinary sentences, and parts of speech, it varies greatly. In addition, the notation sway is intense, especially when including former pseudonyms. The generation of events plays an important role in the concept described in this paper, and the acquisition of elements such as verbs is essential for the generation of the event. It is conceivable to extend the language notation dictionary of the integrated narrative generation system and manually correct the elements. Correspondence with sentence pattern: In this study, an element is inserted into a sentence pattern by forcibly generating an event (sentence) and ignoring the constraint condition. However, if the event is generated by ignoring the constraint condition, unnatural events (sentences) will be created, which may affect the application of the subsequent gap technique. Elements that can be obtained from haiku are limited, and there is a high possibility that an element will not correspond with the current sentence pattern. Therefore, it is necessary to expand the technique to be able to flexibly deal with combinations of various elements. Correspondence to haiku without a verb: For this study, the authors have assumed that haiku has one or more verbs as a premise. However, haiku without verbs also exists. When dealing with such haiku, it is necessary to first add a verb in some form.

REFERENCES Genette, G. (1972). Discours du recit, Essai de methode, Figures III. Seuil. Hirota, A., Oka, N., Araki, M., & Tanaka, K. (2018, March). Haiku generation by seqGAN that divided learning data set. Proceedings of the Twenty-fourth Annual Meeting of the Association for Natural Language Processing, 1292-1295. Imoto, N., Hisatomi, T., Hori, N., Yamashita, K., & Maruyama, K. (2008). Nihon no koten wo yomu: Vol. 20. Okuno hosomichi Basho / Buson / Issa syu. Shogakukan. Ishihara, Y. (2012). Haiku no tukurikata [How to make a haiku]. Meiji Shoin. Ito, T., Arai, T., & Ogata, T. (2017). Narrative generation from the semantic networks of a haiku. Proceedings of the 31st Annual Conference of the Japanese Society for Artificial Intelligence. Ito, T., Igarashi, K., & Ogata, T. (2018). A consideration on a multiple approach for haiku generation Proceedings of the 32nd Annual Conference of the Japanese Society for Artificial Intelligence. Ito, T., & Ogata, T. (2019). A Framework for Haiku Generation from a Narrative. Journal of Robotics. Networking and Artificial Life, 6(1), 23–26. doi:10.2991/jrnal.k.190531.005 Ito, T., Ono, J., & Ogata, T. (2018). Haiku generation using gap techniques. Proceedings of 2018 International Conference on Artificial Intelligence and Virtual Reality, V-004. 10.1145/3293663.3293666

264

 Haiku Generation From Narratological Perspective

Ito, T., Ono, J., & Ogata, T. (2018). Haiku generation using gap techniques. Proceedings of 2018 International Conference on Artificial Intelligence and Virtual Reality. Masaoka, S. (1966). Haikai taiyo [Important points of Haikai]. Iwanami Shoten. Mastuo Basyo. (1702). Oku no hosomichi [Narrow Road to the Interior]. (https://kotenseki.nijl.ac.jp/ biblio/100000092/) Nitta, Y. (2018). Haiku-like aesthetic sentence composition through the narrative process. In T. Ogata & S. Asakawa (Eds.), Content generation through narrative communication and simulation (pp. 286–309). IGI-Global. doi:10.4018/978-1-5225-4775-4.ch009 Ono, J., & Ogata, T. (2018). Considering a haiku generation process using deep learning. Proceedings of the 28th Annual Conference of the Japanese Neural Network Society, 166-167. Ono, J., & Ogata, T. (2019). A method of haiku generation using deep learning for advertising generation. Proceedings of the 2019 International Conference on Artificial Life and Robotics, 578-580. 10.5954/ ICAROB.2019.OS24-1 Takahama, K. (2009). Haiku no tsukuriyou. Kadokawa Shoten. Wu, X., Klyen, M., Ito, K., & Chen, Z. (2017). Haiku Generation Using Deep Neural Networks. Proceedings of the Twenty-third Annual Meeting of the Association for Natural Language Processing, 1133-1136. Yoneda, K., Yokoyama, S., Yamashita, T., & Kawamura, H. (2018). Generated haiku distinctions by deep learning. IPSJ SIG Technical Report Vol. 2018-ICS-192 No. 8.

KEY TERMS AND DEFINITIONS Cyclic Story Generation: A concept for generation of story and haiku. Cyclic in this case implies a mechanism whereby mutual inputs can be inserted into a production system, and one input can output the other. Haiku: A poetic form common in Japan. It consists of three lines. The first line contains 5 moras, the second 7, and the third 5, for a total of 17 moras). Mora: A sound unit in haiku. In most cases, the authors do not use the term syllable. For example, the word book consists of a single syllable and two moras. Narrative Discourse: Narrative discourse is a concept presented by Genette (1972). This chapter deals with narrative discourse as a technique for transforming the structure of a story. This chapter considers haiku as a discourse technique. For example, it is possible to make a haiku by cutting a story from a particular viewpoint. Alternatively, a specific event is missing, etc. Umeji: A method of composing haiku. First, the first and third phrases are prepared. Then, the second phrase is inserted between them. The second phrase is based on the relationship with other two phrases. For example, the relationship is word familiar, dissimilar, and so on.

265

266

Chapter 8

On Logical Literary Work Generation and More Akinori Abe Chiba University, Japan

ABSTRACT In several papers, the author has discussed the possibility of the computational literary work generation. That was based on Julia Kristeva’s the concept of the intertextuality. This very simple application was achieved by random combination of the selected words/phrases/letters. It is a simple technique such as the cutup method. The author showed even such a simple method could generate literary works that can be regarded as not so bad literary works. However, for more intelligent generation, the author introduced abduction. By abduction, it is possible to guarantee the logical consistency and coherency. However, it does not care the very cognitive aspects such as beautiful, sentimental, and valuable, which will be very important factor in the literature. In this chapter, the author will discuss to add such cognitive aspects to the logical generation of literary works.

INTRODUCTION In several papers, the author has discussed the possibility of the computational literary work generation. The generative process was based on Julia Kristeva’s concept of the intertextuality (Kristeva, 1980). Kristeva defined intertextuality as follows: The word’s status is thus defined horizontally (the word in the text belongs to both writing subject and addressee) as well as vertically (the word in the text is oriented towards an anterior or synchronic literary corpus)... each word (text) is an intersection of words (texts) where at least one other word (text) can be read... any text is constructed as a mosaic of quotations; any text is the absorption and transformation of another. The important point is the phrase “each word (text) is an intersection of words (texts) where at least one other word (text) can be read... any text is constructed as a mosaic of quotations.’’ DOI: 10.4018/978-1-7998-4864-6.ch008

Copyright © 2021, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

 On Logical Literary Work Generation and More

This means that it is possible to generate new literary works by the reconstruction/combination (mosaic of quotations) of the frictions of the previously written literary works. In (Abe, 2005; Abe, 2910a; Abe, 2019), the author has discussed the possibility of the automatic waka (Japanese poem) generation. In the works, the main objective was to show the possibility of using the concept of intertextuality to generate literary works. Accordingly the author used a random combination of words from the widely known Japanese waka (Hyakunin-issyu (百人一首)). For instance, the following wakas could be generated (in Japanese1). 鳴く鹿の 尾上の桜 命にて ながながし夜を 山の奥にも A bleating deer, cherry blossoms in Ogami, it may be our fatal, in a long long night, deep in the mountain. 初霜の 衛士のたく火の 奥山に つらぬきとめぬ 紅葉なりけり The first frost, fire given by guardians, in the secluded mountain, cannot penetrate read leaves. The very simple application was achieved by random combination of the selected words/phrases/ letters. It is a simple technique such as the cutup method. Phrases are previously divided to 5, 7, 5, 7, 7. The phrases were extracted from the original wakas in Hyakunin-issyu. Accordingly, all phrases are the part of the original wakas. Then those phrases are selected randomly and combined according to the rule of waka2. The author showed even such a simple method could generate literary works which can be regarded as readable literary works. Actually computers cannot evaluate the quality of the generated waka. The quality of the generated waka can be determined by human. In addition, the random combination method cannot consider consistency coherency, nor meaning etc. during the generation as it is. Accordingly, for more intelligent generation, the author introduced abduction. By abduction, it is possible to guarantee logical consistency and coherency. However, the generative process does not consider the cognitive aspects such as beauty, sentimentality and value such as preciousness or preference, which are very important factors in literature. In this chapter I will discuss adding such cognitive aspects to the logical generation of literary works. The main purpose of this chapter is building a bridge between a computer and affective aspect of human. This is very significant trial and I will discuss from both computational aspects and authentic aspects. For that, the author explains abduction which can deal with a strict consistency and coherence maintenance. In addition, as a non-intelligent but exciting method, the author will introduce random generation such as cutup method. Then by using the result of previously conducted experiment, the author will show human’s abductive story generation. This may show a certain emotional aspects of human as well. In addition, for the emotional aspects, the author will address the importance of analogical mapping as a pastiche like method.

ABDUCTION In the previous section, the author mentioned literary work generation based on Julia Kristeva’s concept of the intertextuality. Although wakas are constructed as a mosaic of quotations, it was almost a random generation. Though readable literary works could be generated, it is not a logical and intelligent generation. For more logical and intelligent generation, one of strategies is an introduction of abduction. Abduction is one of the human thinking styles and regarded as a creative activity. In this section, the author will briefly show the mechanism of hypothetical reasoning (Theorist) which can be regarded as a computational abduction. 267

 On Logical Literary Work Generation and More

Computational Abduction Peirce illustrated abduction as follows (Peirce, 1955): ...abduction is an operation for adopting an explanatory hypothesis, which is subject to certain conditions, and that in pure abduction, there can never be justification for accepting the hypothesis other than through interrogation. The important keyword of the definition is explanatory hypothesis. That is, abduction is performed as an explanation of something. Peirce would not have assumed the performance of computational abduction. The first achievement of computational abduction was proposed by Pople (Pople, 1973). Pople explored the abductive reasoning process and developed a model for its mechanization, which consists of an embedding of deductive logic in an iterative hypothesis and test procedure. After several years, more logical abduction systems were proposed such as Theorist (Poole et al., 1987) and ALP (Abductive Logic Programming) (Kakas et al., 1992). Theorist is considered as a computational hypothetical reasoning system. A hypothetical reasoning is an explanatory reasoning mechanism that generates (collects, selects) a consistent hypothesis set from hypothesis candidates to explain the given observation. The generated hypothesis set can be regarded as an answer (solution) which can explain the observation. The inference mechanism of Theorist is described as follows: F O. (O cannot be explained only by F.) (1) F ∪ h ⊢ O. (O can be explained by F and h.) (2) F ∪ h □. (F and h are consistent.) (3)where F is called fact which is always true. On the other hand, h is called hypothesis which is not always true and included in the hypothesis base H (h ⊆ H). O is an observation to be explained. □ is an empty set. When F ∪ h ⊢ □, F and h are not consistent. In the knowledge base, inconsistent information is also included. Though details were shown in (Abe, 2019), the author will show a brief application of abduction to the generation of poems. If an observation is “satisfy hungry,’’ when the situation is ¬knife, bitter summer orange(road side), H will be a hypothesis set such as {pick up(bitter summer orange(road side)), peel(nail, bitter summer orange(road side)), eat(bitter summer orange(road side))}. Then the outline of the poem shown in the section 5.2 can be generated.

STORY GENERATION AS A RANDOM GENERATION In the previous section, the author introduced his previous works, for instance, an automatic waka generation. In this section, the author will illustrate a literary work generation by a type of random generation method. In the previous section, the author mentioned the possibility of the automatic (computational) waka (Japanese poem) generation. By using the concept of intertextuality, the generative system was able to generate wakas. These were acceptable in a certain sense. There was an almost random generation or combination of phrases. In the field of the modern literature, this type of generation was conducted as the cutup method. William S. Burroughs mentioned that ‘’[t]he method is simple. Here is one way to do it. Take a page. Like this page. Now cut down the middle and cross the middle. You have four sections: 1 268

 On Logical Literary Work Generation and More

2 3 4... one two three four. Now rearrange the sections placing section four with section one and section two with section three. And you have a new page. (Burroughs)’’ In addition, Burroughs stated “Sometimes it says much the same thing. Sometimes something quite differentócutting up political speeches is an interesting exercise in any case you will find that it says something and something quite definite.’’ For instance, the opening of the “ The Soft Machine’’ (1961) is as follows:

Dead on Arrival I was working the hole with the Sailor and we did not do bad fifteen cents on an average night boosting the afternoons and short timing the dawn we made out from the land of the free but I was running out of veins ... I went over to the counter for another cup of coffee ... in Joe’s Lunch Room drinking coffee with a napkin under the cup which is said to be the mark of someone who does a lot of sitting in cafeterias and lunchrooms ... waiting on the Man ... `What can we do?’ Nick said to me once in his dead junky whisper... It is a rather nonsensical sentence, but it can be acceptable as an English sentence. This “novel’’ has been lauded as having literary value after the cutup. David Bowie also said that “I use it [Burroughs’ cutup method] for igniting anything that may be in my imagination... You write down a paragraph or two describing several different subjects, creating a kind of `story ingredients’ list, I suppose, and then cut the sentences into four or five-word sections; mix ‘em up and reconnect them.’’ (SONGWRITING TIPS: Try David Bowie’s `cut-up’ method of writing lyrics, https://thehitformula.com/2015/09/21/songwriting-tips-try-david-bowies-cut-up-method-of-writing-lyrics-2/, September 21, 2015 (22, Nov., 2017 retrieved)) The important matter is that “for igniting anything that may be in my imagination.’’ That is, by the cutup, a potential expression or emotion might appear. In (Abe, 2005; Abe, 2010a; Abe, 2019), the author also cut the original waka into four-letters clusters (kanji and kana are included) and combined randomly. In this case, all the fragments are not always Japanese words and do not have a meaning. The generated wakas are: るらむ誰 に雲がく る人にせ に朽ちな るも別れ (Ruranu dare Ni kumogaku Ru hitonise Ni kuchina Rumo wakare) かりほの 白妙の衣 がむれば 暮るるも がれつつ (Karihono Shirotaeno koromo Gamureba Kururumo Garetsutsu) あらねど 龍田の川 あぢきな 猪名の笹 あだ波は (Aranedo Tatsutano kawa Ajikina Tyonano sasa Adanamiha) A strange division such as “るらむ誰’’ appears, but the last generated waka can be regarded as a waka composed by a (novice?) human. It happens to be composed of rather meaningful phrases. Actually the phrases in the waka are computationally divided. The original wakas were divided in four letters (including both kana and kanji) cluster, and combined with five clusters. Therefore, the generated waka is rather difficult to understand and seems like nonsense, compared with a waka composed of meaningful phrases. However, it may be obviously composed with very exciting “Japanese’’ phrases. By the cutup method, a potential expression or emotion might appear. The acceptability of the generated works can be evaluated by human. Currently it is rather difficult to build a computational evaluation system. To build it, it will be necessary to model the human acceptance of artworks. In addition, the combination without consciousness and intention can generate exciting works. This might be related to the “Descriptions automatiques,’’ which were performed in the age of Dada and Surrealism (Jenny & Trezise, 1989). 269

 On Logical Literary Work Generation and More

Although it will be difficult to extend the above method to the generation of general literary work, it is possible to generate literary work using the feature of intertextuality. In fact, it is a chance operation mechanism conducted by, for instance, John Cage and Karlheinz Stockhausen. It is addressed that “[t] he basic principle [of chance operation] is to remove one’s own intention from the work and hand that over to the oracle. Intention is always to some extent circumscribed by one’s own tastes and personality, whereas non-intention moves beyond like and dislike and becomes something more resembling an act of nature (https://www.biroco.com/yijing/cage.htm (retrieved 2 June, 2020)).’’ Thus by chance operation, very exciting works can be occasionally generated. Read pointed out that “[a]t the core of their work, OuLiPo considers language not as a veil of truth but as words, arranged within a logical and arbitrary structure of formal constraint. From this perspective, both traditional poetic forms and modern free verse can be seen as systems of composition based not in expressive honesty, but in the art of language. Writing poetry is a word game played according to certain rules (Read, 1988).’’ OuLiPo`s member proposed several argorithms for poem generation. Some are interesting, but rather experimental. Of course this type of generation can be easily conducted by a computer. However, this type of generation will not be intelligent. As an intelligent generation, I previously proposed an abductive generation (Abe, 2019). The author means “intelligent’’ as consistency and coherent. In the following sections, the author will briefly introduce the procedure and the method of the abductive generation. First, the author will show the abductive story generation during an art appreciation. Results from this experiment will show the process of the abductive story generation.

STORY GENERATION DURING THE ART APPRECIATION Previously Tadaki and Abe conducted several experiments ((Tadaki & Abe, 2017), (Tadaki & Abe, 2018), (Tadaki & Abe, 2019), (Tadaki & Abe, 2020) etc.) in order to know how we can appreciate art works. So that Tadaki and Abe controlled the quality and quantity of information, for instance, title, painter, history, and explanation of th