Computing Action: A Narratological Approach 9783110201796, 9783110176285

“Computing Action” takes a new approach to the phenomenon of narrated action in literary texts. It begins with a survey

181 89 2MB

German Pages 362 [363] Year 2003

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Frontmatter
Table of Contents
Part 1. Concepts of Action
Part 2. The Computer-Aided Analysis of Narrated Action
Part 3. An Experiment in Action: Conversations of German Refugees
Backmatter
Recommend Papers

Computing Action: A Narratological Approach
 9783110201796, 9783110176285

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

Jan Christoph Meister Computing Action

Narratologia Contributions to Narrative Theory/ Beiträge zur Erzähltheorie

Edited by/Herausgegeben von Fotis Jannidis, John Pier, Wolf Schmid

2

≥ Walter de Gruyter · Berlin · New York 2003

Jan Christoph Meister

Computing Action A Narratological Approach

Translated by Alastair Matthews With a Foreword by Marie-Laure Ryan

≥ Walter de Gruyter · Berlin · New York 2003

Als Habilitationsschrift auf Empfehlung des Fachbereichs Sprach-, Literatur- und Medienwissenschaft der Universität Hamburg gedruckt mit Unterstützung der Deutschen Forschungsgemeinschaft.

P Gedruckt auf säurefreiem Papier, E das die US-ANSI-Norm über Haltbarkeit erfüllt.

ISBN 3-11-017628-9 Bibliografische Information Der Deutschen Bibliothek Die Deutsche Bibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über http://dnb.ddb.de abrufbar.  Copyright 2003 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin Dieses Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Printed in Germany Einbandgestaltung: Christopher Schneider, Berlin Druck und buchbinderische Verarbeitung: Hubert & Co., Göttingen

Table of Contents List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vii vii

Foreword by Marie-Laure Ryan . . . . . . . . . . . . . . . . . . . . . . . . . . ix Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii Part 1 Concepts of Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Action as Project and Construct . . . . . . . . . . . . . . . . . . . . . 1.2 Action: From Word to Concept . . . . . . . . . . . . . . . . . . . . . The Etymology of the German Word Handlung . . . . . . . . Action as a Concept of Poetics . . . . . . . . . . . . . . . . . . . . . . Towards a Concept of Aesthetic Action . . . . . . . . . . . . . . . Action I: Singular Agential Activity . . . . . . . . . . . . . . . . . . Action II: Complex (Multi-agential) Sequence of Events . . Action III: Discursive Meta-Activity . . . . . . . . . . . . . . . . . . 1.3 Philosophical Definitions of Action . . . . . . . . . . . . . . . . . . The Commonsense Concept of Action . . . . . . . . . . . . . . . . Analytical Definitions of Activity and Action . . . . . . . . . . . Transcendental Definitions of Action . . . . . . . . . . . . . . . . . The Constructivist Concept of Action . . . . . . . . . . . . . . . . Hegel: The Philosophical Value of Aesthetic Action . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 The Elements of the Action Construct . . . . . . . . . . . . . . . . Lotman’s Definition of the Event . . . . . . . . . . . . . . . . . . . . Structuralist Approaches: Event, Transformation, Move . . The Event as a Modal Change of State . . . . . . . . . . . . . . . The Event as a Construct of Reception . . . . . . . . . . . . . . . Event, Property and Matter . . . . . . . . . . . . . . . . . . . . . . . . Conditions of an event . . . . . . . . . . . . . . . . . . . . . . . . . . . Class-Homogenous events . . . . . . . . . . . . . . . . . . . . . . . . . Class-Heterogenous events . . . . . . . . . . . . . . . . . . . . . . . . The Status of the Translation Rule . . . . . . . . . . . . . . . . . . . World Knowledge and the event . . . . . . . . . . . . . . . . . . . . Constructing events: An Example . . . . . . . . . . . . . . . . . . . matter and focus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 3 29 29 31 40 42 45 45 48 48 49 57 74 80 84 90 91 95 103 107 109 117 119 124 127 129 131 133

vi

Table of Contents

state of affairs and object . . . . . . . . . . . . . . . . . . . . . . Constructing an object event . . . . . . . . . . . . . . . . . . . . . . Constructing a discourse event . . . . . . . . . . . . . . . . . . . . The event Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Referential Nature of the event . . . . . . . . . . . . . . . . . Definitions of the Episode . . . . . . . . . . . . . . . . . . . . . . . . . . The Formal Definition of the Episode . . . . . . . . . . . . . . . . The Semiotic Square: A Reappraisal . . . . . . . . . . . . . . . . . . The episode: From Semiotic to Episodic Square . . . . . . . . The episode as a Three-dimensional Construct . . . . . . . . .

133 135 139 142 144 151 160 164 174 180

Part 2 The Computer-Aided Analysis of Narrated Action . . . . . . . . . . . . 2.1 Narratology and Literary Computing . . . . . . . . . . . . . . . . . Narratology and the Cognitive Sciences . . . . . . . . . . . . . . . The Limits of Practical Implementation . . . . . . . . . . . . . . . 2.2 The EventParser Program . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Constructing episodes and actions . . . . . . . . . . . . . . . . . The Syntagmatic Link . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Ontic Link . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Semantic Link . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The episode Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 The EpiTest Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Action Potential and action Product . . . . . . . . . . . . . . . . . 2.6 Practical Analysis Using EpiTest . . . . . . . . . . . . . . . . . . . . .

199 201 207 209 214 224 226 227 228 235 235 242 248 253

Part 3 An Experiment in Action: Conversations of German Refugees . . . . 3.1 Critical Approaches to the Conversations . . . . . . . . . . . . . The Compositional Logic of Goethe’s Conversations . . . . . Philosophical Discourses in Goethe’s Conversations . . . . . . 3.2 Describing the Conversations . . . . . . . . . . . . . . . . . . . . . . . The Isochronous Constructs . . . . . . . . . . . . . . . . . . . . . . . . The Isochronous and Anisochronous Constructs . . . . . . . . 3.3 Explaining the Conversations . . . . . . . . . . . . . . . . . . . . . . . Goethe and the Unfathomability of the Weltbegebenheiten The Logic of action in the Conversations . . . . . . . . . . . . . 3.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

257 259 264 271 275 280 284 287 288 300 303

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 Auhor Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329 Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333

List of Tables Table Table Table Table Table Table Table Table

1.4.1: 1.4.2: 1.4.3: 2.3.1: 3.2.1: 3.2.2: 3.2.3: 3.2.4:

Statements in an object event . . . . . . . . . . . . . . . . . Statements in a discourse event . . . . . . . . . . . . . . . The event Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . The episode Matrix . . . . . . . . . . . . . . . . . . . . . . . . . Analysis of Isochronous Constructs in EpiTest . . . . . Combined Analysis of Constructs in EpiTest . . . . . . . Summary of Isochronous Analysis in EpiTest . . . . . . Summary of Combined Analysis in EpiTest . . . . . . . .

138 140 142 235 278 279 280 285

List of Figures Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig.

1.4.1: 1.4.2: 1.4.3: 1.4.4: 1.4.5: 1.4.6: 1.4.7: 1.4.8: 1.4.9: 1.4.10: 1.4.11: 1.4.12: 1.4.13: 1.4.14: 1.4.15: 1.4.16: 1.4.17: 1.4.18:

Fig. Fig. Fig. Fig.

2.2.1: 2.2.2: 2.2.3: 2.2.4:

Structure of Narratives . . . . . . . . . . . . . . . . . . . Structure of Arguments . . . . . . . . . . . . . . . . . . . . . . . The logical square . . . . . . . . . . . . . . . . . . . . . . . . . . . Greimas’ semiotic square . . . . . . . . . . . . . . . . . . . . . . Order of operations in the semiotic square . . . . . . . . Tree diagram for the event construct . . . . . . . . . . . . event construct 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . event construct 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . Semiotic square for event construct 1 . . . . . . . . . . . . Semiotic square for event construct 2 . . . . . . . . . . . . event constructs 1–3 for Forster’s minimal story . . . event constructs 1 and 2, Goethe’s Fairy Tale . . . . . Explicit episode for event constructs 1 and 2 . . . . . Virtual episode for event constructs 1 and 2 . . . . . . Subtext of virtual episodes and events . . . . . . . . . . Redundant event construct . . . . . . . . . . . . . . . . . . . . Temporal axes of the episode construct . . . . . . . . . . Congruence of object and construct time in an episode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . EVENT PARSING window . . . . . . . . . . . . . . . . . . . . . . . . EVENT DEFINITION menu in operation . . . . . . . . . . . . . PREDICATE DEFINITION window (expositional predicate) EVENT PARSING window with predicate list . . . . . . . . .

159 159 175 177 179 182 184 185 185 186 187 189 189 190 191 192 194 196 216 217 218 219

viii Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig.

List of Figures

2.2.5: 2.3.1: 2.3.2: 2.4.1: 2.4.2: 2.4.3: 3.2.1: 3.2.2:

window (dispositional predicate) Logic of Second-Order Synonyms . . . . . . . . . . . . . . . Instantiation of Second-Order Synonyms . . . . . . . . . EpiTest flowchart . . . . . . . . . . . . . . . . . . . . . . . . . . . . EpiTest user interface . . . . . . . . . . . . . . . . . . . . . . . . . List of episodes generated by EpiTest . . . . . . . . . . . . Graph of Isochronous Analysis . . . . . . . . . . . . . . . . . Graph of Combined Analysis . . . . . . . . . . . . . . . . . .

PREDICATE DEFINITION

220 221 222 243 245 246 281 285

Foreword by Marie-Laure Ryan Propagated by countless works of science-fiction, a popular myth of the digital age predicts that the artificial intelligence of computers will soon exceed the natural intelligence of humans in all areas of mental activity. In recent years, the writings of cyber-gurus have given scientific credibility to the myth. In The Age of Spiritual Machines, for instance, Ray Kurzweil, a computer scientist with significant technological achievements to his credit, writes that by the year 2029, ‘Cybernetic artists in all the arts—musical, visual, literary, virtual experience and all others—[will] no longer need to associate themselves with humans or organizations that include humans. Many of the leading artists [will be] machines’ (Kurzweil 1999:223). The belief that computers are capable of mastering all forms of intelligence is so deeply ingrained that in 1993, the New York Times ran a story about a romance novel allegedly written by a computer in California without questioning its authenticity. The claim was subjected to a devastating critique by the distinguished Artificial Intelligence scholar Douglas Hofstadter which must have given a terminal case of writer’s block to the cybernetic author: we are still waiting for opus II. To dissipate the hopes of a take-over of the literary domain by computers in a foreseeable future (as well as the fears, for the idea of superhuman machine intelligence is both fascinating and disturbing), all one needs to do is look at the current state of automated story-generation, computerized text-processing and dialogue systems. Even though story-generating programs, from Jim Meehan’s Talespin (1981) to Selmer Bringsjord and Dave Ferrucci’s Brutus (2000) have steadily increased the sophistication of their algorithms, none of them is able to produce more than one type of narrative pattern (by this I mean action types such as mystery stories, fables, quest narratives, or tales of betrayal), and none of them has produced a narrative that people would truly want to read for its artistic merit. The cognitive processing of narrative texts by Artificial Intelligence algorithms is equally primitive. Though computers can extract information from individual sentences and simple texts by using syntactic parsers, a built-in lexicon, semantic networks, and a library of stereotyped scripts, they still lack the huge database of life-experience and cultural knowledge, as well as the ability to process ellipses, allusions, and

x

Foreword

figural language, through which humans interpret narrative texts logically, symbolically, and aesthetically. At the other end of the spectrum of opinions are the traditionalists, who believe that the literary imagination and the mechanical symbol manipulation of which computers are capable are incommensurable modes of thinking, and that computers have nothing to teach us about narrative. The cognitive deficiency of the computer is wittily satirized in Italo Calvino’s 1979 novel If On A Winter’s Night a Traveler. One of the characters, Lotaria, is writing a thesis on the work of a prolific Irish novelist, Silas Flannery. To save herself reading time, and to achieve scientific objectivity, she feeds the novels to a computer program, and the machine outputs a list of words in order of frequency. Of a text that produces a list dominated by the words ‘blood, cartridge belt, commander, do, have, immediately, it, life, seen, sentry, shots, spider, teeth, together’ Lotaria peremptorily declares: ‘There’s no question. It’s a war novel, all action, brisk writing, with a certain underlying violence’ (Calvino 1981:187). From another computer output, she knows, without having read the input novel, that ‘we are dealing with a full-blooded story, violent, everything concrete, a bit brusque, with a direct sensuality, no refinement, popular eroticism’ (Calvino 1981:188). In this double-pronged satire, Calvino pokes fun less at the crudeness of the computer’s analysis (a domain that left ample room for improvement when the novel was written) than at the self-delusion of the literary scholar, whose attempts to eliminate human subjectivity from the act of reading results in hasty and superficial judgments worthy of the most temperamental of literary critics. The moral of the story: all reading is interpretation, and you cannot replace interpretation with a computer. Yet this moral can be heeded without giving up on the computer as an instrument of narrative analysis. Interpretation may be largely determined by the individual experience of the reader and by particular cultural environments, but the cognitive manipulation of concepts and symbols also involves logical rules and universal processes which operate according to definable algorithms. If this assumption holds true, computer simulations should provide an invaluable means of testing hypotheses regarding the nature and mode of functioning of these processes. As Jan Christoph Meister writes, ‘most narrative theories are developed and tested on the basis of a single illustrative example which later provides an ideal breeding ground for sweeping generalizations, dogmatic theories, and vitriolic polemics’ (p. 201f.). The computer provides a potent antidote against this kind of theoretical laxity. It forgives neither imprecise definitions nor inflated claims, and it possesses the uncanny ability to confront us with situations that we had not anticipated. Since computer programs are designed to run on a variety of inputs, they offer a reliable test of the

Foreword

xi

generality of theoretical concepts. An idea based on a single example is likely to make the program crash, or to make it produce garbage, on any text other than the one that it originally described. And finally there is no better teacher than the computer when it comes to revealing the layer of implicit knowledge that underlies the interpretive process. Since the machine cannot rely on life experience, all the knowledge that it uses must be anticipated by the programmer and written into the software or into its database. But the benefit of directing our attention to what we would normally take for granted has also its drawbacks, because it means that the design of reasonably powerful story-generating or narrative-processing program would require a huge number of narratologists-programmers working as a team. We can only dream of what could be achieved, if entire companies of the size of Microsoft or Apple decided to devote their efforts to narrative computing! Until Bill Gates or some other industry giant decides to include narratology among his philanthropic projects, scholars interested in narrative computing will have to chose between two options: pretend that the economic means are available, and play the role of virtual project manager by outlining comprehensive algorithms without intent to implement them; or work from the bottom up, by limiting the ambition of the project to the solving of narrowly defined narrative issues, in the hope that these modest programs will eventually find their place as subroutines in the grandiose projects envisioned by the first breed of theoreticians. It is in this second tradition that Jan Christoph Meister has chosen to work. The purpose of the software presented in this book is to measure what I will call (though Meister does not use the term) the density of the narrative texture of a text. Traditional narratological approaches divide the text into events, episodes and action (or plot). In this terminology, an episode consists of several events, and an action consists of several episodes. It is tempting to assume that in a well-constructed narrative every event fits into an episode and every episode fits into a global action; but the textual reality offers a much more complex picture. Some narratives may integrate a high proportion of events into episodes and action, while others may represent many isolated events that fail to contribute to a global pattern. The former are goal oriented and economical, the latter digressive and contemptuous of narrative teleology. Gerald Prince has suggested the term “degree of narrativity” to account for these differences in narrative texture, but until Meister’s program, the only test for measuring “degree of narrativity” was the proportion of stative and active predicates contained in the text. In this line of reasoning, a story with many non-integrated events would achieve the same degree of narrativity as a story with a tightly woven network of semantic relations between events.

xii

Foreword

As this sketch of Meister’s project suggests, the basic unit of his program is the event. But the common narratological use of the term—which associates event with some happening, textually represented by an active predicate—is too loose to turn “event” into a working concept. A given happening, such as a tree being knocked down by the wind, can mediate between different states of affairs, and participate, in Meister’s definition, in different “events”: killing somebody, blocking a trail, or creating a bridge over a river. All these distinct events could in fact take place simultaneously. In Meister’s definition, “eventness” does not therefore reside in the nature of a particular happening, but in a relation between distinct states of affairs. An event takes place when a specific object is affected with distinct properties—antonymic predicates—at different, not necessarily adjacent points in time. Whereas the informal conception attributes a quasi objective, text-internal status to events, Meister’s definition presents them as an interpretive reader construct. A given passage can participate in many different events, and the predicate that describe the changes of state need not be present in the text. In contrast to the model that analyzes narrative as a rigid, unidirectional chain of state-event-state-event, this model presents events as networks of relations that criss-cross the text in many directions, and that potentially exceed the number of happenings. How can this conception of event form the basic building block of a computer program? As Meister observes, computers can only work with a precisely annotated text, this is to say, with a mark-up language; but marking up the text in terms of event constructs is an interpretive task that remains beyond the capabilities of present-day literary computing. The originality of Meister’s software resides in a bold move that makes virtue out of necessity. If the mechanical mode of operation of computers only enables them to detect what is objectively “in” the text, and if events are interpretive constructs rather than text-internal features, why not ask readers to create the mark-up themselves? The reader will analyze the text into events, and the computer will do the rest of the work, by trying to combine events into episodes (defined as two events entering into a semantic configuration inspired by Greimas’ semiotic square) and by trying to fit episodes into an action, which Meister conceives as a group of episodes ‘sequentially ordered in the semiotic continuum of the text’ (p. 250). (The particulars of this definition remain to be worked out.) The first module of Meister’s program, EventParser, prompts the reader to select a passage of text, to identify a focal object within these passages, to associate this object with a predicate, to select another passage where the same object is associated, in the mind of the reader, with an antonymic predicate, and to offer antonyms for both of the predicates involved in the event. Repeated for the entire text, this operation produces a list of events, syno-

Foreword

xiii

nyms and antonyms, that forms the input for the second module, EpiTest. This module analyzes the files, detects the semantic relations that allow the building of events into episodes and the integration of episodes into a sustained action; and calculates the degree of “event integration” and “episode integration” of the text. This division of labor between the human user and the computer is not merely a strategy born out of necessity, it also carries intriguing theoretical implications. The interpretation of narrative texts is presented as a collection of cognitive processes that operate on two levels. On the microlevel—the level on which users assign predicates to the state of affairs they judge narratively important—interpretation is highly dependent on individual factors, such as the reader’s knowledge and personal values. Leaving this level of interpretation to the human user accounts for its idiosyncratic character. On the macro-level, by contrast, interpretation is governed by semantic patterns, inference rules and combinatorial operations which present sufficient regularity to be simulated by the computer. If two different readers happen to produce similar interpretations on the level of events, Meister’s system predicts that they will produce equivalent constructs on the levels of episode and action. By the same reasoning, if you and I agree on the particular contents “all men are mortal” and “Socrates is a man,” we will agree, on the basis of the same general rules, that Socrates is mortal. And so will the computer, given the same input. Interpretive variety will however reappear on the meta-level, when readers evaluate the significance, or symbolic meaning, of “Socrates is mortal.” Let me conclude that for those who prefer ideas to quantitative analysis and speculative to empirical approaches, this book has much more to offer than the description of a computer program and the discussion of a test case, Goethe’s collection of stories Conversations of German Refugees. Meister’s tracing of the notions of action, episode and event through the history of philosophy, poetics and narratology takes us on a grand tour of Western thought that leads from Aristotle to Hannah Arendt and from Kant to Greimas. To those who believe that mere machines do not belong in such august company, Meister’s project replies that through its implacable logic, which forces its users to constantly revise their assumptions, algorithms and definitions, the computer fully deserves to be enrolled in the collective effort of philosophy and narratology to sharpen intuitive concepts into useful analytical tools. Indeed, what Meister’s book demonstrates is that the computer’s role in narratology—and for that matter in literary studies—may well be that of a meta-tool that helps us to refine older tools.

Acknowledgments This book originated in a research project which began to crystallize in my mind in the late 1980s. I would not have managed to sustain this interest over a period of more than ten years and transform it into a cohesive piece of research—part of which you have in front of you, part of which is available for download on the Internet—had it not been for the support of a number of individuals and institutions to whom I am deeply grateful. Among the latter, my particular thanks go to the University of the Witwatersrand, Johannesburg, the Anderson Capelli Foundation, and the Centre for Science Development of South Africa. These three institutions enabled me to conduct a pilot study during a sabbatical in 1993, and to meet with colleagues in North America and Europe who introduced me to the exciting field of humanities computing. From 1998 to 2000, the University of Hamburg and the Deutsche Forschungsgemeinschaft, Bonn (DFG) supported my work with a generously endowed habilitation grant. This stipend enabled me to conduct the research itself and write a habilitation thesis on the project, which was officially accepted by the University of Hamburg in September 2001. The DFG has also provided significant financial support towards the cost of publishing the book you are now holding. Institutional and material support is extremely important for Humanities research. Yet there is another type of backing even more fundamental to its true success: the intellectual and emotional companionship of fellow human beings. This is not the place to name those who are closest and most dear to me, of parents, partner, family, and friends who have helped me in my endeavours—often without knowing it, always without expecting any kind of return. Not least, through our shared experiences, they have helped me remember that the reality of human life is more important than any of the abstractions to which it is all too easy to succumb in the arts and sciences alike. More at home in an academic publication are my thanks to those who have assisted in the production of this book in a professional context. In this respect, my thanks go to my academic colleagues in North America, Europe, and South Africa, to whom I am indebted for support, criticism, and helpful suggestions during the research phase. A big ‘thank you!’ is also due to my former students at Wits University and Hamburg University: I have gained enormous benefit from the perceptive and enthusiastic curiosity

xvi

Acknowledgments

and interest that they brought to the seminars which dealt with various aspects of the theory and methodology of the project. Alastair Matthews I wish to thank for translating and editing the extensively revised version of the original German habilitation thesis in which my research eventually culminated and on which this book is based. Alastair deserves praise for doing decidedly more than ‘just’ translating the manuscript—which was itself a complicated enough task, considering the range of conceptual references and methodological paradigms covered in the study. My thanks go to Christine Putzo for helping prepare the final English manuscript—no mean feat, considering the amount of German quotes for which English translations had to be sourced and the anarchic charms of a bibliographical database which has, over the past decade, migrated between several different formats. Despite these acknowledgments, there is one respect in which I can claim sole responsibility for the book: any remaining factual mistakes, errors, and contradictions in the reasoning which underlies the study. This, I hope, is an unequivocal enough attribution of potential shortcomings to myself for me to conclude with a final word of thanks to my teacher. It is to Günter Dammann that I owe my interest in narrative theory and narratology, in structuralism and in Goethe’s Conversations of German Refugees. More importantly, I thank him for being a friend and scholar whose work I have always found exemplary, the product of a rare combination of intellectual and personal integrity that makes the humanities worthy of the name. Jan Christoph Meister

Knysna and Hamburg, 2003

Introduction The following introductory pages contain a brief outline of the three major themes of this book. They are, as the title suggests, the themes of computing, narratology, and action. After considering computing, we shall put the abstract before the concrete by commenting on narratology and action in that order. Although this contrasts with the word order in the title of our study, the book itself will, I hope, make up for the imbalance—a good third of it is dedicated to a philosophical criticism of the concept of action before we move on to a discussion of its application in a narratological and computational context. First, however, a few words should be said about the spirit in which the book has been translated from the original German version. At each stage, the overall aim was to make the text as accessible as possible to English-speaking readers, while always maintaining the necessary theoretical clarity and accuracy. This is particularly apparent in our treatment of quotations from authors and critics who write in a language other than English. Whenever possible, we have provided the equivalent passage from a reliable English translation. When, however, no translation has been published, or the existing translation was unobtainable or felt to be potentially misleading, we have provided our own translation. In those cases where the wording of the original German text is of paramount critical importance, it has been reproduced alongside the translation. This will allow readers who are familiar with German to compare the translation with the original and discover the nuances and allusions which escape what even the best translation can say. Computing The realities of the twenty-first century and the age of information technology might give one to believe that literary theory must soon, if it has not already done so, succumb entirely to the accuracy and objectivity supposedly promised by numerical data analysis. But such an analysis of the contemporary situation is a distorted exaggeration—although computational approaches to parsing and analysing texts are no longer the anathema they once were, there has not been a change in mentality such as would cause a full-scale paradigmatic shift in the critical community.

xviii

Introduction

It is more appropriate to speak of coexistent paradigms that complement one another. This mutual tolerance is reflected in the wider context of the developing framework of humanities computing, a new methodology which has its own distinctive theoretical and technological features but is nonetheless firmly anchored inside the humanities. Literary computing, the processing and analysis of literary texts by computational means, is one subsector of this new field. The methodological and epistemological status of humanities computing is still emerging as I write. Is it a discipline? A methodology? A method? Or no more than a tool?1 The same issues and the same questions apply to literary computing, or, as it is known in the German-speaking world, computational philology (Computerphilologie), the terminological counterpart of computational linguistics (Computerlinguistik). Even if literary computing has not yet been properly defined, however, it is clear that the pragmatic uses (and restrictions) of computer-aided textual analysis have been sufficiently well-documented to allay even the most traditional critic’s misgivings about the use of computers. Indeed, the employment of computers in the scholarly analysis of texts and languages is anything but new. For a long time, linguistics, corpus linguistics, and stylistics in particular have all employed computational methods to handle the characteristics and phenomena of written and spoken language which lend themselves to formal definition and description. These approaches include the analysis of metre and the calculation of morpheme and word frequencies. The computer-aided analysis of formally representable textual characteristics such as these began some fifty years ago, and its methods have increased in popularity and interest in the recent past as hardware resources have become more and more accessible to those outside the scientific community where they originated. The project presented in this book, however, has little in common with the above approaches apart from its technological foundations and the abstract principles with which it is implemented. My primary concern is a philological one in the traditional sense of the word. For our purposes, empirical textual features that can be described and analysed are of interest only in so far as they have a function in producing the synthetic meaning of a text. Meaning is a high-level concept and not to be confused with denotational reference; it defines the generalized anthropological and semiotic function of the text as a whole rather than the abstract linguistic or concrete pragmatic functions of the individual elements and sequences of elements in it. Meaning is the product of the complex and extensively 1

Among the most recent contributions are McCarty (2002), Meister (2002), Orlandi (2002), Rockwell (2002), and Unsworth (2002).

Introduction

xix

recursive interaction of empirical textual features on the one hand and idiosyncratically and historically predefined structures and knowledge on the other. When we talk and speculate about meaning, when we analyse and describe how it emerges in a textual representation, we must treat language in terms of dynamic, multidimensional processing models rather than unidirectional functions and two-dimensional structures. The question of how meaning is created when humans process texts— more specifically, when humans process narratives (texts which contain temporally indexed representations of events)—is not the exclusive concern of hermeneutically orientated disciplines. In our case, for example, the approaches of artificial intelligence are obviously of particular interest. When, however, the literary theorist ventures into this latter field, it becomes painfully obvious that there is a fundamental problem inherent in employing computer-aided analysis to investigate symbolic systems. If we are to preserve the paramount importance of the concept of meaning in criticism, we must recognize that literary computing will, by definition, be required to reconcile two conflicting epistemologies: that of the quantitative, numerical paradigm and that of the qualitative, hermeneutic paradigm. In so far as it employs computational methods, literary computing draws on the quantitative paradigm, where data is processed by explicitly defined algorithms which transform input data into uncontradictory, unambiguous output data. In general, the algorithms themselves are semantically neutral and not influenced by the data they process. Nothing could be less true of the hermeneutic paradigm, in which it is equally important that the processor is context-sensitive and self-aware. In the hermeneutic paradigm, we have to ensure that it is possible to access dynamically redefined frames of reference. Perhaps more importantly, the whole purpose of processing a text in this paradigm is precisely not to transform input into output in an unambiguous, predictable, and experimentally verifiable manner. It is rather to produce a unique and completely new interpretation that has never occurred before. From a theoretical perspective, a piece of information is meaningful in the emphatic sense of the word if it is a singular phenomenon which is more than just a mechanically reproducible result, more than the result of a process that simply encodes or decodes signifiers. However, the divide between the two paradigms can, in theory, be bridged. To a certain extent, computational algorithms can be programmed to consider the data they transform from input into output and the way in which they do so. This makes them self-aware and context-sensitive and thus able to emulate human intelligence. Furthermore, if we design them to perform recursive combinatorial transformational operations rather than unidirectional ones, we enable them to emulate the capability of the human mind to explore the potential meanings of textual data. The result is that

xx

Introduction

the use of computers in literary criticism can enable us to overcome two of the most fundamental shortcomings of traditional theories: first, the inability to base individual interpretations of a text on properly consistent empirical descriptions of the phenomena in that text; second, the failure to analyse sufficiently large corpora in the methodologically consistent way that is necessary if generalizations about works, epochs, or genres are to be based on inductive inferences rather than impressionistic observations and normative declarations. Narratology It is at this point that narratology comes into play. As the reader will recall, the accurate description of the formal features of literary texts was the goal of the programme of the Russian Formalists and the structuralist narratologists who later adopted and refined their theories and concepts. The high-level semantic and aesthetic functions of literary texts, it was thought, could eventually be traced back to formal features of the symbolic material in the text and the structures in which it appeared. The narratological theories and concepts that preceded the post-structuralist movement are steeped in formalist methodology and therefore particularly suited to being modelled, explored, and applied using computational means. In the context of this observation, ‘computational’ need not necessarily be understood in the technical sense of a concrete computing device. Granted, the computational modelling of a narratological concept should eventually mean implementing the model in a concrete piece of software. More important, however, is the reformulation of the concept itself as something that can be computed in the first place. Textual phenomena of narratological interest are rarely located below the level of sentences (or rather propositions). Because of this, they will always be phenomena which can occur only when the sentences (or propositions) are cognitively processed in the human mind. This, essentially, is why scholars like David Herman argue that narratology should be fundamentally redefined as a subdiscipline of cognitive theory. After all, the processing of textual information, particularly the temporally indexed events represented in narrative texts, is arguably the most complex approximation we have of how humans process information in the real world in real time.2 2

David Herman’s comprehensive study Story Logic. Problems and Possibilities of Narrative (Herman 2002) was published after the manuscript of this book was completed. It is not, therefore, considered in any detail in the following pages. From my perspective, there are considerable resemblances between our approaches. The major methodological difference would appear to lie in the fact that I, unlike Herman, am somewhat sceptical

Introduction

xxi

The reorientation of narratological theory and methodology as part of cognitive theory is bound to have significant consequences. Human cognitive processing never takes place in a context-free environment. If narratologists are to take this insight seriously, they will have to rethink some of the most basic concepts—such as function, event, and action—which they have inherited from the formalists. And so, we suddenly find the narratologist and the computing humanist in the same boat. This is why there is no need to go overboard and suddenly redefine everything in terms of scripts and frames alone and reject the formalist paradigm altogether. In cognitive processing, scripts, frames and knowledge interact with logical universals; the same applies to the processing of narrative representations. If this were not the case, it would be a complete mystery as to how we can arrive at even remotely identical interpretations of identical narrative data. The approach of literary computing towards a specifically narratological phenomenon must, therefore, be capable of relating the abstract logical model of the phenomenon to a process model of how instances of the phenomenon are read. What’s in it for narratology? Recall that even the most radically structuralist approaches to narratology have, so far at least, failed to supply studies in which a well-defined methodological and taxonomical system has been consistently applied to an adequately large corpus of narrative texts. Propp’s morphological and functional analysis of one hundred Russian folk-tales seems to have remained a singular example. As this book will attempt to show, computational technology may allow us to overcome the impasse. Although the practical application of the theory described in this study will be restricted to six rather short narratives, its empirical methodology can easily be applied to much larger corpora without further modification. Two software tools—EventParser, a markup tool, and EpiTest, an analytical tool—were developed to accompany this book and are a crucial part of our project. Both applications, as well as extensive documentation and record files, are available for download.3 However, the above utilitarian argument does not do full justice to the way in which literary computing approaches narratological problems. The true relevance of literary computing is methodological rather than empirical in nature. One of the biggest problems for scholars in the humanities is establishing the validity of their hypotheses—contrast the situation in the natural sciences, where we can perform experiments to test whether

3

as to whether concepts of cognitive theory (e.g. script and frame) can be directly applied in the context of narratology. Also, by studying the story, Herman is examining a highlevel concept, whereas the scope of my study is restricted to basic readerly constructs. See .

xxii

Introduction

a hypothesis is valid or not. Of course the matter becomes appreciably more difficult in both contexts when we start dealing with abstract and hypothetical entities, and even worse when we consider the fundamental relativity of observation that is an inescapable part of all perception and cognition. On the other hand, Occam’s razor and the appeal to logical consistency are universal criteria that can, irrespective of discipline, be applied to almost any theoretical construct. But we should not be under any illusions—more often than not, it proves extremely difficult to apply these criteria to the soft and impressionistic theorems that abound in the humanities in general and literary studies in particular and which hardly ever satisfy the Popperian test of falsifiability. Against this background, the decisive methodological role of the computer proves to be that of a ‘meta-instrument’ (Orlandi 2002:53) rather than a tool. In other words, the computer helps us to simulate an objective test-bed for our ideas about how narrative processing works.4 Algorithms are unforgiving not only when it comes to dealing with obvious contradictions but also when it comes to processing ill- or under-defined input. It may require considerable effort to express narratological concepts in the form of program code that can actually run on a machine, but the investment of time and energy is rewarded each and every time the program malfunctions, crashes, or descends into an infinite loop, for, by doing so, it highlights a flaw in the reasoning behind our theory as it stands. Action As the title of the book suggests, ‘action’ is the central narratological concept which is investigated in this study. Since Aristotle’s Poetics, scholars have discussed the question of what narrated action is, what its elements are, and how we manage to read and narrate it. The first part of our study will discuss these questions from the perspectives of philosophy and narrative theory before presenting a possible new theoretical definition of action. In the second part, I shall attempt to reformulate the preliminary definition in the context of narratology and apply it in the context of literary computing. In the third and final part of the book, I shall describe the demonstration analysis of a concrete literary text. I have chosen Goethe’s Conversations of German Refugees of 1795 as my example text for two reasons. First, this cycle of novellas has generally been considered interesting precisely 4

We must, of course, remember that this test-bed is not really objective and is simply a construct which is employed for methodological purposes.

Introduction

xxiii

because it is very difficult to establish the coherence of its overall action and the actions contained in its component novellas. Second, I intend to show, against the background of a detailed analysis that will test the concepts and tools developed in the preceding parts of the book, that the Conversations do not only contain, or represent action and actions but are in fact a discourse on the problem of how we narrate and read action itself. In order to show this, I shall interpret the data generated by the computational analysis of the text in the light of Goethe’s ideas on the epistemological difference between how we perceive natural phenomena on the one hand and social and aesthetic phenomena on the other.

Part 1 Concepts of Action Poetry depicts movements and, implicitly by way of movements, bodies. Laokoon (Lessing 1973, 5:59)1

1

‘Die Poesie schildert Bewegungen, und andeutungs Weise durch Bewegungen, Körper.’

1.1 Action as Project and Construct Action is both a key concept of literary poetics and aesthetics and one of the most prominent concepts discussed in post-Enlightenment philosophy. Thus, in his introduction to the Naturrecht, Fichte makes the following categorical statement: ‘The I itself is nothing other than acting that takes place upon itself’ (Fichte 2000:3; ’Das Ich selbst ist nichts anderes als ein Handeln auf sich selbst’, Fichte 1979:1).2 The metaphorical use of the German preposition auf (whose literal meaning is ‘on’ or ‘upon’) may strike us as odd but is highly appropriate in that it at once evokes the image of the I as a construct which is built on a material foundation. The rather obscure meaning that Fichte meant to communicate with the image only becomes apparent if we rephrase the sentence as follows: ‘The I is nothing but a product of acting that is about itself.’ In the sense of self-conscious and self-referential doing rather than objectivized behaviour, acting—or ‘activity’, as we shall call it—thus becomes the key criterion for the existence of an I. What type of doing, we must then ask, satisfies this criterion? Does it have to deal with external objects, or are internal acts of thinking and imagining in themselves sufficient to make an I emerge? And if the latter is indeed the case, can we then draw the conclusion that without acting there is no being, ago, ergo sum? Questions such as these point to the important role which action and the ability to act play in the self-definition of human beings. In the context of this study, however, we shall focus on the relevance of the concepts in fictional rather than factual contexts. Even in the latter case, though, it still seems plausible to apply Fichte’s observation. For example, most readers will only make the effort to create mental images of a fictional character if and when that character is somehow seen to be involved in, or at least affected by, the narrated action. In more general terms, the aesthetic relevance of a literary narrative is to a large degree as contingent on the action represented in it as it is on the manner in which the representation takes place. We might therefore state that, like the I, the narrative is only in so far as it is on, upon, and about action. On the other hand, if we state that narrative texts are based on action we seem to be resorting to nothing more than a figure of speech. Any act 2

As early as the Grimms’ Wörterbuch, this dictum is quoted under the headword Handlung (‘action’).

4

Part 1 Concepts of Action

of narration is of course the product of empirical speech acts. And since narratives can deal with almost anything, concrete as well as abstract, why not also with the very concept of action itself? The fact that action need not just provide the content of a narrative but can also become its actual subject matter does not automatically mean that the narrative act in itself is by necessity a self-reflexive action in the philosophical sense. However, for our purposes, Fichte’s aphorism can be interpreted not only in terms of speech act theory, but also in the sense that it points to a certain logic in the constitutive process which brings action into being. Note that in this interpretation we refer to the narrative text as something that acts3 on action rather than on activity. Action and activity are two entirely separate concepts. The word ‘activity’ refers to an ongoing process, whereas the word ‘action’ denotes the result of such a process. Our commonsense understanding of action is based on the intuitive knowledge that, with hindsight, individual instances of activity can generally be subsumed under the abstract notion of a complex action. Conversely, we intuitively believe that there can be no action which has not been preceded by at least one instance of activity. This ex negativo definition of action in terms of its dependency on prior instances of activity seems only natural in that it reflects the logic which governs the real world as we experience it. But that is not necessarily how the world objectively is, for activity and action are in themselves abstractions, not names for empirical entities. In other words, they do not refer directly to objects or even states of affairs in the empirical world, but rather to the particular way in which we as human beings organize our perception of successive states affairs and, in the present case, our perception of the aesthetic representation of successive states of affairs. The qualities which the two terms imply may seem evident in the context of language and the imagination, but they nonetheless lack the kind of absolute evidence, empirical or ideal, that exists independently of observation and description. It is for this reason that the following attempt to redefine the concepts of action and activity does not presuppose any particular theoretical definition of action and instead takes a phenomenological and analytical survey as its starting point. This survey aims to identify the distinctive features of the phenomena of perception to which we refer using terms such as ‘activity’ and ‘action’, and, in particular, the relevance of these phenomena to the cognitive processing of literary narratives. We shall approach our definition of the concept of literary action by starting from the periphery and moving inwards:

3

I use the verb ‘to act’ in the sense of a doing concerned with something, not in the sense of representing or role-playing.

1.1 Action as Product and Construct

5

we begin with an analysis from a commonsense, phenomenological, and analytical perspective (chapter 1.1), followed by an etymological (1.2) and then philosophical discussion (1.3). Finally, we arrive at a definition of the concepts of event and episode as fundamental constituent elements of all literary action (1.4). What, the reader may ask, is the point of going to all this trouble? What is wrong with deriving a theoretical definition of the aesthetic concept from our commonsense understanding of action as the result of an intentional doing attributed to an agent? I believe that, in the context of literary and fictional narratives, the supposedly self-evident intentional definition of action brings us no further than the dead-end of a mimetic theory which conceptualizes action as one of many (abstract or concrete) referential objects that are symbolically represented in a text. In other words, such a definition inadvertently portrays action as something that pre-exists outside the narrative text and is either depicted in it in the form of fiction, or made empirically present in the form of an aesthetic speech act at the level of discourse. These, however, are essentialist premises which are irreconcilable with our narratological interest in formulating a definition of literary action which takes into account theories which are based on constructivism and reception theory and model how readers process narratives. I intend to explore the concept of action neither in purely intentionalistic terms, nor idealistically as an autonomous aesthetic abstraction, but ultimately as a concept which is self-explanatory in the most literal sense of the word, as a concept, that is, which is closely related to the concepts of both the real and the fictional self. Who acts in literary activity, in literary action? On what ontological levels are such actors to be found? What conditions of possibility are shared by activity as a process and action as the product of a sequence of such processes? And what are the consequences when these processes become self-referential, when activity takes place on action and makes transparent the very mechanisms by which action is perceived and constructed step by step in the reader’s mind? It is in this respect that Fichte’s dictum will be relevant to our exploration of the concept of literary action, which we shall investigate from the perspective of a narratological interest that focuses on the problem of how to conceptualize literary action against the background of theories such as reception theory, constructivism, and Greimassian semiotics. Put in less abstract terms, our overall interest is twofold. First, we are concerned with the question of how and under what narrative constraints and conditions a recipient constructs a mental image of a certain type of event complex that fulfils the defining criteria of a literary action. Second, we shall explore the extent to which the process of construction is itself a real action whose ultimate objective is to interpret the event complex as a meaningful structure.

6

Part 1 Concepts of Action

The question of the subject of literary activity and action provides an approach to our problem which does not entail any undue objectivization of the matter under investigation. One could, for example, adopt a radically positivist stance and assert that the medium itself, in most cases a text or a book, is by definition the only material entity that is a necessary prerequisite of literature about or containing an action. Yet a text or a book as such would only qualify as an actor in the metaphorical sense expressed by the German sentence Das Buch handelt von X ‘The book is about X, deals with X’, where the use of the third person singular of the verb handeln ‘to act’ results in the personification of the inanimate Buch as the subject, the acting entity. In a highly condensed form, this personification encapsulates the complexity of the temporally differentiated empirical actions of coding and decoding with which we handle and experience a text as a means of symbolic representation. Thus, neither did the text that you happen to be reading at this moment write itself, nor will it transform itself into concepts and mental images of its own accord. I, the author, am writing word by word, sentence by sentence, in a spatio-temporal position which is logically and chronologically anterior to that of you, the reader. On the other hand, you are decoding a text which does not even exist for me at my point in time; or rather, it exists only as a project which is yet to be completed. Moreover, for you I am (hopefully) a real person, whereas I can at best speculate on the hermeneutic, critical, and experiential conditions which might govern your reading process. I can therefore enjoy the privilege of having the first word, which, as a consequence of the temporal and logical distance between the acts of production and reception, is counterbalanced by the normative final say, the source of the reader’s ultimate power. And so it is neither the book nor the text that acts; it is rather you and I that interact in, with, and about it by exchanging words, sentences, arguments, reservations, and mental images across the logical and chronological hiatus without which the semiotic system cannot function. That, however, is about as far as one can legitimately stretch the analogy between action as found in a book and action as attributed to its author and reader. A concept of literary action rooted primarily in discourse theory would be as inadequate as an approach which focuses on the best way of decoding action-related elements in a narrative (who or what acts or becomes the patient of activity inside the fictional world?). Such approaches are inadequate because the object domain of literary theory consists of non-expository texts, fictional narratives. Reference is just one semantic aspect of this particular class of texts, in which it plays no more than a minor role. Instead, the primary function of this class of texts is to evoke a fictional domain which is structured in such a way that we can

1.1 Action as Product and Construct

7

engage with it interpretively and construct a meaning rather than decode an extratextual denotation.4 The source of the aesthetic and, ultimately, pragmatic (educational and normative) relevance of a fictional narrative is not its elementary, quasi-denotational meaning but its systematic meaning. Seen in this light, the issue at hand immediately presents itself as one of much greater complexity than was initially apparent, for we now have to explain and define the concept of literary action from the perspective of the role it plays in supporting the non-referential semantic function that is the hallmark of literary narratives. Writing and reading, that is to say, the production and reception of texts, are two of the many activities which humans can perform. Given that these two activities are concerned with texts which themselves narrate fictional activities, we can refer to them as meta-activities. Once they have reached completion, such meta-activities acquire a holistic quality and are transformed into instances of completed action. This transformation parallels the synthesis which takes place when we consider a narrative’s fictional events and acts in retrospect and convert the fictional activity represented inside the text into an integrated functional action complex. This latter type of action can thus be referred to as an object-activity (the activity in the text) which provides the basis for the meta-activity (the reading of the text).5 As the concept of action is yet to be properly formulated and will only be defined formally at a later stage, we must, for the time being, adopt a pragmatic convention whereby both processes, meta-activity and object-activity, are referred to as actions when they have been permanently or temporarily completed. It should be remembered that, as a result of this convention, the precise meaning of the term ‘action’ will vary depending on the context in which it is used. Although this approach means that completed processes of writing and reading are subsumed under a single concept of meta-action, the two remain clearly distinguished at a phenomenological level, for it is only the former, productive activity that leaves its mark on the text, the common object

4

5

I understand the contrast between meaning and denotation in terms of Frege’s distinction between ‘Sinn’ and ‘Bedeutung’ as interpreted by Gabriel (1975), who applies it to literary semantics and defines ‘Bedeutung’ as ‘fictional reference, denotation’ and ‘Sinn’ as ‘meaning’ in the sense of the ethical, moral, and aesthetic relevance of a work of art. The latter notion is non-referential and abstract but nonetheless real in that it can be experienced intersubjectively. The concept of object-activity refers both to the activities of fictional agents and to events which lack an agent and are portrayed simply as contingent transformations in the narrated world. In current theories of interactive digital storytelling, the distinction that we have made here is reconceptualized as that between content action and user action (Paul and Fiebich 2003).

8

Part 1 Concepts of Action

of the two processes. This mark is the presence of a narratorial voice.6 A further feature of the literary narrative text is that it may itself thematize the difference between the roles played by its producers (author and narrator) and its recipients (readers and critics). This feature sets literary narrative texts apart from most (but not necessarily all) dramatic texts. Literary texts can employ techniques unique to narration—changes in time, mode, and perspective—in order to highlight the ontological divide which separates reading, narrating, and what is narrated. In addition, the textual product of the narrative process draws attention to the epistemological problem of how to distinguish instances of meta-activities inside and outside the text from object-activities inside its world. The more conspicuous the narrative authority becomes, the more aware the reader becomes of his dependency on it: he realizes that any and all knowledge that he has about the fictional world has been acquired indirectly. Our acts of reception and cognition are, therefore, anything but autonomous acts. They are always based on foreign epistemic processes grounded in an antecedent narrative act which they nonetheless logically transcend in its entirety. The same Plato who saw the writers of fiction as liars and wanted them banned from the utopian state as a matter of principle also appreciated the fact that the epic narrator acknowledges, if only at a formal level, the contingent and construed nature of the mental images he evokes. We can, after all, identify the person of authority who is responsible for reporting what is narrated in any given narrative text. The concept of this person, however, both points to the possibility of deception (the persona is, as the Latin word indicates, no more than a mask of identity) and indicates that any information narrated about a fictional event is passed on to us by the mediating authority in the medium of sound (per son); that is to say, the information is related symbolically. To complicate matters even further, the authority from whom narrative discourse originates can relate the same object—a fictional event—from two logically exclusive perspectives (in practice, they are typically combined to some degree rather than being absolute opposites). By virtue of the syntagmatic principles of language, the narrator invents himself as an agent or witness who, whether directly involved in the affairs reported or simply perceiving them from outside, can never know more than what has already happened. Irrespective of what a narrator may factually say, know, or suggest, and irrespective of the authorial power asserted, the language of a text always creates the 6

See Kindt and Müller (1999, 2002) for a detailed discussion of the concept of the implied author and the methodological conditions under which it can be applied as a metaphorical equivalent of textual intention, and thus as a concept that belongs to the domain of interpretive theory rather than text-based analytical narratology.

1.1 Action as Product and Construct

9

illusion that report and event are formally coincident. As a speaker, the narrator acts step by step, sentence by sentence, and, in this respect, he seems to be subject to the same laws of sequentiality as those that govern the natural human perception of real events. Yet, as a completed metaactivity, narration must chronologically succeed the events about which it narrates. As a completed action, narration thus reveals a logical and constructive knowledge about an entire complex of events which were not only preconceived but also preinterpreted. The narrator must have had access to this knowledge in the form of a complex whole rather than as a succession of individual events. This applies, moreover, even in those cases where the narrator explicitly claims that he does not know any more than the characters in his narrative. The narrator has a paradoxical double role because of his formal position both inside and outside the sequence of narrated events. The situation is mirrored in the role of the reader. We immerse ourselves in narrated events and start imagining them as an open-ended process in order to experience suspense. This behaviour presupposes the suspension of our knowledge that the order of events is preordained and their outcome already established.7 The various mental images which the text evokes appear in our consciousness in a sequential order which is ultimately supported by nothing more than the symbolic material of the language with which they are represented. Nonetheless, the reader embraces the illusion that the various denoted events unfold dynamically as part of an open-ended process in the fictional world where they take place. We consciously enjoy this illusion, despite being constantly reminded that it is no more than that, an illusion. The reminders take two forms: the recurring signals which indicate the presence of a narrator, and the formally indisputable empirical fact of our position, which lies ontologically outside and logically above the text. Thus, even the most naive of readers will recognize, sooner or later, that we experience enjoyment by temporarily treating the apparently open-ended activities and events in the narrative text as facts, and that, in doing so, we approach an aesthetic representation as if it were a real one.8 We attain this insight as soon as our attitude to the narrative changes and we turn our interest from the constituents and fictional development of the imagined reality in the narrative to the sense and meaning of what is narrated. This, however, is precisely the moment at which we cease dealing with activities and events and become involved with action instead. In contrast to fictional activities, 7

8

See the discussion in Ricœur (1984:59f.) on the connection between the dechronologization of narrative structures and the revelation of their logic. The phrase ‘as if’ stems from Vaihinger (1927), who argued that the employment of fictional entities is an epistemologically valid exploratory mechanism and of fundamental importance to both the arts and the sciences.

10

Part 1 Concepts of Action

which are bound to the mode of referential signification, the category of action is not all that distant from meaning, the defining category that sets literary narratives apart from expository texts.9 Taking the self-referential nature of the concept of literary action one step further, we can also apply it to the meta-activity of scholarly discourse about literature, such as that in which we ourselves are currently involved. The exercise illustrates yet another aspect of the logic of action. The very first sentence of this chapter is symptomatic of the unique way in which action comes into being qua activity. Our first sentence is the product of a completed speech act and thus an action in which a particular activity has solidified. At the same time, it indicates that at least one possible alternative activity, if not an infinite number of them, has not been realized. For example, from a rhetorical perspective I might have been well-advised to begin with a proper captatio instead of in medias res. With the very first sentence, I have wasted this chance once and for all and unwittingly added a first entry to the countless list of actions that will be excluded from this text. Moreover, the list will increase in length with every new sentence I write. Like all action, the discursive activity by means of which an argument is developed highlights the consequences of a choice between alternative possibilities. It implies accepting one thing and declining another, yet it can never encompass all the imaginable possible activities in which we could take part, nor all the knowledge we could possibly have, nor all the utterances we could possibly make. No matter where and how we choose to act, the virtual and the real will always be inseparably intertwined in the processes in which we act. Like many fundamental dependencies, this is not normally one that we have to consciously remember in the pragmatic context of everyday life, where the choices we have declined only become interesting when we realize that the ones we have accepted have led us into conflict or been the cause of failure. In classical times, Aristotle realized that one of the most notable and cognitively and didactically productive functions of the epic or dramatic narrative is its ability to make us speculate on what was not done, what 9

According to Grabes (1993) the different attitudes to narrative sequentiality, one illusionary and the other critical, point to two fundamental and complementary aspects of the reading process and perception in general. In terms of the psychology of perception, textual comprehension and communication are fundamentally dependent on the sequential formation of meaning before we engage in the synthesizing operations that enable us to store processed information in memory. However, this synthesizing in turn presupposes that we suspend temporal order and, therefore, the principle of sequentiality. Although Grabes is referring primarily to the non-critical reading of text, he also mentions one of the methodological consequences which apply at the meta-level: the predominance of spatial over temporal metaphors in the various narratological taxonomies. On the latter point, see also Ryan (1992).

1.1 Action as Product and Construct

11

might have been, by means of artificially constructed conflict situations. However, Aristotle paid little attention to the fact that a narrated character cannot legitimately be held responsible for the choices he makes. The acts of the characters in fictional activities are analogous to the acts of real people in empirical activities in that the choices in which they originate can only be investigated ex post facto, that is, once they have been performed and transformed into completed actions. In the context of the everyday empirical world, we attribute such choices to actors who are an immanent part of the world in which they, and we, live. In contrast, the constitutive acts of choice which result in a character’s initiation of or failure to initiate an activity are completely external to the fictional world. They take place outside of it even though they apparently result in a seamless integration of the three ontological domains of diegesis, narration, and reception. A choice may be manifested as a fictional event, but the process of choosing as such transcends it. Whatever happens in the fictional world, it is ultimately the result of choices made in the real one during the actions of production and reception. This last point is particularly important for the methodology of our investigation. I believe that a truly narratological theory of action should be able to model fictional action as a narrative project and, simultaneously, as a receptive construct. In particular, both these aspects must be considered if a narratological theory of the episode is to be formulated. By this I mean a theory that deals not only with the complexity of the phenomenon as a project and construct but also with the importance of the virtual domain, that is, the domain of activities and actions which consists of positive choices and their unrealized alternatives. The phenomenon of literary action can no longer be described from a purely positivist perspective but should be modelled as an object best approached by more differentiated means. The virtual domain is, however, also the domain of vast numbers: it is comprised of an enormous set of theoretically conceivable variants of activities and actions, the size of which is practically impossible to grasp. The investigation of this domain calls for tools which allow us to analyse a structure whose complexity we cannot even begin to understand using traditional hermeneutic methodology (i.e. the analysis and exemplary interpretation of specific texts). The transition from formalist narratology to literary computing presented in the second and third parts of the present study is an attempt to bridge the gap between qualitative (hermeneutic) and quantitative (numerical) analysis which we have outlined above. A key idea to bear in mind during the following pages is the idea that a tension exists between the possible virtual instances of an action and the action that we actually perceive, be it an aesthetic project or a receptive construct. This idea lies behind each stage of our discussion, from

12

Part 1 Concepts of Action

the definition of our basic concepts (state, sequence of states of affairs, event, episode, and, last but not least, action itself) to the development of a theory of literary action and the practical application of the concepts and the theory behind them. I believe that this tension constitutes perhaps the most fundamental necessary condition of fictional action, fictional events, and the epic representation of fictional happenings. As a logically necessary condition of action, the tension is usually manifested indirectly, most often in the form of a fictional psychological disposition which controls the motives, intentions, and aims of the fictional characters. Readers have the impression that the events which take place in a fictional world are, to a greater or lesser degree, open-ended or defined by agents’ intentions. The impression is not created by the fictional happenings as such, but is rather the result of an ontology constructed by the narrator and his recipients. Seen from a logical point of view, therefore, the course taken by the activities and happenings inside the fictional world is not just a hypothetical reminder that everything could have turned out differently. With every development that takes place, it confirms the fact that the virtual world is the necessary prerequisite of the actual world and will always retain its power. The meaningfulness of the fictional world is secure only when we are able to conceptualize its constituent events and the order of the elements contained in it as the function of a narrative or receptive choice that is fully flexible. The crucial shortcoming in reconstructing the fictional world in a psychological or anthropomorphic manner is that such methods tend to obscure the essential fact that the world’s events and elements are being narrated, read, and therefore constructed. By necessity, this results in the separation of fiction from the significant act of choice and implies that diegesis is an autonomous, static, and thus in two ways deterministic quasi-reality devoid of choices. There are two effects of denying that the fictional world is constituted by choice and identifying the fundamental tension between the actual and the virtual on which it is based. First, this affects the narrated action abstractly by negating its meaningfulness; second, it affects the recipients by robbing them of the opportunity to contrast the fictional world with the modern secular world, which is by definition a contingent reality devoid of meaning and teleological choices. As result, it is impossible to turn to the hypothetically meaningful structure of a fictional world in order to explore new ways of creating meaning for our own one. There is nothing to be learnt from the action and happenings that take place in a fictional world that has been reduced to the state of a deterministic parallel reality governed universally by immediate necessity. The completed world, the completed action, and the necessary, final, and conclusive interpretation of both cannot lead to anything but the end of the fictional world as a work of art.

1.1 Action as Product and Construct

13

The aesthetic and cognitive power of the fictional narrative can only be experienced while the abstract complementary relationship between the real and the virtual is kept alive, whether consciously or unconsciously. All criticisms of mimetic realism from Plato to Brecht can be understood as warnings against the use of aesthetic methods which obscure the interdependency of the real and the virtual and thus hinder literature from performing its proper task of showing how the production of meaning and sense is based on narrative and receptive choices. It is tempting to go further and apply this Platonic criticism to the modern phenomenon of virtual realities, which present themselves as autonomous separate realities. But it is hardly worth the effort to do so, for the concept of virtual reality is terminologically contradictory because the opposition between the virtual and the real on which it depends does not actually exist. In logical terms, the virtual is as real as anything else and in no ways conflicts with the ontological conditions that govern the real world. The only predicate it lacks is that of existence, but this is not a serious deficiency, particularly when we remember that we are dealing with fictional worlds in the first place. Likewise, the positive assertion of a fictional world as a possible world does not result in the dichotomous juxtaposition of virtuality and reality.10 If we want to define the virtual in an ex negativo manner, we cannot do so by contrasting it with the actual or real using the possibility predicate, for the latter is in fact shared by both domains. The opposite of the virtual is rather the antimodel of the possible world, that is to say, a model whose logical and rhetorical ontology defines what never will, can, or could be possible. The theoretical distinctions between the concepts of the virtual and the real and between the concepts of the real and the impossible do not present any considerable difficulties. However, the practical application of these distinctions to narrative and aesthetic subject matter is somewhat more challenging, for here the concepts form a continuum instead of contrasting clearly with one another. Prince (1988) proposes an important classification of the various narrative and linguistic phenomena which express this continuum, which ranges from an open mode of fiction to an unqualified anti-fiction.11 His first category, the open mode, covers the overt omission of fictional objects by the narrator. It is common knowledge that no narrative can provide a complete representation of all the elements contained 10

11

The concept of the possible world is meant here in both the philosophical sense explained by Rescher and the narratological sense defined by Doležel, Pavel, and Ryan. Prince (1988) here returns to an idea he had already raised in a previous publication (Prince 1982:129). In contrast, Chatman (1988) confines the nonnarrated to the expression plane of discourse. For Chatman, a story is nonnarrated if the reader is unable to identify a narratorial instance in it.

14

Part 1 Concepts of Action

in a world; even so, Prince’s first and most simple category is anything but unproblematic. In normal circumstances, we can turn to a specific world knowledge in order to resolve most cases of the ellipsis of fictional objects, events, and perceptions. However, resolving an ellipsis in this way does not provide us with a reliable means of identifying the motives behind the omission, which can take place for any number of reasons, ranging from moral to ethical to technical considerations (examples of technically necessary ellipsis can be found when suspense is created and when information could only be related by adopting an inconsistent narratorial standpoint). Prince therefore proposes categorizing each case of omission according to its position between two extremes: the normatively unnarratable and the nonnarrated, which is omitted by the narrator for reasons of economy. Despite the distinction between them, both concepts can be subsumed under the wider category of rhetorical negativity, of which the standard examples are phenomena such as ellipsis, analepsis, and the like. Generally speaking, these phenomena occur when information is withheld because it is felt to be incompatible with the narrator’s intentions, would result in formal inconsistency within the narrative, or conflict with certain (e.g. social, cultural, and aesthetic) norms. Irrespective of the particular ellipsis in question at any given time, the recipient of the narrative will always be able to retrieve the omitted information by way of inference. Such omissions are therefore examples of Ingarden’s characteristic points of indeterminacy, which we perceive as challenges rather than hindrances. Rhetorical negativity is, therefore, a temporary rather than absolute phenomenon: anything that is implicitly asserted as not narratable will eventually be inferred by the reader, if not indeed narrated to him. Prince’s domains of the unnarratable and nonnarrated, which I have combined in the concept of rhetorical negativity, comprise an important part of the virtual side of diegesis. They cover the objects, states of affairs, and events which may, or indeed do, exist but which (for whatever reason) have not been actualized by the narrator and must in part, therefore, await actualization by the reader. The second kind of negative fictionality, on the other hand, is absolute and irreversible in nature. Prince refers to it as the disnarrated. When a narrator narrates something about a fictional world, he does more than just imply the presence in it of those things which are part of it but cannot be mentioned for various reasons. Every utterance that a narrator makes also contributes to the definition of what is not part of the fictional world because it cannot, as a matter of principle, exist or take place in it. Prince develops this concept by referring to the enumerative etymology of the French verb raconter (the idea is equally applicable to the German verb erzählen which also highlights the role of re-‘counting’ in narration):

1.1 Action as Product and Construct

15

In underscoring the displayable nature of narrative, its qualities of an object calling for contemplation, exploration, and response, the disnarrated shows that narrative is not only a matter of counting, accounting, and recounting, but also one of discounting. It insists upon the ability to conceive and manipulate hypothetical worlds or states of affairs and the freedom to reject various models of intelligibility, of coherence and significance, various norms, conventions or codes for world- and fiction-making. It institutes an antimodel in terms of which the text defines itself and indicates the aesthetics it develops and espouses, the audience it represents and aspires to, the matters, topics, and configurations this audience takes to be tellable. Prince 1988:6

Prince’s definition of the disnarrated is based on the discursive function of the syuzhet: the disnarrated serves to draw our attention to the aesthetic self-definition of a text. In a sense, then, the distinction between the category of the disnarrated and that of the nonnarratable and unnarrated is suspended once again. We can prevent this by restricting the term ‘disnarrated’ to the context of the fabula, thus emphasizing that what we refer to as disnarrated is omitted not temporarily, as in rhetorical negativity, but as an absolute and insurmountable ontic negativity which is a fundamental part of the design of the fictional world and which neither narrator nor reader can revoke. Prince himself alludes to such a narrower definition of the disnarrated when he refers to Ryan’s concept of the virtual narrative and Bremond’s triad of action logic (Bremond 1966, 1973).12 According to Bremond, any fictional happening that a text narrates can be described as the consequence of a schematic process which is only apparent in the logic of the fabula’s action. The process is one of choosing between the logically complementary possibilities of enacting (‘passage à l’acte’) and not enacting (‘non passsage à l’acte’) a potential situational transformation (‘éventualité’).13 It is the non-enacting of events that constitutes the distinctive domain of the disnarrated and thus identifies what I see as the ontic negativity of fiction (as distinct from its rhetorical negativity). Events alone are singular in their chronological position; by occurring at a given point in time, they can by implication rule out the occurrence of that which cannot have happened concurrently and cannot be imagined to have done so. In other words, only events can properly be disnarrated; because nothing in one object logically precludes the existence of another object, there is no way of asserting the non-existence of objects by way of

12

13

In Prince’s view, the formal plot model in Ryan (1986) expresses the ‘tellability’ of a narrative as a ‘function of unrealized strings of events’ (Prince 1988:5). The role of choice is also emphasized by Juri Lotman, who defines an event as something that happened and is conceived of in the light of the possibility of it not having happened (Lotman 1977:238f.). We shall discuss Lotman’s concept of the event in chapter 1.4.

16

Part 1 Concepts of Action

implication. Nonetheless, the disnarrated remains a function of the narrative and its discourse structure: the logical primacy of the fabula should not be misinterpreted as an ontological primacy. Both the fabula and its negative counterpart, the disnarrated, are abstractions which are based in the syuzhet and have no existence beyond their concrete instantiation in a narrative. With the exception of their virtual appearance in the formal rules of grammar, we will never find them separated from their narrative realization.14 One of the more controversial debates in narratological circles has its roots in the confusion of the ontological and logical arguments concerning the status of narrated action. Smith (1980) criticized narratologists for designing the majority of their theories and models against the influential background of the paradigm of the non-fictional, historical narrative. Common to this form of narrative is the fact that they typically have a documented original text which allows us the objective identification of the real sequence of events. As a result, Smith argues, the plot (= syuzhet) can be described as a variation of the original story (= fabula). In the case of the fictional narrative, however, she argues that narratologists inadvertently draw an impermissible analogy between fictional and factual narratives when they refer to the dichotomy between plot and story, for here there is no ‘original story’ in the first place. Smith’s criticism provoked a considerable number of responses.15 Genette was among those to note that the dichotomy of histoire and récit is of an idealized nature and does not presuppose the independent existence of the former, especially since the latter itself employs various means (ellipsis, analepsis, and prolepsis) to indicate that it is not a strictly chronologically ordered sequence of events.16 Cohn (1990) has added an ontological argument to supplement this phenomenological one. She points out that, while the historical narrative is based on documentary evidence, the fictional narrative posits its own fabula as its basis and presents itself, often rearranged, in the form of discourse:

14

15

16

Prince (1998) turns his earlier definition of the disnarrated on its head and states that it represents ‘what did not happen but could have’ (Prince 1998:46). Referring to Coste (1989), he concludes that we should conceive of narrated action as ‘the presence as opposed to the absence of alternative courses of action for the narrative participants’ (Prince 1998:46.). The argument, seen in more general terms beyond the narrower topic of action, concurs with that of Ryan (1991). Examples include Goodman (1981), Chatman (1990), Cohn (1990), and Genette (1990). For histoire, read ‘plot’ or ‘fabula’; for récit, read ‘story’ or ‘syuzhet’. As Genette (1990: 759f.) rightly points out, the rhetorical techniques of ellipsis, analepsis, and prolepsis can also be employed in non-fictional narratives.

1.1 Action as Product and Construct

17

A novel can be said to be plotted but not emplotted: its serial moments do not refer to, and cannot therefore be selected from, an ontologically independent and temporally prior data base of disordered, meaningless happenings that it restructures into order and meaning.…Story and discourse, as narratologists have repeatedly stressed…are conceived as synchronous aspects of fictional texts, with no presumption of priority of story over discourse. Cohn 1990:781

This is why most grammars of the logic of narrative and action are fundamentally descriptive rather than generative in nature. They present a model of their object domain which is intended to explicate its constituent processes and elements but does not necessarily present an algorithm that would allow us to generate fictional activity and action, let alone a complete narrative text. It is regrettable that this distinction has frequently been overlooked since post-structuralism opened the debate on narratological universals.17 So far, we have concentrated primarily on action as a project of narration. We shall now attempt to shift our attention to a preliminary concept of action as a construct, as the product of a choice between the actual and the virtual which takes place in the reader. One might well ask whether this aspect of the problem merits a more detailed analysis at all; the practice of summarization, long-established in the naive and critical reception of texts, certainly seems to suggest otherwise. Summarization rests on the methodological premise that what has been told is the equivalent of a specific complex of events which can, in principle, be unambiguously reconstructed.18 Readerly choices would not appear to have much of a role to play in this situation. Note, however, that the question of what a narrative contains must be distinguished from that of why it contains what it does. The ‘why’ is directed at the semantic function of the action in a narrative, and that function, of course, can only be established by means of interpretation. Indeed, the fact that interpretation inevitably leads to disagreement is practically the defining principle of literature and the disciplines of literary criticism and theory. Yet even scholarly, critical readers work on the general assumption that the epic action in some way constitutes a specific structure which can be proven to exist by citing the sequence of fictional events as evidence. For this reason, conflicting hermeneutic interpretations of a given narrative frequently refer 17

18

Tolliver (1997) is a recent proponent of criticizing classical narratology for positing narrative universals. Only in fantastic literature are we routinely confronted with worlds that contain conflicting ordering principles at the event level. In all other epic genres, any difficulty in deciding on the real sequence and order of events can be put down to the behaviour of the narrating authority, who can supply incomplete, faulty, or even deliberately contradictory reports.

18

Part 1 Concepts of Action

to the narrated action as the final and absolute measure of their validity. The minimum criterion of any interpretation seems to be that it should be at least compatible with what has happened in the fictional world and been reliably explicated as an action. Against the background of the question of the accuracy of a high-level interpretive hypothesis, action thus assumes factual status as evidence which must not be ignored by any reader who subscribes to the principle of rational interpretation. The method of ex negativo definition, which we exploited when discussing action as a project of narration, can also be employed to approach the concept of action as a construct of reception. Since Ingarden’s work, it has been common knowledge in narrative theory that the fictional world is by definition characterized not only by the material it explicitly includes, but even more importantly by patchy descriptions, schematizations, and omissions. The effect of such points of indeterminacy on how readers receive literary texts has been most informatively investigated by Iser (1970). In the context of German narrative theory, Stanzel argued—long before the watershed of cognitivism and constructivism—that if literary theory was to understand the issue properly, it would have to examine how readers generate complementary story elements and sequences (Stanzel 1977; 1995:118f.). In the contemporary literature, the phenomenon appears to be covered by the disciplines of cognitive theory, literary didactics, and constructivist literary theory, each of which deals with it from a different theoretical angle (Mandler and Johnson 1978; Lehnert 1981; Jahn 1997; Nünning 1990). The concept of literary action as such, however, is not debated in these contexts and has often just been borrowed from the social sciences.19 It is therefore appropriate to broaden our perspective and address the question of how and under what systematic conditions readers are able to make positive inferences about a narrated action. Narration is a finite act; action, as a construct that is generated on the basis of the narrated text, must therefore incorporate a finite set of fictional activities and happenings. This is the fundamental reason why we find it possible to ask ourselves exactly what action has been narrated when we finish reading a text. If we had a rule-governed procedure for reconstructing what happens in a text on the basis of its content alone, we would theoretically be able to make unambiguous statements about the action of the text by identifying two things: first, the elements which make up the set of fictional events contained in its world; second, the combinatorial rules according to which they can be systematically manipulated to produce an action. This idea lies at the heart of Propp’s formalist programme. By 19

This is the case, for example, in Kindt and Schmidt (1979), Schmidt (1979), Hauptmeier and Schmidt (1985), Barsch (1992), and Verdaasdonk and van Rees (1992).

1.1 Action as Product and Construct

19

developing the notion of the function in his Morphology of the Folktale, Propp constructs a fully text-based model for describing the fabula. The key feature of his approach is his plan to treat the elements of action structures in terms of their function rather than their content. This allows him to assemble a functional description of the action structure of each particular text. Propp’s analysis of Russian folktales leads him to conclude that each concrete action narrated by the texts in his corpus represents a partial realization of an archetypal action model. Since the set of action functions is finite and their combinatorial variants restricted by certain rules, the set of theoretically possible actions is also finite. However, a number of theoretically possible function combinations are not realized in the corpus of one hundred folktales which Propp examines. Despite his commitment to text-immanent interpretation, therefore, he cannot escape the fact that external and contingent conditions affect action design. Even Propp shows that the theoretical range of possible action constructs is subject to cultural restrictions in practice. However, the existence of superimposed selective principles is not in itself sufficient to justify rejecting the concept of fictional action as a text-immanent entity. Several critics of Propp, among whom Lévi-Strauss (1984) is most prominent, have pointed out that his methods lead to an inappropriately linear description of action. Tomashevsky’s approach, on the other hand, captures not only the principle of the syntagmatic order of a fabula but also the principle of the thematic and systemic combination of events: The theme of a work with a story represents a more or less homogeneous system of events that develop from one another and are connected to each other. We refer to the totality of events in their mutual internal connectedness as the fabula. Tomashevskij 1967:134 (my translation)20

Since Tomashevsky, this ‘totality of events,’ the sum of which constitutes the pure fabula, has been called a syuzhet when it appears artistically arranged in a work of art and not simply as ‘an entertaining chain of events…with a beginning and an end’ (Tomashevskij 1967:136; my translation).21 The introduction of the concept of syuzhet led to considerable terminological confusion as a result of the inconsistent use of paired concepts such as theme and syuzhet, motif and event, which are used as synonyms the one

20

21

The current English translation of Tomashevsky’s text (Tomashevsky 1965) is based on an earlier edition of the Russian original and does not contain this passage. Note that Tomashevsky understands the event as an anthropocentrically defined transformation between situations. Again, this passage is not included in the current English translation (see note 20).

20

Part 1 Concepts of Action

minute and antonyms the next.22 Despite this confusion, the examples in Tomashevsky’s systematic motif classification do show clearly that the motif should be understood as a thematic rather than logical element. The motif is a descriptive concept whose function lies in the hermeneutic reconstruction of the syuzhet; it identifies something that can be described as movens, something that introduces motion into the overarching complex of the fabula or syuzhet.23 This may help explain the lack of categorial homogeneity in Tomashevsky’s examples, which include both fictional objects, such as ‘revolver’, and descriptions of cognitive processes, such as ‘recognition’.24 Although the concept of the motif might appear to have practical relevance for the description of action as a construct of reception, its use actually leads to a not insignificant methodological problem. While the motif does indeed facilitate the inductive reconstruction of action in the sense of the fabula, it is based on the assumption that the fundamental systematic structure of a literary text can only be found at the syuzhet level. Propp’s notion of the function, on the other hand, is less prescriptive; it permits the methodologically consistent reconstruction of narrated action as a sequence of transformations of fictional states, regardless of the order of the syuzhet.25 A second danger inherent in the use of the motif concept is that it can easily leads to a psychologizing approach which confuses narratorial, if not authorial, motifs at the syuzhet level with the psychological motivations of fictional characters.26 This can only lead to the confusion of the concept of narrated action with that of mimetically represented activity. 22

23

24

25

26

Tomashevsky himself states, only a page after his first definition of fabula, that the story [fabula] is ‘the aggregate of motifs in their logical, causal-chronological order; the plot [syuzhet] is the aggregate of those same motifs but having the relevance and the order which they had in the original work’ (Tomashevsky 1965:68). Note, however, that the reference to the etymology of movens leads to a further contradiction because Tomashevsky actually distinguishes between dynamic motifs, which change the situation, and static motifs, which have a neutral influence on the situation. Examples of the former are the deeds and actions of heroes; of the latter, descriptions of nature, of locations, people, their characters, and so on. See Tomashevsky (1965:68). Doležel (1980) is an interesting attempt to overcome the terminological confusion in Tomashevsky’s approach and, at the same time, combine Tomashevsky’s microstructural approach with Propp’s macrostructural approach. As Kafalenos (1997:473) points out, Propp himself does not distinguish between fabula and syuzhet; genre conventions mean that they are isomorphic in his corpus of fairytales. Belknap (1985:242–51) compares the approaches and concepts which developed in Russian Formalism for theoretically modelling the basic units of literary plots. He identifies two principle methodological approaches. The first aims to develop a generative plot model; it conceives of all empirical plots as permutations of a single archetypal plot. The second approach, in contrast, is an analytical attempt to break complex phenomena down into their constituent elements. For Belknap, each approach is based on a different underlying world model. He believes that the concept of plot is derived primarily from Veselovskij’s definition of syuzhet in the sense of a complex of motifs. As a result, formalists drawing

1.1 Action as Product and Construct

21

Let us summarize the key findings of the above discussion. The advantage of Propp’s model lies in the fact that it allows us to derive fictional happenings, including fictional activities, deductively from the superior structure of the functionally structured action. The model’s major disadvantage is that it depends on a sequential ordering principle which gives the impression that any complex action in a literary text is basically no more than a linear sequence of events and, as such, equivalent to the fictional activities performed by the fictional characters. Tomashevsky’s concept of the motif avoids this linear reductionism but does so by turning a deductive explanation into an inductive explanation, thereby running the danger of falling foul of another kind of reductionism, that of modelling action as a product of the fictional intentions of fictional characters. In other words, neither Propp nor Tomashevsky allows us to differentiate clearly between the concepts of activity and action. This is more than just an abstract terminological problem. Propp’s functionalist approach must be seen as standing in the tradition of the aesthetic concept of action that was first put forward by Aristotle in his Poetics. Aristotle stated categorically that ‘[the persons] do not act in order to portray the characters; they include the characters for the sake of the action’ (Aristotle 1984:2320 = 1450a, 20f.).27 In this particular instance, Aristotle is not referring to the mythos, the overall action of a tragedy. ‘Action’ is meant instead in the sense of the position, happy or unhappy, to which the behaviour of an individual human caught up in a complex set of events and circumstances eventually and unwittingly takes him. For Aristotle, the constant features of a human being’s moral character and his cognitive faculties are the key factors which determine this process. However, no person acts solely on his own; and, as the ramifications and interdependent consequences of an individual’s behaviour cannot be foreseen, the mythos will always be superior to the fate of that individual. Or, put the other way round, ‘there are many actions of one man which cannot be

27

on Veselovskij used the concept of the motif to mean a ‘minimal unit’ of plot, ignoring the fact that Veselovskij himself had already referred to the motif in a double sense—as an ‘atomic perception of motifs as the many irreducible units out of which a plot may be assembled’ and an ‘alchemical perception of a motif as a single unit which may grow into a plot through the use of repetition rules’ (Belknap 1985:244). Tomashevsky subsequently adopted the dualism and ensured its survival. Against the background of these facts, Belknap (1985:245) argues that the concept of plot is based on the basic ideas of set theory in that all plots (sets) can contain a number of microplots (subsets). In my view, Propp’s concept of the function can be interpreted in a similar manner. It is not an indivisible unit but has the dual purpose of capturing an individual move (what Propp calls a khod in the sense of an individual transformation event) and the entire logically and functionally connected complex of events and happenings. The Greek original of this passage (Aristoteles 1982:20) refers to ‘praxeis’ (note the plural form) rather than ‘mythos’. This double nuance is lost in the English translation.

22

Part 1 Concepts of Action

made to form one action’ (Aristotle 1984:2322 = 1451a, 18f.). With these words, Aristotle questions whether a person’s individual activities, which are aimed at achieving a particular biographical objective, ever become part of the ordered complex of the overall action. One might even ask whether it is correct to accept the assertion, which Aristotle proclaimed as a poetic norm and Propp tried to prove empirically, that a text has a singular and coherent action at all. This doubt does not relate to the unity of plot (i.e. action) that is required by classical dramatic poetics; it is directed not at the level of poetics but at the more fundamental level of logic. Schmid (1982, 1984) in particular has shown that the action which is eventually narrated is the product of a multi-level process as far as the theory of its constitutive logic is concerned. The process begins with the artist’s inventio of a complex event. The next step is the selection, formation, and segmentation of the elements of the event to give the dispositio, the prerequisite for the narrative representation of the complex event as a meaningful artefact. This structural and material basis provides the foundation for the composition, the narrative itself, which integrates ‘the constituent parts of the event in their artificial order’ and is subsequently verbalized in the elocutio, the ‘presentation of the narrative’ (Schmid 1982:94; my translation). Schmid’s four-level model of the narrative is therefore one which captures both the logic and the process of composition; it does not treat them in terms of constructivist, reader-reception, or communication theory. Only indirectly does Schmid answer the question with which we are here concerned, the question of the origin of the ordering principle which organizes disjunct occurrences into a structured complex event as early as the level of the ordo naturalis (i.e. prior to any narrative transformation of fabula into syuzhet). Schmidt appears to identify the primary ordering principle as the artistic transformation of the initial selection of basic events into a structured sequence. The inherent order of the basic events prior to such a transformation is no more than potentially significant: The events are not devoid of meaning but are overspecified with meaning potentials. They are actualized by narrative transformations; the actualization results from the selection of material equivalences and the emphasis of selected relations by means of equivalences from other [thematic or positional] kinds of arrangement. Schmid 1982:99 (my translation)

Schmid’s observation provides a starting point for a discussion of the degree, if any, to which readers are able to make choices when they reconstruct action constructs in an interpretive activity which is based upon, but nonetheless transcends, the earlier designing activity of a narrator. My hypothesis is that, because the constituent events of an action are semantically over-determined, they offer additional combinatorial possibilities

1.1 Action as Product and Construct

23

beyond those that the narrator chose to use: there are virtual alternative combinations which can be explored and compared with the actual narratorial model of the action. In addition to semantic over-determination, there is a second variable that produces variance in how readers construct action constructs: the reader’s inherent ability to draw on individual and cultural contextual knowledge. The situation is best demonstrated by analysing a textual fragment which shows a high level of internal structuring: E. M. Forster’s frequently cited example of a minimal narrative The king died, and then the queen died of grief (Forster 1974:93).28 In this example, two basic events, (1) The king died and (2) The queen died, are combined in a threefold way: logically (and), temporally (then), and causally (of grief). Forster was of the opinion that it is essentially the last of these three, the element of causality, that transforms the story of two successive events into a true plot. There are two objections which can be raised to this analysis. Genette (1988:20) formulates the more radical one when he points out that Forster’s example is not a minimal story but actually one element of the set of complex and interesting stories; a truly minimal story would require nothing but The king died. At a fundamental level, Genette is of course correct, for every grammatically complete verb phrase implies a theoretically possible potential transformation.29 Thus, The king died obviously implies that the king must have been alive beforehand. On the other hand, the completion of the ellipsis (with, for example, The king was alive) is hardly sufficient on its own to turn the individual event into a sequence of events that could be rendered in the form of a narrative. The equivalence relation that Genette draws between sentence and narrative conflicts with the general categorization of different definitions of narrative which he makes at the outset of his study: first, narrative in the sense of ‘the signifier, statement, discourse or narrative text itself’; second, narrative in the sense of the ‘signified or narrative content,’ for which Genette proposes reserving the term ‘story’; third, the ‘producing narrative action’ which he terms ‘narrating’ (Genette 1980:27). For our purposes, we should focus on narrative in the second sense, bearing in mind also the distinction between the generative concept of narrated events and its equivalent constructive concept of action. Both denote the same (ideal) object, seen from the different perspectives (narrative and narrated events respectively) of the constructing producer 28

29

The example continues to feature prominently in contemporary introductions to narrative theory. See for example Martinez and Scheffel (1999:109f.), where it is employed to illustrate the distinction between event and narrative (‘Geschehen’ and ‘Geschichte’). Such verb phrases do not, however, necessarily require an ontic subject. For example, despite the absence of an ontic subject, the sentence It is raining implies the possibility of a transformation into It is no longer raining.

24

Part 1 Concepts of Action

and the reconstructing recipient who aims to integrate the event material which he finds into an action. My objection to Forster’s postulate depends on a more detailed logical analysis of his example sentence The king died, and then the queen died of grief. Significantly, none of the three conjunctions between the two events (and, then, and of grief) can be removed without dissolving the minimal story into two disjointed descriptions. Even the unassuming conjunction and is indispensable in that it is the only proof that the king’s death and the queen’s grief are contiguous. Contiguity in the sense of the adjacency of elements is of course the weakest relation that can exist between two distinct perceived facts, and it is immediately called into doubt by contingency. This is why the text needs to strengthen the hypothetical, purely formal logical connection by asserting a temporal then relation. As a result, the two events become connected in two ways, and this connection is now a potentially causal one. In order for the assumption of causality to become acceptable, further evidence is required; it is provided in the form of the explicit causal conjunction of grief. Contiguity, temporality, and causality—and, then, and of grief—form a hierarchy of logical propositions that finally results in the integration of the two events into an action. Furthermore, the modest little word and is important because it activates the frame of ideological reference without recourse to which a reader cannot plausibly construct the simple causal interpretation of Forster’s minimal story that we all find so obvious: The queen died because of her grief about the king’s death. What at first sight appears to be a fully text-based reconstruction of a single action turns out to be dependent upon our access to two sets of context factors. Only under certain conditions will a reader be able to reverse the transformational process described in Schmid’s fourlevel model and reconstruct a pre-existent complex event (‘Geschehen’) on the basis of its final manifestation in the textual narrative. The conditions in question are found only when the contextual knowledge of reader and narrator is completely equivalent. In practice, such complete equivalence is the exception rather than the rule. The more typical incongruity generates variance between the narrator’s construction of action and the recipient’s reconstruction of action. The first type of contextual knowledge involved, empirical world knowledge, is indispensable and always active. In the present example, it includes knowledge of such things as the fact that entities like kings and queens are animate and can therefore be born and die. This knowledge confirms that there can be a possible world inhabited by a king and a queen where the former can be transformed from being alive to being dead. This we might call an ontic probability assessment. It is followed by a more detailed analysis in the form of a transformational probability assessment that focuses on the reasons put forward for the

1.1 Action as Product and Construct

25

postulated causal link. At this point, we are bound to meet a problem in our reconstruction of the action of Forster’s minimal story. We all know that it is rather improbable (though not impossible) for a person to die of grief. The problem is exacerbated when kings and queens are involved, for we know that their marital relationships are typically a consequence of dynastic interests rather than romantic affection. Nonetheless, the problem is nevertheless easily solved; all we need to do is make recourse to the second category of world knowledge which affects the process of reconstructing an action. This type of knowledge we might call cultural knowledge. One element of this cultural knowledge is the fact that narrative representations are not necessarily restricted to empirical entities but may, to varying degrees, present fictional, experimental entities and the relations and transformations between them. Our cultural knowledge also contains the awareness that there is a specific genre of fictional texts in which the fictional situation that is problematic for our empirical world knowledge is the rule rather than the exception. The genre, of course, is the fairytale, according to whose principles the combination of romantic and dynastic facts is anything but counterfactual. This, in the final analysis, is how and why we are able interpret Forster’s The king died, and then the queen died of grief as an action construct equivalent to The queen died because of grief about the king’s death in which the elements that are empirically plausible (context one) are reconciled with those that are aesthetically permissible (context two). The example demonstrates that even a so-called minimal story requires the reader who wishes to construct an action to take into account various types of world knowledge in order to resolve inferences and find a plausible way of handling inconsistencies or contradictions at the level of basic events. Reading and reconstructing an action that is the appropriate equivalent of a narrated story is far more complex than we might suppose, and, particularly when understood as a hermeneutic process, it turns out to be based on implicit presuppositions which we normally ignore. Even the construction of causal connections between basic events can turn out to be dependent on specific pieces of contextual knowledge. This is the basic reason why nobody can force a reader to draw on a particular context and derive a particular action construct. The reader’s answer to the question of what the action of a given narrative is will always be one among many possible answers. Let us suppose that I, the reader, am unable to explain the strange fact that an aristocrat dies of romantic grief by using knowledge of genre conventions—perhaps I simply have no access to knowledge of this kind. In no way does this call the successive order and contiguity of the two basic events into question. It does, however, mean that I am unable to construct a valid causal connection between them—unless, of

26

Part 1 Concepts of Action

course, I simply accept the explanation offered by the text without further ado and complement my world knowledge accordingly without actually resolving the contradiction. In the long run, however, this may prove to be a problematic strategy, for I am bound to end up introducing contradictory facts into my world knowledge. It is at the point where the contiguity of and then can no longer be automatically translated into the causality of and therefore that we realize the possibility and necessity of thinking about action not only in terms of something that has been signified but also in terms of something that signifies and calls attention to the fact that every reading of an action is a constructive act based on numerous premises. This holds true not only for the reading of literary action but also for our encounters with action in everyday life, where, just as in literary narratives, there will never be a single definitive action construct which integrates all the elements of a given set of events. From a genetic point of view, it is of course plausible to assume that the narrator or author of a fictional narrative has carefully selected the events which his story presents. This means that our reception is governed by the heuristic premise that the narrated events are likely to form a complex and coherent structure. Nonetheless, the essence of this heuristic premise is that we can find a structure, not the definitive structure (which, of course, does not exist). We have still to address the conditions under which the meta-activities of production and reception can be thematized as actions by the narrative text. In addition to their fictional signification, narrative texts also refer metonymically to something which is not just a linguistically evoked mental image but an empirical fact, a concrete and real activity which takes place in our own world: the completed process of narration itself. Action is thus found at two distinct levels, the object-level of narrated events and the metalevel of the action of narration. We have shown above that the decoding of narrated events and action at the first-order level of the narrated world’s fictional objects and occurrences is context-dependent. The decoding of second-order actions, on the other hand, is governed by different logical conditions. Although the distinction between the two levels of action is usually defined in ontological terms, there is a second, and perhaps more fundamental, difference between them. This new difference stems from the logical ambivalence of the concept of action itself. Taken in the theoretical Aristotelian sense, action (i.e. plot) as mythos refers to a global structure which is, from the perspective of the characters involved, either the result of contingent processes or governed by a teleology inaccessible to human beings. Taken in the anthropomorphic sense, however, action denotes an intentional doing. Narration falls into the latter category because it denotes not something that actually happens but rather the intentional artistic design

1.1 Action as Product and Construct

27

and production objectified in a narrative. This premise is the general foundation on which all hermeneutics rests. From classical prescriptive poetics to contemporary narrative theory, the narrator’s utterances have always been conceptualized as intentional in this broader sense. While it does not require us to subscribe to a strict theory of authorial intention, we must recognize that even a discrete intention of which the narrator is unaware is sufficient to make the narrative act an intentional one. Indeed, without the assumption that a narrator is capable of making intentional utterances, it hardly makes sense to define the non-expository narrative text as one constituted by propositional narrative statements which, because they are declared fictional, are exempt from having their truth-values validated. The intentionality presupposition is one of the basic conventions on which the reception of literature, indeed any symbolic system, is based. Categorizing something as information implies thinking of it as message sent with an intention. This feature sets information apart from indication, which lacks intrinsic meaning and may even have been generated by chance and without signification in mind, as we sometimes (rightly or wrongly) believe to be the case in works of art produced by mentally handicapped people (such productions are often categorized as symptoms rather than messages).30 Literary theory is widely characterized by a tendency to clarify the semiotic status of a text by attributing it to an intentionally acting uttering instance, as is perhaps best demonstrated by the ongoing debate surrounding the concept of the implied author. To deny that it is possible for a literary text to be generated intentionally and purposefully leads to a disturbing aporia: we must either accept that the text speaks to us in the mythical sense of the world speaking to mankind, or rather God speaking to us via His book (i.e. the world and its content), or fall into the delusion that the text, which is ultimately only a medium, possesses ontological autonomy and speaks for itself. And that means that we end up talking to ourselves. In order to avoid this aporia, we must conceptualize the act of narration as more than a simple happening and see it as a purposeful act analogous to the acts which we perform as readers. The analogical projection of this intentionality hypothesis often transcends the context of the narrative authority when it is applied retrospectively to the fictional object-level. The German language offers linguistic evidence to support this argument. Even in cases where a series of narrated events is not attributed to an intentionally acting originator, we still refer to it as a Handlung, an action in the sense of mythos. The anthropomorphic origin of the term is demonstrated by its etymological root concept, Hand (‘hand’). At this point, we encounter the vital distinction, dependent on intentionality, which must be 30

See Eco (1981:38–42) on the distinction between sign, indication, and symptom.

28

Part 1 Concepts of Action

made between praxeological and literary concepts of action. Since all our ideas about the order and causality which govern human behaviour and the empirical world are of a historical nature, it makes sense to assume that this distinction too is the result of specific cultural and ideological conditioning. When we still believed in God as demiurge and director of events (Weltenlenker), all that happened in the world, whether a contingent occurrence or caused by humans, was an ultimately intentional act. In this cultural context, therefore, the concepts of action and occurrence did not denote fundamentally different classes of happening (intentionally caused happenings versus contingent happenings). Instead, there were two coexistent models of explanation for the same set of observed facts, one rooted in the world and the other transcendent. The latter model has generally been considered obsolete since the Enlightenment—and yet we continue to use ‘action’ in the pre-Enlightenment sense to refer to something that, seen in terms of its own internal logic, is a non-homogenous sequence of fictional occurrences, doings, and events. However, we never use the term ‘action’ in this way to refer to empirical and intentional acts of reading and narrating. There are two ways to explain this surprising philosophical inconsistency. The first is that we, living in the post-Enlightenment era, have decided to project the medieval ordo concept onto aesthetic representations of event complexes and promote the narrator and/or recipient to the status of an intentionally acting demiurge. The second is that the aesthetic concept of action actually has nothing in common with the commonsense interpretation of the word. It can be defined in formal terms without recourse to the criterion of intentionality, and, in this respect, the opposition between immanence and transcendence set up by the Enlightenment is destroyed. If this is the case, a fictionally presented series of events that has been integrated to form an action should be appreciated as an aesthetic experiment in which the fundamental problem of modernity, the loss of faith in an all-encompassing intentional cause for everything that takes place in the world, has long since been overcome. In the end, the interest of the concept of action for literary theory will prove to be inseparable from broader questions relating to the correlation between the aesthetic and ideological functions of fictional action narratives. In order to consider critical and praxeological definitions of the basic terms, the notions of action and activity must first be considered from the perspective of their historical development. For this reason, we shall first discuss the etymology of the two concepts, taking Handlung—the German equivalent for action—as an exemplary case. We will then try to sketch how the word gradually developed into a more narrowly defined concept of key relevance to the poetological as well as to the critical discourse on literary texts.

1.2 Action: From Word to Concept The Etymology of the German Word Handlung Analysing a word’s etymology is no substitute for an accurate definition of its meaning. Despite this caveat, the following excursus on the historical development of the German word Handlung will illustrate the complexity of the word and its meanings and should also provide us with a preliminary insight into the range of concepts which the word has denoted at different stages in its history. Current dictionaries suggest a variety of English translations for Handlung: ‘act’, ‘action’, ‘activity’, ‘deed’, ‘plot’, ‘storyline’, ‘exercise’, and ‘shop’. No more than a glance at some of the standard works of German etymology is needed to illustrate the extent to which the word has been affected by changing sociohistorical attitudes to human activity.31 The original meaning of the word is reflected in the stem of the Old High German verb hantalon, which contains the image of a feeling, moving, and manipulating hand or Hand, thus showing the original meaning of the word as ‘to manipulate’. Modern German expresses this sense using a verb such as behandeln or bearbeiten. Interestingly, the prefix be- is a relatively late addition; even in Luther’s time, it was still possible to refer to the Handeln of a person or object without making the transitive nature of the action morphologically explicit. This is similar to the use of the verb to handle in contemporary English, which can refer transitively both to concrete (Handle this package with care) and abstract entities (I can handle this problem myself) and can even be transformed into a noun (handling). Handling has become a familiar Anglicism in the German language, particularly because it can only be translated with somewhat unwieldy compounds such as Handhabbarkeit and Bedienerfreundlichkeit. However, the English handling not only refers to an activity as such but can also denote the mode, result, or expertise associated with it, as in Her handling of the situation was highly competent. The gist of this sentence is not that someone manipulated an object in a concrete fashion; it is rather 31

The following works were consulted: Etymologisches Wörterbuch des Deutschen (EWD 1993), Deutsches Wörterbuch (Paul 1992), Wörterbuch der deutschen Gegenwartssprache (WDG 1969), Deutsches Wörterbuch von Jacob Grimm und Wilhelm Grimm (Grimm 1984), and Historisches Wörterbuch der Philosophie (Ritter 1974).

30

Part 1 Concepts of Action

a statement about the ultimate effect of the woman’s activity. Although the connotation of an action performed on an object is preserved by the morphological link with the verb to handle, the primary function of such uses of handling is to provide a qualitative assessment of the particular way in which a person chooses to deal with a given situation. In this respect, English better preserves the pragmatic context which, as the shared rood hand- implies, was originally present in both languages. To express an intransitive human activity in English, we normally select a different verb altogether and use the subject-orientated to act, which originates in the Latin actus. In German, on the other hand, it was possible as early as the twelfth century to use the verb handeln ‘to act’ intransitively in the sense of doing, completing, producing, or pursuing something, as described in the Grimms’ dictionary (‘etwas thun, vollbringen, verrichten, betreiben’). The Grimms note that, in all of these uses, ‘the original association of work done by hand has receded or disappeared altogether’ (‘der engere sinn der handarbeit zurücktritt oder untergeht’ Grimm 1984, 10:col. 374; my translation). This is why German allows us to use the verb, not the noun, to refer to an instance of handeln that can be right or wrong, successful or unsuccessful, but need not necessarily deal with something or somebody. Modern German, therefore, can still express the abstraction inherent in a modal or normative judgement without having to use one of the verb’s nominal forms such as der Handel, das Handeln, or die Handlung. Of these nouns, Handel is a relatively late addition to the language which first appears in Middle High German, where it has a wide range of meanings. The word could refer to a manual doing, a process, a state of affairs, a legal case or dispute (with the plural händel), slyness, and, finally, an economic transaction or the object (commodity) of such a transaction. Hantalunga, the first noun to be derived from the original verb, was available to speakers in Old High German and underwent only minor change to become handelunge in Middle High German. Its modern phonological descendant is Handlung. Unlike the etymologically more recent Handel, however, Handlung and the many compounds of the root Handel- used to be almost completely restricted to the semantic field of objects, events, and facts related to economics or justice. The number of such compounds is impressive; the tenth volume of the first edition of the Grimms’ Wörterbuch contains fifty-two compound nouns with Handel- as their root, ranging from Handelsamt to Handelszweig. A similar situation exists with words containing the root Handlung-. Among them, confusingly enough for narratologists and literary theorists, the Grimms’ Wörterbuch even lists the word Handlungsplan, whose first occurrence it locates in a text by the German author Wieland. It is defined as the ‘plan zu dem gebäude einer

1.2 Action: From Word to Concept

31

kaufhandlung’ (Grimm 1984, 10:col. 408). This is hardly unambiguous and can mean both the abstract ‘plan of action for a purchase’ or the more concrete ‘architectural design for a business building’—neither of which is particularly poetic. We can draw the above observations together by concluding that, for a long period in its history, the German verb handeln was able to alternate flexibly between transitive and intransitive usages without any change in its form. The meaning of the related nominal form Handlung suggested intentionality from an early date as a word for referring to a specific form of goal-orientated activity in the social world. In contemporary German, the situation has been reversed. The verb now requires obligatory prefixes and prepositions in order to mark whether it is meant in a transitive or intransitive sense, whereas Handlung can mean both the individual activity of a person and a highly complex structure of cause and effect which in certain conditions unites several such instances of individual activity, be they real or fictional. Luther was the first to use the verb handeln in the abstract metaphorical sense of ‘dealing with an idea’ when he said of himself that he too was ‘counted among those who deal with God’s scripture on this earth’ (‘gerechnet unter denen, die göttliche schrift jetzt auf erden handeln’; my translation). However, according to the Grimms’ Wörterbuch, the first use of Handlung in the narrower figurative sense of ‘narrated action’ occurs in Lenz, who asked: ‘Wovon handelt das Buch?’ (‘What is the book about?’; my translation). This question becomes more common towards the end of the eighteenth century as the use of Handlung to refer to the sequence of events depicted or narrated in a dramatic or epic text finally begins to spread into non-specialist language. At first sight, this new meaning of the term Handlung appears to embody like no other in the German language the Romantic doctrine of the autonomy of art. In fact, however, this meaning is really a logical stage in the development of a far older tradition of reflection on poetic theory.32 Action as a Concept of Poetics Until the late eighteenth century, the German word Handlung was primarily used as a technical term to refer to individual scenes or acts in a play. In this usage, the word denotes the elements out of which the staging or narrating process is composed (not the entire complex of events which 32

This comes close to providing proof that, at least in this particular instance, etymology is a historical surface phenomenon.

32

Part 1 Concepts of Action

takes place in the staged or narrated world).33 In effect, this definition of Handlung as a dramatic technical term applies the word as a translation of the Latin actus. It competed with the metonymical use of Handlung to refer to the sequence of events in a play as a whole or, at times, the play itself. In the mid-seventeenth century, Gryphius speaks of the sequence of events in a play as the ‘complete action’ (‘ganze[n] Handlung’; my translation). Moriz Heyne, editor of the relevant volume of the Grimms’ Wörterbuch, considered this formulation important enough to be highlighted as a special case: ‘but apart from this [Gryphius’s] technical meaning, handlung had already been used previously in a more original way to refer to the sequence of what happens in a drama’ (‘aber abgesehen von dieser technischen bedeutung war handlung schon früher in frischerer weise auf die reihenfolge dessen, was in einem drama vorgeht, bezogen worden’ Grimm 1984, 10:col. 405; my translation). Only with Lessing was the terminological confusion among German critical writers finally overcome. As a critic and practising dramaturge, Lessing’s key principle was to consider dramatic Handlung from a functional perspective, that is, in terms of the representational and denotational intention of the text. With reference to the moral tale (a popular eighteenth century epic narrative genre which expounded moral and ethical norms by means of an exemplar), for example, we read in the famous Hamburgische Dramaturgie: We are satisfied if this intention is fulfilled and it is the same to us whether this is so by means of a complete action that is in itself a rounded whole, or not. Lessing 1962:10034

Different criteria, however, apply to the drama, which is not designed to exemplify a moral proposition but primarily intended to create an aesthetic effect (psychological or mimetic plausibility) and which therefore requires ‘a certain integrity of action, a certain harmonious end’ (‘eine gewisse Vollständigkeit der Handlung, ein gewisses befriedigendes Ende’ Lessing 1973, 5:93; translation from Lessing 1962:101). For Lessing, the classical unity of action (plot) and the unities of space and time (required because of the pragmatic limits on staging complex events, if nothing else) are more than formal restrictions imposed from outside; they should be treated as key productive principles exemplified by the authors of classical antiquity. Lessing treats aesthetic action as more than just the mimesis of 33

34

Goethe, for example, used the word in this sense in his review of a play in 1773. See “Neue Schauspiele, aufgeführt in der Kaiserlich Königl. Theatern zu Wien. Preßburg. Erster Band” (Goethe 1995, I.37: 227). ‘Wir sind zufrieden, wenn diese Absicht erreicht wird, und es ist uns gleichviel, ob es durch eine vollständige Handlung, die für sich ein wohlgeründetes Ganzes ausmacht, geschieht oder nicht’ (Lessing 1973, 5:93).

1.2 Action: From Word to Concept

33

individual activities with varying degrees of embellishment; it is the ideal of a meaningfully ordered complex sequence of events. As Lessing notes in his Abhandlungen über die Fabel, there is a marked difference between this concept of action and the pragmatic everyday meaning of the word: ‘In vernacular usage, an action typically refers to something which is performed as a consequence of a certain intention’ (‘dem Sprachgebrauche nach heißt gemeiniglich das eine Handlung, was einem gewissen Vorsatze zu Folge unternommen wird’ Lessing 1973, 5:377f.; my translation). Thus, the defining criteria of a real empirical human action are that it must be motivated by an intention and directed towards a goal. These features can characterize an isolated doing or an entire ‘sequence of movements’ (‘Reihe von Bewegungen’ Lessing 1973, 6:594; my translation).35 The aesthetic concept of action is very different: ‘objects which follow each other, or objects whose parts follow each other, are generally called actions. Actions, therefore, are the true subject matter of poetics’ (‘Gegenstände, die auf einander, oder deren Teile auf einander folgen, heißen überhaupt Handlungen. Folglich sind Handlungen der eigentliche Gegenstand der Poesie’ Lessing 1973, 6:103; my translation).36 Lessing’s definition of action as a concept of the drama allowed this more restricted and specific use of the word Handlung to become established in German as an aesthetic concept. It stands in the same tradition as the older discussion of mythos in Aristotle’s theories and brings us to the end of our study of the etymological and historical background of the word Handlung. We have arrived at the first explicit definition of Handlung in a specifically theoretical context, where it is treated as equivalent to the concepts of mythos and aesthetic action. The meaning of the word in this context is no longer bound to its original pragmatic association of manual doing and is instead abstractly defined in terms of structural arrangement and meaning. Despite its significance, Lessing’s definition is dogged by a fundamental problem which, even today, continues to present itself in all attempts to define action. The problem concerns the question of whether action is always, as the vernacular use of the word assumes, accompanied by intentionality. Should not this empirical concept of action as intentional doing be distinguished from an aesthetic concept of action in which events are not reconstructed in terms of the motivations and goals of their agents but by necessity interpreted ex post facto from an external perspective? 35

36

This quote is from Lessing’s posthumously published work on topics similar to the subject matter of the Hamburgische Dramaturgie. Lessing repeatedly attempted to put forward an even clearer definition of the concept of action. In the process, he found it important to point out the differences between the modes of aesthetic representation of literature and painting. We return to this point on page 108.

34

Part 1 Concepts of Action

In such an aesthetic concept of action, events initially perceived in a chronological sequence are assigned a hypothetical abstract order which cannot be empirically or objectively proven and whose validity depends solely on hermeneutic plausibility and consistency. Lessing took this path and argued that we should make a fundamental distinction between two separate notions of action. There is a praxeological, empirical notion of action, which refers to the real world of human experience; and there is an aesthetic notion of action, which refers to fictional event complexes. Lessing found this distinction particularly important for aesthetic reasons; he argued that the empirical concept of action forces us into an aesthetically crude materialism if we use it in the context of a work of art: Are there not, after all, critics who associate…such a material concept with the word ‘action’ that they can find action only where bodies are active in such a way that they effect some kind of change in space. They cannot see action in a tragedy except when the lover falls to his knees, the princess faints, and the heroes fight….It has never occurred to them that each internal emotional struggle, each chain of thought…is an action too. Lessing 1973, 5:373 (my translation)37

If we follow Lessing and include the thinking of thoughts among the phenomena covered by the concept of action, we can no longer apply the intentionality criterion in aesthetic contexts. Lessing’s contemporary, Johann Jakob Engel, on the other hand, does not see the aesthetic concept of action as constituting a separate category of its own. He argues that it arises from internal differentiation within the empirical concept of action. Accordingly, he observes that ‘there is action in a poem only if we see a change in it which is brought about by the activity of a being who acts with intentions in mind.’ (Engel 1774:15, my translation). However, Engel also refers to a concept of action similar to that proposed by Lessing, for he points out that it is a fact of aesthetic reception that we as readers eventually find ‘the unity of action in the unity of the change produced; if we work backwards from this change to seek the causes that can explain it to us as the whole that it is, then all the answers we find will be part of this one action.’ (Engel 1774:19, my translation.)38

37

38

‘Gibt es aber doch wohl Kunstrichter, welche einen…so materiellen Begriff mit dem Worte Handlung verbinden, daß sie nirgends Handlung sehen, als wo Körper so tätig sind, daß sie eine gewisse Veränderung des Raumes erfordern. Sie finden in keinem Trauerspiele Handlung, als wo der Liebhaber zu Füßen fällt, die Prinzessin ohnmächtig wird, die Helden sich balgen….Es hat ihnen nie beifallen wollen, daß auch jeder innere Kampf von Leidenschaften, jede Folge von Gedanken…eine Handlung sei.’ See Bachmann-Medick (1989) for a more detailed treatment of Engel’s contribution to the debate.

1.2 Action: From Word to Concept

35

Aristotle himself seems to have been aware of the difference between a distinct non-intentional aesthetic concept of action and an aesthetic concept with intentional and non-intentional variants. He understood the activity of fictional agents as mimesis of the intentionally motivated doings and behaviour of real persons. His poetics, however, classifies such doings and behaviour as events; by definition, this means that they are not associated with fictional individual intention and tied to a superindividual aesthetic intention instead. In the Poetics, Aristotle accordingly declares that the most important part of the process of creating an action narrative in a drama is the combination of the incidents of the story. Tragedy is essentially an imitation not of persons but of action and life.…In a play accordingly they do not act in order to portray the characters; they include the characters for the sake of the action. Aristotle 1984:2320 (= 1450a, 15–21)

Aristotle’s list of the six elements of tragedy gives precedence to the mythos in the sense of a ‘structure of events,’ a state of affairs which Manfred Fuhrmann has aptly labelled the ‘primacy of the action structure’ (Fuhrmann in Aristoteles 1982:110, note 12; my translation).39 Aristotle’s emphasis of a feature of structure rather than content is a direct consequence of his functional definition of tragedy in terms of its psychological effect on the audience. This definition is reflected in his statement that the audience of a tragedy is most affected by two particular components of the mythos, peripeteia and recognition. The depiction of characters and circumstances may interest us, but, according to Aristotle, the only thing that emotionally affects us is the changing fate which the structure of the overall action imposes on these characters and circumstances. The Poetics treats this definition in terms of psychological effect as a normative postulate as soon as it appears. Consider how Aristotle praises the exemplary poignancy and coherence of the mythos in Homer’s Iliad and Odyssey: in his view, Homer’s mythos, centred around a pivotal event, could be realized in the form of the singular action of a tragedy, something that other dramatic and epic writers could never achieve because they tend to combine networks of parallel actions and subplots. With Aristotle, the concept of action is elevated to categorial status, even though its substance has not yet been defined. It provides him with a common factor and aesthetic criterion which can be applied to all literary genres; it is a categorial postulate stated in the same breath as the definition of those genres in terms of their distinctive methods, subject matter, and modes of representation. 39

The elements of tragedy are plot (i.e. mythos), character, thought, verbal expression, music, and spectacle.

36

Part 1 Concepts of Action

The decision to employ action as the central normative concept in poetics must be seen both as the result of a change in philosophical paradigm and as an event which took place against the background of developments in cultural history. Aristotle himself was to some extent responsible for the former; for much of its length, his Poetics is a response to Plato’s ontological condemnation of art, which, in accordance with Platonic idealism, denounces it as a source of doubly false images. According to Plato, the empirical objects which our senses perceive as existing in our world are themselves nothing more than transient, inconstant, and therefore deficient representations of eternal and immutable divine ideas. Representations of such objects in the arts are therefore doubly transient, unreliable, and false. Plato makes an axiomatic distinction between the deceptive and incomplete world of the transient and the true and perfect world of the eternal. Provided that the proper conditions are met, this need not prevent a sequence of transient phenomena from having philosophical value. As Fuhrmann has shown with reference to the Phaedrus, the relationship between the idea (the eternal) and the phenomenon (the transient) can be treated negatively as one of deficiency or positively as one of participation and mediation. For Plato, this mediation between reality and ideas, between the sensual and the spiritual, was ‘a privilege of the beautiful, and the beautiful alone can make possible the ascent of the soul from the world of transient phenomena to the intransient world of divine noumena’ (Fuhrmann 1992:82; my translation). In Plato’s view, however, the art that he saw around him certainly did not meet the requirements which followed from his conception of the beautiful as an intermediary between the sensual world and the spiritual world of the intellect. In Plato’s eyes, art did not try to mediate between these worlds directly by accessing the ideas themselves; instead, it confined itself to representing the reflection of ideas in phenomena. Although tautological, this is the essence of the argument behind his condemnation of artists and writers as the deceitful producers of doubly false illusions. It is interesting to examine the consequences of Plato’s ontology for the idea of action as an aesthetic concept.40 Being facts of perception, real empirical actions can only be perceived as sequences of states. And, because they are continually changing configurations of individual objects of perception, they exemplify the principle of transience in two ways: they are dynamic structures and are composed of dynamic phenomena. Only from a retrospective, reflexive point of view can they be interpreted as teleologically governed and thereby brought to a certain degree of timelessness,

40

Kauffmann (1993) provides a systematic analysis of Plato’s theory of action.

1.2 Action: From Word to Concept

37

the hallmark of eternal ideas.41 However, action as a dynamic process of activity belongs in the world of illusion even when it takes place outside a fictional context. It is therefore the utter opposite of both static divine ideas and human happiness.42 This is the philosophical background to Aristotle’s postulation of the primacy of action in his Poetics. He decides to do so in order to reject the radical Platonic dichotomy of idea and phenomenon. Introducing the concept of entelechy (the theory that the idea is realized in the phenomenon) allows dialectal mediation between the two ontological spheres which Plato treated as absolute opposites. Only against this background can action be seen to have any philosophical value. It ceases to be a deceitful and illusory sequence of phantasmagoria and is now nothing short of the process with which ideas are realized and manifested in reality. As a result, interest in the artistic mimesis of action wins an astonishing philosophical legitimacy. According to Aristotle, it is the artistically crafted mythos that, unlike the supposedly true to fact representations of historiography, enables the elimination of things that may have happened accidentally but were basically logically irrelevant. This refinement helps to emphasize the character of pure timelessness which is the source of the dignity of a singular and perfectly coherent action. Aristotle’s theoretical interest in the concept of action is not only intended to overcome Plato’s ontological dichotomy between idea and matter; it also rests on the process of cultural history which led to the emergence of the classical Greek tragedy on which the Poetics focuses. Although Aristotle does not influence this process, he reflects on and theorizes about it. For a long period, the precursors of the classical Greek tragedy paid little attention to the representation of action (Barthes 1985:63–88).43 They were based not on the symbolic representation of a fictional activity but on a complex type of empirical speech act on stage that took the form of a narrative or commentary delivered by the chorus. The dithyrambs are one example of this. They evolved towards the end of the seventh century b.c. and were introduced to Athens by Thespis in around 550 b.c. They consisted mainly of a performance by the chorus, who danced around the 41

42

43

Fuhrmann (1992:87) has examined the consequences of applying the Platonic scheme of a three-level ontology (idea, phenomenon, and phantasmagoria of the phenomenon) to Homer. He concludes that one would have to state that Odysseus blinded Polyphemus with a view to the idea of blinding him, that Homer represented the representation of this idea, and so forth. Plato defines happiness as lasting possession of the beautiful and the good (Fuhrmann 1992:83), while action—and even more so second-order imitated action—is change and therefore the antithesis of the eternal divine truth. The French original of this text dates back to 1965—Barthes’s structuralist period, in other words.

38

Part 1 Concepts of Action

altar of Dionysos, and included soloists, but not actors in the true sense of the word. The chorus retained the role of collective protagonist in the satyr plays, a Doric import which, from the fifth century onward, was seen as the obligatory conclusion to all theatrical performances of three tragedies. Even in the early stages of the classical Greek tragedy, the episodes which represent actions are closely bound into a structure of discursive components whose function is distinctly rhetorical rather than mimetic.44 In general, therefore, early Greek theatre had hardly anything resembling an action in the sense of a structured sequence of events that organizes the entire text and affects the mode of (re)presentation in it. Instead, as Barthes has described, the earlier forms contain the fixed alternation of the spoken and the sung, of action and commentary. Indeed, there may be better reason to call it ‘narrative’ than ‘action’; in tragedy (at least), the episodes (our acts) are far from representing actions, that is, immediate modifications of situations; the action is generally refracted through intermediate modes of exposition which distance it by telling it: narratives.…This narrated action is periodically suspended by the choral commentary which obliges the public to collect its thoughts in a simultaneously lyric and intellectual mode. Barthes 1985:67f.

Only towards the end of the fifth century b.c. did the chorus decline in importance as the leader of the chorus, the exarchon, develops into the first actor. This was a decisive step towards the emergence of theatrical illusion supported by other mimetic techniques (costumes, masks, and stage mechanisms). It was probably the introduction of a second actor (attributed to Aeschylus) that made the representation of dialogue on stage possible for the first time ever (Grant 1991:76f.). Simultaneously, the original liturgical function of the tragedy began to recede. The older forms had focused on the ritual assertion of theological and cosmological axioms by applying them aesthetically to a complex of events that was predominantly narrated, but not actually depicted on stage. Now, however, the thematic focus shifted to the event complex as such, which was communicated by means of theatrical presentation and understood as represented action. The technique of choreia (the unity of poetry, music, and dance, the three forms used to express commentary and representation) is forced into the background as precedence is given to the illusion of a separate reality; or, as Barthes puts it, ‘what happens on stage is no longer the sign of reality but its copy’ (Barthes 1985:86). By asserting the primacy

44

The actions are framed by prolog, exposition, parodos, stasima, and exodos.

1.2 Action: From Word to Concept

39

of action, the Poetics provides the theory to accompany what had already established itself as theatrical practice over the course of centuries. 45 One would therefore expect Aristotle’s discussion to contain examples from tragedies that date from the second half of the fifth century b.c. and were produced after the liberation of the action of the tragedy from the choreia principle. Aristotle does indeed mention individual authors such as Aeschylus and Euripides in order to substantiate particular arguments in his Poetics. But he also refers to much earlier writers and works, even, in fact, to genres like the dithyramb or the nombos, which at first glance seem to be characterized by the complete absence of an action structure. To make matters even more confusing, Aristotle elevates Homer’s Iliad to normative status—it is an epic, not a drama, and at least three hundred years older than the majority of the tragedies discussed in the Poetics. It almost seems as if these texts, which vary substantially in their genres and dating, are comparable solely because the concept of action is applied to every single one of them alike. If anything, this would seem to be a circular argument, especially given that the concept of action itself is still being discussed and formalized in the Poetics. But the tautology is only an apparent one. We have already noted that the Aristotelian definition of action is not restricted to the abstract construct of the complex sequence of events in their totality but also refers to the lower level at which individual characters are involved in activities without having any knowledge of or insight into a teleology which governs their lives. It is for this reason that Aristotle is able to consider dance and music among the forms of mimesis: ‘For even he [the dancer], by the rhythms of his attitudes, may represent men’s characters, as well as what they do and suffer’ (Aristotle 1984:2316 = 1447a, 27f.). Each of these kinds of representation uses its own particular methods to portray various aspects of human behaviour, and it is this narrower, phenomenologically defined notion of action as an empirically observable doing or activity that allows all works of art to be compared regardless of their genres and forms. From this point of view, every work of art contains representational elements which can potentially be integrated to form a complex, teleologically orientated overall action in the sense of the higher-level logical concept which is normatively applied in the Poetics. The Aristotelian concept of action does not claim aesthetic autonomy and complete independence from the empirical concept of action and its 45

Although the Poetics, at least in those parts that have been preserved, deals mainly with the tragedy and mentions the epic only briefly in chapters 24–26, scholarship has considered Aristotle’s remarks on topics such as the aesthetics of effect, formal features such as completeness and closure, and the logic of action as universally relevant and not restricted to the drama.

40

Part 1 Concepts of Action

associated intentionality criterion. Instead, its key feature is that it differentiates between two distinct levels of action. It can denote (a) the mimetically represented behaviour of fictional characters who believe themselves to be acting intentionally or (b) the complex structure of events which has been intentionally designed by an artist so as to be meaningful (i.e. the mythos that draws together all instances of individual behaviour, be they intentional doings, or contingent or non-intentional occurrences). The phenomena which fall under level (a) of action can be described in analytical statements as the strategic projects of individual agents—provided, of course, that such agents are acting intentionally. Level (b) of the action concept, however, is concerned with a synthetic construct which provides its components, the basic actions of individual fictional agents, with a meaning that far transcends the strategic calculus of those same fictional agents. Aristotle himself points to the implications of this double concept of action when he states that ‘there are many actions of one man which cannot be made to form one action’ (Aristotle 1984:2322 = 1451a, 18f.). The distinction between the analytic and synthetic concepts of action which we find in the Poetics is of considerable relevance to modern literary theory in general and narratology in particular. Towards a Concept of Aesthetic Action Despite Aristotle’s repeated reference to criteria of aesthetic effect, a primarily normative concept of action is developed in the Poetics. This normative approach presupposes that the design of the mythos is subject to a global intention concerning the message and effect which the poet wants his work to have. In the context of a work on poetics, this is a plausible, perhaps even unavoidable, assumption. In a hermeneutically orientated scientific theory of literature, however, it is methodologically problematic to subscribe blindly to this maxim. Even if we accept that the work of art is globally governed by the external intentions of the artist who creates it, this does not make it legitimate to hermeneutically construct an unbroken intentional chain of cause and effect from singular fictional doing and behaviour right up to the level of the factual aesthetic activity which empirically manifests itself as a narrative speech act. As external observers, we believe we are entitled to reconstruct the logical project of a fictional intentional activity, despite the fact that it may well appear completely differently from the internal perspective of a fictional character, who may see it as a mere coincidence, a consciously intended result, or even the unintended effect of his own doings, to mention but a

1.2 Action: From Word to Concept

41

few possibilities. While it is heuristically legitimate to respond by treating the activity of fictional agents as if we were observing it from a viewpoint insight the diegesis, this approach too is methodologically problematic. One reason for this is the trivial fact that, because neither the activity nor its agents are real, we cannot call the latter to account for what happens in order to verify our hypotheses about their motives. More important, however, is the fact that we become interpreters, and are thus logically confined to an external perspective, as soon as we move beyond the simple reconstruction of the causal and chronological order of events and hermeneutically enquire about their meaning. Every element of perception will always appear as seen from the perspective of an overarching synthetic framework. Like the empirical author, the recipient does not just exist outside the fictional events in an ontological sense; he also acts logically above them, (re)constructing and assembling them for his own purposes by means of propositions and judgements which assume the idea of the synthetic mythos required by aesthetic convention. The problem can be seen as analogous to that which we encounter when observing the empirical behaviour of real human beings and which prompted Bourdieu (1979) to introduce the concept of habitus into sociology and ethnology. He points out that, as far as empirical human behaviour is concerned, we must always be careful not to confuse the logic of description with the logic of the thing described. Just because an external and culturally distanced observer can reconstruct a behaviour as a (socio)logical project of intentional action this does not necessarily mean that the behaviour has the same status in the eyes of those who are members of the culture and act habitually according to its rules. They may see it simply as a necessary consequence, as an enactment of what was always contained and anticipated in the present in which they live. For Bourdieu, this corresponds to Husserl’s distinction between different ways in which humans perceive time. Bourdieu describes how Husserl distinguishes a relationship with the future which one can call project and which posits the future as future, that is, as a possibility constituted as a possibility, which can therefore become reality or not become reality, and, in contrast to this, another relationship with the future which Husserl calls protension or pre-perceptive anticipation. Bourdieu 1998:145 (my translation; italics in the original)

As literary theorists, we cannot even begin to ask how far the singular behaviour of fictional agents constitutes a protension in the context of their world—methodological necessity forces us to describe their behaviour as a project. Nonetheless, it is still relevant to ask ourselves to what extent our own acts of description are protensionally determined in the sense

42

Part 1 Concepts of Action

of surreptitiously orientating themselves around a model of the synthetic complex action. A second important difference between narratology and poetics lies in the way they understand the concept of the event. When we talk about an event from the Aristotelian standpoint (i.e. in a discussion of the globally intended meaning and aesthetic effect of a work), we treat it as something which has a normatively postulated function as an element of the mythos. There is no need to distinguish analytically between static states, arbitrary events, intentionally caused transformations, and so on. The distinction is, however, vitally important in literary theory, particularly in constructivist models of reception, which are designed to elucidate how readers (re)construct the semantic machinery of a complex action. It is all the more important because it has never been sufficiently investigated. As long as over twenty-five years ago, the action theorist Waltraud Brennenstuhl commented on the disconcerting fact that an action as understood by literary criticism can be an action in the sense of the theory of action, or made up of several such actions, but it can also be a mixed sequence of actions, non-actions, natural events, social events, historical events, fateful occurrences, happy coincidences; in short, an action can be constituted by ‘things’ that can ‘happen’, things that have an aspect of eventfulness in some way or another. Brennenstuhl 1975:3 (my translation)

This distinctly confusing multitude of event-like phenomena which literary critics have traditionally subsumed under the concept of action is in urgent need of systematization. If nothing else, we must, I would argue, distinguish three key meanings of the term ‘action’ in literary criticism. Action I: Singular Agential Activity Seen from the perspective of reference, as a quasi-perception of fictional objects which appear in changing states of affairs, a narrated action initially appears to be a depiction of events which have been brought about by individual agents. We understand each such event as the result of the intentionally governed doing of a fictional agent who behaves in a certain way or plans and eventually enacts one such doing or another. At this most basic level, then, it seems quite natural to project the criterion of intentionality, which is of the utmost importance when we talk about action in everyday contexts, onto the aesthetic domain of fiction. In practice, however, the intentionality criterion, which is assumed to apply to the commonsense notion of activity and has special relevance to the explanation of the concept in analytical philosophy, has a consider-

1.2 Action: From Word to Concept

43

ably limited role in the naive and critical interpretation of literary texts, even when all we are trying to do is explain the doings and behaviour of a single fictional character as an instance of activity. We frequently treat the behaviour and doings of a fictional agent as actions in cases where we ourselves would certainly not want to be held legally responsible for them as the consequences of consciously made decisions.46 Literary criticism has found it difficult, if not downright impossible, to arrive at a clear-cut distinction between intentional activity, ordinary behaviour, and the particularly problematic case of a fictional agent’s failure to act. One reason for this is that only one of the many possible narrative perspectives, that of an authorial narrator who gives direct insight into the psyche of his characters, can provide us with reasonably reliable information about an agent’s motives and intentions. To make matters worse, we soon realize that such explicitly narrated motives and intentions frequently fail to tell the entire truth about a character. More often than not, they turn out to be part of a carefully planned strategy of information disclosure which characterizes the voice of the narrator and qualifies what he narrates. 46

This is particularly obvious if, as a methodological counterexample, we examine the legal practice of reconstructing a person’s subjective responsibility for his actions. The legitimacy of this practice depends on the unambiguous definition of the relevant terms. Berkemann (1984) has shown that the theoretical concept of action in jurisprudence has little in common with the ordinary descriptive term, which refers to any instance of observable behaviour. The legal concept does not postulate intentionality as an absolute criterion and an objective property; instead, it treats it as a relational term which becomes applicable only when observed behaviour, a hypothetical guiding intention, and an externally based normative evaluation of that behaviour can be connected with one another. This is the necessary condition for moving from the theoretical concept of action to the classificatory concept on which any normative system of action types must be based. In this respect, the legal concept of action turns out to be affected by a tension not unlike that which characterizes the aesthetic concept. On the one hand, we need to take into account the perspective of the agent, whose intentions, cognitive state, and evaluation of situational circumstances must be reconstructed in order for us to establish whether the objective outcome of a doing can subjectively be classified as an action for which responsibility should be accepted. On the other hand, we must also consider the external perspective of society, which classifies individual actions in terms of an action typology in order to establish regularity and predictability, irrespective of the guilt and intention which may be associated with actual individual instances of activity. However, the problematic nature of all intentional concepts of action is not just due to the relative nature of the criteria by which another person’s behaviour can objectively be described as intentional. Equally problematic are the questions of how extensive any such intention is in real terms—where to draw the boundary between separate individual activities—and the extent to and the conditions under which such individual activities should be regarded as constituting a superior complex action, in which case it is of course no longer legitimate to hold an individual legally responsible for the individual component actions (in the German legal literature, this problem is known as ‘Idealkonkurrenz’ and has been discussed for at least fifty years, as documented in contributions ranging from Honig 1925 to Puppe 1975).

44

Part 1 Concepts of Action

No narrator, not even the most realistic one, can present every possible piece of information; by definition, therefore, the information that we do receive is selective and partial. From this it follows that the narrated behaviour and the narrated activities of fictional characters are no more than partial representations. Even in the most promising case, that of the autobiographical first-person narrator, we remain dependent on our ability to complement the explicitly narrated facts by extrapolating from them or supplementing them by referring to our world knowledge. Time and time again, we find ourselves engaged in circular reasoning as we form hypotheses about a character’s overall disposition as it develops with each successive event so that we can identify a potential intention behind each and every instance of the character’s activity that we perceive, or, more accurately, is narrated to us. Similarly, the inadequate transparency of intentions has a problematic effect on any attempt to reconstruct the intentional foundations of behaviour in the real world. Mental facts outside our own consciousness can only be accessed indirectly by drawing conclusions from empirically observed events. When reconstructing human behaviour, we must therefore use a methodological framework completely different from that used when reconstructing sequences of physical events.47 The explanations of action suggested by the social sciences (and, more generally, all methodologically comparable hermeneutically orientated explanations) are primarily historical, classificatory rationalizations, even if they superficially claim to be truly logical explanations. At heart, they are ideological in nature, for they presuppose that the agents to whom they ascribe intentions and behaviour have acted rationally. This, however, is not an objectively verifiable assumption. It is an unavoidable discursive convention without which there could be no meaningful communication (i.e. communication based on understanding) about empirical human behaviour at all. Moreover, only rarely does a single principle control how we reconstruct the logic behind fictional events as the intentional action of a particular 47

Pettit (1982) outlines the differing objectives which distinguish the reconstruction of facts in hermeneutics from that in the natural sciences. The models developed by the natural sciences generally serve a predictive purpose and can be verified empirically. The epistemological or ontic status of the events observed is not in question. Such models are therefore characterized solely by the abstract categories of temporal succession and causal relation. In contrast, the reconstructions of the hermeneutics and the social sciences presuppose an intentional concept of action and a theory of rational man. When human behaviour becomes the object of empirical observation, it is by default declared to be intentional; only subsequently are the concrete subjective aims and motives that fit in with this axiom identified. An observed event is thereby elevated to the status of a legitimate object of social discourse, and only then, as an action, can it be discussed in any meaningful way.

1.2 Action: From Word to Concept

45

agent. In reality, we almost always build such arguments using a complex combination of reasons which can be causal, intentional, psychological, and so on. All such reasons are acceptable providing that they serve the overriding purpose of making the behaviour of a fictional character understandable, interpretable, and thus accessible to rational discourse. Action II: Complex (Multi-agential) Sequence of Events The second concept of action employed by literary criticism refers to something more than the behaviour of the individual agents represented in a narrative: the entire complex of interconnected behaviour and events which contribute to the dynamic nature of the narrated world. Since classical times, the multi-agential narrative of action in this wider sense has typically revolved around the tension between two factors: first, intentional doing which can be identified as the action of a particular agent; second, external factors and event sequences which are beyond the individual’s control in that they are either the intentional behaviour of other agents or accidental occurrences which (seen from inside the fictional world) can be interpreted only as contingent events which lack any intention whatsoever. Action II is thus equivalent to the concept of mythos in Aristotelian poetics. As far as the intentionality criterion is concerned, this concept of action often appears to be defined ex negativo. The failure of an agent’s behavioural planning in a fictional world, the clash between intentional doing and external factors, is elevated to the status of an aesthetic metaevent in the real world, in which cultural convention dictates that all aesthetic products are meaningful and a sign of an artist’s intention to express himself. Seen from the perspective of the fictional agents, events negate the intentions of their agents, who are demoted to the role of mere patients. At the meta-level of the aesthetic activity in which real author and real reader are involved, this negation is considered meaningful and thus simultaneously preserved and deproblematized. Action III: Discursive Meta-Activity The third concept of action which is often encountered in literary criticism refers to the empirical activity of the real producers and recipients of literature. In formal terms, Action I and Action II deal with the object domain of fictional reference: the symbolic representation of individual intentional behaviour and complex, interconnected happenings in a fictional world. By definition, both are equally well the concern of mimesis. Our third kind of action is very different in this respect.

46

Part 1 Concepts of Action

One of the consequences of the linguistic revolution of the twentieth century has been the introduction of ideas based on speech act theory into modern literary criticism. Narratology and the study of rhetoric have reformulated the definition of action to make it a concept that is no longer restricted to denoting abstract (or fictional) objects but can also describe the process by which literature itself comes into being. The act of narration itself is now seen as what Schmid has called an ‘action of narration’ (‘Erzählhandlung’; my translation), and Stierle has even coined the phrase of the ‘text as action’ (‘Text als Handlung’; my translation), which equates the entire text with a complex speech act. In both cases, we are referring to actions which, in an ontological and logical sense, take place not within but over the realm of the fictional world. Reader-response theory and the cognitive revolution have increased our understanding of action in the context of the aesthetics of production and rhetoric by adding to it the dimensions of communication, reception, and cognition as relevant frameworks for theoretically explaining the concept. Consequently, the act of reception also falls under the widened notion of what constitute instances of literary activity. Empirical literary theory has argued for a complete and radical reorientation of literary criticism per se, which it suggests should abandon the traditional philological paradigm and adopt that of the social sciences. The redefinition of action plays a vital role here: literary criticism as conceptualized by the Siegen School and others is no longer the study of texts; it is the study of the empirical human actions which are triggered by and revolve around texts. However, it seems to me that action at this point becomes too loosely defined to be of viable use to literary criticism—for example, the Siegen model of the action levels of the literary system focuses almost entirely on the real (empirical and historical) human actions that take texts as their object in one way or another; the idea of action as a referential object or fictional construct no longer has any role to play. We can now summarize the conclusions of our brief survey. Action I is the singular, agential activity which we explain as a mimetic representation of the intentional behaviour of a fictional agent. Action II refers to the mythos in the Aristotelian sense, that is, the complex and multi-agential sequence of events described in a narrative. From the point of view of literary criticism, this notion subsumes instances of intentional agential behaviour and non-intentional events and states in the fictional world. Action I and Action II together form the domain to which we referred above as object-action. Action III, in contrast, refers to all instances of discursive meta-activity, that is, to narrative speech acts, acts of reception, and the entire range of para-, intra-, and metatextual human behaviour that is characteristic of

1.2 Action: From Word to Concept

47

literate cultures. It is beyond the scope of our study to cover the entire semantic field covered by all three definitions of action; instead, we shall concentrate on the domain of Action I and Action II. Action III, and thus any uses of the concept which draw on paradigms such as speech act theory, discourse theory, and social studies, are only of secondary importance for our purposes.48 We are required to focus on Action I and Action II because the theory which we shall develop is primarily intended to model the interaction between the phenomenologically definable features of the text and the habitually and logically determined constructive acts of the recipient. These constructive acts take place at the basic level; they can, and indeed must, therefore be modelled as such before we can even begin to consider the historical and social factors which affect how we read action, just as they affect all our other empirical activities. Although we still lack a theoretical definition of action as a concept of narratology and reception theory, it should now be clear what is not necessary or sufficient for such a definition: the intentionality criterion central to the anthropocentric commonsense notion of action. Any student of literature who talks about the action of a fictional narrative text is making a statement that can be reduced neither to the identification of a fictional agent nor to the identification of a change of state in the fictional world with a necessarily intentional cause. This does not, however, mean that we should ignore all intentionalist approaches to the concept of action (e.g. the findings of analytical philosophy). Let us recall our key question: how and under what narratively defined constraints do recipients generate mental representations of a symbolically represented combination of events which can be classified as a literary action (Action I and II)? We can safely assume that the constructive acts of the recipient with which we are concerned include both analytic and synthetic judgements about fictional events. The philosophy of action, therefore, presents us with at least three profitable starting points for further discussion: the analytic, transcendental, and constructivist definitions of the concept of action.

48

The close relationship between intentional concepts of action and models derived from speech act and discourse theory is best demonstrated in the work of van Dijk. They are linked by the premise that ‘narrative discourse may be conceived of as a form of natural action description’ (van Dijk 1976:287). To my mind, such a description of ‘natural action’ presupposes a culturally determined, intentional concept of action.

That which gives the extraordinary firmness to our belief in causality is not the great habit of seeing one occurrence following another but our inability to interpret events otherwise than as events caused by intentions. It is belief in the living and thinking, as the only effective force—in will, in intention—it is belief that every event is a deed, that every deed presupposes a doer, it is belief in the “subject.” Nietzsche 1968:295

1.3 Philosophical Definitions of Action The Commonsense Concept of Action What is an action? A sunrise, for example is not; turning on the light in a room, on the other hand, certainly is. Generally speaking, an event can be considered an action when it has manifested itself in an objective result and been caused by the intentional behaviour or doing of a rationally planning (i.e. anthropomorphous) agent.49 It is of little or no importance whether the actual effect caused is identical to the one originally intended: I might decide to turn on the light, flick the switch, and thereby unintentionally cause the bulb to blow. If this happens, the action I have performed is blowing the bulb rather than turning on the light—but it is not the outcome of what I have done that counts. The original motivation behind what I have done is what allows my deed to be attributed to me as an action that I have performed. From an intuitive perspective, this definition of action seems perfectly plausible, particularly given the fact that rationality and individuality are the definitive defining characteristics of the place of humans in modern society. From a philosophical point of view, however, this commonsense definition exposes us to a considerable danger: it is easily misunderstood as stating that intention is an ontological prerequisite of action rather than a conventionally determined attribute of it. It is this erroneous post hoc ergo propter hoc argumentation that Nietzsche has in mind when he argues that the concept of action has been distorted by a naive belief that only living and thinking things can be agents, a conviction inseparable from the 49

Topitsch (1975) analyses the world view which is derived from the model of the artefact and which uses analogy to interpret and explain empirical reality as an intentionally constructed reality. Aristotle’s practical syllogism is the most absolute expression ever of the anthropocentric intentional model of action. This may explain why references to the Aristotelian model are so important in structuralist and semiotic theories of action and narration, in particular those of Greimas, which we shall discuss in more detail in chapter 1.4 below.

1.3 Philosophical Definitions of Action

49

more abstract conviction that causality is an ontological principle. Taking Nietzsche’s argument to its logical conclusion, we have no alternative but to acknowledge that the all too familiar uncritical use of the intentional concept of action is only one manifestation of a far-reaching universal intentionalism that requires each and every observed happening to be interpreted as of causal origin, each and every instance of causality as the expression of an intention, and each and every intention as the expression of a consciousness. The end result is that our world must be devoid of contingency, that all happenings and all events in it must be read as the acts of conscious beings. Nietzsche’s cultural criticism identifies very real dangers inherent in the commonsense understanding of action. However, he provides few conceptual starting points from which to develop a more adequate definition of the concept. For a start, he never even mentions the word ‘action’; instead, he employs a number of circumlocutions such as ‘occurrence’, ‘following another’, ‘events’, ‘causality’, ‘intentions’, ‘effect’, ‘will’, and ‘interpretation’.50 The terms in this list illustrate what may well be a symptomatic unwillingness to confront the heart of the matter. Nietzsche’s aphorism is a cutting reflection on the problematic nature of the concept of action, but it is of no use as a philosophical guide to how we might go on to investigate the issue and clarify it in more detail. The present chapter of our study will investigate a number of different philosophical approaches to the concept of action. Most of these methods focus on action in the sense of a phenomenon that occurs in the real world rather than as an aesthetic category. We shall not treat these two contrasting interpretations as fundamentally incompatible; in fact, our discussion aims to explore how we might derive the aesthetic concept of action from definitions of the pragmatic, empirical concept. Analytical Definitions of Activity and Action The Representational Content of Activity The approach to the definition of action in analytical philosophy does not focus on the phenomenological aspect of the problem (what is an action, and by what features can it be identified?); instead, it considers the conditions under which it is correct to apply the term ‘action’ and why they 50

In the original German the terms used by Nietzsche are: ‘Vorgang’ (‘occurrence’), ‘Hintereinander’ (‘following another’), ‘Geschehen’ (‘events’), ‘Kausalität’ (‘causality’), ‘Absicht’ (‘intentions’), ‘Wirkung’ (‘effective force’), ‘Wille’ (‘will’), and ‘Interpretation’ (‘interpretation’). See Nietzsche 1977, 3:501.

50

Part 1 Concepts of Action

permit us to do so.51 In this theoretical framework, calling an occurrence or happening an action means assigning a particular kind of description to an empirical event; the description in question explains the event as the consequence of the intentional doing of an agent. Thus, Davidson (1971: 5f.) argues that a happening is an action or behaviour if and when it can be described in terms of an intention. Intentionality is the key criterion for distinguishing between two different ways of interpreting events, the ‘reconstruction of an action-based causality of actions and [that of] a locomotive causality of events’ (Merkel 1983:116; my translation). There are at least two reasons why it is hard to apply this intentionalist approach to action as an analytical concept to the aesthetic object domain and the systematic categories of Action I and II as described above. First, the analytical concept of action primarily concerns the connection between empirical events and agents, not the more particular, far more complicated case of the link between fictional agents and occurrences in a fictional world. By definition and aesthetic convention, fictional agents are not ontological but functional entities: whatever they do and however they behave must ultimately be evaluated as serving the purpose of communicating a message to the recipient. Thus, readers assume that all fictional behaviour and events to be governed by the ontologically foreign intentions of a real agent who is external to the fictional world. These aesthetic intentions are located in the real world and are as unfathomable to the fictional persons as the cosmological intentions (assuming that there are any) revealed in the laws of nature are to us. In other words, because of the ontological hiatus dividing the domains of effect and intention, the latter must always be inaccessible to a being that exists in the universe of effect. Second, in addition to the inaccessibility of intentions due to the ontological and semiotic status of the fictional world, there is a related methodological problem. In analytical philosophy, action is conceptualized primarily from the point of view of Selbstverstehen, our understanding of our own behaviour.52 This approach is particularly obvious in Danto’s 51

52

Perhaps the most concise analytical definition is Rescher’s ‘canonical description of an action’ (Rescher 1967:215). Rescher lists a series of questions which, taken together, capture what we should require from the sensible use of the term in a commonsense context: the description of an anthropocentric happening in as comprehensive a way as possible. This description centres on an active agent whose behaviour is intentionally determined and contextually evaluated. In this respect, Rescher distinguishes between the opposing principles of the causal (cause-orientated) and the intentional (intentionorientated) explanations of action. The distinction between Selbstverstehen and Fremdverstehen (between understanding one’s own behaviour and another’s behaviour) was introduced into the philosophical debate by Alfred Schütz in order to distinguish the two basic methodological perspectives from which we perceive action. We shall consider Schütz’s argument in more detail below.

1.3 Philosophical Definitions of Action

51

work, which is generally concerned with how to formulate adequate descriptions of one’s own subjective behaviour. This approach, however, is irreconcilable with our approach to interpreting foreign behaviour, including the particular case of fictional foreign behaviour. When interpreting such behaviour, the attribution of intention to an agent can only be justified or refuted circumstantially; we cannot derive our evidence from direct insight into the intentions of the agent concerned, and it can only take the form of inferences drawn from the context in which a particular event took place and the consequences that followed it. In other words, this forces us to argue that an action has taken place by using synthetic rather than analytic arguments. Despite the two facts which make the analytical concept of action incompatible with our aesthetic object domain, the manner in which it is defined is, at least in part, functional, and therefore relevant for our purposes. As a first step in exploiting this relevance, we shall examine action as it appears from the narrower methodological perspective of Selbstverstehen. If we accept it as a fact that the criterion of intentionality can only be properly tested in private, self-reflexive situations, why do we go to the trouble of distinguishing two types of our own subjective behaviour when we discuss it, one intentional and the other governed by contingency? What is the purpose of this distinction if its empirical basis is by definition incommunicable? The answer is provided by Wittgenstein, who points out that we do not describe our own behaviour as an action (as opposed to unintentional behaviour) in order to make a statement which attributes a certain essential quality to our past doing. Rather, we describe our own behaviour as an action in order to initiate what one might call a linguistic game. In other words, our attribution of intention, and hence of action, to ourselves serves a communicative function, irrespective of the transparency of the attributed intention. When we characterize our past behaviour as intentional in a conversation, we do so in order to let our conversation partner learn about our reactions to what happened, not necessarily about our motives for making it happen (Wittgenstein 1958: 167). Thus, Wittgenstein concludes in the Logical Investigations: Why do I want to tell him about an intention too, as well as telling him what I did?—Not because the intention was also something which was going on at that time. But because I want to tell him something about myself, which goes beyond what happened at that time.—I reveal to him something of myself when I tell him what I was going to do. Wittgenstein 1958:167, § 659 (emphasis in the original)

In pointing out this communicative function in the way we attribute intentions to ourselves, Wittgenstein transcends the methodological

52

Part 1 Concepts of Action

paradigm of Selbstverstehen. However, the above quotation also makes clear that, for Wittgenstein, the content matter communicated cannot be reduced to a historical intention but concerns a fragment of information whose validity is untouched by historical contingencies: a revelation of the agent’s inner self to the addressee. The idea that we do not attribute reasons to events simply in order to reconstruct mechanical causality is also present in Danto’s philosophy of action. Like Wittgenstein, Danto attaches considerable importance to the fact that reasons have an anticipatory character and transcend the present by binding the past and the future together, whereas causes are disinterested in the future, which is actually invisible to them. This leaves us with the question of how the relationship between reasons and the future functions. Danto believes that human activity is fundamentally representational, that it is triggered by and accompanied by ideas of how we would like the world to appear as a consequence of our deeds (Danto 1976:12). Reasons refer to a content that both logically and temporally transcends them, and this reference to an anticipated and imagined content is how they create actions. Actions therefore have a semantic as well as pragmatic function. They supply an ex post facto argument which proves that the content, originally no more than imagined and anticipated, is actually a real fact and a true idea. For Danto, a performed action occurs when an idea about how the world should be is converted into reality by the influence of the person in whom it originated. It follows from this that we can only describe an empirical event as an action if we perceive it as having reached closure. Conversely, speaking of an action is equivalent to knowing, declaring, or positing the representational endpoint of an event. The stimulus (i.e. the anticipated result) that triggers a process of activity is exhausted the moment that we begin talking about its product, an action. In effect, Danto is saying that the word ‘action’ can only be used meaningfully if it is used to refer to instances of singular and completed activity. It does not make sense to talk about an incomplete process of activity as an action. Furthermore, and this is of particular importance in the context of literary criticism, it does not make sense to use ‘action’ to describe a complex of fictional happenings and events that progressively brings itself into being. In Danto’s theory, the only true action in literature is an author’s intentional speech act which has been completed and therefore shows that the author’s image of how the real world should appear as a result of his behaviour has become reality as a new aesthetic artefact in the world. If we do as Danto suggests and trace the concept of intention back to the concept of rationale, and the concept of rationale back to the concept of an imaginary idea realized by agential doing, we find that here, in the representationalist core of the analytical concept, there is a formal analogy

1.3 Philosophical Definitions of Action

53

with the aesthetic concept of action. However, the aesthetic concept of action is not concerned with providing an inductive and analytical description of an individual doing performed with the aim of realizing an agent’s subjective idea of how the world should be. Instead, the interest of the aesthetic concept of action is directed at a complex idea which contains a possible world and all its agents, in other words, an Aristotelian mythos. In order for this mythos to represent an aesthetically plausible systematic whole, its component actions must be semantically determined from a point above the fictional agents and their individual ideas of what their deeds should achieve. This is where the narrator and the author come into play. At the same time, however, aesthetic plausibility is endangered when the mythos appears to be nothing but an effect caused by its author’s intentions. However, these requirements for the aesthetic plausibility of a mythos must be qualified by the caveat that the criteria are to be understood as historical rather than universal conditions. From the Analytical to the Systematic Concept of Action A deductive, systematic approach to the definition of action is not confined to such disciplines as literary criticism, history, sociology, and transcendental philosophy. As early as 1979, von Wright, originally one of the most committed proponents of the intentionalist approach, turned away from the intentionalist paradigm (von Wright 1979:422f.). In his opinion, neither the strictly intentional explanation of action (which refers exclusively to subjective reasons and intentions) nor the strictly causal counterexplanation (based solely on empirical causes) provides a convincing solution to the problem. In fact, von Wright came to consider the dichotomy of empirical cause and logical reason itself misleading (von Wright 1979:426). He observed that the explanations of actions are not based on singular empirical observations but constitute conceptual statements which refer to a wider network of interrelated facts. This is why he saw deduction rather than induction as the key principle that should underlie all inferences about the logic of an action: The criterion for validating statements such as: ‘A did x because he was ordered to do so’ is a kind of coherence of facts. The facts are: A did x, and he was ordered to do x. The criterion for validating the because-relation lies not in a relation, in a correlation between these two facts seen in isolation, but in a wider complex of facts. von Wright 1979:427 (my translation)

Von Wright therefore advocates a cautious revision of the intentional, analytical concept of action. The criterion of coherence which he postulates is ultimately a criterion of plausibility, and it should be applied relatively

54

Part 1 Concepts of Action

and with reference to an composite explanatory complex, not absolutely and to singular explanations of action. The discussion in Moya (1990) takes a further step towards the hermeneutic problem so important in literary criticism. For Moya, the philosophical problem of the concept of action is more than just a conceptual gap that can be filled by the discoveries of an analytical approach. It is the manifestation of a metaphysical problem that affects the human condition itself. Moya argues that we need to decide whether a subjective concept of action that presupposes the will and autonomy of human agents can possibly be upheld in the context of a world where the natural sciences hold sway and conceive of it as ‘a network of causally related events where no room can be found for agents, for beings capable of initiating new causal chains’ (Moya 1990:9). Danto had attempted to resolve this problem, which he saw as one of infinite regression rather than determinism, by introducing the idea of the basic action as the immutable fundamental instance of direct activity. Moya takes up this argument again but inverts it as he does so. For him, the paradigmatic, fully autonomous human action is not constituted by a freely made movement of the human body in which will and deed coincide; it is exemplified rather by the symbolic act, the meaningful action. Despite the fact that, here too, practical behaviour is determined by the human will, this type of action also creates a meaning that transcends its actual context. (Moya 1990:37ff). Meaningful actions are therefore the only actions which must by necessity be performed by agents; the results attributed to actions of this type can never be the outcome of contingent, non-intentional events, unlike spontaneous movements of the body, which can legitimately be misinterpreted as basic actions. What, then, does Moya have in mind when he talks about meaningful actions? I am thinking of…actions that can be said to have a meaning or a symbolic content, such as bidding in an auction, holding a lecture, voting, making a chess-move, signalling for a turn when driving or greeting a friend. Now, what is the difference between this kind of action and the examples of actions we have been analysing so far, such as raising one’s arm or killing someone? These latter actions have results, namely one’s arm raising or someone’s death, that, being happenings, can [also] occur with nobody’s raising his arm or killing someone. Moya 1990:38

Symbolic actions, meaningful actions, are therefore pure actions in the strictest sense. We do not recognize them as actions ex post facto on the basis of the results which they bring about; we identify them as actions in and as themselves (Moya 1990:39). They coincide with and are therefore identical with their own consequences; their apparent ‘results cannot, strictly speaking, occur without the action of which they are results be-

1.3 Philosophical Definitions of Action

55

ing performed’ (Moya 1990:38; emphasis in the original). According to Moya, the true principles of human action can be elucidated neither by the positivist reductionism of physical basic actions (bodily movements) nor by the causal derivation of events from causes. We can discover the principles of action only when we realize that true activity is defined by the fact that it transcends actual doing and that this transcendence carries meaning. At this point in his argument, Moya approaches Danto’s thesis of the representational nature of human action and Wittgenstein’s identification of the communicative purpose of self-attribution of intention. As we have seen, Danto treats transcendence of situation as tied to the phenomenon of intentionality, the pursuit of a goal. Moya likewise is of the opinion that ‘being intentional is the criterion of agency’ (Moya 1990:167). But how can we reliably determine when an intention is actually present? After all, it is easy to erroneously read intentionality into a number of forms of animal behaviour. We need a way to identify instances of behaviour which are necessarily accompanied by intention. This, according to Moya, is the case if and only if an intention occurs not only in the form of a premeditated intent but also in the form of a commitment to engage in subsequent activity after the present activity. Moya’s concept of action is defined by this mutually implicative ability to have significative function and enter into a future-orientated commitment. As far as the agent is concerned, this means that only a subject able to make a commitment to cause signification is in a position to perform an autonomous (pure) action. Conversely, it is legitimate to try to identify an agent who consciously brought about a perceived event if and only if that event is considered to be meaningful in the first place: Meaningful actions imply…the existence of subjects able to commit themselves to do things in the future. I shall contend that this notion of commitment is essential to our notion of agency and to the important sense of the distinction between actions and happenings. Moya 1990:46

How can this approach, which is basically concerned with an ethical definition, be applied in the context of literary criticism? The crucial point to consider is its fundamental re-evaluation of the intentionality criterion. The classic analytical definition of action postulates that the attribution of behaviour to an agent as his action can only take place when a motivation can be identified in the agent in question. Moya, on the other hand, proposes a distinction between action and happening which makes no reference to any such secondary motivational factors. For him, an action is defined by its semantic function, not by the particular way in which it comes about. Intentionality is a consequence, not a cause. Moreover, the meaning of an event which is observed and categorized as an action

56

Part 1 Concepts of Action

transcends its agent’s individual pragmatic aim of realizing a subjective representation of the world, an image of how the world ought to be. In Moya’s approach, meaning is restricted to the self-redefinition of the behaviourally active agent who makes a commitment. This restriction follows from Moya’s methodological focus on Selbstverstehen. If we broaden the perspective to include modes of Fremdverstehen, it is clear that acts of signification and acts of commitment, irrespective of their subjective content and intent, imply the existence of symbolic systems and normative systems respectively, both of which are composed, and constantly updated, intersubjectively. By introducing the category of meaning into the philosophical discussion, Moya presents us with a unique opportunity to mediate between the various concepts of action, agential and systematic, analytic and synthetic or transcendental. The methodological benefit is considerable, for it means that we no longer have to explain the characteristics which set actions apart from ordinary events in terms of qualities which by necessity can only be attributed to the agents of actions. Such explanations are particularly problematic in the context of literary criticism because of the factual inaccessibility of the motivations of fictional characters which we described earlier. This is no longer a problem in our new functional and semantic approach, which defines action not by means of an analytically identified motivational or logical condition but by means of the interpretively attributed meaning that distinguishes events from actions. The crucial factor is the external realization of a semiotic function by the recipient or spectator, not the reconstruction of the internal (psycho)logical condition of the agent.53 As soon as we attribute a semiotic quality to an observed change in state of affairs, be it an event, happening, behaviour, or activity, we engage in an assignment of values that begins to transcend observation. In asking what the thing that happened might mean, we are no longer involved in the analytical description of a process and have already begun to evaluate its semantic value. Seen in this way, every happening that we consider sufficiently semantically marked is by definition an action. This, of course, begs the question of what we mean by ‘sufficiently’. We can rephrase the question as follows: how do we identify the qualitative frame of reference which is required by the semantic evaluation of events and happenings? We are now in a position to formulate a tentative definition of action as any happening that can be interpreted as meaningful, irrespective of whether we mean ‘meaningful’ in the context of an individual actor or a systematic whole in the sense of the Aristotelian mythos. Broad and 53

Danto mentions the semantic function of activity, but he identifies it with the way that our behaviour realizes the representation which guides an action.

1.3 Philosophical Definitions of Action

57

unspecific as it may be, this preliminary (re)definition sheds new light on the deeper epistemological problem associated with the concept of human action. Our definition is not beset by the danger of interpreting all happenings as intentionally caused. It heeds Nietzsche’s warning and resists the temptation to treat the world as a place governed by universal intentionality. Instead, however, we face the menace of universal semiosis, a menace rooted in our inability to accept the world as a place that is devoid of meaning by its very nature. It would seem that we have the unenviable task of choosing between determinism and nihilism, two equally unappealing solutions to our problem. But there is another possibility: we can think of human action not simply in terms of individual acts which are representational in Danto’s sense and pursue short-term tactical goals, but as part of a long-term strategy which introduces meaning into the world. The universal semantic function of human activity is the focus of transcendental definitions of action, and it is to them that we now turn. Transcendental Definitions of Action Action as the Exposure of the Subject From Danto to Moya, analytical philosophy has discussed a concept of action which applies to singular, spontaneous instances of human activity rather than large-scale complexes of connected events. In the context of an analytical explanation of the concept of action, it may be acceptable to ignore the elaborate strategic, psychological, and motivational calculations with which we balance decision and desire when planning complex extended sequences of several component actions. In the context of an explanation of aesthetic action, however, it is clearly not permissible to neglect this important aspect of the problem. Moya, and even more so Danto, are guilty of such an omission, as is particularly apparent if we, like Bubner, read such approaches against the analytical grain. He identifies a distinct parallel between Danto’s idea of the basic action and the Aristotelian concept of praxis, which is, in contrast to pragmatic goal-orientated poesis, the true entelechial realization of human existence. Bubner therefore derives the analytical distinction between basic and non-basic actions from Aristotle’s ontological distinction between poesis and praxis. He argues that, whereas chains of non-basic actions in the sense of poesis are teleologically constrained by the outcome they are meant to produce, basic actions are analogous to instances of praxis, which is a process that is its own completion in that its aim is realized in the act. The attainment of practical goals coincides with this process so that no produced object remains after the closure of the praxis. Bubner 1976:70 (my translation)

58

Part 1 Concepts of Action

Unlike representationalism, this approach does not understand action as the realization of an agent’s idea of how the world should be. It is an essentialist model and sees action as the exposure and existential realization of the acting person’s existence in the world. It is of particular interest to literary criticism and fiction theory not least because it is clearly extensible to the existence of fictional agents. Like real agents, they too can only be existentially realized, albeit in a mimetic manner, if their behaviour and activity have been consigned to the past and thus made narratable. The entelechial interpretation of the completion of action as the completion of life is most prominent in the work of Hanna Arendt, and the discussion that follows is extensively based on Arendt (1979). She defines activity as a second start analogous to human birth, a new beginning that sets an open and dynamic process in motion. The end of the process can only be theoretically anticipated as the death of the subject and is otherwise hidden and unpredictable. Although the ultimate goal of the process cannot be identified, the process itself can be functionally defined: activity effects the ‘exposure’ (‘Enthüllung’; my translation) of the subject, which reveals itself to other subjects through its own behaviour and communication. Only after its eventual conclusion does the process acquire its ultimate meaning when all its constituent moments are integrated in the form of a narrated ‘Geschichte’ (‘story’ or ‘history’). This history of a life exposes a ‘who’, the agent’s subject, and it does so even though we cannot possibly identify an author who could have composed the narrated process on the basis of a plan or model (1979:29). The meaningful whole of an actor’s biography is necessarily concealed from the actor himself and will only reveal itself to a narrator who is not involved in it. This is why Arendt argues that transparent sequences of human activity executed according to a concrete plan and with a concrete and conscious goal in mind are instances not of activity but of production. Such production is of an instrumental nature; that is to say, it is determined by a particular relation between means and aims. True activity, on the other hand, serves the abstract global goal of affirming the existence of the actor by exposing him. This makes it perfectly legitimate for Arendt to take the domain of intentional doing and behaviour into account even though her specific concept of action is defined in transcendental terms.54 In her view, intentional doing should not in itself be classified as activity, for it refers only to extracts from the global event sequence which are

54

To avoid any confusion, we should note Arendt’s somewhat idiosyncratic terminology. She refers to the dualism of ‘Tun/Handeln’ (‘doing/activity’) in the sense of Schütz’s distinction of ‘Handeln/Handlung’ (‘activity/action’). The concept of ‘Handlung’ (‘action’) as such is not distinguished in Arendt’s approach.

1.3 Philosophical Definitions of Action

59

clearly delineated chronologically and causally and which represent the greatest level of complexity which the agent can grasp. In contrast to true activity, our intentional doing makes no reference to a complex biography beyond its own limits; its potential is exhausted once it has reached its practical goal, which it does without bringing the existence of its subject to an end. Arendt thus defines the concept of activity in two ways: as a process by which the subject exposes itself, and, in a transcendental or final manner, as the completed object of a biography which can only be narrated posthumously and from an external perspective. It is of fundamental importance that the agent himself has no insight into this overarching activity; for him, even the doings which are completed during his lifetime are no more than plans of production enacted in the sense of Aristotelian poesis. The transcendental dimension of our activity is beyond the reach of the imagination, which can neither grasp the absolute openness of the process nor comprehend the absence of an authorial instance responsible for its events. Since classical times, this double void, in which we see the quandary of activity and human existence itself, has resulted either in the hypostatization of a historical or divine subject or in the reduction of the concept of action to the idea of instrumental production. Arendt’s approach is relevant to our study in several respects. Admittedly, it does not clarify the conditions under which a happening can be identified as the activity of an agent. However, it places us in a position where, for the first time, we can distinguish the different existential positions from which such a happening can be perceived and described. Basic intentional doing, the production of a result according to a plan, and global activity which exposes the self all refer to one and the same anthropomorphic happening. Within this common domain, they cover subfields of various logical and temporal extension. From Arendt’s perspective, the distinction between Selbstverstehen and Fremdverstehen amounts to more than just the difference between the dispositions of a strategically orientated agent and an epistemologically orientated spectator. The epistemic inaccessibility of our own activity is nothing short of an ontological limit that defines existence. This calls to mind Heidegger’s postulate of the ‘Sein zum Tode’ that structures time yet can only be grasped from a position beyond time and therefore beyond all existence. Activity and Action: The Dialectic of Subjective and Objective Exposure One of the conclusions to be drawn from Arendt’s work is that we should no longer treat the two basic modes of interpreting agential happenings, Selbstverstehen and Fremdverstehen, as diametrically opposed methodological approaches. Instead, the two terms should be seen as dialectically

60

Part 1 Concepts of Action

related. The mutual connection between the subjective and objective interpretation of a sequence of happenings has an equivalent in the concept of action itself: as far as agential happenings are concerned, the activity/action dichotomy does not present us with an either/or choice—both concepts are interrelated. The two terms are logically linked, and this logical pairing lies behind the differing methods of the two fundamental ways of describing happenings. This raises the question of what activity means in relation to action and vice versa. Alfred Schütz’s distinction between Selbstverstehen and Fremdverstehen is crucial in this regard in that it implies a systematic difference between the terms ‘activity’ and ‘action’ as notions to be defined in terms of one another. This sets his work apart from most philosophical treatments of action which either use the terms synonymously or distinguish them in a vague and inconsistent manner. Only when it is explicitly pointed out that the notions of activity and action are not identical do we realize that we are concerned with a dualism between two concepts, each of which represents an equally possible way of describing agential happenings. The dualism consists of a concept concerning a process on the one hand and a concept concerning a project on the other; it is, in fact, the same dualism that we encountered in our brief etymological discussion. We shall soon see that the epistemic dialectic of Selbstverstehen/Fremdverstehen leads into a hermeneutic dialectic of two complementary concepts of meaning, one of which is concerned with processes and the future and the other with finality and the past. Schütz’s definition of the concept of meaning is based on Husserl and, more importantly, on Weber’s sociology, but despite these influences he substantially revises the work of his predecessors. Weber points to the connection between human activity and the constitution of meaning, but Schütz criticizes him for focusing solely on the perspective of the acting subject and thus on the dimension of the subjective meaning of social behaviour (i.e. the meaning perceived by the actor or actors themselves) (Schütz 1974:14). The main drawback in Weber’s approach, Schütz argues, is his disregard for the difference between activity and action at the object level and the difference between Selbstverstehen and Fremdverstehen at the methodological level: Weber makes no distinction between activity as a process and completed action, between the sense of creating and the sense of what is created, between the sense of one’s own activity and that of others, between the sense of one’s own experiences and those of others—no distinction, ultimately, between understanding oneself and understanding others. Schütz 1974:15 (my translation)

1.3 Philosophical Definitions of Action

61

In his Der sinnhafte Aufbau der sozialen Welt, Schütz attempts to deal with both these problems by arguing that meaningful phenomena in the social world vary intersubjectively because they are interpretive phenomena which are thetic in nature (Schütz 1974:16). What is needed, therefore, is an analysis of meaning as a process rather than as a static object. In this context, we cannot distinguish between activity and behaviour by saying that one creates meaning while the other does not. The distinction between the two lies instead in the ways in which they create meaning (Schütz 1974: 28). Behaviour is a basic-level process in which we experience meaning spontaneously but not consciously. Only retrospectively can we become consciously aware of the meaning which has already manifested itself in spontaneous activity: Only when instances of behaviour, or at least their initial phases if they are divided into phases, have ended…can they be set apart from the sea of all other mental experiences as clearly distinct and retrospectively identifiable. Schütz 1974:73 (my translation)

The transition from behaviour to experience is made possible by ability of our internal consciousness of time to adopt a retrospective perspective. This ability also enables us to move from the notion of activity to that of action. An activity is thus a behaviour which can be described as planned and designed. From the perspective of Selbstverstehen, that of the agent who subjectively interprets his own activity, the design takes the form of an action which is counterfactually imagined to be already completed. Schütz refers to this counterfactual anticipation as a way of thinking modo futuri exacti. From the objective perspective of Fremdverstehen, on the other hand, we must accept that the unity of activity is a fundamentally subjective idea—assuming that the observer does not simply assert an action unit which cannot be objectively identified (Schütz 1974:82).55 In Schütz, the meaning of an activity is its predesigned action as seen from the subjective perspective of planned protension (Schütz 1974:79). Only action, which is understood as already completed, not activity, can be logically pre-arranged and temporally anticipated in this way:

55

This restriction allows Schütz, to some extent at least, to circumvent the problem which we noted above in our discussion of Danto’s representationalist approach: Selbstverstehen and Fremdverstehen can assign different representations to an observed behaviour, but the logical stringency of an externally assigned representation can only be evaluated with reference to the systematic context. This allows competing interpretations of a happening to coexist.

62

Part 1 Concepts of Action

What is designed (‘pre-remembered’) is not the activity that performs something step by step, but the action which is the goal of the activity and designed to be realized by way of it.…It can therefore be seen that a phased process of intended activity can never be isolated from its intended action. The reflexive view of true memory can only see action as a completed product of activity; it is blind to activity as a process. By the same token, the reflection of preremembering can grasp imagined actions, never imagined activity. Schütz 1974:78f. (my translation)

The temporally differentiated perspectives of consciousness and perception do not just mark the difference between activity and action; they also provide us with a way of distinguishing between conscious and unconscious activity. In Schütz’s paradigm, we can only speak of an activity as conscious if the activity has already been completed, that is, if it has already become an action. This applies not only to individual histories and biographies but also to the historiographical interpretation of facts in order to shed light on the discrepancy between what was historically desired and historically achieved (Schütz 1974:86). Husserl’s theories about the principles which structure the human consciousness of time are the ultimate source of the analytical differentiation between activity and action in Schütz’s approach, which has important implications for the issue of the free will of agents. Schütz argues that the popular spatial metaphor of making a choice between two separate ways or paths is not applicable to choosing an action; in a temporal sense, there can be no alternative between actions before an action has actually taken place, that is, before one of the paths of activity has been followed to its conclusion.56 We can summarize the two alternative types of action distinguished by Schütz as follows: –



56

In the mode of Selbstverstehen, or subjective understanding, action is the design which precedes a person’s behaviour and which he imagines as already completed. In the mode of Fremdverstehen, or objective observation, action is an interpretation of foreign behaviour which a person has observed and whose start and end point he, the objective observer, has subjectively asserted. He happens to interpret this behaviour as having reached its conclusion. Simultaneously, by analogy to the subjective action described above, he assumes that the foreign behaviour was preceded by a project on the part of its agent. Schütz believes that the erroneous idea of choosing an action has affected deterministic and non-deterministic theories alike. Drawing on Bergson, he attributes the mistaken analysis to the retrospective projection of a state of mind after the completion of an action onto a point in time that objectively precedes the choice which the agent made (Schütz 1974:94).

1.3 Philosophical Definitions of Action

63

In neither case is it possible to access the internal logic and dynamics of activity, which provides itself with new possible meanings as it takes place as a process. Its internal logic and dynamics cannot, therefore, be anticipated or grasped in retrospect. In a similar way to Arendt, Schütz explains the activity/action dualism as an epistemological and hermeneutic one; we can only recognize and understand a sequence of agential happenings if it has lost the character of a process because our interpretive intervention has given it a beginning and an end. In objective terms, there is no action which in itself can claim to be complete; all there is is infinite activity. However, neither Arendt nor Schütz discusses the conditions under which the singular events which comprise an observed foreign action can be interpreted as more than just a sequence of happenings between two arbitrary points A and B. Both approaches focus on the motives behind such an interpretation rather than the logic which makes it possible. Unlike Arendt, Schütz places the perspective of Selbstverstehen and the subject’s self-perception in the foreground, insisting on the primacy of the subjective perception that enables the perceiving subject to register any perceived event as his own experience and thus transform it into an action. Once again, we find ourselves reminded of the words of Nietzsche with which we opened this chapter, for as soon as we begin to relate singular happenings to ourselves and endow them with the character of experiences, their neighbouring events surreptitiously turn into elements of an action. Schütz was aware of this problem, but he did not interpret it as a compulsion to interpret each and every happening as an intentional doing in a global teleological context. Instead, he saw it as the expression of a need to create meaning as early as the very basic level of the singular perception of events: The reflexive view which focuses on a completed, distant experience and thus isolates it from all other experiences as one whose duration is well-defined creates that experience as one that is meaningful. Schütz 1974:95 (my translation)

As we have seen, Schütz provides us with a much-needed clarification of key concepts, particularly in the context of the distinction between action and activity. The question is now: can we go one step further and apply the methodology of his approach in the context of literary criticism? At first sight, there does appear to be a parallel between social scientists (who seek to objectively explain empirically perceived human behaviour as action) and the recipients of fictional narratives (who seek to interpret narrated behaviour as a global meaningful whole). But, tempting as it may be to conclude that the problems of comprehension and interpretation in the social sciences are analogous with those of literary criticism, such an assumption is not without its dangers. As is the case with the common-

64

Part 1 Concepts of Action

sense and analytical concepts of action, uncritically transferring Schütz’s concept to literary theory implicitly elevates the experiences of fictional agents to the status of quasi-empirical facts, thus ignoring the distinctive semiotic signature of narrative events. Or, if not that, it distorts our perspective so that we cease to discuss action in the particular fictional text we are studying and talk instead about the real human action which is concerned with texts in general and is already studied in other contexts such as speech act theory and reception theory. The concept of action as Schütz defines it cannot be directly applied to the happenings contained in fictional narratives. Nonetheless, in demonstrating that we need to read action in order to create meaning, he proves that, theoretically at least, the distinction between real and fictional happenings is irrelevant. Schütz explores our need for meaning not from the perspective of logical necessity but from the perspective of sociological requirements and functionality, most notably in the co-ordination of human activity. Reading action into perceived events is understood as a constructive activity which has a rich potential for producing coexistent alternatives and is by no means tied to a preordained result. Seen as open processes, activity and happening present us with possibilities for generating polythetic interpretive hypotheses. They produce a theoretical multiplicity of behaviour and experience which becomes a monothetic synthesis of action and meaning, both of which owe their existence to the identity of the acting and perceiving self. Or, as Schütz says: The entire process of experience to which the term ‘activity’ refers, from the planning of actions to the completion of concrete actions which can be identified in retrospect, is a synthetic mass which is built out of ordered polythetic acts. When it has run its course, therefore, it is visible to a one-dimensional glance, monothetically as is always the case in the naive and natural view of the world. The moment activity is completed, regardless of the phases in which it came into being, it becomes homogeneous from planning to execution because it is being observed by the one-dimensional glance of the I. Schütz 1974:93 (my translation)

The Logic of Construction: The Transcendental Approach The retrospective re-evaluation of past activity as action is the decisive act by means of which meaning is created. This re-evaluation logically requires the stable identity of the perspective from which the activity is perceived and then interpreted as a coherent and completed whole. But what if the dependency relation operates in completely the opposite direction and the identity of the perceiving subject is not a condition but a consequence of the interpretive act by which activity attains the status of an action? Might there be an epistemological, and perhaps ultimately

1.3 Philosophical Definitions of Action

65

transcendental, necessity that requires us to read action into our activity? Can we presuppose the identity of the self that synthesizes instances of activity into the meaningful whole of an action, or is not rather the identity of the self dependent on this synthesis which affirms the identity of the perceiving and the acting self? The question seems to be wholly philosophical in nature, for it obviously concerns our attitude towards cases of real empirical activity, be they attributed to ourselves or external subjects. However, by drawing on the metaphor of reading an action, we can easily relate the philosophical question to our original problem of literary theory. Our perception of real behaviour and our reception of narrated fictional behaviour are both affected by the same basic problems of interpretation. In both cases, our methods face the intrinsic difficulties of adopting the position of Fremdverstehen towards complexes of partially agential and partially agential events. Despite this parallel, however, we still have to face the anything but trivial question of why the recipients of fictional narratives employ a concept of action that allows them to integrate even completely non-agential happenings into a global action model equivalent to the fabula of the entire story. This question is particularly pertinent given the fact that the commonsense distinction between intentional activity and non-intentional behaviour is conventionally and silently accepted in real life as perfectly rational and self-explanatory. The problem can be rephrased in more fundamental terms. We need to decide on the authority to which we should attribute ontological primacy in the understanding of processes of happenings. Does primacy lie with the existence of agents and patients who can appear in successive states of affairs, or does it lie with the transformations which affect the state in which a person or thing exists and which manifest themselves as ontological facts in sequences of states of affairs? More succinctly, do events presuppose the existence of objects, or is it the other way round such that can objects only exist in relation to events? Let us begin by considering the Kantian approach to our problem. Kant’s point of departure is the idea that activity and perception are essentially equivalent. This allows him to define action by analogy to apperception. In transcendental apperception, individual, temporally differentiated perceptions are transformed into a temporally consistent and numerically identical meaning; they become a thought.57 The same applies to actions; 57

On this point see Rohs (1986), who comments that ‘apperception does not link perceptions but transforms them into something that is no longer a perception: into thoughts....The intertemporal, numerical identity of meaning must be understood on the basis of the intertemporal numerical identity of the apperception that accompanies perceptions’ (Rohs 1986: 224f.; my translation).

66

Part 1 Concepts of Action

in being meaningful, they are distinguished from natural occurrences by the property of invariance in the thought attributed to them. As Rohs has noted, ‘meanings, thoughts belong to actions too. Actions are not sentences, but they can be “comprehended” like sentences’ (Rohs 1986: 225; my translation). Thus, in order to become meaningful, individual actions must be subsumed under the unity of an apperception. This unity is the result of an identity of motives which transcends individual activities. The concept of action is thus based on the concept of the self, or, as Rohs puts it, ‘the motives can only be the same in the different phases of the action because an established and constant self is present within them’ (Rohs 1986:227; my translation).58 According to this interpretation of Kant, human activity must be seen in contrast to natural occurrences. However, such an argument is anything but unproblematic. As Gerhardt (1986) has shown, Kant’s pre-critical and post-critical writings apply the same concept of Handlung to natural occurrences and human behaviour alike. The concept refers to any type of event and does not make any further distinctions, requiring only that the event can be traced back to a causal relation.59 Thus, even the pre-critical Kant found himself faced with the question of how spontaneous and free human activity can be possible at all when it must be explained in the same way as causally effected natural happenings. Without rejecting the causal concept of action per se, therefore, Kant distinguished between two types of causation that can be attributed to perceived happenings. The distinction between various types of action is then derived from this distinction between different types of causes. A free (i.e. human) action is one that cannot be reduced to empirical causes and is derived from a cause of reason, a principium internum which is not externally attributed to the causing substance but immanent in it: In no way does the spontaneity made possible by the principium internum lead to an action of a particular kind; it only brings about a different attribution of the action. The basis of its definition, but not the action as such, is different. The causing substance too must be thought of as a different one, but the action schema, the relation of a real effect to an associated cause, remains valid. Whether free self-determination…exists or the course of events in the world is 58

59

Willaschek (1992) follows a similar line and points out the etymological root of Kant’s concept of Ursache in German metaphysics. He argues that Kant treats the fundamental difference between natural and human causes leading to an effect as a consequence of the directly experienced identity of the apperceiving self whenever it is engaged in self-apperception. The apperception of external happenings, on the other hand, cannot guarantee the identity of the perceiving self in this unquestionable way. Kant, as is typical of his time, uses Handlung as a translation of Latin actio to denote any ‘happening sub specie a consequence’ (‘Geschehen sub specie einer Folge’ ‘Gerhardt 1986:100; my translation).

1.3 Philosophical Definitions of Action

67

uniformly determined, nothing changes the fact that everything in nature (and therefore in man) is action. In any case, the course of events in our world is made up of actions. Even creation itself is interpreted as an actum creationis matching the model of action. Gerhardt 1986:107 (my translation)

Thus without affecting the reigning natural laws, individual phenomena can be seen in relation to other, empirically non-existent causes. In fact, the hypotheses of empirical and non-empirical cause are ultimately equally valid (Gerhardt 1986:128). Free human activity guided by reason simply denotes a subclass of the wider concept of action; it denotes a happening which is no longer understood in terms of external (physical or psychological) determination but which owes its occurrence to a free act of the will. More precisely, the happening is described as being rooted in such a wilful decision (Gerhardt 1986:121). This idea has far-reaching consequences when applied globally to empirical natural happenings. It allows us to give such happenings a wholeness and teleological coherence without which they would be automatically incomprehensible to man (Gerhardt 1986:124). Yet the sequence of cause and effect that constitutes an empirical event and is then reinterpreted synthetically in terms of some underlying reason is always and inevitably embedded in a world made up of a course of events which is fundamentally infinite and devoid of meaning. Moreover, the same inevitable contingency applies to the very act of reinterpretation which represents the primary example of an act of free will. In this context, neither the objects nor the results of free acts of apperception and interpretation are philosophically relevant. What counts is rather the transcendental side effect: the self-affirmation of a self that conceives of itself as numerically and temporally identical at every stage in the sequence of its apperceptions. This particular aspect is emphasized by Kaulbach (1979), who identifies a fully developed theory of activity in Kant’s philosophy. According to this theory, we meet transcendental activity when objects of perception are subject to the necessary conditions of cognition and practical activity. On the other hand, we find that instances of practical activity themselves testify to the fundamental role of the transcendental subject as the source of the possibility of cognition. In both contexts, the subject affirms itself as one that is constructively active, as one that has ontological primacy and is substantial in that, unlike the objects with which it deals, it is not affected by change: The acting subject makes itself the spokesman and guardian of the reason which proclaims the laws and practical cohesion of the world; practical reason designs the cohesion of the world so that the acting person can immerse himself in it. Kant is saying that reason becomes directly practical through activity. Kaulbach 1979:660 (my translation; emphasis in the original)

68

Part 1 Concepts of Action

Kaulbach interprets Kant’s dialectical combination of transcendental and practical activity as an attempt to use a dual concept of action to mediate between the two sides of Aristotle’s distinction between praxis, the principle of self-realization, and poesis, the principle of production.60 Kaulbach concludes that, for Kant, not only I think but also I act must be able to accompany each and every mental construct of the human mind (Kaulbach 1979:651). In the context of this conception of activity, does the subject of an action posit and found itself or does it expose and reveal itself? Should Kant’s concept of action be read as entelechial or constructivist?61 The question is less overwhelming if we narrow our scope and consider the implications of the problem for literary theory rather than philosophy. It is clear that fictional activity does not possess any real substance in the world in which it is stated to have occurred. But, despite this, it is able to evoke the mental image of one or more practically acting subjects in the recipient. Accrediting these fictional subjects with the ability to act transcendentally (the ability to conceive of a coherent model of the world in which their fictional acting occurs) would amount to asserting that fictional agents can have a real consciousness of their own fictional existence. The idea, though intriguing, simply replaces an already difficult problem with an impossible one. We have no choice but to reject it. The unsolved question of the ontological implications of interpreting happenings and instances of activity as fictional actions is perhaps best handled by posing it in relation to the reader of action. Seen in this light, the apperception of a non-empirical fictional complex of events and occurrences and their deliberate, totally free (empirically undetermined) interpretation as a synthetic action constitutes a perfect example of free transcendental activity on the part of the recipient. The process of interpretive synthesis through reading enjoys the same freedom as cognitive processes which are directed at the empirical world and able to impose an order on phenomena despite deterministic natural laws. In the paragraph ‘Clarification of the cosmological idea of a freedom in combination with the universal truth,’ Kant writes that 60

61

The argument in Wiehl (1976) also hinges on the analogy between activity and perception in Kant’s writings. Wiehl, however, emphasizes the character of an open process that is inherent in both. Bittner (1986:21) intends to show that Kant understood activity as proceeding in the manner of a syllogism, that is to say, as ‘an externalization of the envisaged law. The effect which someone causes simply brings to light what was already stated in the law of which he is and was aware’ (my translation). The argument, which Bittner himself finds contradictory, originates in the Aristotelian theory of action. Gerhardt, on the other hand, sees Kant’s concept of action as a precursor of a modern constructivist notion and likens it to Lenk’s concept of the interpretive construct. See Gerhardt (1986:116, note 13) and Lenk (1979:279–350).

1.3 Philosophical Definitions of Action

69

the condition of a successive series of occurrences could itself be empirically unconditioned. For here the condition is outside the series of appearances (in the intelligible) and hence not subject to any sensible condition or to any determination of time through any passing cause. Kant 1998:543 (= A 552)62

To the reader, the narrated happenings of a narrative amount to more than just a sequence of events, acts of behaviour, deeds, and suffering arranged in the and then progression of a linear sequence that draws together a number of quasi-intentional actions attributed to fictional agents. We interpret the sequence as a transcendental synthetic action that reflects the necessary condition of cognition to which we are subject: the identity of our self, which must remain itself throughout the sequence of apperceptions in its entirety, be their content real or fictional, concrete or abstract. When it is impossible to synthesize events and happenings into an action, on the other hand, the fictional world in which they are located becomes meaningless in a double sense. Not only does it lose its referential function with respect to human reality as experienced by the reader; it also fails to perform the transcendental function of asserting the reader’s identity. If we accept this epistemological argument derived from criticism of Kantian philosophy, we must conclude that the impulse which motivates us to read activity and happenings as action is not to be found in fictional entities, persons, or the situations in which they occur. Instead, it is immanent in the real act of perception and cognition, which is both an activity, in that it projects, and an action, in that it retrospectively asserts the self-definition of the receiving subject to be identical over time. The proof of our identity derived from this synthesizing reading of events that take place in a world, be it real or fictional, may actually be based on a logical circle, as Kant implies in his criticism of the ‘Paralogism of Personality’ (Kant [1924]/Kant 1998:A 442), in which he characterizes the coherence of the perceiving conscious not as a consequence of our concrete acts of cognition, emphatically not as proof of our identity, but as a logical precondition of our ability to perceive anything in the first place. The distinctive persuasiveness of narrated events and activity is illustrated by our readiness to read coherence into considerably disordered representations of fictional worlds (if all else fails, this coherence manifests itself only at the level of discourse). Might it not be the case that this persuasiveness points to a related, aesthetic paralogism that successfully tempts 62

‘Daß ‘die Bedingung einer sukzessiven Reihe von Begebenheiten selbst empirisch unbedingt sein konnte. Denn hier ist die Bedingung außer der Reihe der Erscheinungen (im Intelligiblen) und mithin keiner sinnlichen Bedingung und keiner Zeitbestimmung durch vorhergehende Ursache unterworfen’ (Kant [1924]:618 = A 552).

70

Part 1 Concepts of Action

us to mistake the logical necessity which governs our acts of aesthetic cognition for ontological proof of the existence of their fictional content? This view is supported by the fact that the synthetic interpretation of a fictional world is not normally restricted to a simple outline of its abstract logical coherence but also motivated by an interest in the moral, ethical, and philosophical message supposedly conveyed by all works of art. In the context of this argument, coherence of action suggests teleological order and thus compensates for what, apart from the collapse of the naive certainty of his own identity, is arguably man’s most shattering loss since the Enlightenment. Kant’s transcendental concept of action has many implications for the theory and aesthetics of reception beyond those which we have discussed here. It anticipates several ideas that feature in cognitivist and constructivist approaches to explaining action, some of which we shall consider in more detail below. But we must also remember that the Kantian approach has its dangers if applied uncritically to the needs of literary criticism. Foremost among these dangers is the fact that it leads to the reification of fictional entities which are thereby elevated to the status of quasi-empirical objects of perception and cognition. In the Kantian model, events can only manifest themselves with the attachment of different predicates to entities which are numerically identical across successive segments of time. Again, this implies that the entities (objects, agents, and patients) involved in happenings have ontological primacy while events (and their quantitatively more complex descendants such as actions) are mere secondary, inference-based abstractions. This may be plausible, but it confronts us with a paradox inherent in the reading of action in, or into, fictional narratives, a paradox which calls the relevance of the Kantian theory into question. How can we perceive states of affairs and infer events, let alone mentally construct entire actions, when the world in which they supposedly happen is by definition devoid of the existence of any entity, devoid, in fact, of existence itself? And how can we become convinced of the existence of our own real selves, even if only paralogically, by perceiving and rationalizing that which we know neither existed nor happened?63 Events and Existence: Towards an Event-Based Ontology Both the above questions alert us to the danger of misreading, albeit subconsciously, the logical cohesion of events in a fictional world as an ontological assertion of our own identity and existence. A radical way to avoid 63

Admittedly, this provocative formulation is somewhat extreme. Fictionality does not imply the factual non-existence of the entities it represents; it only requires a commitment to refrain from claiming or demanding proof of the factual existence of those entities.

1.3 Philosophical Definitions of Action

71

the risk of this misinterpretation is to invert the relation between events and existence. Among others, Abel has expressed this provocative thesis; he asks whether ‚events, understood as the primarily temporal occurrences to which we refer predicatively in nominalized propositions, function as the basic arguments of our ontology, or are they always derivatives of the schemata inherent in our language and basic concepts? (Abel 1985:157f.; my translation). Abel’s discussion refers to Davidson, Quine, Kant, and Hegel, before concluding somewhat contentiously: The primacy of becoming over ontological being is dominant throughout. Becoming must no longer be conceived of as a facet of mutable being. Instead, it has existence in itself, and being is actually an abstraction of becoming. Abel 1985:184 (my translation and emphasis)

This thesis turns commonsense ontology on its head; Abel’s main evidence to support is drawn from modern quantum physics, which conceives of physical objects as a series of temporally conjoined events at a microscopic level (Abel 1985:160).64 Drawing on Davidson, Abel believes that the same is true in philosophy, for it is a fact that a large proportion of what we think and say cannot be understood without assuming the ontological presence of events (Abel 1985:163f.). Most obviously, the idea of a sequence in time, indeed the very idea of time itself, can only be expressed if we recognize the existence of events. But what are events; or, more accurately, what form does their existence take? In their article in the Stanford Encyclopedia of Philosophy, Casati and Varzi give a concise up-to-date summary of the various distinctions that set events apart from objects, that is, from existent things: Although not undisputed, some standard differences between events and physical objects are commonplace in the philosophical literature. First, there is a difference in mode of being: material objects such as stones and chairs are said to exist; events are said to occur or happen or take place…. Second, there are differences in the way objects and events are said to relate to space and time. Ordinary objects have relatively clear spatial boundaries and unclear temporal boundaries; events have relatively unclear spatial boundaries and clear temporal boundaries. Objects are invidiously located in space—they occupy their spatial location; events tolerate co-location…. Objects can move; events cannot…. Objects are continuants—they are in time and they persist through time by being wholly present at every time at which they exist; events are occurrents—they take up time and they persist by having different parts (or ‘stages’) at different times. Casati and Varzi 2002 (emphasis in the original)

64

For a more detailed discussion of the correlation between the concept of the event in philosophy and the natural sciences, see Higginbotham et al. (2000).

72

Part 1 Concepts of Action

Abel identifies a more general ontological distinction and emphasizes that the philosophy of the event which he proposes is by necessity an ontology without substance or substrate (Abel 1985:171). This is easy to demonstrate by nominalizing a simple clause like Man thinks into Man’s thinking and inserting it into a sentence: Man’s thinking occurs. Although the sentence has a grammatical subject, it certainly does not refer to an ontological subject in the traditional sense of something existent. The example shows that Abel’s philosophical idea of an event-based ontology cannot but affect the definition of concepts such as the subject and the self. Indeed, Kant before him had already realized that the self is an absolute principle devoid of substance. Abel’s model takes this desubstantialization of the subject even further and conceives of thought in terms of an event-based ontology: If the I think and the synthesizing effect of categories are to be conceived of as actions, and if actions are a species of events, not the other way round (!), then what we mean by the word ‘I’ is a function whose argument is of the event type not the object type. The I-function must be treated as an event function. Abel 1985:172 (my translation; emphasis in the original)

The reference to the concept of action shows the importance of Abel’s idea of an event-based ontology for our discussion. It implies that the existence of the fictional world is ontologically established not by narrative propositions that assert the existence of fictional entities (both inanimate and animate) in it, but by the linguistic representation of events. Furthermore, the logically necessary first event is not a fictional event located within the fictional world; it is a perfectly real meaningful action in Moya’s sense: the cognitive act with which the recipient posits his self as the frame of reference. Thus, with his own cognitive act, the reader supplies the logically necessary first event that demonstrates the real existence of his real world and the fictional existence of the narrated world. Without this cognitive act neither fictional nor empirical existence can be conceived of at all, let alone considered meaningful. Let us turn from these far-reaching philosophical implications to the role played by events inside the fictional world in constituting the existence of that world. In our interpretation of Abel, these fictional events (rather than fictional things) are what make up the fictional world as an intelligible, meaningful one. Beneath the level of events, neither the fictional world nor its linguistic narration makes sense to us. But where exactly is the dividing line between a purely theoretical virtual being and a being that has become practical in the sense of the Aristotelian concept of praxis and the defining characteristic of an event? Regarding the minimal conditions for the linguistic representation of a potentially eventful existence, Plato pointed out the fundamental difference between the simple naming and

1.3 Philosophical Definitions of Action

73

proper predication of objects. He prescribed that proper discourse should combine the naming of an object with the description of that same object by attributing predicates to it. If this requirement is not met, we are left with simple naming. Neither nouns nor verbs amount to proper discourse if they are simply presented one after the other; only when combined do they constitute proper discourse, for it is in such a combination that the ‘action or inaction or existence of anything that exists or does not exist’ can be represented (Plato 1961:435). For Plato, being and action can only be communicated through discourse: When one says ‘a man learns,’ you agree that this is the least and first of sentences, do you not?...For when he says that, he makes a statement about that which is or is becoming or has become or is to be; he does not merely give names, but he reaches a conclusion by combining verbs with nouns. That is why we said that he discourses and does not merely give names, and therefore we gave to this combination the name of discourse. Plato 1961:435

Thus, Abel’s event-based ontology indirectly illustrates the central argument of Plato’s criticism of mimetic representation as the deceptive use of language to attribute existence to the non-existent. That which exists can only be perceived once it surfaces in a context of movements or happenings in which we can identify it as something that is moved or static, altered or unchanged. It cannot be perceived in isolation outside of such a context. And the poet, the cunning sophist, exploits the dependence of our cognitive faculties on events by turning it to his advantage: rather than directly postulating the existence of the non-existent, he imitates an event, knowing full well that the recipient will unwittingly complement it with the image of an agent or patient involved in the imitated event. Fictional speech thereby obscures the fact that the ontology of fiction does not presuppose substance; it tempts and encourages us to interpret fictional (insubstantial) events analogously to real (substantial) events. Moreover, unlike the persons and things we encounter in the real world, the beings whose existence is postulated in a fictional mode are typically presented in more than a single momentarily perceived event; they reappear throughout entire event sequences which, taken together, constitute a complex whole. When synthetically interpreting the overall meaning to be attributed to the multitude of narrated events, recipients cannot but consider the recurrence of identical things and agents in changing situations and predications. In doing so, readers dignify those things and agents as existences which not only exist but whose existence is itself deemed to be meaningful. Thus, although it is not even remotely ontologically necessary, fictional entities are paralogically treated as substantial; we think that something happens, and, more importantly, that it occurs in a meaningful synthetic overall context.

74

Part 1 Concepts of Action

And we conclude that the transformational processes we have observed did indeed involve some thing that does, albeit fictionally, exist. We have seen that, from a transcendental perspective, the reading of action amounts to more than just the attachment of an aesthetic surplus value to the representation of happenings that take place in possible worlds. It also points to a way out of the dilemma in which we find ourselves when we realize that our conscious sense of our own identity is logically dependent on our awareness of the changes in, if not the non-identity of, the objects that we perceive. With that, we have arrived at the true sense of the aphorism with which we began this philosophical discussion, in which Nietzsche criticizes the human inability to interpret happenings as anything other than intentional actions. However, his words also point to another facet of the concept of action which we have not yet considered in the detail it deserves: the fact that action is interpretive, that actions are hermeneutic constructs built by means of synthetic judgements. The Constructivist Concept of Action Despite their differing emphasis, the philosophical approaches surveyed so far are agreed on one fundamental principle: ‘action’ is not a term which denotes an empirical phenomenon but a concept which stands for a particular kind of synthetic interpretation and description of empirical (or fictional) phenomena. These phenomena can appear as isolated individual events in a sequential order or can be seen together as a structured action which consists of a complex of paradigmatic relations. We have repeatedly found that ‘action’ refers not to a particular kind of logically distinct bundle of events in itself, but rather to the description of a bundle of such events which attributes an intention, meaning, or reason to them. This interpretation is found in Davidson’s theory, which argues that a person is deemed to perform an action when what he does can be described as intentional; in Arendt’s entelechial approach, where the exposure of an acting existence comes to an end only with its posthumously narrated biography; and in Kant’s fundamental postulate that happenings can be described both as conditioned by the laws of nature and as the results of a transcendental causality free from any such conditions. In all these theories, the concept of action refers to a hermeneutic entity rather than an empirical one. There are only two hypotheses which postulate the existence of a type of event which constitutes action in itself. The first is Danto’s theory of the basic action, in which the individual’s will and deed are inseparable from one another (Davidson 1977:286); the second is Bourdieu’s concept of

1.3 Philosophical Definitions of Action

75

habitus, with which he argues that empirically observed social behaviour is based on a superindividual logic of activity which automatically raises such behaviour to the status of a logically ordered action (Bourdieu 1979, 1998). Interestingly enough, these essentialist approaches also argue that the logic of the behavioural actions to which they refer is inaccessible to both external observers and the agents themselves. In Danto’s basic actions, we act directly without our physical behaviour being bound into causal chains which we could analyse and describe in a rational way. Bordieu’s concept of habitus, on the other hand, must be seen in the context of his methodological criticism of the rationalizations which social science and social anthropology employ in order to explain the social behaviour of others. He argues that these rationalizations stem from the observer’s interest in identifying a particular, familiar type of action logic; this makes the observer unwilling to accept that the observed behaviour is actually guided by a blind dynamic of activity to which not even the actors themselves have access. The ideas of Danto and Bordieu on the theory of action amount to an agnosticism which prevents us from discussing any further the cases which define the spectrum of agential events that can be considered actions. We can reflect neither on instances of an agent’s spontaneous physical movement at one end of the scale, nor on complex social interaction at the other. Accepting Danto’s or Bourdieu’s argument leads to the conclusion that, even if action does exist in some phenomenological sense, we can neither know nor understand it; the result is that the whole debate about the essence of what an action is becomes irrelevant. And that brings us back to our starting point, the argument that it does not make sense to talk about action as an object that one can identify within the world. By taking the metaphor of ‘making sense’ literally, we are reminded that all the approaches we have discussed employ action as a descriptive hermeneutic concept. It is clear that the shared premise of all philosophical discussions of action is the recognition that action, as a particular way of describing happenings, quite literally makes—produces—sense. It does so by attributing an observed happening with a disposition towards a meaningful content which is structured before or after the happening itself chronologically takes place. This meaning can be seen as the realization of an idea about how we would like the world to appear as a result of our action (Danto); as the result of our symbolic commitment to future activity that characterizes meaningful action (Moya); as the exposure of the existence of an individual subject (Arendt); or, in the social sphere, as the result of the design of an activity, either projectively to postulate meaning or retrospectively to give meaning (Schütz). However, in the final philosophical reckoning, we must agree with Kant that the making of sense

76

Part 1 Concepts of Action

through action is based on the transcendental postulate of a perceiving and acting self. Generally speaking, the approaches discussed above treat action as an interpretation of sequences of occurrences; depending on the particular methodological framework in question, these sequences are seen as consisting either of singular events and behavioural instances or of complex substructures of happenings and activity. Regardless of such differences, each approach conceives of action as a construct which, although it necessarily requires the empirical basis of events, happenings, or behaviour at the objective level, is equally dependent on the presence of a competent interpreter who perceives the basic material. This twofold requirement of action as a construct makes it possible for the philosophical concept of action to be compatible with the concept of action in literary theory. Seen from the perspective of either discipline, action is, as far as its logical form is concerned, a synthetic judgement in the sense of the Kantian definition of synthetic empirical judgements which ‘add to the concept of the subject a predicate that was not thought in it at all, and could not have been extracted from it through any analysis’ (Kant 1998:141 = B 11).65 In the philosophical field, Hans Lenk has given particular attention to the constructive nature of action. In Lenk (1979), he coins the programmatic phrase of ‘action as an interpretive construct’ (‘Handlung als Interpretationskonstrukt’; my translation). His approach aims to prove once and for all that the essentialist concept of action is misguided and inappropriate. Like us, Lenk begins by discussing whether it is possible to define action in terms of intentionality; he concludes that the attribution of intentionality to an observed behaviour is a sufficient but by no means necessary condition of activity. While phenomena which can be described as intentional behaviour are, in principle, instances of activity, not all instances of activity are members of the special class of intentional behaviour.66 The restricted applicability of the intentionality criterion is a first indication of the fundamentally descriptive (as opposed to denotational) nature of the concept of action; this becomes even more apparent when we begin to deal with the question of how actions can be individuated.

65

66

‘Sie fügen dem Begriff des Subjekts ein Prädikat hinzu, welches in jenem gar nicht gedacht war und durch keine Zergliederung desselben hätte können herausgezogen werden’ (Kant [1924]:61 = B 11). ‘Subject’ is here meant in the logical rather than grammatical sense. ‘We must therefore conclude that intentionality or the possibility of there being an intentional description cannot in itself be regarded as the absolute criterion for activity.…Not all action is intentional.…Purposeful [intentional] activity should be seen as a case—even if a special one—of activity in general’ (Lenk 1979:284f.; my translation).

1.3 Philosophical Definitions of Action

77

Actions are made identical or non-identical not because of any of their substantial attributes (e.g. the combination of cause and effect, the identity of their agents) but rather by the fact that they appear as the bearers of identical or non-identical descriptions. This observation brings us to the central thesis of Lenk’s theory of action: objectively observable actions do not exist; what exists is only a set of perceptions which can be described as actions. At its core, therefore, action is an interpretive construct, or rather an action construct, as Lenk puts it with a critical allusion to Danto’s concept of the basic action: Actions are not spontaneous movements of the body but interpretive constructs, interpretations of (usually observable) movements. They come about as the result of an interpretation which is dependent on requirements of context, situation, recipient, addressee, and, ultimately, resolution. They are constructs which can be understood as actions only when they have a particular description. Actions should not be conceived of as objectively observable events—they are explicitly not observable movements; rather they ‘are’—formulated somewhat loosely—interpreted events; strictly speaking, they are not events at all but really interpretive constructs which can (often ‘must’) be combined with (predominantly) observed events (movements, for example). Lenk 1979:293 (my translation)

For Lenk, even activity is existent as such only at the level of interpretive description. This reformulation of the concept of action in the context of a theory of description affects not only the attitude of scientific observers towards movements and changes but also the attitude of agents towards their own doings. They too can conceive of action only from the epistemological position of an external observer. It follows that Schütz’s epistemological distinction between Selbstverstehen and Fremdverstehen is no longer relevant to the constructivist concept of action. Furthermore, not only does the agent retrospectively interpret his own behaviour as action in terms of his own context; the external observer projectively treats instances of foreign behaviour as action and does so in terms of the situational, normative, and historical context applicable to his position as an interpretive agent. In terms of their ability to generate retrospective and projective action structures, the roles of the acting subject and the interpreting subject are mutually interchangeable. A further aspect of this conception of action as an interpretive construct makes it possible—at last—to free action of its illusory substance and understand it as a structure which is an integral part of a dynamic interpretive context. Interpretive action constructs combine facts and observations not only of current interpretive constructs but also of adjacent ones which are already historical; such historical constructs can contain conventions, classifications, norms, descriptions of situations, and so on. This means that the motivations and motives of action become part of

78

Part 1 Concepts of Action

the description of behaviour as an action. When we refer to such motives, we are concerned not with the identification of objective facts but rather with the interpretation of psychological dispositions which may have been brought about in the agent by the agent himself, or by someone else. Lenk’s constructivist concept of action would seem to lend itself like no other to the needs of literary criticism. Among other things, its appeal lies in the fact that it allows us to avoid many of the difficulties which are normally entailed by applying an essentialist, or realist, concept of action to narrated worlds. For example, the question of the assumed motives and intentions of fictional agents no longer needs to be addressed in terms of referential truth and can be considered instead in terms of systematic plausibility and consistency in the context of a particular action construct. Furthermore, treating action as a hermeneutically derived statement about object-level narrative statements means that the constructivist concept of action is methodologically located at the level of metalanguage. This proves that it is futile to attempt to draw a clear-cut ontological distinction between real and fictional actions. Both real and fictional actions are hypothetical syntheses of event material, irrespective of the fact that this material enters our consciousness through empirical (real) perception in the former and symbolic (fictional) perception in the latter. The ontological difference lies in events, not actions, and the key difference as far as real and fictional actions are concerned is not ontological but systematic. A real action is a dynamic construct which can trigger events within a synthetically conceptualized world. It has a reflexive potential not to be found in the fictional action construct which, while it can have an aesthetic or cognitive effect on the real world of reader or author, cannot exert any influence on its own domain, the fictional world. We interpret fictional happenings as actions on the basis of a predefined and finite set of narrated events which are completely blind to the interpretive meta-event of which they will eventually become part. In other words, narrated events are absolute events: simply by virtue of being narrated, any event that is narrated is postulated to have receded into the past. It has thus become something which the recipient can cognitively process but to which he can no longer practically respond. As we have known since Plato, such mimetic absolutism is not without its dangers: literary action is an interpretive construct which, while the narrator provides the material for it in the form of propositions, is ultimately the responsibility of the reader in whom it is realized under the influence of other constructs which are largely drawn from the empirical circumstances of his own real existence. This reference to the process of construction brings us to what can be seen as the crucial problem in the constructivist concept of action as Lenk describes it. Even if action is indeed a purely interpretive construct, it is

1.3 Philosophical Definitions of Action

79

obvious that the necessary conditions for the generation of the construct are not limited to the three listed by Lenk, which are (1) it must be possible to attribute the act to an agent; (2) it must be possible to attribute the faculties of consciousness and control to that agent; and (3) the interpretation must have a normative frame of reference. If we were to follow Lenk here, we would have to accept the implication that action is defined primarily by conventions and that our use of the concept of action must therefore conform to socially accepted interpretive rules rather than universal formal or logical criteria. Yet concepts such as agent and control (used by Lenk himself in (1) and (2) above respectively) imply logical rather than conventional predication: the agent is the causal originator of the event, and his control of his own behaviour is expressed in his ability not to act, in other words, in his ability to negate the existential predicate of a process. In a similar vein, Lenk’s model makes the scope of the interpretive construct entirely dependent on the interpreting subject, more precisely, on the conventions to which that subject subscribes. This means that phenomena such as openness and closure can only be analysed in the sense of explaining the conventional rules in question. But what is the material basis of these interpretive conventions? Is there any inherent quality in observed events and happenings which could help us evaluate the plausibility and acceptability of a given interpretive construct? Or is it rather a case of anything goes, such that, at least in theory, anything and everything can be interpreted as an action? If all happenings can be (or become) actions simply as a result of an interpretive decision, we have no choice but to accept a relativity of description. It is telling that Lenk advocates such a position more and more in his later works of the 1990s.67 His terminologically succinct slogan ‘action as an interpretive construct’ eventually becomes completely abstract: it excludes the categories of sense and meaning, despite the fact that Lenk explicitly conceives of action as a hermeneutic phenomenon. Constructivism could certainly redefine the concept of literary action as the interpretive construct of a global action which is generated by the recipient on the basis of entities and processes that exist in the fictional world. However, such a definition would leave two key questions unanswered. First, what logical and semantic properties allow a world and the objects and processes 67

The concept of the interpretive construct recurs in Lenk (1993, 1994, 1995, 1997). Significantly, Lenk (1993) has the title Interpretationskonstrukte: zur Kritik der interpretatorischen Vernunft. (Interpretive Constructs: The Criticism of Interpretive Reason). In his later works, Lenk attempts to formulate a theory of interpretive constructs which he hopes will result in a unifying epistemological theory capable of bridging the gap between the ‘two cultures’ defined by C. P. Snow. Eventually, everything becomes an interpretive construct for Lenk—including the concept ‘interpretive construct’ itself.

80

Part 1 Concepts of Action

it contains to be integrated into such an interpretive construct? Second, what does the interpreter gain from constructively synthesizing fictional elements of perception? We shall attempt to answer the first question in chapter 1.4 by mapping the constructivist approach onto a narratological and semiotic model of the minimal action for which the term ‘episode’ will be proposed. An answer to the second question can be found in Hegel’s discussion of action as an aesthetic concept. Hegel: The Philosophical Value of Aesthetic Action Unlike the approaches we have discussed so far, the final theory in our survey treats action, whether in the sense of empirical or aesthetic phenomena, as a historical concept. For Hegel, the acting and behaving subject of the premodern and classical eras was a heroic character who answered ‘for the entirety of his act with his whole personality’ (Hegel 1975, 1: 188). Only in modern times did man learn to identify a particular subset of what he does as action. As Hegel puts it in the passage we have just quoted, from this point onward man ‘enters to his own account only what he knew, and, on the strength of this knowledge, what he did on purpose and intentionally’ (Hegel 1975, 1:187f.). In this distinctly intentionalist definition, action is one particular aspect of the general concepts of doing and behaviour; in the end, one particular aspect of the concepts of happening and event. The validity of this definition is generally accepted in jurisprudence, where it is treated as derived from the moral philosophy’s presupposition of a free will that is aware of its own existence and of that of other free wills.68 However, the reference to the intentionality criterion is less important for the classification of doing as an action in an aesthetic sense, where we should concentrate instead on the analogy between Hegel’s definitions of action and cognition.69 According to Hegel, who defines the concept in an essentially anthropocentric fashion, 68

69

See Hegel’s remarks in the Grundlinien der Philosophie des Rechts, §113: ‘The expression of the will as subjective or moral is action. Action contains the following determinations: (a) it must be known by me in its externality as mine; (b) its essential relation to the concept is one of obligation; and (g) it has an essential relation to the will of others’ (‘Die Äußerung des Willens als subjektiven oder moralischen ist Handlung. Die Handlung enthält die aufgezeigten Bestimmungen, a) von mir in ihrer Äußerlichkeit als die meinige gewußt zu werden, b) die wesentliche Beziehung auf den Begriff als sein Sollen und g) auf den Willen anderer zu sein’ Hegel 1954, 7:172f.; translation from Hegel 1991:140). ‘An action requires [according to Hegel]—and here too nothing distinguishes it from the theoretical concept of cognition—a concrete situation in which the undefined generality of the state of the world specifies or stylizes itself to fit an individual action or individual action context’ (Wiehl 1971:147; my translation).

1.3 Philosophical Definitions of Action

81

singular, subjective action formally originates in a situation of conflict.70 Without a conflict, without a clash of particular interests and aims, the state of the world is ‘devoid of situation’ (‘situationslos’; my translation). Inventing and depicting such states of the world, which are the opposite of action, is the primary concern of arts such as sculpture and architecture. In epic and drama, however, it is imperative that we trace the chain of situations leading from the cause of a conflict to its eventual solution. It is clear that, in this context, action cannot be reduced to the subjective reasoning of individual agents. Aesthetic action, the mythos, amounts to more than the concept of specific and conscious doing; it also includes the Other of subjective reason: the powers (originally fate and the gods, today replaced by history, society, impulse, and the subconscious) which influence the overall development of action and cannot be influenced or even perceived by the individual. The superindividual powers join the characters of individual agents in guiding the successive states of the world which, taken in their entirety, represent a homogeneous closed action. However, aesthetic action does much more than juxtapose subjective reason and autonomous powers. For Hegel, the cognitive importance of aesthetic action, which places it on a par with the theory of cognition itself, arises from the fact that the production and reception of an aesthetic mythos both express a claim to participation in the absolute reason which organizes every complex and closed overall action. At the level of this absolute reason, the antinomy of human free will and superindividual powers is dialectally resolved. Hegel identifies the drama as aesthetic action’s most radical claim to autonomy: it is the genre in which aesthetic action resists reduction to a mimetic function. The action in a drama is, it would seem, the sum total of the actors’ speech acts. Once the narrative authority has receded and disappeared into the background, the aesthetic illusion attains perfection: the recipient can no longer trace the aesthetic action back to the subjectivity of a lyric or epic I. Its part has now been taken over by an abstract subjectivity in the action as a whole, a subjectivity which takes the form of absolute reason and is ideally suited to identificational participation.71 If the effect of dramatic action were confined to presenting the recipient with the material and pretext for enacting fantasies of authorial 70

71

The following summary is heavily indebted to the discussion of Hegel in Wiehl (1971: 145ff.). See also Stegemann (1999:71–102). This subjectivity of absolute reason in the unity of action must, when compared with the individual consciousnesses of the acting individuals, appear as a power which, unlike the power of the epic narrator, will remain anonymous and all the more alluring to the audience. The anonymity and invisibility of this power only add to the seductiveness of its invitation to the audience to share in it, itself, and its absolute knowledge’ (Wiehl 1971:157; my translation).

82

Part 1 Concepts of Action

omnipotence, it would be hard to see why Hegel does not arrive at the same conclusions as Plato. However, Hegel sees the aesthetic illusion and its radical claim to autonomy not as a purpose in itself but rather as a way towards understanding the truth. Hegel argues that aesthetic action, when made reflexive and considered in full awareness of its illusory nature, allows recipients, whose participation in its supposedly objective reason is itself illusory, to understand that illusion is nothing short of a necessary category of cognition itself. This is how we gain the opportunity to proceed to perceiving that truth which lies beyond illusions, most importantly beyond the illusion embodied by all that is unequivocal and unambiguous. Although the system presented in Hegel’s Aesthetics culminates in observations specific to the drama, he nonetheless considers the epic to be the primary medium of the aesthetic cognition so critically important, for, unlike the drama, the epic can integrate even the category of chance into its action: [E]ven if in drama external circumstances are operative, they still can only count through what the will and the mind make of them and the manner in which a character reacts to them. But in epic, circumstances and external accidents count just as much as the character’s will, and what he achieves passes before us just as what happens from without does, so that his deed must prove to be conditioned and brought about just as much by his entanglement in external circumstances. Hegel 1975, 2:1070

The drama may seem to be more absolute and to present a world with a higher degree of coherence than the epic. As far as the representation of the overall logic of events in our empirical world is concerned, however, the drama is outclassed by the epic because epic poetry moves in the element of an inherently necessary total state of affairs, and nothing is left to the individual but to submit to this fundamental situation, i.e. to what is, be it adapted to him or not, and then suffer as he may or must.…This destiny is the great justice and it becomes tragic not in the dramatic sense of the word in which the individual is judged as a person, but in the epic sense in which the individual is judged in his whole situation; and the tragic nemesis is that the greatness of the situation is too great for the individuals. Hegel 1975, 2:1071

Aesthetic action, and in particular the aesthetic action of the epic, has a double value for Hegel. On the one hand, it is a medium of philosophical criticism; it expresses the necessarily illusory nature and dependency on perspective and interpretation which characterize subjective human activity and cognition. The dependency is most obvious when both fail. As Wiehl comments, in Hegel ‘neither the illusion of art nor the illusion

1.3 Philosophical Definitions of Action

83

of reality, to which it is analogous, has the meaning of simple illusion or deceit’ (Wiehl 1971:162; my translation). On the other hand, aesthetic action (that which is designed as and represented in a mythos) is in itself nothing but a material form and product of human activity; at the same time, it is one of the few real actions, if not the only one, to stem from a truly free will as the expression of an ago which is on equal footing with its cogito. For Hegel, therefore, aesthetic or dramatic action, unless it is consumed merely for its mimetic illusions, represents a twofold human claim to truth and autonomy. From his perspective, the concept of aesthetic action is a dialectical one. Not only is the mimetically represented activity of fictional agents confined to the periphery of Hegel’s consideration—even the mythos itself is eventually made the subject of a non-aesthetic definition when it is conceived of as an aesthetic means which the cunning of reason employs to let us attain philosophical truth. In constructing a mythos, we perform the privileged act with which we express our free will to achieve this truth (Wiehl 1971:170). The Hegelian concept of aesthetic action has little to do with mimesis or, indeed, with any symbolic or referential concept of action. It focuses instead on the cognitive and philosophical relevance that the production and reception of art has for the individual. This follows from the nature of Hegel’s thought in general, as can be demonstrated by an extract from the Phenomenology of Spirit. Hegel defines true action as the type of individual doing with which an individual consciously presents himself in the world by causing something to come about in it. This process, however, necessarily entails the individual’s alienation from his inner self and his being as such: Speech and work are outer expressions in which the individual no longer keeps and possesses himself within himself, but lets the inner get completely outside of him, leaving it to the mercy of something other than himself. For that reason we can say with equal truth that these expressions express the inner too much, as that they do so too little: too much, because the inner itself breaks out in them and there remains no antithesis between them and it; they give not merely an expression of the inner, but directly the inner itself; too little, because in speech and action the inner turns itself into something else, thus putting itself at the mercy of the element of change, which twists the spoken word and the accomplished act into meaning something else than they are in and for themselves, as actions of this particular individual....The action, then, as a completed work, has the double and opposite meaning of being either the inner individuality and not its expression, or, qua external, a reality free from the inner, a reality which is something quite different from the inner. Hegel 1977a:187 (= 312; emphasis in the original)72 72

‘Sprache und Arbeit sind Äußerungen, worin das Individuum nicht mehr an ihm selbst sich behält und besitzt, sondern das Innere ganz außer sich kommen läßt und dasselbe

84

Part 1 Concepts of Action

In the strictest sense of Hegel’s philosophical definition, action is tied to the individual. Only at a higher level do we encounter the work, the complex whole of an overall action, where the latter is the objective combination of all the actions performed by individuals and all the contingent coincident occurrences. For this reason, aesthetic action is necessarily illusory, for the function of the absolute reason into which we can gain insight by constructing a mythos is not simply to reconstruct the logic of events in terms of chains of effect that are partially causal and partially free. Seen as a product, as a completed overall aesthetic action, the mythos can only mimic the truth of the world at large, be that world real or fictional. However, seen as a process, as a truly free and subjective act which is subject only to formal, logical conditions and gives rise to this same illusory truth, aesthetic activity is true because it is real. This process of constructing, the perilous and paradoxical attempt to arrive at a synthetic interpretation of events as an action is the only activity, the only free doing, through which we can bring ourselves ‘zur Wirklichkeit’ in the double sense of ‘being true’ and ‘being real’. This aesthetic activity must, therefore, be evaluated as an alienating hypostatization. The mythos which we, the aesthetically acting recipients, construct as an action is one of the linguistic and practical expressions in which, as we read above, ‘the individual no longer keeps and possesses himself within himself, but lets the inner get completely outside of him, leaving it to the mercy of something other than himself’ (Hegel 1977a:187 = 312). Conclusions How and under what narratively established constraints do recipients generate the concept of literary action in their minds? It should by now be clear that the answer to this question is bound to depend, more than anything else, on what we understand by action.

Anderem preisgibt. Man kann darum ebensosehr sagen, daß diese Äußerungen das Innere zu sehr, als daß sie es zu wenig ausdrücken; zu sehr,—weil das Innere selbst in ihnen ausbricht, bleibt kein Gegensatz zwischen ihnen und diesem; sie geben nicht nur einen Ausdruck des Innern, sondern es selbst unmittelbar; zu wenig,—weil das Innere in Sprache und Handlung sich zu einem Anderen macht, so gibt es sich damit dem Elemente der Verwandlung preis, welche das gesprochene Wort und die vollbrachte Tat verkehrt und etwas anderes daraus macht, als sie an und für sich als Handlungen dieses bestimmten Individuums sind.…Das Tun also, als vollbrachtes Werk, hat die doppelte, entgegengesetzte Bedeutung, entweder die innere Individualität und nicht ihr Ausdruck oder als Äußeres eine von dem Innern freie Wirklichkeit zu sein, welche ganz etwas anderes ist als jenes’ (Hegel 1977:235 = 312).

1.3 Philosophical Definitions of Action

85

The survey in this chapter has shown that there is a common denominator shared by all the philosophical definitions we have considered: ‘action’ is not a term which denotes an object in the world; instead, it characterizes a particular way of talking about the world. Within this very broad definition, a narrower narratological concept of action seems to be best served by the constructivist theories. They treat action as a particular kind of human interpretive construct that attributes observed events with an agent (not necessarily human) who is characterized by the qualities of control, consciousness, and rule-based (or rule-negating) behaviour. This constructive idea of action covers only a small subsection of what we intuitively mean when we talk about action; it represents the category which we have defined as Action I. The action construct with which literary theory is concerned, however, also includes the category of Action II and thus encompasses the entire complex structure which traditionally falls under the concept of mythos. To complicate matters still further, we are searching for a concept of action which can not only be explained theoretically but also put into narratological practice in the form of controlled analytical procedures which provide verifiable descriptions of the relevant features of narrative texts. Most importantly, literary theory must not ignore the functional aspect of action as a construct that creates meaning. Thus we must give particular attention to the fact that the reading of action (be that action our own or foreign, real or fictional, past or future) is an act of semantic construction. Reading action amounts to employing what I shall call a heuristic algorithm; the term is intended to indicate the interpretive and combinatorial processes which it involves. In this sense, reading action means instantiating the variables (e.g. agent, patient, cause, effect) of an algorithmically based heuristic model of action and filling them with concrete empirical or fictional data. It is probable that a minimum number of such variables must be instantiated before our heuristic algorithm can begin to function and generate a meaningful partial construct and then, eventually, a global construct which integrates the highest possible number of basic events just like an overarching mythos does. The extension of this construct is, it would seem, entirely up to the interpreter. This cannot, however, also be true of its logical structure; if it were, the concept would be meaningless for our purposes. Apart from enabling us to perceive and understand the world, our cognitive acts, and in particular our synthesizing constructive acts, have a crucial reflexive function: as Kant has shown, they help establish our sense of identity. As far as modelling the process from the reader’s perspective is concerned, this is perhaps the most fundamental point which we must take into account. By reading action in (or indeed into) a sequence of ac-

86

Part 1 Concepts of Action

tivities, we assure ourselves of our own identity, of the fact that we have remained the same throughout a sequence of perceptions and doings.73 From a Hegelian perspective, however, the subject’s interest in reading aesthetic action is simply a manifestation of the cunning of reason which employs the aesthetic construct as a means of making the fundamentally illusory character of all acts of perception and cognition transparent and performing the only truly free human act. This free act consists of generating the illusory construct itself. The construct becomes an externalization of its constructor; it is a construct which tells us at least as much about the logical and axiomatic conditions which govern the constructor’s mind as it does about the logic of action which applies to the fictional and narrated world. In the end, seen as a constructive phenomenon of reception, narrated action turns out to be of far greater philosophical depth and profundity than we would ever imagine from our commonsense understanding of the concept, which reduces it to effects of suspense, closure, and identification. This constructivist dimension, it is hoped, will always be given due consideration in the narratological model for describing and combinatorially analysing action which is presented in the coming pages. The model is not, however, intended to make any contribution to the philosophical debate on the concept of action; our study stands in the tradition of structuralist and formalist narratology and is concerned with tracing aesthetic effects back to concrete elements and structures in the empirical narrative text. The logic of plot and story has interested pre-deconstructivist narratological research mostly as a subject for a theory of poetics or, more abstractly, semiotics. One of the few integrated models to be found is that of Schmid (1982, 1984), who describes the process of the formation of fictional objects and the text as a whole from a combined generative and semiotic perspective. There is an obvious parallel between Schmid’s generative approach to the phenomenon of action and the constructivist approach suggested by Lenk. Both conceptualize complex action as a phenomenon dependent on a change of level in the system, a change from the level of ‘Geschehen’ to that of ‘Geschichte’ (Schmid; perhaps best translated as ‘happenings’ and ‘story’ respectively). This change of level presupposes a progression from the sequential order of singular events to their meaningful paradigmatic 73

The importance of the epistemological approach is a key methodological difference between the current study, in which it features strongly, and Peter Brooks’s Reading for the Plot (Brooks 1984). Brooks too emphasizes the surplus semantic information which is generated by the synthesis performed by the reader of a narrative. However, he refers mainly to the methodological paradigm of psychoanalytical theory, specifically the integrating power of eros, to explain the principle motivation which drives our ‘reading for the plot.’

1.3 Philosophical Definitions of Action

87

structure. Neither of these models, however, discusses the conditions which must be met in order for such a generative or constructive transformation to be possible, nor how this transformation can be traced in a step by step manner. Once we have accepted the philosophical position that actions do not exist as such, it is merely a matter of personal preference whether we choose to focus on the constructive activity of the producer (author) or the recipient (reader) of an action. The theorists of classical rhetoric, indeed, would hardly have been able to make sense of the distinction; the idea of ex nihilo aesthetic production was unknown to them. The author was not considered to be a creator; he was treated as an inspired medium (the poeta vates), or, to put it in Platonic terms, an aesthetic artisan who found, selected, and arranged his material in a process guided by inspiration or technique. From its very beginnings, therefore, poetic speculation has been familiar with the idea that aesthetic artefacts are, in a logical sense, always secondary texts which can only occur in an intertextual context. This recognition calls into question the Hegelian idea of art as the only medium in which truly free human action can be experienced, but it does not affect the Kantian epistemological argument of the role of reading action in forming the self. Our approach will not consider how writers preform actions; we shall focus instead on the process with which readers construct them. It should be noted that a narratological exploration of this process requires an approach which proceeds inductively, and, as far as possible, does not presuppose a universal philosophical concept of meaning. Our first step is to explain the lowest common denominator of every definition of action, the concept of the basic event. In the second step, we define the episode as the minimal construct which can be built using event material. Step three explores the combinatorial possibilities that are inherent in episodes and can result in the formation of complex action constructs. I intend to capture the relationships between events, episodes, and actions by means of a combinatorial model in which the middle layer, the episode, is central logically as well as structurally. The scope of this approach is of course modest compared to other narratological or hermeneutic models of action which address far greater questions. It cannot provide any insight into the psychology of the narration and reading of action and is restricted to an exploration of the fundamental necessary conditions which apply to the reading of action right from the very basic level of the minimal constructs which can then be integrated to form more elaborate structures. These elaborate structures would obviously require the use of an action concept more sophisticated than that to be developed in the coming chapters.

88

Part 1 Concepts of Action

In the end, then, we will not discover a radically new concept of action that consigns our predecessors to the history books. Against the background of the complex philosophical debate on action, which we have only been able to outline here, literary theory is in no position to claim that any one of the many ways of talking about action alone is relevant to its purposes. Conversely, it would be ill-advised to declare that any one particular philosophical concept of action is valid and postulate that only phenomena which fall under this particular concept constitute instances of literary action. The concept of action that I hope to describe is not normative; it is an exploratory, combinatorial concept and intended to allow us to understand why narratives contain such a wide range of possible actions. At the same time, I believe that Kant was right to point to the epistemological necessity of reading at least one such possible action in order to stabilize the identity of the self in the course of the perceptive and cognitive process. This does not, however, necessarily amount to engaging in metaphysical speculation. Studies in cognitive psychology, cognitive science, and reception psychology all underline the fact that the human mind can only retain a finite set of unrelated data. This data must be continually synthesized and compressed if we are to reduce its complexity and create space for new data. Seen in this light, the human ability to read action in(to) events is simply a very efficient way of dealing with the problem of limited memory. Remembering this pragmatic aspect is just as important as discussing more elaborate philosophical reflections on the concept of action, which is, therefore, necessary in more than one sense. Finally, as it is hoped the following pages will demonstrate, my interest lies in a constructive rather than constructivist concept of action. The latter adjective implicitly raises the phenomenon and its functional description to the status of a scientific methodology; it expresses a dogmatic need for systematization which can only further confuse what is already a complicated issue.74 Hegel was right to drive home the fact that the synthesis of fictional events and the building of what we can now call action constructs is one of the most demanding of all human cognitive achievements. All that is far-reaching, elegant, and creative in the human mind cannot but run the risk of being suggestive. That is why we opened this chapter with Nietzsche’s polemical words, which accuse us of being bent on reading intentional action into anything and everything. But if we read Nietzsche against the grain here—and he would probably have wanted it that way—we see that his fundamental criticism also has a positive side. It testifies to the astonishing human ability to identify con74

In other words, I shall use ‘constructive’ in the descriptive sense found in Eco (1979) rather than in a normative sense.

1.3 Philosophical Definitions of Action

89

nections even within a set of totally separate events, and to do so in a cognitive environment where neither the things nor the events perceived are in the least bit real but merely point to a world which is by definition possible, not real. Even in this context of a conventional non-reality, we are willing to experience the effects of the mental constructs we create. And those of us who are prepared to admit that they gain pleasure from reading narratives will confirm that these effects can be just as impressive and far-reaching as those derived from the constructs we generate on the basis of empirical perceptions of the real world.

1.4 The Elements of the Action Construct In this chapter, we shall discuss the theory of the epic action construct which is used in the next two parts of this book. My theory does not seek to explain the structure and logic which govern the composition of complex actions which encompass entire narratives. It is instead an attempt to characterize the smallest possible unit of closed action (where closure has yet to be defined.) This minimal construct is the synthetic interpretation of a complex happening, and it functions according to the logic of action. I propose that we adopt the term ‘episode’ to refer to it. As the word ‘episode’ has many different, not always mutually compatible meanings in both technical and vernacular usage, we need to define more accurately what it means in the context of our study. One of the major concerns of this chapter is therefore to provide a considerably more precise definition of the episode in which the original meaning of the word is respected. The decision to focus on the episode as the smallest possible coherent action construct has implications for the terminology and methods we use. In general, applied narrative theory affirmatively defines or posits a global concept of action; I intend to replace this with an inductive concept which is more suitable for subsequent use in literary computing. Before we can formulate the conditions which govern the inductive process of building the narrative action construct, we need to define its components. We begin, therefore, by examining the event, the perceptive phenomenon that is the smallest element shared by all theoretically possible action constructs. Our second step is to investigate how basic events can be combined to form minimal action constructs. To distinguish the more general use of ‘event’ and ‘episode’ from their use as constructive terms, these words will be identified typographically (as event and episode) when they are used specifically as constructs of reception as outlined in the theory of the following pages. Our field of inquiry stands in a long tradition of research into the logic behind the narration of events and happenings. The formulation of a precise definition of the phenomena of event and action has long been an important concern of narrative and dramatic theory; it is of considerable relevance to semiotics and, more recently and importantly, cognitive science and psychology. A comprehensive overview of research in these areas would fill a book in itself, so we shall consider only those approaches which are directly relevant to the model we develop in the course of this chapter.

1.4 The Elements of the Action Construct

91

Lotman’s Definition of the Event The definition of the event developed by Juri Lotman has come to enjoy near-canonical status as a result of being employed for over three decades in a significant amount of work in tacitly hermeneutically orientated narrative theory (as opposed to more descriptive narratological approaches, such as that of Gerald Prince, which adopt rigidly formalist methods).75 As is well known, Lotman defined the event in normative and semantic terms as ‘the crossing of that forbidden border which the plotless structure establishes’ (Lotman 1977:238). For him, the axioms which govern the fictional world are expressed in the delineation of semantic fields which represent semantically homogeneous topographical, pragmatic, ethical, psychological, or cognitive spaces. These axioms are normative postulates, and each of the syuzhet’s events represents a violation that calls these axioms into question.76 Events are therefore differentially defined. Even if they are presented as no more than self-contained boundary crossings, they nonetheless call the order of the entire fictional world into question. It is likely that narrative theorists have found Lotman’s approach attractive because his definition draws on the universal complex of values and axioms that is narratorially posited in the narrated world. This means that his model meets a key requirement of hermeneutic approaches: the requirement that the analysis and description of the object under investigation (i.e. the narrative text) should be directed as far as possible towards providing a global interpretation which is based on the categories of the sense and meaning of what happens in the narrative. Lotman’s event concept allows interpretations to be developed rigorously (and thus in an objectively verifiable way) on the basis of an examination of the text alone. Or rather, it seems to do so. In methodological terms, it deceives us into crossing a border of our own and moving from a formal method of semantic description to a normative method of interpreting content. Wolf Schmid emphasizes the formal content of Lotman’s event concept by pointing out its conceptual parallels with Danto’s definition of the basic conditions of narration, which in Schmid's words amount to an ‘opposition between the situations or states of mind of a subject (x) at two points in time which are connected by something which happens to (x) between 75

76

A concise and informative overview of Lotman’s theory can be found in Schmid (1992: 107–9). Lotman (1977:238f.) contrasts syuzhet texts with syuzhet-less, or mythological texts, which do not narrate new developments in a changing world but portray the cyclical iterations and isomorphisms of a closed universe whose order they fundamentally confirm. The modern syuzhet text was produced by the two typologically primary text types influencing and distorting one other.

92

Part 1 Concepts of Action

them’ (Schmid 1992:107f.; my translation).77 However, not only is this modest definition anthropocentric; it is heavily dependent on a number of assumptions, at least when we try to apply it productively to textual analysis. Schmid postulates no fewer than seven additional conditions which are necessary for an event to be present in an aesthetic environment. It is not hard to see that we can only judge whether a narration of happenings fulfils these conditions once the whole text has been read and at least a rough interpretation of it been formed.78 For methodological reasons alone, Lotman’s definition can be of no use in an inductive model of the event, for in such a model we can assume neither that there is a global frame of reference (itself always constructed interpretively), nor that there is a predesigned normative semantic space. Renner (1983) is the most ambitious attempt to date to find a way out of this methodological dilemma and apply Lotman’s model in a rigorously inductive fashion. Renner interprets semantic space not as a global prespecified quantity but as a frame of reference which must be reconstructed step by step as the act of reading progresses. But even this often fascinating approach, for all its formal and theoretical ingenuity, is ultimately unable to deal with the basic problem: the fact that it is impossible to circumvent the reference to a pre-existent normative framework in Lotman’s concept.79 Indeed, on a wider scale, it has hardly been noticed that Lotman’s wellestablished event concept is critical rather than analytic; it is clearly based on a semiotic description of the criteria that a text must meet in order to 77

78

79

Danto’s logical triad has been put forward in similar forms by other theorists. As Stempel (1973) explains, it can be realized by a sequence of just two sentences. Stempel details a number of conditions required by the minimal schema of a narrative sequence: the subject in which the change takes place must be identical; the content of the narrative statements must be compatible; the predicates must form a contrast, indeed an opposition; and the facts must be arranged in chronological order. Schmid’s seven conditions are (1) reality (the change taking place in the fictional world cannot be simply desired, dreamt, imagined, or hallucinated); (2) relevance (it must be significant); (3) resultativity (the change must have actually occurred before the end of the story); (4) unpredictability (it must come as surprising, unexpected); (5) irreversibility; (6) non-iterativity (it must be unique); (7) consecutivity (the change must have implications for the thoughts and deeds of the subject) (Schmid 1992:108f.). Renner (1983) explains the concept of space, which Lotman uses more metaphorically, in terms of set theory. He sees semantic space as composed of the set of ordering theorems which express the statistically regular conditions to which the fictional world and its happenings are subject. Ordering theorems are not, however, pre-existent postulates; they must be abstracted out of the narrated happenings by the interpreter. However, this model too requires that the formation of ordering theorems logically precedes the identification of events. In this theory, therefore, the concept of axiomatic ordering is a prerequisite of the event rather than created by it. Moreover, Renner retains Lotman’s idea of evaluating happenings semantically. In other words, he, like Lotman, does not treat all happenings as events; only those happenings which represent a statistically significant deviation from the norm can be considered events.

1.4 The Elements of the Action Construct

93

have a properly interesting syuzhet. It is easy to see that the conceptual roots of Lotman’s concept lie in the idea of originality common in the nineteenth and late eighteenth centuries, with which it shares the belief (superficially semiotic, but ultimately cultural and ideological) that art can only be a significant event when it seeks out, portrays, or stages conflict with social or aesthetic norms in the sense of the crossing of a border. This kind of differential definition of the event is directed at the syuzhet and attempts to characterize semiotically what Goethe referred to with appropriate ambivalence as the ‘unheard-of occurrence’ (‘unerhörte Begebenheit’; my translation), a sequence of happenings that has not previously been handled aesthetically and is therefore ‘unheard-of’ in a literal sense while also being an ‘unheard-of’ (i.e. provocative) challenge to established norms. If we now defy Lotman and the theorists who support him, and insist on a purely formal definition of the event, two objections must be reckoned with. First, signification is by definition only conceivable under the condition of semantic difference. One way or another, if we are to be able to define a narrative event as a meaningful one, we seem to be required to postulate a system of values defined by their content, be the value system a semantic space, a set of fictional axioms, or even an ordered structure to which a narrated occurrence can be related and by means of which it can be differentially defined. This argument is persuasive, but must be put in perspective. Approaches such as Lotman’s assume a highly complex, aesthetically normative concept of meaning, while it is our belief that the potential signification of an event is located as low as at the microlevel of the sequence of states of affairs, which is purely syntagmatic and thus essentially oblivious of any involvement it may have in a system. In only two cases can differential reference be made to a high-level construct such as a paradigmatically ordered semantic space. Either a sufficiently complex set of axioms must be available in the world (i.e. we must know what kind of sequence of states of affairs is normal in the fictional world and thus uneventful in Lotman’s terms), or it must be possible to draw on a generic topos or a motif (e.g. the search, the choice, the battle, etc. in the case of the fairytale) for heuristic assistance. But meaning in the sense of such a high-level construct is far beyond the scope of our investigation, which is directed specifically not at the content of event and action constructs but at the fundamental conditions which make it possible to construct them in the first place. One possible way out of this problem is offered by atomic theories of meaning, whose initial assumption is simply that every atom of meaning—linguistically represented by words and cognitively represented by the mental concepts attached to them—always has a specific opposite. Indeed, it is important for the process of inductive event understanding

94

Part 1 Concepts of Action

that readers are able not only to recognize the individuating property of a (real or fictional) state of affairs but also to identify the opposite of that property immediately and accurately. From our basic perspective, however, it is of little importance what systematic complexes and paradigms these pairs of opposites form; we can be content just to register their internal interrelationships as a basic condition required by all events. The second objection to formal definitions of the event concerns the dynamic nature of reception processes. Frequently, we only realize in retrospect that something is a narrated event; again and again, we jump in and out of the narrative syntagm in order to re-evaluate narrative statements which we have already processed. As readers of real and fictional processes, we are by no means restricted to identifying events procedurally and sequentially. We are perfectly well able to put initially unconnected, self-contained states of affairs on hold until we are able to evaluate them retrospectively as events which make up an action by referring to the paradigms that have since been successively established during the course of narration and reception or drawn from our world and textual knowledge. Thus, even if there is not a theoretical principle which requires us to read events after they take place, we frequently do so anyway in practice, by referring to a semantic space that has taken shape since we first encountered them. If this is so, why continue to insist on an inductive model of how we perceive events? This objection cannot be rebutted. In fact, it should be reformulated even more radically. Any reading and perception of events is unquestionably an act of interpretation in itself and thus a second-order happening which chronologically and logically follows the reported or perceived occurrence. The intuitive belief that we register basic real or fictional events in real time is an illusion. There is a difference between the initial comprehension of an event, supposedly as it happens, and the subsequent retrospective re-evaluation of the event from the perspective of a paradigmatic knowledge which has become more comprehensive in the intervening period. However, the difference is one of degree, not of principle. In both cases, the perceiving subject can only recognize as an event that which has, albeit tentatively, already become part of a synthetic interpretation. We have touched on the question of the ontic status of events in the context of our philosophical discussion of the concepts of action and event and shall return to the issue again below; for the time being, let us simply say that, unlike Lotman and the theorists who follow him, we assume that it is genetically impossible to construct the structure of the complex semantic space of a narrative without the basis of events which are apparently directly perceived. For when there are no basic events, the fictional world completely lacks the immanent diachrony and

1.4 The Elements of the Action Construct

95

Hegelian ‘situational quality’ (‘Situationshaftigkeit’; my translation) which together provide the foundation required by more abstract concepts such as systematic structures and semantic spaces and their boundaries. It is important to note that recipients do employ retrospective methods at the second stage of reception in order to identify the crossing of boundaries and the questioning of previously established normative values which, at a meta-level of the system, are typical of the syuzhet. However, in doing so, the recipients direct their interest at aesthetic phenomena far more complex and meaningful than those that could properly be covered by a formal concept of the event. Indeed, we are concerned no longer with a question of meaning but with the issue of the sense of the narration or perception of an event. Lotman’s idea of semantic space as the boundary of possibilities in the syuzhet possibilities is, therefore, in no way a basic category, in no way a concept which we must continually bear in mind if we are to gain any understanding of the event. It is rather an aesthetic construct which is based on the interconnection of basic events and comes into being only through such interconnection.80 Structuralist Approaches: Event, Transformation, Move A number of authoritative approaches take Propp as a model in their structural description of the internal logic of the fabula’s action. Bremond is one such critic; his interest is directed less at an event’s semantic effect (the particular values it normatively postulates, confirms, or questions) and more at its function in the development of the overall event complex and the formation of the complex action.81 In his model, the basic binary structure of a singular event is not normatively defined as thesis and antithesis or negation and affirmation; it is defined modally to correspond with the dichotomy between the possible and the real. Every situation, every state of affairs has the potential to be involved in an event; the key question is therefore, what formal conditions are necessary if this potential is to be realized so that states of affairs can be linked together to form the basic sequence of an event? As Bremond puts it,

80

81

On the relationship between historical models and the reception and narration of events in historiography, see among others (for once avoiding a reference to Hayden White) Greimas (1973), Koselleck (1973, 1973a), Lämmert (1973), Stempel (1973) and Lübbe (1977). See in particular Bremond (1964, 1966, 1973). For a criticism of Bremond’s approach and the project of generative narrative grammars in general, see Scheerer and Winkler (1976). Bremond’s approach plays a particularly important role in Pavel’s move grammar (Pavel 1976, 1985).

96

Part 1 Concepts of Action

what are the minimal conditions to be met by a temporal segment in order for it…to be rendered extensively in a narrative?…The basic sequence typically articulates itself in three principal moments, each of which holds an alternative: – a situation which ‘opens’ the possibility of a behaviour or an event (with the proviso that this virtuality is actualized); – the conversion of this virtuality into an act (for example, the behaviour which responds to the incentive contained in the ‘opening’ situation); – the final result of the action which ‘closes’ the process by way of a success or a failure. Bremond 1973:32 (my translation)

Bremond illustrates the basic unit of narratable happenings, the séquence élémentaire, as a serious of dichotomous choices:

after Bremond 1973:131

As with Propp, Bremond treats the acting subject as a variable; his approach is concerned not with the logic of activity but with the logic of action, which instrumentalizes figures and happenings for its own purposes. Todorov takes this abstraction even further; he believes that the event is the minimal unit of every happening and is formed analogously to linguistic affirmative clauses. His point is that the description of action should be orientated around the predicate and take the form of propositions, for he considers that the grammatical subject of an affirmative clause is ‘devoid of internal properties, for these derive only from a temporal junction with a predicate’ (Todorov 1977:111).82 He concludes: The agent cannot be endowed with any property, but is rather a blank form which is complete by different predicates. The agent has no more meaning than a pronoun such as ‘he’ in ‘he who runs’ or ‘he who is brave’.…We shall therefore keep description solely within the predicate. Todorov 1977:110f.

Todorov’s objective is to create a descriptive system which can be used as a framework for classifying two different classes of predicate. The first class comprises the narrative predicates that are used in stative or iterative episodes which describe a state of equilibrium or disequilibrium; these 82

Todorov’s argument is based on Tesnière’s dependency grammar (Tesnière 1959, Manjali 1997).

1.4 The Elements of the Action Construct

97

predicates are referred to as narrative adjectives. To the second class belong the narrative predicates which describe a change of state and therefore a transformation; they are referred to as narrative verbs. Todorov argues that these predicates are comparable with verbs which describe the passage from equilibrium to disequilibrium. Several narrative grammars take up the idea of treating the event as a special kind of narrative statement which mediates transformationally between two stative propositions. One example is Prince’s model, which differs from Todorov by explicitly returning to the idea that there is a link between the affirmative statement and the grammatical subject, which is thereby identified with the subject of the narrative action. Prince takes the event as the basic unit of a story and describes it in the form of an basic affirmative clause. According to Prince, a minimal story consists of exactly three events, of which the first and third are static events (they describe a state) and the second an active event which effects the transformation from the first to the third state. Prince’s model, one of the most stringent of all narrative grammars, confronts us with a problem which is at first glance nothing more than a matter of terminology. As we have seen, he describes both static and transformational propositions as events.83 The three events must be arranged in a fixed chronological sequence; in order to achieve this, they are connected by two temporal conjunctive features. In addition, the second and third events must be connected by a causal conjunctive feature. Finally, the third event must be identifiable as an inversion of the first. Thus, the full definition of the minimal story is as follows: A minimal story consists of three conjoined events. The first and third events are stative, the second is active. Furthermore, the third event is the inverse of the first. Finally, the three events are conjoined by three conjunctive features in such a way that (a) the first event precedes the second in time and the second precedes the third, and (b) the second event causes the third. Prince 1973:3184 83

84

They are indeed events insofar as they are manifestations of narrative speech acts, but this is not the interpretation Prince has in mind. Prince’s next step is to widen the scope of the grammar from the kernel simple story to the simple story, that is, a story which turns a kernel simple story into a plot by means of specific technical narrative transformations such as zeroing (the elimination of redundant events), coding (the generation of a story-specific code by using multifunctional events which are themselves the precondition of zeroing), and converting the natural event chronology into a narrative one. The transformations that take place at this level are applied to individual events and therefore referred to as singular transformations. The transition from simple story to complex story then takes place at the highest level of the system. The complex story is composed of a combination of several simple stories. The transformations executed at this stage do not affect individual events but entire stories; Prince refers to them as general transformations and describes them as grammatical rules.

98

Part 1 Concepts of Action

Apart from the contradictory description of the initial situation as a stative event, this formulation reflects an ontic premise whose implications could not be greater: the assumption that stative states of affairs and the dynamic transformations between states of affairs are located at one and the same object-level. This amounts to the assumption that events exist, just as states do, as fictional entities which can be referenced by narrative propositions. Todorov takes a markedly different attitude; there is good reason for his metaphorical distinction between narrative adjectives and narrative verbs. Furthermore, we have him to thank for a more precise clarification of the concept of the transformation. At first sight, his description appears relatively trivial and reads: ‘Two propositions may be said to be in a relation of transformation when one predicate remains identical on both sides’ (Todorov 1969:224). That is to say, there is a principle such that a proposition composed of argument and predicate can only be transformed by modifying the argument while preserving the predicate: P1(A1) becomes P1(A2) (a simple transformation) or P1(P2(A1)) (a complex transformation). Simple transformations modify or complete the argument of the proposition, while complex transformations extend the original predicate by adding, or grafting onto it, a second predicate, which thereby acquires the logical function of a complex argument. This allows us to define the minimal and the maximal conditions of a transformation: On the one hand, there is not yet a transformation if the change of operator cannot be clearly established. On the other, there is no longer a transformation if instead of two ‘transforms’ of one and the same predicate we find two autonomous predicates. Todorov 1969:22585

Todorov uses the term ‘predicate’ in a primarily logical rather than grammatical sense. His work is written against the background of a model in which predication is seen as the logical connection of a particular term (the argument) with a general term (the predicate). Todorov deviates slightly from this convention by explicitly restricting occupation of the argument position to figures, the fictional concrete entities in narrated worlds.86 The ontic status of the transformation, or event, in this model is particularly important for our purposes. It is clear that Todorov does not view the 85

86

I prefer the term ‘argument’ to Todorov’s ‘operator’, which is misleading in terms of predicate logic. Predicates have arguments and are connected by operators. According to Quine (1960:100–114 = §§21–23), on the other hand, all word forms and concepts can appear as predicates; there is no isomorphic mapping of grammar onto logic. With the help of grammatical means, general, predicative terms (e.g. adjectives) can be converted into singular arguments of the second order (abstracts).

1.4 The Elements of the Action Construct

99

transformational event as an occurrence which takes place in the fictional world and affects the concrete entities (figures) in it; he sees it as a process which connects universals (predicates) with one another in the logical space above the quasi-empirical fictional world. Prince draws on this approach in Narratology (Prince 1982), the revised version of his earlier work, which now bears the significant subtitle ‘The Form and Functioning of Narrative’. In it Prince presents a reformulation of the event concept in terms of predicate logic; this, he hopes, will allow narratology to move from the structural to the logical analysis of the semantics of action narration. The real meaning of the ideas and title of Prince’s earlier A Grammar of Stories is now apparent: from the beginning, it was a grammar of the narration of action, not a grammar of action and the events of which it is made up.87 Thus, the category of sense is now paramount: Characteristically, narrative presents more than temporal sequences of states and actions (involving some kind of conflict): it presents temporal sequences of states and actions that make sense in terms of a human project and/or a humanized universe. Prince 1982:148

With that, we have closed the circle and once again arrived at the question of whether the event is the expression of a human project and can only be perceived deductively by being aesthetically reconstructed, or whether the project is created when the event takes place, be it the production or reception of narrated happenings and their constituent events, and does not exist prior to it. Can individual events, irrespective of the formal description we 87

This is particularly apparent in the context of Prince’s explanation of the third substitution rule for narrative propositions (rule 1 states that they are replaced systematically from left to right, rule 2 that neighbouring propositions must be chronologically ordered and non-identical). In rule 3, Prince postulates that ‘a proposition can be a substitute for the element N prop stat mod (modified stative narrative proposition) if and only if its initial argument is identical to that of the proposition substituted for the corresponding N prop stat (stative narrative proposition) and its predicate is a modification of that of the latter proposition’ (Prince 1982:93). More simply, this means that if the kernel narrative consists of three statements, of which (1) and (3) are stative and (2) is active (in the sense that it converts statement (1) into statement (3)), (1) and (3) must have the same argument, and the predicative definition of (3) must be a modification of that of (1). On the one hand, this is indeed an insightful refinement; in a concrete narrative, (1) John is happy can be converted by (2) He falls ill into (3) John is unhappy, but apparently not into Bill is unhappy. In actual fact, however, not only the predicates of propositions but also their arguments can be transformed in a story. (1) John is happy (2) He falls ill, and (3) Bill is unhappy also form a story, but here we are obviously confronted with a case of the ‘zeroing’ of a redundant proposition (see Prince 1973), for the connection between (1) and (3) is now only implicit (there is some kind of causal connection between (1) and a zeroed proposition Bill is happy).

100

Part 1 Concepts of Action

apply to them, enter into any semantic connections, or do they inherently contain some kind of parameter which specifies a priori how and into what larger structures they can subsequently be combined? Lévi-Strauss discussed this problem in his analysis of the Oedipus myth. He argued that the human project is present from the start as a fixed procedural constant which is able to impose itself on every narrated event; we are, after all, always concerned with reconciling opposite terms which reflect the fundamental aporia of all human existence. According to Lévi-Strauss’s view, actions and events are nothing but an obvious expression of this stative universal structure from which diachrony has been removed. The criticisms of this model are well known and do not need to be repeated here. We shall turn instead to a countermodel which attempts to link the different models of action and the event as found in pragmatics and logical semantics without resorting to the fatalism of structuralism and the philosophy of history. This new model is the move grammar which Pavel develops in his book The Poetics of Plot (Pavel 1985).88 Its basic idea is borrowed from game theory, according to which a game is a rule-governed, teleologically directed process in which two or more competing players make tactical moves in pursuit of an agreed objective, continually influencing one another as they do so. This makes the overall strategic situation a dynamic one and forces the individual players to continually re-evaluate the situation and react to it with new legal moves. Although they know the aim of the game, the players cannot plan more than a few moves ahead in their efforts to reach it. Pavel describes the individual transformations of state in the fictional world analogously and uses the term ‘move’ to refer to them. In doing so, his model draws on Bremond (from whom he takes its tripartite structure) and Todorov (the source of the transformation concept). Unlike Bremond and Todorov, however, Pavel argues determinedly that the subject of action should be reintroduced into the theory of plot because ‘plots as strategic clashes cannot be reduced to sequences of anonymous actions; a proper understanding of plot includes knowledge of the person or group who performs an action, the reason for it, and its effect on the overall strategic configuration’ (Pavel 1985: 88

Pavel uses the term ‘plot’ in this book without explicitly considering the familiar narratological distinction between story and plot (fabula and syuzhet). He develops a structural descriptive model which lies somewhere between an abstract representation of the fabula, or story, and an exact analysis of the narrative technique of the plot. This does not present any difficulties in the context of Pavel’s example analyses of Marlowe’s dramas; the narrative manipulations of the basic story involved are rarer in the drama than the novel, and Marlowe’s dramas, apart from relatively transparent chronological shifts, do not employ extreme manipulations of the story. At a more general level, Pavel’s approach joins the structuralist works which see the transition from story to plot as a process of successive transformations.

1.4 The Elements of the Action Construct

101

14.) Let us examine Pavel’s definition of the move in more detail. It reads as follows: As in game-theory, a Move is the choice of an action among a number of alternatives, in a certain strategic situation and according to certain rules.…The main criterion for an action to be considered as a Move is its impact on the overall strategic situation. An action is a Move if it either, directly or indirectly, brings about another Move, or if it ends the story. Pavel 1985:17

Moves consist of two components: a problem that leads to or forces a choice between alternative actions, and a solution that (as a fictionally performed activity) represents a protagonist’s tactical response to the problem. As in transformational generative grammar, tree diagrams are an ideal way of visually representing individual moves and showing how they can be combined with and embedded in other moves:

after Pavel 1985

Structures such as the above reflect the narrative causality by which the solution to Move 1 leads into the problem of the subsequent move, Move 2. The model captures the formal, syntactic structure of the plot, but, in its present form, it does not have anything to say about the internal logic of the individual moves. For that, a notation is required which can reflect different perspectives towards the overall strategic situation. Pavel justifies this requirement pragmatically: It seems to me unlikely that one could go much further in eliminating the concrete narrative content than the reduction of the plot to a hierarchical structure of moves. To obtain a specific plot from these empty trees, one needs at least a set of semantic elements that ‘fills’ them with content.…It will be assumed that the semantic structures proposed operate on the syntactic basis offered by the Move-structure. Pavel 1985:42

For this reason, Pavel introduces the semantic category of the narrative domain, which is defined as the set of moves which are initiated by a particular protagonist (or group of protagonists). The moves are indexed

102

Part 1 Concepts of Action

with the name of the acting person concerned. As a rule, narrative texts contain several such domains, each of which can be characterized by the ontological, epistemological, axiomatic, and pragmatic norms and rules which apply to the protagonist concerned. Compared to other narrative grammars, Pavel’s move grammar has considerable advantages, not least of which is its concise and intuitive terminology, which is far more easily applied than the formal language of, say, Prince. The move grammar allows us to portray at a glance the highly complex arrangement of all the individual narrative actions which constitute a plot, and it allows the logical causal order and the narrative order to be handled in a single model. In addition, it provides a way to create a typology of plots, as Pavel demonstrates with examinations of epochal styles and wide stretches of the history of literature and its forms (see also Pavel 1990). The fundamental weakness of the model, besides its obvious suitability for dramatic rather than epic texts, lies in the key concept of the narrative domain, with which Pavel introduces instances of the fictional person, the character, into the narrative grammar. This forces us to adopt a naive, realistic view of the fictional subject and not only takes us back to before the approaches of Greimas and Bremond’s Logique du récit but also contrasts with Propp’s concept of the function. Pavel is basically concerned with making the logical progression of the action comprehensible by deriving it from the psychological and tactical situations of its protagonists and antagonists. For example, at one point in his example analysis of Marlowe’s Tamburlaine we read ‘The whole episode serves the interests of Tamburlaine and will accordingly be represented as a Move belonging to Tamburlaine’s domain’ (Pavel 1985:27). Subversively, a psychologizing concept of action is established; it reduces action to a mimetic reflection of the intentionalistically interpreted activities of figures.89 If we are to adapt the move model for use in an expressly non-intentional concept of action, it is precisely this reductionism that we must avoid at every point, right down to the level of terminology. The objective of our adaptation must be to combine the strengths of the 89

Pavel points out the possibility of identifying propositions and maxims that are of a general nature as well as those that are relatively tightly bound to the protagonists. There is no doubt about the existence of such ‘general regularities’ (Pavel 1985:46), and Pavel is certainly correct when he states that ‘the maxims and regularities involved in the semantic domains contribute to the intelligibility of Moves, by making clear the conditions under which Problems arise, Auxiliaries can be sought for, and Solutions devised’ (Pavel 1985:47). But if these ‘conditions’ are truly general in nature, then either they are not domain-specific, or the concept of the domain has been misdefined such that the unity of a domain is established not by the protagonist or character, but by the nature of the specific and general maxims and the relations between them by means of which one domain can be distinguished from another.

1.4 The Elements of the Action Construct

103

move grammar with an approach based on cognitive theory. The move grammar will contribute the principle, borrowed from game theory, of portraying action as a dynamic system (i.e. a system which is both a syntagmatically ordered sequence and a paradigmatic structure of events which mutually influence one another). The approach of cognitive theory will make the basic moves and the complex overall action comprehensible not as the intentional products of fictional actors but as constructs of the real recipient. The Event as a Modal Change of State Ryan’s deliberations on a theory of the ‘Modal Structure of Narrative Universes’ become particularly significant at this point in our discussion, the more so because Ryan was one of the first to move from text-centred theories of plot to using a theoretical computer model to capture the complex processes of narrative comprehension (Ryan 1985, 1991).90 Ryan’s model draws directly on Pavel’s move grammar and Doležel’s work on the semantics of possible worlds; she describes the individual narrative plot as a complex array of possible worlds, where each such world contains a partial modal description of the narratively evoked fictional universe (the hypothetical sum of all the partial descriptions or worlds). Accordingly, the narrative universe contains an actual world, which has absolute validity, and a wide range of hypothetical worlds, which are only valid in terms of the knowledge and beliefs possessed by a figure. The latter set of worlds includes intention-worlds and epistemic or knowledge worlds on the one hand, and model worlds on the other. Ryan uses the term ‘knowledge world’ to denote the subjective representation of the actual world in a figure’s consciousness. The intention-worlds of figures are aimed at directly modifying the actual world, while model worlds are concerned with an ideal description of the world ‘as it ought to be’ which is not directly relevant to the action (Ryan 1985:726).91 According to Ryan, the narrative plot can thus be described as the changing of the complex relations which govern the entire system of all these worlds and their interaction with one another, where the change occurs in pursuit of a goal with an intention in mind. Seen from the perspective of an intentionally acting agent, such an intervention in pursuit of a goal is the equivalent of a move 90 91

See also chapter 2.1 below. Some particularly significant members of this latter class are wish-worlds (Ryan 1985: 726), worlds of moral values, obligation-worlds and complex alternate universes (Ryan 1985:728–30), which are layered and structured accordingly (as, for example, in the case of the phantasmagorical construction of a separate world by a figure or a reference to a world in a book in the world, that is, an intertextual allusion).

104

Part 1 Concepts of Action

in Pavel’s model. The theoretically significant distinction which follows is that between the narrative state and the narrative event as constituent elements of the narrative plot: If a narrative state is a system of relations between the actual narrative world and the relative worlds of the characters’ domains, a narrative plot is the Movement of these worlds within the global narrative universe. From the viewpoint of the characters, the goal of the ‘narrative game’ is to make the actual world coincide with as many as possible of their private model worlds.…Whenever some of the propositions of a private world…become unsatisfied in the actual world, the narrative universe falls into a state of conflict. This event can be visualized as a satellite of the actual world leaving its orbit.…By this definition, conflict is not simply the complication or thickening of the plot that occurs between exposition and resolution, but a more or less permanent condition of narrative universes. Ryan 1985:732f.92

Consequently, apart from the extreme case of an eschatological or apocalyptic resolution, there is no situation in which all the conflicts that can be represented in a modal description are resolved. Instead, for Ryan, the narrative of a plot has reached its end when all the fictional agents who are able and willing to take part in actions have no more possible productive actions to make. Analogously to Pavel’s move grammar, the realization of such possible actions should be thought of as a transition between states. Narrative states alone, therefore, are not sufficient to constitute a plot; only when they are connected and changed by narrative events does the plot come into being. This complex dynamic machinery is continually evaluated and re-evaluated by the reader as reading takes place. Ryan assumes that we subscribe to an intentional concept of action, as becomes clear from her subdivision of narrative events into happenings and moves (Ryan 1985:731, 739). Although this methodological premise makes Ryan’s model incompatible with our approach, her study ends with an outline of a hypothetical means of investigation which seems ideally suited to our needs. Ryan considers it perfectly conceivable that an ‘“algorithm” for the processing of narrative discourse’ could be formulated; this algorithm would proceed in nine steps and, both as a model of the natural (human) reception of narratives and as a conceptual specification

92

It is clear that Ryan does not find the dichotomous distinction between state and event sufficiently precise. She subsequently subdivides the latter into narrative events and nonnarrative events and postulates that ‘an event moves the plot forward, and is consequently narrative, when it creates a change in the relations among the constituent planets of the narrative universe… . For an event to be narrative, then, it must either have an impact on the fulfilment of a model-world requirement, or create a new requirement which will determine the character’s choice of an I(ntention)-world’ (Ryan 1985:737).

1.4 The Elements of the Action Construct

105

for a hypothetical computer model, would describe the progressive mental composition of complex narrative universes during the process of reception. What would this program offer us? A program following this algorithm would be able to answer questions about the story at every stage of its reading. As it processes new information, it would correct or complete its representation of previous states and events, so that queries about the same episode made at different moments of the reading would potentially yield different answers.…A program with this ability to form hypotheses, chain events backwards, reconstrue game-plans, and revise its representation of past states or events would fulfill one of the most ambitious projects of textual semiotics: it would capture—as Stanley Fish (1970:14) puts it—‘the temporal flow of the reading process’ and ‘the successive responses of the reader as he goes through the text. As an algorithm for a computer program, this outline is still a utopic construct. Ryan 1985:753f.

The model which Ryan proposes here has a number of problems, among them its treatment of the distinction between state propositions and event propositions, its exclusion of non-narrative events, and, most importantly for our purposes, its foundation on an intentional concept of action. Nonetheless, it is a model whose theory takes into account the potential and highly complex mutual relations between the diverse epistemic or normative worlds (or, better, world models) that can be generated using the perspective of a narrator or actant respectively. The project of a theory that is outlined conceptually and then put into concrete practice in a computer program which mechanically reads action is described as a ‘utopic construct’ for good reason. Ryan refers to it as such not because our technology is not sufficiently advanced to handle the complexity involved (it most certainly is), but because the processes of evaluation and sorting which it requires involve fragments of knowledge that cannot easily be formally defined. After all, the project would assume on the one hand, the existence of a program giving the computer the ability to fully understand natural language on the sentential level, and on the other, a huge data base approximating the knowledge stored in human memory. Its implementation in an artificial intelligence project would require severe limitations on the topic of the story and on its linguistic manifestations. But despite these difficulties for a concrete implementation, the above algorithm suggests that from the present analysis of the modal structure of narrative universes we can derive a set of systematic and reasonably mechanical procedures for organizing narrative information, locating conflicts, advancing explanations for the behavior of characters, and reducing the text to its most important components. Ryan 1985:754

106

Part 1 Concepts of Action

Ryan returns to this ‘utopic project’ again in Ryan (1991); as far as the practical implementation of the theory in a computer model is concerned, the spotlight has now shifted from a descriptive to a generative approach. Although she retains the idea of a formalism for the semantic mapping of plots, her ultimate intention is now to employ the possible worlds approach in the service of a ‘Heuristics of Automatic Story Generation.’93 As was the case with earlier canonical artificial intelligence projects which attempted to use narratological theories to model the cognitive processes of plot reception and construction (Galloway 1979, Lehnert 1981), Ryan does not consider converting her theoretical model into a working computer program. In any case, the benefits of the program would never have been enough to justify the effort needed to make it work. In contrast, this chapter of our study develops two concepts, the event and the episode, which are later put into use in two fully functional software applications, EventParser and EpiTest. This is achieved at the cost of a considerably narrower definition of the phenomena which are modelled: the fictional event, the episode (sequence of fictional events), and (to an extent) the fictional action. There are three main differences between our model and the models of Ryan and Pavel: our approach is based on a non-intentional concept of action; it is formally derived from sequences of state propositions alone; and, instead of algorithmically and prescriptively modelling discrete interpretive or evaluative stages of interpretation, it provides us with a set of descriptive instruments and tools which capture complete concrete acts of human interpretation. In conjunction with the basic data which they generate and a theoretical definition of the minimal action structure, these tools allow the exploration and analysis of combinations of the virtual actions embedded in the propositional event material of a narrative. This means that our project turns away from a purely theoretical definition and model of narrated action to embrace a multi-method approach. Although this approach suggests the presence of a theoretical and conceptual framework, it nonetheless encompasses concrete, empirical acts of reception which can only be reduced to set patterns using very general rules. The aim of theoretically modelling human interpretive acts with the aid of a computer is replaced by the far more modest one of designing terminological and technological tools which, at the cost a considerably narrower scope of study, will allow us to record such acts of interpretation with unprecedented precision and also to investigate their semantic potential.

93

See Ryan (1991:233–57), where she discusses a series of algorithms (including talespin, telltale, and universe) for automatic story generation.

1.4 The Elements of the Action Construct

107

We shall now attempt to define a term for the basic happening which can become an action in a fictional world. Up to this point, we have described a series of terminological antecedents; from the passage a l’acte (Bremond) to the transformation (Todorov), the event (Prince), and the move (Pavel). By and large, they can all be derived from Propp’s concept of the function. We shall take this same concept as our starting point but develop if from a new perspective. The Event as a Construct of Reception As the basic component of action, the event can be considered from the point of view of two basic definitions. These definitions are concerned with two different constructs which can be realized by the recipient; the constructs are formally identical but are found at different functional levels of narration because of their different roles in narrative theory. All relatively complex phenomena of narrative theory have two aspects, as is reflected in the use of terminological distinctions such as narrative time and narrated time, narrative and discourse, and, arguably most fundamentally of all, fabula and syuzhet. The same is true of the event, where we must distinguish between the object event and the discourse event. An object event is an occurrence which manifests itself in the objects which we can locate at the level of symbolically represented fictional objects. A discourse event occurs with each successive linguistic or narrative act which tells us about significative objects and the fictional changes in their properties. From its beginning, the debate about the concept of the event in poetics has been influenced by intuitive convictions which approach the relationship between the object and discourse levels from the perspective of reference theory. Thus, any poetics that is even remotely mimetic takes as its starting point the axiom that a narrative of events is founded on the secure base of a concrete or abstract reality of events which is logically established prior to the narrative and whose causal and chronological order is specified without linguistic expression. Examples of this assumption can be found in the earliest records of philosophical and aesthetic criticism of and reflection on poetics; it is made not only by Plato, but also by his successor, Aristotle. It is true that Aristotle’s poetics, orientated around aesthetic effect, does replace the strict criterion of referential truth with the more abstract and flexible criteria of what is in principle possible and probable. But even from this perspective, the narrated event continues to be seen as the content and object of aesthetic perception, and its aesthetic relevance can only be conceived of given a historically factual or logically virtual archetype and real or possible happening. Only with the radical

108

Part 1 Concepts of Action

reduction of consciousness to language is this interpretation reversed. With transcendental philosophy, Romanticism predicts what eventually became the defining tenet of post-structuralism and deconstructivism (and a welcome addition to the institutional self-esteem of textual criticism): the assertion that the discourse level cannot be circumvented and is itself a plane of reference. The event can now exist only when comprehended in the discursive meta-event, the representative speech act. It seems that language is the first and indispensable source of logically possible events. The world which is not yet structured by language, on the other hand, has no events, no occurrences, and certainly no such thing as the aesthetically evaluated ‘unheard-of event’ called for in Goethe’s definition of the novella which we quoted earlier. According to the new axiom, the idea that the world contains events is no more than an illusion created by the syntagm of changing descriptions of states. Of the two mutually exclusive approaches outlined above, neither the mimetic nor the semiotic approach does justice to a constructive concept of the event, or, for that matter, of any aesthetic phenomenon. There is good reason for the tendency of more advanced models of narrative theory to use interconnected, dialectical definitions of aesthetic perceptions, the phenomena of which they connect to the object and discourse levels. Genette’s analysis of narrative frequency documents this shift in method in terms of the category of time (Genette 1980:113–60). More fundamentally, we should recall Dorrit Cohn’s forceful rejection of the idea that the key feature of the novel is that it is based on an ‘ontologically independent and temporally prior data base of disordered, meaningless happenings that it restructures into order and meaning.’ According to her argument, story and discourse are really concerned with ‘synchronous aspects of fictional texts, with no presumption of priority of story over discourse’ (Cohn 1990: 781). If we choose to adopt this view, we must consider the consequences for our conception of the event. In epic texts, fictional object events can only exist given the presence in linguistic form of discourse events which are turned into mental conceptions of objects during reception. Narrative discourse events, on the other hand, can never occur without objects; when narrating, we necessarily narrate about something which could also have been represented in another language, genre, or even another symbolic system. Of course, narrative discourse can elliptically avoid the objects it marks as events, or it can indicate their presence ex negativo, or it can even make itself a reflexive object of itself. Nonetheless, in all such cases, the discourse refers to an independently given or posited object of the narration or symbolic representation contained in the discourse. Whether this object also fulfils the strict criterion of being empirical (sensually perceptible) turns out to

1.4 The Elements of the Action Construct

109

be a redundant question in the restrictive context of fictional speech. The objectiveness of the narrated event is exhausted in the objectiveness of an ordered mental complex whose plausibility depends solely on its logical coherence, not on any kind of reference to the real world in the ideas which make up the complex. The situation is exactly the opposite in the case of the empirical, real event; here, the redundant question is whether what our senses have perceived as a happening which has factually happened is actually possible (i.e. whether it has a reference in logical space). The intention behind such a question is well expressed by the colloquial German expression Nicht sein kann, was nicht sein darf (‘What is not possible cannot be’): doubts about empirical and logical validity are never directed at the event as an object in itself (unless we are interested in the theological or spiritualistic proof of miracles); they are directed instead at the reliability of our methods of perception and the plausibility of our mental and theoretical representations of the object in question. If, as a signified of event narratives, the narrated event has now become an object (even if only a mental one), this is only partially due to the symbolic referential function which, like all semiotic systems, our language possesses by definition. Arguably more fundamentally, this objectiveness owes its existence to the logical relation of mutual implication between event and object and determines the epistemology of the narrative of events. What a narrative of events really signifies is precisely not the abstractly imagined happening itself, but the object(s) involved in this happening. This indirect objectiveness of the event will guide the development of a definition of the term ‘event’ which diverges from that of the approaches surveyed so far. Event, Property and Matter Let us pause briefly to consider the etymological roots of the word ‘event’. It points back to the Latin evenire (or its nominal form eventus), where the prefix e- connotes the emergence, development, or appearance in the spatio-temporal continuum of a quality which was inherent in the object of which it is predicated. Thus, the English term ‘event’ bears the active entelechial characteristic of the realization of virtual qualities, while the definition which we are developing shows the passive characteristic of being subject to changing, externally caused predications.94 94

As always, we must remember that etymological definitions are generally language-specific and selectively exploit just one of several conceivable historical derivations. Take Ereignis, the German word for ‘event’. Its prefix shows that it involves the gradual exhaustion of a potential, an act such as Er-schöpfen ‘exhausting’, Er-fahren ‘learning’,

110

Part 1 Concepts of Action

But what kind of successive acquisition of properties is involved in events? And to whom do events belong? In its most fundamental sense, the term ‘event’ refers to an objective phenomenon, an external occurrence in the empirical world, that is, it refers to the logical complement of that which is subjectively registered (i.e. in the consciousness of a perceiving subject) as a phenomenon or sequence of phenomena (occurrences) bound to a constant object. By their very nature, phenomena and perception are anthropocentrically defined concepts: they stress the way in which objects are mentally represented, the form and the process by which being enters consciousness. The fictional event, on the other hand, although it too only exists for us in the form of a perception, does not, in the first instance, affirm the existence and identity of the perceiving subject; instead, it refers to a process which logically requires the existence of a concrete object in which it can take place and manifest itself. In view of this mutual dependency of event object and event, Abel suggests adding an event-based ontology to the traditional existential ontology (Abel 1985:172f.).95 We cannot explore here whether this is a legitimate philosophical reorientation that adds to our understanding. Nonetheless, in the practical context of our narratological enquiry, a new approach using an event-based ontology does throw new light on one of the oldest of all problems in poetics. This is the problem of the relative hierarchical position of fictional substance and fictional event in the aesthetics of production and reception. In fact, the new approach supports a long-established postulate of poetics, for it uses the logic of fiction to underline the necessity of the primacy of action over figures which Aristotle attempted to prove with two functional arguments. Aesthetically, he used the criterion of the action’s effect on the audience; philosophically and ethically, he argued (unlike Plato and by referring to the category of truth) that the real achievement of the fictional action narrative lies in its entelechial revelation of superindividual truth. Here, the approach of an event-based ontology points out the fundamental ontological dialectic of existence and event on which fictional narratives depend. It makes clear that the fictional text asserts the existence of fictional objects and persons in order

95

Er-leben ‘experiencing’, or Er-zählen ‘narrating’ which necessarily takes place cumulatively and step by step, not in the mode of a single complex perception. In addition, the Er- indicates that the process is also one of positive or negative acquisition, essential self-change. Erfahren is different from lernen ‘study’, erleben from sehen ‘see’, erzählen from reden ‘talk’. Unlike processes of sensory perception, processes of the epistemic acquisition of properties temporarily call the boundary between the epistemic subject and the apperceived object into question and make the subject itself accessible. Our perceptions remain external to us, but what we learn is internalized by us, shapes us, and defines our identity. See also chapter 1.3 above.

1.4 The Elements of the Action Construct

111

to use them to demonstrate possible events, happenings, and ultimately action, and, conversely, that the recipient has to acknowledge the possible existence of fictional actors and objects if he is to accept the existence of the fictional events asserted by the text as possible and involve himself in them. The intradiegetic relationship between fictional event and fictional being is therefore analogous to the extradiegetic relationship between action construct and (receiving) I. The fictional event marks the beginning of the constructive spectrum, the complex action its end. Arguing equally paralogically, the former construct amounts primarily, but not exclusively, to a postulate of the fictional existence of the actors and objects of the action, while the latter construct seems to prove the unchanged identity of the real consciousness which finds an affirmation of itself in its ability to synthesize action logic. The dialectic of the related basis of event and existence can be substantiated semiotically as well as epistemologically. We find an early example of this is in Lessing’s Laokoon. For Lessing, the principle (i.e. semiotic) difference between painting and literature is most clearly expressed by the different possibilities for representing actions in the two disciplines. No art can completely reproduce every aspect of the spatial and temporal situation of a given object. While painting is able to capture a multitude of spatial dimensions but only one temporal position in a synthetic representation, writing can portray a body (an object) over the course of time but does so at the cost of losing track of the object’s many concurrent spatial dimensions. This difference is due—Lessing’s argument is important here—not to some kind of ontic difference between the two arts, but to the different nature of the signs which are available to them.96 Thus Lessing realizes that it is precisely in the context of aesthetic representation that the poetic portrayal of action connotes the presence of a body, just as the representation of a body can, but need not, connote the presence of an action.97 Lessing continued to use a mimetic concept of representation as the basis for his deliberations; they therefore suggest that certain aspects of an object or fact exist prior to and outside of its representation. The event-based ontology of our reception theory approach goes a decisive step further: it emphasizes that the event’s true function lies in the ontology of fiction. In other words, once recipients accept the convention of the event, they are able to accept, without having to turn to real referents in the real world

96

97

Lessing develops this idea most concisely in the second and third chapters of the Laokoon; see Lessing 1973, 6:564f. The extension of Lessing’s concept of action into a dualism of static and moved action which removes the need for the caveat is described by Mendelsohn in Lessing (1973, 6: 566, note *).

112

Part 1 Concepts of Action

for support, the assertions made by the narrative about the existence of narrated things or persons in the fictional world. In fact, the participation of fictional existences in events is a necessary requirement of aesthetic illusion. Only when something appears to us in a fictional event as the bearer of changing features (predicates) does it confirm its existential character, which is otherwise no more than hypothetical. When the event is understood as a construct of perception and reception in this sense, we shall refer to it as an event, which we provisionally define as follows: Provisional Definition: an event is the assignment of distinct properties to an event object which remains identical over the course of time.98

From a logical point of view, the question of what comes first (the event or its event object) is meaningless. Each assumes the other; the dialectic cannot be circumvented. But in terms of the aesthetics of reception and the logic (or ontology) of fiction, things are different, as we know from Propp and his definition of the narrative function, if not indeed from even earlier critics. As a simple sequence of distinct predications, it is clear that an event can manifest itself in any object whatsoever. I can narrate about

98

From a purely formal point of view, we can distinguish between the extensional and intensional content of the event concept. A narrative of events does not just refer enumeratively to distinct sets of time/space/objects/properties; it also formally connotes the mental construct of the unchanging existence of (at least) one event object. Brooks (1984) and Coste (1989), both clearly influenced by Lacan, claim to see a narrative paradox in this, given that a narrative statement asserts at once the identity and difference of the I narrating itself. Like most such paradoxes, this one is nothing more than a variant of the problem of private speech; it is contrived using the special case of reflexive narration that thematizes the narrating I. And besides, the fact that one and the same external object can appear under changing (narrative) descriptions and yet appear identical would probably not have been sufficiently dramatic on its own to justify proclaiming the discovery of a narratology that claims to be ‘one of the best ways of understanding the individual perception of mortality and its consequences,’ as Coste (1989:11) puts it with all seriousness. (Bal 1992 contains a venomous criticism of this approach under the illuminating title “Narratology and the Rhetoric of Trashing”). If we disregard these metaphysical adventures, however, we do find that Coste’s work concurs with our thesis of an identity postulate which is formally implied by the narration of events. Coste uses arguments from communication theory to arrive at the formulation: ‘In order to understand “X has changed”, I have to accept that X remains X whether p or—p is true about it’ (Coste 1989:9). In this case, the identity of the argument of two successive propositions is a necessary logical prerequisite of the event, but not one on which the coherence of discourse about complex sequences of events and states depends. Van Dijk in particular points this out and notes that ‘referential identity is merely one aspect of textual coherence; it is neither necessary nor sufficient’ (van Dijk 1980:36), for it is perfectly possible that coherence is created not by the identity of the referents but by that of the predicates or the modes of the propositions.

1.4 The Elements of the Action Construct

113

events which have real or fictional objects, but just as easily about those that take an object variable as an ontological placeholder (There was once an x / then the x died).99 The event structure is preserved and exists in this form regardless of whether or not the variable x is fictionally instantiated (and the placeholder filled by an object). Conversely, I cannot narrate about an object without making it the subject of a proposition and assigning at least one property to it (a predicate such as being, fictional existence, or even absence, non-existence, etc.). But if I predicate something of this object, I automatically introduce the object as the object of at least one virtual event which can later suspend or transform the first predication. Thus, no object about which I narrate can be found alone and prior to events; even the simple assertion of an object’s fictional existence must be seen as an eventful fact from the point of view of the theory of perception, not only because such an assertion is itself always a discourse event, but also because it amounts to the virtual transformation of the existential predicate and is thus an event which involves the object.100 These ideas are important for narratology, and clarifying them with the methods of an event-based ontology not only means investigating, in the strictly philosophical sense of ontology, the existential conditions of fictional events; it also requires us to consider the foundation which these events provide for the existence of fictional things. In another respect, too, we must consider the concept of an event-based ontology in more detail. By convention, the objects of non-referential, fictional narration can be considered as possible, but not as true objects; they are the purely imaginary content of speech. This alone, even without the assumption of a constructive concept of action, is enough to direct a new approach of event-based ontology towards an epistemological rather than metaphysical definition of the event. Our interest is directed not at the question of what fictional events and objects are and under what conditions they can be, but rather at how and under

99

100

As already mentioned, the most emphatic development of this idea can be found in Todorov, whose concept of the narrative transformation explicitly refers not to a change in the properties of fictional persons or objects, but to a change in the properties of (grammatical) predicates (i.e. verb phrases) in which the predicate itself necessarily remains identical while being transformed modally or by the addition of extensions (such as auxiliary verbs) (Todorov 1969:224). In this sense, the convention of narration can be seen as a requirement to predicate individuating predicates of (fictional) things, persons, and other objects of perception. We need, however, to investigate whether this is not also the case in other genres and text types. In lyric texts, for example, enumerations of generic terms without accompanying predicates are perfectly legitimate, but do these terms not actually appear in the predicate rather than the argument position and therefore have the purpose of effecting the predication of the lyric I?

114

Part 1 Concepts of Action

what conditions we as the recipients of a narrative text imagine them to be existent and in this unconventional sense perceive them.101 Bearing this in mind, we must consider again the distinction between object events and discourse events which we introduced above and relate it to our narrower definition of the event construct. Since formalism, the standard method of distinguishing fabula and syuzhet has been to examine the different ways in which they structure narrative time and narrated time in the form of event sequences. In the fabula, the chronological order of event elements is governed by fictional chronology and causality; in the syuzhet, it is governed by a rhetorical intention which can be traced back beyond fiction to the perfectly real causality which controls situations of aesthetic production and reception.102 We might imagine that a corresponding distinction pertains in the case of the event. If this were so, object events would be those events in which the chronological sequence of changing properties follows the fictional world’s internal chronology (i.e. the narrated time of the fictional world). For example, a hero could be rich at point t in time, but poor at point t + 1, where the sequence of all possible points t1…tn in time is congruent with the natural course of time in the fictional world. discourse events, on the other hand, would be created by a sequence of attributions of changing properties to objects, where the temporal structure of the sequence is congruent with the syntagmatic system of the language of narration, congruent, that is, with a narrative time which encompasses the continuum e1…en. Whether a sequence of states of affairs constitutes an object event or discourse event then depends basically on which of the two temporal continua contains the temporal indices of the elements in the sequence. However, such a systematization using the criterion of time is not only insufficient; it is also based on problematic assumptions. There is one and only one continuous, unilinear time: no matter how complex, anisochronous, or iterative the narration, the beginning of every narrative will always be temporally 101

102

On this fundamental modification of the concept of ontology, see among others Rapaport and Shapiro, who, in the context of their artificial intelligence project of modelling cognitive processes, introduce a corresponding distinction between metaphysical and epistemological ontology: ‘Corresponding to the purely or metaphysically ontological question, “What are fictional objects?,” we ask the epistemologically ontological question, “How does a cognitive agent represent fictional objects?’. And corresponding to the purely ontological question, “How are properties predicated of fictional objects?”, we ask the epistemologically ontological question, “How does a cognitive agent represent the predication of properties of fictional objects?”’ (Rapaport and Shapiro 1995:109). Models of cognitive theory, such as that of Rapaport and Shapiro, must of course be seen against the background of their indirect, behaviourist use of basic cognitive data, as is pointed out in Simon and Wallach (1999). The terminology chosen by Schmid (1982) gives particular emphasis to the rhetorical aspect of his model.

1.4 The Elements of the Action Construct

115

located after the beginning of the narrated happenings, and the end of the narrated happenings will always be temporally located before the end of the narrative.103 It is true that narrated time and narrative time both consist of sequences of points in time which can be coextensive, partially overlapping, or even completely separate. But the individual points in time out of which these sequences are formed are always located on one and the same temporal plane and form a continuum in it. Were this not so, it would be meaningless to talk about the effects of accumulation, dilation, iteration, among many others. There would also be fatal consequences for the current practice of back-translating the ordo artificialis of the syuzhet into the fictional ordo naturalis of the fabula, which relies on the tacit assumption that the syntactic and the causal, chronological order are mapped onto one another by aesthetically discretionary transformations rather than denotational relations. It would no longer be possible to reconstruct a natural sequence of causal chronology unproblematically, for doing so would involve interpretation. Thus, while the distinction between narrative time and narrated time is unquestionably useful for analytical purposes, it must not be misinterpreted as implying the existence of two ontologically distinct and autonomous times. It follows that we cannot distinguish between the two kinds of event by assuming that they have different temporal specifications. Both the object event and the discourse event have structurally identical ordered sequences of predicates which embody instances of the acquisition of properties. However, the two event types are distinguished from one another by the fact that the acquisition processes concern subjects (event objects) which are located at completely different ontic levels. discourse events are created not by the changing predicates predicated of fictional event objects (things, persons, actors, etc.) but by the changing predicates of knowledge or experience which are acquired by an agent or patient who acts in an increasingly real manner. We can list, in hierarchical order, the following objects which take shape during the course of discourse events, the source of their constant individual existence: the narrating figure, the narrator, the author, and, finally, the recipient himself.104 In a 103

104

Genuine prophecies are an interesting exception to this in so far as they narrate what can by definition only happen after the end of the narrative. Prophecy must not be confused with prolepsis, which Genette (1980:79f.) says (without good reason, in my view) borders on anachronism when it appears together with analepsis as remembered prediction or anticipated memory. Genette’s approach is orientated around discourse theory and establishes the being of the narrator as a function of voice (and expression). We can now place an approach couched in terms of perception theory alongside it: every speech act of a narrator can be traced back to a (fictional) act of perception: the narrator reports on something he (allegedly or actually) saw, experienced, learnt, or at least imagined to himself.

116

Part 1 Concepts of Action

discourse event, one of these authorities asserts itself as an identical experiential focus throughout the sequence of unconnected predicates predicated of arbitrary fictional objects. The most obvious and concrete way in which this can happen is the creation of a first-person narrator; the most abstract and subtle is the inversion of the event structure itself so that, instead of various properties being attributed to an identical thing over the course of time, an identical property in the continuum of experience is predicated of various things. If this happens, it is no longer the narrating authority that adopts an identical existence throughout the sequence of predicates; it is the authority of the hierarchically and historically final perceiving consciousness itself, the instance of the recipient. The transition from syntagm to theme and paradigm, from fictional object event to real discourse event is thus marked by a fundamental change in focus. We switch from observing the sequence of changing predicates predicated of (fictional) event objects to treating the sequence of statements as a sequence of predicates predicated of our own receiving consciousness, which now adopts changing states of knowledge and consciousness. object events and discourse events, therefore, are no more hermetically different from one another than narrated time and narrative time. They are synthetic constructs which are built out of the same set of material supplied by narrative statements; from the point of view of an event-based ontology, however, they are aimed at affirming categorially different hypotheses of existence. Viewing them in this way means that it is problematic to use the conspicuously reifying phraseology adopted above, which treats the existence of event objects as confirmed in events. Similarly to Pavel’s move grammar, which uses the concept of the domain to re-establish the place of the intentionally acting figure, this kind of language implies the autonomous existence of a (fictional) actor and subject, given that he is assembled and constructed only as the reception of events progresses. For this reason, we shall now introduce the more abstract concept of event focus. The form of the event focus is influenced and affected, with a greater or lesser degree of persistence, by the text, the constructor (i.e. recipient) always has the last word. By positing an event focus, the interpreter specifies the hypothetical event object that is to appear as the bearer of the changing predications in an event construct. Correspondingly, we reformulate our preliminary definition of the event as follows: Definition: by an event we understand the attribution of distinct properties to an identical event object under a stable event focus.105

105

In this very general form, our definition is identical with a number of other suggested definitions, such as that of Stempel (1973).

1.4 The Elements of the Action Construct

117

Conditions of an event We shall now consider the conditions which must be met in order for a complete event to be constructed. For the present, we shall confine ourselves to the investigation of object events. In formal terms, our inductive concept of the event initially requires no more, and no fewer, than two non-identical statements of predication: (a) (b)

x(P1, t1)—read: ‘there is an x with property P1 at point t1 in time.’ y(P2, t2)—read: ‘there is a y with property P2 at point t2 in time.’

These two statements of predication can be synthesized to form an event under certain conditions, namely: (a) Order: the temporal index (chronological position) of t2 must succeed that of t1.106 (b) Identity of event focus: the variables x and y must be instantiated by an identical hypothetical event object. (g) Non-identity of predicates: P1 and P2 must be different. So far, our formal descriptions have—hopefully—corresponded to our intuitive understanding of an event as a change applied to the properties of an object. The object itself, however, does not adopt this transformation of properties; it does so only from the perspective of our perception. This is the basic problem of practically implementing an event-based ontology, and the relatively easily tested conditions (a), (b), and (g) are, unfortunately, not sufficient to deal with it. P1 and P2 must not only be non-identical; in addition to meeting this criterion, they must also be comparable with one another in some way. The comparability can be asserted directly, based on logical reasoning, or learnt. 106

It seems intuitively obvious to assume that the natural chronological order of the components of an event defines the essence of that event. In theoretical terms, however, this assumption is questionable. For example, the transformation poor/rich and its inverse transformation rich/poor are, as far as the elements they combine and their ordering are concerned, not only structurally but also functionally identical: they represent the same semantic distinction. Chronology is not relevant to singular events considered in isolation; it only becomes relevant in the context of the combinatorial possibilities made available by transformations; these combinatorial possibilities can be forward links to subsequent events and backward links to previous ones. The chronological order of statements of states of affairs no longer defines the event itself (the formal criterion of different temporal indices is sufficient for that); instead, it limits the possibilities of connecting the event into the superstructure of the overall sequence. A footnote is not the place to discuss the philosophical problems concerning the presumably natural event chronology; for our purposes, it is sufficient to note that we should avoid speaking of the perception of quasi-natural (and thus presumably epistemically unproblematic) points in time. Instead, and more cautiously, we should employ interpretively assigned temporal indices.

118

Part 1 Concepts of Action

Let us begin by considering the last of these three possibilities, learned comparability, which we shall explain with what may initially appear to be a somewhat extreme example. If a protagonist is initially poor, but later possesses the property of being able to talk to fish, it is at first glance hardly conceivable that two such disparate situations could be made to belong to a single event construct. The reader who is familiar with the fairytale Vom Fischer un sine Fru, however, knows that both descriptions can quite easily be connected via a sequence of linking events, given that able to talk to fish leads to being rich in this text (thus establishing a perfectly plausible change from poor to rich). This knowledge, once gained, does not fade; it describes a valid event which we treat as remaining possible for as long as we can imagine. Quite apart from its ethical and moral axioms, this fairytale of the Grimms also implicitly teaches us about a transformational rule (albeit one that is to a certain extent fantastic, in fact fairytale). According to this rule, the predicate poor can be transformed into the predicate able to talk to fish. Granted, in Vom Fischer un sine Fru, this transformation is derived somewhat indirectly and is anything but a basic constituent event of the action structure. In this particular instance there is no construction of an event in real time as the narrative unfolds; we are confronted immediately with a highly complex action construct. But nonetheless, once it is learnt or constructed by a recipient, this action is absorbed into our cultural world knowledge in the abbreviated form of a set of abstract transformational rules. We can retrieve it from memory and use it to build basic event constructs on future occasions. Just like all other relations of opposition and equivalence between the terms of symbolic systems, the semantic relations between the predicates that make up events are learnt and culturally acquired; they are the products of historical and biographical processes, and only in exceptional cases are they likely to be innately predetermined universals. Our reference to the learnability, the intertextual transferability, and the conventional nature of the transformational relations between predicates has still not answered the fundamental question: what are the logical conditions which govern a sequence of predications which come under an identical event focus and are thereby bound to an identical event object? Before we specify the detailed mechanics of the comparability condition (d) to be formulated below, we shall return yet again to our first, long-suffering textual example and formally describe the event it contains:107

107

In order to retain a stable event focus, we must extrapolate the prior state The king lived from Forster’s The king died. Then the queen died.

1.4 The Elements of the Action Construct

119

(First) the king is alive = king(alive, t1) (Then) the king is dead = king(dead, t2)108

In the transformation from alive to dead, we meet the simplest and most unambiguous type of comparability-preserving non-identity: P2 is a function of P1. In the most general definition of mathematics or set theory, a function F assigns to every element of a set A (the domain) a unique element of a set B (the codomain). It is easy to illustrate how and according to what rule this takes place in the present example: an element of A (alive) is translated into the element of B which is its exact opposite (dead). Despite their obvious non-identity, P1 = alive and P2 = dead can, therefore, be compared because they can be translated into one another by the negation rule F. This brings us to a general definition of our comparability condition: (d)

Predicate mapping: P1 must be translatable into P2 according to a rule F. Class-Homogenous events

Predications which are combined in pairs and which satisfy condition (d) as well as the first three criteria can now be combined to form an event construct. At this point, we must make a fundamental distinction between class-homogenous and class-heterogenous constructs. Our above example The king was alive. (Then) the king died is an example of the translation of a predicate P1 (alive) into a class-homogenous predicate P2 (dead). What is the basis of this homogeneity? We shall not approach this question from the point of view of the philosophy of language (or any other kind of philosophy); instead, we shall examine it from a pragmatic perspective. The following discussion takes it as given that the formation of classes or paradigms not only expresses the structure of the world as we perceive it but also drives the process with which we structure it. According to the generally recognized conventions of non-metaphorical expression, predicates like alive and dead can, at least in our language, be assigned only to terms which denote organic objects, that is, things whose nature can be predicated as biological. The term biological is a classificatory predicate, 108

The assumption of an axiom of chronological sequentiality is clearly apparent here. According to our modern world knowledge, people live before they die. Applying this axiomatic postulate to our example, however, results in a dangerous contradiction with the axioms of the genre of the fairytale, from which the example was taken in the first place—in the world of the fairytale, it is accepted that people can quite easily be brought back to life (even though a death still presupposes a life).

120

Part 1 Concepts of Action

not an individuating predicate. There is an infinite number of (fictional) objects to which we can assign the generic predicate biological; the individual biological things, however, are distinguished from one another by being biological in different ways—human, animal, old, young, dead, alive, male, female, to name but a few.109 In order for two biological objects to be distinguished from one another, the objects must be differentiated in at least one way, that is, by at least one individuating predicate. The sum of all the predicate classes which can be drawn on for the predication of an object therefore contains the set of theoretically possible individuating predicates of that object.110 All the elements of a given predicate class (i.e. the individuating predicates which are members of that class) can appear as completions of the generic class predicate. Thus, in the present example, kings are biological things and therefore belong to a class of objects of perception to which, in addition to the highly general biological attribute, individuating predicates such as dead or alive can be assigned in order to create the potential basis for the formation of an event construct. In fact, without individuating predicates, there are not just no individual things; there are also no events. It is of course quite possible to apply class predicates without using a supplemental individuating predicate, in which case the class predicate is applied to a generic term. But if we apply a class predicate to an object to be individuated without also adding a specific element of the set of predicates attached to the class, we arrive at a redundant expression of the form x = king(biological, – , tn). The first and second non-redundant predications, the necessary prerequisites of an event construct, must therefore have the form ‘event focus(predicate class, predicate, temporal index)’ or, in the current example: (1)

(2)

109

110

The king was alive. [event focus = king (predicate class = biological, predicate = alive, t1)] Then the king was dead. [event focus = king (predicate class = biological, predicate = dead, t2)] The distinction between generic and individuating predicates corresponds to the distinction in van Inwagen (1977) between the metafictional predication of a fictional object as a fictionally established object of perception and the intrafictional ascription of properties which then distinguish the object inside the narrated world. See also the critical discussion in Rapaport and Shapiro (1995:110f.). In a very abstract sense, we could see this as the creation of a virtual semantic space, for the conventional determination of what classes of predicate are acceptable leads to the exclusion of other classes of predicate. Every assignment of a predicate from a false or unacceptable predicate class would then constitute the crossing of a border in Lotman’s sense. A talking stone per se would be just as much an event as a sleeping beauty who

1.4 The Elements of the Action Construct

121

In its general form, our translation rule (d) requires that P1 must be translatable into P2 according to a rule F if an event construct is to be formed. We can now specify what F is in the case of a class-homogenous construct, which, in addition to the condition of non-identity, requires that conditions d1 and d2 below must be met: (d1) P1 must be mapped into PC and onto P2 by a rule F. (d2) P1 and P2 must be elements of the same class of predicates PC. If we were to continue developing these ideas in terms of set theory, we would need, among other things, to distinguish between the isomorphic and non-isomorphic mapping of P1 onto P2. If we try to apply such formalizations to concrete processes of reception and event construction, however, they give rise to the illusion that predicate classes, their respective elements, and the rules that translate between them are invariant and universally available, whereas in reality they are not. The formation of event constructs is a dynamic process, which means that during it we can draw on existing predicate classes, elements, and translation rules, as well as making new classes and defining new rules. A translation rule which is firmly established inside a homogeneous predicate class from the very beginning is so only because of convention and encompasses only a fragment of what is theoretically possible. For example, there is no doubt that the biological predicate class, according to our definition above, contains elements such as alive, warm, human, male, old, female, dead. The negative mapping of P1 onto P2 according to a translation rule makes it acceptable to combine many terms of this class into pairs of opposites which can be drawn on at any time (e.g. dead/alive, male/female). Even inside a class, however, it is perfectly possible for new, previously unknown mappings to be found, such as dead/warm, which borders on the addition of the element cold to the predicate class. The predictable objection is that this translation is effected by resorting to a second, supplemental rule, the rule of equivalent mapping, which goes beyond the negation rule and states that warm = alive and cold = dead are valid inside this predicate class. I do not find this convincing, for the relations of opposition and equivalence

had been lifeless, but undead, for a hundred years—provided that neither stone nor sleeping beauty is met in a fairytale. These examples show that the question of the acceptability of a particular predicate class cannot be answered in absolute terms, only, at least as far as narrated worlds are concerned, with reference to ontological norms such as those of genre. In addition, though, I feel that the simple singular use of an unacceptable predicate class such as x = king (inorganic, of stone) is enough to create a discourse event even though no transformation of the king, and therefore no object related event whatsoever, has manifested itself in the fictional world.

122

Part 1 Concepts of Action

in which the elements of a binary table of predicates can participate are just as much a question of the cultural, aesthetic, and ideological conventions stated or assumed as they are a consequence of the real necessity of synthesizing statements of predication into an event construct. In short, the formation of these constructs obeys a logic that is not fixed but dynamic and combinatorial. This logic is fascinating not because of predetermined necessities but because of constantly changing logical possibilities, for the latter are the ultimate basis of the cognitive diversity created by reading events and actions. We can gain some impression of the dimensions of this diversity with the help of a simple calculation. Let us assume that our king is defined in a sequence of sentences with different temporal indices using four predications which draw exclusively on elements of the biological predicate class. That is to say, the narrator reports that the king is first alive, then full up, then warm, and then finally dead. In theory, alive can enter into three combinations; our king could undergo a transformation from alive to full up; or from alive to warm; or from alive to dead.

Or the king could remain alive and change from full up to warm; or from full up to dead.

And if this is not the case, he can still undergo the change from warm to dead.

Thus, using the formula x = n(n – 1) / 2, from a single event object and four successive predications (n = 4), we obtain a maximum of six possible binary constructs. There is no question that our preference here is to read the transformation from alive to dead as an event and treat the predicates full up or warm as adjectives that extend alive, with which they are therefore synonymous in terms of event logic. We do not interpret them as elements of independent new predications of an object (the king) that was previously predicated as alive. So, we combine several predicates under a single temporal index and ascribe their sequential order to the syntactic constraints of language, which, by its very nature, requires the clauses to follow one another (assuming, of course, that the linguistic form of the individual clauses allows this and does not explicitly assert a relation of causality or temporal succession between the properties). Significantly, however, our preference for one of the six theoretically possible constructs is, at least at a basic level, the outcome of a free choice rather

1.4 The Elements of the Action Construct

123

than a logical necessity.111 When I construct only one of six theoretically possible events, I make a decision with far-reaching consequences for the combinatorial possibilities which will be subsequently available and the dimensions of the resultant action construct.112 Choosing one of six possible events does not sound particularly dramatic until we consider that the number of theoretically possible combinations increases exponentially with every additional predication. If our king is the subject of, say, six successive predications instead of four, then we have fifteen virtual binary event constructs, considerably more than even the most ambitious reader could ever instantiate. This is clear even before we take into account the additional combinatorial possibilities that result from the compound predications (e.g. The king was alive / then he was full up and warm) which appear in a complex action narrative. The contrast between the number of virtually possible combinatorial event constructs and the significantly smaller number of constructs which can actually be realized in reality gives us an idea of the quantitative magnitude of the construction of basic events which takes place in textual interpretation. It also underlines the fact that object events and discourse events are members of a multiplicity defined by the tension between its combinatorial basis and the pragmatic limits of perception. Both a happening involving a (fictional) object and the subsequent interpretation of that happening in the construction of an event are productive in terms of the ontology of events: they are processes which confirm the existence of the object. Even an example as simple as that of the binary, class-homogenous construct discussed above is enough to illustrate the pragmatic cognitive condition which is necessary for us to identify the constant being of an event object throughout a sequence of discrete perceptions. This condition 111

112

As soon as we start to consider more complex, contextually bound sequences of predication, we must consider logical conditions such as the requirement that a negative mapping of alive onto dead or dead onto warm must not represent a contradiction. There is a long history of investigation into the question of how many alternative constructs of a given complexity humans can instantiate and how long they can then be mentally retained. It has been the from different angles in cognitive theory, cognitive psychology, and artificial intelligence research. Hasselberg-Weyandt (1985) gives an overview (now in need of revision) of the relevant psychological research on story retention; a similar study in reception theory can be found in Löffler (1988), who examines the question of how story structures are perceived. The more specialized question of the extent to which reception processes require a polyvalence convention is investigated by Meutsch and Schmidt (1985). The contrasting method of point-driven understanding is outlined in Vipond and Hunt (1987). The discussion in Rapaport and Shapiro (1995) is concerned specifically with the field of literary cognitive processes and fictional narratives; they present a fascinating prolog computer model of the cognition processes involved. Recent developments in cognitive theory and artificial intelligence research are outlined in, among others, Simon and Wallach (1999).

124

Part 1 Concepts of Action

is the precise, predominantly culturally determined selection of a small number of available constructs from the enormous set of theoretically possible virtual constructs. Our construction of the fictional existence of narrated actors and our experience of our own real existence as interpreters are both made possible by this selective reduction. Our ability to read and construct events which are subsequently combined into a complex action is thus due to something more than a free choice to enjoy an aesthetic experience—the epistemological and pragmatic conditions of perception to which the process of construction is subject are one and the same as those which are so crucial in allowing us to experience the essential identity of our own conscious existence in reality. Class-Heterogenous events The importance of arbitrariness in construction and selection is even more apparent in the case of class-heterogenous constructs. This type of construct requires that, in addition to the non-identity of the predicates P1 and P2, both d1 and (instead of d2) a new rule d3 be satisfied: (d1) P1 can be mapped into PC and onto P2 by a rule F. (d3) P1 and P2 are elements of different predicate classes PC1 and PC2. We have already met an example of such a class-heterogenous event construct in the fairytale Vom Fischer un sine Fru. However, because the connection between the predicates poor and able to communicate with other species is somewhat difficult to express in the formal language of a hypothetical construct, we shall turn once again to our king for help in providing two new example sentences: The king was full up. Then the king was curious. In class-homogenous constructs, the domain and the codomain are generally identical: theoretically, P1 can be translated into every element of its own class apart from itself.113 In class-heterogenous event constructs, on the other hand, we are concerned with non-identical predicate classes: 113

We could omit this restriction, thereby raising the possibility of defining indeterminate events—which can be narrated, at least at the level of object events, by redundant or iterative predications—as the result of the mapping of a predicate onto itself. In addition, the introduction of indeterminate events would make it possible to use the terminology of an event-based ontology to describe the familiar fact that, once narrated, the predication of an event object generally remains in force until it is explicitly transformed (once our king is alive, he will not leave this world and quietly retire: so long as he is not expressis verbis narrated as dead, we can quite reasonably assume that he is still alive). As I have said, this is a theoretically interesting possibility, but for the sake of clarity we shall not explore it further here.

1.4 The Elements of the Action Construct

(3)

(4)

125

The king was full up. [event focus = king (predicate class = biological, predicate = full up, t1)] Then the king was curious. [event focus = king (predicate class = cognitive, predicate = curious, t2)]

There are three ways in which an event can be derived from these two statements. The first is to assume that we are dealing with an instance of ellipsis: the narrator has neglected to tell us that, before point t2 in time, the king was not only full up but also not curious. This resolution of the ellipsis to complete the narrative would enable the class-homogenous mapping of a hypothetical not curious onto the explicit curious. Furthermore, the king would remain full up and, if the narrative were not continued, would remain so for the foreseeable future. This is good for the king, but not particularly good for our efforts to prove that we can, indeed must, globally synthesize predications into an event construct which is as coherent as possible. We have indeed formed such a construct at the insignificant cost of making a single additional assumption, but this has left us with a stranded element in our example, the first predication, (3), which is now essentially redundant and serves no function whatsoever. This amounts to dealing with our problem by ignoring part of it, and that is not a satisfactory solution. The second alternative is to retain the additional assumption outlined above and extend it by saying that the king was not curious because he had previously been full up. This introduces two supplemental hypotheses: first, that the narrative is elliptical; second, that there is a causal connection between having a full stomach and not being curious. To fully solve the problem, we have then to formulate a third supplemental hypothesis and assume another ellipsis: at some point before point t2 in time, the king must have ceased to be full up. We have therefore had to formulate three supplemental hypotheses in order to be able combine two simple predications into an event. This does not only make the king hungry; it also leaves the recipient clamouring for more as he runs out of assumptions to draw on for assistance. Let us now consider the third possibility. It does not require supplemental hypotheses to be made about elliptically omitted states of affairs and hypothetical causal relations in the fictional world; instead, it postulates a new class-heterogenous translation rule: being full up is defined simply as the metaphorical opposite of being curious. He who is biologically full up lacks the cognitive hunger of curiosity. The new translation rule F which has just been stated is, in this case, easily identifiable as the expression of a stereotype which sees physical asceticism as the way to cognitive

126

Part 1 Concepts of Action

riches—a characteristic motif of Judaeo-Christian religions. However, it is also conceivable, at least theoretically, that a recipient or narrator required to compose a coherent event construct could independently create such a class-crossing translation rule and thereby prepare the way for its subsequent evolution into a stereotype. This is the case, for example, in the structuralist reading which Claude Lévi-Strauss applied to the Oedipus myth. When first presented, it was a matter of controversy, but it can now be considered required knowledge in members of the academic intelligentsia. In quantitative terms, the introduction of class-homogenous translation rules leads to a significant expansion of the set of virtual event constructs. In our example of a class-homogenous construct, we were dealing with four predicates of the biological class (alive, full up, warm, dead) and six possible combinatorial variants. With the introduction of metaphorical translation rules which cross classes, the classes of biological and cognitive predicates become compatible; we must therefore include the four cognitive predicates (curious, clever, educated, not curious) in our calculations. The number of elements which can be combined is merely doubled to eight, but the number of possible binary combinations of them is more than quadrupled: it leaps from six to twenty-eight. At this point, there are two possible objections to be dealt with. One is practical, the other methodological. First, not all the combinations which can be theoretically generated can be practically realized as events: other, more important guiding factors constrain the set of possible constructs more and more as the complexity of the text increases. These factors may, for example, include assumptions about the plausibility of possible constructs which contradict psychological norms or the laws of nature. Second, it may be thought that our demonstration of the combinatorial basis of the set of virtual constructs suggests it is possible to calculate all the possible theoretical meanings of a text. On this point, the practical application of our model in part 3 will conclusively allay any suspicion that we are engaged in the formulation a hermeneutic algorithm; in a dynamic model of action, there is no place for the idea that a text has a finite number of meanings. The action constructed by a recipient in a particular case must always be seen as the expression of a differential rather than definitive concept of the text’s meaning. Theoretically speaking, reading an event or an action means rejecting an infinitely large number of equally possible alternative constructs. In fact, therefore, it is the long-established custom of turning to one or two canonical reconstructions of the fabula whose methods are speculative and suspect, not the new approach we propose. The hermeneutic process begins, as our discussion has shown, not at the relatively late stage of engagement with the syuzhet. It begins with the event, the most simple construct of any narrative.

1.4 The Elements of the Action Construct

127

The Status of the Translation Rule As the example of the class-heterogenous event construct has just shown, we need a more precise definition of the concept of the translation rule which we have introduced. The translation rule is not a generally valid, ahistorical law; it is a perceived or assumed regularity. One possibility is that this regularity is predetermined, in which case it is stored in the fields of cultural and, in the case of literature, genre-specific knowledge to which the recipient has access.114 Being dead is the opposite of being alive; one is translated into the other by a translation rule. While we may think that we just know such things, we must learn them in some way or another. In its most rigorous form, our pool of knowledge consists of constant pairs of opposites (e.g. dead/alive, poor/rich, big/small, 0/1, etc.) which can be grouped into static culture-specific or even universal paradigms in the manner of Lévi-Strauss. Since his work, the paradigm has grown to be an indispensable theoretical construct. It is the most effective conceptual tool for drawing at first sight non-identical predicates and objects together in a single class. It achieves this by using a common factor shared by all elements and a constant transformational rule. The resultant classes consist of (binary, ternary, etc.) groups of elements. But paradigms are abstractions. As classes, they assume the existence of elements (which, relative to the class concept, are concrete), not the other way round. In other words, human readers never, in normal circumstances, perform the semiotic equivalent of swimming on dry land by learning culturally and/or universally valid paradigms and their inventories in order to apply this knowledge to distinguish events from other things. There is no doubt that our knowledge of the existence and content of such paradigms plays an 114

As far as this area of knowledge is concerned, Lotman’s concept of the event as a crossing of the borders of semantic space can still, if indirectly, be reconciled with the concept of the event which we are describing. However, a more fundamental difference between the two approaches also emerges, for Lotman’s semantic space is not defined, say, by the complete set of abstract, bidirectional translation rules; it is concretely brought into being by the predicated world. In the context of the current example, being alive would be known and assumed as the standard state; it would, more or less, be the default setting in the fictional world; being dead would then constitute a considerable deviation from it. Quite apart from the question of how great this deviance must be in order to be qualitatively considered an event in Lotman’s sense, I see two key problems here. First, there is the issue of genre ontology (the normal state of the fairytale world, for example, is universally known to be different from that of the realistic novel); second, the theoretical exclusion of redundant happenings (i.e. those which confirm the default setting), which obviously cannot have the character of events. This is consistent only if we, like Lotman, define the event from the perspective of an aesthetically satisfying syuzhet and thereby formulate an aesthetic, critical concept of the event rather than a formal one.

128

Part 1 Concepts of Action

increasingly important role in our life as we grow older—this is why we increasingly read events (be they symbolically represented or experienced in real life) against the background of the classes which we have previously formed and which allow us to store our knowledge in the form of formal representations. However, reading events in the narrower sense (i.e. forming event constructs) is at its core an inductive and exploratory process. When we read a hypothetical event into successive descriptions of an object which are non-identical as well as having different temporal indices, we examine the changing predicates of the object with a view to building paradigms out of them.115 Only in the second instance is the translation rule, whose consistent applicability to the predicates defines the event according to condition (d), formulated a posteriori on the basis of the observation of empirical regularities. In the first instance, the rule expresses an exploratory hypothesis whose accuracy is determined only as the process of constructing events and action proceeds. The regularity of the translatability of predicates can be just as much normative as thetic in origin; the translatability of a predicate P1 into a second predicate P2 can be learnt from a textual example, learnt from the socially conditioned interpretation of events in the world in which we live, or introduced on an ad hoc basis: as long as we treat it as a rule by applying it consistently throughout the construction process, it must be valid for us. Event constructs, therefore, are not only based on knowledge. They also, at least potentially, make knowledge. Whether it is retrieved from existing memory or created for the first time, this knowledge, articulated as the application of a translation rule, develops into a complex matrix which

115

Lévi-Strauss’s hypothesis of the universal structure of myth can be understood both as a structure of myth which is identical across cultural and geographical space or as a structuring process which uses different means to generate structures which are comparable despite cultural and geographical differences. Seen in this way, the universal nature of myths is guaranteed not by their product (the structure built using existential predicates such as dead/alive, above/below, etc.) but by the fact that the theoretically open process of structuring produces a specific structurally homogenous set of culture-specific results. It is interesting to consult Lévi-Strauss’s own discussion of this alternative interpretation—in the context of his criticism of Propp’s Morphology of the Folktale, for example. Here, he emphatically asserts that, although words have a fundamentally arbitrary status in the semantic system of language, they are not arbitrary but semantically pretuned when they function as the meaningful elements of fairytales and myths (Lévi-Strauss 1984:188). In our context, this means that, in so far as the linguistic predication of objects employs words whose class membership is prespecified, we can only read events which do not contradict this semantically pretuned class membership. At any rate, Lévi-Strauss’s formulation leaves open the theoretical possibility that arrangements can be of equal value; in fact, the possibility must be left open in so far as words which symbolize meaningful elements (mythemes in Lévi-Strauss, predicates in our terminology) can never be counted as members of just one single class.

1.4 The Elements of the Action Construct

129

opens up a practically inexhaustible range of combinatorial possibilities for building event constructs. World Knowledge and the event In addition to the distinction between class-homogenous and class-heterogenous event constructs, we can classify events according to whether they are projective or protensional. This new distinction does not systematize the semantic class membership of predicates of states of affairs; it is concerned with the constructive heuristics with which interpreters identify the transformational relations between such predicates. The terminology is borrowed from Husserl’s distinction between project and protension as the two basic ways in which we conceive of future time; it is not, however, intended to suggest that there are phenomenological parallels between our usage and Husserl’s, or that our approach and Husserl’s have a common philosophical basis. Husserl’s terminology hints at the basic similarity of the two kinds of construct; they both begin with ‘pro-’, which indicates their anticipatory function of strategically predicting consequences and sequences of events in order to help us in our behavioural calculations. However, this use of ‘pro-’ is actually deceptive, for it makes us confuse the role of a conscious and distanced constructor of action with that of the agent in fictional activity, when that activity is actually inaccessible to us, not just for the relatively banal reason that it is merely narrated and fictional, but also because it is recounted, formed out of a pool of states of affairs which can no longer be expanded.116 In the event, therefore, we reconstruct the course of the states of affairs as a future which we pretend to be open but which is really already known to us. The illusion of fictional action is thus twofold: not only does it assert a world which consists of fictional objects and sequences of states of affairs; it also necessitates a mode of interpretation which creates the illusion that the irreversibility of real time is suspended for the duration of the reception of an action. Our construction of events and actions is affected by an initial methodological decision that reflects wide-ranging premises of Weltanschauung, ideology, and philosophy which underlie the way we understand action. This decision takes the form of a choice between the model of causa finalis and the model of causa efficiens, that is, between the principles of finality and causality. We adopt Sachsse’s understanding of causality and finality; by causality we understand that ‘in the course of events, certain states have certain necessary consequences’ (‘im Ablauf 116

Our words about our own, real actions are also subject to this problem.

130

Part 1 Concepts of Action

der Ereignisse bestimmte Zustände notwendig bestimmte Folgen haben’) and by finality that ‘processes reach a final state which is determined by the initial state’ (‘Prozesse einen Endzustand erreichen, der durch den Anfangszustand festgelegt ist’ Sachsse 1987:13; my translations).117 This contrast is the basic principle behind the difference between projective events, which are based on assumptions of universal causality, and protensional events, which assume that there is a principle of final ordering behind the entelechial signature of the states of affairs which they contain. Thus, a projective event interprets a fictional happening as an open sequence of fictional states which obeys causality (i.e. is subject to natural, psychological, and sociological laws) but is not governed by an essential determinism which reveals itself only in specific objects. The hypothesis that the construction of projective event constructs is rule-governed is of particular importance because it makes us less dependent on the presence of narrative objects. The central role in a projective event is quite clearly played by the predicates, which should fit together according to a causal regularity asserted as true and should be read as doing so. Alive becomes dead, young becomes old; behind every such abstract relation of opposition between two terms, we can always find the assumption of a transformational regularity whose origin can lie either in the fictional world or in the contextual knowledge which has been used. In terms of the methods of interpretation, this assumption has a generally valid heuristic function, for it allows us to extrapolate, even from a highly elliptical narrative, an object in which the anticipated transformation is able to manifest itself. Unlike projective event constructs, protensional event constructs do not assume a causal or empirical regularity to which they then assign the appropriate objects (or placeholders for objects not yet found). A protensional event is a finally orientated construct and takes the object and predicate of the second, dispositional state of affairs as its starting point in order to discover the first predicate which, in the sense of a necessary cause, must have been transformed into the second one that is concretely present. When a protensional event is constructed, the fictional happening is treated as the connection of two states in the world, of which the second (the effect) follows the first (the cause)—not according to a law known in advance, but as an example of an entelechial development which potentially indicates a previously unknown law of connection. The projective event leads into the construct of an action which by necessity obeys a regularity which is known when it is constructed; the protensional event, on the other hand, reveals the work of a principle of final ordering which stimulates the

117

See Sachsse (1987:6ff) on the Aristotelian distinction between the four kinds of cause.

1.4 The Elements of the Action Construct

131

discovery of a new regularity in the happenings of the fictional world and the action. The projective event confirms the world knowledge which we use in the process of construction; a protensional event enriches it with the hypothesis of a new regularity of which we do not yet know whether it has a de facto claim to be a universal law or whether it simply denotes an end and idea which are essentially linked only to the particular things (objects) involved in a particular concrete sequence of states of affairs. Ultimately, it is protensional events that make the reading of action an inductive and dynamic process from which we learn something instead of just finding confirmation of the world knowledge we already had. As mentioned above, the class-homogenous/class-heterogenous and projective/protensional distinctions are located at different levels. It will only be possible to apply the former in the remainder of our study; the latter will have to remain purely theoretical. Even so, we should note that the two planes do overlap to a certain extent. For the recipient of a narrated happening, the construction of class-homogenous and, at the same time, projective events provides temporary relief from the stipulation that the object in question must be recognized as a familiar one. Abstraction liberates us from the specific and opens up the possibility of completing and normalizing points of indeterminacy and cognitive grey areas by referring to our knowledge of the normal way of things. On the other hand, the recipient who generates class-heterogenous events of a protensional kind is fully dependent on being able to argue the identity of the object in both states of affairs. The class-homogenous and projective construction of events can therefore be taken as a marker of deductive reading of action, which is defined as tending to affirm pre-existent world knowledge, while protensional and class-heterogenous construction can be treated as a sign of inductive reading, which is aimed at dynamically enriching our world knowledge.118 Constructing events: An Example We shall now consider a concrete example and demonstrate step by step how events are constructed during the process of reception. Our example is taken from the Fairy Tale of 1795 with which Goethe concluded the cycle of novellas called Conversations of German Refugees. The text begins as follows: 118

It would be of considerable interest to investigate the extent to which individual texts encourage these different kinds of construction and force us to adopt one or the other of the two receptive stances.

132

Part 1 Concepts of Action

By the great river, which was newly swollen with heavy rain and overflowing, the old ferryman, weary from the toil of the day, lay in his little hut and slept. In the middle of the night loud voices wakened him; it seemed that travelers wanted to be ferried across. Goethe 1989:70119

According to the definitions we have outlined in the preceding pages, the construction of events requires us to combine elements of the linguistic statements contained in the text. We shall now narrow our systematic perspective to examine both the specific elements which are involved in this combinatorial constructive process and the rules which govern it. So far, we have used the term ‘predication’ to refer to what we can see as the constructive event molecule (the linkage of at least one event object to a property attributed to it). From now on, however, we shall no longer refer to abstract predications; instead, recognizing the ability of fictional speech to evoke mental images, we shall now speak of statements of (fictional) states of affairs, whose constructive equivalents are imagined states of affairs.120 Before investigating the virtual connections in which the initial statement of state of affairs can take part, we must first consider the constituents of which it is composed.121 119

120

121

Unless explicitly stated otherwise, this reference applies to all quotations from Goethe’s Fairy Tale in the remainder of this chapter. Our decision to use the term ‘statement of fictional state of affairs’ should be seen in the context of Wittgenstein’s attempt to treat the relation between reality, the (logical) image of reality, and the reality of the logical image as one between homologous states of affairs in which matter appears in particular situations. This of course assumes a certain willingness to read the Tractatus (Wittgenstein 1961, particularly §§2.11–15) against its neopositivist grain and apply Wittgenstein’s idea to a linguistic statement that is fictional and does not make a claim to truthfulness. Instead of ‘states of affairs’, we could have chosen the term ‘narrative proposition’ (as used by Prince, Todorov, Ryan, and others). It is true that the latter term is more obviously connected to linguistics and formal logic, but there are two reasons why it is problematic to apply it in the context of a discussion of the epic event. First, both ‘narrative’ and ‘proposition’ imply the existence of an identifiable subject of the utterance, and thereby give more weight to the aesthetics of production than the aesthetics of reception, where in reality the latter should be the focus of our attention in a constructive concept of action. Second, the use of the word ‘narrative’ to qualify ‘proposition’ is anything but unambiguous. Does ‘narrative proposition’ mean a proposition formulated in the mode of narrative (a narrating proposition which is distinguished from other kinds of proposition by its form) or a proposition which represents something and is made in the symbolic medium of narrative (the narration of a pre-existent proposition which could also have been expressed in other ways)? The term ‘statement of fictional state of affairs,’ on the other hand, stresses both its components (the matter, the equivalent of the formal argument in the logic of fiction, and its content, its predicate) and the fictional status (which by definition eliminates any claim to truthfulness) of the linguistic statement. Terms drawn from van Dijk (1980) and Linell and Korolija (1997) also appear in small capitals in the original.

1.4 The Elements of the Action Construct

133

matter and focus Every statement of fictional state of affairs is a statement about a piece of matter, an object of perception which is located in the domain of perception in question and is identified by particular properties at a particular point in time. matter is anything that has the potential to appear as an event object; the term is, therefore, not restricted to concrete entities and can also denote abstract or imaginary ones. However, before matter can be perceived as the bearer of changing particular properties during a sequence of statements of states of affairs, it must be demonstrated that it has stable generic predicates. These generic predicates provide the basis for the definition of a new focus which may become functional within an event construct. In epic texts, the criterion for the generic predication of fictional matter is met when matter is marked by a particular class concept; in our example, this is the concept of ferryman. Ferryman is the name of an occupation which can be decomposed into a large set of subfeatures. To be a ferryman, matter must meet at least the following criteria. It has to – – – –

be human, be male, have a job involving transportation, work on a boat as part of this job. state of affairs and object

The particular attributes assigned to the particular ferryman we meet in Goethe’s Fairy Tale distinguish him from the infinite number of other elements of the set of ferrymen who are classified by the same generic term. The sum of the particular properties attributed to the ferryman in a given statement turns the matter, which has so far been considered under an abstract focus, into a hypothetical object embedded within an event construct. Goethe’s ferryman is assigned individuating properties which show us three things: the kind of properties he himself has, how he is different from the other ferrymen in the set of all possible worlds, and how he is different from the other individuated matter in the particular fictional world of Goethe’s Fairy Tale. In our first example sentence, we find that, without exception, all the relevant individuating predications are contained in the main clause. It portrays this ferryman as the particular one who, in the situation with temporal index t1, – – –

occupies or possesses a hut, is lying in this hut, is weary,

134 – –

Part 1 Concepts of Action

is old, is asleep.

We shall pass over the logical inconsistency in the narrator’s description of the ferryman as simultaneously weary and asleep. It is obvious—that is to say, our knowledge of the causal logic of the world tells us—that he is asleep because he was previously awake, at which time he felt tired. The meaning of the predication of the ferryman as asleep has nothing to do with a hypothetical statement which claims that this particular ferryman has the remarkable ability to sleep and simultaneously be consciously aware of his own weariness. It is rather the case that the ferryman is sleeping at the present time because he became weary earlier, and he became weary earlier because of the ‘toil of the day.’122 The syntactic structure of the first sentence identifies the ferryman as its grammatical subject and also, therefore, as the hierarchically dominant matter in the complex statement of state of affairs. The component statements which refer to the ferryman can be divided into two categories. The first category contains predications whose unambiguous purpose is to individuate him as a specific piece of matter, that is, to attribute properties to him which are not already implied by the generic term ferryman. The second category, on the other hand, contains predications that are linked to the generic predicate ferryman, which they repeat but also, and more importantly, modally qualify. For example, all ferrymen work on rivers or lakes (and not, say, in the desert). But not all rivers are great and swollen, not all ferrymen work hard to the point of exhaustion, not all ferrymen live right by the river, and so on. Both categories, ferryman and river, have elements which allow matter that is initially generically defined to be described more particularly as possessing one or more individuating 122

It is interesting to consider the interaction of syntax and logic in the German original, ‘An dem großen Flusse, der eben von einem starken Regen geschwollen und übergetreten war, lag in seiner kleinen Hütte, müde von der Anstrengung des Tages, der alte Fährmann und schlief. Mitten in der Nacht weckten ihn einige laute Stimmen; er hörte, daß Reisende übergesetzt sein wollten’ (Goethe 1981, 6:209). As a circumlocution for work, Anstrengung des Tages is not only a highly compressed expression of the causal history which has led to the present state of affairs. It also, in the syntagmatic relations of the sentence, anticipates the generic predication which the initially general and non-specific matter undergoes when the generic occupational term Fährmann is assigned to it at the end of the sentence: ferrymen are among those people who are involved with rivers and lakes not for their own enjoyment but for the purposes of earning a living (i.e. work). The locative predicate an dem Flusse has a similar function of syntagmatically predicting and logically affirming the subsequent generic predicate Fährmann. For its part, the generic river is described as individuated matter which is ‘groß’ and ‘von einem starken Regen geschwollen und übergetreten.’ In this case, the same holds true for the syntax of the English translation—the locative predicate by the river anticipates the generic predicate ferryman later in the sentence.

1.4 The Elements of the Action Construct

135

states. In principle, it is of little significance whether this individuation takes place by the introduction of supplemental predicates or by the partial modification of the generic predicate itself. In the practical context of event construction, however, it emerges that the former method generally provides combinatorial possibilities different from and more plentiful than those of the latter. In any case, it is clear that only the combination of generic and individuating predicates allows matter considered under a focus to be available as a potential event construct’s object.123 Constructing an object event Our theoretical discussion of the concept of the event has shown that the allocation of individuating predicates to the appropriate predicate class is of key importance in determining the translation rules which can be applied in a given case. Consider again the opening main clause of the Fairy Tale: By the great river…the old ferryman…lay in his little hut and slept.

This statement of state of affairs has the status of the exposition of a possible event construct. In a particular order, it provides the potential event elements which would be needed for the formation of such a construct: the hypothetical object and its predicates. From now on, we shall refer to the first of two statements as the expositional statement of state of affairs, which, if we are to build an event construct, we must be able to convert into a second statement, the dispositional statement of state of affairs. The material of the exposition provides the starting point for a multitude of hypothetical event constructs, while the disposition is the result of a choice for which the recipient is responsible and which is expressed in the construction of one or more concrete events. This process is, of course not only subject to the decisions of the recipient; it is also subject to the linguistic material and its structure (the words and their syntax) which the text or the narrator has provided. In the case of the present example, which contains just two statements of states of affairs, the situation is straightforward: only the second sentence is eligible for consideration as the dispositional statement. The term ‘dispositional’ is borrowed from rhetoric (dispositio) and intended to remind us that the recipient building an event construct, like the orator composing a speech, cannot avoid referring to a prestructured 123

The event object is potential in so far as we cannot yet tell whether one of the individuating predicates will be subjected to a transformation in the course of the narrative, that is, whether the object will take part in an event at all.

136

Part 1 Concepts of Action

selection. A second rhetorical connotation of the term is even more important. Dispositio refers less to the result of the structuring of cognitive, linguistic, and imaginary material, and more to the continual, dynamic nature of the structuring process.124 When the recipient evaluates a statement of state of affairs as dispositional, that statement becomes a double stimulus: first, a stimulus to convert the virtual event construct contained in the preceding exposition into a real one; second, a stimulus to investigate the dispositional statement of state of affairs to see if it itself has expositional content and the potential to enter into further combinations which satisfy the logic of events.125 We shall now define the expositional statement of state of affairs in formal terms. The following notational conventions will be used: SAexp Fexp

= =

PCA…Z IP1…n Texp

= = =

expositional statement of state of affairs. expositional focus (i.e. matter as defined by a generic predicate). predicate class A…Z. individuating predicate 1…n. temporal index of the expositional situation.

Concrete expositional statements can now be represented in the following general notation: SAexp

=

Fexp[(PCA, IP1) … (PCZ, IPn); texp]

In our example sentence we are confronted with an instance of multiple (as opposed to single) predication. For convenience, we list the individual predications in the following form: SAexp

=

[ferryman; (relational, by the river); (physical, lying); (relational, in the hut); (physical, old); (cognitive, asleep); texp]

So far, we have created three predicate classes. Although their class terms (relational, physical, cognitive) have been chosen to be sufficiently general, their specification is ultimately arbitrary. They are specified according to the personal world knowledge of the individual and the conventions to which he subscribes, not by reference to epistemological categories which could 124

125

The meaning of the German word Disposition retains the idea of transitoriness in phrases such as etwas zur Disposition stellen ‘to make something available’, which highlights the tentative, temporary nature of having something at one’s disposal and how it demands to be used and flexibly adapted. See Lausberg (1990:§§46–90) on the concept of the dispositio in rhetoric.

1.4 The Elements of the Action Construct

137

be claimed to be universally valid.126 Each of the individuating predicates listed above provides at least one theoretical starting point for a virtual transformation of state of affairs which could take place in our object, the ferryman. Each element of the list also points to a practically boundless set of implied predicates which could have fallen victim to ellipsis.127 In the present case, we find the following dispositional statement of state of affairs:128 In the middle of the night loud voices wakened him; it seemed that travelers wanted to be ferried across.

In our formal notation, this is represented as SAdisp

=

[ferryman; (cognitive, awake); (relational, loud); (physical, hearing); (cognitive, requested); tdisp]

The formal notation reveals that, as in the case of the expositional statement of state of affairs, we have imposed two interpretations on the sentence. First, we have translated the verb ‘wakened’ into the adjective ‘awake’ and compressed the subordinate clause ‘that travelers wanted to be ferried across’ into ‘requested’. Second, the decision to allocate the predicates to their respective classes is an interpretive decision. Both these interpretive steps are motivated by the objective of generating an event from the statements of states of affairs. They are hermeneutic acts and are saved from being decisions of arbitrary expediency not by the presence of specific rules of mapping and assignment, but only because they are subject to more general methodological rules which must be consistently followed at every stage in the process of event construction.129 The possibilities for interpreting the sentences of the text in terms of such a construct can be represented in the form of the following tabular comparison of the contents of expositional and dispositional statements: 126

127

128

129

In a practical context, the philosophical foundation of a heuristic set of categories is less important than the need for the posited categories to be used consistently rather than being replaced or extended whenever it is found convenient. Alive is one example of such an implicit predicate; someone who is old and asleep is obviously not dead. We would be forced to apply this argument retrospectively to the initial state of affairs if we found a subsequent sentence of the kind of But the next morning, the ferryman was dead. In our theory, therefore, implicit predicate, differ from explicit predicates in that they are not the logical prerequisite of an event construct but its logical consequence: their effect on the logic of events is that they force us to reinterpret recent expositional statements of states of affairs. Strictly speaking, these are two separate statements of states of affairs, for we are dealing with two main clauses. I have combined the two clauses solely in order to facilitate their formal representation. More specifically, we are concerned with three such methodological rules.

138

Part 1 Concepts of Action

Table 1.4.1: Statements in an object event (1) Linear text processing. When identifying object and discourse events alike, we must not deviate from syuzhet, the concrete order of the narrated statements of state of affairs as they are presented to us in the text. The interpretive reconstruction of the quasi-natural chronology of the ordo naturalis (fabula) is of little importance in the generation of event constructs; whether narration is isochronous, as in the present example (sentence 1 presents the ferryman as asleep, sentence 2 as awake), or analeptic (as in the hypothetical example In the middle of the night voices wakened him. Weary from the toil of the day, the old ferryman had slept in his hut by the river), in both cases, structurally identical, albeit inverse event constructs are produced. (2) Specification and use of a finite set of predicate classes. A predicate class is defined as a semantic hyperonym which covers the semantic field of a theoretically open set of individuating predicates. In the current example, we have used three classes so far (the classes of relational, physical, and cognitive predicates); in the course of the practical application of our theory in part 2, we shall introduce six further classes (emotional, material, social, aesthetic, moral, and philosophical predicates). As interpreters of events, we are able to decide the class to which any given concrete predicate should be allocated. Theoretically, multiple allocations or allocations which differ between statements of states of affairs are acceptable. However, practical experience in the analytical program EventParser, which is based on the theoretical discussion here, has shown that, as a rule, the allocations chosen by interpreters are stable throughout the interpretive process. (3) Conceptual representation of predicates in the form of adjectives. By convention, the argument position of a propositional statement of state of affairs is filled by a fictional object of perception which appears as matter and is described in the event construct as the bearer of changing predicates. The narrative text itself is not restricted to explicit adjectival predication of matter. With respect to the social semantic field (i.e. the predicate class social), we could perfectly well describe the ferryman using the predicate isolated in the first sentence of our example and its opposite integrated in the second sentence. By doing so, we would interpret the text as connoting these predicates in the complex situations it describes. We are completely free as regards the content (i.e. the actual allocation) of predicates and are required only to express these interpretations in a formally consistent manner. This allows us to ensure that all individual interpretations are formally comparable.

1.4 The Elements of the Action Construct

139

At first glance, the above may appear to be a somewhat laborious reformulation of what any reader will identify as the event contained in these two (or three) sentences: The old ferryman was woken. However, this periphrastic statement is elliptical and obscures a number of characteristics of this particular event construct which are clearly visible in the representation we have provided above. In addition to the identity of the matter in both statements of states of affairs, we note first that, out of a total of three predicate classes, only one, the cognitive predicate class, is treated as functional in the event. Inside this class, we have translated the element asleep into the element awake (as highlighted in bold in the table). Now, different or additional transformations are also perfectly conceivable. For example, is hearing not a metaphorical antonym of asleep, and could our paraphrase not just as easily run At first the old ferryman was asleep, then he heard something? This would give us a categorially heterogenous event instead of a categorially homogenous one. It is clear from our table, however, that we have given precedence to the construct which has class-homogenous predicates and is rhetorically highlighted by the chiasmus of asleep (in final position in SAexp) and awake (in initial position in SAdisp). And our tabular description also shows that it is simply not possible to realize some of the hypothetical event constructs suggested by SAexp—for the time being, at least. The ferryman is clearly still lying (physical predicate) in the hut (relational predicate), and he will presumably remain old (physical predicate) for the rest of the story. Constructing a discourse event In an object-centred event construct, identical matter appears as the object and common factor in two statements of states of affairs. discourse events, on the other hand, are characterized by the fact that their constituent statements of states of affairs do not have an identical focus, yet can compensate for this lack by having not only identical predicate classes, but furthermore by presenting actual predicates whose logical relation to one another is of a paradigmatic nature. Such logical relations are often further emphasised by specific textual cues. Our example text allows us to demonstrate this with reference to the predicate old and its predicate class (physical). Shortly before the end of the Fairy Tale, we read: ‘Don’t you know,’ she replied, ‘that you have also grown younger?’ ‘I am glad, if to your young eyes I appear to be a sturdy youth.’

140

Part 1 Concepts of Action

Note that the old man who is here described as having become younger is not our ferryman but another character, the man with the lamp.130 However, because of the confusion caused by Goethe’s liberal use of ‘the old man/woman’ (both the ferryman and the man with the lamp are referred to as ‘the old man’), the above quotation invites us to look out for a possible paradigmatic connection. According to such a reading, the narrative here completes the hypothetical event construct it initiated in the old ferryman by translating the physical predicate old into its class-homogenous opposite, young. This means that, in the first sentence of the Fairy Tale, we have an expositional statement of state of affairs which opens up the possibility of generating a discourse event which spans twenty pages—almost the entire text, in fact. Both in terms of the text’s aesthetic coherence and the identity-forming synthetic achievements of reader and narrator, this goes far beyond anything that could be achieved by the constructive formation of a simple object event based on the man with the lamp. The new discourse event can be schematically illustrated as follows:

Table 1.4.2: Statements in a discourse event

130

As has been noted in the secondary literature (e.g. Brown 1975), such descriptions as ‘the old man/woman’ (‘der/die Alte’) contain references which extend beyond this particular narrative to the narrative in which it is embedded, the Conversations of German Refugees, whose narrator is also called ‘the old man’ or ‘the old priest’. In terms of the internal form of our particular narrative, however, these descriptions are interpreted as the basis for a practically overwhelming range of potential connections in its action. From what we know of the Fairy Tale, this is no chance occurrence but part of the work’s intended effect.

1.4 The Elements of the Action Construct

141

In this metaphorical reading of the discourse event, the potential event prepared by the expositional statement By the great river…the old ferryman…lay in his little hut and slept.

is realized in the dispositional statement of state of affairs: ‘Don’t you know,’ she replied, ‘that you have also grown younger?’

The ability of readers to form this paradigmatic construct, if they are able to do so at all, depends not least on whether it is possible for them to retain the individuating predicate of sentence (A) unchanged in memory until sentence (B) just before the end of the text, twenty pages later. It is also conceivable, as noted above, that we could construct an object event based on the man with the lamp as an action (or, more precisely, as an event construct) running in parallel to the as yet incomplete transformation of the ferryman. This would be achieved by interpreting (B) as a dispositional statement of state of affairs made in reply to an earlier exposition (A + n) (not shown here), and in this case (B) would more clearly fit the syntagm of the narrative. Individual preferences and cultural conventions may well influence the decision between the paradigmatic option, the discourse event, and the syntagmatic option, the object event. However, we note again that there is no fundamental, theoretically demonstrable principle which requires us to settle on a particular event construct. The opposite is in fact the case: we must always bear in mind the possibility that object and discourse events can overlap and overlie one another. We now return to the apparent non-identity of the object in a discourse event. In strict terms of the ontology of events, this non-identity is no more than an illusion—if there is to be a perceivable event at all, a change of predicates must obviously take place in some identical subject. This illustrates the key advantage of distinguishing the focus from the object: discourse events, as described above, do not take individuated objects as their focus; in them the latter consists rather of abstract entities behind the fictional figures: groups of figures, spaces, narrator figures, narrating authorities, and, at the highest and most abstract level, the perceiving consciousness of the recipient. When this consciousness can find no other way to synthesize the statements of states of affairs of an epic text into the constructs of event and action, it has no choice but to attribute the statements of states of affairs to itself as predicates of its own changing states of knowledge. This brings us to an important boundary. When we find that only discourse events can be produced, we must question whether it is appropriate to assign our text to the epic genre at all, for at this point we are dealing with a text which does not narrate events but which is, for us in our own world, an event in the real and everyday sense of the word.

142

Part 1 Concepts of Action

The event Matrix We shall now draw together in a single matrix all the theoretically possible ways of forming event constructs. This matrix will allow us to generate the combinatorial variants which result from the interpretive assignment of relations of identity and non-identity between the constituent elements of expositional and dispositional statements of states of affairs.

Table 1.4.3: The event Matrix

The matrix shows that a pair of three-part statements of states of affairs can only be connected to form an event when they are neither completely identical (a) nor completely non-identical (d). Signification and existence in general can only be grasped in terms of difference, and this is true also in the particular case of the event. The use of the terms

1.4 The Elements of the Action Construct

143

‘antithetic discourse event’ and ‘thetic discourse event’ for (e) and (f) respectively requires explanation. Type (e) is antithetic in so far as it identifies an antithesis between asleep and awake, irrespective of the focus under which the two statements of states of affairs are seen. Type (f) is thetic because it amounts to the formulation of a normative assertion: if I posit the recipient as the real focus (i.e. the focus which is connoted as identical in both states of affairs despite the denoted non-identity of their matter), then the two statements of states of affairs predicate of the recipient two pieces of knowledge distinguished by different temporal indices. The event in this case appears to consist of nothing more than the accumulation of knowledge on the part of the recipient; if we went no further than this highly abstract definition, we could make any pair of statements of states of affairs whatsoever into an event. Thus, (f) covers only those pairs whose predicate classes and predicates are both identical. This is the reason why we describe the event construct as thetic, for not only are arbitrary pieces of knowledge about the fictional world, disparate in category and/or content, accumulated under the recipient’s focus, but, in addition, a paradigm is constructed by reference to the equivalence criterion, which in this instance is the predicate tired. We have already mentioned that this is the last resort of a recipient who pursues synthesis in the form of an event construct; the event in such instances is the construction of the paradigm itself. Straining the concept to its limits as it does, variant (f) marks the most extreme form of what can be considered an event in an event-based ontology. The formation of this kind of construct lies, as we have said, in the realm of purely paradigmatic synthesis. Granted, hardly anyone reads epic texts as event narration in this way. But even so, just because the occurrence of a borderline case such as this is statistically rare, this does not permit us to ignore it in our theory. Neither the boundary between the object event and the discourse event nor that between the discourse event and the reception event can be captured purely in terms of an event-based ontology; in this respect, our approach is clearly deficient. Regardless of our theory, the borderline phenomenon of the thetic discourse event is an indication of the lengths to which recipients will go in their continual struggle against the semantic horror vacui. It is a struggle which the recipient tries to conclude in his favour at each stage of reception with an act of synthesis in which statements of states of affairs become object events and discourse events. When we perform this synthesis, our tolerance of the unconnected and the unconnectable is most fundamentally limited by the pragmatic factors of cognition, for the brain can only keep a limited number of distinct conceptual perceptions available for reference at any one time. Beyond this, however—and here

144

Part 1 Concepts of Action

we return to an idea hinted at at the end of the previous chapter, where it had not yet taken concrete shape in narrative theory—we find that the real limit of what our minds can bear is reached when we are confronted with a narrated world in which no state of affairs fits in with another, in which no state of affairs reflects, negates, takes up, or transforms another; in short, when we are confronted with a fictional world in which there is utterly no possibility of an event being found at all. And it is this semantic deficiency that defines the ontology of a world in which we as interpreters cannot make anything our own, not even the consciousness of our own selves as the identical subjects of this shattering perception of all-encompassing isolation. The Referential Nature of the event We have defined the event, the one element common to all theoretically possible action constructs, with reference to three mental constructs which are generated by the recipient under the guidance of textual and contextual factors. These mental constructs are fictional matter and the two (if not more) fictional states of affairs in which it functions as a shared focus. What, then, do the constructs of matter and states of affairs mean, what do they refer to apart from a singular event? From the point of view of existential ontology, they have the function of making a statement which asserts the positive or negative existence of a distinctive object in the fictional world. Like all objects individuated by particular characteristics, this object can undergo a change of state or property. matter attached to an initial state of affairs, then, is the minimal unit from which we can derive the textually evoked idea of a fictional existence before any synchronic events have taken place.131 Most fundamentally, then, narrated states of affairs denote the existential conditions, postulated in the narrative and quasi-empirically reconstructed during the reading process, of matter as it is instantaneously captured in isolated individual perceptions of phenomena. From the perspective of the construction of hypothetical object events, fictional objects, persons, and even abstract entities (ideas, emotions, etc.) appear as matter; from the perspective of the construction of hypothetical discourse events, on 131

In a more precise, philosophical sense, we must follow Wittgenstein (1961:7–9) and treat the difference between any matter (or thing) and its state of affairs as one that is completely abstract and contradicts the phenomenology of our mental ideas. Matter (taken in the general sense of the word, and not just in terms of our narrower narratological definition) never occurs outside states of affairs; even in logical terms, all matter has the unrealized potential of being embedded in various states of affairs.

1.4 The Elements of the Action Construct

145

the other hand, they (in this case with the exception of abstract entities) appear as narrating and receiving figures or people. It is clear that matter refers beyond itself; by connecting states of affairs into a sequence of states of affairs and constructing an event, we enrich the text with metastructures which, as synthetic mental constructs, are located above and outside that which the concrete linguistic sentences have posited as distinct existences and momentarily perceptible states in the fictional world. At this metafictional, metaconstructive level we are inevitably forced to turn to our world knowledge and assumptions about the translation rules which it is acceptable to use when forming events.132 It is true that narrative texts can force their recipients into constructing a particular event construct; even so, without the cooperation of a reader as a compliant constructor, there will be no events in the fictional world in the first place. In short, in our theory there is no such things as an event in itself; events are idealized objects above and not in the fictional world and are resistant to narrative denotation. This thesis is probably the most fundamental difference between our approach of an event-based ontology and previous definitions of the event, which are almost always formulated in terms of the aesthetics of production. With few exceptions, formal narratological theories about the existence of events in the fictional world are based on a class of narratorial metapropositions which mediate transformationally between particular elements of the basic level (the narratorially posited propositions of existence, state, and property). Examples of this are the models of Prince (1973, 1982), which differentiates between stative and active events, and Ryan (1985), in which the distinction between state propositions and event propositions plays an important role.133 In such models, the transformational process itself is

132

133

See van Dijk (1980:228–49) on the dependency of text-orientated receptive and cognitive processes on world knowledge. Ryan postulates that ‘The claim that a narrative plot is a temporal succession of different states of affairs mediated by events is one of the least controversial of narrative theory’ (Ryan 1985:717; my emphasis). As a matter fact, the claim is not quite as unproblematic as Ryan suggests; the distinction between static states of affairs and dynamic events was questioned, for example, by Todorov as long ago as 1969. Also, I believe that Ryan causes herself unnecessary difficulties with the event concept in particular when she subsequently introduces the category of narratively represented non-narrative events, which may be plausible in terms of the aesthetics of reception but is logically meaningless. The category is intended to cover changes of state in which the interpreter can find no relevance to the progression of the plot and which could therefore have been omitted without significant consequences (Ryan 1985: 737f.). This, like Lotman’s concept of the event, confuses an aesthetic interpretation, dependent on a final global evaluation, with a purely formal process of categorization which takes place successively in the course of reception alone.

146

Part 1 Concepts of Action

understood as one that is propositionally denoted. This assumption may be acceptable under the premise that the text contains events in so far as they are signifieds in which the author narratively encodes a statement which he intends to make with the linguistic signifiers in the text. The same assumption (which can be traced back to Aristotelian poetics) can be found in less formal models such as those of Lotman, Stierle, and Schmid. The approach has met with philosophical approval, not least because of Danto’s tripartite definition of the conditions of narration as the situation or state of a subject at point t1 in time, an opposite situation or state at point t3, and an event which takes place at t2 between t1 and t3 and which turns the first state into the second. Schmid summarizes this structure effectively by calling it a logical triad, but he also joins Stempel (1973) in pointing out the contradiction inherent in the fact that this triad can be realized in a sequence of just two sentences (Schmid 1992:107). Does the event, whose presence Danto argues to be a minimal condition for narration to take place at all, not have an essentially bipartite instead of tripartite structure? The confusion here is, of course, due to the fact that we have inadvertently adopted the definitions of narrative and the event as they appear in reception theory. In this context, the neglected central proposition, which Danto requires for the transformation to be denoted, is not denoted by a manifest narratorial statement but can be interpretively (re)constructed by the recipient because it is connoted resultatively by the descriptions of the initial and final situations. Be it real or fictional, an event in itself quite simply cannot be linguistically formulated or propositionally asserted without placing it at the same level as mentally represented states of affairs. This requires an act of synthesis which is the privilege of the recipient. Even narratorial statements whose content is an unfinished process (Brutus began to kill Caesar) denote not an event but the first step on the way to one; so far, Caesar is not dead and Brutus has not stopped stabbing him. Events (and therefore events) require the presence of completed transformations; in this strict sense, our example contains an event only to the extent that the initiation of an attempt to murder marks the end of a previous state (not wanting to murder). But the narrator did not say that and can hardly have intended to suggest it. It is a perplexing state of affairs; we can see the picture before our eyes as Caesar’s blood pours out amid plenty of action, but this is not enough to be formally considered a completed happening and is therefore not (yet) an event. As processes, transformational processes can point only to temporal continua: they begin after the initial state and stop before the final state. At the core of our problem, therefore, lies a lack of precision in temporal indexing, for, as a process, an event is by definition represented by

1.4 The Elements of the Action Construct

147

something (i.e. the temporal continuum) which cannot be bound to a specific point in time. If a narrator wanted to assert that the fictional world contained such a process with propositional content, he would require at least an assertive statement located at a completely different level of formal complexity than that of the fundamental propositions with which the existence of fictional things and states of affairs can be asserted.134 In any case, propositional statements about existents belong, because of their content and, more importantly, their form, to a different class than propositional statements about events; problematic consequences result if the attempt is made to merge these classes. Our approach, on the other hand, does not expose itself to any such difficulties in the first place, for it essentially conceives of the event as a synthetic product which is made not by the narrator but by the recipient. In this sense, when we claim that events are present, we have in mind quasi-propositions which are not themselves present in the text but which the assertions of the text prompt us to form. Irrespective of their transformational content, events are synthetic constructs because they connect t1 and t2, two distinct points in time, with one another. Before an event is formed, narrative statements of states of affairs are temporally indexed, but their conceptual content is presented to us synchronically (unless the excessive use of temporal deictics or a rigid compulsion to narrate isochronally makes possible their arrangement in unbroken order). With the formation of the first event construct, however, we establish the principle of the diachronic ordering of states of affairs. Narratively asserted states of affairs continually motivate us to move beyond synchrony to diachrony, for they not only denote the state in question but also connote its logically implicit potential for change. What is now so must, at least in theory, have the potential to be (or have been) different at another point in time.135 The formation of a real, diachronically arranged event construct by the recipient amounts to the realization 134

135

In contrast to our understanding of the proposition as a temporally indexed assertion of a fictional state of affairs, Quine sees propositions as eternal sentences, abstract entities that are valid irrespective of the references of their associated linguistic sentences and the circumstances in which they are uttered. The temporality which Quine argues to be absent from the proposition does not concern the inherent temporal structure of the concrete, singular content of the utterance; it concerns the eternal validity of the abstractly conceived utterance itself. In this sense, we read that ‘reports and predictions of specific single events are eternal too, when time, places, or persons concerned are objectively indicated rather than left to vary with the references of first names, incomplete description, and indicator words’ (Quine 1960:193f.). In the terminology of Bremond’s action logic, this means that a pure (fictional) existence itself is an éventualité which indicates, in the sense of a hypothetical passage à l’acte, the possibility of it being turned into a non-existence.

148

Part 1 Concepts of Action

of what was initially present in virtual form in the expositional state of affairs. Such a construct thereby makes it possible for the recipient to realize two superior semantic functions which amount to more than the postulate of fictional existence which we have observed until now in matter and the state of affairs. The event construct now establishes a reference which is partly reflexive, partly paradigmatic. It is reflexive in so far as a new perspective of event-based ontology is revealed, in which, for the first time, we are able to experience the fictional identity of matter (not just its hypothetical fictional existence). For only that which can be comprehended as a self (the subject of a proposition) in at least two non-identical descriptions (predications) can acquire the feature of identity and be an individuated and identical object of perception.136 On the other hand, the new (relative to the diachronic transformation of states of affairs) paradigmatic referential function provided by the event is directed not at the subject, matter, but at its predicate.137 Here again we find the syntagmatic realization of what was originally virtually contained in the raw data, the expositional state of affairs, the realization of a transformation of state in a property, the predicate assigned to matter. In this sense, event constructs explain not only the ontic postulates which assert the fictional existence of fictional objects and processes but also the semantic postulates which structure the world of the epic text and make it possible for it to have reference at all. In other words, only by constructing events are we able to give the predicates of the individual statements of states of affairs a differential definition anchored in the fictional domain.138 Statements of states of affairs have a twofold referential function when they are placed in the synthetic context of an event. Formal existential postulates about the existence of fictional objects turn into postulates of identity which are defined by their content and require the contrastive foil of a change in the properties of fictional matter in order for any as136

137

138

In the extreme case of redundant narration, the difference between the two descriptions can be reduced to the non-identity of the temporal index of the two statements of states of affairs. The predicate is that singular or complex individuating property which is assigned to a general term (matter) in the course of a predication (Quine 1960:96). We can, of course, identify singular predicates (more concretely: adjectives, adverbs, modal operators, etc.) by referring to lexical and grammatical knowledge which is purely linguistic and lies outside the fictional world. Inside this world, however, a property can assert itself as a property, even if just in formal terms, if and only if it has already been subjected to a change. In terms of form and content, if we are to realize the predicate as a predicate during the reception of statements of states of affairs, we have to turn to a paradigmatic class which is either already present outside the text or needs to be built on the spot inside it.

1.4 The Elements of the Action Construct

149

sertion of identity to be made. Qualitative postulates which join an object to a predicate of as yet undetermined content in a singular state of affairs become definitions of the property itself which operate in terms of difference and are thus concerned with its content.139 At this point, we ask what brings about the next step, the connection of events into sequences of events, that is to say, into complex interpretive constructs of narrated action and read action. The answer to this question is provided by the methods of our approach, which resembles that of other constructivist studies of reception. One example is Kafalenos (1997), who, like Pavel (1985), takes up Propp’s function model. By the function, Kafalenos means an ‘interpreted event, an event interpreted according to its consequences, and I assign the act of interpretation…to the reader’ (Kafalenos 1997:470; my emphasis). Used in this way, Propp’s function transcends the systematic dichotomy of fabula and syuzhet, for the function as an ‘interpreted event’ can now be equally functional at the levels of both fabula and syuzhet: The possibility of using one set of terms to analyze both trajectories is the direct result of the principle that underlies the definition of the function. Because a function expresses an interpretation of an event in the sequence in which it is perceived, a given event can be interpreted as a function in any sequence in which the event occurs. Kafalenos 1997:470

The functional nature of the event, therefore, has two aspects. It is assigned a role by both the logical construct of the entire action (fabula) and the logic of its narratorial presentation (syuzhet). Indeed, it becomes a systematic interface between the levels of fabula and syuzhet. When the reader retrospectively categorizes an event from the perspective of an effect which has a more recent temporal index, this does not constitute the discovery of a causally prior event proven with the invalid argument of post hoc, ergo propter hoc. Instead, a (re-)evaluation takes place, the formal basis of which is inherent in the inescapable presence of the sequence of cause and effect. This re-evaluation selectively suspends the systematic hiatus between synchrony and diachrony:

139

At this stage, we may well be agreed that our perception of fictional matter is dependent on its inclusion in an event construct. But do we correspondingly form semantic paradigms only when we construct events, or do we not actually construct events with the help of semantic paradigms? Even if the event- or action-orientated reception process leads to the enrichment of paradigms with new elements, or even to the formation of new classes, it nonetheless assumes an initial awareness, drawn from the world knowledge acquired in real life, of the opposition and equivalence relations which apply between basic predicates.

150

Part 1 Concepts of Action

An effect does not cause a prior event. An effect indicates the significance of a prior event by placing it in a sequence in which the prior event can be interpreted according to its consequences. The ‘discursive forces’ of the sjuzhet that Culler discerns, I argue, do not alter the principles of causality; rather, they guide perceivers to establish the function of the event in question, to interpret it according to its consequences. Kafalenos 1997:471f.

In several example analyses, Kafalenos demonstrates how, as reading progresses, a shift takes place in the functions which the reader assigns. What is initially (i.e. with reference to the sequential order of the syuzhet) categorized as primary motivation can subsequently be re-evaluated as a merely preparatory state of affairs (Kafalenos 1997:475). Furthermore, there is no canonical reconstruction of the sequence of functions; in fact, Kafalenos considers her approach an attempt to utilize the concept of the function in the differential analysis of interpretations ‘at the very basic level of interpretation’ (i.e. the level of the construction of the fabula) (Kafalenos 1997:477). In her view, events are fundamentally polyvalent (although we should still continue to bear in mind our tendency, demonstrated by cognitive psychology, to retain our first evaluation of an event). We must also consider the role of virtual unnarrated or unnarratable events, which are logically implied by an effect but not represented anywhere in the syuzhet. Her definition of ambiguity as the ‘coexistence of mutually exclusive fabulas in one subject’ (Kafalenos 1997:442) indicates that not all the virtual connections suggested at the event level are positively sanctioned by the syuzhet. Instead, in Kafalenos’s view, the order of the syuzhet actively influences what recipients construct. Used constructively in this way to analyse reception and applied primarily to the fictional object event asserted by the narrator, Propp’s somewhat aged concept of the function takes on a surprisingly modern character. The reformulation of the concept of the function in order to analyse reception also deals with Lévi-Strauss’s structuralist objections to Propp. If we follow Kafalenos and define the hermeneutic event construct as the equivalent of Propp’s function, the latter loses the qualities of monofunctionality and lack of ambiguity which it had in Propp’s theory. Propp’s aim was to derive the fabulas of an entire corpus of individual texts from an abstract generic template. Only while we continue to pursue this aim are singular events subsumed under a single specific event type and apparently unambiguously defined function. For Propp, the action which the magic fairytales of his corpus report in various ways is an unproblematic abstract object whose existence is assumed without further ado. As a result, the fictional events reported by the fairytales have a correspondingly unproblematic character. But if we treat the event as a construct of

1.4 The Elements of the Action Construct

151

reception and define the concept of the function from this constructive perspective, we realize that the level of events does not document a single definitive action. events as functions are multifunctional constructs of reception which provide the basic material for a multitude of possible action constructs, some coextensive, others mutually exclusive. In practice, however, only a tiny fraction of what is theoretically possible is ever actively exploited in reception. There is no doubt that, as Kafalenos notes, the structure of the syuzhet has a persistent influence on the formation of action and even event constructs during reception. The mechanisms which narrative texts employ to constrain the virtually possible to what is aesthetically permissible or authorially intended continue to be a rewarding field of research, particularly in the case of the phenomenon of narrated action. Relatively little attention, however, has been given to the perhaps more fundamental question of the conditions under which the multitude of virtual possibilities comes into being in the first place before it is subsequently narrowed down. It is to this question that we now turn. How can basic action constructs be generated on the basis of a given set of events, and what logic governs the functional polyvalence of the constituent events in the various alternative action constructs? To find the answer, we need to introduce a new topic into our discussion, a structure of intermediate complexity located between the basic construct of the event and the complex construct of the global action. This intermediate concept, the smallest possible closed action construct, is the episode. Definitions of the Episode Proposing to redefine the concept of the episode for our own particular purposes immediately exposes us to two fundamental criticisms. First, is it not counterintuitive to place the episode at an intermediate level of the system, between the constructs of event and action, given that entire actions can sometimes be seen as episodic? Would the episode, or the equivalent construct we describe, the episode, not be better located above the level of action? Second, does our proposal to define the episode in terms of a constructive reception theory not blatantly contradict the familiar application of the term in aesthetic criticism, in which sense it has been used since Aristotle, established in the critical literature, and justified by the history of the concept itself? We begin by dealing with this latter question; in answering it, we shall show that the former question is actually irrelevant. Even a cursory glance at the wide variety of vernacular uses of the word ‘episode’ is enough to show that our commonsense understanding of the

152

Part 1 Concepts of Action

term is extremely vague. Outside of aesthetic contexts, we use ‘episode’ to refer mostly to perceived or represented event complexes which are generally negative deviations from the norm: the occurrence of illness; erratic fluctuations on the stock market; periods of political and social unrest; and even moments of unexpected happiness which we, perhaps in a defiant anticipatory gesture of resignation, secretly treat as occurring simply to lift us up higher so that the malevolent gods can take increased pleasure in watching our eventual fall to the normal level of existence. Such kinds of episode have the gratifying quality of coming to an end and thus being impermanent; for this very reason however, they have the unpleasant habit of recurring from time to time. The episode can, therefore, be unique in content, but not in form, for every occurrence which autonomously repeats itself iteratively or cyclically and is in some way unusual is, in principle, an episode. Thus, in the widest sense of everyday usage, an episode is a completed but for the most part repeatable happening which is distinct from what normally happens. When referring to aesthetic objects in non-technical contexts, on the other hand, we treat the episode as something which, in some way or another, resists the large-scale plan of a meaningful whole. More clearly than when used in everyday speech to refer to empirical states of affairs, this use of the word reveals that the episode is actually a hermeneutic rather than epistemic phenomenon. Depending on one’s philosophical taste, things episodic can be seen as chic and highly appropriate to our modern awareness of the limited potential of our postmodern knowledge (take, for example, the revival of episode film); or they can be seen as the defeatist expression of a culture whose attention span has decayed to such an extent that its capability to synthesize formal units is limited to textual and visual fragments of the standard found in televised soap operas. Both these uses of ‘episode’, one positive and the other pejorative, lack a satisfactory critical or theoretical foundation. In fact, they are used more in the metaphorical manner which, perhaps surprisingly, is typical of the way the term is used in literary criticism. In recent literary criticism and narrative theory, we find the term being applied in ways which are insufficiently defined and, in fact, often contradictory. Eberhard Lämmert can be cited as an example of the widespread confusion. He places the episode in the categories of ‘additive interpolations,’ ‘forms of consecutive connection,’ and ‘interpolated analepsis’ (Lämmert 1955:44, 7, 113; my translations). Such inconsistent theoretical treatment of this narratological concept is problematic not least for Lämmert’s own purposes, and so he attempts to characterize the term more precisely with a quotation from Batteux, who defines the episode as connected to the main action in an additive or correlative fashion. This

1.4 The Elements of the Action Construct

153

allows Lämmert to credit it with deliberately constructed symbolism and a self-sufficiency which interrupts the tension of the text. As can be seen, the place of the resultant concept in Lämmert’s system is ambiguous not least for methodological reasons; as it is clearly explained in terms of its aesthetic effect as a captivating digression, a contradiction is bound to arise when it is then defined semiotically as the design of a fabula with a contrastive or additive relation to the main action and included in a phenomenological catalogue of the elements of narration.140 Let us turn, therefore, to the use of the term in the more distant past. It originated in theories of normative poetics, where it played a central role. ‘Episode’ and its associated adjective ‘episodic’ have a long history which stretches back to about a century before Aristotle. Protagoras was the first to use the term epeisodion to refer to a meaningful compositional element used as part of an action to provide the motivating connection between two adjacent strands of the plot.141 In Aristotle, the noun epeisodion is already being used in two ways, and in his work we also find ‘episodic’, the pejorative adjectival form which has survived to the present day. Dammann summarizes how the author of the Poetics uses the concept as follows: (a) Quantitatively, only when referring to tragedy: to denote the historically later scenes of dialogue between actors which are inserted between the songs of the chorus. (b) qualitatively, when referring to both tragedy and epic: to denote the concrete scenes, characteristically and organically connected, into which the writer divides his elaboration of the sketch of the underlying fabula. In addition, Aristotle uses the adjectival form of the term in a negative sense: (c) to refer to the disjointed nature of the parts: ‘I call a plot [mythos] episodic when there is neither probability nor necessity in the sequence of its episodes [in sense (b) of the word].’ Dammann 1982:69 (my translation)142

The pejorative critical use of ‘episode’ in the sense of ‘non-essential added scenes,’ as definitively paraphrased in Else (1957:182, note 199) has remained familiar to the present day. But, as Nikau (1966) has shown with meticulous attention to philological detail, this interpretation of the word can only be derived from the adjectival variant (c) and has nothing in com140

141

142

The episode is ambiguously semantically evaluated by Max Lüthi, among others. He treats it both as an elementary motif and as the frame for several individual motifs. See Dammann (1982:69f.) on this and other examples of the use of the term. According to Nickau (1966:159), who finds the first use of epeisodion in an interpretation of Homer by Protagoras (Pap.Oxy.2 (1899) p. 68 col. XII lin.20–24). The embedded quote is from Aristotle (1984:2323 = 1451b, 34f.).

154

Part 1 Concepts of Action

mon with the true concept of the epeisodion in classical poetics—it does not take into consideration the philosophical background against which Aristotle must have intended the term to be understood. To do proper justice to the concept of the epeisodion, we must consider the close connection between Aristotelian logic and poetics. Both share the idea of immanent meaning which can be set out syllogistically in logic and mimetically in poetry (Gaede 1977:19–25). The individual epeisodion is thereby raised in status; it becomes a poetic syllogism which helps translate the abstract material of the action plan into the particular material of mimesis. Of key importance here is the philosophical assumption that the overall aim of the complex, complete action is an inherent part of each individual epeisodion. Thus, the epeisodion is not a disjunctive, non-essential episode in the contemporary sense of the word; it must be understood, analogous to logical deduction, as a necessary narrative statement which contributes to the unity of the poem. But, as Gaede argues by returning to Aristotelian logic in his interpretation of the Poetics, epic unity means that the beginning and end of something do not transcend it but lie in it itself. Unity is therefore realized in complete closure in which nothing apart from its own premises are needed for it to be apparent. Thus, behind the narrower definition of tragic mimesis, there lies the model of complete closure. Gaede 1977:23 (my translation)143

From the point of view of the theory of production, or the internal evolution of a work, the Aristotelian epeisodion denotes the basic mimetic construct with which the epic poet begins to convert the abstract structure of the fabula into the linguistically and conceptually realized narration of the whole. The epeisodia are the indispensable scenes which make up the work. In the first stage of production, the poet forms a general plan; this is followed by the invention of the characters, after which the epeisodia are worked out by the poet following the plan developed earlier in stage one (Nickau 1966:161). The epeisodion is thus anything but a non-essential digression; in fact, it marks the first and most basic realization of the abstract action plan. Nickau accordingly concludes that the term epeisodion denotes the implementation of all the units of the action (Nickau 1966: 163).144 We can see, therefore, as early as sense (b) of the Aristotelian term 143

144

‘Anfang und Ende einer Sache diese nicht transzendieren, sondern in ihr selbst liegen. Die Ganzheit ist deshalb im vollkommenen Schluß verwirklicht, in dem nichts außer den Prämissen notwendig ist, um ihn evident zu machen. Das Modell des vollkommen Schlusses steht darum hinter der engeren Bestimmung der tragischen Mimesis.’ Nickau’s definition of the epeisodion as a concept of epic poetics places the episode at the same level that it has when it is used as a dramatic technical term in the sense of ‘act’ or ‘scene’ (i.e. Dammann’s sense (a) above). In the context of Aristotle’s systematics and the rhetorical intention of the Poetics, this is convincing. In historical terms, however,

1.4 The Elements of the Action Construct

155

epeisodion, a concept which clearly takes into account the constructive nature of the work of art. Unlike digressions and interpolations, which are episodic in a negative sense, epeisodia are not non-essential, and nor is their function in the action sporadic in the sense of motivationally or causally linking the syntagm of happenings at separate points. Instead, each and every epeisodion points to the complex mythos. This shows that the concept of the epeisodion as we find it in Aristotelian poetics is far more suited to a model of the constructive concept of a basic action than the modern word ‘episode’, which vacillates between descriptive content and normative connotations and, in the worst cases, has degenerated into an apologetic or critical term with which readers describe their impressions. Nonetheless, it would be risky to develop the constructive concept of the episode in our event-based ontology solely on the basis of the Aristotelian epeisodion. Quite apart from crossing a hiatus of over two millennia, we would face the particular problem of reconciling Aristotle’s theory of production with our theory of constructive reception. It is, therefore, particularly important to remember that the use of the word ‘episode’ in a constructive, positive sense that can be traced back to Aristotle was subject to a renaissance and re-evaluation when popular narrative forms became the object of critical study in the nineteenth century. As well as the fairytale and the folktale, entire genres like the courtly romance and the picaresque novel made it possible for critics to appreciate the episode as a constituent unit of action narratives and see the syntagmatic connection of episodes as one of the principles of the such narratives (Dammann 1982:70f.). The ancestry of this constructivist concept of the episode as a principle of narration can be traced back even further. For example, in the Renaissance, it was manipulated with virtuosity by authors and obviously enjoyed by their critical audiences, as Riggs observes: The Renaissance critic did not think of the episodes as an interlocking chain of incentives, motivations, and explanations which assist the serial development of the main plot.…The relationship is a logical one, but it is to be explained by a logic of resemblance rather than a logic of cause and effect. The aptness of the analogies, rather than the intrinsic vividness of the material, is what ‘delights’ the discerning spectator. Even though the episodes do not assist the the analogy is slightly contrived. The episode of the early Attic tragedy, in particular, was more a narration of and commentary on action than a mimetic portrayal of it, so, at that stage, the episode did not have such a fundamental, normative significance. As a purely descriptive term, it denotes a compositional element which is framed by prolog, parodos, and exodus and interrupted by numerous chorus parts. (Barthes 1985:65–69). Haidu arrives at similar conclusions by arguing, more semiotically than logically, against the pejorative use of ‘event’; he presents as evidence the claim that, as early as Aristotle, ‘episodes can be viewed as a necessary mode of textualization of the abstract plot’ (Haidu 1983:657).

156

Part 1 Concepts of Action

serial development of the plot, they are meant to tell us something about it. The episodes remind us that other versions of this story are conceivable, that other outcomes might have been possible. Riggs 1973:171

If the Renaissance saw the episodic as the aesthetic expression of a welcome philosophical emancipation from established doctrine and the prescribed avoidance of ambiguity, this judgement was reversed once again in the seventeenth and eighteenth centuries. The episodic structure of both the German and English novel shows clear parallels to the insecure world view of the time.145 The collapse of the idea of philosophical unity brings with it the downfall of the idea of aesthetic unity. The epeisodion through which the coherent whole was syllogistically and poetically revealed has become the episodic narrative which is symptomatic of a consciousness which is capable only of fragmentary perception. In Kant’s philosophy, as in classical aesthetics, a new synthesizing force is found, in which the postulate of the closed work need no longer assume that the coherence of things teleologically exists. Instead, the subject itself is provided with an attitude which is capable of removing fragmentation. Thanks to a restored capability to believe in transcendental explanations, the connection between perception and aesthetics was reassessed again by the Romantics.146 Finally, Hegel returned to Aristotle’s premise of the closure of the whole, but in doing so treated it emphatically as a closure which must be produced aesthetically to allow the reconciliation of mind and nature. The loss of the ability to find narrative closure has continued to be celebrated and lamented alternately to the present day, not just in philological but also in historical disciplines. The above historical outline makes two things clear. First, over the course of time, the premodern concept of the episode has become increasingly directed at thematically and motivationally focused event sequences whose dimensions and complexity transcend those of the conventional 145

146

See Riggs (1973) and Pavel (1995), who compares the action structures of English and French neoclassical/Renaissance dramas. On the English novel of the seventeenth and eighteenth centuries, see Beasley (1988). There are many different methodological perspectives from which the issue of episodic narration can be discussed. Some examples are Bloomfield (1980), who examines its stylistic function and sees it as a purely superficial phenomenon; Dodd (1981), whose discussion concentrates on episodic constructions in the drama and how the recipient connects distinct episodes; Taylor (1971), who uses the picaresque novel to illustrate the possibility of constructing a typology of genre-specific episode templates; and, finally, Sugars (1996), who discusses the correlation between episodic and chronological structure. The list could be extended still further, but even these few examples are sufficient to illustrate the wide range of contexts in which the phenomenon of episodic segmentation can be heuristically relevant. See Frank (1984) on fragmentation in Romanticism.

1.4 The Elements of the Action Construct

157

epic or dramatic scene and lie above the basic narrative unit that Aristotle originally called an epeisodion. Second, as a structural feature of the symbolic representation of groups of events, the episodic can no longer be understood in terms of the mental assumption that there is a logically closed coherent world which can and (as a normative consequence) should be narrated as one that is closed. The episode has become a purely aesthetic phenomenon; the modern episode has nothing left in common with the logically and ontologically based Aristotelian epeisodion apart from their shared etymological stem. The history of the concept of the episode is not the only reason why it is untenable to cling to the absolutism of its contemporary, aesthetically normative definition. A glance over the philological horizon reveals that the principle of modular segmentation applies not only to the narration of world and action but also to the much more basic levels of cognition and the pragmatically functional linguistic representation of empirical reality. This is why discourse analysis, cognitive science, and psychological theory all use the term ‘episode’ to denote basic thematic structures of minimal complexity. For example, discourse analysis argues that the multi-party conversations of everyday communication have a modular construction which is based on discourse episodes whose overall coherence is, as a rule, preserved despite their segmentation. In their Topical Episode Analysis (TEA) model, Linell and Korolija (1997) introduce a total of eight kinds of episode change, which can be identified by linguistic markers that also occur in narrative discourse.147 Significantly, these markers are formally defined as conjunctive, not disjunctive, features; total incoherence is practically inconceivable, given that the concern of discourse analysis is the pragmatic question of how the overall coherence of discourse remains preserved despite the changes of topic, or even speaker, from episode to episode. Nonetheless, as the authors note, this of course still leaves some vexing questions open: What is ‘real’ incoherence like in discourse and interaction? Or do our minds always build coherence, even when we are faced with what seems textually totally incoherent? Linell and Korolija 1997:199

Like modern linguistics, discourse analysis, and cognitive theory, studies of older languages also take as their starting point a formal definition of the 147

In Topical Episode Analysis, the boundaries of discourse episodes are determined according to four main criteria: change of referents, which are thematized in new situations; change of the participants in the discourse; change of prosodic level; and change of the contextual resource domain. The classification of the possible types of episode change which the authors derive is of note not least because it could be applied analogously in narrative theory. See Linell and Korolja (1997:176–88).

158

Part 1 Concepts of Action

episode which treats it as a modular unit of a progressively constructed whole. Aristotle’s basic idea, which introduced the constructively produced epeisodion as a correlate of the basic logical structure (which he saw as entelechially based), has therefore remained relevant in studies which follow a descriptive approach instead of being normative or poetically prescriptive. A contemporary example of this is van Dijk’s ambitious attempt to put forward, to quote the subtitle of his book Macrostructures, an ‘Interdisciplinary Study of Global Structures in Discourse, Interaction, and Cognition’ (van Dijk 1980). Van Dijk basically treats discourse, interaction, and cognition as processes which take place in the zone of tension between basic propositions on the one hand and determining superstructures (world knowledge and conventions) and cognitive sets (normative attitudes of participating subjects) on the other. The purpose of these processes is to constructively build the synthetic macrostructures without which neither discourse, interaction, nor cognition can be meaningfully organized or semantically functional. Van Dijk’s macrostructure occupies the same position in the middle of the system as that which we have chosen for our episode, that is, it is above the level of the basic states of affairs but below the level of the complex global structures. Because of this similarity, it will be worthwhile to examine van Dijk’s hierarchical system in more detail. At the lowest level, we find propositions. They can combine to form complex propositions which express a fact, which van Dijk defines as ‘a cognitive representation of one state, event, process or action’ (van Dijk 1980:21).148 The fact in van Dijk’s system is therefore analogous to our event. A fact-sequence is then a ‘representation of a whole scene, episode, discourse’ which has been drawn together into a macrofact (van Dijk 1980:21). In the broadest sense, a singular fact corresponds to a sentence, a fact-sequence to a sequence of sentences. The constructive combination of facts to produce the macrostructure of a diachronic fact-sequence or a synchronic macrofact has to observe a certain coherence condition: we must turn to the superstructure of our world knowledge in order to investigate the extent to which there is a relation of conditional, generic, or logical derivation between the basic facts under discussion (van Dijk 1980:33–36).149 In the case of event sequences, the scale of conditional connections ranges

148

149

A survey of the components of the fact can be found in the diagram in van Dijk (1980: 22). According to this diagram, at least eleven atomic propositions are required in order for the categories which make up the fact (i.e. action, participants (agent, patient, goal), world, time, and place) to become functional. ‘fact’ has small capitals in van Dijk. Here, as before, it is somewhat unclear whether van Dijk is treating the fact-sequence and the macrofact as constructive or representational phenomena.

1.4 The Elements of the Action Construct

159

from weak ‘compatibility or copossibility’ to strongly causal ‘necessity’ (van Dijk 1980:33; emphasis in the original). Sequences of states, on the other hand, can be considered connected when the individual states appear in an identical world and situation and when hierarchical relations such as ‘general-particular, whole-part, and container-content’ can be formed between them (van Dijk 1980:35; emphasis in the original). The system is designed primarily from the perspective of discourse analysis, which would seem to make it more suitable to be borrowed by production theory (as opposed to reception theory) for the purposes of defining the concept of the episode. However, the comparability between van Dijk and our approach which we asserted above is distinctly more apparent from the perspective of cognitive theory, to which van Dijk himself subsequently draws attention. Here, he places the global structure of a complex narrative (the narrative) at the same level as the global structure of a logical argument (the argument). This is clearly demonstrated in the following pair of structural models, which I have placed alongside one another for easy comparison. The central terms for our purposes are highlighted in bold.

Fig. 1.4.1: Structure of Narratives150

150 151

After van Dijk 1980:116. After van Dijk 1980:119.

Fig. 1.4.2: Structure of Arguments151

160

Part 1 Concepts of Action

As can be seen from figures 1.4.1 and 1.4.2, the narrative episode is explicitly interpreted as analogous to the logical inference; this interpretation stands firmly in the tradition of the Aristotelian epeisodion. The same tradition is also the home of the concept of episodic memory in cognitive theory, which uses it to refer to the class of acts of memory which operate not on the basis of sequentially stored raw data but on the basis of data structures which have been cognitively processed (i.e. internally connected and/or globally interpreted).152 Van Dijk’s starting point is an investigation in cognitive theory with which he hopes to shed light on the surface level logic of the processes of discourse, interaction, and cognition in which macrostructures are assembled by applying certain macrorules.153 Our attention, however, is directed at the logical, semantic deep structure of the narrative episode. Van Dijk’s approach aims to provide a detailed description of the complex interdependencies between empirically registered facts; preformed macrostructural templates stored in memory (schemata, frames, and scripts); contextually determined normative superstructures; and the physiological conditions of human cognitive operations (retrieval, reproduction, and reconstruction). We, however, are investigating the necessary conditions for the creation of coherence and meaning in and by episode constructs, and we hope to derive these conditions from the inherent logical relations between the constituent events (i.e. facts) of which an episode is composed. To do so, we shall have turn to a more abstract model than van Dijk’s. The Formal Definition of the Episode Attempts to define the episode in formal terms can be found in various studies of narrative grammar, irrespective of whether their disciplinary focus is that of narrative theory, discourse theory, or cognitive theory. In 152

153

For example, Shen (1985) discusses how the recipient divides the (re)constructed narrative units into a hierarchical system in order to differentiate formally identical units according to their informative value. Shen (1990) then examines the special role which the identification of episodes plays in remembering stories; Noice and Noice (1993) investigate how dramatic episodes are remembered. From the point of view of empirical cognitive psychology, a more cautious evaluation of the relevance of structure building for remembering stories seems appropriate. For example, Hasselberg-Weyandt argues that content, not structure, is primarily retained, although the structural context of such content is nonetheless necessary for it to be remembered (Hasselberg-Weyandt 1985: 104). Van Dijk postulates three macrorules (deletion, generalization, and construction), all of which can ultimately be derived from the underlying pattern of the logical syllogism.

1.4 The Elements of the Action Construct

161

most of them, however, either their categorial apparatus does not give the episode a clear position, or their definitions return to the commonsense, psychologizing concept of action and measure the formal coherence of an episode’s event structure according to whether a fictional content (usually the motive or intention of an agent) which can only be discovered by interpretation has been realized. Todorov, a narratologist, succumbs to the first of these shortcomings. He uses the term ‘episode’ to refer both to a singular description of a state and to a basic transformation of state (‘passage from one state to another’; Todorov 1977:111). The latter definition reappears in Shen’s XBar Story Grammar, in which the episode is defined as ‘the basic action unit, assigned to an “agent” who attempts to solve a problem’ (Shen 1988:650; small capitals in the original). Pavel’s move grammar, which we described earlier, combines the methods of Todorov and Propp but restricts the scope of the basic episode, or move, to the extension of a singular transformational event in the sense of Todorov’s definition. In Rumelhart’s story schema, we find the opposite situation. Here, there is a more extensive concept of the episode which treats it as a basic four-place sequence of activity, cause-problem-try-outcome, which includes the causal originator (cause) of the problem. According to the schema model (Rumelhart 1975), the episode can cover both basic and complex event sequences in so far as each of its four components can subsume one or more subordinate episodes. Mandler and Johnson (1978) follow Rumelhart and treat the episode as a recursive category which allows embedding. They extend Rumelhart’s model to a chain of five elements with which, instead of inductively describing the activity of a fictional agent in terms of action logic, they hope to be able to deductively construct thematically coherent sequences of fictional happenings from the recipient’s perspective. Episode → Beginning Cause Development Cause Ending Mandler and Johnson 1978:345

This approach treats the closure of the event structure as something which is brought about synthetically during reception. In this respect, it anticipates the constructive definition we hope to provide, but it falls short of our requirements in one key respect: for Mandler and Johnson, the sequence of happenings in this kind of episode does not cover the entire fictional object domain of a complete autonomous story.154 In their view, there 154

According to Mandler and Johnson, a text must contain at least four propositions in order to be treated as a narrative. These propositions are represented by the nodes setting, beginning, development, ending. See Mandler and Johnson (1978:345; small caps in the original).

162

Part 1 Concepts of Action

are two required elements in a closed narrative: a setting (a scene that introduces protagonists, antagonists, and context) and an event structure (which consists of at least one episode) (Mandler and Johnson 1978:343f.). In fact, it is only the setting that guarantees the coherence of the episode—the elements of the latter can be linked only because of the fact that they can be treated as parts of an identical segment of the fictional world. We find this idea in a different form in Haidu (1983), who postulates that the episode can be formally defined as a syntactical unit of narrative structure, modular in form and serial in content, that (qua episode) is subsidiary to a principle of textual coherence located not in the referential function of a category like ‘character’, but on the semantic level (insofar as the text is coherent). Haidu 1983:680

As a simple syntactic unit which can only be conceived of as a coherent unit of meaning ex post facto (i.e. from the synthetic perspective of the completed overall story), this concept of the episode depends on the assumption of a superordinate normative paradigm. It follows that generative narrative grammars are unable to define the nature of this kind of episode inductively by using context-independent, formally defined syntactic rules with which narrative units are derived and combined analogously to linguistic sentences. This is, in fact, a general property of the majority of narrative grammars. In various forms, they outline a purely syntactic rule for generating narrative sentences; it basically describes the process of transforming a first proposition into a second one. Nothing in the sentences generated by this rule, however, indicates the kind of action which they can or must combine with one another to form. Only by reference to an external normative system whose categories can be assigned to narrative sentences as and when they appear can it be decided what the cause, problem, effect, and so on actually is. Even in its most abstract form, which makes no reference to concrete information drawn from intertextual allusions or learnt world knowledge, the normative system is typically based on the strategic action structure, a concept which describes activity as an intentional agential doing aimed at solving a problem (Shen 1988: 639–42).155 Given what we have said above, will our definition of the episode be able to avoid resorting to a teleology of purposeful doing such as is key to the naive intentional concept of action? At first glance, a recent proposal

155

This observation applies to the inductive generative story grammars of Rumelhart and Prince and to Pavel’s move grammar in that they derive constituent events from a logic of problem-solving decisions as paradigmatically described in Bremond (1966, 1973).

1.4 The Elements of the Action Construct

163

by Tjupa seems to provide a way out of the dilemma. Tjupa attempts to define the episode as a unit of the what he calls the current segmentation of a narrative discourse (i.e. the segmentation manifested at the surface level of fictional happenings).156 In doing so, he describes the episode as a linguistically realized textual segment which is characterized by the unities of time, place, and character. The boundaries between the individual episodes of a text are manifested in a) a change of space; b) a shift in time; c) the appearance or disappearance of a character or group of characters. Any one of these features is sufficient for the formation of a new episode; the quantitative extent of changes of space and shifts in time is of no importance. Tjupa 2000:1 (my translation)

This definition of the episode is based on a spatial metaphor, the segmentation of a sequence of happenings into coherent scenes. Tjupa’s concept of the episode is largely equivalent to what Mandler and Johnson mean by setting. It avoids referring to an intentional concept of action, but only at the cost of introducing a new difficulty. In Tjupa, events are explicitly not involved in episodes as constituent elements; instead, they mark the boundaries between them, a situation which originates in Tjupa’s use of Lotman’s concept of the event. Thus, according to Tjupa’s definition, the term ‘episode’ denotes a complex of contemporaneous fictional states of affairs and is thus nothing more than the contrastive foil against which events and, ultimately, a complex action unfold. This is of no use to our aim of defining the episode as a dynamic structure and minimal action which is built out of events. None of the approaches we have discussed so far has been able to provide a satisfactory explanation of the episode as a formally coherent and semantically connectable basic action in the sense of the Aristotelian epeisodion. This leaves us facing two tasks. First, we have to determine the logical semantic criteria which necessarily govern the integration of events into the receptive construct of an episode. Second, we need to explain what makes it possible for an individual episode to be linked to other episodes and contribute to a complex action construct.

156

The current segmentation of the text is determined by the syuzhet. Tjupa distinguishes this from the virtual segmentation, which is the abstract sequence of events (in Lotman’s sense of the event as the crossing of a semantic boundary by a figure), that is, the fabula of the narrative text. See Tjupa 2000:1.

164

Part 1 Concepts of Action

The Semiotic Square: A Reappraisal If we assume that the semantic structures developed above the level of the individual words of a narrative text are not merely products of coincidence, an unrestricted aleatory logic or combinatorial process is not sufficient to explain the formation of constructs such as action and the episode. This applies equally to the processes of production and reproduction (interpretation); in neither case can we arbitrarily connect anything with anything else. That is to say, we cannot arbitrarily combine any narrated state of affairs with any other state of affairs, nor can we freely combine the resultant events in any order whatsoever so as to produce one or more episodes. One of Propp’s well-known central theses is that, even when seen in isolation, the constituent elements of a narrative contain features which predetermine their place in a global structural plan which is yet to be realized. His Morphology of the Folktale argues not only that the magical fairytale has a finite set of functions, but also that these functions can appear only in particular positions in narrative discourse. It remains disputed whether there is a finite number of constituent elements on which all action narratives can draw. But when a specific textual corpus is considered (and a descriptive perspective thereby adopted), it is virtually impossible to refute Propp’s conclusion that the narratives of a corpus are constructed using a finite pool of narrative elements. A similar judgement (qualified by the same reservations about shifting from theoretical argument to empirical observation) applies to the assumption that these constituent elements are in addition endowed with what we might call an action specification gene which is presumably associated with a particular genre and determines the position and logical function which a given element can occupy and perform in an action. Analogously to Propp, we shall base our argument on the theory that neither in narration nor in reception are the states of affairs and events in a fictional text subject to a blind combinatorial process which takes a brute force approach and works through all theoretically possible arrangements and permutations before making a selection from these possibilities under the guidance of arbitrarily specified external (aesthetic, philosophical, rhetorical, etc.) criteria. We shall assume instead that the combinatorial constructive formation of episodes (and subsequently actions) is governed by both the semantically ambivalent, purely formal criteria of a mathematical system and the more restrictive criteria of a semantic system. In addition, however, the debate surrounding Propp’s model has shown that the formalist model of the composition of narrative structures runs the risk of falling victim not only to the Scylla of arbitrariness but also

1.4 The Elements of the Action Construct

165

to the Charibdis of inevitability. Nothing has provoked so much criticism of the Morphology of the Folktale as the thesis that each individual function is by necessity assigned one and only one unalterable position in the chain of thirty-one functions. The hostility to this thesis is rooted in a misunderstanding of the status which Propp attached to the Morphology; the generalization of the observations he made on the basis of a particular textual corpus in order to produce a cross-generic narrative theory is really the work of Propp’s interpreters, not Propp himself. Even so, the idea of narrative inevitability, of whatever kind, has quite rightly been called into question (Pavel 1988). Converting regularities detected in dubious empirical observations into postulates of theoretical regularity is one of the favourite and most questionable habits of literary critics.157 As far as our present discussion of the theory of the episode is concerned, our objective is explicitly not to arrive at a formula which would, without exception, allow one and only one episode to be built out of two particular events at any given place in the continuum of fictional happenings. The best chances of increasing our understanding of the interpretive event construct lie not in deductive or generative axioms but in a combinatorial system which can be heuristically evaluated. Bearing in mind our investigation of the ontology and perception of the concept of action, we need now to investigate the necessary conditions of the construction of the episode (as opposed to the sense of the construct in sociology or reception psychology). We shall assume that the key function of the episode construct is to create a kind of meaning which lies above the construct’s individual fictional reference to its constituent elements but which is not thereby relegated to being an abstract categorial concept or an arbitrary thematic bracket established ex post facto. We propose that the meaning of the episode is neither a static product nor a reference which has been determined in advance and needs only to be decoded. Instead, the semantic function of the episode as a construct of interpretation takes the form of a process which manifests itself in the synthesizing effect with which this construct, located at an intermediate level of complexity, contributes to the formation of the structure of the overall narrative action. Working inductively, we construct episodes out of semantically biased events in accordance with a particular compositional logic, but we do so without having committed ourselves to the unwavering pursuit of a particular overall action to its completion. To search for an immanent logical and semantic necessity in an aesthetic action is to ignore the fact that such an action is not just a construct of 157

The fundamental methodological criticism of this practice in Pasternack (1975) is still valid today.

166

Part 1 Concepts of Action

interpretation but also a historical construct of interpretation. The same events could theoretically have been combined into different episodes, the same episodes into different actions, but, under the particular conditions of cultural and cognitive selection present in a given case, only a tiny subset of the theoretically possible constructs is acceptable. Thus, if a certain inevitability is, metaphorically and ex post facto, attributed to an aesthetic action narrative, it is attributed in a normative thetic manner and should not be misinterpreted as an absolute logical property. A number of critical methods have the potential to mislead us into adopting the latter, mistaken attitude—for example, the descriptive systems typical of narrative theories which are based on poetics and focus on production, aiming to classify the narrative text by means of a closed set of descriptive terms and categorize the relations between its parts and the whole; and the historical systems of literary history, in which every motif and every text can only be seen in terms of preformed intertextual connections. Both these approaches blind us to the exploratory, knowledge-producing aspect of reading events, episodes, and ultimately actions, none of which are formed or conceived in advance. Our aim, therefore, is to define the episode as a construct which explores meaning by combining certain semantic atoms—events—into the minimal closed sequence of an action. In this respect, our model of the episode is based on Greimas’s model of meaning which is strikingly expressed in his well-known semiotic square. The adaptation of Greimas’s model requires further comment, not only because it is of central importance to our approach, but, above all, because with it we are adopting a model which is an obvious archetype of classical structuralism of the 1960s and therefore rather unfashionable at the present time. If, apart from that of Roland Barthes, any name is representative of the unmatched rigour of the structuralism of this period, it is that of Greimas. Mentioning him cannot but provoke a fundamental methodological protest: is it not a contradiction in itself to use Greimas’s orthodox structuralist model of the semiotic square as the foundation for a constructive and exploratory model of the episode? Greimas’s position as a semiotic structuralist, if not in more extreme terms a committed disciple of Hjelmslev and a linguistic formalist par excellence, is most clearly illustrated by his much-quoted essay “Elements of a Narrative Grammar,” in which he draws explicitly on Propp, LéviStrauss, Bremond, and Saussure.158 Thirty years after it was first published, 158

The original article is “Élements d’une grammaire narrative” in Greimas (1970:157–83). In the following discussion, I draw on both the German translation (“Elemente einer narrativen Grammatik”) by Irmela and Jochen Rehbein (Greimas 1972:47–67) and the English translation by Paul J. Perron and Frank H. Collins (Greimas 1987:63–83).

1.4 The Elements of the Action Construct

167

Greimas’s central thesis that narrative surface phenomena can be derived from a universal semantic deep structure remains sufficiently controversial to provoke knee-jerk protest and criticism of his work. Greimas’s ideas have not been widely taken up by German theorists, but his combined inductive and generative approach has met with considerable interest among AngloAmerican critics. Greimas’s theories, particularly those in his early writings from the 1960s, have been criticized in both circles less because of the idea itself of extending the linguistic dichotomy of surface and deep structure to narrative texts, and more because the resultant theory cannot be put into practice, let alone tested. Even Greimas’s most committed supporters have to accept that he does not describe any clear rules for mapping the deep structure of logic onto the surface structure of discourse. This obvious shortcoming prompted Pavel to compose a detailed assessment of the programme of narrative semiotics, in which he draws his arguments from methodology and the history of theory. He sees Greimas’s semiotic square as a classic example of the programme’s uncritical adoption and synthesis of three sets of theories: (1) Saussurean linguistics, the source of the theorems of the non-referential nature of language and the concept of meaning as consisting of binary oppositions; (2) Lévi-Strauss’s analysis of myth; and (3) Propp’s writings, which the Greimassian school has exploited while completely ignoring the fact that, as we noted above, Propp’s Morphology puts forward a genre-specific hypothesis rather than a generalizable theory of narration (Pavel 1988:598ff.). Greimas’s key idea, using a generative trajectory to model the derivation of the textual surface structure from the semiotic square, is dismissed by Pavel as untenable. He argues that Greimas completely fails to explain the underlying concept of generation sufficiently, as a result of which it is not acceptable to employ the concept at all. To him, Greimas’s approach is essentially a scientistic theory of knowledge; in his view, its generative pretensions need to be abandoned and the formalist approach applied with much greater flexibly.159 Adopting a somewhat less polemical approach than Pavel, Ohno (1995) does not examine the theoretical framework as such, but rather the key concepts involved in the Greimassian model. She argues that three essential paradigms are central to the Greimassian analysis of meaning: reference, difference, and standardisation. Like Pavel, Ohno points out a number of weaknesses and inconsistencies in Greimas’s model; unlike Pavel, she does not question the validity of the model itself. Instead, she argues for a looser application of the semiotic square, to which she gives 159

‘Formal systems [are] not the only way to knowledge, narrative semiologists may choose to stress the most valuable aspect of their discipline through more open hermeneutic practices, the use of heuristic models or any other softer kind of rational argumentation’ (Pavel 1988:605).

168

Part 1 Concepts of Action

a key role as ‘an aid to interpretation in various analytical contexts, for example, in the interpretation of texts and other semiotic complexes in the widest sense of the word’ (Ohno 1995:33; my translation). The aim of this last section of part 1 is to apply Greimas’s semiotic square in precisely this way, as a descriptive heuristic model rather than an inductive generative one. Standing the model on its head in this rather audacious way is legitimate not least because such a move is suggested by the arguments of Bremond, whom we have to thank for what is, as far as narratology is concerned, the most penetrating analysis to date of Greimas’s modèle constitutionnel. Bremond’s early criticism already concentrates on what he feels is Greimas’s untenable claim of taking the deep structure of the semiotic square and deriving from it a logically inevitable sequence of narrative statements at the surface structure of discourse. Nonetheless, Bremond believes that the model itself is still worthy of our attention: What is our conclusion? In our opinion, the ‘constitutive model’ which inspires Greimas is not radically false or dysfunctional in its applicability to the structures of stories. Rather, since the essence of the notion of narrativity itself has not been clarified, the model is put to the test under the conditions of stereotypical assumptions which disguise rather than reveal the combinatorial possibilities of the language of narrative. Bremond 1973:98f. (my translation)160

Bremond is particularly concerned with keeping combinatorial possibilities open, not least because this is in the interests of an aesthetics of freedom which he considers threatened by Greimas’s pretentious emphasis on logic. Thus, he calls for a revision of Greimas’s modèle constitutionnel, stating: For him [Greimas], the essence of a narrative is concentrated in a game of atemporal concepts which transcends the development of their related events; for us, it is inherent in that development itself and varies between a multiplicity of possible futures, one or the other of which is chosen in order to continue and finish the narrative. For us, the impression of freedom, of truth, of beauty…is not an illusion that masks the ‘game of semiotic opposites.’ If there is a game at all, then, instead of being subordinate to these constraints, it is a game ABOVE them, a liberating experiment which exploits and goes beyond them. Bremond 1973:101 (my translation)161

160

161

‘De là, que conclure? Non pas, à notre avis, que le ‘modèle constitutionnel’ dont s’inspire Greimas est radicalement faux ou inopérant dans sons application aux structures de récit. Mais que, la notion même de narrativité n’ayant pas été dégagée dans sons essence, ce modèle est mis à l’épreuve sur des stéréotypes qui voilent, au lieu de les manifester, les possibilités combinatoires de la langue du récit.’ ‘L’essentiel du récit se concentre pour lui [Greimas] dans un jeu de concepts intemporels, transcendant au devenir des événements racontés; pour nous, ils est immanent

1.4 The Elements of the Action Construct

169

Bremond’s rhetorical comparison of form and content and inversion of the logical relation between them is intended as an aesthetic response to the logical semantic determinism of the modèle constitutionnel. Those who join Greimas in arguing that all possible meanings are prefigured in an abstract sense can give the work of art no more credit than that due to it as the creator of an illusory game in which the apparent openness of the possibilities for narrating different meanings is no more than a masquerade. The alternative suggested by Bremond is that we should understand the aesthetic meaning of action structures (récit) as the product of a free, emancipating act of combination which is based on, yet at the same time liberates us from, the abstract and uninteresting level of semiology that consists of the logical conditions of meaning. If the fabulas of action narratives are subject to semantic determinism in Greimas’s model, Bremond sees the fabula as the exact opposite: a product of aesthetic emancipation which applies the principles of openness and combination so as to articulate itself above and beyond prespecified patterns of meaning. The generic above/below arrangement asserted by Greimas is not only turned on its head but also transformed from a logical into an aesthetic arrangement.162 From this perspective, the objections of later critics to Greimas’s excessive semiological formalism and his logical and terminological inconsistencies appear, if not downright pedantic, then at least highly contrived. For not only is it possible to read Greimas against the formalist and determinist grain—his work positively invites us to do so. Apart from obvious logical inconsistencies,163 Greimas’s key essay “Elements of a Narrative Grammar” prompts us to take just such an approach. It contains a striking contradiction between spatial metaphors and methodological self-de-

162

163

à ce devenir même, et variable selon qu’entre plusieurs futurs possibles, tel ou tel est choisi pour continuer, puis finir le récit. Pour nous l’impression de liberté, de vérité, de beauté…n’est pas une illusion qui masque le ‘jeu des contraintes sémiotiques.’ Si jeu il y a, celui-ci n’est pas subi, c’est un jeu SUR les contraintes, l’expérience libératrice qui les exploite et les sur-monte.’ I have quoted above from chapter 4 of Bremond’s Logique du récit (Bremond 1973), originally published in the journal Semiotica (vol.V/1972, 262–82). Bremond had been critical of Greimas from the beginning, as his essay “Le message narratif” (Bremond 1966, also in Bremond 1973:11–47) shows. It refers to Greimas’s lectures of 1963–64, which were themselves subsequently published in Sémantique structurale (Greimas 1966); see Bremond (1973:11, note 4). It is interesting to consider the chronology of these publications; as can be seen, Bremond’s criticism of Greimas in 1973 anticipates the post-structuralist criticism of scientism, to which formalist narratology and semiotics in their entirety were subsequently subjected. We shall examine the inherent contradictions of the semiotic square in more detail below (e.g. the inverted implication relations between the sememes s1 < –s2 or –s1 < s2). See also Ohno (1995:331) and Kalinowski (1981).

170

Part 1 Concepts of Action

nunciation.164 As he himself declares, Greimas hopes that his work will contribute to the general semiotic theories which ‘attempt to account for the articulation and manifestation of the semantic universe as a totality of meaning belonging to the cultural or personal order’ (Greimas 1987: 64). In the context of such a programme, meaningful narrative structures are given logical primacy because they exist independently of the media in which they are manifested. For this reason, Greimas believes that we need a fundamental methodological reorientation which liberates us from inductive approaches. He calls for a change in methodological position to enable the use of deductive methods in future research: Formerly we could believe that the linguistic project consisted in putting into place a combinatory or generative mechanism that, starting with simple elements and original kernels, would account for the production of an unlimited number of utterances, with the latter in turn being transformed and combined in order to create successions of utterances in discourse. Now, on the contrary, we have to conceive of ab quo instances of generation of signification such that, starting with agglomerations of meaning that are as little articulated as possible, we can, as we descend through successive levels, obtain more and more refined significative articulations. Greimas 1987:64 (my emphasis)

Let us summarize: the ab quo of signification is a meaning which is logically both primary and assumed but not yet articulated, that is, it has not yet been formulated in successive linguistic statements and is initially pre-existent in synchronic form alone. The process that mediates between it and a manifest, linguistically articulated ad quem is what Greimas calls a trajectory. Narrative structure provides the vehicle for this deductive process, which starts from above (from the abstract, logically superior formal meaning) and moves down (towards the diachronic, syntactically articulated meaning manifested in discourse). Note that the metaphors ‘above’ and ‘down’ are used in a logical sense, as methodologically befits the project of a deductive, descriptive approach. However, it is easy to misinterpret them as signs of genetic formalism because of the fact that Greimas chooses to refer to the logically superior structure as a fundamental structure, or, even more confusingly, as a deep structure. Terminology, methodology, and metaphor contradict one another. In this respect, the criticism of Bremond which we quoted above touches on an obvious weakness in Greimas’s formulation, if not in his concept itself. Moreover, as Quéré and others have pointed out, the interpretation of Greimassian

164

The French version was first published 1969 in the journal L’Homme, vol.9:71–92. The essay was then included in Greimas’s book Du sens I (Greimas 1970). In the following, I quote predominantly from the English version (Greimas 1987).

1.4 The Elements of the Action Construct

171

semiotics and narrative grammar is heavily dependent on the particular critical context in which they are dealt with. In Anglo-American contexts, the narrative grammar is, first and foremost, interpreted inductively and generatively in a way which is clearly based on the model of the practical syllogism (an attitude which may be seen by its supporters as justified by Greimas’s increasing interest in a modal or actant-based grammar in the course of his work). In the context of continental Europe, on the other hand, Quéré (1992:xi) considers that the emphasis lies in evaluating the narrative grammar as a hypothetical deductive model. What then, we may ask, is the real purpose of Greimas’s work? Is the modèle constitutionnel meant to be generative or descriptive? The answer to the question differs depending on the methodological and theoretical context in which it is posed. What is certain, however, is that the descriptive approach only comes to the fore in Greimas’s later writings when we survey his work from a historical perspective.165 This shows us that, even in its less rigid interpretation, Greimas’s semiotics is a heuristics rather than a set of axioms of meaning and understanding. This descriptive approach is compatible with the non-intentional concept of action we outlined earlier. As before, fundamental importance is attached to the assumption that the anthropomorphic character of all fictional activity in eventful narratives is due to the logic of how we speak about meaning, not to the logic of meaning itself: When developing models of description of narrative structures, it is necessary once again to identify two levels: an abstract deep level and a more concrete surface level. The difference between the two is that the surface level is an anthropomorphic level, because all syntax of natural languages is anthropomorphic. There exist subjects, objects, beneficiaries; qualifications are attributed to subjects, for example.…The deeper level we try to establish is the level of abstract operations, that is to say, operations in which the operating subject is no longer a human subject but, just as science demands, a substitutable subject. This is what guarantees the transmissibility of scientific knowledge. Often people do not understand the necessity I felt to posit the existence of this deep abstract level.—As to the semiotic square, it could be a square or a cube or a circle. The shape is of no importance whatsoever. Greimas and Ricœur 1989:554f.

165

Looking back over the development of semiotics, Greimas himself has clearly stated that his model was developed in order to identify the ‘minimal conditions for the appearance, apprehension, and/or production of meaning’ (Greimas 1989:539). Thus, as he himself says, his work in his early period is concerned with clarifying the necessary (semio)logical conditions of meaning, not with proposing a generative grammar of meaning or narration which could be applied in algorithmic form. See also “On Narrativity,” Greimas’s controversial discussion with Ricœur (Greimas and Ricœur 1989).

172

Part 1 Concepts of Action

The anthropomorphic surface level of discourse is where Greimas locates mental perceptions of activity as a sequence of acts of intentional agential doing. In addition, this surface level also bears, and thus indirectly anthropomorphizes, the sequences of non-agential states of affairs in which fictional activity at the discourse level of the narrative is sequentially embedded. The ‘deep abstract level’ to which Greimas refers above is the very same level as that where we believe the principle of action is located as a logical structure which is a logical predecessor of the discourse level of an eventful narrative but is not in itself a necessary and sufficient condition for the instantiation of a specific structure and set of variables at the surface level of discourse. Here, Greimas is following the central thesis of his “Elements of a Narrative Grammar”: the transition from the deep grammatical level of the syntactic operation which characterizes the semiotic square to the surface grammatical level of syntactic doing is a transformational process of three stages (what Greimas calls a ‘conversion’). The subject of the doing is a variable which is initially ignored and only later employed, at the figurative level, as a human or ‘anthropomorphised’ subject (Greimas 1987:71). It is important here to keep the genetic and logical aspects clearly distinct from one another: what appears before us in the form of an eventful narrative at the surface level of discourse can be derived from an abstract logical form by means of a heuristics such as that provided by the semiotic square. Such deduction provides a virtual and abstract description of the concrete object which we find before us (the fictional happening). Now, this model can indeed be converted into an inductive model in which narratable constructs of fictional happenings can be generated using the concrete semantic elements (figures, properties) provided by the text. But in addition to this basic operation, the inductive process is influenced by other conditions which are heuristically inaccessible and condition precisely what constructed happening is ultimately narrated. Interpreting Greimas in this way, we should note, does not necessarily mean giving the late period of his work priority over the early period. Even in “Elements of a Narrative Grammar,” we can see a significant discrepancy between Greimas’s statements about deep grammar and surface narrative grammar. On one side of the divide is the descriptive deductive approach, on the other the inductive generative method which Greimas later abandons. This is the reason why Greimas describes the semiotic square, a model of deep grammar, as a ‘taxonomic schema’ and the deep grammar in general as a construct of ‘conceptual nature’ (Greimas 1987: 69f.). On the other hand, the grammar of surface structure, to which the second part of his study is devoted, has an increasingly inductive, generative character. This is due not least to the terminological decision to use

1.4 The Elements of the Action Construct

173

the term ‘anthropomorphic doing’ to describe narrative utterances. This opens the door to the confusion of doing and linguistic doing, even through Greimas himself makes considerable efforts to prevent such confusion in his argument (Greimas 1987:71ff.). At the highest level of Greimas’s system, that of the mental perceptions of fictional figures, a narrative utterance such as The king died evokes the idea of a fictional doing that stands in a relation of reference to a happening which is possible in our concrete world. As the result of a speech act, on the other hand, the narrative utterance itself marks a linguistic doing which is undertaken with the intention of communicating information. It is at precisely this point, when Greimas introduces the category of the intention to inform, that the descriptive model becomes generative and the deductive methods inductive. We shall now attempt to model the basic structural element of action— the minimal ordered sequence of events which I propose to describe with the redefined episode concept—analogously to the “Elementary Structure of Signification” (Greimas 1987:49). In doing so we make two assumptions which should be emphasized again here. First, our express concern is to formulate a model of the construction and narration of action which is not a generative theory of production but a heuristic model within a wider theory of construction and reception. We hope to understand how and under what conditions episodes can be read in narrative texts and then combined with one another; it is not our intention to provide a statement about how narrative discourse is made. Second, we attach particular emphasis to the modal verb ‘can’ in the preceding sentence—our model is intended to help us delineate the range of theoretically possible episode and action constructs. Nothing could be further from our intentions than to tie the diverse empirically observable interpretations of a narrative text to the Procrustean bed of universal semiotic necessity. Like all acts of construction, the construction of episodes as we understand them is subject to two laws: the law of a supratemporal logic whose principles control the smallest action that can be read in and out of events, and the law of a historical norm which selects what individuals and societies want, ought, and are able to read and construct under the conditions which apply in any particular case. Furthermore, the reading of the semiotic square proposed here—as a descriptive model which can only cover and represent one of the many aspects of generative logic—does far more justice to Greimassian semiotics than might first seem to be the case with respect to the “Elements of a Narrative Grammar”. In his essay “The Interaction of Semiotic Constraints,” published a year earlier, Greimas himself explicitly pointed out that, as a rule, the actualization and manifestation of a particular meaning is the result of a historically conditioned selection from the considerably larger set of

174

Part 1 Concepts of Action

theoretically possible meanings that a single semiotic system can generate (Greimas 1987:48–62).166 For Greimas, this constraining historicity is essentially an interference phenomenon; he interprets it as the result of the layering of concurrently active semiotic structures and their influence on one another, as the result, that is, of a hierarchically arranged structural complex which is described with the concept of epistemy. Greimas’s friend, Roland Barthes, has demonstrated that such a polysystematic model of meaning can be put into practice and has shown how to apply it to the interpretation of literary texts. He does so in S/Z, a book which we are accustomed to read as the farewell to a strict and scientistic structuralism, a structuralism with which no name is more synonymous—dubiously so, as we have now seen—than that of Greimas. The episode: From Semiotic to Episodic Square We turn now to the semiotic square itself. Greimas (1987) claims that we are dealing with an arbitrary model, one whose ‘shape is of no importance whatsoever’ for his theory (Greimas and Ricœur 1989:554f), but we can safely disregard this assertion as one made simply to defend himself against possible criticism. His approach undertakes to bridge the gap between semiotics and logic; because of this, the square is more than just an illustrative model: it provides the basis of the methodological analogy between the logical and semiotic syllogism which is the key idea in Greimas’s theory (Quéré 1992:x). Not without reason is Greimas’s semiotic square a variant of the logical square attributed to Aristotle but first documented in Apuleius in the second century a.d. (Ohno 1995:322, 334).167 Greimas’s taxonomic schema of signification has its predecessor to thank not only for its four-part structure, which is divided into an upper and lower pair of opposing terms, but also for the idea that there is a logical relation which links the upper and lower levels of the system by relating the terms deductively from the upper to the lower level and inductively from the lower to the upper level. A particular cause for criticism of Greimas’s semiotic square has been the way in which he describes the relations along the vertical axis of his model. Our attempt to reformulate the semiotic square as an episodic square will deviate considerably from 166

167

See in particular Greimas’s deliberations on Hjelmslev’s concept of ‘usage’ (Greimas 1987:59ff.). The logical square was widely used in scholasticism; in Peter of Spain’s Tractatus duodecim (Strasbourg 1514), for example, it is used as a square of oppositions (Piltz 1982: 68f.). To my knowledge, Greimas does not refer to Aristotle or Apuleius in any of his central works.

1.4 The Elements of the Action Construct

175

Greimas in this respect and thereby avoid the peculiar confusion of logic and semiotics which gives rise to the—in my view unjustified—generative interpretation of the square. We begin with the logical square, which illustrates the relations which can be drawn between different types of categorical statements. Apuleius describes it as shown in figure 1.4.3.

Fig. 1.4.3: The logical square168

Even in the form in which it is found in Apuleius, the model does not only distinguish qualitatively between the positive propositions (A, I) on the left hand side and the negative propositions on the right hand side (E, O); it also, simultaneously, distinguishes between the universal and particular propositions (A, E and I, O respectively) which are derived from them.169 For each pairing of possible truth-values, relations of implication and exclusion can be established; thus, for example, either one of the two opposites A and E can be true, or neither of them, but never both at once. Greimas transfers this schema to the structure of signification, replacing the two-place propositions (A, E, I, O) with the basic variables s1, s2, –s2, and –s1, which are placeholders for sememes, the morphemes of meaning. For him, a necessary requirement of meaning is a relation of 168

169

According to Apuleius; after Ohno 1995:322. The original can be found in Apuleius, Opera quae supersunt III. A more detailed explanation of the square can be found in Ohno (1995:322): ‘The letters A I E O—introduced in the Middle Ages—are taken from the Latin verb-forms “affirmo” (“I confirm”) and “nego” (“I deny”) and refer to the quantity (all/some) and quality (positive/negative) of propositions with the same subject and the same predicate. The division of the propositions into universal propositions (A-propositions: “All S are P.” E-propositions: “All S are not P.”) and particular propositions (I-propositions: “Some S are P.” O-propositions: “Some S are not P.”) originates in Aristotle’ (my translation).

176

Part 1 Concepts of Action

opposition S located at the upper level of the system, which the semiotic square reserves for universal propositions. The normative postulate that the structure of all meaning is fundamentally based on oppositions brings Greimas to the decisive conclusion that the postulate can be applied selfreflexively to meaning itself: the concept of meaning is meaningful only when we can place it in opposition to a concept of non-meaning. This opposing concept of non-meaning finds expression at the lower level of the system, which is reserved for particular propositions (I, O) in the logical square and designated as the semantically neutral –S axis (as opposed to the upper, meaningful S axis) in the semiotic square. The logical correlation between the universal and the particular has become the semiotic correlation between the meaningful and the semantically neutral. Even at this early stage, we can tell that the conversion of the logical square into a semiotic square is not going to be entirely unproblematic. Greimas is unable to make his schematic analogy work in practice. In the logical square, four separate propositions are correlated so that we can judge their respective truth-values. The possible range of truth-values is limited to a greater or lesser degree depending on the particular position of the proposition involved (A, E, I, or O). In particular, as mentioned above, only one of the two universal propositions can be true, but it is perfectly possible for both to be false without the square losing any of its logical significance. In the semiotic square, things are different, for it is stricter in the sense that there is a necessary requirement that all the variables (s1, s2, –s1, –s2) must be filled with values at any one time in order for it to be meaningful, the functional equivalent of the logical square being true. However, because of this requirement, the question of the truth of the individual elements of the semiotic square (unlike the propositions of the logical square) does not even arise: they are simply there. For this reason, in Greimas’s early practical examples, the variables contain not two-place predications such as alive(Socrates) but single-place predicates such as alive, dead, and the like.170 There is good reason for this, for a predicate with an empty or false opposite would by definition be semantically empty. In fact, such a dysfunctional opposite term would bring about the collapse 170

Greimas seems to have silently passed over this fundamental difference. In his illustrations of the semiotic square which are based on examples of Lévi-Strauss and Propp, the sememe is immediately replaced with predications like permitted relations versus unaccepted relations. This is of course consistent, in so far as the practical application of the square requires both axes to have a shared factor which is predicated differently in s1 and s2 (or –s1 and –s2) in order for axial homology is to be established. It is questionable, however, whether this conceptual borrowing from Lévi-Strauss is really necessary at all; in any case, it means that Greimas obscures the crucial boundary between logic and semiotics which he insists on elsewhere (Greimas 1971). See, in particular, Kalinowski (1981) for an effective criticism of the logical inconsistencies in the semiotic square.

1.4 The Elements of the Action Construct

177

of the entire semiotic square; the logical mechanics of meaning are, after all, immune to distinctions of the true/false type and permit not only false but also ambivalent and categorially heterogenous attributions (i.e. the drawing of metaphors). Only because of this are we able, for example, to assign the values left-wing to s1 and reactionary to s2 in order to use their supercategorial opposition relation to construct a new systematically formed semantic field in which the complementary terms progressive and right-wing also have a semantic position. Nonetheless, despite this reservation, the model of the semiotic square does not contain any contradictions, at least not in its abstract form where the sememes of which it is composed are represented by the variables s1, s2, –s1 and –s2. Greimas gives the annotated illustration shown in figure 1.4.4.

Fig. 1.4.4: Greimas’ semiotic square171

The relation of implication is the subject of a somewhat laconical remark in a footnote: Although the existence of this type of relation seems undeniable, the problem of its orientation (s1 → –s2 or –s2 → s1) has not yet been settled. We shall not raise this issue here since its solution is not necessary for its demonstration. Greimas 1987:228 (note 1 to Greimas 1987:49) 171

After Greimas 1987:49. From the context, it is clear that the terms –s1 and –s2 have been placed in the wrong position in the original illustration. This has been corrected in the above diagram.

178

Part 1 Concepts of Action

Despite this claim, the note actually concerns the key point which has repeatedly marked the failure of attempts to apply the semiotic square by mapping it onto a narrative surface level (i.e. a concrete sequence of narrative statements). The logical square allows the individual propositions to be integrated into an unambiguous hierarchical system: on the upper axis, we find two axiomatic, logically exclusive universal propositions; on the lower axis, two particular propositions which enter into two-way affirmation relations along the vertical axis. Along the diagonals, the propositions enter into a relation of mutual exclusivity. The crux of all this is that the semiotic square can preserve the logical dichotomy of universal and particular propositions and the difference between relations of deductive and inductive reasoning only at the cost of changing the above/below relation (which is reversible in a logical sense) into an irreversible, unidirectional below/above reference. This is even more important if we want to understand the square not just as a static taxonomic model but also as a model of a dynamic process, as a semiotic map, so to speak, according to which the process of meaning orientates itself, a map, moreover, which can only be linguistically drawn (narrated and/or read) in a sequential and therefore diachronic manner, emphatically not in a formal synchronic manner. In his “Elements of a Narrative Grammar,” Greimas calls this process the ‘narrativization of taxonomy,’ about which he writes: We can see that the taxonomic model, because of the stability of the relations that define its structural terms, can be considered a primary nucleus of an elementary morphology. Nevertheless, an examination of the conditions under which meaning is apprehended shows clearly that if signification…appears as an articulation of stable fundamental relations, it can also be represented dynamically, if one considers it as an apprehension or production of meaning by a subject. Taking into account this dynamic aspect, one can establish a network of equivalences between the fundamental constitutive relations of the taxonomic model and the projections of these same relations, or operations, this time having to do with the already established terms of this very same elementary morphology, operations of which the regulating mechanisms would constitute syntax. Thus, contradiction, as a relation, enables us to establish binary schemata at the taxonomic level. As a contradictory operation it will, at the syntactic level, negate one of the terms of the schema and at the same time affirm its contradictory term. Such an operation, when carried out on terms already invested with value, results in the transformation of contents by negating those that are posited and installing in their place newly asserted contents. Greimas 1987:68f. (emphasis in the original)

At the basic taxonomical level, Greimas continues to define signification and articulation as axiomatically specified relations between sememes. But his interest is now directed at the syntactic process of the symbolic surface level, where abstract relations are converted into concrete, semantically

1.4 The Elements of the Action Construct

179

productive operations. These operations, Greimas argues, must be carried out in a specific order by the producers and recipients of a meaning. The narrative conversion of abstract relations into representational operations is a process which, it must be stressed, does not simply reproduce the axiomatically predetermined static opposition between the contrary terms s1 and s2; the function of the representational operations is rather a connective one designed to maximize the meanings they reveal. For this reason, they mainly take place along the model’s diagonals of dynamic contradiction. The contradiction operation is followed by a move along the vertical axis; given that Greimas refers to this move as a presupposition operation, it is clear that he sees it as a deictic operation which takes place inductively, moving from below to above in order to reveal a further piece of semantic information: The contradiction operation that, for example, by negating the term s1 at the same time posits the term –s1, must be followed by a new presupposition operation that gives rise to, and joins to the term –s1, the new term s2. Thus, syntactic operations are not only oriented, but also organized in logical series. Greimas 1987:69

We are now in a position to illustrate how the semiotic square operates as a dynamic model. For the purposes of illustration, colour predicates have been assigned to the sememe variables in figure 1.4.5, and roman numerals indicate the order in which the operations take place.172

Fig. 1.4.5: Order of operations in the semiotic square

172

See also Ohno (1995:330). Following Greimas (1987:69), Ohno illustrates the relations/ operations along the diagonals using double arrows. Greimas believes that the direction in which these schematic relations can be narratively operationalised is arbitrary. I have confined our illustrations to the representation of the variant s1 ⇒ –s1 for simplicity’s sake.

180

Part 1 Concepts of Action

This reformulation of the taxonomic model as a dynamic model is the starting point that provides the basis for the rest of Greimas’s study, in which he traces operational performance, by which we mean the logic which governs how individual narrative operations are linked into chains of complex syntagms, and the conditions under which these syntagms can attain semiotic closure. However, Greimas’s method has a premise which, as we mentioned above, makes the section of his study which investigates these issues incompatible with the principles of our theory of action and an event-based ontology. In Greimas, the operations can only be narratively represented in anthropomorphized form (more specifically, as the portrayal of a conflict between intentionally acting subjects). This parallel with intentional models of action is particularly apparent when Greimas portrays the circulation of semiotic elements (the values which are assigned to the abstract sememe variables s1, s2, –s1, and –s2 in the narrative process) as an ordered sequence of modal predications which are bound to an actant (Greimas 1987:74–83). At this point, therefore, we must either reject Greimas’s model in its entirety or commit ourselves to using it in a more eclectic fashion. I shall opt for the latter alternative; in what follows, neither Greimas’s actantbased semantics nor his deliberations on the exchange of values will be mentioned. In their place, let us return again to the two questions by which our investigation is driven. What logical semantic criteria underlie the integration of component events into the reception construct of an episode? And what characteristic makes it possible for a singular episode to be linked to another one so that, together, they can begin to form a complex action construct? If we cannot answer both these questions in the same breath, we should at least approach them from the same methodological perspective, and that perspective is the representation of event constructs in the heuristic model of the semiotic square. The episode as a Three-dimensional Construct As we have seen, the episode, even in the everyday sense of the word, is more than the linguistic representation of a sequence of events perceived in a particular order; that is, it is more than just a simple happening. The patterns according to which the individual elements of a happening can be combined are of many different types; the possibilities range from a causal order determined by the principle of cause and effect which seems to be anchored in the distinct states of affairs themselves, to a purely epistemological combination which is completely external to the perceived elements of a happening and is determined by the chronology of perception and cognition.

1.4 The Elements of the Action Construct

181

The former hypothesis, syntagmatic order, is preserved in the episode construct, which also adds a hypothesis of systematic order above it. In the traditional context of normative criticism, as opposed to our constructive hermeneutic context, the episode’s principle of two levels of ordering corresponds to the conventional requirement that a narrated happening must have a certain degree of closure in its aesthetics or the logic of its action. In general, a happening can be considered closed when a specific pragmatic objective has been reached or realized at the internal level of fiction (i.e. when the protensional or projective tension identified by Husserl in the process of action has been resolved). This concept of closure treats coherence criteria as anchored inside a happening and their fulfilment as, in the widest sense of the word, an entelechial process in which the happening finds its way to itself, so to speak. We find the concept in use as early as Aristotle, although Aristotle, and Plato before him, emphatically stressed that it involves an entelechial effect which is aesthetically produced by mimesis and owes its existence to a superordinate calculation of rhetoric or the aesthetics of effect. In reality, however, the coherence of a narrative happening only appears to be an effect produced by the successful conclusion of the step by step progress of a fictional event towards its teleogically predetermined outcome. The logical situation is actually the other way round: only with the postulate of coherence are the conditions established under which the content of actual fictional events can come to an end at all. In and of themselves, individual events contain enough that is to give them a definite and necessary end, purpose, or objective. That is not to say, however, that we should go overboard in this respect; it goes without question that the text is an aesthetic artefact which will always be the starting point and plane of reference which we cannot circumvent. episodes and action may well be predominantly synthetic interpretations of happenings constructed by the reader, but, as interpretations of happenings, they are necessarily based on the textual medium of symbolic representation. There is no way round this essentialism; the action-orientated interpretation of narrative texts is not a creation ex nihilo but an act of reconstruction, no matter how much this reconstruction deviates—as it always will—from the hypothetically and pragmatically postulated initial construct, whether the latter be conceived unconsciously, intentionally by the author, intertextually, or whatever. With these qualifications, the semiotic square turns out to be a perfectly practicable heuristic model of how we read and uncover the closure of a sequence of events, where by ‘uncover’ we do not mean making an unexpected discovery, but rather identifying, to a greater or lesser degree unconsciously, something that has always been there, often as one op-

182

Part 1 Concepts of Action

tion among many. What we identify is expressed at the level of fictional happenings but really external to it. This is true, and particularly so, in the case of the episode, the smallest possible closed sequence of events. Moreover, when we treat an episode presented to us in the mode of a fictional report as closed, we do not mean to give it a meaning located at the level of its own fictional reference to things and figures, for the episode itself, even in each of its individual constituent events, is an abstraction of the fictional states of affairs. Identifying a closure which misleadingly seems to be organically contained in fictional happenings is really an ingenious attempt to elevate events metaphysically, that is, to project onto the fictional world an interpretive metaconstruct whose complexity and level of abstraction come very close to the fundamental essence of meaning itself. Instead of pursuing this idea further, I shall now attempt to represent visually—if that is not too paradoxical, considering the abstractness of the ideas involved—the transition from the level of the epistemic construct of the event to that of the even more abstract metaconstruct of the episode. As a first step, the definition of the event which we developed earlier in this chapter is reproduced as a tree diagram in figure 1.4.6.

Fig. 1.4.6: Tree diagram for the event construct

Figure 1.4.6 illustrates the structure which applies to the object event and discourse event classes. As we noted above, the two can be differentiated internally. In the former, we can distinguish between class-homogenous

1.4 The Elements of the Action Construct

183

and class-heterogenous object events; in the latter, between antithetic and thetic discourse events. For the sake of clarity, I shall confine the following illustrations to the first of these variants: the class-homogenous object event.173 For our examples, we shall return to the beginning of Goethe’s Fairy Tale, which I have condensed into four basic statements of states of affairs. Doing so means that I have interpreted rather than described the text. My division of our example, therefore, cannot and is not intended to make any claim to absolute validity—neither in its delineation of the sections of the text nor in the descriptive terms which it attributes to the statements of states of affairs. Statement of State of Affairs 1.1: ferryman (social predicate, isolated) By the great river, which was newly swollen with heavy rain and overflowing, the old ferryman, weary from the toil of the day, lay in his little hut and slept.

Statement of State of Affairs 1.2: ferryman (social predicate, requested) In the middle of the night loud voices wakened him; it seemed that travelers wanted to be ferried across. Stepping outside he saw hovering over his moored boat two large will-o’-the-wisps, who insisted they were in a great hurry and wished they were already across.

Statement of State of Affairs 2.1: ferryman (material predicate, providing [service]) The old man pushed off without delay and rowed across the river with his usual skill…and soon reached the other side.

Statement of State of Affairs 2.2: ferryman (social predicate, receiving [payment]) ‘This is for your trouble!’ cried the travelers, and as they shook themselves, glittering gold pieces tumbled into the damp boat. 173

The class-homogenous object event corresponds to what is intuitively the most basic and realistic way of reading a narrated happening: as a change in the properties of a concrete fictional object. From a theoretical point of view, however, it actually presents us with the event type that has the greatest requirements, not the simplest one. The construct of a class-homogenous object event must fulfil a total of three criteria. Its focus, which covers multiple states of affairs, must be created by an object which is evaluated as identical; the individuating dispositional and expositional predicates must be counted as belonging to the same generic class; and the content of these predicates must be evaluated as non-identical. Thus, this construct must fulfil the positive identity criterion in two ways: its constituent propositions must agree in both their arguments (identity of the objects) and their predicates (identity of the predicate classes). Only the thetic discourse event is positively defined in a similar manner (identity of the predicate classes and predicates). In it, however, only the predicate is involved; the argument is not considered.

184

Part 1 Concepts of Action

From statements of states of affairs 1.1 and 1.2 above, we can build the categorially homogenous event construct in figure 1.4.7.

Fig. 1.4.7: event construct 1

The event in figure 1.4.7 affects the ferryman, whose situation changes from one of social isolation and non-communication (he is sleeping in his hut) to one of being requested by the travellers (he steps out in front of his hut and hears the will-o’-the-wisps asking to be taken across the river). Similarly, statements of states of affairs 2.1 and 2.2 produce a second categorially homogenous event construct shown in figure 1.4.8. This second event construct uses the predicate class material instead of social. In this second sequence of states of affairs, the ferryman undergoes a material change; his situation changes from being a person who provides a service to being an agent who receives something in return. At this stage, it is of little importance for our purposes that the ferryman is not satisfied with his reward and wants, as we know from the text, to be paid with vegetables (three cabbages, three artichokes, and three large onions, we recall). Mapping these events onto the semiotic squares in figures 1.4.9 and 1.4.10 does not in itself present any considerable difficulties. Starting with the terms introduced in the above constructive synthesis (highlighted in bold in figures 1.4.9 and 1.4.10), it is easy to fill in the two missing terms (dotted lines) in a further act of interpretation. In object events, the

1.4 The Elements of the Action Construct

185

Fig. 1.4.8: event construct 2

focus term by definition denotes identical matter in the constituent statements of states of affairs, so we do not need to consider the ferryman in our notation any more. The conventions of our diagrams deviate from Greimas’s model in one respect: in our illustrations we place the term of the first state of affairs in the bottom left rather than the top left. As we have abandoned the hierarchy of logical levels anyway, this change may initially seem to be a somewhat unnecessary innovation; the reasons for it will be explained below.

Fig. 1.4.9: Semiotic square for event construct 1

186

Part 1 Concepts of Action

Fig. 1.4.10: Semiotic square for event construct 2

The heavy diagonal arrow from the bottom left to the top right of our illustrations represents the transformation of state of affairs which is discovered during the interpretive process. Unlike bidirectional logical and semiotic relations, this transformational relation is unidirectional (despite the fact that it is reversible from a semiotic point of view). Thus, the squares in figures 1.4.9 and 1.4.10 provide a visual representation of three logical relations and one dynamic relation, which is manifested in surface structure as a transformation.

It will not have escaped the reader’s attention that the visual representation we have chosen is ideally suited to facilitate the integration of events 1 and 2 in an episode in which all four positions of a single semiotic square are filled. This is true—and yet not true. For, even in the special case of seamlessly interlocking narratorially postulated states of affairs, is it not possible for such an integration to be realized unproblematically. Even the exhaustively debated example of Forster’s famous (or perhaps infamous) minimal story The king died and then the queen died of grief is enough to show that the final state of the first transformation must be translated into a new initial state if the second transformation is to be connected to the first. To repeat what was already demonstrated in chapter 1: this translation is performed by the modest explanatory addition of grief, which allows us to read the death of the king not just as a fictional cause but, above all, as the semantic equivalent, validated by our knowledge of the

1.4 The Elements of the Action Construct

187

real world, of a grief which ends up sending the queen to her grave after her husband.174 What Forster’s artificial example also illustrates is that there is a type of minimal story in which the two constituent events share (at least) one semiotic term. This term functions as the common factor in the two semiotic squares, which can, although they do not merge with one another, therefore be placed congruently over one another. Systematizing this, Forster’s story must be represented with three squares, given that it contains three transformations and two changing event objects. Our next illustration, figure 1.4.11 contains a corresponding three-dimensional model of the sequence of events. To simplify the diagram, I have abbreviated king and queen to K and Q, and the predicate classes biological and emotional to bio and emo respectively. The two axes on which the points of connection are found indicate the equivalences posited during interpretation.

Fig. 1.4.11: event constructs 1–3 for Forster’s minimal story

174

To the extent that our knowledge of genre-specific norms also belongs to our world knowledge, it is of little consequence that this translation hypothesis has only limited applicability in the fairytale. The knowledge itself is, even when it refers to fictional worlds and the conventions of literary genres, a real one in every way.

188

Part 1 Concepts of Action

As can be seen, we have had to complete the minimal story with a number of propositions which are elliptically omitted in the text itself: the king and queen must have been alive before they can die, and the queen must first have been assigned the attribute not-sad before a transformation to sad can take place. The unused positions of the various semiotic squares are filled with non-instantiated variables (U, V, W, X, Y, Z). The model has been influenced by our knowledge of the causal logic of the fairytale world: we know that it is possible to die of grief. The introduction of this world knowledge makes it possible to connect and interpret the complex sequence of states of affairs as a superordinate sequence of events. Furthermore, by exploiting this possibility, we enrich our logical and semiotic world knowledge with the entire series of equivalence relations which must be posited, even if implicitly, in our complex event construct in order for it to be completed as a logically closed frame of reference (represented by the dotted lines in the diagram). This construct begins with the explicitly posited equivalence between alive and not-sad and, in the above model, ends only with the opposition relation between the abstract variables Y and Z. However, the sequence of states of affairs portrayed in the diagram in figure 1.4.11 does not yet meet the requirements of the coherence criterion; this can be seen from the fact that it is not yet possible to form a single complete semiotic square which combines the event constructs that have been assembled on three different levels. The geometric manipulation that will allow the events to be combined is clear: the third event construct needs to be rotated 180 degrees around its vertical axis. The rotation represents a further, crucial interpretive act: it means positing a chronological order at the middle level of the sequence of constructive events. This chronology is not the elementary chronology of the fictionally empirical sequence of states of affairs, nor that of the real empirical interpretive process which necessarily takes place in real time (as is expressed by the spatial layering of the three event constructs in our model). Rather, this chronology involves the introduction of a fictional temporality which represents a necessary intermediate stage on the way to the highest ranking metaconstruct, the syuzhet. It is just as necessary to refer to this middle temporality of the event sequence if we want to back-translate a syuzhet into a logically anterior fabula without thereby reducing it to the fundamental chronological order of perception itself.175 175

As I have made clear earlier, this is not meant to suggest the presence of an ontically autonomous fictional temporal space. Constructing a fictional chronological order amounts to reconstructing, at the time of reception, a causal prehistory of narration and reception which extends without interruption to the time at which the reconstruction itself is performed. From the perspective of cognitive theory, therefore, the reader’s efforts to chronologically structure the fictional world have the indirect effect of establishing a firm temporal foundation for the here and now of narration and reception/interpretation.

1.4 The Elements of the Action Construct

189

To illustrate the rotation, I shall leave Forster’s artificial example text and return to our concrete example from Goethe’s Fairy Tale. Figure 1.4.12 shows a model of the event constructs we assigned to our earlier extract. It is based on figures 1.4.9 and 1.4.10 above, but it contains a rotated second square, which is therefore structurally indexed as temporally posterior. As the ferryman is the matter predicated in both events, we have not represented him in the diagram.

Fig. 1.4.12: event constructs 1 and 2, Goethe’s Fairy Tale

Our two event constructs can now be integrated into two episodic squares. The first is built out of the terms of the event constructs which are actually present and is therefore referred to as an explicit episode square.

Fig. 1.4.13: Explicit episode for event constructs 1 and 2

190

Part 1 Concepts of Action

In figure 1.4.14, on the other hand, we are dealing with the metaconstruct of an episode generated from the complementary logical terms that are implied by, but not propositionally represented in, the basic events. These complementary terms produce a virtual episode.

Fig. 1.4.14: Virtual episode for event constructs 1 and 2

It is this virtual construct that is of paramount importance for the construction of further explicit event and episode constructs. It enriches the states of affairs which form the first two events with connotations that give them incomparably greater flexibility and connective potential than they have during the construction of the first-generation (i.e. explicit and virtual) event constructs. This is the real reason why reading episodes is a dynamic process. Reading is a constructive task which combinatorially tests a given set of atoms (states of affairs) and molecules (events) for possible connections which are then stored in the latter by means of indices in a prespecified base of world or textual knowledge. And reading brings with it the continual and exponential growth of this knowledge base itself. A virtual construct like that in figure 1.4.14 can be cognitively assembled at any time. Such a construct can then become a subtext located beneath the explicit construct that precedes it, as illustrated in figure 1.4.15. This provides us with multiple arrangements of ordered sets of points of connection which we can use to assemble virtual events and episodes and integrate them, together with coherent explicit episodes, in the complex overall construct which will ultimately become an action. The virtual episode can, as its middle position in figure 1.4.15 illustrates, be used as a heuristic tool which allows us to work backwards and discover new possibilities for connecting terms. That is, it allows us to discover what was previously no more than a virtual event. In our present example, we can see that the state of affairs of social isolation can,

1.4 The Elements of the Action Construct

191

Fig. 1.4.15: Subtext of virtual episodes and events

in theory, be equivalent to that of material requesting. Hypothetically at least, this means that the transformation from social isolation to material providing must also be a conceivable event construct. On the grounds of clarity, we shall not consider further implications of this kind here (hence the use of the variables Y and Z). Instead, we shall put our hypothesis to the test straight away: is there any occasion in the Fairy Tale where this theoretical transformation takes place in our ferryman? Indeed there is, not much later on in the text, when the ferryman has accepted the willo’-the-wisps’ promise to give him the outstanding payment and sets off from the bank again. Statement of State of Affairs 3.1: ferryman (social predicate, isolated) But he heard them not, and continued his course. When he had reached a point lower down, on the same side of the river, he came to some rocks which the water was unable to reach, and proceeded to bury the dangerous gold.

192

Part 1 Concepts of Action

Statement of State of Affairs 3.2: ferryman (material predicate, giving) Observing a deep cleft which opened between two rocks, he threw the gold into it…

Statement of State of Affairs 4.1: ferryman (social predicate, isolated) …and returned to his dwelling.

Figure 1.4.16, described as a redundant event construct, is intended to reflect the perplexing situation in which we become lost in a circle of action logic. However, despite its logical flaw, figure 1.4.16 still represents a textually adequate interpretation in so far as it confronts us with the first of many instances where Goethe’s narrator demonstrates his enjoyment in caricaturing stereotypical plots and malevolently setting up loose ends and misleading conclusions—something happens, but the story does not go on, at least not yet for our ferryman.176

Fig. 1.4.16: Redundant event construct

176

Here, Goethe caricatures the repetition of motifs which is typical of the fairytale. See Dammann (1990) on the concomitant linking principle, hardly supported by the logic of the action, of persistent figures in the Fairy Tale. Dammann interprets it as an allusion to the poetics of courtly Baroque novels.

1.4 The Elements of the Action Construct

193

Let us review the key differences between the model we have outlined and its Greimassian antecedent, beginning with the logical relations between the individual terms (states of affairs) in event and episode constructs. Here, as mentioned above, our episodic square is more flexible than the semiotic square, for it does not assert a categorial hierarchy between an upper relation of contrariety and a logically subordinate relation of subcontrariety.177 Instead, our model postulates a reversible relation of mutual implication between the two levels. Put somewhat loosely, four states of affairs can form an episode if they can be combined in pairs into two event constructs which can then be integrated into a complete square of mutual opposites, contradictions, and implications. In contrast to the Greimassian square, event-forming transformations are only to be found on the diagonals of our model and thus always treated as bound to narrative contradictions. ‘(Contra)diction’ is meant in an absolutely literal sense here: only the propositions and transformations expressed in narrative sentences (i.e. natural language) can be provide a basis from which event constructs can be constructed.178 event and episode constructs do not directly involve the axiomatic contrariety represented by the upper and lower horizontal double-headed arrows or the implications symbolized by the vertical lines in the original Greimassian model (see Fig.1.4.4). Instead the relations represented and symbolized by these arrows are purely abstract and exist logically behind and chronologically before any narrating speech about the fictional world. Thus, their existence is not connected to actual event and episode constructs produced by readers; this logical independence is what makes it is possible to narrate more than just a single story for every axiom. This does not affect the silent completion of narrative reports that takes place by filling in elliptically omitted states of affairs and is regularly performed so that complete event constructs (at the very least) can be constructed. The second aspect of our model which can profitably be compared with that of Greimas is the way in which chronological relations are represented. Our three-dimensional is clearly more effective in this respect, for it repre177

178

It is obvious that Greimas has transferred this distinction directly from the logical square to the semiotic square. He increasingly tends to ignore it as his interest becomes directed at the question of the mediation between deep structure and narrative surface structure. In my view, Greimas has created an unnecessary systematic burden for himself here. See the criticism in Kalinowski (1981) and the suggestion of Douglas (1997) that the counterproductive link back to the logical model of Aristotle and Apuleius should be abandoned. Here I follow Stempel (1973) and Schmid (1992)—without, however, adopting their ontologizing reference to the required reality of reported happenings. Everything—even the dreams of a fictional figure—that appears in a text is real in so far as it is linguistically posited by that text.

194

Part 1 Concepts of Action

sents not only the logical relations between narrative assertions of states of affairs and their implications, but also the double chronological order which underlies any sequence of narrated or received assertions of states of affairs. Depending on whether our interest is directed primarily at the syuzhet or the fabula, the temporal axis runs along the three-dimensional plane of depth or the two-dimensional plane of height respectively.

Fig. 1.4.17: Temporal axes of the episode construct

The three-dimensional model in which the individual squares are arranged on top of one another represents the temporal depth created by interpretive reading, which is a succession of discrete empirical acts of perception, construction, and interpretation. In terms of production theory, the resultant temporal depth is that of the syuzhet; in terms of reception theory of reception, it is that of interpretive constructs. As we are most interested in the latter aspect here, we refer to this depth, the third dimension, as construct time. On the other hand, the fabula, which (at least theoretically) manifests itself in an uninterrupted succession of individual states of affairs and events, has a two-dimensional planar time. As fictional object time (i.e. time that passes in the textually represented world), it is modelled on the Y axis, the axis of states of affairs. The actual textual statements which represent these states of affairs, however, are arrayed in horizontal succession along the X axis of reading time. In this respect, our model assumes that the sequential reading process cognitively processes literary representations in a manner equivalent to that of the sensual perception of objects in the real world (i.e. in a way which does not necessarily reflect the objective passage of time).

1.4 The Elements of the Action Construct

195

The treatment of the category of time in the three-dimensional model proposed here is more economic and consistent than most, if not all, previously developed narrative grammars. We have previously criticized the ontologizing language which, from Müller to Genette, distinguishes narrative time and narrated time as if we were concerned with two distinct kinds of time. One of the advantages of our model, is that it provides a visual demonstration of the extent to which the two times ultimately fit together as a continuum. This, in fact, is why we made the key decision to part company with Greimas and represent the connections between states of affairs using rising diagonals inside the square. If we are to grasp the two temporal axes, object time and construct time, in one glance, then we must necessarily see them as running in parallel and from below to above. As far as the z-axis of construct time is concerned, this is an optical illusion, but it simply cannot be avoided in observation. In this way, our model establishes an appropriate analogy for the cognitive diffusion of the empirical and fictional temporal continua which takes place in our minds when we attempt to read more than just one singular event in a sequence of states of affairs. It is precisely this that makes the constructive formation of episodes so amazingly enjoyable and rewarding; to use a visual metaphor again, it is a journey into the vertical where we rise above the sequentiality of happenings and see them in a bird’s eye view which turns planar sequences of states of affairs into actions that, somewhere in their three-dimensional depth, have the potential to connect themselves to our real activity as recipients and create meaning for us. This places the episode right on the border between two- and three-dimensionality in the system, between the two forms in which the fictional happenings can be seen: as a hypothetically postulated sequence of abstract states of affairs combined at most in pairs on the one hand, and as a meaningfully ordered structure which acquires the quality of a mental representation on the other. When we read an episode, we establish a logical connection from the final state of affairs back to the initial one. Only by adopting this three-dimensional perspective can we come to terms with the fact that, by making the logical connection, we have forced a numerical sequence into a semantic circle. In our model, this is expressed by the back-to-front arrangement of the second event square. Here, without a doubt, we have reached the limits of our ability to visualize a process which the producers and recipients of action narratives perform without the slightest difficulty in practice. We shall draw our discussion to an end by considering once again, in figure 1.4.18, the construct derived from opening paragraph of Goethe’s Fairy Tale in which the four constituent statements of states of affairs are integrated into an episode.

196

Part 1 Concepts of Action

Fig. 1.4.18: Congruence of object and construct time in an episode

Statement of State of Affairs 1 By the great river, which was newly swollen with heavy rain and overflowing, the old ferryman, weary from the toil of the day, lay in his little hut and slept.

Statement of State of Affairs 2 In the middle of the night loud voices wakened him; it seemed that travelers wanted to be ferried across. Stepping outside he saw hovering over his moored boat two large will-o’-the-wisps, who insisted they were in a great hurry and wished they were already across.

Statement of State of Affairs 3 The old man pushed off without delay and rowed across the river with his usual skill…and soon reached the other side.

Statement of State of Affairs 4 ‘This is for your trouble!’ cried the travelers, and as they shook themselves, glittering gold pieces tumbled into the damp boat.

Our integrated model is intended to represent the complex, logically recursive process of making events out of states of affairs and episodes out of events. Like all analogies, it has the potential to be misunderstood, to be read as saying something which it is not. It should, therefore, be

1.4 The Elements of the Action Construct

197

explicitly said here that our intentions do not extend to revealing the grand discovery of a universal law in the world of symbolic phenomena in general or the action narratives that we are studying in particular. The three-dimensional model which we have described is, at this stage, nothing more than a heuristic tool, a theory which guides the search for structures it has defined in advance. To draw this chapter to an end, we can summarize that, to carry out this search, our model defines the episode as a semiotic square. All the terms in the square are instantiated and they are connected by events along both diagonals. The semiotic square is a metaconstruct of a purely conceptual nature, and we can only see it when we have freed ourselves from the purely numerical sequence of states of affairs and made semantically comparable event squares logically congruent by superimposing them on one another. It is of little theoretical significance how many event levels are stacked together; in practice, however, the gaze of any interpreter can reach only as far as a certain depth that is determined by cultural and subjective factors. In particular, the connectivity and comparability of perceived phenomena, the essential features without which we could not semantically correlate them in our minds, are not contained substantially in the phenomena themselves. Comparability is posited or learnt and can therefore only be tested by making reference to a world knowledge which is both different in every single one of us and ready to be expanded at any time. The proof of the pudding is in the eating. That is just as true of heuristic methods as it is of anything else. The reader deserves our congratulations for persevering and digesting a somewhat hefty helping of terminology. In part 2, we shall move on to what will hopefully be a more appetizing prospect: the application in literary computing of our newly developed concepts of action logic.

Part 2 The Computer-Aided Analysis of Narrated Action The Professor…said, perhaps I might wonder to see him employed in a Project for improving speculative Knowledge by practical mechanical Operations….Every one knew how laborious the usual Method is of attaining to Arts and Sciences; whereas by his Contrivance, the most ignorant Person…may write Books in Philosophy, Poetry, Politicks, Law, Mathematicks and Theology, without the least Assistance from Genius or Study. He then led me to the Frame, about the Sides whereof all his Pupils stood in Ranks. It was Twenty Foot square [and] composed of several Bits of Wood, about the Bigness of a Dye.…They were all linked together by slender Wires. These Bits of Wood were covered on every Square with Paper pasted on them; and on these Papers were written all the Words of their Language in their several Moods, Tenses, and Declensions, but without any Order…. The Pupils at his Command took each of them hold of an Iron Handle, whereof there were Forty fixed round the Edges of the Frame; and giving them a sudden Turn, the whole Disposition of Words was entirely changed. He then commanded Six and Thirty of the Lads to read the several Lines softly as they appeared upon the Frame; and where they found three or four Words together that might make Part of a Sentence, they dictated to the four remaining boys….Six hours a-Day the young Students were employed in this Labour; and the Professor shewed me several Volumes in large Folio already collected, of broken Sentences, which he intended to piece together; and out of those rich Materials to give the World a complete Body of all Arts and Sciences. Gulliver’s Travels (Swift 1995:173-175)

2.1 Narratology and Literary Computing The traditional student of literature would be hard put to find a more disparate pair of disciplines than literary criticism and information technology. The former manipulates symbols hermeneutically, the latter mechanically, and they would seem to rest on irreconcilably different methodological and philosophical paradigms. Any attempt to combine them, we might imagine, would be ridiculous enough to rival any of Swift’s satires, not least the absurd ‘project for improving speculative knowledge by practical mechanical operations’ which Gulliver encounters at the Lagado Academy. If the reader has found himself nodding in agreement with the above paragraph, he should pause to consider that obvious ‘facts’ such as this can be the most misleading of all our assumptions. The late 1960s, for example, witnessed the collapse of classical structuralism (the origin of classical narratology) when a fundamental misconception at its very core was exposed.1 S/Z, Roland Barthes’s famous work of 1970, takes the dogmatically scientific models which portray themselves as objective sciences of text and meaning—exemplified by the generative grammars of linguistics and narratology—and places them before a mirror which reflects the true nature of their vainly repressed subjectivity. It is this return of subjectivity to the reading process that strikes us as the groundbreaking achievement of S/Z when we look back on it today. However, Barthes was really using the book to make a far more profound methodological and programmatic statement. At the deeper level, S/Z contains a concrete, often self-reflexive interpretive process which represents a uniquely rigorous attempt to explain the multiple meanings of narrative texts as the effect of a dynamic process. In this complex process, text, reader, and interpretation interact simultaneously in various codes and join together in a single interpretation under the influence of complex semiotic and cognitive chains which cause the number of potential meanings to increase in exponential stages. S/Z is a classic demonstration that, however hard we try, we will never be able to uncover the objective meaning of a textual code by analysing it systematically with the theoretical and conceptual tools of structural semiotics. When stripped of its systematic exterior, the fact of the matter is that this 1

The distinction between classical and postclassical narratology was introduced by David Herman (1997:1049f.) and is derived from the distinction between classical structuralism and post-structuralism.

202

Part 2 The Computer-Aided Analysis of Narrated Action

kind of analysis follows a series of elaborate, highly idiosyncratic steps in pursuit of what is, and always will be, an elusive goal. It cannot be denied that Barthes was right to condemn the rigid systems of early structuralism for claiming absolute scientific validity. However, S/Z fails to deal convincingly with the resultant question: if there is no such thing as absolute validity, how can we judge the relative value of different, but methodologically consistent, interpretations of a text? This practical problem first began to emerge in the mid-1960s; ever since, the theory of narratology and semiotics has advanced more rapidly than critical practice, which has not yet arrived at a methodologically consistent means of analysing and describing concrete texts. In the heyday of the structuralist movement, formalist studies in the style of Propp attracted considerable interest, and Gerald Prince, for example, took the development of a rigorously designed and strictly formal narrative grammar forward into the 1980s. But despite such work, there is a dearth of studies which inductively analyse the fabula of an actual fictional narrative text, let alone the fabulas of an entire corpus; in fact, we can count such studies on the fingers of one hand.2 To an increasing degree, Bremond’s Logique du récit, Todorov’s Grammaire du Décameron, Barthes’s S/Z, and Genette’s Figures all complement their theoretical and taxonomical arguments by discussing examples which demonstrate specific features and thus illustrate the case in point at any given time instead of providing a systematic proof of the writer’s overall grammar or system. Only in Pavel’s Poetics of Plot does Propp’s formalist tradition of fabula analysis survive in any recognizable form. In general, therefore, most narrative theories are developed and tested on the basis of a single illustrative example which later provides an ideal breeding ground for sweeping generalizations, dogmatic theories, and vit2

This does not include analyses of artificially constructed example texts or sentences which have been formulated simply in order to illustrate a particular grammar of narrative or action (e.g. Prince (1973, 1982)). As we would expect, almost all studies in structuralist narratology and literary criticism, whether introductory or more involved, discuss disconnected illustrative examples; representative works are Chatman (1969), Titzmann (1977), Rimmon-Kenan (1983), and Bal (1988). Martinez and Scheffel (1999) is the most recent introduction in German. From the linguist’s point of view, Toolan (1988) is particularly noteworthy for its thorough methodological overview of the pre1988 literature. There are few examples of formal analysis of extended textual passages, at least at the syuzhet level; they include Todorov’s Grammaire du Décaméron (1969) and, of course, Pavel (1976, 1985). The reader may also wish to consult Renner (1981, 1983) and Hallstein (1997). As noted above, Renner (1983) is an unusually rigorous investigation of the details of Lotman’s semantic definition of the aesthetic event. Only marginally relevant to the present study, Hallstein (1997) analyses the structure of time in E. T. A. Hoffmann. Narratological and semiotic studies which stand in the Greimassian tradition would appear to show the greatest commitment to stringent formalization—see, for example, Perron (1996). Lee (1992) is little more than a curiosity.

2.1 Narratology and Literary Computing

203

riolic polemics. Indeed, the remarkably persistent popularity of Genette’s theories is due not least to the fact that he keeps his concepts and categories general enough to be suitable for further use, despite demonstrating them with just a single text. This was, and still is, most agreeable to practitioners of traditional literary criticism, for, as Graves and Frederiksen (1991) have shown, expert readers do not stick rigidly to a step-by-step receptive process but receive the text as synthetically as possible: after digesting individual elements of the text, they integrate them into complex hypothetical units of meaning more readily than normal readers.3 Thus, crawling along a trail of linguistic minutiae is as objectionable to the literary critic as leaping at hermeneutic explanations is to the linguist, for whom such a move is both premature and lacking in sound methodological justification. It must not, of course, be forgotten that the narrative and its component levels, the fabula and sjuzhet, are highly complex objects. In contemporary narrative theory at least, the detailed theoretical analysis of their fundamental structure grinds to a halt not for lack of commitment or suitable evidence but rather because of the limitations of the available resources. Nonetheless, we cannot avoid the conclusion that narrative theory is justly stigmatized for failing to exploit what could provide the most convincing proof of its legitimacy and relevance: the abundance of empirical objects (i.e. texts and corpora) against which the various theories could be tested. The situation is considerably different in folklore studies, an important methodological precursor to narrative theory in terms of the formal analysis of narrative representations. Here, we are concerned primarily with texts, such as fairytales and myths, which are preserved in oral transmission; such texts possess a clear cultural function but do not have an identifiable individual author or narrator whose message and authority can be investigated. Because of this, it is practically impossible for folklore research to restrict its methodological perspective to the aesthetics of production, whereas critics of literary narratives, which basically involve individually authored texts, do so routinely. Whether critics have dispensed with the authorial instance altogether (as in the school of French structuralism) or concentrated on the role of authorial intention (as is typical of German narrative theory represented by Lämmert, Stanzel, and others), the end result is the same in each case: a theory of narration couched in terms of classical poetics has now been joined, at least in the most promising instances, by a systematically descriptive theory of what is narrated.

3

See also Löffler (1988), who illustrates how recipients gradually move from reconstructing a story to evaluating it, and Harker (1989), who describes the process of text reception from the perspective of information theory.

204

Part 2 The Computer-Aided Analysis of Narrated Action

Yet even this two-pronged methodological approach does not fully capture an important idea suggested by the grandfather of literary criticism, Aristotle himself, the idea that there is a systematic connection between poetics and rhetoric, in other words, between the aesthetics of production and the aesthetics of effect. Such a synthetic combined perspective was in fact practically ignored in most narrative theories before Genette. The theory and aesthetics of reception and the historical trends in the reception of a text are not conventionally seen as narratological topics, and, until only a few years ago, it was taken for granted that the cognitive processing of narratives is the concern of cognitive theory rather than narratology.4 The situation has of course changed significantly with the recent shift towards constructivism in narrative theory, which is perhaps best demonstrated by Herman (2000, 2001,2002), who suggests that we should reconceptualize narratology as a cognitive science.5 Similarly, Fludernik’s Towards a Natural Narratology is generally seen as one of the most determined attempts to turn narratology on its methodological head in a pragmatist way. The central category in her model is experientiality, by means of which ‘reading patterns for typically human predicaments’ are explored and tested (Fludernik 1996:25). However, Fludernik’s conception of the text as a medium, based partly on cognitive psychology and partly on communication theory, exposes us to the danger of identificational reading, by which we mean reading that partially reconstructs the psychological and behavioural patterns of fictive figures for didactic reasons as opposed to synthetically assembling an experimental semantic construct for aesthetic and pragmatic reasons. It is therefore not surprising that Fludernik openly disputes the primacy of action (plot) by stating that in her model ‘there can be narratives without plot, but there cannot be any narratives without a human (anthropomorphic) experiencer of some sort at some narrative level’ (Fludernik 1996:13). More relevant to the present study is an approach sometimes termed applied narratology. The origins of applied narratology lie in the collaboration between researchers in formalist narratology and artificial intelligence which began in the late 1970s and is particularly widespread in North America. As early as 1979, Patricia Galloway put forward the theoreti4

5

The models of cognitive theory, on the other hand, have long been prepared to adapt story grammars from narratology; they have even developed their own (Mandler and Johnson 1978; Rumelhart 1980; Shen 1988). Early examples of designing computeraided models can be found in Galloway (1979, 1983). Lehnert (1981) is one of the most methodologically interesting approaches in this area. A ‘cognitive shift’ had previously been advocated by Nünning (1990). Its influence can be observed throughout the extensive body of constructivist-style literature that has emerged over the past ten years or so. See also Jahn (1997).

2.1 Narratology and Literary Computing

205

cal outline of a strictly formalist project in which she proposed using a computer to test a grammar of plot. The grammar describes the structure of fictional narrative texts as a sequence of transaction units. Transaction units are ‘exchanges—of material goods, attitudes, and items of information—between the characters in the course of the story’ (Galloway 1979: 53). In Galloway’s project, concrete texts are described and marked up in a strictly sequential manner. First, individual words of the text are classified according to eight grammatical categories. Second, the reader examines the sequences of words for closed transaction units and defines their extension. In the third and final step, the distribution of the basic units that have been identified is analysed statistically in order to detect patterns in their location, recurrence, and combination with other types of unit. Related methods can be found in the approach of Wendy Lehnert, who also draws on Rumelhart’s work. Lehnert’s analysis of plot is based on the affect states of acting persons; these states are divided into positive events (events that please), negative events (events that displease), and mental states (which are affective reactions to events and functionally realized as the motivations of actions) (Lehnert 1981:295). The development of a story can now be described in its most basic form as a linear sequence of affect states, whereby a separate descriptive sequence must be assembled for each fictive person. In Lehnert’s model, affect states can be combined into triadic primitive plot units by reconstructing the causal links between a group of three elements. Lehnert’s system identifies a total of four causal relations (motivation, actualization, termination, and equivalence). It can be shown that this allows thirty-six theoretically possible paired arrangements. In practice, however, they can be reduced to fifteen variants, thus providing a typology of the primitive plot units which supply the building-blocks for the assembly of complex plot units. Once a story has been described using the inventory of primitive plot units and the often idiosyncratic complex plot units, summarization takes place in the final stage of the analysis. Lehnert argues that if we compare her theoretical model with empirically observed (natural) summaries of a story, we will find that the best summary is the one that contains the greatest number of pivotal plot units (a pivotal plot unit is a plot unit which allows conclusions to be made about other, causally subordinate elements; the best summary is the summary which most adequately reflects the causal logic of the plot structure). Thus, the functionality of a given summary structure is determined primarily by how well it reflects the level of narrative cohesion in the story in question. Lehnert understands cohesion not as a prescriptive criterion which refers to concrete world knowledge but as a preference which models our expectations about the well-orderedness and organization of stories. Her model therefore provides us with a formal description

206

Part 2 The Computer-Aided Analysis of Narrated Action

of cohesion, which is defined as resulting from the formation of families and clusters around particular plot units (Lehnert 1981:316f.). As can be seen, Lehnert’s theory is a highly ambitious one; it concludes with a summary of method which reveals her fundamentally cognitivist agenda: There is infinite variation in the number of plots that are possible, and people can understand a story with a new plot line whether they’ve seen a similar plot before or not. This suggests that plot recognition must be based on bottom-up processing, rather than top-down analysis. Lehnert 1981:329

The greatest advantage of her approach, Lehnert believes, is that she describes a bottom-up process of systematic plot description which could be easily converted into an algorithm suitable for use in a computer program. However, like many other artificial intelligence projects, her study deals with the construction of a purely theoretical model. The practical implementation of this model—for example, in the form of a working descriptive, analytical, or generative software application—lies outside the scope of her work. At first sight, a similar qualification applies to Ryan’s discussion of creating an artificial intelligence simulation of the modal possible worlds model of the narrative text. Ryan’s ideas are developed in individual essays and drawn together under the significant title Possible Worlds, Artificial Intelligence and Narrative Theory (1991). Unlike Lehnert’s study, Ryan’s theory is developed in an explicitly interdisciplinary context which combines the methods of cognitive theory and narratology. As early as her self-contained essay “The Modal Structure of Narrative Universes” (Ryan 1985), which sets the theoretical and methodological scene for her later book, Ryan outlines a nine-step algorithm for processing narrative discourse. She intends the algorithm to be used in two ways: first, as a descriptive model which replicates the natural receptive process that takes place in human readers; second, as the theoretical specification and conceptual blueprint for a computerized artificial intelligence model which simulates the formation of successive complex narrative universes during reception (Ryan 1985:749ff.). The same project is described, albeit in a somewhat different context, in Ryan 1991. Clearly influenced by the rapid development of hyperfiction, Ryan now shows a more pronounced interest in the practical implementation of generative algorithms in computational models, but she remains justifiably sceptical about the likelihood of achieving this goal in the foreseeable future.

2.1 Narratology and Literary Computing

207

Narratology and the Cognitive Sciences: Contrasts and Connections We have now reached the point where we can no longer avoid confronting two fundamental methodological questions. First, what is the connection between the studies in cognitive theory which we have mentioned and our narratological interest in the theory of action? Second, and more generally, how compatible, if at all, are the objectives and research methods of the two disciplines, and to what extent, if any, is it possible for them to influence and inform one another? We shall begin by considering the latter question. Literary criticism and narratology deal with text-based cognitive processes and are therefore concerned with one particular aspect of cognitive science, the processing of symbolically represented information. However, if we consider the apparent similarity more carefully, it is clear that the two disciplines employ fundamentally incompatible concepts of the text and linguistic sign. Cognitive modelling tests the hypotheses of cognitive theory on the basis of sample natural language texts. Because these texts contain expositional discourse, they require the use of a referential concept of the sign and therefore subscribe to a common convention of everyday language use whereby linguistic signs and the texts which they constitute should, and do, refer to empirical objects and perceptions, whether directly or via intermediate concepts. While it does not follow that this is the only concept of the sign recognized by the cognitive sciences, it is nonetheless the one which they adopt by default, and it is characteristic of the majority of texts which they have studied to date (we shall describe one of the few exceptions to this generalization below). A completely different concept of the sign is associated with non-expositional discourse and the texts which are the primary object of study in aesthetic, philological literary criticism and classical narratology. Of course the linguistic sign has a referential function when it is used in the mode of fictional literary speech—but this referential function is intrinsically different from that of the cognitivist sign: the reference of the literary sign is directed at a completely different class of objects (possible rather than empirical worlds) and has a fundamentally different purpose (it has a subjective relevance which is aesthetically or philosophically realised rather than a reference which is empirically and pragmatically functional). So, even if certain schools of literary theory and the cognitive sciences share a common interest in analysing and modelling how humans understand texts, their methods assume completely different object domains and text types which in turn imply completely different concepts of the sign. This is no chance occurrence—it follows from the differing objectives of the two disciplines. The cognitive scientist is primarily interested in how

208

Part 2 The Computer-Aided Analysis of Narrated Action

inferences and associations develop inside a cognitive system; how their development is linked to schematized experiences and remembered knowledge; and finally how these internal processes are coordinated with the processes of empirical perception. The literary critic, on the other hand, takes the primary cognitive activity of perceiving and processing basic meanings for granted; he focuses instead on how and under what conditions a recipient or producer forms hypotheses about the complex overall meaning of an intricately prestructured and semantically over-determined semantic complex (i.e. the literary text). In a word, the interest of literary theory is directed at the interpretive product and how it is formed rather than the niceties of the cognitive processes which precede these high-level interpretive operations. In the face of this substantial difference, one might well wonder whether any of the apparent similarities between the two disciplines stand up to detailed scrutiny. Let us consider the formal description of textual features, which has always been one of the central concerns of narratology. Since Propp, the investigation of formal description has been accompanied by an interest in theories which could employ it in order to model the process of narratio formation. Narratology may therefore have been the first school of literary and textual criticism (as distinct from pure linguistics) to adopt methods and aims which mark it out from the hermeneutic paradigm and give it the theoretical opportunity to draw on the analytic paradigm of empiricism. Moreover, several concepts which originated in narratology (most obviously plot, but also episode) have been adopted by the cognitive sciences; this suggests an affinity between the two disciplines which runs deeper than the differences outlined above. On the other hand, we must remember that studies in cognitive theory can use the term ‘plot’ to mean something completely different from its original, narratological definition; in general, cognitive theory does not consider the narratological origins of its borrowed terminology.6 Yet, despite this caveat, the models of cognitive theory that we have considered (e.g. Lehnert and van Dijk) are linked to narratology by more than superficial terminological resemblances. Parallels can, for example, be detected between the various systems involved, as is the case with Bremond’s action logic and Pavel’s move grammar, which are obviously analogous to van Dijk’s macrostructure model. Besides sharing terminological features, all three models present their terminological systems as transformational environments by employing tree diagrams, which had previously been used by Lévi-Strauss to analyse action structure in a similar fashion. 6

Rumelhart (1975), one of the first scholars to consider Propp’s approach from the perspective of cognitive science, is an exception.

2.1 Narratology and Literary Computing

209

We are left to conclude that only contributions such as those of Lehnert and Ryan can be considered truly interdisciplinary in the sense that they contribute to our knowledge of narratology and cognitive theory alike. True interdisciplinary models of this kind—and the present study does not claim to be among them—show clearly that we have to do more than taxonomically define a descriptive terminology if we are to turn a descriptive narratological model into a constructive model of a cognitive process. At least an equal amount of effort must be put into accurately modelling the separate stages of the process in which the descriptive terms are practically applied. This is the particularly true if our ultimate aim is to convert the descriptive model into a concrete program (i.e. a sequence of systematically executed heuristic and classificatory instructions). Considering Ryan’s approach also helps us to understand the limits of what can be feasibly accomplished using computational methods. The nine steps of her algorithm involve the classification of concrete narrative sentences using descriptive terms—a quick glance is enough to show that this is only possible if the linguistic data to be classified can be semantically interpreted (Ryan 1985:749ff.). It follows that such a classificatory process can be efficiently automated in a computational text parser if and only if a set of unambiguous classificatory rules or a normative referential lexicon is available. Now, it may well be theoretically possible and computationally viable to use these auxiliary requirements during the description of expositional texts, where disambiguation is made easier by clear pragmatic contextual references. However, narratology and literary criticism are concerned primarily with texts which contain a narrative representation of fictional events, where normative classification in this manner is not possible—it would violate the fundamental hermeneutic assumption that fictional literary texts can be produced, and must be received, without relying on unambiguous referentiality. As a key principle of aesthetic production and reception, referential ambiguity is therefore both a requirement and a limitation of practical implementation and must be kept in mind in what follows. The Limits of Practical Implementation We shall now use a computer program to apply the terminological apparatus that we developed in part 1 above, according to which matter and state of affairs are basic concepts, and event, episode, and action are the constructs derived from them. Unlike the above-mentioned artificial intelligence simulations of human cognitive processes, we intend not to design a theoretical specification but to assemble a set of practical tools

210

Part 2 The Computer-Aided Analysis of Narrated Action

by developing a pair of application programs, EventParser and EpiTest. As its name implies, EventParser is a tool for decomposing textually represented complex narratives into their constituent events.7 These events are the indispensable basis of every concept of fictive action generated by a recipient. The program is not a parser in the strict sense of the word—the decomposition of the text is not automated but performed by a natural reader who carries out the necessary interpretive operations. EventParser is essentially a mark-up tool—a technical aid which we can employ to provide digitized texts with standardized mark-up and annotation. Unlike sgml files, the mark-up is not inserted into the electronic text itself but written to a record file which is relationally linked to the object text.8 EpiTest subjects the record files produced by EventParser to combinatorial analysis and uses its algorithms to generate episodes and actions (chains of linked episodes), the two types of construct which are built out of basic events. The operation of both programs will be described in detail shortly, after we have said a few words about the methodological scope of the software and the computational narratology of action which it represents. As we have already made clear, the present study and its practical tools are not intended as a contribution to cognitive theory; they fall firmly within the remit of constructivist narratology. The comparative analysis of virtual and real constructs which range from the state of affairs to the episode will require a bottom-up approach which takes the event as its basis and therefore starts from a level that is normally ignored by literary theorists when they analyse fabulas and actions.9 It goes without saying that a computer-aided approach should remove much of the tedium inherent in the study of such a fundamental level by automating a significant 7

8

9

EventParser is based on the MoveParser program described in Meister (1999). The new name of the application is intended to highlight the fundamental change in the underlying conceptual foundations. It is no longer based on Pavel’s move theory and now uses the event concept we have developed above. There are several reasons for rejecting sgml- or xml-compatible encoding in our approach and implementing a proprietary solution instead. Foremost among them are considerations of a practical rather than a theoretical nature (for example, our solution allows us to: produce compact record files in which metadata is encoded in a form which immediately complies with prolog syntax; differentiate between different types of metadata; and store and integrate different versions of metadata clauses). Even so, it should not be forgotten that our decision has obvious disadvantages, of which platformdependency is the most significant. See also the distinction in cognitive theory between subject-centric, inductive processes of (bottom-up) comprehension and quasi-objective object interpretations which are deduced (top-down) from elements of world knowledge assumed to be valid and relevant. See Christmann and Groeben (1999:145–48), Strohner (2000:261–74), and Scherner (2000: 186–95).

2.1 Narratology and Literary Computing

211

amount of repetitive textual description and accelerating the quantitative and statistical analysis of textual and metatextual data. But despite the obvious reduction in the practical workload of the human narratologist, particularly when studying event and episode constructs, the automation has a relatively trivial methodological effect—the computer does little more than accelerate what humans are able to do anyway and does not bring about a qualitative improvement in our method. It can in fact be argued that we are concerned not with a trivial effect but with a trivialization short and simple, given that an automated analytical method based on an emphatically quantitative system would seem to exclude the use of qualitatively (i.e. semantically) defined categories. As our earlier philological and theoretical discussion has shown, narrated and read actions are complex interpretive constructs: they have a quantifiable material foundation in the signs of the text but also, and more importantly, require the use of a subjective pool of world and rule knowledge. This knowledge, which is external to the text and its empirical signs, cannot be represented in any known formal notation. It might seem that computers can do nothing more for literary criticism than provide a trivial increase in speed. However, this is not the case; the true benefit of computers lies in the fact that they force the critic to adopt an unprecedented level of formal stringency and methodological rigour. In this sense, the computer becomes what McCarty has called the ‘perceptual agent’ of the literary critic: The promise of the new (technology) is thought to be quantitative: the new thing will do the old job faster, more efficiently, and more cheaply; it will increase productivity, and so forth. Tools, however, are perceptual agents. A new tool is not just a bigger lever and more secure fulcrum, rather a new way of conceptualizing the world, e.g. as something that can be levered. McCarty 1991:74

It may come as something of a surprise to learn that the most important feature of our ‘new tool’ is the fact that its analytic and hermeneutic operations are unable to handle any data types which have not been defined in the underlying theory and considered in the design of the program’s algorithms—computers are not intelligent enough to cope with such problems by generating ad hoc hypotheses of their own.10 Any problems which occur during program development are thus particularly helpful because they identify specific gaps in our knowledge and theory. Designing an algorithm is considerably more time-consuming than traditional

10

However, as we shall soon see, there is nothing in theory which prevents such algorithmic programs from having the ability to learn.

212

Part 2 The Computer-Aided Analysis of Narrated Action

self-correcting critical practice, which deals with inconsistent data and theoretical shortcomings as it goes along, but the reward is an unrivalled theoretical robustness and methodological consistency. This qualitative benefit is the primary advantage of using the computer as a perceptual agent to help formulate and test narratological theories. In the longer term, this may translate into an additional quantitative improvement by enabling us to conduct narratological analyses of larger samples such as entire corpora. Indeed, complex analyses and calculations have been made possible by digitally represented literary texts since the 1970s. Some of the studies in question, such as those which investigate authorship, themes, and styles, have clear hermeneutic benefits and thus do more than just generate concordances, indices, and registers. In our case, however, the situation is somewhat more complicated. The retrieval and analysis of tokens in the concrete linguistic material of the text has long been standard practice in computational linguistics; it is now equally possible, at least in the rudimentary form of a search command, for users of commercial electronic editions of published texts. However, it is not so easy to search for semantic units in a similar fashion—such units must first be manually identified and marked in the text. The interpretive annotation of an electronic text in this way is known as tagging, which can only be automated to a limited extent. In theory, to be sure, there is nothing to prevent us from defining a search algorithm that automatically tags every occurrence of, say, the first of Propp’s thirty-one functions. All this requires in theory is an abstract formal representation which anticipates every possible way that literary language can represent the state of affairs ‘One of the members of a family absents himself from home’ (Propp 1984:26). In practice, this means that our formula has to consider such possibilities as ellipsis and metaphorical, metonymical, or even intertextual substitution; and this, for the time being, is more than literary computing is capable of doing. Thus, building an automatic text tagger which automatically marks up or tags meaningful units (including elements of the fabula) has so far proved to be beyond what is reasonably possible in literary computing. However, once an electronic text has been tagged, or an equivalent relational database set up, the analysis of the resultant representation does not present any such insurmountable difficulties—an analysis of distribution, for example, simply involves expressing, refining, or reformulating a particular question in formal terms. As will become clear in the next two chapters of part 2 and the practical application of our model in part 3, the present study aims to go a step further: as well as analysing descriptive formal records, we shall explore their combinatorial potential. We hope to use EventParser to capture concrete interpretive constructs; subsequently, and

2.1 Narratology and Literary Computing

213

more importantly, we hope to generate additional, theoretically possible virtual constructs with the help of the second program EpiTest. Only by considering the virtual domain can we properly understand the semantic potential of the episodes and actions which are encoded in the basic events of the received object, the narrative text. For the purposes of our narratological approach, therefore, we shall treat literary computing as a heuristic methodological tool with which we can test our narratological theory of action in an experimental computational implementation. This has two key benefits: it will help us refine our theoretical framework and will provide us with tools with which to investigate concrete literary data. The computer is a test-bed for our theory and not, we emphasize once more, a Swiftian machine for manufacturing interpretations ‘without the least assistance from genius or study.’

2.2 The EventParser Program The most important lesson to be drawn from our theoretical discussion in part 1 is that a fictive happening, even in its most basic form (X and then Y), never enters our consciousness in and of itself; it can do so only in the form of an interpretive construct. This basic construct of action logic, which we refer to as the event, corresponds to an assertion by the narrator or an assumption by the reader that two causally or chronologically linked states represent a transformation in the properties of a fictive object. Points (a) to (d) below will serve as a brief reminder of how events are defined and an outline of how they are encoded (the construction of events is known as encoding in EventParser terminology). Note that for brevity’s sake EventParser (as well as the EpiTest program to be described in chapter 2.4) use shortened terms and refer to the two types of state of affairs (i.e., expositional vs. dispositional states) as two statuses, to a fully defined expositional state of affairs as exposition and to its counterpart as disposition. 11 (a) It must be possible to assign every predicate to a pre-existent predicate class. (b) A focus must be present, where a focus is an identifiable, narratively based perspective of perception of an object in the narrated world. (c) Under this focus, the state of a fictional object of perception must be seen to be distinguished by predicate1 at point t1 in fictional time. Such a complex of object and predicate is referred to as a fictional state of affairs. The first state of affairs in an event construct is termed the expositional state of affairs (= exposition). (d) Under the same focus, or a different focus which has an equivalent epistemological function (i.e. which fulfils the same epistemological truth conditions), the state of the identical object from (c) must be seen to be distinguished by predicate2 at point t2 in fictional time.12 This is the dispositional state of affairs (= disposition). 11

12

Program specific terms appearing on the user interface of EventParser and EpiTest (in particular those defining windows and buttons) are generally identified by small caps in italics, e.g. EVENT PARSING window. The use of EventParser has to date been confined to object events; the following discussion therefore assumes the epistemological congruence of focus and object. However, we must not ignore the theoretical distinction between the two, and nor should we forget the presence of the discourse event as a theoretical concept.

2.2 The EventParser Program

215

Like all symbolic systems, narrating texts are characterised by a referential economy that suppresses information which is redundant or relatively easy for recipients to deduce from contextual evidence. To enable our theory to cope with this fact, we must formulate a supplementary rule which governs what kind of object can be involved in an event construct: (e) The fictional object in an event construct must fall into one of two classes. The first class consists of fictional objects which are referenced directly (e.g. by personal names or pronouns) or indirectly (e.g. by iterative or durative phrases such as during or at once) in a literal narrative assertion of their existence. The second, less obvious, class is composed of fictional objects whose existence at point t1 in time can be hypothetically postulated on the basis of points of narrative indeterminacy by back-projecting it from t2 (the dispositional state of affairs). It should be emphasized that rule (e) is not bidirectional: it cannot be inverted so as to allow us to look into the future beyond an expositional state of affairs. In other words, the rules of interpretation allow us to assume that, if the text documents the present existence of an object, the object must also have existed virtually in the fictional past. We cannot, however, speculate about how, if at all, such an object will exist in the fictional future after the most recent description of the state of the fictional world in the narrative. (The only exception to this is when the narrative itself opens the fictional world to the future, for example, by making prophetic predictions.) We cannot test an assertion that epistemologically (i.e. in terms of focus) transcends the interpreter’s maximum possible stable knowledge state. It follows that we cannot falsify action constructs whose formation depends on such an assertion; they are therefore beyond the scope of our approach. We can now turn to the practical description of EventParser. Apart from the welcome window, which appears when it is opened, the program contains a total of four windows. These windows provide an interface with which the user can designate and mark up the events in a text and make declarations to elucidate the world knowledge on which his definition of events depends. (1) The EVENT PARSING window is the heart of the text-processing interface. (2) The PREDICATE DEFINITION window allows the user to define new predicates. (3) The EVENT BASE FILE window allows the user to view and select encoded events.

216

Part 2 The Computer-Aided Analysis of Narrated Action

(4) The SEMANTIC DICTIONARY window lists all the descriptive terms currently in use. The list represents the program’s knowledge database, which expands dynamically as a text is processed. Once the text to be processed has been loaded, it is displayed in the EVENT PARSING window. The user can then process the text by performing a prescribed series of definition operations. The help panel on the right-hand side of the window displays instructions which guide the user through each stage of the mark-up process. The first step in defining our first event is to mark the text which contains its exposition. We do this by clicking and dragging with the mouse to select the appropriate range of characters as illustrated in figure 2.2.1.

Fig. 2.2.1:

EVENT PARSING

window

The help area now asks us to use the EVENT DEFINITION menu to define the focus. The EVENT DEFINITION menu is the key user interface element involved in allocating descriptive terms. It displays lists from which, depending on the stage in the mark-up process, we can select focus descriptors or predicates that describe the expositional and dispositional statuses. When the user loads a text for the first time, the list of narratorand actant-focus descriptors is empty, so, as with all such lists, the user must enter the appropriate term manually in the top field of the list. The

2.2 The EventParser Program

217

program automatically saves terms entered in this way and displays them in the list the next time it is displayed. The status options are slightly different, for they require one of the nine semantic classes to be selected before the class-specific lists can be viewed. As these latter lists are initially empty, we must enter the necessary predicate terms one by one when we start processing a new text.13 The program does not restrict the type and scope of the user’s predicates; thus, when using EventParser to encode a text, especially for a series of experiments, the researcher should ensure the comparability of the terms by establishing appropriate conventions in advance.14 When a descriptive term is assigned, it is displayed in the relevant field of the grey panel just above the text listing (as is already the case with our ferryman in figure 2.2.2). Isolated errors can be corrected by using the REDO buttons next to each field. Alternatively, the user can reject the entire event definition by clicking the CANCEL EVENT DEFINITION button; each stage in the event definition process must then be repeated from scratch.

Fig. 2.2.2: 13

14

EVENT DEFINITION

menu in operation

Note, however, that predicate lists from previous sessions can be reused in new texts simply by copying the *.lbf files. For example, in a test run with students and lecturers at the University of Hamburg, it was decided that only adjectives and adverbs could be used as predicate terms.

218

Part 2 The Computer-Aided Analysis of Narrated Action

The screenshot in figure 2.2.2 shows the state of the program when the user has defined ferryman as an actant focus, confirmed the definition by pressing ENTER or clicking the NEXT STEP button, and opened the list of nine expositional status predicate classes in the EVENT DEFINITION menu. The class terms are prespecified; in the current version, they were chosen on a purely intuitive basis—one of the shortcomings of our theory and methods as they stand. In part 1, we decided that the isolated predicate should be assigned to the exposition selected in figures 2.2.1 and 2.2.2. If we click on the social option in the class list, an empty list appears (not shown in the screenshot). We enter our new term isolated at the top of this empty list and confirm the new entry by pressing ENTER. The PREDICATE DEFINITION window is then displayed as in figure 2.2.3:

Fig. 2.2.3:

PREDICATE DEFINITION

window (expositional predicate)

In the PREDICATE DEFINITION window, EventParser prompts the user to enter the opposite term which accompanies isolated. The layout of the window shows the upper axis of a semantic square in the making. We define integrated as the opposite term of isolated and by doing so give the program a first, small indication of the normative and cognitive frame of reference which applies to the event being defined. After entering the new opposite term and clicking the BACK TO EVENTPARSER button, we are returned to the main EVENT PARSING window.

2.2 The EventParser Program

219

The fields in the grey panel above the text listing now display the current state of the event definition: we can see that the focus is ferryman and that the expositional status of the object is described by the isolated predicate, which is a member of the social predicates class. The definition of the selected segment of text as an exposition is now complete. Note that the numerical indices of the first and last characters in the selection (18 and 177 respectively) are displayed in the lower right-hand corner of the window under the heading CURRENT STRING. The program’s help panel now prompts us to repeat the process by selecting and defining the dispositional state of affairs in a similar manner. It is up to the individual interpreter to decide where this second state of affairs is located and how wide the scope of its definition should be. In chapter 1.4, we decided that the disposition is contained in the following segment: In the middle of the night loud voices wakened him; it seemed that travellers wanted to be ferried across. Goethe 1989:70

We also decided to define a categorially homogenous dispositional status. As illustrated in figure 2.2.4, we therefore provide the disposition with the requested predicate, which belongs to the same class (social) as the expositional predicate.

Fig. 2.2.4:

EVENT PARSING

window with predicate list

220

Part 2 The Computer-Aided Analysis of Narrated Action

As this involves the use of a descriptive term which the program has not encountered before, we have to enter it manually in the list box on the right-hand side of the main window, which now also contains integrated and isolated, the terms that we entered when defining the exposition. As before, we are prompted to define the new requested predicate in the PREDICATE DEFINITION window, which now displays a semantic square in which three positions have already been filled. The first pair of opposite terms, isolated and integrated, has moved to the bottom of the square; the new term, requested, occupies the top right-hand corner. By defining its opposite as ignored in the top left-hand field (see figure 2.2.5), we can complete our first semantic square. In the newly created square, the diagonal from isolated to requested represents the real event construct, while the diagonal from ignored to integrated represents our first virtual event construct (that is, an implicitly suggested event which is possible according to our world knowledge and compatible with the narrated world as we have read it).

Fig. 2.2.5:

PREDICATE DEFINITION

window (dispositional predicate)

221

2.2 The EventParser Program

When we return to the EVENT PARSING window, EventParser gives us the opportunity to add the event we have just defined to the event database. We do this by clicking the ADD TO EVENT BASE button, after which we are presented with a list of all the events that have been defined (see figure 2.2.6). Clicking on any event in the list displays its associated segments of text.

Fig. 2.2.6.

EVENT BASE

window

Clicking the RETURN TO EventParser button returns us to the main window, where a range of additional menus and buttons can be used to invoke further functions. In particular, the KNOWLEDGE BASE menu allows us to examine the knowledge database that is incrementally expanded as successive predicates are declared. The knowledge database is saved in a file with the .sbf suffix (sememe database file) and is displayed in the SEMANTIC DICTIONARY window, where semantic terms appear in the order in which they were entered or generated. In the example screenshot in figure 2.2.7, the knowledge database contains two pairs of opposites, two pairs of direct antonyms, and two pairs of direct synonyms. The antonyms and synonyms are direct because their elements belong to the same predicate class.

222

Part 2 The Computer-Aided Analysis of Narrated Action

Fig. 2.2.7.

SEMANTIC DICTIONARY

window

We have now reached the end of our discussion of the main components and principles of the EventParser software.15 The program allows us to define events in accordance with the theoretical and methodological criteria set out above and makes it possible to save formally consistent mark-up data (in event database, list database, and sememe database files) for subsequent differential and relational processing. The program creates the list database and sememe database files behind the scenes; they are then automatically combined to produce a master file with the suffix .esf (event-sememe file) in the format required for analysis in EpiTest. Superficially, EventParser is little more than a simple mark-up tool, but it also has a far more important function: at every stage, it forces the user to explicitly define the semantic terms being used and thus make clear the symbolically represented world knowledge which influences the 15

A number of auxiliary functions (see the panel at the bottom of the EVENT PARSING window) have been added to assist the user in dividing texts into events. Searching for user-defined characters or strings (PHRASE MARKERS) takes some of the effort out of selecting segments of text (PHRASES). A second search option is provided by the STRING SEARCH field, in which the user can enter a string for which to search by scanning forward from the insertion point. Unlike the PHRASE MARKERS field, the STRING SEARCH option does not alter the selection; its primary function is to locate specific names, adjectives, or similar features.

2.2 The EventParser Program

223

encoding process. In the next section of our discussion, we shall turn to EpiTest and the question of how to design an automated tool for analysing the combinatorial potential of events. What distinguishes a pair of events which can be linked to form an episode? What kind of relations exist between two such events? And do interpreters prefer some ways of linking events to others? Only when we have considered these questions in a theoretical context will we be able to devise generative algorithms which answer them. The algorithms will be used in EpiTest to search for virtual episode constructs and action metaconstructs in EventParser record files. In the process, it will become clear that formally represented world knowledge, a side effect of EventParser mark-up, is the cornerstone of combinatorial analysis in EpiTest.

2.3 Constructing episodes and actions Individual states of affairs can only be connected to form an event if there are certain formal or logical semantic links between them. Our theoretical discussion has shown that a similar requirement applies to the subsequent connection of two events to form an episode. The connection between the events in an episode construct (and thereafter in an action construct) can be decomposed into a combination of three interrelated features: (a) syntagmatic linkage, (b) ontic linkage, and (c) semantic linkage. Different combinations of these three criteria produce different functional variants of the episode. Each of the variants represents a cognitive synthesis performed by the recipient; as synthetic variants, they assume the fulfilment of certain logical semantic requirements before them and trigger new synthetic effects after them. The reason for this is that each and every episode construct expresses an implicit hypothesis about the principle which binds together the world, or at least the events embedded in it, as an action. We are therefore concerned with the necessary conditions of a fundamentally important construct. In the study of the aesthetics of reception, the construct is significant because it effects the synthesis of symbolically represented states of affairs into complex event sequences. More importantly, however, the construct provides the perceiving and receiving subject with the opportunity to objectify its own assumed identity in a fictive object, sometimes an event, ideally an episode, or, more ideally still, a complex action. Later in this chapter, the functional diversity of the episode will be illustrated in a matrix and demonstrated with a series of textual examples. First, though, we must consider the principles which control how events are combined to form episodes. This is best achieved with the aid of the abbreviated formal notation which is central to the combinatorial algorithms of the EpiTest application. The essential features of the notation are perhaps best illustrated with a brief example event; it will be familiar to the reader from our discussion of EventParser, where we have seen the recipient encode a concrete event at the beginning of Goethe’s Fairy Tale. In this event, an object (the ferryman) appears in a situation of social isolation which then becomes an event by being transformed into a situation where the ferryman is socially engaged. Behind its user-

2.3 Constructing episodes and actions

225

friendly interface, EventParser reduces the segments of text marked and semantically described by the user to the formal notation of the following multiple place proposition: event ('ferryman', 'social', 'isolated', '18', '177', 'social', 'requested', '178', '283'). Our formal notation has the following general form: event (Object, PCexp, Pexp, StartExp, EndExp, PCdis, Pdis, StartDis, EndDis). The variables Object, PCexp, Pexp, and so on contain the following pieces of qualitative and quantitative data which are required to provide a complete representation of an event: Object PCexp Pexp StartExp

EndExp

PCdis Pdis StartDis

EndDis

The object or subject of the event. Specifies the matter which appears in the two predications correlated by the event. exposition predicate category. The semantic category which contains the predicate that describes the exposition. exposition predicate. The predicate assigned in the exposition. Start of the exposition. The numerical index of the first character in the segment of text selected as the exposition by the recipient. End of the exposition. The numerical index of the last character in the segment of text selected as the exposition by the recipient. disposition predicate category. The semantic category which contains the predicate that describes the disposition. disposition predicate. The predicate assigned in the disposition. Start of the disposition. The numerical index of the first character in the segment of text selected as the disposition by the recipient. End of the disposition. The numerical index of the last character in the segment of text selected as the disposition by the recipient.

These notational conventions were designed with two purposes in mind: first, they provide a way to represent event mark-up; second, they reflect the features which determine whether events encoded by the recipient can be connected to form complex episodes. We shall now consider the principles which control the connection process.

226

Part 2 The Computer-Aided Analysis of Narrated Action

The Syntagmatic Link: Isochronous and Anisochronous Episodes Syntagmatic order is a purely formal requirement. It refers to the sequence of characters that make up the text and stipulates simply that the end of the first event (i.e. the final character of its disposition) be located before the beginning of the second event (i.e. the first character of its exposition). In formal terms, EndDis1 < StartExp2. This criterion makes a considerable number of events eligible for inclusion in complex synthetic episode constructs—it accepts every pair of events that the recipient has encoded in a linear order and separated clearly from one another by avoiding overlaps between disposition1 and exposition2. Less technically, this is equivalent to assuming that the logic of action can link anything and everything in the fictive world if it happens in a clearly sequential order and is narrated in that same order (i.e. isochronally).16 Being purely formal, the criterion of syntagmatic order strikes us as naive because of its vagueness and total ignorance of content. Moreover, it assumes that isochronous narration is used consistently throughout all narrative texts. But despite these shortcomings, we shall see that there are times when isochronous syntagmatic sequentiality becomes an indispensable requirement of episode construction. This is the case when the identity of the object of the two events in an episode is uncertain. The criterion of syntagmatic sequentiality is therefore of little use as an independent, absolute requirement; it is really a last resort, a position to which we can fall back during construct formation in order to compensate formally for semantic underspecification. 16

We have not previously placed any such formal restriction on the semantically defined internal structure of events. For the sake of consistency, however, the algorithm which generates isochronous episodes does place a condition of sequential order on the internal composition of events. It should be noted that this algorithm permits partial overlapping in events and episodes: the disposition of an event must commence after the start of that event’s exposition but need not await the end of the exposition. In formal terms, StartExp1 < StartDis1 and StartExp2 < StartDis2. As a rule, event constructs are formed directly from the linguistic material of the text which the reader receives in a linear order. Practical experimentation with EventParser has shown that even expert readers rarely exploit opportunities to link states of affairs anisochronally (retrospectively) into an analeptically constructed event. It would therefore appear that analeptic action constructs are generally realized at the more complex level of the episode. As a result, the distinction between isochronous and anisochronous event constructs is of little practical relevance for our purposes. Even so, we must assume that the text and its generic markers are the primary factors which control a recipient’s preference for linear event construction. Analytic narratives (for example, the classic detective story) live from the fact that they are able to hone their analeptic structure down to a particular retrospectively specified motif (the hidden expositional state of affairs) which constitutes a discovery far more forceful and effective than a prehistory that can be deduced or (re)constructed with relatively little effort.

2.3 Constructing episodes and actions

227

To conclude, two main classes of episode construct can be distinguished according to the nature of the syntagmatic link between their events. In an isochronous episode, the second event must not start before the end of the first event has begun: StartDis1 StartExp2. In an anisochronous episode, on the other hand, the second exposition must be located before the first exposition in the semiotic continuum of the text: StartExp2 < StartExp1.17 The Ontic Link: Object-Identical and Object-Neutral Episodes The isochronous and anisochronous classes of episode construct are defined by a semantically neutral criterion which considers only the numerical order of the sections of text which the recipient marks as the exposition and disposition of the events concerned. However, testing our second criterion, ontic linkage, requires the reader to explicitly interpret the content of the text. This brings the object into play as a fictive entity in the ontology of action. The object allows an event-defining transformation of states of affairs to be identified, and it is the shared element of any two events which are connected to form an episode. The criterion can thus be represented formally as the condition that Object1 = Object2. It is obvious that, in itself, this criterion is hardly sufficient to define what reasonably experienced readers intuitively classify as episodes and actions. The simple fact that the same fictive person or object is manipu17

This means that the narration of the second event must begin before the start of the event which actually precedes it in the fictive chronology. This definition deliberately prohibits the complete coextension of exposition2 and exposition1 that would be theoretically possible if we formulated our condition as StartExp2 < StartDis1. When complete congruence occurs between the expositions of two events received one after the other, we are concerned not with isochrony but with a polyvalence of action logic where the recipient has assigned two successive distinct dispositional states of affairs to one and the same exposition. A few words should be said about the conversion of these theoretical definitions into software algorithms. While the definition of the isochronous episode has been reproduced almost verbatim in prolog syntax, there are certain cases where our program does not specify explicit criteria for the anisochronous episode and uses an ex negativo definition instead. For example, the events of isochronous episodes are required to consist internally of isochronally narrated states of affairs. Anisochronous episodes, on the other hand, require neither isochronous nor non-isochronous internal composition; in fact, anisochronous narration often combines events which have been internally narrated in a fully isochronous manner. In theory, such an ex negativo definition can produce two sets of results. EpiTest does not generate double output in practice because of the nature of prolog: the language’s structural and syntactic features (recursively processed ‘else if’ branches and ‘set of’ predicates respectively) allow pairs of combinatorial variants to be merged into a single output list.

228

Part 2 The Computer-Aided Analysis of Narrated Action

lated in a particular sequence of fictive events is not enough to create a plausible episode context in the logic of action, at least not a context which meets the requirements of modern causal logic or the demands of contemporary aesthetics. Nonetheless, our theory must still consider the difference between object-identical and object-neutral episodes. As with the syntagmatic link, failure to fulfil this naive ontic criterion can be tolerated only if other links are activated in its place.18 The Semantic Link: Categorially Homogenous and Heterogenous Episodes When we construct actions, we intuitively know that we cannot connect two linguistically represented events to form a larger structure of action logic if the material of the first event is completely different from material of the second—they must be comparable in some way. The comparability test has two aspects: first, it must consider the objects of the two events; second, it must compare the events in terms of the classes to which their predicates belong. Against the background of the action concept of our event-based ontology, where the object is simply a function of its predications, the second aspect of the comparability test is clearly more important than the first. When readers read events, they decide what predicates belong to what classes, and their decisions therefore have consequences which extend far beyond the immediate context of event construction. This is why EventParser requires the user to assign every predicate he employs to one of nine heuristic classes. Thus, there are two features to note when the recipient encodes an event. Most obviously, the recipient interprets the content of the text in order to find a dispositional predicate which is distinct from the predicate of the preceding exposition and therefore sufficient to define a clear change in state of affairs. Second, we gain an insight into the axiomatic world knowledge which the recipient employs. In the simplest case, the interpreting reader builds the difference relation of an event using predicates which are available in a single category where they constitute (to some extent at least) a natural pair of opposites. For example, alive and dead are two elements of the class of biological predicates, in which they enter into a relation of contrariness with one another; at this stage, they are completely different from nor18

We have deliberately used the phrase ‘object-neutral’ instead of a stronger form referring to differing objects. The latter, negative formulation of the object criterion would bind us solely to genre- or epoch-specific antithetic action sequences which necessarily require the presence of non-identical protagonists and antagonists. In our weaker definition, this antithetic model is only one of several permitted models.

2.3 Constructing episodes and actions

229

mative predicates such as, say, redeemed and damned.19 This mundane situation becomes interesting when a reader treats the two different sets of material as a pair of opposites in a receptive event construct; in our example, this means identifying the change in property from alive to redeemed as an event. We cannot investigate the details of how this classcrossing connection is made; the reader should simply note that there is a close relationship between the elliptic, synthetic conceptual combination of distinct states of affairs and the cognitive application of metaphor by postulating analogy. The nature of this relationship is of considerable interest to metaphor theory and cognitive theory.20 For our purposes, the first significant point to consider is the distinction which our system makes between the two kinds of connection that can be formed across the boundary of a pair of events (i.e. between the disposition of the first event and the exposition of the second event). The predicates can be linked by categorially homogenous or categorially heterogenous connections (PCexp = PCdis and PCexp z PCdis respectively).21 As we have seen above, EventParser requires the user to enter opposite terms for every previously unknown predicate which appears during event encoding. The program uses this data in two ways. First, it writes the explicitly defined antonym (an opposite in EventParser terminology) to the sememe database file (*.sbf); second, it invokes a semantic algorithm which generates and represents the derived synonyms and antonyms in a single structure, the Greimassian square. The synonyms and antonyms thus produced are saved as prolog clauses in the sememe database file. Categorially homogenous antonyms and synonyms are treated as direct and saved in the forms dirsynonym (A, B) and dirantonym (A, -A) respectively; categorially heterogenous antonyms and synonyms are treated as indirect and saved in the forms indsynonym (A,B) and indantonym (A, -A) respectively. The opposites, defined directly by the user, and the antonyms, defined by EventParser, can be ignored for the time being as far as the episode is concerned; they are relevant only to those events which are based on a 19

20

21

See below for a further discussion of the problems inherent in juxtaposing such axiomatically extreme values. Note the situation in discourse analysis, where we meet the concept of the coherent topical trajectory of discourse episodes. In Linell and Korolija (1997:167f.), for example, the coherent topical trajectory is explicitly treated as a communicatively produced construct which is assembled by the discourse participants and conditioned by contextual resources. For simplicity’s sake, we shall ignore the theoretically possible distinction between episode constructs which have a categorially homogenous basis throughout their constituent events and episode constructs which are categorially homogenous only at the point where event1 borders on event2.

230

Part 2 The Computer-Aided Analysis of Narrated Action

relation of semantic opposition.22 However, both types of synonym pairs, dirsynonym (A, B) and indsynonym (A, B), are of primary importance for the episode, in which distinct events are connected on the basis of equivalence relations. We shall begin by considering two extreme cases in order to identify the weak and strong definitions of synonymy as it pertains to episode constructs. We meet the weak definition in what appears to be a bug in EventParser: under certain conditions, the program is unable to generate any synonyms at all. Such a situation occurs when, during the encoding of an event, the disposition is not only defined as categorially homogenous (belonging to the same semantic class as the exposition) but defined as such by means of a term which the user entered as the explicit opposite of the exposition predicate in the previous stage of encoding. This reuse of an axiomatically opposite term immediately after its creation causes the distinction between the levels of the semantic square to collapse—it levels the difference between the upper level of explication and the lower level of implication. There is no technical reason which prevents us from updating EventParser to handle this special case by making the user define additional explicit (context-independent) synonyms of the expositional and dispositional predicates. This would ensure that it remains possible to connect the events concerned into an episode. In reality, however, the apparent bug in the software indicates a unique, in some ways almost hermetic, kind of reading which converts sequences of states of affairs into isolated, unconnected monadic events which not only exemplify but also hypostatize the axioms involved. This kind of thetic event construct, which completely annuls its initial state of affairs without dialectically expanding its potential semantic connections, is an example of zero episodic content. The reader, moreover, should not attempt to overcome a case of zero episodic content by artificially providing his own context-independent interpretation of the event construct. Supplying the missing synonyms in an ex post facto semantic enrichment would be an act of meta-interpretation with no basis in the initial reading of the event. An event is, after all, defined by the antithetic juxtaposition of two successively perceived states of affairs and their predicates, whose comparability depends on purely formal features (their objects or their predicate classes), not on the semantic equivalence of their content.23 22

23

However, we shall return to opposites and antonyms again below because of the key role that they play in the automatic generation of metasynonyms. In the more extreme terms of a negative definition, it could be argued that an event construct can treat any semantic term which is not considered during its construction as a synonym of the first predicate. This dialectical manoeuvre, however, brings us no nearer to solving the practical problem—not because the set of synonyms of the first term grows to an enormous size, but because its composition will change with every event and never attain stability.

2.3 Constructing episodes and actions

231

In other words, a dispositional predicate cannot acquire semantic surplus without being involved in an event construct. A predicate’s semantic surplus is anything which the predicate makes available for semantic connection apart from itself and its opposite. It is not an abstract feature on a par with the predicate’s abstract membership of class or paradigm, and it never occurs in a functionally neutral environment; instead, semantic surplus is actually identical with the semiotic surplus that develops as an implication during the reading (i.e. construction) of an event. By inadvertently overspecifying itself, the dispositional predicate expresses the perpetuality of the semiotic process described as infinite semiosis by Eco (1981:168f.). Conversely, the reading of action ends before it has even begun if a recipient’s reading constructs instinctively short-circuit the syntagmatic sequence of states of affairs in order to make reference to the paradigmatic arrangement of an underlying set of axioms. A concrete example of an event with zero episodic content is found if we take the sentence The king died and then the queen died and encode it in such a way that we extract from it the single proposition event (queen, living, dead). It has always been felt that the action logic of this proposition is deficient, but no satisfactory account has yet been put forward to explain why. We are now in a position to provide the explanation that has previously been lacking: the proposition is unsatisfactory because the recipient creates it by directly invoking the axiomatically posited pair of opposites opposite (living, dead). As a result of the direct levelling of the semantic profile, neither a categorially homogenous (direct) nor a categorially heterogenous (indirect) synonym can be assigned to the predicates living and dead. Consequently, the event event (queen, living, dead) cannot be included in an episode which is formed on the basis of a synonym or indsynonym relation between predicates. This exclusion can only be overcome in two circumstances: first, if the dispositional predicate (dead) appears in identical form as the expositional predicate of a subsequent event;24 second, if the expositional or dispositional predicate is linked to 24

An example of this can be seen below in the definition of a concrete episode of category eiso3. From an aesthetic point of view, there does seem to be a case in which the recipient of a literary narrative finds it easier to handle absolutely identical terms of action logic: if the object nexus is absent, they can be treated as a kind of restricted or over-determined synonymy. An artificial example should illustrate this. The sequence of states of affairs The husband was poor / The husband was well-off / The wife was well-off / The wife was poor intuitively encourages the formation of a cyclical action construct. This is not true of the sequence of states of affairs The husband was poor / The husband was well-off / The husband was well-off / The husband was poor. Here, nothing happens between the first and second states of affairs of being well-off. The fact that readers accept a pragmatic fictional ellipsis is less significant than the fact that they are able to compensate for the semantic underdeterminacy which arises from the anaphoric use of the term well-off. It can be assumed that recipients shift the location

232

Part 2 The Computer-Aided Analysis of Narrated Action

a third term in another event, thereby exchanging its redundancy for the ability to engage constructively with the semantics and logic of action.25 We now turn from zero episodic content to the opposite extreme, where the dispositional predicate of the first event and the expositional predicate of the second are stored in our database as a categorially heterogenous pair of synonyms indsynonym (Pexp, Pdis). Here, it is theoretically possible to construct an event. On the other hand, we are also confronted with a new danger: constructing the event means making a categorially heterogenous connection, and the latter implicitly amounts to correlating entire semantic paradigms. If we match one predicate with another simply because it is convenient to do so, it follows that all members of the class of the first predicate can be matched with all members of the class of the second predicate. If distinct semantic classes are treated as functionally equivalent in this manner, the number of theoretically possible eventcombining complex structures will increase almost exponentially as the number of events increases. For example, a beta version of EpiTest contained a purely formal combinatorial algorithm that completely ignored contextual factors and limiting conditions. The algorithm used recursion to derive synonyms of the third to the nth order from the user-defined terms in the *.sbf file; that is to say, it generated synonyms of synonyms of synonyms and so on until all possible permutations were exhausted. The results were striking: given a sufficiently large set of initial terms, the context-independent algorithm produced result sets of an exponentially increasing size. This extreme form of philological ‘semantic crunching’ devours processing resources as rapidly as any of its number-crunching counterparts in the more exact sciences. However, a brute force approach of this type, which throws computing power into the combinatorial maelstrom in order to make connective possibilities available at all costs, leaves us with little more than the semantic equivalent of a Pyrrhic victory. A context-independent combinatorial semantic system is ultimately self-defeating, for it will eventually make it possible to correlate just about anything with everything. In our case, this means that it will become possible to connect any two states of affairs to form a hypothetical event and subsequent episode. On the other hand, the automatic generation of second-order synonyms (synonyms of synonyms), at the very least, is indispensable. In mixed semantic combi-

25

of the semantic transformation to the meta-level of narrative happenings and treat the repetition as narratorial amplification. Thus, to ensure that action can still be read, the recipient must intervene and artificially raise the barrier between the two pragmatic and discursive levels of action. The addition of grief in Forster’s minimal story creates semantic surplus and activates connective potential in this very way. It should, therefore, not be dismissed as a supplementary causal complex of questionable value from a relatively trivial realistic perspective.

2.3 Constructing episodes and actions

233

nations, for example, second-order synonyms allow us to identify hidden three-dimensional implication relations between two distinct semantic squares.26 This can be abstractly illustrated as follows:

Fig. 2.3.1: Logic of Second-Order Synonyms

Internal synonymy relations exist inside each square and are denoted by the vertical arrows; in addition, however, a new synonymy relation is established when the two squares are considered together. It originates as a horizontal link between the two upper pairs of terms. -Exposition1 and -Exposition2 are functionally equivalent because they share with -Disposition1 an implication relation. The latter term has been newly entered in square 1 and is therefore a predefined part of our knowledge when it comes to be used again in square 2. The same is true of the elements +Exposition1 and +Exposition2; they are both synonyms of the same term, +Disposition1. We can now fill out the abstract framework using a pair of squares built with natural language adjectives:

Fig. 2.3.2: Instantiation of Second-Order Synonyms

The implication to be derived from these two squares will be familiar to anyone who is used to reading fairytales: ugly corresponds to evil, beautiful to good, and ugly is the opposite not only of beautiful but also of good. And each of the terms is obviously also a functional synonym of itself. Yet, 26

In a mixed semantic combination, the expositional term is new (entered for the first time), and the dispositional term is old (already stored in the knowledge database), or vice versa.

234

Part 2 The Computer-Aided Analysis of Narrated Action

obvious as this logic is to humans, it means nothing to an algorithmically driven machine: a computer program can recognize implications such as this only if we enter them explicitly in its knowledge database—unless, that is, we supply it with a procedure for automatically generating secondorder synonyms and opposites. EpiTest, in fact, has just such an algorithm for generating second-order synonyms (termed metasynonyms). The algorithm operates by analysing the implication relations contained in the semantic terms of the *.esf files generated by EventParser. If second-order relations are identified, the program automatically makes the necessary additions to its dynamic knowledge database before proceeding to compute episode and action constructs. EpiTest displays the metasynonyms that are generated during each cycle in a log window and saves the results in a separate file called metaterms.msf for later examination. It is clear, then, that we need to be able to extend the semantic database combinatorially; the operation of the extension process, however, must always be limited by certain constraints. This shows how hard it is to model, even approximately, the network of relations of semantic equivalence and opposition which make up human world knowledge and are necessary for the reading (i.e. construction) of episodes and actions. In addition, we see that questions such as this, which initially seem to involve nothing more than pragmatic technicalities of algorithm design, can also reveal hitherto unknown aspects of the many assumptions and requirements which govern natural interpretive processes. In such cases, using computers achieves more than a technological acceleration and quantitative gain: it opens up fresh epistemological insights by forcing us to adopt a new methodological rigour. We discover things we have forgotten to explain properly, and we find the opportunity to learn from our mistakes as we build our models of literary theory. To return to our original problem, the formation of categorially heterogenous constructs: it is clear that we require a limiting condition similar to that which applies when the syntagmatic sequentiality criterion is suppressed in anisochronous connections. The limiting condition is necessary in order to ward off the threat of universal semiosis which would ultimately cause the self-destruction of all meaningfulness. The condition which governs the indirect categorially heterogenous connection of events to form an episode can be stated as follows: the object of the two events must be directly identical. This restriction ensures that we cannot pair events with one another simply by placing any predicate of a given class alongside any member of another predicate class. The two predicates can be correlated if and only if they are explicitly bound (by the recipient at least) to the same object. And so we see again that the act of construction is inseparable from the process of interpretation.

2.3 Constructing episodes and actions

235

Closure In our theoretical discussion of the episode in part 1, we considered the intuitive requirement that episodic sequences of states of affairs must have closure. We proposed that this coherence criterion is satisfied when two distinct event squares can be integrated into a single complete semiotic (episodic) square. This condition is automatically fulfilled by all events which have been encoded in EventParser and successfully connected by the EpiTest algorithms: as soon as the semantic predicates of disposition1 and exposition2 are identified as equivalents, either directly (via an identical term) or indirectly (via a shared synonym), the co-terms in the two distinct semiotic squares (opposites, antonyms, and synonyms) are also equivalent. In this respect, the closure condition does not need to be defined explicitly—it is resolved as a side effect of our model’s highly abstract approach. It may seem scandalous to deal with coherence, one of narrative theory’s favourite problems, in such a frivolous way. Is that a weakness of our model? Or is it an indication of the illusory nature of the problem itself? The reader is left to his own conclusions. The episode Matrix Our model recognizes five combinations of the above criteria which are sufficient for a pair of events to be connected in an episode. The connection is made at the boundary between the second and third states of affairs in the logic of action (disposition1 and exposition2 respectively). The five resultant episode types, eiso1–3 and eaniso1–2, are shown in table 2.3.1.

Table 2.3.1: The episode Matrix

236

Part 2 The Computer-Aided Analysis of Narrated Action

We shall now illustrate these five episode types using examples from Goethe’s Fairy Tale. We provide the relevant quotations from the text and use formal notation to describe the events they contain and the episodes that can be built out of them. The first type is eiso1, the isochronous, object-identical, categorially homogenous episode. exposition1 ‘By the great river…the old ferryman…lay in his little hut and slept.’27 disposition1 ‘In the middle of the night loud voices wakened him; it seemed that travellers wanted to be ferried across.’ event1

event ('ferryman', 'social', 'isolated', '18', '177', 'social', 'requested', '178', '283').

exposition2

‘The old man pushed off without delay and rowed across the river…while the strangers hissed at one another in an unfamiliar…language.’

disposition2

‘He was gone and did not hear them.…There among towering rocks he found an enormous cleft, tossed in the gold, and returned to his hut.’

event2

event ('ferryman', 'social', 'with company', '450', '615', 'social', 'at home', '2542', '2836').

An episode of category eiso1 can be built from these two events because: (a) the object is identical in both event clauses; (b) the constituent states of affairs of both events are consistently isochronous both inside the events and in their overall order; (c) the hypothetical constructing recipient has assigned all the predicates to the social class and, in addition, defined the second expositional predicate with company as a categorially homogenous (direct) synonym of the first dispositional predicate requested, as a result of which the semantic database will contain a prolog clause dirsynonym ('with company', 'requested'). On the basis of this data, the EpiTest algorithms generate the following complex episode and store it in a new file with the suffix *.con (construct file): episode

27

episode (eISO1, 1, 'ferryman', 'social', 'isolated', '18', '177', 'social', 'requested', '178', '283', 'ferryman', 'social', 'with company', '450', '615', 'social', 'at home', '2542', '2836').

This and all subsequent quotations are taken from the English translation in Johann Wolfgang von Goethe: Conversations of German Refugees (Goethe 1989:70ff.). After installing the downloadable files obtainable at the respective text base file to be loaded into EventParser for a re-run of our experiment can be found under c:\compact\Fairytale\fairy1.tbf (assuming that the default installation settings were used.)

2.3 Constructing episodes and actions

237

We now consider eiso2, the isochronous, object-identical, categorially heterogenous episode. It is our first example of a categorially heterogenous episode, and is based on the same textual starting point as the categorially homogenous episode described above. However, in this case, the expositional state of affairs is described in terms of physical or biological predicates (PCexp = physical), and the dispositional state of affairs is described in terms of cognition or perception (PCdis = cognitive). exposition1

‘By the great river…the old ferryman…lay in his little hut and slept.’

disposition1

‘In the middle of the night loud voices wakened him; it seemed that travellers wanted to be ferried across.’ event ('ferryman', 'physical', 'asleep', '18', '177', 'cognitive', 'hearing', '178', '283'). ‘Stepping outside he saw hovering over his moored boat two large will-o’-the-wisps, who insisted they were in a great hurry and wished they were already across. The old man pushed off without delay...’ ‘“Old man! Listen, old man! We have forgotten the most important thing!” He was gone and did not hear them.’ event ('ferryman', 'social', 'reacting', '290', '486', 'social', 'unaffected', '2470', '2576').

event1 exposition2

disposition2 event2

Once these passages have been described using EventParser, only one extra operation is needed to make the boundary between disposition1 and exposition2 meet the criteria required for episode construction. This extra operation means generating a pair of second-order synonyms metaindsynonym ('hearing', 'reacting'). This is essential in order for the synonymy requirement to be fulfilled. The synonym pair is derived algorithmically in EpiTest, which scans the *.esf file, identifies the shared opposite (unaffected) of hearing and reacting, and therefore evaluates the latter two terms as synonymous. Unaffected appears in two opposite clauses in the relevant section of the file: opposite ('hearing', 'ignoring'). opposite ('reacting', 'ignoring'). dirantonym ('reacting', 'unaffected'). dirantonym ('ignoring', 'affected'). dirsynonym ('ignoring', 'unaffected'). dirsynonym ('reacting', 'affected'). opposite ('unaffected', 'affected'). The resultant second episode is represented in formal notation as:

238

Part 2 The Computer-Aided Analysis of Narrated Action

episode

episode (eISO2, 2, 'ferryman', 'physical', 'asleep', '18', ‘177', 'cognitive', 'hearing', '178', '283', 'ferryman', 'social', 'reacting', '290', '486', 'social', 'unaffected', '2470', '2576').

The third type of episode, eiso3, is the isochronous, object-neutral, categorially homogenous variant. events cannot normally be connected to form an episode if their objects do not supply a connection between them. However, this ontic deficit can be overcome if the predications stand in strictly isochronous order and are categorially homogenous. We shall illustrate this using an example from our text which begins with an event whose encoding is similar to that of eiso1 but contains withdrawn instead of isolated and cooperative instead of requested. exposition1

‘By the great river…the old ferryman…lay in his little hut and slept.’

disposition1

‘Stepping outside he saw hovering over his moored boat two large will-o’-the-wisps, who insisted they were in a great hurry and wished they were already across. The old man pushed off without delay.’

event1

event('ferryman', 'social', 'withdrawn', '18', '177', 'social', 'cooperative', '290', '486').

exposition2

‘“I am forever in your debt,” said the serpent….“Ask of me what you will! Anything in my power I will do for you.”’

disposition2

‘“This service,” replied the serpent…, “I cannot perform at once. Unfortunately, the fair Lily lives on the other side of the river.”’

event2

event ('snake', 'social', 'cooperative', '6607', '6767', 'social', 'unable', '6974', ‘7122').

The indices of the beginning and end of both events show clearly that they are in isochronous order, and both events have been classified using elements of a single predicate class (social). The semantic connection between them is established by the adjective cooperative. It is applied to ferryman and snake and functions as the predicate of disposition1 and exposition2. It is worth noting that both scenes refer to the river and the desire of the will-o’-the-wisps to cross it. The similarity in setting can hardly be a chance occurrence—our text guides the process of reception and construction using scenic pairs and the fields of semantic association they anaphorically re-evoke. EpiTest generates the following episode and stores it in the construct file (*.con suffix):

2.3 Constructing episodes and actions

episode

239

episode(eISO3, 3, 'ferryman, 'social', 'withdrawn', '18', '177', 'social', 'cooperative', '290', '486', 'snake', 'social', 'cooperative', '6607', '6767', 'social', 'unable', '6974', '7122').

That completes our inventory of the three isochronous episode types. We turn now to the two anisochronous types, beginning with eaniso1, the anisochronous, object-identical, categorially homogenous episode. exposition1

‘With a graceful bow the young gentlemen took their leave. The serpent was happy to be rid of them, partly to enjoy her own light, partly to satisfy a curiosity that for some time had been tormenting her. In the rocky clefts, where she often crawled about, she had made a remarkable discovery.’

disposition1

‘She believed she could now illuminate this wondrous subterranean vault with her own light, and hoped all at once to become fully acquainted with these peculiar objects.’

event1

event ('snake', 'cognitive', 'unacquainted', '8575', '8873', 'cognitive', 'acquainted', '9798', '9966').

The tense change in the final sentence of the expositional state of affairs signals that we are going to encounter an analepsis sooner or later; only with the beginning of disposition1 (as identified here) do we return to the isochronous course of the narrative. Bracketing the analepsis with the first event like this is expedient because it allows expectation and protension to help us encode the analeptically narrated event: what we have received and construed as an event must, in some way or another, connect to the end of what we have already read but not yet encoded as an event. Here is the embedded analepsis, whose exposition overlaps the disposition of the event which is received first but located second in the fictive chronology: exposition2

‘In the rocky clefts, where she often crawled about, she had made a remarkable discovery. Although she had to creep through these chasms without any light, she could distinguish objects well enough by touch.’

disposition2

‘But to her great amazement, inside a rock enclosure sealed off on all sides, she had felt objects…. All these perceptions she now wished to bring together at last through her sense of sight, and to confirm what she only conjectured.’

event2

event ('snake', 'cognitive', 'perceiving', '8785', '8991', 'cognitive', 'investigating', '9266', '9966').

240

Part 2 The Computer-Aided Analysis of Narrated Action

We have encoded event1 and event2 in such a way that they fulfil all the necessary conditions for the backward (anisochronous) connection of exposition1 and disposition2 according to the criteria of the eaniso1 episode type—the objects and predicate classes are identical; exposition1 appears before the end of disposition2 in the semiotic continuum of the text, even though disposition2 comes first in the ordo naturalis of the action logic; and the first expositional predicate unacquainted can be interpreted as semantically equivalent to the second dispositional predicate investigating, which is logically posterior but textually anterior to disposition1. The following resultant episode is assembled on the basis of the *.esf file, again with the help of a shared opposite term bound to each predicate in an opposite clause: episode

episode(eANISO1, 4, 'snake', 'cognitive', 'perceiving', '8785', '8991', 'cognitive', 'investigating', '9266', '9966', 'snake', 'cognitive', 'unacquainted', '8575', '8873', 'cognitive', 'acquainted', '9798', '9966').

We now come to eaniso2, the anisochronous, object-identical, categorially heterogenous episode. This is the last of the five ways of connecting events to form an episode. For the most part, eaniso2 corresponds to eaniso1, except for the fact that it does not require that the predicates in the states of affairs of disposition2 and the adjoining exposition1 belong to the same semantic class. exposition1

‘In this cleft lay the lovely green serpent, who was awakened from her sleep by the coins clinking down. She no sooner caught sight of the shining disks than she immediately devoured them with great relish.’

disposition1

‘No sooner were they swallowed than she felt the gold melting in her entrails…and to her great joy she discovered that she had become transparent and luminous.’

event1

event ('snake', 'physical', 'inconspicuous', '2842', '3046', 'physical', 'transparent and luminous', '3139', '3368').

exposition2

‘Long ago she had been assured…’

disposition2

‘…that this phenomenon was possible.’

event2

event ('snake', 'philosophical', 'doubting', '3369', '3399', 'philosophical', 'having been assured', '3399', '3432').

The episode construct below clearly assumes that the reader is prepared to perform a number of interpretive operations: he must resolve an ellipsis in event2 by assuming that, as the snake has been assured, she must

2.3 Constructing episodes and actions

241

have expressed doubts earlier, and he must agree that an agent assured of something is the opposite of a physically inconspicuous agent (we cannot assure a person if we are not aware they are present). Once again, these complex interpretive operations are made possible by the combinatorial richness of the semantic lexicon. The resultant episode is: episode

episode(eANISO2, 5, 'snake', 'philosophical', 'doubting', '3369', '3399', 'philosophical', 'having been assured', '3399', '3432', 'snake', 'physical', 'inconspicuous', '2842', '3046', 'physical', 'transparent and luminous', '3139', '3368').

2.4 The EpiTest Program The five examples we have just considered are not randomly generated episode constructs. They were built in EpiTest, our second software tool, by processing ten events which had been defined in advance for the express purpose of illustrating the episode type matrix. Many of them, for example eiso1 and eaniso2, are more or less compatible with an intuitive concept of the episode as a minimal action with some kind of closure. Others may strike the reader as too artificial, but, in general, the examples are sufficient to show that instances of all five theoretical episode types can be found in real texts. The EpiTest program was developed with a specific task in mind: combinatorial analysis of *.esf files produced by EventParser. The heart of the program consists of a set of algorithms which search saved databases of events for connections which correspond to the categories in our matrix. EpiTest was written in prolog (Programming in Logic), a programming language which has long been regarded as ideal for developing artificial intelligence models and is also one of the standard tools of computational linguistics. It is only a short step from the current applications of prolog to using the language to model the cognitive processes involved in the production and reception of narrative texts, but only in isolated cases has this actually been done.28 28

“An Introduction to a Computational Reader of Narrative” (Shapiro and Rapaport 1995) is one example which is particularly relevant to our approach. The article describes the SNePS system (an ‘intensional, propositional, semantic-network, knowledge-representation and reasoning system’) and its practical implementation in CASSIE (‘Cognitive Agent of the SNePS System—an Intelligent Entity’). Two aspects of the project deserve special attention. First, the knowledge base of the SNePS system is specifically designed to represent intensional mental substance (the mental concept of an object and its properties), not extensional semantic content (content bound to a referential fictional object). By representing entire narratives in this way, SNePS compiles a description of the metapropositions which are considered to be true and immanently identify the narrative system as such. Second, SNePS combines object knowledge and rule knowledge in a single level—it does not maintain separate registers for rules and facts but represents rules as conceptions of facts of logical dependence, inference, and so on. This means that knowledge of rules and knowledge of propositions are directly linked to one another, can be directly combined with one another, and can extend one another directly. This is in fact a defining feature of intuitive human reading: beings and happenings in the fictive world are expressed as propositional statements of states of affairs and read as knowledge about the rules of the fictional world. Shapiro and Rapaport argue that the flat knowledge system of their model is a necessary prerequisite for describing the

2.4 The EpiTest Program

243

The code listing of EpiTest version 3.85 (the version designed and used when writing the present study) runs to about 2,240 lines including comments.29 The comments document the route taken through the code during execution and should help readers who are not familiar with prolog to understand how the program and its various models operate.

Fig. 2.4.1: EpiTest flowchart 30

29

30

backward connections of the dynamic reading process in a theoretical system and modelling them in cognitive theory. A logical programming language like prolog is ideal for implementing the methods of such an approach. All languages make it relatively easy to code deterministic algorithms which make the computer find a suitable solution for a given problem; prolog’s distinctive strength lies in its outstanding support for recursive algorithms. Backtracking therefore allows combinatorial analysis to be used not only to identify a single ideal solution but also to list all the possible solutions generated in reply to a given query. In addition—and this is particularly important for the SNePS architecture—prolog allows knowledge (data specified in propositional clauses of the form fact (object, property1,…, propertyn)) and program code (instructions of the type if A do B else do C) to be dynamically extended, reduced, modified, and so on during the course of program execution. Prolog’s support for recursion and dynamic extension makes it possible for the language and any program written in it to be described as intelligent to the extent that they provide a way of modelling continual learning. As noted above, the Shapiro and Rapaport approach works towards modelling a computational reader, not a story generator. In this respect, it is an extension of Ryan (1985) and reception theory’s counterpart to Lebowitz (1985). Comments are sections of a source code listing inserted by the programmer to clarify his code to the reader—or indeed sometimes to himself! Comments are completely ignored during program execution; in prolog syntax, they are marked by a preposed % or enclosed in /*…*/. The flowchart illustrates how our software exploits the combinatorial capabilities of prolog. EpiTest first builds a list of all the episode constructs that fit the appropriate

244

Part 2 The Computer-Aided Analysis of Narrated Action

However, it will probably be easier to digest the flowchart shown in figure 2.4.1. With the help of the diagram, we can describe a complete execution cycle of EpiTest as follows: 0. Synonyms, antonyms, and event constructs are read from the record file (broken lines). 1. metasynonyms are generated in Algorithm 1 (Module 6 in the program listing). They are then added to the program’s dynamic knowledge database (solid line). 2. episode constructs are generated by testing for combinatorial connections in Algorithm 2 (Module 4). The constructs are saved in the *.con construct file and added to the dynamic knowledge database (dotted line). 3. action constructs are generated by combinatorially connecting episode constructs in Algorithm 3 (Module 5). The first action construct is saved in the dynamic knowledge database before the program recursively checks whether further episodes can be added to the action chain. Complete action constructs are saved in the *.con construct file. 4. The constructs that have been generated are subjected to quantitative statistical analysis in Algorithm 4. The reader may well ask why we should spend a considerable amount of time and effort designing these computational algorithms in order to implement what is already a highly abstract formal system of narrative theory and action logic. Some quantitative data may help illustrate the potential benefits. In our illustration of the different types of episode, we analysed the combinatorial potential of ten events and identified five acceptable episodes which can be produced from them. If we enter our ten basic events in the sample file fairy1.esf and analyse this file in EpiTest, the program generates the same five episodes as those we found manually. It also produces 246 metasynonyms which allow it to uncover hidden semantic connections and more than quintuple the size of its semantic database. On this extended basis, EpiTest generates forty-one additional, unpredicted episode constructs, each of which fully satisfies the criteria of one of our five categories eiso1–3 and eaniso1–2. Now, these results could have been worked out manually, given a few sharp pencils, plenty of paper, and even 31

criteria and then initiates a second recursive process for building action constructs. However, EpiTest is not really an intelligent program—although it adds newly generated factual knowledge (constructs and metasynonyms) to its knowledge database, it does not generate dynamic rule knowledge of any kind.

2.4 The EpiTest Program

245

more patience—but not in the 0.33 seconds which our program needed to complete the task.31 When it has finished processing a file, EpiTest presents the display shown in figure 2.4.2.Fig. 2.4.2: EpiTest user interface

Fig. 2.4.2: EpiTest user interface

The various features of the program can be briefly described by considering the nine buttons in the screenshot reproduced in figure 2.4.2. The four buttons in the panel on the upper left-hand side relate to the database file being evaluated (in this case, it is fairy1.esf, which was previously produced in EventParser). CONSULT allows the user to open a new file; EVENTS lists the current file in a text field; SYNONYMS displays a similar list of categorially homogenous (direct) and categorially heterogenous (indirect) synonyms; METASYNONYMS calls the algorithm for generating metasynonyms (see above) and lists them in a text field. To the right of these last three buttons, the program displays the number of events that have been read and the number of user-defined and program-generated semantic terms. After the program has read and prepared a *.esf database, the user should specify what kind(s) of combinatorial analysis are to be performed. 31

System specification: 2.66 GHz Pentium IV processor with 512 MB RAM, Windows XP.

246

Part 2 The Computer-Aided Analysis of Narrated Action

Depending on the checkboxes that have been selected, EpiTest will generate isochronous episode constructs, anisochronous episode constructs, or action constructs. The buttons in the top right-hand corner concern the actual generation, saving, and display of these constructs. CONSTRUCT initiates the generation process and asks us to define the name of the results file (*.con suffix). Once the program has completed its calculations and displayed the message ‘Database analysis completed,’ the results can then be displayed in text fields. EPISODES opens a list of the episode constructs that have been found; ACTIONS shows the possible ways of linking these episode constructs in the form of a prolog predicate action(1, 2, 3,…, n), where each of the numbers inside the parentheses is the index of an episode which makes up the action. SAVE allows the results of the various calculations to be stored in a separate *.daf (data analysis) file. This file can be opened later in any text editor and contains a one-page statistical summary of the results. EXIT quits the program. The panel on the bottom right-hand side displays the number of episodes that were generated in each category and the total number of action constructs they produced. In the lower left-hand corner, we can see the time taken by EpiTest to complete its calculations. When EpiTest has finished generating episode constructs on the basis of the fairy1.esf file and stored them in the fairy1-aniso.con results file, the user can click on EPISODES to display a list of the episode constructs which were found, as shown in the screenshot in figure 2.4.3.

Fig. 2.4.3: List of episodes generated by EpiTest

2.4 The EpiTest Program

247

In this case, the original fairy1.esf file contains only ten events, but, even so, the episode list displays a total of forty-two virtual episode constructs. Representing EpiTest output in a list format like this makes clear that a considerable number of episodes (theoretically possible ways of combining events) can be identified by using a computer to analyse reception records (*.esf files produced in EventParser). The output does not lend itself to qualitative analysis—it is not a reading whose merits can be evaluated by critics, and the methods behind it cannot be judged against the aesthetic standards of a hermeneutics of interpretation. It is obviously meaningless to ask which of the forty-six episode constructs or thirty-eight action constructs in our data is more correct, more satisfying, or even better in the sense of being more faithful to the original text. Qualitative evaluations, where the original text of the literary narrative is paramount, might therefore seem irreconcilable with the quantitative calculations on which our model is based. But this is not the case—by introducing the concept of virtual action potential, we will we able to distinguish individual texts or even entire corpora from one another by studying the values of this new statistic.

2.5 Action Potential and action Product The action potential which is described in this chapter has only one constant factor: the literary text, a concrete object which consists of a finite set of symbolic signs that are arranged, so the empirical evidence tells us, according to certain syntactic and grammatical rules. The signs must be processed in a number of ways—through reading, interpretation, recombination, evaluation, comparison, and so on—before we can even begin to speak of concepts, let alone actions. The signs are processed under the influence of many variables—the reader’s ability, expectations, intertextual and world knowledge, and aesthetic norms, to name but a few. These variables differ not only from reader to reader (i.e. from constructor to constructor) but also inside individual reading subjects and receptive processes, where closer observation shows that they can vary dynamically from text to text and even from sentence to sentence. With each successive sentence, the reader knows more, discovers new associations, and expects different things. Reception in general, and the reception of actions in particular, is fundamentally dynamic in nature and cannot be reconstructed in a supposedly ideal reading situation in order to facilitate the experiments of our computational approach. However, readers do not evaluate literary narratives on a purely subjective basis by considering the richness of their action logic, the coherence and depth of their chains of fictive happenings, and the originality and elegance with which they combine events. Readers also take part in discussions with other readers. Theoretically, the semiotic process may continue into infinity as Eco has described, but, in practice, we will terminate it on the basis of our pragmatic needs. The same, I believe, applies to the discussion, be it naive or critical or philosophical, of the action potential of a narrative text. Only if the reading (receptive construction) of action can create synthesis, meaning, does it make sense to try to define action potential in theory and measure it in practice. The closer a receptively constructed action clings to the referential framework of fictively represented actors, things, situations, and transformations, the more it is reduced to nothing more than a denotational function; but the more an action distances itself from the fictive substance borne by a possible world and is condensed into a sequence of transformations which can be reconstructed as abstract formal propositions, the more it becomes a hollow, empty shell of meaning, the illusory meaning of a receptive happening that portrays itself as the only

2.5 Action Potential and action Product

249

true action. From this perspective, the empirical reading and construction of action can be seen as an attempt to find a way through the world of latent possibilities contained in a network of events. Our definition of the episode as the smallest such action construct provides us with a unit with which to measure and quantify this world of possibilities. We shall begin by considering the upper and lower theoretical values that can arise. If only a single event is present, it is impossible to form an episode at all. Once two events, the minimum quantity needed for an episode, are present, there are two ways of combining them. The definitions below assume that our two events are (1) Egg from chicken and (2) Chicken from egg. (a)

(b)

Hyperdetermination: every event capable of entering into a connection can enter into one and only one such connection. Thus, we must read our two events either as (1) + (2) or as (2) + (1). The formula required in this case is simple: for n events, the minimum number of episode constructs (Epmin) is given by the formula Epmin = n / 2. Hyperconnection: every event can connect with every other event to form a pair, but not with itself. The ontic primacy of chicken over egg is just as conceivable as the primacy of egg over chicken, and there are two possible episodes, (2) + (1) and (1) + (2) respectively. For n events, the maximum theoretical number of episode constructs (Epmax) is given by the formula Epmax = n(n – 1).

In a practical context, these simple formulae can be applied to obtain numerical values which describe the minimum and maximum levels of episodicity in a concrete text. They delineate the spectrum of event combinations which are theoretically possible on the basis of a particular reading before syntagmatic and semantic criteria are considered. Putting the formulae into practice using the demonstration data of the ten events discussed above, we find that there is a theoretical minimum of 10 / 2 = 5 episodes and a theoretical maximum of 10 * (10 – 1) = 90 episodes. The actual number of virtual episode constructs which can be produced will lie somewhere inside the theoretical range of five to ninety; in the present case, it is forty-six. Expressed as a percentage, this gives us an event integration of just over 54 percent. In less formal terms, this means that (a) the events in the text can be combined into eighty-five theoretically possible distinct episode constructs; and (b) of these eighty-five theoretically possible constructs, forty-six (i.e. 54.1 percent) can be virtually instantiated on the basis of the semantic and formal predications with which the events have been encoded. EpiTest calculates this value for us and displays it in the upper right-hand corner of the window under the heading % EVENT INTEGRATION.

250

Part 2 The Computer-Aided Analysis of Narrated Action

The action potential of a text, which basically measures the degree to which a given reading can be translated into overall action constructs, can now be evaluated in terms of the combinatorial affinities between atomic event constructs and molecular episode constructs. However, we shall not stop here; our quantitative measure of action potential will also consider the scope for episodes themselves to be joined together in action logic. This means that, having considered local (episode-forming) event synthesis above, we must now consider global (action-forming) event synthesis as well. Ideally, we would define a series of semantic criteria similar to those used in the episode matrix; this would allow our theory to represent our intuitive knowledge that the scope of a particular action narrative in the form of an action construct depends primarily on the presence of a dominant semantic category, or theme. However, such a refinement is beyond the remit of the present study, and we must be content with a provisional, pragmatic criterion instead. EpiTest’s combinatorial algorithm is therefore based on the following plausibly simple, broad rather than narrow, definition of the action construct: two episodes form an action if and only if they are sequentially ordered in the semiotic continuum of the text. This test therefore does no more than provide a solution to the following question: assuming that episode a is the first in an unbroken chain, what episodes (b, g, d,…) can be linked to a in an isochronous order? The answer obviously depends on the connective potential of the initial construct a in each case.32 In the case of our demonstration file, we obtain a striking result: despite the high event integration of 54 percent, the forty-six virtual episode constructs cannot produce an action list with more than two elements. Consider again the second row of figures in the top right-hand panel of the EpiTest window in figure 2.4.3. On the right of the ACTION button, 32

Successive new connections will obviously become possible as we move through the second, third, and subsequent positions in the list. The corresponding action lists are represented as actlist (1, 2, 3, …, n) in the program syntax. However, EpiTest generates them with an algorithm that is partially deterministic rather than fully recursive: it takes from the database only the first episode construct which isochronally follows its predecessor on any occasion. A fully recursive algorithm would generate a huge number of combinatorial variants, and we are not yet in a position to place such a burden on reader or computer. Practical tests have shown that it takes long enough to calculate deterministic action constructs alone—of the 0.39 seconds needed to process the Matrix.esf file, only about 0.05 seconds were required to assemble the episode constructs; the rest of the time was spent building action lists. Processing Unterhaltungen.esf, the complete mark-up file for the Conversations of German Refugees (our example text in part 3), took over four hours on a 266 MHz Pentium II processor. (Advances in computer technology since the time of writing will have reduced this figure considerably by now.) If nothing else, these figures illustrate the magnitude of the cognitive task which humans perform so easily and find so deceptively simple when they read action.

2.5 Action Potential and action Product

251

we can see numerical values for the length of the shortest action chain generated, the length of the longest action chain generated, and the average length of all generated action chains. In the present example, the value of each statistic is two. Under the heading % EPISODE INTEGRATION, we can see the percentage of virtual episode constructs which are combined in the longest generated action chain. In our example, we have obtained a very small value: 4.3 percent. Although this result may seem somewhat disconcerting at first, it is easily explained. Because our example episodes reused the same segment of text several times, it follows that the possibilities for building extended action chains must be severely limited. We are concerned with a narrative text that has been read as having ten events, on which small basis a considerable number of possible episodes can be formed, but the text remains episodic in the non-technical sense of the word. When we consider the opposite situation, it becomes even more apparent that the episodicity of a text is related to the number of event, episode, and action constructs which it contains. Texts with a low level of event integration but a high potential for action synthesis are weak in action, not because they are episodic like the above example, but simply because they lack available episodes. If a text has low levels of both event integration and action synthesis, it will lack connectable event constructs and therefore the raw material of a proper story in the first place. The final case consists of texts which when read produce a high level of event integration and a high level of virtual episode integration; the most extreme example of such a text is one in which all the constituent events are arranged in an uninterrupted succession which consists of and then connectors. We are now in a position to abandon the interim terms ‘action potential’ and ‘episodicity’. Each of the four representative text types described above can be quantitatively represented as the product of the event integration and episode integration percentage values. In the demonstration file, this value, which we shall refer to as the virtual action product, is 54.117 * 4.347 = 235.294. The EpiTest window displays this figure under the final heading in the top right-hand panel. Episode constructs support an almost unlimited range of possibilities; the action product is a measure of how much of this potential can be synthesized logically in a global action construct (the latter is, as noted above, defined in very broad formal terms). It should be emphasized again here that our final quantification of the action product is in no way meant to be an objective measure of the coherence of a given narrative text. This is obvious not least because in itself a number like 235.294 says absolutely nothing. Is 235.294 a lot? Is it a little? The question can be answered only in differential terms, never

252

Part 2 The Computer-Aided Analysis of Narrated Action

absolutely. For the time being, then, this modest number is nothing more than the result of a cognitive formula which is evaluated when we form episode and action constructs during the comprehension of texts and the reading of action. It is a formula that involves many different variables and dynamic linking processes of which we can be sure that only a fraction are known to us. Nevertheless, it is a fact that we can obtain results perfectly compatible with human intuition by applying our theoretical arguments and the practical model which culminates in the action product which correlates real (empirically identified) event constructs and algorithmically derived virtual episode and action constructs. The reader will see this for himself in the final part of our study, and if that fails to convince him, he can experiment further with EventParser and EpiTest on data of his own choosing. He might, for example, follow the author and decide to test the programs by processing an extract from Lewis Caroll’s Alice in Wonderland which was deliberately encoded to produce a narrative with a single uninterrupted strand. This extract represents the opposite of the above extracts from the Fairy Tale. Because of its typically episodic combination of high event integration and low episode integration, the Fairy Tale example yielded an action product of 235.294. Compare this with the results of processing the Alice.esf file, which yields an action product that is almost ten times higher than that of Matrix.esf. Like the latter, Alice.esf contains just ten event constructs; but it permits the formation of an action chain which contains a maximum of nineteen isochronally joined episodes.33

33

Alice.esf is included with the downloadable files which accompany the book.

2.6 Practical Analysis Using EpiTest In the preceding pages, we have described a first, modest attempt to put our theory into practice, and the reader may well be eager to proceed at once to part 3 and the full-scale application of our software to a real text. However, it is well to remember that our model and methods are not without their limits. These weaknesses are particularly obvious when we come to deal with the anisochronous event types. Readers can process the semantics of anisochronous event narratives effortlessly and intuitively because they use the framework of a prespecified comprehensive and wellstructured knowledge context. This is very different from the incrementally assembled set of axioms which EventParser stores in its *.sbf files. The example episode for eaniso2 (see above) should illustrate the difficulties which can result. Assuming that narration (and therefore reading) is strictly sequential and does not employ embedding techniques, the prehistory supplied by event1 (‘Long ago she had been assured that this phenomenon was possible’) must be encountered after event2 (when the snake devours the gold and becomes transparent and luminous). The order in which the acts of reception (i.e. event construction) take place is irrelevant in our theory and the EpiTest algorithm with which we are concerned. So far, so good. Now, event1 and event2 must have explicitly connectable predicates if they are to be linked semantically by EpiTest. The practical problem is that recipients tend to provide the events with such predicates only if they have mentally arranged the analeptically narrated events in the order of the fictive ordo naturalis beforehand. The eaniso1 example encourages this, to be sure, but the bracketing and assistance it provides are the exception rather than the rule. Narratologists typically treat the practical receptive necessity of returning to the ordo naturalis of the fabula as evidence that the ability to reconstruct a quasi-natural chronological order at the diegetic level is a necessary prerequisite for identifying a consistent action order. This may well be reasonable from the perspective of an intentional or causal definition of the concept of action, but things are different if event, episode, and action are defined semiotically as dynamic constructs of reception. Every state of affairs that we encounter in the empirical or fictional world is not just projectively configured in anticipation of a possible change which it could undergo in the future; it is also the base of a retrospective semantic

254

Part 2 The Computer-Aided Analysis of Narrated Action

protension which can make it the result of just such a change which has already taken place. Between the theoretical possibility of this semantically preconfigured construct and its concrete instantiation there lie many worlds, a marvellously comprehensive knowledge of the semantic relations which can accumulate through the symbolic representation of the phenomena of at least one possible world. In its present state, EpiTest has no access to the complex heuristic world knowledge of the semantic ordo naturalis which is constantly available to natural readers. Until this shortcoming is rectified, the rare occasions when EpiTest does combine anisochronous events into episodes and actions will be nothing more than flukes. In order to overcome this obvious weakness in EpiTest and the theory behind it, we would have to commit ourselves to using concepts from cognitive theory such as the script and the frame. These concepts postulate a schematically represented mental knowledge of standardized situational and pragmatic patterns which is constantly accessible to recipients; the knowledge comes into its own when the recipients need a heuristics with which to explain protensionally (anisochronally) arranged semantic event terms.34 In contrast to this ideal solution, we have employed something of a stopgap measure in the practical application of the EventParser and EpiTest programs to our example texts, the six individual narratives in Goethe’s Conversations of German Refugees. A maximum of ten pairs of opposite terms have been formulated in advance for each of EventParser’s nine heuristic categories. Taken together, these terms are an intuitively plausible model—albeit an admittedly crude one—of the knowledge context which I believe to be of crucial importance in the identification of anisochronous episodes. The results of the analysis will be described in the third and final part of the book after a discussion of the critical background against which our narratological study in literary computing makes sense. Only one point remains to be made before we can continue. There are at least four different combinations of text and reader in which EventParser and EpiTest can be used to mark up and combinatorially analyse record files. Together, the four combinations represent a typology of action analysis; each of them concerns a different methodological problem: (a)

34

One text and one reader. In this relatively simple situation, we are usually concerned with differential analysis of the components of a single more complex text. A typical topic of study would be: what are the differences between the action products of a set of subnarratives embedded in a single overall narrative? A concise overview of this subject can be found in Schnotz (2000:501–3); see also Christmann and Groeben (1999). Access to frames and scripts plays a decisive role in the cognitivist redefinition of narratological models suggested in Herman 2002.

2.6 Practical Analysis Using EpiTest

(b)

(c)

(d)

255

One text and n readers. Research of this type usually analyses statistical patterns in how different readers interpret a single text. A possible research topic might be: given a single text and two groups of readers, how and why does the mean action product of the readers in one group one differ from that of the readers in the other? N texts and one reader. In this case, we have a single reader whose individual way of reading action provides a means of analysing historical developments in a genre or the works of a given author. It is basically a variant of (a); a typical theme might be: how do differences in action product relate to the early and late novels of a particular author? N texts and n readers. This case involves the differential analysis of a text sample using multiple readers. The approach might be most profitable in a study where action is just one aspect of a broader empirical analysis of the style of an individual author or movement.

As can be seen, the four different types share the use of differential analysis to explore empirical data. This, the hallmark of empirical research, sets our computational approach to action narratology apart from other approaches, which adopt the methods of subjective criticism rather than objective investigation and are modelled on traditional hermeneutic textual exegesis, no matter how deceptively scientific their elaborate formal systems may seem. The reader will, we hope, be struck by the similarity between this empirical aspect of our methodology and the conclusions of our earlier philosophical discussion, where we argued that it is impossible to find an essentialist definition of action, be the latter practical or aesthetic. There is no point in asking what action, even literary action, actually is. But it is certainly worthwhile asking how and why we believe that a certain amount of action is present in a given narrative, and it is these questions that we shall consider in our case study of type (a) in part 3.

Part 3 An Experiment in Action: Conversations of German Refugees In seeking to penetrate Kant’s philosophy, or at least apply it as well as I could, I often got the impression that this good man had a roguishly ironic way of working: at times he seemed determined to put the narrowest limits on our ability to know things, and at times, with a casual gesture, he pointed beyond the limits he himself had set. He had no doubt observed man’s precocious and cocky way of making smug, hurried, thoughtless pronouncements based on one or two facts, of rushing to hasty conclusions by trying to impose on the objective world some notion that passes through one’s head. Thus our master limits his thinking person to a reflective, discursive faculty of judgment and absolutely forbids him one which is determinative. But then, after he has succeeded in driving us to the wall, to the verge of despair in fact, he makes the most liberal statements, and leaves it to us to decide how to enjoy the freedom this allows us. “Judgement Through Intuitive Perception” (Goethe 1988:31)

3.1 Critical Approaches to the Conversations Had it not been for the fame of their author, the Conversations of German Refugees would probably have faded into literary obscurity soon after they were first published in 1795. Most contemporary readers viewed the work as one of Goethe’s less accomplished creations, and many dismissed it as a complete failure with respect to both form and content. Goethe, it seemed, had completely disappointed his audience by presenting them with an unconvincing attempt at uniting several morally dubious French and Italian novellas in a single work, adding merely two genuine creations of his own and embedding the entire collection in a frame narrative. One of the few voices of encouragement came from Schlegel; in 1796, one year after the serialization of the Conversations in the Horen, he argued forcefully that we ‘cannot fail to note the accomplished and lively dramatic skill, right down to the smallest of the small stories that are recounted,’ (my translation) with which the embedded narratives reflect the formal principle of the group discourse.1 The discourse is set in the fictional conversation situation created by the frame narrative, which introduces the characters as members of a circle of noble storytellers who have fled the events of the French Revolution. Even if few shared the enthusiasm with which Schlegel praised the Conversations, the Fairy Tale with which the series ends did attract considerably more interest and appreciation than the work as a whole. Indeed, the devotees of the Fairy Tale were not dissuaded even when Goethe himself expressed his amusement at their insistence on searching for a definitive interpretation of the story. The fascination with the Fairy Tale which overcame Goethe’s contemporaries was to mesmerize later critics in much the same way. Take, for example, Lucerna’s comprehensive interpretation of the Fairy Tale in terms of the symbolism of natural philosophy (Lucerna 1910). Although it undoubtedly has its benefits, her contribution is of most interest to us because it marks the beginning of a long line of studies dominated by the ‘interpretive compulsion’ (‘Auslegungssucht’; my translation) which Goethe had prophetically ridiculed more than a century 1

See Schlegel’s review of the first volume of the Horen in the Jenaer Allg. Literatur-Zeitung (January 1796). The quotation here is from the reprint in Goethe (1981, 6:608). A representative selection of contemporary responses to the Conversations can be found in Voßkamp and Jaumann (1992:1523–43).

260

Part 3 An Experiment in Action

earlier.2 Only in 1971 did Fink provide the obsessed critics with a lasting cure for their affliction, before Brown (1975) finally returned the Fairy Tale to its proper place as an integral part of the Conversations after almost two centuries of isolated existence at the hands of symbolist critics.3 But we should not be too hasty to complete our survey of the critical tradition. Stepping back to 1855, we read of the Conversations as a whole that ‘the novellas have a polished and agreeable form, but the actual content of the stories, which frequently involves the occult and supernatural material that we would expect in such a period of upheaval, is little more than trivial; the Fairy Tale in particular we can gladly leave to its interpreters’ (Gottschall 1855, quoted in Baumgartner 1886:144; my translation). Gottschall’s terse and uncompromising appraisal is echoed by Baumgartner, whose sarcastic opinion is typical of the attitude to the Conversations that prevailed into the second decade of the twentieth century: That was all that the ‘first’ German writer was able to contribute to the first volume (1795) of the ‘first’ German journal. Indeed, he could have continued to pour ‘precious’ French novellas, crude memoirs, and low-life literature into the Horen for years, no doubt adding a few personal confessions to boot. But even that would have been too much effort for him to put into such an insignificant project—in his contribution to the 1796 volume, he contented himself with a bit of translation. Baumgartner 1886:144 (my translation) 2

3

The pejorative phrase ‘interpretive compulsion’ was actually coined by Schiller, who uses it in a letter to Goethe (29 November 1794; Voßkamp and Jaumann 1992:1509) while the latter was working on the Conversations. The letter probably alludes to an earlier conversation with Goethe, and it should be remembered that, as an editor, Schiller had a vested interest in public reaction and must have welcomed any work which tapped into the interpretive compulsion of his readership. Indeed, his jestingly ironic words suggest that he was not immune to it himself. The letter in question is important for our argument and will be discussed in detail below. This did not stop Mommsen (1982) from returning to the symbolism of the Fairy Tale. Mommsen intends to provide a concrete critical foundation for what is a somewhat far-fetched, if attractive, thesis about the biography of the two great German authors. Mommsen’s theory is that the Fairy Tale is full of allusions to Schiller’s aesthetic theory because Goethe wrote it to help Schiller out of a creative crisis brought on by excessive theoretical reflection. However, the real background to the Conversations is probably far more prosaic. Gaier (1987) gives a convincing and finely detailed account of the diplomacy and flattery, sometimes bordering on self-humiliation, which Schiller employed in order to secure Goethe’s co-operation. Gaier also describes the amount of philosophical misunderstanding which can be observed in the exchanges between the two authors. Baumgartner, a committed supporter of Schiller, was well aware of the situation a century earlier and paid special attention to the pressing material circumstances that made Schiller call on Goethe for assistance—which Schiller repaid by presenting Goethe with fait accomplis such as publishing the entire Fairy Tale in a single issue of the Horen against the author’s wishes. This is one particular example of the manipulative and tactfully persuasive pressure which Schiller was able to apply to Goethe (Baumgartner 1886:135ff.).

3.1 Critical Approaches to the Conversations

261

Gundolf (1920) was the first critic to interpret the apparent aesthetic deficiency of the Conversations as an authorially intended effect. This allows us to reconcile the disjointed nature of the work with the idea that Goethe was a writer of meaningful texts. Gundolf’s argument returns to the text itself and proposes a new subdivision of its components: The introduction to the conversations must be distinguished from the collection of novellas which it frames.…The introduction is a political frame in which a series of poetic narratives are embedded, and, most importantly, the frame was present before the picture appeared inside it.4 When they were conceived, the Conversations were intended to elucidate political problems. Only later, when Goethe’s interest turned away from political themes, primarily because of Schiller, did the plan change to writing a collection of novellas. As a result, the introduction was left stranded as a political frame narrative. The political preamble, accompanied as it is by the apolitical novellas, no longer looks like the introduction to a political discussion. Instead, it seems to be a warning against such a discussion…politics has shifted from providing the motivation for the discourse to being a technical device for employing embedded narration. Gundolf 1920:489 (my translation and emphasis)

Gundolf’s theory is that the Revolution theme in the frame narrative is functionally equivalent to the plague theme in Boccaccio’s Decameron. He argues that in both cases the frame demarcates a zone of stable atemporal vision which is sealed off from the harsh reality of contemporary events by the power of art. While the novellas move away from the frame narrative during the discourse of the Conversations, its textual themes return, albeit under a different set of normative conditions, in the Fairy Tale at the end. Furthermore, Gundolf believes that the genre of the fairytale is the ‘most irresponsible work of the imagination.’5 In accordance with this interpretation of the genre as a symptom of escapism, Gundolf dismisses the Fairy Tale as an unsuccessful attempt by Goethe to reject reality once and for all and escape from the ravages of history for good: ‘Thus, at those 4

5

This particular claim does not stand up to closer scrutiny. From what we know of the discussions between Schiller and Goethe during the conception of the Conversations, it appears that the Prokurator novella was the first material that Goethe proposed including in the narrative—see the relevant passage in Schiller’s letter to Goethe of 28 October 1794: ‘So I would remind you of your idea of reworking Boccaccio’s “Tale of the Honourable Attorney”’ (‘so erinnere ich Sie an Ihre Idee, die “Geschichte des ehrlichen Prokurators” aus dem Boccaz zu bearbeiten’ Voßkamp and Jaumann 1992: 1508; my translation). Although factually erroneous, Schiller’s belief that the Prokurator novella belonged to the quintessential Romance novella cycle is conceptually important because it may well have given rise to the idea of embedding several novellas in a frame narrative. ‘Unverantwortlichste Produkt der Einbildungskraft’ (Gundolf 1920:489; my translation). Note that the phrase ‘work of the imagination’ is itself taken from the Conversations.

262

Part 3 An Experiment in Action

times when Goethe felt that the Revolution was a bad dream, he would irresponsibly try to relate and interpret it in fairytale works’ (Gundolf 1920:492; my translation).6 This argument came to characterize an entire school of criticism which saw the Conversations as the quietistic reaction of an escapist Goethe to the events of the French Revolution. The views of this tradition were still influential as late as 1979, when Geißler declared that ‘the unity of the Conversations lies in their lack of meaning, which is itself produced by the escapism and depoliticization needed to preserve a fellowship which subjective opinion would otherwise threaten to destroy’ (Geißler 1979:41; my translation).7 Even today, it is still fashionable to attack the work for its apparent lack of political commitment. However, as early as 1927, Tornius introduced a second, more evenhanded line of interpretation in his contribution to Goethes Leben, the biography of Goethe founded by Wilhelm Bode. Tornius notes a parallel between the moral of the narratives embedded in the Conversations and the renunciation motif in Faust. We can see, he argues, a ‘fairly conspicuous emphasis on moral qualities’ in the text, which becomes most obvious in the Prokurator novella (Tornius 1927:35; my translation.)8 Tornius interprets this as a sign of the influence of Schiller or the principles of Kantian morality as seen from Schiller’s perspective; he suggests that this theme is then reworked artistically in the Conversations, particularly in the dialogues contained in the frame narrative (Tornius 1927:35).9 Tornius’s theory marks the first step on the way to the eventual fusion of the moral and aesthetic approaches to interpreting the Conversations in Raabe (1939). There are two key achievements in Raabe’s combination of the two different paradigms. First, he draws the frame narrative, the embedded narratives, and the Fairy Tale together under a single general theme which unites the individual textual components of the Conversations. This theme is the defeat of the monstrous, that which the human soul cannot grasp and control, by love. The monstrous can appear in social form (as a violation 6

7

8

9

Landauer comes to a similar conclusion and sees the Fairy Tale as ‘a work in which reality is transformed but never lost sight of’ (‘ein Stück verwandelte und wohlgeborgene Wirklichkeit’ Landauer 1921, quoted in Voßkamp and Jaumann 1992:1539; my translation). Beyer (1981) is even more steeped in political correctness when he observes with regret that Goethe’s work fails even to reach the level of Brecht’s Conversations among Exiles. English translations of the Conversations generally refer to the Prokurator as ‘lawyer’ or ‘attorney’. In his treatment of the Fairy Tale, Tornius was one of the first critics to warn expressly against engaging in yet more disputes about interpretation: ‘There is nothing more than idle speculation to be gained by engaging in arguments about whether one particular reading or another corresponds most to the author’s intentions’ (Tornius 1927:99; my translation).

3.1 Critical Approaches to the Conversations

263

of social rules or, in extreme cases, as revolution) and in aesthetic form (as a non-classical, ill-proportioned work of art). The second outcome of Raabe’s synthesis of the moral and aesthetic approaches is that it makes us aware of a logical principle which underlies each stage in the progression of the narrative as a whole. For the first time, the arrangement of the narratives in the Conversations can be seen as representing an ordered development from ‘the “monstrous” to “education”’ (Raabe 1939:27; my translation). Furthermore, Raabe makes the inspired suggestion that the disastrous failure of the educational project on the first storytelling evening was artistically designed with a deliberate purpose in mind: it is meant to let the listeners inside the story and the real readers outside it indulge their aesthetically and morally reprehensible fascination with the monstrous. Only in the Prokurator and Ferdinand novellas of the second storytelling session do we encounter the more desirable progress which overcomes the monstrous with the moral and the moral with the religious. In the final stage of Raabe’s interpretation, the Fairy Tale acquires the function of a medium for textual self-reflection—it neutralizes the monstrous by means of irony and overcomes the moralizing of the moral tales with its reference to the redemptive power of love.10 10

Raabe’s reference to signals of irony anticipates Rindisbacher (1994), who argues that the Conversations construe the novella as an ironic genre in accordance with Schlegel’s definition. The idea that the Conversations are the first real German contribution to the genre of the novella is widespread (e.g. Kunz 1968; Müller 1969; Jacobs 1981), but, like the genre term itself, this view has met with considerable disapproval, particularly among Anglo-American critics—see Ellis (1986), who considers Goethe’s Novelle, and Weing (1998), who places particular emphasis on the link to Aristotle. Henel (1985) proposes a new definition of the genre by drawing on the distinction between plot and story. According to this definition, the novella is the first non-lyric genre in which paradigmatic and discursive ordering principles take priority over realistic mimesis. In terms of genre history, this thesis may appear somewhat tenuous (the interpretation in Käuser 1990 is a representative criticism), but it could nonetheless provide the starting point for an interesting contrastive analysis of the narratives (i.e. the stories recounted on the first evening) and the novellas in Goethe’s Conversations. For example, a particularly high degree of freedom in chronological ordering can be noted in Goethe’s own creation, the Ferdinand novella, where it is obviously intended to build up to what has since come to be known as the ‘turning point’ (‘Wendepunkt’). This is most apparent in the introductory narratorial commentary, ‘I will skip over many scenes from his childhood and relate only one occurrence that illuminates his whole character and that marked a distinct epoch in his life’ (Goethe 1989:56). The word ‘epoch’ may not appear to be particularly special, but, nonetheless, it has its own unique history. Rennie (1996) shows how Goethe distinguishes between chronos, which is governed by the laws of nature as an incoherent (or at least unknowably coherent) sequence of natural events, and kairos, which is governed by the destiny under whose sway human existence authors itself as forward- and backward-looking ‘moments’ (‘Momente’). It is such ‘moments’ that make up ‘epochs’ by marking the biographical caesuras with whose help a limitless chronological sequence of existential ‘instances’ (‘Augenblicke’) can be idealized, above all narratively, as a set of coherently organized

264

Part 3 An Experiment in Action

Raabe’s interpretation founded an interpretive tradition which continues to be developed by contemporary critics and whose evolution is too rich and varied for us to consider in detail here. Taken in their broadest sense, his ideas provide the basis for the work of Jürgens (1956), Fricke (1964), Müller (1969), Fink (1971), Stammen (1990), Niekerk (1995), Bauschinger (1985, 1997), and Hoermann (1996). Geißler (1979) and Beyer (1981) can also be included in this list, although they are characterized by a conscientious re-evaluation of the Conversations as apolitical reactionary literature. There are also several informative studies which provide essential background information about the history and adaptation of the sources: Kraft (1935) on the Bassompierre narrative, Brandt (1983) on the Prokurator novella, and Ziegler (1984) and Hoffmann (1985) on the Antonelli narrative. However, we must now turn to two different lines of criticism which are of more direct relevance to our study than those described above. First, we shall consider the descriptive (as opposed to interpretive) analysis of the compositional logic which underlies the narrative of the Conversations. This approach was founded in the twentieth century by Goldstein (1906) and taken up again by Lockemann (1956). We shall then discuss the work of critics who have examined the philosophy of the Conversations; the individual emphasis in these studies varies from aesthetics to the philosophy of history to epistemology, but all aim to explain the principles in terms of which the Conversations should be properly understood.11 The Compositional Logic of Goethe’s Conversations Lockemann (1956) has described a systematic classification of the different ways in which a cycle of novellas can be embedded in a frame narrative. His typology rests on two key oppositions. The first of these is the dichotomy between functional frames (‘Zweckrahmen’) and legitimizing frames (‘legitimierende Rahmen’). Lockemann describes functional frames as typical of novella cycles, where they unite the individual embedded narratives as a functional whole by illustrating the overall narrative intention

11

temporal spaces. Bearing this in mind, it is perfectly possible to link the two moral tales which Goethe based on Romance tradition with Henel’s thesis of the specificity of novellas based on the biographically epochal ‘unheard-of event’ (‘unerhörtes Ereignis’). This brief description of the main trends in Conversations criticism is perhaps most useful in conjunction with other such summaries; see, for example, the relevant passages in Ziolkowski (1958:57ff.); Bauschinger (1997:244–48), the most comprehensive summary to date; and, particularly useful, Dammann (1990:1–5). To my knowledge, Hoffmann (1985:106–9) is the only study which treats the critical tradition of the nineteenth and early twentieth centuries in any real detail.

3.1 Critical Approaches to the Conversations

265

which lies behind them. Legitimizing frames, on the other hand, do not illustrate the purpose of the narration; they are designed instead to reinforce the credibility of what is narrated. The function/legitimization dichotomy is joined by a second dichotomy between closed and open frames; this opposition is concerned less with the internal unity of the fictional narrative and more with the way that the narrative guides the external reader who interprets it. Lockemann describes a closed frame as ‘a frame which is sealed off from reality on every side and encourages the reader to place himself in a preformed context, such as the social circle of the narrative’s audience’ (Lockemann 1956:209; my translation). Open frames, on the other hand, allow the reader to step outside the fictive reception situation and emancipate himself from it through criticism and reflection. We shall not concern ourselves here with the validity of Lockemann’s classification of frame types; what is important, however, is the first example text with which he illustrates it: the Conversations. Goethe’s text provides such a rewarding object of study for the typology because of Lockemann’s formal interpretation of the text, which is highly original and has no clear theoretical antecedents.12 Although he never explicitly states that he is adopting this formal interpretation, Lockemann must have realized that the Conversations contain examples of many combinations covered by his typology. While Raabe’s influential study considers the progression in the content of the storytelling sessions, Lockemann’s theory is the first to draw attention, albeit in passing, to the formal progression in the Conversations. The progression starts with the closed functional frame of the overall narrative sequence and ends with the open functional frame (which Lockemann might have called an open null frame) of the Fairy Tale. Although the Conversations are followed by further sample texts which illustrate Lockemann’s typology, there is no doubt that he could easily have found more to say about his first example if he had examined it in more detail. In particular, there is ample material for discussion in the Klopfgeist (Noisy Spirit) narrative, where the main narrator character of the frame departs from his normal role by becoming a listener and the source of an ironic commentary. In places, the Klopfgeist story descends to the level of the most banal frame type of all, a fragile legitimizing single frame in which a personal narrator is forced to begin by reeling off assurances about the credibility of what is narrated (in this particular case, it just so happens that the narrator is Lockemann’s fictional namesake 12

This judgement may need to be qualified on the basis of a closer examination, not possible in the context of the present study, of the older literature, particularly Goldstein (1906).

266

Part 3 An Experiment in Action

Fritz…). As if that were not enough, we are also dealing with a deus ex machina that produces empirical evidence with suspicious ease and acts out acoustic effects to its heart’s desire. One might say that Lockemann is caught unawares by this unexpected type of narrative frame, which is simultaneously legitimizing and an ironic caricature of openness. Lockemann is thus the first critic of the Conversations to employ a set of formal criteria to analyse the principles behind the construction of the overall text. With good reason his work quickly became a precedent for further studies. Consider, for example, Ziolkowski (1958), which followed two years later. Ziolkowski’s primary aim is to illustrate a hypothesis of genre history and show that Goethe takes the novella form exemplified by his Romance models and attempts, ultimately without success, to adapt it to the German language during the series of narratives in the Conversations. Ziolkowski’s debt to Lockemann lies in the dependence of his argument on a preliminary formal description of the text, which he divides into three formal units: two ‘fantastic anecdotes,’ two ‘love anecdotes,’ and two ‘moral tales’ (Ziolkowski 1958:70).13 Lockemann’s influence rescues Fricke (1964) from the fate of becoming just another reworking of the interpretation which is exemplified in Raabe (1939) and Jürgens (1956), developed more recently in Bräutigam (1977) and Mommsen (1982), but actually rooted firmly in the nineteenth century. This traditional interpretation sees the Conversations against the background of Kant’s ideas and, depending on the critic in question, Goethe’s praise or criticism of Schiller. The intention of the text is then to depict an incremental moral development which should be striven for during the narrative and receptive process. The innovative aspect of Fricke’s contribution to this tradition lies in his treatment of the opening narrative. Fricke evaluates the stylistic caesura, which has often met with disapproval elsewhere, as an intentional artistic formal feature rather than a deficiency.14 The caesura splits the Antonelli narrative into a veritable ‘little novella’ (‘kleine Novelle’) in its first part and a second part which ‘is not far removed from satire and unconcealed mockery’ (Fricke 1964: 278; my translation). Fricke argues that the partition is actually the result of a well thought out rhetorical decision—the main purpose of the first storytelling session is to provide an ironic exposition of thematic material, so it is perfectly appropriate that it contains exemplary illustrations 13

14

Ziolkowski justifies (again, formally) his exclusion of the Fairy Tale from this structure by arguing that it belongs to a different genre. Ziolkowski (1958:65f.) had previously examined the Antonelli narrative in more detail and decided to use the term ‘exposition’ to refer to it, but only in a self-contained sense that applies to the themes of the individual narrative itself, which like Pongs he includes in the minor category of a ‘fantastic anecdote.’

3.1 Critical Approaches to the Conversations

267

of the possibilities and conditions for both the success and the failure of the aesthetic and educational project (Fricke 1964:278). Hoffmann (1985) subsequently counters this thesis with several convincing arguments relating to the sources of the Conversations.15 Nonetheless, Fricke’s choice of method proves that a formal description of the text can supply the interpretive paradigm which we need if we are to treat the apparent aesthetic flaws in the Conversations as a deliberate feature of the text. In this light, the true significance of Fricke’s choice of title becomes clear—not without reason is his essay a study of the ‘Sinn und Form’, the meaning and form of the Conversations. The formal mode of interpretation applied by Lockemann, Ziolkowski, and Fricke sets their work apart from other similar studies. The same formal factors, however, play a large part in the major disadvantage shared by all three approaches: the decision to set the Fairy Tale apart from the rest of the Conversations.16 This unmistakable gap in the theory is not repaired until Brown (1975), who analyses the role of cyclical form in the Conversations. Like many others, Brown identifies a thetic content in the overall text; she interprets it as celebrating the reconciliation of a divided society by means of individual self-sacrifice and commitment to the common cause. She backs up her hypothesis of cyclical, or spiral, progression in content with a new formal analysis which is the first of its kind to include the Fairy Tale. Each narrative unit occupies a relatively autonomous position in the overall structure of the Conversations, and Brown describes the order of these self-contained textual segments as follows: the introductory expositional frame narrative; three narratives on 15

16

Hoffmann proves that the Correspondance littéraire contains the text which Goethe in all probability used as his source and performs a precise textual comparison which goes far beyond the preliminary work in Ziolkowski (1958). Among Hoffmann’s own conclusions, one in particular is worth noting here: while the portrayal of supernatural events constitutes only a third of the original text, Goethe expands it to make up almost half of his version (Hoffmann 1985:127). The wider programme of Hoffmann’s study is to return the Antonelli narrative to critical favour; he argues that in Goethe’s skilful hands it is transformed from a recounted individual case into an exemplary event (Hoffmann 1985: 142). See also Söring (1981), who asserts the symptomatological socio-historical dignity of the ghost stories, and von Wilpert (1991), who tries, with debatable success, to read the Antonelli novella as a political narrative and interpret its formal break as a symptom of the unbridgeable divide between the ancien régime and the post-Revolutionary era in social history (von Wilpert 1991:200). Ziegler (1984) will be helpful if the reader prefers to consult a more balanced, descriptive survey of the sources. Müller (1969) lies at a tangent to our differentiation between studies of form and content. He hopes to provide a positive re-evaluation of the Conversations in their contemporary context and also bases his argument on an initial theoretical segmentation; he sees the narrative sequence as a series of two ‘morally indifferent ghost stories,’ two ‘chronicle anecdotes,’ and two ‘didactic moral tales,’ all crowned by the Fairy Tale (Müller 1969: 159).

268

Part 3 An Experiment in Action

the first evening with negative connotations for the aesthetics of production and reception; two positive ‘poetic efforts’ from the main narrating character (the Abbé) on the second evening; and finally the Fairy Tale. The Fairy Tale not only intensifies the symbolic nature of the content but also, and this must be Brown’s most significant insight, takes the frame triad (Carl, the Abbé, and Luise) and reflects it ironically in a new triad at the figural level (the foolish giant, the insightful old man with the lamp, and the simple old man with the black hand). As a result, Brown can show, formally at least, that the narrative process does return to its starting point, albeit by being dialectically resolved as symbolism. She concludes: ‘Thus the cycle insists upon a much tighter unity than the critics have accorded it.…It must be read with the same attention to relationships between parts as a four-line poem, for all the pieces reflect upon and correct one another.’(Brown 1975:3217) Dammann (1990) stands firmly in Lockemann’s tradition of form-based criticism yet takes this approach to a far more ambitious methodological level in that he is the first critic to approach the Conversations from the perspective of established narratological or semiotic concepts. Dammann borrows the distinction between reinforced and non-reinforced narratives from Greimas, whose metaphor of reinforcement (‘armature’) he explains as follows: Narratives with reinforcement are…those in which the narrated action is subjected to a high degree of structuring in the form of equivalences…narratives without reinforcement are those in which this is not the case. From the perspective of the producer, a general lack of structuring results in a story without reinforcement having two kinds of openness: the narrated action can begin and end with relative arbitrariness, and it can be relatively easily filled with elements (persons in the ‘characters’ parameter, episodes in the ‘action’ parameter, and objects in the ‘locality’ parameter). A complete absence of structuring leads to a third, additional kind of openness, this time on the part of the recipient: a story without reinforcement offers the reader the chance to engage with the narrated world by examining and discussing it. Dammann 1990:9 (my translation)

17

This formal approach to the meaning of the Fairy Tale has recently been taken further in Hoermann’s highly original study (Hoermann 1996). He too sees the Fairy Tale text as a reflection of the general situation in the Conversations, with the first part corresponding to the frame narrative and the second to the embedded exemplary novellas. Then, Hoermann argues, Goethe decides against returning to the wider context of the frame which encompasses the Fairy Tale: ‘Goethe’s pedagogical realism has evolved into a kind of ironic idealism, with the aesthetic reality of the comic performance being the only surviving element of the pedagogical demonstration, which structurally has, in fact, yielded a “play-within-a-tale-within-a-storytelling circle”’ (Hoermann 1996:87).

3.1 Critical Approaches to the Conversations

269

For Dammann, the reinforced (structurally supported) and non-reinforced (structurally unsupported) types of narration define the frame of reference inside which narration has positioned itself since the Enlightenment. This is not the place to discuss Dammann’s lucid explanation of the role of genre history and the edifying textual examples with which he illustrates his thesis. Instead, we shall concentrate on his preliminary conclusion, which is of crucial importance for our detailed formal analysis of the Conversations. Dammann describes how a pronounced tendency to use structurally unsupported narration begins to develop during the eighteenth century, when it was particularly widespread in moral tales published with a clear didactic moralizing intent. The concomitant openness is a logical necessity of such didactic narratives, which deliberately operate using the frame fiction of a narrating and receiving community which improves its own education by discursively considering the content of the text.18 A real audience will take the call for didactic moral reflection seriously if and only if the fictional reality is related in the mode of a quasi-empirical representation in which no authorial explanations of meaning is present. Dammann argues that the individual narratives in the Conversations are arranged to challenge the predominant trend in genre history by announcing a commitment to structurally supported narration. His formal analysis divides the sequence of constituent narratives as follows: unstructured narration on the first evening (the Antonelli, Klopfgeist, and Bassompierre narratives); structured narration on the second evening (the Prokurator and Ferdinand novellas); and hyperstructured narration on the third evening (the Fairy Tale). The action model of the Fairy Tale is derived from the courtly historical novel of the Baroque period, where the fictional action is related word for word as a set of seamlessly interlocking narrative foci (figures and places). The result is that it is generally possible to dispense with the anaphora and temporal deictics that would otherwise establish coherence with an extradiegetical complex of abstract focus bearers such as time, space, theme, and so on. Dammann explains the ordering principle of the Conversations with a rigorous, insightful study of the various narratives and novellas, concluding with the Fairy Tale itself. However, one aspect of the otherwise convincing argument is conspicuously problematic—while Dammann’s findings are in themselves perfectly acceptable, the intentionality from which he derives them is not. Suppose we insist that the formal principle behind Goethe’s narrative cycle is a literary and ‘overtly partisan attempt’ 18

Recall Goethe’s admiration for the work of Justus Möser; see in particular Goethe’s 1823 essay “Justus Möser: Superstition and Poetry” in his Essays on Art and Literature (Goethe 1988:205ff.).

270

Part 3 An Experiment in Action

(Dammann 1990:20; my translation) to restore the damaged credibility of reinforced narration (i.e. narration whose aesthetics and semiotics have been preconfigured by the author). As Dammann himself notes with some concern, the consequences of such an interpretation are anything but insignificant: ‘Goethe’s essay about what prose narratives should be at the end of the eighteenth century would appear to plunge its author into what, from the perspective of genre history, is a disconcertingly reactionary role’ (Dammann 1990:20; my translation). Dammann attempts to deal with this problem in a finely detailed analysis of the treatment of traditional motifs in the Fairy Tale and the Ferdinand novella. He argues that the impression that both narratives simply recycle inherited material is misleading and superficial; in reality, Goethe updates his sources to reflect contemporary Enlightenment thought. Thus, the triad which we find in Ferdinand is indeed a feature familiar from the typical erotic triangle of the Schwank novella—but Ferdinand is modernized by the modern subjectivity of its eponymous central figure.19 And the Fairy Tale climaxes in general contentment and a static happy ending (just what feudalism would need to ensure its continued survival for the foreseeable future)—but Goethe ends the story on the more dynamic note of an open-ended process of continual exchange and interaction. Despite attempts such as the above to handle the difficulties which result, the weak point of Dammann’s otherwise consistent and persuasive study still lies in his attribution of somewhat improbable intentions to Goethe. For example, he argues that Goethe intended to make a monothetic statement of poetic theory which affirms pre-Enlightenment narrative traditions and essentially renounces the principles of Goethe’s own creative work. Such an interpretation is indeed possible, but only if two conditions are fulfilled: first, Goethe must have deliberately set himself the objective of engaging in a self-reflexive discussion of genre theory; second, he must have intended to resolve the resultant theoretical debate by means of an ironic metacommentary. It is not hard to imagine Schiller or the Romantics engaging in ironic, critical self-reflection at this abstract level of poetics, but is it really likely that Goethe, who tended to steer clear of debates on aesthetic theory, would have done so? Dammann’s semiotic and essentially narratological approach marks the culmination of formal criticism of the Conversations to date. And, 19

By modernizing the subject in the Ferdinand novella, Goethe violates the traditional rules of his adapted genre. This forms the first part of Dammann’s restoration hypothesis, for which some additional support can be found in Neumann (1984). Neumann uses Foucault’s concept of discourse to support his thesis concerning the conditions which defined the subject in the late eighteenth century. Niekerk (1995) is a recent comprehensive treatment of the logic behind the concept of the subject in the Conversations.

3.1 Critical Approaches to the Conversations

271

as such, it highlights a fundamental methodological problem which dogs the methods of every formal analysis of the Conversations: how can we explain the deficiency of a text when the deficiency itself is an assumed, unproven one? Unlike his predecessors, Dammann sees the deficiency of the Conversations in terms of social and genre history rather than normative aesthetics, but he is still left facing the very same problem. It can be put more pointedly as follows: given an apparently inferior narrative text, how can we describe it in such a way that we have a firm formal basis with which to interpret it as an innovative contribution to the development of a genre? More directly still, we might ask—particularly of those who claim to have discovered that Goethe was actually using the Conversations to make a satirical or ironic criticism of Schiller’s aesthetics—under what methodological conditions is it acceptable and reasonable to attribute a literary text with a global authorially intended statement of poetic theory in the mode of narratorial speech? We shall return to this problem after we have described the central theme in the theoretical discourse which many scholars have found reflected in the Conversations. Philosophical Discourses in Goethe’s Conversations Bräutigam (1977) was one of the first critics to devote a study to the analysis of the aesthetic and philosophical programme of the Conversations. His essay stimulated a series of studies and interpretations with a similar focus, such as Mommsen (1982), Witte (1984), Voßkamp and Jaumann (1992), and Bauschinger (1997). Like Bräutigam himself, all these critics examine how the Conversations are related to the situation in which they were produced. That is to say, the critics all aim to decide whether the content of Goethe’s narrative series was intentionally composed so as to contain a commentary on the aesthetic and moral didactic programme of the Horen, the journal in which Schiller published the Conversations and his own Ästhetische Briefe. The consensus has emerged that the best practicable heuristic apparatus can be assembled by combining judicious reference to the contemporary background with the comparison of three text groups: the epistolary correspondence between Goethe and Schiller which accompanied the Conversations project; Schiller’s Ästhetische Briefe; and the narrative texts which were submitted in several instalments by Goethe and together make up the complete text of the Conversations.20 20

In addition to Trunz’s commentary in the German edition (Goethe 1981, 6:611–23), see in particular Voßkamp and Jaumann (1992:1505–13) and Bauschinger (1997) for more recent discussions of the history of the composition of the Conversations in German.

272

Part 3 An Experiment in Action

Although there is considerable disagreement about the specific nature of the message that Goethe intended to communicate, it is generally accepted that the Conversations should be seen as a narrative reply to Schiller’s theoretical deliberations. Furthermore, all critics are more or less agreed that Kantian moral philosophy is the shared paradigm in terms of which we should approach the philosophy of Goethe and Schiller. This is significant because it is Kant’s moral philosophy that gives aesthetic instruction in social values a dignity which extends beyond the limited historical context of our two authors. However, the Kantian connection which establishes a superficial link between Goethe and Schiller also lays a dangerous methodological trap in our path. The detailed explanation of the potential danger by Gaier (1987) is unlikely to be bettered. Gaier reads the Conversations as a satirical imitation of Schiller’s ninth Ästhetischer Brief. With unrivalled expertise, Gaier paints a sobering picture of just how much the two correspondents, Goethe and Schiller, differed in their interpretation of key concepts such as that of the aesthetic judgment of taste or Geschmacksurteil (Gaier 1987:264) and the problematic status of objects of perception.21 We shall return to this point later. By demonstrating that the terms used by Schiller and Goethe are at least partially incompatible, Gaier forces us to adopt a far more sceptical attitude than before to the value, other than as heuristic devices, of approaches to the Conversations which analyse their origins or contextualize them by means of textual comparison. In a recent essay, Müller (2000) identifies a defining theme in previous research by noting that, for some time, critics have tussled with the links between Goethe’s Conversations on the one hand and Schiller’s Ästhetische Briefe and programme for the Horen on the other. Under Gaier’s influence, Müller then identifies a key methodological shortcoming in such approaches: the problem of validating interpretive hypotheses. The problem presents itself because we have yet to develop an effective means of deciding which of the several competing interpretations is correct (Müller 2000:166). This is not the place to discuss whether future critics will find it as easy as their predecessors to dismiss undesirable dissenters like Müller as purveyors of self-defeating theoretical arguments. What cannot be denied, however, is that Müller has every right to reinterpret the frequently discussed renunciation motif of the Conversations as a postulate of critical ethics which is thematized in the text and also acts as a normative restriction on our critical engagement with it:

21

See Gaier (1987:228) on the interpretation of the term ‘problematic’.

3.1 Critical Approaches to the Conversations

273

As the narrator makes clear to the reader (but not to the persons of the frame action, among whom only the Abbé knows this), individual facts and occurrences must be noted and respected; but they do not provide us with a focal point around which we can organize our interpretation of our own life, or someone else’s life, as a process of education. Müller 2000:171 (my translation)

The essence of Müller’s argument can be paraphrased by saying that the idea of education in the Conversations as a whole and the Fairy Tale in particular does not have any concrete substance. By this we mean that the text with which we are concerned does not propagate absolute social values but fosters a system of thought which enables man to question his own nature critically when he reflects on such values. It should be noted that this thesis does not amount to the methodological agnosticism which critics of inconvenient voices such as Müller would have us expect. However, despite the attractiveness of his theory, Müller has, I believe, omitted an important stage in his argument. Before we can consider the development of a self-critical thought process which allows the individual to attain social competence, there are more fundamental philosophical problems to be confronted. For example, how does the mind comprehend the external object domains of empirical nature and culture? When we perceive objects and events, what makes us think that they are connected in a particular way, and how do we reach the conclusion that they have developed in a process whose logic we can reconstruct? These questions raise issues which suggest that we should take a fresh look at the Conversations from a new, epistemological perspective. After all, there is no reason why Goethe’s stated intention to play with the interpretive compulsion of the reader should be interpreted simply in terms of contemporary themes such as the Revolution or the moral principles required for rational discourse to take place in an enlightened society. Indeed, Goethe himself believed that natural philosophy, not the philosophy of history, is the greatest source of rational human insight. For him, scientific experimentation is the key source of rational knowledge, while historiographical and hermeneutic knowledge can usually be inferred by directly interpreting the facts. In the following pages, we shall explore the possibility of reading the Conversations as an epistemological experiment, paying particular attention to the lack of formal homogeneity which characterizes the text and presents its critics with their greatest challenge. Before we begin, it may be helpful to recall a famous dictum from Wittgenstein’s Philosophical Investigations. Wittgenstein believes that explanation and description, the two basic ways of approaching an object of study, are locked in permanent combat with one another because explanation produces nothing but statements of belief and therefore has

274

Part 3 An Experiment in Action

little affinity with true science. The conclusion appears to be obvious: ‘We must do away with all explanation, and description alone must take its place’ (emphasis in the original). However, when Wittgenstein looks more carefully, he realizes that the semantic emptiness of purely descriptive science is illusory. So he continues, ‘And this description gets its light, that is to say its purpose, from the philosophical problems,’ and makes clear that the problem is necessarily based in language, the mode of the symbolic representation of meaning (Wittgenstein 1958:47). With that, it becomes clear that the question of how we might best describe the formal uniqueness of the Conversations is really a philosophical problem which is fundamentally inherent in the object itself. Bearing this methodological observation in mind, we shall now perform a detailed descriptive analysis of one particular aspect of the Conversations. Our description will be based on a computer-aided investigation of the action product yielded by my own reading of Goethe’s Conversations as an experimental series. The action logic of the sequence of individual narratives will be shown to display a continually increasing level of connectivity which culminates in the compulsively hyperconnective Fairy Tale. The data generated by our reception record files therefore bears the empirical footprint of a semantic effect of the game of compulsive interpretation masterminded by the narrator. Because of this, we shall then move from description to explanation in order to understand the philosophical and methodological argument at the core of the intentions behind the Conversations.

3.2 Describing the Conversations On 6 June 1830, thirty-five years after the publication of the Horen, Goethe wrote of the Fairy Tale to Thomas Carlyle, ‘When the work of the imagination has been channelled and guided, it compels reason to derive something regular and consistent from it; and reason, unable to resist the challenge, pursues that something without ever finding it’ (Goethe 1981, 6:610; my translation).22 This little aphorism could easily be placed at the beginning of just about any of the many explanations and interpretations produced by literary critics—including the present one. Goethe may not promise us much chance of lasting success, but we shall try nonetheless to ‘derive something regular and consistent’ from what is, after all, a work of Goethe’s ‘channelled and guided imagination’ and therefore designed to have a particular effect on us. In describing this effect, we shall employ the concept of action developed in part 1 and the computer software presented in part 2 which formally describes and combinatorially analyses narrative action potential in terms of numerical action product. The experiment was configured as follows: 1. The entire German text of the Conversations (including the Fairy Tale but not including the sections of frame narrative) was divided into a total of seventeen text files with the suffix *.tbf . The individual narratives were divided on purely quantitative grounds in order to make sure that they did not exceed the EventParser file size limit. The only exception to this was the Antonelli novella, which was split into a love anecdote and a supernatural anecdote as proposed by Fricke (1964: 278).23 As the analysis in this chapter is inextricably linked to a reading of the original German text, the *.tbf files were named after titles of the narratives in the German version, and the German titles will be used throughout the remainder of this chapter. 2. The seventeen *.tbf files were then encoded with EventParser. A total of 249 event constructs were defined during the encoding process. Because the encoding results from a single reading by a single natural reader (me), it represents one highly subjective reading. It is not, however, a

22

23

‘Eine geregelte Einbildungskraft fordert unwiderstehlich den Verstand auf, ihr etwas Gesetzliches und Folgerechtes abzugewinnen, womit er nie zu Stande kommt.’ In the German edition, this division lies at line 39 in Goethe (1981:151).

276

Part 3 An Experiment in Action

self-fulfilling prophecy which presents an individual’s interpretation of the text in the guise of an objective formal description. This is due to two key features of the processing procedure: first, our theory ensured that the role of human interpretation did not extend beyond the minimal construct, the event; second, the seventeen text fragments were encoded in a random order which differs from that of the actual syuzhet.24 3. The standardized term files Standard.lbf and Standard.sbf were loaded before each of the seventeen fragments was processed. This preconfiguration ensured the formal comparability of the record files and supplied a standardized stable world knowledge in the form of prespecified descriptive predicates. The use of supplemental descriptive terms (mainly adjectives and adverbs) was deliberately kept to a minimum; nonetheless, the terms defined on an ad hoc basis almost always outnumbered the standardized ones. The results of this first phase of processing were stored in seventeen record files with the suffix .esf. 4. A list of all the terms entered or derived during processing was copied into each *.esf file before it was combinatorially analysed in EpiTest. This allowed semantic connections to be made between the record files despite the fact that they had been produced by an arbitrary segmentation of the text. 5. In the next step, EpiTest was used to analyse two sets of *.esf files. First, the software examined the seventeen *.esf files produced for the arbitrarily defined text segments. The seventeen original *.esf files were then merged into seven new *.esf files (one for each of the six individual narratives and a seventh for the entire text): Antonelli.esf, Bassompierre.esf, Klopfgeist.esf, Prokurator.esf, Ferdinand.esf, Maerchen.esf and Unterhaltungen.esf. This second set of seven files was then analysed in EpiTest. Each group of files was passed through EpiTest twice. In the first pass, the program was instructed to generate isochronous episode and action constructs (these can be identified by the prefix iso- in the relevant *.con files). In the second pass, both isochronous and anisochronous constructs were generated (prefix aniso-). The tables and illustrations in the following pages are based on a statistical analysis of the results as saved in the relevant results files (*.daf). The

24

The order of encoding was as follows: (1) Maer3.tbf, (2) Klopfg.tbf, (3) Proku3.tbf, (4) Anto1.tbf, (5) Maer6.tbf, (6) Ferdi3.tbf, (7) Bass1.tbf, (8) Ferdi2.tbf, (9) Maer4.tbf, (10) Proku2.tbf, (11) Anto2.tbf, (12) Maer2.tbf, (13) Proku1.tbf, (14) Bass2.tbf, (15) Maer1.tbf, (16) Maer5.tbf, (17) Ferdi1.tbf.

3.2 Describing the Conversations

277

processing time varied considerably, from the 0.1 seconds required for the combinatorial isochronous analysis of the simple Klopfgeist narrative (the story of the noisy spirit) to the 14,472.7 seconds (over four hours) needed for the combined isochronous and anisochronous analysis of the entire Unterhaltungen, during which 7,680 individual episode and action constructs were generated.25 All the text base files and report files processed and generated in the course of the experiment have been made available for download so that readers can examine them in their own time.26 After familiarizing themselves with the procedure by reproducing our experiment, the more adventurous readers could then move on to adapting it to facilitate investigations of their own—it would be particularly interesting, for example, to analyse the English translation of the Unterhaltungen and compare the results with those for the German original. This and a wide range of other projects can be undertaken with our two software tools, EventParser and EpiTest. Three values calculated by EpiTest are of particular interest and are explained below; we shall then proceed to discuss and interpret the results of our experiment. 1. % event integration (EPISODE onstructs / EPISODE potential) * 100 This is a percentage value which represents the degree to which the event constructs can be integrated into virtual isochronous and/or anisochronous episodes. The percentage is not calculated on the basis of the number of event constructs themselves because that value depends on the reader’s earlier interpretive acts and would therefore give a misleading result. Instead, we subtract the minimum theoretical number of possible episode constructs from the maximum theoretical number of possible episode constructs to obtain a figure that represents the theoretical potential of episodes. The number of episode constructs which are actually produced lies somewhere within this range.

25

26

Since the processing time required is largely dependent on the hardware of the host system, the relation between the two extremes is more important than the absolute values. However, absolute values are interesting to note where machine generated output is concerned. At times, the temporary action lists and prolog episode predicates can fill almost 2 MB of RAM during processing, and the resultant aniso-Unterhaltungen.con file occupies just over 1 MB of disk space with 500 printed DIN A4 pages of episode clauses and action lists which, if back-translated into natural language narratives, would occupy at least ten times as much space. In comparison, the Unterhaltungen themselves span approximately 125 pages. This quantitative difference should give some impression of the complexity of potential action that narrative texts can imply rather than denote. These are obtainable at

278

Part 3 An Experiment in Action

3.2 Describing the Conversations

279

280

Part 3 An Experiment in Action

2. % episode integration (maximum ACTION length / EPISODE onstructs) * 100 This is a percentage value which represents the amount of episode constructs which can be combined in isochronous action constructs. We calculate the percentage using the number of episodes in the longest action chain divided by the total number of episodes. 3. action product EVENT integration * EPISODE integration Unlike the previous two results, this value is not a percentage. It is a measure of the overall theoretical connectivity of a text and represents the degree to which the text allows the reader-defined events to be integrated in episodes and those episodes to be integrated in action constructs. The Isochronous Constructs It is a somewhat daunting task to identify general trends in the mass of raw output data presented in tables 3.2.1 and 3.2.2. Interpreting the results will be much easier if we confine ourselves to the six individual narratives, focus on the three key results described above, and represent them in the form of a graph. We shall begin by considering the 1,937 isochronous episode constructs which EpiTest generated from our seventeen *.esf record files. In this first part of our discussion, we shall pay particular attention to the level of event integration. Table 3.2.3 summarizes the results, which are then plotted in figure 3.2.1.

Table 3.2.3: Summary of Isochronous Analysis in EpiTest

3.2 Describing the Conversations

281

Fig. 3.2.1: Graph of Isochronous Analysis27

For the most part, the event integration plot (diamonds) corresponds to our intuitive impressions. The value for the bipartite Antonelli novella is held down somewhat by its problematic appendage; it therefore joins the Prokurator and Ferdinand novellas in a group of three narratives which are closest to the average level of event integration. Bassompierre can be seen as the closed narrative whose content is most tightly organized and least disrupted by narratorial reflection. It represents a striking peak on the event integration plot. Of the four narratives of fictional biography, the Bassompierre anecdote is most persuasive at encouraging the reader to construct an episodic reading. Unlike the other three, it does not contain any real moral, or more correctly moralizing, superstructure. The Klopfgeist narrative, on the other hand, lies at the opposite end of the scale in this and all other respects. The loosely connected enumeration of acoustic phenomena in its narrative can only be encoded in the form of isolated epistemic event constructs, if that, which stubbornly resist being semantically combined as episode constructs. The reason for this has nothing to do with Klopfgeist’s relatively small number of event constructs (eleven). Table 3.2.1 shows that the first fragments of the Bassompierre and Ferdinand narratives have almost the same number of event constructs 27

To improve the clarity of the graph, the initial value for the Antonelli novella in column one of table 3.2.3 (9.669) was used as the initial value for the plots of columns two and three. Thus, all figures in column two were multiplied by 4.093, and all figures in column three by 0.42403. This mathematical adjustment means that the plots all start from the same point and that the path of the plots of columns two and three is raised and lowered respectively. The scale on the Y axis refers only to the percentage values for the first plot (event integration) because this is the value which is of most interest to us. Table 3.2.3 gives the data in its absolute, unadjusted form.

282

Part 3 An Experiment in Action

(twelve); despite this, they yield a far larger quantity of episode constructs (sixty-four for Bass1.esf and twenty-one for Ferdi1.esf, as opposed to three for Klopfgeist.esf). Thus, the events encoded for Klopfgeist cannot be dysfunctional because of their quantity. The real reason is that they are semantically redundant—the empirical cycle of continual recurrence does not provide the logic of action with the meaningfully arranged happenings that it needs in order to be effective. The Prokurator and Ferdinand narratives are both examples of the genre of the moral tale, and there are clear parallels between the motifs and conception of both stories. Their event integration values are closer to the average than all the other narratives, and, as we would expect, the two texts have only slightly different levels of event integration. The Märchen is very different, for its level of event integration is only slightly greater than the minimum represented by Klopfgeist. One hundred and six event constructs were encoded in the Märchen record file by the recipient. The size of this figure suggests that a correspondingly large number of episode constructs will be generated. A quick glance at table 3.2.1 is enough to confirm that an unsurpassed maximum theoretical number of episode constructs is indeed produced: 11,130 far outshines the next highest number (2,352 for Ferdinand) and would seem to represent a considerable amount of material for the logic of action to work with. Yet the analysis of the Märchen record file produces no more than a meagre 563 isochronous episodes. Expressed in less mathematical terms, this means that as the number of individual events perceived in a possible world increases, it becomes progressively harder for the semantics of action to bind that world together as a coherent whole. It is extraordinarily difficult for a narrator or reader to absorb and cognitively synthesize an overcrowded event pool. As we shall now see, the same applies to the more abstract combination of episode constructs into actions. The second plot (circles) describes the maximum degree to which virtual episode constructs can be integrated in complex action constructs. The most conspicuous gap between episode integration and event integration can be observed in the Bassompierre narrative—although the Bassompierre record file yields a very large number of episode constructs, these episodes cannot be converted into a similarly high number of action constructs. In fact, the percentage yield of action constructs for Bassompierre is just less than that for Antonelli. The episode integration statistics therefore provide further evidence to support the theory suggested above, which treats the Bassompierre anecdote as a classic, representative example of a text with a high level of episodicity: readers can generate many self-contained episode constructs on the basis of the text, but these constructs do not bond well with one another when combined in the logic of an overall

3.2 Describing the Conversations

283

action. The episode constructs in the Antonelli narrative, on the other hand, are ideally suited for further integration in action constructs, and the Prokurator novella is the next most efficient text in this respect. These results may suggest that the first embedded narrative is intended to have an expositional function which is then dissolved, equally intentionally, when we enter the world of the ghost story. The Antonelli, Bassompierre, and Prokurator narratives all have similar levels of episode integration. This may be due to the similar ways in which they were composed—as we know from previous work on the background of the Unterhaltungen, Goethe wrote these three narratives by creatively modifying Romance sources (which is very different from the origins of the Klopfgeist anecdote, which, Charlotte von Stein tells us, Goethe reproduced almost word for word from memory after hearing it in Weimar).28 The potential for episode constructs to be integrated in action constructs is almost identical in the Antonelli and Prokurator novellas. If we are to interpret this as an intentional result of the author’s aesthetic craftsmanship, we must explain why the two lowest points on the episode integration plot are marked by the very parts of the text where the author was not restricted by following a model source. The explanation can be found in table 3.2.1: in our experiment, Ferdinand and the Märchen were read in such a way that they contain a considerably higher number of virtual episode constructs than any of their counterparts. Once again, we see that the legacy of the reader’s constructive acts can have far-reaching implications for the logic of action. We shall return to this idea below. Our third plot (triangles) shows that the action product (event integration * episode integration) generally rises and falls in conjunction with the event integration level. The high level of event integration in the Bassompierre narrative now makes up for its low level of episode integration. Bassompierre presents us with a mathematical phenomenon for which there is no obvious and convincing phenomenological interpretation; the only conclusion that might be drawn is that this first-person narrative produces a relatively high action product because of its relatively simple theme and narrative technique. The Prokurator narrative fits in with the general pattern thanks to the interplay between the calculations which underlie the action product. It contains extended passages of narratorial reflection and commentary and, above all, extensive dialogue between its characters. Such material cannot be converted into fictionally referential event and episode constructs by 28

See Hoffmann (1985) on Antonelli; on the Bassompierre and Klopfgeist narratives and the Prokurator novella, see the commentary in the German edition (Goethe 1981, 6: 626ff.) and Brandt (1983).

284

Part 3 An Experiment in Action

the recipient. It is not surprising, therefore, that Prokurator can generate only a rather small number of virtual episodes in comparison with the two subsequent texts. The small size of this number ensures that the action product has an acceptable value. The action products for the Ferdinand novella and the Märchen, on the other hand, are the latest in a series of disappointing results for these two texts, Goethe’s two original creations in his narrative experiment. The action products of these stories plunge to a depth not far short of the minimum marked by the Klopfgeist anecdote. The action logic of the narrative series, it would appear, builds to a clearly defined anticlimax. The Isochronous and Anisochronous Constructs We shall now discuss the results of the second stage of our analysis, in which EpiTest generated and evaluated isochronous and anisochronous episode constructs. There is a sound pragmatic reason for undertaking this second analysis: until now, EpiTest has not generated a single action chain of more than four episodes in length. It is not hard to explain why this is so—the more extensive the overall semantic potential of a narrative text and the more specific the semantic encoding of its constituent events, the greater the gap between the exposition and disposition of the individual episodes when only isochronous constructs are being generated. This gap can become so great that only a small number of semantically connectable episode constructs of classes eiso1–3 can be generated. The effect is intensified in the Unterhaltungen novellas by the hypotactic narration typical of the genre, which is designed around a climactic resolution held back until as late as possible. Conversely, more compact episode constructs are produced by the compact plots of the trivial paratactic reports, which progress towards the resolution of tension with as little disruption as possible. There is also a theoretical justification for the second part of our analysis. If we are to define the semantic significance and formal cohesion of virtual actions differentially, we must be able to measure them as objectively as possible and with as few prior assumptions as possible. In practice, this means that we need a frame of reference which does not prematurely narrow down the theoretical possibilities by means of an ordering criterion which lacks a proper formal basis. Take the criterion of formal sequentiality, which requires that the constituent episodes of an action construct must stand in unilinear sequence. This criterion has nothing to do with the abstract syntagmatic order which defines the semiotic continuum of a text. Rather, it represents a hermeneutic manifesta-

3.2 Describing the Conversations

285

tion of our deep-seated, intuitive normative conviction that the symbolic representation (the narrative) and the fictive object to which it refers (the sequence of events) inherently unfold in a parallel chronological course. We can gladly leave it to learned experts on philosophy, religion, and physics to decide whether things are like this in the real world; what matters for our purposes is that analeptic and proleptic disruptions of the illusory ontic norm are par for the course in narrative texts and, indeed, in most oral reports of any kind. We have taken account of this fact by using EpiTest to perform a combined isochronous and anisochronous evaluation of the seventeen *.esf record files for the Unterhaltungen. The results are presented in table 3.2.4 and figure 3.2.2.

Table 3.2.4: Summary of Combined Analysis in EpiTest

Fig. 3.2.2: Graph of Combined Analysis29 29

The values of all three plots have been adjusted to start from a single point represented by the action product of the Antonelli narrative. To achieve this, the data in table

286

Part 3 An Experiment in Action

In this second discussion, we focus on the value of the action product. It therefore replaces event integration as the reference scale on our graph’s Y axis. We shall confine our discussion to the third plot (triangles), which describes the overall results of the combinatorial analysis of the seventeen record files in EpiTest. For the first three components of Goethe’s narrative experiment (the Antonelli, Bassompierre, and Klopfgeist narratives), the overall pattern matches that the results of the isochronous analysis described above. After the rise and fall of the first three points, however, we see a sustained increase in the action product of the Klopfgeist, Prokurator, and Ferdinand novellas. They do not come close to matching the high standard set by the Bassompierre narrative, which is confined to pure action narration; on the other hand, it is encouraging that Ferdinand, Goethe’s first independently designed novella and a counterpart to the Romance models he adapted elsewhere in the Unterhaltungen, almost exactly matches the ideal, average action product of the six narratives. The status of the Märchen is now even more disappointing—it has the smallest amount of action of all the narratives in the experiment, outdoing even the distinctly banal Klopfgeist narrative. To put it bluntly, the Märchen’s record file, Maerchen.esf, yields the lowest action product of all; in formal terms, this means that the event constructs encoded in Maerchen.esf have the lowest level of cohesive integration in the action logic of the text. How might we interpret the significance of this result for the Märchen? In purely numerical terms, the second analysis lends further backing to the impression created by the first that the Märchen falls victim to its own complexity. It has 106 basic event constructs; they can produce a theoretical maximum of 11,130 possible episode constructs and a range of 11,077 distinct significant episode constructs constituting its theoretical episode potential. Thus, although the Märchen produces an impressive 1,075 episodes, this amounts to less than 15 percent of the theoretically possible number. And although the second analysis of the Maerchen.esf record file does produce a longer maximum action chain (sixteen episodes) than the first, this falls a long way short of the Ferdinand narrative whose maximum action chain of 243 episodes is fifteen times greater.

3.2.4 was manipulated as follows: column one was multiplied by 27.384, and column two by 26.951.The values on the Y axis apply only to the action product and should be multiplied by 100. Table 3.2.4 gives the data in its absolute, unadjusted form. Note that the comparison of these figures with the data documented in the downloadable *.daf-files (generated by EpiTest in our actual experiment) may show slight deviations which result from the rounding of numbers.

3.3 Explaining the Conversations The semantically neutral algorithms of EpiTest generate raw numerical data which we have listed in our tables and illustrated in the more convenient and stimulating form of our two graphs. The time has come, however, to put an end to the increasingly convoluted periphrasis of our numerical findings. At some point or another, we will have to recognize that the quantitative description and analysis of empirical objects has its limits; in order to progress beyond them, we must abandon our tortuous speculation about the possible correspondences between numbers and meaning and show a new readiness to take a hermeneutically unavoidable risk. As with so much of the ground we cover in this chapter, Goethe has been here before us, and here, as will often be the case in the coming pages, he shows us the way in his essay “The Experiment As Mediator between Object and Subject.” In it he writes that in the mathematical method we find an approach which by its deliberate and pure nature instantly exposes every leap in an assertion. Actually, its proofs merely state in a detailed way that what is presented as connected was already there in each of the parts and as a consecutive whole, that it has been reviewed in its entirety and found to be correct and irrefutable under all circumstances. Thus its demonstrations are always more exposition, recapitulation, than argument.…We can see the great difference between a mathematical demonstration which traces the basic elements through their many points of connection, and the proof offered in the arguments of a clever speaker. Although arguments may deal with utterly separate matters, wit and imagination can group them around a single point to create a surprising semblance of right and wrong, true and false. Goethe 1988:16

Unlike the mathematician, I will not be able to reduce the numbers and arguments presented in chapter 3.2 to such a ‘single point’ in the following pages. Although personally willing for this failing to be put down to a lack of ‘wit and imagination’ on my part, I should like also to point out that there is a certain amount of methodological justification behind it. A study in literary computing that uses the adjective ‘literary’ with any degree of seriousness must never forget that we should apply numbers as part of an interpretation, not extract it from them. Until this point, an unbridgeable gap has separated the numerical results of our combinatorial analysis of action product from the interpretation of the Conversations in terms of Goethe’s philosophical thought. We are now in a position to supply the

288

Part 3 An Experiment in Action

missing conceptual link; it takes the form of Goethe’s belief that we cannot understand the logic behind the history of the real world. This belief explains the logic behind the fictional events in his cycle of novellas. Goethe and the Unfathomability of the Weltbegebenheiten One of the issues treated in the Conversations is obviously the question of how to come to psychological terms with contemporary events of the French Revolution. However, Goethe’s narrator makes it perfectly clear that this objective lies far beyond what can be achieved by means of the aesthetic creation of meaningful representations: I have been in this world a long time and have always taken an interest in what happens to different people. I have neither the strength nor the courage to review the history of the world at large, and isolated historical episodes confuse me. But of the many personal histories, true and false, that circulate in public or are whispered about in private, some have a greater, more genuine charm than the mere novelty; some amuse us by an ingenious twist; some reveal for a moment the innermost secrets of human nature; and others delight us by their bizarre absurdities. Goethe 1989:26

Two years after the Conversations, Goethe returned to the project of aesthetically combing personal histories and historical episodes or Weltbegebenheiten in the verse epic Hermann and Dorothea. Both works share the general theme of flight and exile caused by war, but the later text shows a twofold innovation compared to the Conversations. The content shifts from a noble to a less refined bourgeois milieu, and the form exhibits a considerably more rigid structure. By necessity, the verse epic concentrates on fictional representation and cannot provide its figures or narrator with the self-reflexive commentaries that would reveal the constructive nature of the interpretive process. As a result of its strict formal configuration, Hermann and Dorothea can hone the historical episode down to an exemplary personal history, thus creating meaning through reductionism in a way that would be utterly impossible in the Conversations. And by no means did Goethe’s interest in the problem end here, as is demonstrated by the later Elective Affinities, which is carefully calculated to thwart any attempt at reading a closed interpretation into the text.30 Goethe’s continued fascination with the problem of personal history and the historical episode is not surprising when we consider the complex typology of 30

See here Beddow (1984), who gives special consideration to the un-ended plots in Goethe’s Die Wahlverwandschaften.

3.3 Explaining the Conversations

289

possible narrative objects suggested by the above quotation. The history of the world at large subsumes historical episodes, to which in turn the entire diverse spectrum of personal histories is subordinate, and they too can be true or false, innovative or traditional, intellectual, merry, or just plain silly. The point is that all this appears as an empirical reality before it is reworked in narrative form—it appears as something which ‘happens to different people,’ something in which the narrator, as an observer, has ‘taken an interest.’ There are two ways to approach the problem of how a narrative can possibly do justice to the convoluted empirical data of the real world. One is deductive, the other inductive, and Goethe’s narrator begins the typology in the above quote with the former (the historiographical method of handling the history of the world at large). Now, we know from Goethe’s own words that he was extremely sceptical about the legitimacy, indeed the very possibility, of writing history in a way that creates meaning. As Bubner (1993) has shown, Goethe would have considered it absurd to attempt to write a comprehensive history of mankind, particularly in an age when it was no longer possible to accept old certainties such as the teleological order of the cycle of eternal recurrence and the providential assurance that every individual story is an integral part of the universal world history. Stripped of our former security, we find that a near-insurmountable obstacle prevents reason from grasping what happens in the world. The obstacle in question is the recognition, now unavoidable, that the happenings in the world are, at least in part, contingent (Boyle 1993: 164f.). Goethe himself puts this as follows in his Theory of Colours: Whatever the conditions in terms of which we consider the development and deeds of mankind, both will vary with time and country, on a small and a large scale which influence one another proportionately and disproportionately. This is the source of all that is incalculable, unfathomable in world history. Law and coincidence interact with one another, but the person who observes them is often misled into confusing them. This is particularly apparent in the work of biased historians, who, most often unconsciously but skilfully enough nonetheless, exploit the uncertainty to their advantage. Goethe 1981, 14:49 (my translation)31

31

‘Man mag sich die Bildung und Wirkung der Menschen unter welchen Bedingungen man will denken, so schwanken beide durch Zeiten und Länder, durch Einzelheiten und Massen, die proportionierlich und unproportionierlich aufeinander wirken; und hierin liegt das Inkalkulable, das Inkommensurable der Weltgeschichte. Gesetz und Zufall greifen ineinander, der betrachtende Mensch aber kommt oft in den Fall, beide miteinander zu verwechseln, wie sich besonders an parteiischen Historikern bemerken läßt, die zwar meistens unbewußt, aber doch künstlich genug sich eben dieser Unsicherheit zu ihrem Vorteil bedienen.’

290

Part 3 An Experiment in Action

The references to the observing subject and the partisan historian show that here too Goethe intends to draw our attention to the methodological and phenomenological aspects of historiography’s most basic problem. Crucially, Goethe does not believe that the same a priori methodological limitation applies to the natural sciences, which alone can contribute to the development of a ‘culture of knowledge instinctively and for its own sake’ (Goethe 1981, 14:47; my translation).32 Goethe sees this quest for insight as the most important of all. However, he immediately shows his realistic streak by adding that it has always been motivated by pragmatic needs rather than the desire for knowledge in itself. From the methodological point of view, therefore, the study of nature does involve a certain amount of pragmatic (as opposed to inherently necessary) subjectivity. From the more fundamental phenomenological perspective, the situation is very different. If anything expresses a perceptible regularity, it is nothing other than the course of nature itself, for nowhere else do we find the old principle of a divinely ordained cycle of eternal recurrence preserved in a form that we can still recognize today. Human history, on the other hand, can always be interpreted in two basic ways: superficially as a causal chain of instances of spontaneous human behaviour and transcendentally as a sequence of happenings whose logic lies in a by definition impenetrable plan which, for all we know, could have been designed by God or equally well by nature.33 The very nature of the object of historiography means it is fundamentally impossible for us to perceive it in a properly scientific manner. Thus, for Goethe, we are wasting our time if we try to identify some epistemological principle which would allow us to found a new, scientific historiography (Rennie 1996:121). Koselleck has coined the term ‘anthropology of historical experience’ (my translation) to denote the unique methodology implied by Goethe’s concept of history. Koselleck argues that Goethe’s anthropological approach is striking because it uses an inductive method more at home in the observation of nature: ‘To observe and go from sense to perception, to imagine and go from perception to judgement, to conviction—that could be the essence of the formula for revealing historical insight before the passage of time obscures it from our gaze [once again]’ (Koselleck 1993:33; my translation). It is tempting to conclude that all is not lost and that there is at least a chance that the methods of the natural sciences, or rather natural anthropology, might show us how to capture the history of the world at large in a narrative of human history. However, as Käuser (1990) has 32 33

‘Kultur des Wissens durch inneren Trieb um der Sache selbst willen.’ A similar theme is prominent in von Graevenitz (1993), whose interpretation of Goethe’s poem ‘Pandora’ (1807) is also based on the philosophy of history.

3.3 Explaining the Conversations

291

convincingly shown, Goethe would have found it simply unthinkable to apply the methods of one discipline to the object of another in this way. Käuser therefore considers the problem from the opposite angle by taking the approach of genre theory rather than natural philosophy or the philosophy of history. He sets out to explore Goethe’s idiosyncratic concept of the genre of the novella and concludes that its rules are formally based on Goethe’s belief in the ‘inherent value of sensual physical phenomena’ (Käuser 1990:160). The key feature of Goethe’s definition of the novella eventually crystallizes as the ‘unheard-of event’; Käuser argues that the term originally denotes an empirical phenomenon which is unheard-of in the literal sense of never having been perceived before and whose truth Goethe believes is anchored in its inexplicable, empirical reality as a phenomenon rather than in the more superficial context of its conceptual definition. For Goethe, knowledge of this kind of truth is identical with neither the truths of philosophy nor the findings of the natural sciences; the theory and method behind it lie rather in the anthropological discipline founded by Herder, which, as Käuser argues, ‘becomes an innovative science of observation in the eighteenth century by means of an empiricism which sets it apart from both the rationalism of the Enlightenment and the idealism of Kant’ (Käuser 1990:160). The key point to note here is that Goethe does not take the simple approach of applying the empirical methods of the natural sciences to mental and social phenomena. Instead, he calls for the creation of knowledge to be divided into two separate methods, those of scientific, morphological study of the objective natural world on the one hand and those of anthropological, aesthetic study of subjective human reality on the other.34 Let us review our brief study of the distinctive features of Goethe’s idea of history, which turns on the question of whether we can gain insight into the logic of world events (‘Weltgeschehen’), and if so, how. We have found three answers to this question, two negative and one more encouraging. We now know what Goethe did not have in mind when he wrote the sequence of narratives in the Conversations. They cannot be designed to provide a deductive history of the world at large which synthetically subsumes the many different personal histories. Nor can they be designed to narrate inductively outwards from the empirically secure personal histories and naively apply the methods of morphological nature study to the analysis of psychological human problems. (Recall that such methodological abuse of the object was the reason for Goethe’s aversion to contemporary histori34

Käuser (1990:164ff.) goes on to link the dualism of rational and anthropological methods of discovering knowledge with the distinction between the genres of the novella and the novel in Goethe.

292

Part 3 An Experiment in Action

cism.) What is needed instead is a narrative experiment whose methods are indeed based on the model of perception of the natural sciences but also, and more importantly, combine a new anthropological aesthetics and empiricism more appropriate to their object. In the ideal case, this will make it possible to ‘reveal for a moment the innermost secrets of human nature.’ Note that this takes place ‘for a moment,’ that is to say, as a quasi-empirical insight into a truth which has the power to convince us before, and indeed without, the conceptual definition that would elevate it to the status of an idea. The views of Goethe and Schiller, the two partners in the Horen project, divide over precisely this point, the move from perception to concept. To illustrate this, we shall return to the letter from Schiller which we quoted above. In this letter, the editor of the Horen provides a critical commentary on the first narrative section of the Conversations (i.e. the first part of the introductory frame narrative), which was submitted by the author and subsequently published in the first issue of the journal in January 1795. Schiller considers ‘it all very sensibly introduced’ (‘das Ganze sehr zweckmäßig eingeleitet’) but finds fault with the division that results from the fact that the second part of the frame narrative will only appear later, in the second issue of the Horen: The reader is provided with too little material to consider and is consequently unable to evaluate properly how what is said relates to the work as a whole. It would therefore have been preferable if the first narrative could also have been included in your submission. Schiller to Goethe, 29 November 1794 (my translation)35

Schiller also takes issue with the frank and presumptuous exchanges between Luise and the Abbé, and criticizes the obvious contemporary allusions incorporated in the acrimonious dispute between Carl and the privy councillor. He fears that these features could cause offence to his readers and mislead them into taking the entire story literally, or, as we would say in more modern critical jargon, reading the text as one that references contemporary reality: A number of features, in particular the somewhat laborious opening of the narrative, suggested to me that you intend to give your readers the impression that they are dealing with something that really happened. As you are going to play with their interpretive compulsion anyway in the subsequent narratives, would it not be better to do so from the beginning and problematize the vehicle 35

‘Der Leser zu wenig auf einmal zu übersehen bekommt und daher nicht so imstande ist, die notwendigen Beziehungen des Gesagten auf das Ganze gehörig zu beurteilen. Es wäre daher zu wünschen gewesen, daß gleich die erste Erzählung hätte können mitgegeben werden’ (Voßkamp and Jaumann 1992:1509).

3.3 Explaining the Conversations

293

itself so as to prevent others from making the same mistake that I did? You must not forget that I too feel compelled to interpret your work. Schiller to Goethe, 29 November 1794 (my translation)36

What would Schiller have preferred to have been sent? First, if Goethe had also supplied the first embedded narrative, he could have made it clear that the frame narrative represents an introduction and commentary. This would have reduced the danger of his readers believing that they were reading historical facts which would have been particularly disagreeable because of their political nature. Second, Goethe should have problematized ‘the vehicle itself’ (i.e. the frame fiction); the most likely reason for this is that it would have prevented even the most careless of readers from being confused by the text.37 The point is that, contentious as it may sound, Schiller the editor is perfectly happy for Goethe to play with his readers’ interpretive compulsion—but only, of course, if the rules of the game are clearly explained in advance and the use of surprise tactics is prohibited. The final sentence in the above quotation can be read in two ways: Schiller is either stylizing himself as an ideal reader with the critical insight needed to pinpoint the shortcomings of Goethe’s contribution or ironizing himself in a rather feeble attempt to soften the impact of his criticism. Our argument does not require us to decide which of these interpretations is correct;38 far more important is the fact that the extract reveals a fundamental misunderstanding between Goethe and Schiller whose subsequent repercussions for the composition of the Conversations cannot be underestimated. Gaier (1987) examines the conceptual divide between the two authors in detail. Superficially, the discrepancy turns on the concept of problematization, but its real cause is the fundamental difference between the ways in which Goethe and Schiller interpret Kant. Schiller understands Kant in an idealistic, subjective manner, while Goethe reads his work in terms of 36

37

38

‘Ich glaubte aus einigen Zügen, besonders aus einer größeren Umständlichkeit der Erzählung am Anfange, schließen zu können, daß Sie die Absicht haben, die Vermutung bei dem Leser zu erwecken, daß etwas wirklich Vorgefallenes im Spiele sei. Da Sie im Verlauf der Erzählungen ohnehin mit der Auslegungssucht Ihr Spiel treiben werden, so wäre es wenigstens nicht übel, gleich damit anzufangen und das Vehikel selbst, in dieser Rücksicht, problematisch zu machen. Sie werden mir meine eigene Auslegungssucht zugute halten’ (Voßkamp and Jaumann 1992:1509). The term ‘vehicle’ had previously been used by Goethe in a letter to Schiller dated 1 October 1794, in which he writes that he is planning ‘vehicles and masks with which and behind which we will be able to manipulate the audience in a number of ways’ (‘Vehikel und Masken, wodurch und unter welchen wir dem Publico manches zuschieben können’ Voßkamp and Jaumann 1992:1508; my translation). Hoermann (1996) discusses the significance of the ironic metaphor of the ‘masks’ in the Fairy Tale. In fact, Schiller establishes a double role for himself in the letter, making some comments as an ‘editor’ (‘Redakteur’) and others as an ‘ordinary reader’ (‘bloßer Leser’).

294

Part 3 An Experiment in Action

an epistemological, empirical paradigm. We have already hinted at what Schiller meant when he called on Goethe to ‘problematize the vehicle’—the fictional status of the frame narrative and its function as an expositional commentary should have been made clear, the premature misinterpretation of the text as a factual referential report prevented, and a clear indication given that the Conversations should be interpreted symbolically. What Goethe actually understood by problematization, however, had nothing in common with the criticism of narrative expression that Schiller had in mind. Goethe thought in terms of natural philosophy and read Kant’s work from its perspective. He therefore believed that it is not simply the ways in which we talk about objects that must be problematized; from his point of view, the objective objects themselves are ontically and therefore epistemologically problematic in so far as an immeasurable spectrum of aspectual ‘Vorstellungsarten’ (perhaps best translated as ‘conceptualizations’ or ‘forms of conception’) can be assigned to every object. Gaier explains the key term ‘Vorstellungsart’ as follows: In his early works on morphology after the Italian journey, Goethe the natural scientist begins to use ‘Vorstellungsart’ as a term for a scientific hypothesis which stands in antinomy to or is dialectically related to an opposing hypothesis because neither of the two on its own is enough to provide a satisfactory explanation of a given ‘problematic’ object. Gaier 1987:223 (my translation)39

Gaier goes on to argue that the philosophical misunderstanding became aesthetically productive when Goethe took up Schiller’s suggestions. The result was that Goethe revised the entire conception of the Conversations, inserted the Antonelli anecdote at the beginning, and postponed the Prokurator novella (which he had already mentioned to Schiller) until nearer the end. This process created the cycle of novellas in the form in which we know it today. Taking the enumeration and testing of experimental hypotheses (equivalent to ‘Vorstellungsarten’) as his model, Goethe provided Schiller with a series of individual narratives which are related differentially to one another in such a way that they intensify each other more and more.40 39 40

This is one of several instances where Gaier draws on Kleinschnieder (1971). Cf. Gaier (1987:230). We cannot discuss here how robust Gaier’s thesis is in its treatment of the origins of the Conversations, but it is certainly developed with great care on the basis of substantial textual examples, and Voßkamp and Jaumann (1992:1511) at least consider it worthy of consideration. However, it cannot be denied that there is a more immediately obvious indication of the formal principle that Goethe had in mind: the reference to the Decameron in the preliminary discussions with Schiller, if not simply the use of a frame narrative in the introduction. In my view, Gaier’s thesis that the Conversations are legitimized solely by their discursive connection to the Horen programme (Gaier 1987: 248) is similarly redundant: it is unnecessary rather than wrong as such.

3.3 Explaining the Conversations

295

By comparing the thought of Goethe and Schiller, Gaier’s study not only elucidates the development of the Conversations but also improves our understanding of the content of the finished work. It is without doubt one of the most insightful and careful interpretations of the text to date. However, it leaves a key question unanswered: assuming that Goethe was indeed using a series of scientific experiments as a means of problematization, why did he subject them to the purely aesthetic principle of formal closure for which a novella cycle strives? And why, if he was therefore aiming to compose a closed cycle of novellas, did he not actually manage to do so?41 Before we can begin to answer these questions, we must deal with a more fundamental theoretical problem: what is it that internally distinguishes a narrative cycle from a narrative series? (Note that a frame narrative which is taken up at the beginning and end of a work is a purely external feature and therefore not an adequate solution to our problem). According to Pickerodt, serial compositions show a continual and progressive thematic development, whereas cycles are organized around a single central theme. He is of the opinion that markers of serial composition predominate in the Conversations (Pickerodt 1994:27).42 As a proponent of this view, he is far outnumbered by the majority of critics, who adopt Müller’s thesis that Goethe consciously turned to the cycle of novellas as his model because it makes possible the ‘discussion dialogues in a storytelling society’ (Müller 1969:162; my translation). However, most supporters of this argument do not stop to consider how sound it actually is. If we take the time to examine the situation more carefully, we see that while Goethe does indeed make extensive use of this formal way to include self-reflexive commentaries, he does not write a complete cycle in the fashion of his Romance models. There are at least two arguments which suggest it is fundamentally impossible that Goethe was trying to write a faithful reproduction of the genre prototype he had inherited. Returning to the frame narrative at the end of the Conversations and rounding them off with a profound closing commentary from the Abbé would have seamlessly connected the fictional reception situation of the storytelling society with the real reception situation of the empirical reader. To do so, however, would have blatantly contradicted Goethe’s own statements to the effect that the closure of a work of art must be internally based. Goethe was convinced that this kind of closure is an aesthetic construct, the logic of which is not objectively 41

42

The idea that Goethe stopped writing the Conversations because he was fed up with the reactions of his readers crops up quite frequently in the literature (e.g. Fink 1971: 117). The predominant final ordering principle is provided by the thetic structure whose content is intended to expound a postulate of poetics (see Dammann 1990).

296

Part 3 An Experiment in Action

drawn from the objects treated by a work of art but rather psychologically inscribed in the mind of the author who handles them. Thus we read of the writer in Torquato Tasso (1790): ‘Only by favour of the Muses will / So many rhymes cohere and harmonize: / And this one impulse dominates his spirit, / To make his poem one harmonious whole./ Not to pile legend upon legend, jumbled, / To charm and entertain, but leave you then / With mere loose words that like miracles fade.’ (Torquato Tasso I.ii.272–77, Goethe 1988.) The metaphorical approach to the problem of closure is later complemented by an explicitly theoretical one during a discussion of the Aristotelian definition of tragedy and the role played by the concept of catharsis. Here, Goethe firmly underlines his belief that the effect of a text on the recipient is a secondary consequence of the formal closure which is a universal feature of all successful compositions. In “On Interpreting Aristotle’s Poetics” (1827), we read: How could Aristotle possibly have thought of the effect—indeed the delayed effect—which a tragedy might have on the spectator, when in his characteristically objective way he was in fact discussing the structure of the tragedy? Impossible! He states clearly and correctly: once tragedy has gone through a series of events arousing pity and fear, it must conclude on stage, before our eyes, with the neutralization and reconciliation of such emotions. Goethe 1986:198

Goethe argues that the aesthetic design process should be concluded ‘on stage,’ by which he means inside fiction and before anything is received or has any effect on anyone. In concrete terms, this means that aesthetic construction must be completed in the action plan of a narrative, and that brings us to the question of what Goethe meant in his much-quoted letter to Schiller of 17 October 1795, where he writes that the Conversations should tail off, ‘one might say, into infinity’ with the ‘work of the imagination’ that is the Fairy Tale (Goethe 1981, 6:606; my translation). The latter phrase is actually a quotation from Goethe’s own words in the final fragment of the frame before the beginning of the Fairy Tale in the Conversations. The passage in question is significant and is therefore reproduced in its entirety below. ‘Can’t you,’ Carl said to the old gentleman, ‘tell us a fairy tale? Imagination is a wonderful faculty, but I don’t like to see it applied to what has really happened. The ethereal forms it creates are welcome to us as a breed all their own; united to truth, it usually brings forth only monsters and then, it seems to me, generally stands in opposition to common sense and reason. It must, I think, attach itself to no object, it must force no object upon us; it should, in producing art, simply play upon us as music does, move us within ourselves, and indeed in such a way, that we forget there is anything outside us that generates this emotion.’

3.3 Explaining the Conversations

297

‘Do not continue,’ said the old man, ‘to elaborate your demands on works of the imagination in more detail. It is also appropriate to the enjoyment of such works that we enjoy them without demands.’ Goethe 1989:69

When we interpret fragments such as this from the frame of the Conversations, it cannot be denied that we must also consider the characters who utter the statements they contain—not all the participants in the discourse of the storytelling society have equal status, and not all contributions to the discourse have equal weight.43 On the other hand, the passage with which we are concerned is obviously marked out from the previous contributions by its conspicuous position as the final part of the discourse whose culmination it can be seen to represent. Previously, Carl has repeatedly come to our attention because of his rather poorly thought-out contributions, but in this case the Abbé does not criticize the content of the norms put forward by Carl; all he says is that he does not want to see them elucidated in any more detail. There are good reasons for this rather puzzling behaviour by the Abbé. His objection is superficially legitimized in terms of the aesthetics of effect—he seems to be expressing concern about the audience’s enjoyment because it is apparently under some kind of threat. But Goethe’s alter ego is just as worried about the potential consequences of Carl’s words for production theory—the narrator does not want the work of his imagination to be restricted by exact plans and prescriptive models. It is obvious, however, that Carl and the Abbé are agreed that the Fairy Tale which Carl has just anticipated will represent a new text type completely different from what has preceded it. Accordingly, but this time explicitly, Goethe describes the Fairy Tale as a ‘work of the imagination’ in his letter to Schiller, thereby setting it apart a second time from the frame and the other narratives in the Conversations. If we now turn to the genre definition which is contained in Carl’s words, we can see that it is an essentially negative one. The exemplary fairytale should reference neither empirical happenings (‘what has really happened’) nor empirical persons and objects, be those objects concrete or abstract (‘It must…attach itself to no object’). In anticipation of the Abbé’s reply, Carl legitimizes his ex negativo definition in two ways. Positively, it is justified by the stipulation of an aesthetic effect which the fairytale 43

Brown makes clear that this reservation must be borne in mind when considering the passage quoted above. In it, statements of literary theory are uttered primarily by Carl, the very same figure who otherwise tends to be the representative of a deep-seated fundamentalist positivism (Brown 1975:20). Conrady (1985:111f.) also reminds us that we must never ignore the obvious discrepancy between the frame characters’ theoretical statements on poetics and aesthetics and the actual poetics of the narratives they recount in practice.

298

Part 3 An Experiment in Action

should create (the work of the imagination should ‘play upon us as music does’); negatively, it rests on a philosophical criticism (combining an aesthetic work of the imagination with empirical truth is dangerous because it produces ‘monsters’ and thus ‘stands in opposition to common sense and reason’). It has become customary for critics to treat this as a dictum of poetics, but we shall concentrate on its epistemological content. In this latter context, it is important to identify what exactly it is that Goethe finds methodologically monstrous about the hypothetical combination of empiricism and the work of the imagination. The monster turns out to be nothing other than the other side of the need for closure that Goethe treats as legitimate in aesthetic contexts but very dangerous nonetheless should it be applied to objective empirical objects. As early as 1792, Goethe shows the influence of Kantian philosophy when he uses his scientific essay “The Experiment As Mediator between Object and Subject” to warn against the uncritical application of the ideal of aesthetic closure to the observation of external nature. The warning is all the more necessary because the vehicle of such a projection is disconcertingly familiar to every one of us—it is the language, the terminology without which the conceptual definition of hypotheses (‘Vorstellungsarten’) would not be possible: Man takes more pleasure in the idea than in the thing; or rather, man takes pleasure in a thing only insofar as he has an idea of it. The thing must fit his character, and no matter how exalted his way of thinking, no matter how refined, it often remains just a way of thinking, an attempt to bring several objects into an intelligible relationship which, strictly speaking, they do not have. Thus the tendency to hypotheses, theories, terminologies, and systems, a tendency altogether understandable since it springs by necessity from the organization of our being. Goethe 1988:14

The psychological ‘tendency’ which Goethe describes here is rooted in the same ‘organization of our being’ that Kant’s Critique of Pure Reason treats under the conditions for the possibility of cognition. Although he considers this organization necessary and unavoidable, Goethe does not take the defeatist view that we must therefore abandon all our observations and conceptualizations to the mercy of subjectivism. In both aesthetics and the study of nature, the risks posed by the monstrous can be minimized if we respond to the conditions for the possibility of cognition by designing and applying a cognitive process which uses rule-governed methods. And it turns out that the Fairy Tale discusses the form that just such a process would need in an aesthetic context. It demands that producers refrain from using referential speech and that recipients show restraint in placing ‘demands on works of the imagination.’ Interestingly, however, the

3.3 Explaining the Conversations

299

Fairy Tale was not the first time that Goethe had treated this theme; he had covered it before in his scientific essay “The Experiment As Mediator between Object and Subject,” which considers the modes of perception of natural science and aesthetics. It treats them as the opposing poles in a methodological dichotomy: Thus in scientific matters we must do the reverse of what is done in art. An artist should never present a work to the public before it is finished because it is difficult for others to advise or help him with its production. Once it is finished, however, he must consider criticism or praise, take it to heart, make it a part of his own experience, and thereby develop and prepare himself for new works. In science, on the other hand, it is useful to publish every bit of empirical evidence, even every conjecture; indeed, no scientific edifice should be built until the plan and materials of its structure have been widely known, judged and sifted. Goethe 1988:13

Taking this methodological distinction as our background, we can now identify a new pattern behind the arrangement of the textual components which make up the Conversations. It is clear that the Abbé is following the model of aesthetic perception when he withdraws alone for his usual walk to plan the Fairy Tale before the last storytelling session. It is no less obvious that the Abbé, Carl, and Fritz, as narrators of the preceding anecdotes and novellas, have shown themselves ready to apply the very same ‘method of co-operative endeavour’ (‘Methode mit Mehreren zu arbeiten’) which Goethe the scientist himself admitted to finding too agreeable to relinquish (Goethe 1981, 13:13; my translation). Furthermore, it is perfectly consistent with the logic of the Conversations as a quasiscientifically configured experiment in serial narration that their material is taken predominantly from older, established traditions, for the value of an experiment lies in the fact that, simple or compound, it can be reproduced at any time given the requisite preparations, apparatus, and skill. After assembling the necessary materials we may perform the experiment as often as we wish.…As worthwhile as each individual experiment may be, it receives its real value only when united or combined with other experiments. Goethe 1988:13

The analogy with which we are concerned should be clear enough and does not need to be forced any further. Even if we can now see the Conversations as based on the formal model of a series of scientific experiments, their overall form is still that of an aesthetic experiment in illusion whose content is in no way restricted to an abstract epistemological problem (the human interpretive compulsion). They are also concerned with more concrete questions of ethics (renunciation and mutual assistance as requirements for the restoration of a social culture), aesthetics and morality (preparation for

300

Part 3 An Experiment in Action

the restored society through narration and discourse), and poetics. This diffuse thematic focus could provide a certain justification for continuing to treat the Conversations as a cycle, were it not for the rather different conclusion which presents itself from a formal point of view. Although the use of the Fairy Tale ending may be irritating when measured against the standards of a cyclically designed narrative, it has a perfectly reasonable epistemological function. The ending of the Conversations not only shows that the experimental series is an open one but also reflects with critical irony the results that the experiment would have if the series were to be (erroneously) assumed closed and thus subjected to an all-encompassing interpretation. It was just such a mistaken assumption that Goethe had in mind in his methodological essay “Judgment Through Intuitive Perception”; in this essay he criticized the habit to make ‘smug, hurried, thoughtless pronouncements based on one or two facts [and rush] to hasty conclusions by trying to impose on the objective world some notion that passes through one’s head’ (Goethe 1988:31). The result of such a misinterpretation would at best be an aesthetically valid idea and hypothesis whose usefulness is a matter of all or nothing—perfectly legitimate as far as aesthetic enjoyment is concerned, but counterproductive with respect to the objective perception of things. The process of objective perception is intrinsically endless precisely because it must proceed methodically towards a complete explication of its object. It is perfectly acceptable for the process of aesthetic perception, on the other hand, to propose a fragmentary sketch of meaning in so far as its methods are expressly designed to reach imperfect conclusions. Like so much we have discussed, this dichotomy too is treated in Goethe’s 1792 essay, which we quote again here: To follow every single experiment through its variations is the real task of the scientific researcher. His duty is precisely the opposite of what we expect from the author who writes to entertain. The latter will bore his readers if he does not leave something to the imagination, while the former must always work as if he wished to leave nothing for his successors to do. Of course, this disproportion between our intellect and the nature of things will soon remind us that no one has gifts enough to exhaust the study of any subject. Goethe 1988:16

The Logic of action in the Conversations In this chapter, we have described a reading of the Conversations which sees them as a series of narrative experiments which systematically test a variety of ways in which narration can be used to create a coherent action logic. The reading that we have proposed ties in with the findings

3.3 Explaining the Conversations

301

of our analysis of the corresponding event data in EpiTest as outlined in chapter 3.2. The series of experiments produces two results, one aesthetic and one philosophical. As for the former, we are presented with a positive example of the standard novella in the form of the Ferdinand narrative. Ferdinand (along with the Fairy Tale and ignoring the frame narrative) represents Goethe’s only individual creation in the Conversations; it therefore demonstrates the success he could achieve even when not working from inherited sources. In Ferdinand, the fictional action creates syntagmatic coherence and is ideally co-ordinated with the arrangement of the narrative’s thematic material. It is an exemplary personal history and, both aesthetically and philosophically, attains the highest level of constructive and aesthetic synthesis that Goethe believed was reasonably possible in a narrated happening, be it fictive or real (e.g. historical). The second conclusion to be drawn from the experiment is that any narrative which has a higher level of event-related synthesis than Ferdinand will either report a banal and morally indifferent occurrence, report an occurrence in an aesthetically banal manner (e.g. the Bassompierre anecdote), or show every sign of claiming to provide an all-encompassing historiographical picture of the history of the world at large. From what we know of Goethe’s views on the philosophy of history, it is clear that he considered it inherently impossible to provide a historiographical synthesis of the history of the world at large; he would thus have treated any apparent such synthesis as a figment of the imagination. This scepticism is of particular importance for the Conversations because their frame refers explicitly to the context of the not-too-distant contemporary past, which the narrator describes simply as a succession of most unhappy days with the gravest of consequences for Germany, Europe and indeed the world at large. Thus, the attempt to provide an all-encompassing synthetic historiographical picture of empirical events meets with aesthetic caricature on the one hand and philosophical and methodological criticism on the other. The primary function of both these responses, caricature and criticism, is to dialectically overcome the principle of mechanical syntagmatic ordering which has been treated as objectively true since the Enlightenment. Goethe proposes replacing it with an ordering principle designed on an anthropological basis. Its truth is subjectively proven, and it identifies primarily paradigmatic arrangements in empirical and imaginary objects. Herein lies the significance of the Fairy Tale for the experiment and argument of the Conversations. The deficiency of its action logic, intentionally designed and effectively produced, makes it a deliberately artificial narrative counterpart to the exemplary personal history which precedes it. The Fairy

302

Part 3 An Experiment in Action

Tale presents a fine picture of the history of the world at large—but it is a world whose individual events are patently not organized according to the rules of a historical or psychological causal Newtonian mechanics which it is possible to reconstruct syntagmatically. The symbolic references are what count in the Fairy Tale and it is they that make it aesthetically and philosophically important. In no way are these references anchored in the things and events themselves, and in no way do they develop entelechially out of them. The recipient produces the symbolic references by arranging particular features of the quasi-empirical constituents of the Fairy Tale world into paradigms as an when he subjectively notices them. Seen in this way, we can think of the Fairy Tale as an aesthetic mirror held up before the series of quasi-scientific experiments that precedes it. It is able to perform this function because it renounces any claim to external reference and tightly embraces hermetic self-referentiality instead. If the things which are brought together in the world of the Fairy Tale mean nothing outside it, the structures manifested in its many different arrangements, that is to say, in the narrated action, are anything but products of aesthetic arbitrariness. Goethe may not have fully understood his Kant, but he certainly knew how to make good use of what he thought he understood. The Conversations begin as an openly self-reflexive and transparent narrative experiment and end opaquely with the Fairy Tale. In this way, they show that Goethe’s study of the ‘good man,’ even if somewhat mischievous, was not without its benefits. What was it that he said about Kant again? But then, after he has succeeded in driving us to the wall, to the verge of despair in fact, he makes the most liberal statements, and leaves it to us to decide how to enjoy the freedom he allows us. Goethe 1988:31

I propose that we should take these words as the epistemological and aesthetic programme behind the Conversations and the formal inhomogeneity that is as perplexing to the reader as it was carefully planned by the author. Our text is designed to experiment with both philosophical action and aesthetic action; the two are intertwined and cannot be separated.

3.4 Conclusion It is clear that there is every reason to read the Conversations as an experimental series in more than one sense. The content of the text treats issues of moral didactics and the philosophy of history, and its aesthetics engage with themes from genre poetics and the aesthetics of production and effect. Thus, the formal principle which underlies the Conversations has two aspects: the work is both a thematically focused cycle and a series of experiments in narrative form. The result of this format is that the scientific epistemology of Goethe the enquirer is confronted by an aesthetic epistemology defended by Goethe the writer. Furthermore, the text which explores these issues is only superficially based on a conceptual model borrowed from the novella cycle. The real conceptual foundation of text is the idea of a series of experiments. In this particular instance, the experiments are designed to investigate what is perhaps the most profound epistemological problem of all, the question of how humans turn what they perceive in a sequential order into a synthetic whole, be it a personal history, a history of the world at large, or an action. As a narrator, Goethe replies that the whole cannot be retold and must be construed. As a natural scientist, on the other hand, he replies that the whole is an object of perception which it is inherently impossible to grasp in its totality but which can nonetheless be approached in controlled steps of approximation. These two same arguments can be applied self-reflexively to the present study, which has, in its own modest way, been an attempt to use new methods to develop, test, and discuss a constructive concept of action and a practical set of narratological tools which can be combined to mediate between object and subject and deepen our understanding of narrative texts. The object of our study was always going to be a narrow one, even when taken in its broadest theoretical sense as the logic which governs how narrated events are bound together in synthetic actions. Even so, it may have come as something of a surprise to find out in this third and final part just how narrow the scope of our investigation actually is—our ideas have turned out to involve just one small aspect of the unexpected philosophical diversity of the text with which we have illustrated them. In a sense, though, our study’s apparent insignificance is the very proof of its success—the true and rewarding depth of the text would have remained unknown had it not been for our theory of action, which has shown the apparent shortcomings of the Conversations to be part of a productive experiment in action narration.

304

Part 3 An Experiment in Action

Literary critics have no doubt been annoyed by our enthusiasm for formal precision and our relative neglect of the philosophy of the Conversations, while empiricists have most probably been shocked by our attempt to bridge the gap between computer-aided narratological textual analysis and philosophical textual hermeneutics. What are we to say to either group of readers? It would be tempting indeed to demonstrate the value and relevance of our methods by drawing everything together in a perfect conclusion—but that is not what truly worthwhile research in literature and philosophy is about. The study that ends with a self-congratulatory QED says more about the problems it has ignored than those it has solved, for true progress can only be achieved by broadening our horizons, asking new questions, and experimenting. The results may not always be what we expect or what we want, but they will always tell us more than we can discover by selectively using the data to prove what was never really in doubt anyway. Vladimir Propp, the founding father of structuralist criticism, probably had a twinkle in his eye when he placed a certain quotation from Goethe’s “Betrachtung über Morphologie” at the beginning of his Morphology of the Folktale. The motto reads that ‘even an unsuccessful experiment…can still be both elegant and useful’ (my translation) and was originally formulated by Goethe during the period when he was deliberating the methodological principles behind his own studies of morphology and plant physiology in the mid-1790s.44 With equal thanks to Propp and the patient reader, I now bring my own experiment to a close with a not dissimilar passage from Goethe’s 1792 essay “The Experiment As Mediator between Object and Subject,” where the author of the Conversations of German Refugees states that once sequential evidence of higher sort is assembled, however, our intellect, imagination and wit can work upon it as they will; no harm will be done, and, indeed, a useful purpose will be served.…Everyone will then be free to connect them in his own way, to form them into a whole which brings some measure of delight and comfort to the human mind. This approach keeps separate what 44

The commentary in Goethe 1981, 13:585f. suggests that the remark was made shortly after 1794. The German original is found in Goethe 1981, 13:127 and reads in context: ‘Sie [die Morphologie] hat den großen Vorteil, daß sie aus Elementen besteht, die allgemein anerkannt sind, daß sie mit keiner Lehre im Widerstreite steht, daß sie nichts wegzuräumen braucht, um sich Platz zu verschaffen, daß die Phänomene, mit denen sie sich beschäftigt, höchst bedeutend sind, und daß die Operationen des Geistes, wodurch sie die Phänomene zusammenstellt, der menschlichen Natur angemessen und angenehm sind, so daß auch ein fehlgeschlagener Versuch darin selbst noch Nutzen und Anmut verbinden könnte’ (Goethe 1981, 13:127). This quote (Propp 1928:5) as well as others included in the Russian text were quietly omitted in the English translation (Propp 1984).

3.4 Conclusion

305

must be kept separate; it enables us to increase the body of evidence much more quickly and cleanly. Goethe 1988:17

If there was one thing that the concepts, tools, and approach developed and discussed in this book were meant to do, it was to ‘keep separate what must be kept separate.’ This may not strike us as a dramatic achievement. On the other hand the phenomenon of the action narrative investigated in this study is one that continually tempts us to forego analytical rigour and rush to synthetic conclusions. Falling prey to this temptation may of course be at the very core of the aesthetic pleasure which we derive from action narratives as ordinary readers. Indeed, as ordinary readers there is no reason why we should forsake this experience—but as scholars of literary studies and narratology we will benefit from delaying the pleasure of projecting synthetic wholeness onto narrated events until we are satisfied that we have sufficiently analysed, described and defined what may eventually be interpreted as a narrative’s action.

References Abel 1985 Abel, Günter: “Einzelding- und Ereignis-Ontologie.” In: Zeitschrift für philosophische Forschung 39, 1985:157–185. Arendt 1979 Arendt, Hanna: “Das Handeln. Die Enthüllung der Person im Handeln und Sprechen.” In: Lenk, Hans (ed.): Handlungstheorien interdisziplinär. Vol. II.1: Handlungserklärungen und philosophische Handlungsinterpretation (Kritische Information 63). München 1979:13–87. Aristoteles 1982 Aristoteles: Poetik. Griechisch/Deutsch. Übersetzt und hg. von Manfred Fuhrmann. Stuttgart 1982. Aristotle 1984 Aristotle: “Poetics.” In: Barnes, Jonathan (ed.): The Complete Works of Aristotle. The Revised Oxford Translation (Bollingen Series 71). Vol. 2, Princeton 1984:2316–2340. Bachmann-Medick 1989 Bachmann-Medick, Doris: Die ästhetische Ordnung des Handelns. Moralphilosophie und Ästhetik in der Popularphilosophie des 18. Jahrhunderts. Stuttgart 1989. Bal 1992 Bal, Mieke: “Review Article: Narratology and the Rhetoric of Trashing.” In: Comparative Literature 44, 1992:293–306. Barsch 1992 Barsch, Achim: “Handlungsebenen des Literatursystems.” In: SPIEL 11/1, 1992:1–23. Barthes 1985 Barthes, Roland: “The Greek Theater.” In: R.B.: The Responsibility of Forms. Critical Essays on Music, Art, and Representation. Translated by Richard Howard. Oxford 1985:63–88. Baumgartner 1886 Baumgartner, Alexander: Göthe und Schiller. Weimars Glanzperiode. Freiburg/Br. 1886. Bauschinger 1985 Bauschinger, Sigrid: “Unterhaltungen deutscher Ausgewanderten.” In: Lützeler, Paul Michael (ed.): Goethes Erzählwerk. Interpretationen. Stuttgart 1985:134–167. Bauschinger 1997 Bauschinger, Sigrid: “Unterhaltungen deutscher Ausgewanderten.” In: Witte, Bernd et al. (eds.): Goethe-Handbuch. Vol. 3, Stuttgart/Weimar 1997:232–254.

308

References

Beasley 1988 Beasley, Jerry C.: “Life’s Episodes: Story and its Form in the Eighteenth Century.” In: Uphaus, Robert W. (ed.): The Idea of the Novel in the Eighteenth Century. Studies in Literature, 1500–1800. Vol. 3, East Lansing, MI 1988:21–46. Beddow 1984 Beddow, Michael: “‘Da wird hinten immer fortgespielt’: Un-ended Plots in Goethe’s ‘Die Wahlverwandschaften’.” In: Publications of the English Goethe Society, New Series 53, 1984:1–19. Belknap 1985 Belknap, Robert L.: “Plotlets and Schemes.” In: Jackson, Robert Louis/Rudy, Stephen (eds.): Russian Formalism: A Retrospective Glance. A Festschrift in Honor of Victor Erlich (Yale Russian and East European Publications 6). New Haven 1985:242–251. Berkemann 1984 Berkemann, Jörg: “‘Handlung’ in der Rechtswissenschaft.” In: Lenk, Hans (ed.): Handlungstheorien interdisziplinär. Vol. III.2: Verhaltenswissenscha ftliche und psychologische Handlungstheorien (Kritische Information 64). München 1984:806–847. Beyer 1981 Beyer, Wilhelm Raimund: “Goethe im Themenfeld von Flüchtlingsgesprächen.” In: Goethe-Jahrbuch 98, 1981:156–178. Bittner 1986 Bittner, Rüdiger: “Handlungen und Wirkungen.” In: Prauss, Gerold/Bittner, Rüdiger (eds.): Handlungstheorie und Transzendentalphilosophie. Frankfurt/ Main 1986:13–26. Bloomfield 1980 Bloomfield, Morton W.: “Episodic Juxtaposition or the Syntax of Episodes.” In: Greenbaum, Sidney/Leech, Geoffrey/Svartvik, Jan (eds.): Studies in English Linguistics. For Randolph Quirk. London/New York 1980:210–220. Bourdieu 1979 Bourdieu, Pierre: Entwurf einer Theorie der Praxis auf der ethnologischen Grundlage der kabylischen Gesellschaft. Übersetzt von Cordula Pialoux und Bernd Schwibs. Frankfurt/Main 1979. Bourdieu 1998 Bourdieu, Pierre: Praktische Vernunft. Zur Theorie des Handelns. Frankfurt/ Main 1998. Boyle 1993 Boyle, Nicholas: “Geschichtsschreibung und Autobiographik bei Goethe (1810-1817).” In: Goethe-Jahrbuch 110, 1993:163–172. Brandt 1983 Brandt, Helmut: “Entsagung und Französische Revolution. Goethes Prokurator- und Ferdinand-Novelle in weiterführender Betrachtung.” In: Impulse. Aufsätze, Quellen, Berichte zur deutschen Klassik und Romantik 6, 1983:37–65. Bräutigam 1977 Bräutigam, Bernd: “Die ästhetische Erziehung der deutschen Ausgewanderten.” In: Zeitschrift für deutsche Philologie 96, 1977:508–539.

References

309

Bremond 1964 Bremond, Claude: “Le Message narratif.” In: Communications 4, 1964: 4–32. Bremond 1966 Bremond, Claude: “La Logique des possibles narratifs”. In: Communications 8, 1966:60–76. Bremond 1973 Bremond, Claude: Logique du récit. Paris 1973. Brennenstuhl 1975 Brennenstuhl, Waltraud: Handlungstheorie und Handlungslogik. Vorbereitungen zur Entwicklung einer sprachadäquaten Handlungslogik (Monographien Linguistik und Kommunikationswissenschaft 19). Kronenberg/Ts. 1975. Bringsjord/Ferrucci 2000 Bringsjord, Selmer/Ferrucci, David A.: Artificial Intelligence and Literary Creativity. Inside the Mind of BRUTUS, a Storytelling Machine. Mahwah, NJ 2000. Brooks 1984 Brooks, Peter: Reading for the Plot. Design and Intention in Narrative. Oxford 1984. Brown 1975 Brown, Jane K.: Goethe’s Cyclical Narratives: “Die Unterhaltungen deutscher Ausgewanderten” and “Wilhelm Meisters Wanderjahre”. Chapel Hill 1975. Bubner 1976 Bubner, Rüdiger: Handlung, Sprache und Vernunft. Grundbegriffe praktischer Philosophie. Frankfurt/Main 1976. Bubner 1993 Bubner, Rüdiger: “Die Gesetzlichkeit der Natur und die Willkür der Menschheitsgeschichte. Goethe vor dem Historismus.” In: Goethe-Jahrbuch 110, 1993:135–146. Calvino 1981 Calvino, Italo: If On a Winter’s Night a Traveler. Trans. William Weaver. New York and San Diego 1981 [1979]. Casati/Varzi 2002 Casati, Roberto/Varzi, Achille: “Events.” In: Zalta, Edward N. (ed.): The Stanford Encyclopedia of Philosophy (Fall 2002 Edition). (seen 19.04. 2003). Chatman 1988 Chatman, Seymour: “Discourse: Nonnarrated Stories.” In: Hoffman, Michael/Murphy, Patrick D. (eds.): Essentials of the Theory of Fiction. Durham, North Carolina 1988:366–379. Chatman 1990 Chatman, Seymour: “What Can We Learn from Contextualist Narratology?” In: Poetics Today 11/2, 1990:309–328. Cohn 1990 Cohn, Dorrit: “Signposts of Fictionality: A Narratological Perspective.” In: Poetics Today 11/4, 1990:775–804.

310

References

Conrady 1985 Conrady, Karl Otto: Goethe. Leben und Werk. Zweiter Band. Summe des Lebens. Frankfurt/Main 1985. Coste 1989 Coste, Didier: Narrative as Communication. Minneapolis 1989. Culler 1980 Culler, Jonathan: “Fabula and Sjuzhet in the Analysis of Narrative.” In: Poetics Today 1/3, 1980:27–37. Dammann 1982 Dammann, Günter: “Episode.” In: Ranke, Kurt et al. (eds.): Enzyklopädie des Märchens. Handwörterbuch zur historischen und vergleichenden Erzählforschung. Vol. 4, Berlin/New York 1982:69–73. Dammann 1990 Dammann, Günter: “Goethes ‘Unterhaltungen deutscher Ausgewanderten’ als Essay über die Gattung der Prosaerzählung im 18. Jahrhundert.” In: Zimmermann, Harro (ed.): Der deutsche Roman der Spätaufklärung. Fiktion und Wirklichkeit. Heidelberg 1990:1–24. Quoted from the corrected version: (seen 25.04.2003). Danto 1976 Danto, Arthur C.: “Vorstellungselemente in menschlichen Handlungen.” In: Neue Hefte für Philosophie 9, 1976:1–16. Davidson 1971 Davidson, Donald: “Agency.” In: Binkley, Robert et al. (eds.): Agent, Action, and Reason. Oxford 1971:3–25. Davidson 1977 Davidson, Donald: “Handeln.” In: Meggle, Georg (ed.): Analytische Handlungstheorie. Handlungsbeschreibungen. Frankfurt/Main 1977:282–307. Dodd 1981 Dodd, William Nigel: “Aspects of the relationship between ‘episode’ and fabula in dramatic texts.” In: Lingua et Stile 16, 1981:471–479. Doležel 1980 Doležel, Lubomír: “Narrative Semantics and Motif Theory.” In: Csuri, Karel (ed.): Literary Semantics and Possible Worlds/Literatursemantik und Mögliche Welten (Studia Poetica 2). Szeged 1980:32–43. Douglas 1997 Douglas, Graham: “Greimas’ Semiotic Square and Greek and Roman Astrology.” In: Semiotica 114/1–2, 1997:1–19. Eco 1979 Eco, Umberto: The Role of the Reader. Explorations in the Semiotics of Texts. Bloomington/London 1979. Eco 1981 Eco, Umberto: Zeichen. Einführung in einen Begriff und seine Geschichte. Frankfurt/Main 1981. Ellis 1986 Ellis, John M.: “How Seriously Should We Take Goethe’s Definition of the Novelle?” In: Goethe Yearbook. Publications of the Goethe Society of North America 3, 1986:121–123. Else 1957 Else, Gerald F.: Aristotle’s Poetics: The Argument. Leiden 1957.

References

311

Engel 1774 Engel, Johann Jakob: “Über Handlung, Gespräch und Erzählung.” In: Voss, Ernst Theodor (ed.): Über Handlung, Gespräch und Erzählung. Faksimiledruck der ersten Fassung 1774 aus der “Neuen Bibliothek der Schönen Wissenschaften und der freyen Künste”. Hg. und mit einem Nachwort versehen von Ernst Theodor Voss (Sammlung Metzler 37). Stuttgart 1964. EWD 1993 Etymologisches Wörterbuch des Deutschen. Berlin 1993. Fichte 1979 Fichte, Johann Gottlieb: “Einleitung.” In: Medicus, Fritz (ed.): Grundlage des Naturrechts nach Prinzipien der Wissenschaftslehre (Philosophische Bibliothek 256). Hamburg 1979:1–16. Fichte 2000 Fichte, Johann Gottlieb: Foundations of Natural Right. According to the Principles of the Wissenschaftslehre. Ed. by Frederick Neuhouser. Translated by Michael Baur (Cambridge Texts in the History of Philosophy). Cambridge 2000. Fink 1971 Fink, Gonthier-Louis: “Das ‘Märchen’. Goethes Auseinandersetzung mit seiner Zeit.” In: Jahrbuch der Goethe-Gesellschaft, Neue Folge 33, 1971: 96–122. Fludernik 1996 Fludernik, Monika: Towards a ‘Natural’ Narratology. London/New York 1996. Forster 1974 Forster, E.M.: Aspects of the Novel. London 1974. Frank 1984 Frank, Manfred: “Das ‘fragmentarische Universum’ der Romantik.” In: Dällenbach, Lucien/Nibbrig, Christiaan L. Hart (eds.): Fragment und Totalität. Frankfurt/Main 1984:212–224. Fricke 1964 Fricke, Gerhard: “Zu Sinn und Form von Goethes ‘Unterhaltungen deutscher Ausgewanderten’.” In: Müller-Seidel, Walther/Preisendanz, Wolfgang (eds.): Formenwandel. Festschrift zum 65. Geburtstag von Paul Böckmann. Hamburg 1964:273–293. Fuhrmann 1992 Fuhrmann, Manfred: Dichtungstheorie der Antike. Aristoteles, Horaz, ‘Longin’. Darmstadt 1992. Gabriel 1975 Gabriel, Gottfried: Fiktion und Wahrheit. Eine semantische Theorie der Literatur. Stuttgart 1975. Gaede 1977 Gaede, Friedrich: Poetik und Logik. Zu den Grundlagen der literarischen Entwicklung im 17. und 18. Jahrhundert. Bern/München 1977. Gaier 1987 Gaier, Ulrich: “Soziale Bildung gegen ästhetische Erziehung. Goethes Rahmen der ‘Unterhaltungen’ als satirische Antithese zu Schillers ‘Ästhetischen Briefen’ I-IX.” In: Bachmaier, Helmut/Rentsch, Thomas (eds.): Poetische Autonomie? Zur Wechselwirkung von Dichtung und Philosophie in der Epoche Goethes und Hölderlins. Stuttgart 1987:207–272.

312

References

Galloway 1979 Galloway, Patricia: “Testing a Theory of Narrative Structure by Computer.” In: Ager, D.E ./Knowles, F.E./Smith, Joan (eds.): Advances in Computer-Aided Literary and Linguistic Research. Proceedings of the Fifth International Symposium on Computers in Literary and Linguistic Research held at the University of Aston in Birmingham, UK from 3–7 April 1978. Birmingham 1979:53–57. Galloway 1983 Galloway, Patricia: “Narrative Theories as Computational Models.” In: Computers and the Humanities 17, 1983:169–174. Geißler 1979 Geißler, Rolf: “Zur Einheit von Goethes ‘Unterhaltungen deutscher Ausgewanderten’.” In: Literatur für Leser 1, 1979:33–44. Genette 1980 Genette, Gérard: Narrative Discourse. Translated by Jane E. Levin. Foreword by Jonathan Culler. Oxford 1980. Genette 1982 Genette, Gérard: Figures of Literary Discourse. Translated by Alan Sheridan. Introduction by Marie-Rose Logan. New York 1982. Genette 1988 Genette, Gérard: Narrative Discourse Revisited. Ithaca, NY 1988. Genette 1990 Genette, Gérard: “Fictional Narrative, Factual Narrative.” In: Poetics Today 11/4, 1990:755–774. Gerhardt 1986 Gerhardt, Volker: “Handlung als Verhältnis von Ursache und Wirkung. Zur Entwicklung des Handlungsbegriffs bei Kant.” In: Prauss, Gerold (ed.): Handlungstheorie und Transzendentalphilosophie. Frankfurt/Main 1986:98–131. Goethe 1981 Goethe, Johann Wolfgang von: Goethes Werke. Hamburger Ausgabe in 14 Bänden. Hg. von Erich Trunz. München 1981ff. Goethe 1986 Goethe, Johann Wolfgang von: Essays on Art and Literature. Ed. by John Gearey. Translated by Ellen von Nardroff and Ernest H. von Nardroff (Goethe’s Collected Works 3). New York 1986. Goethe 1988 Goethe, Johann Wolfgang von: Scientific Studies. Edited and translated by Douglas Miller (Goethe’s Collected Works 12). New York 1988. Goethe 1989 Goethe, Johann Wolfgang von: Conversations of German Refugees. Translated by Jan van Heurck in Cooperation with Jane K. Brown. Wilhelm Meister’s Journeyman Years or The Renunciants. Translated by Krishna Winston. Ed. by Jane K. Brown (Goethe’s Collected Works 10). New York 1989. Goethe 1995 Goethe, Johann Wolfgang von: Goethes Werke. Weimarer Ausgabe (“Sophienausgabe”). CD-ROM, Cambridge 1995. Goldstein 1906 Goldstein, Moritz: Die Technik der zyklischen Rahmenerzählungen Deutschlands. Von Goethe bis Hoffmann. Berlin 1906.

References

313

Goodman 1981 Goodman, Nelson: “The Telling and the Told.” In: Critical Inquiry 8/4, 1981:799–809. Gottschall 1855 Gottschall, Rudolf v.: Die deutsche Nationalliteratur in der ersten Hälfte des neunzehnten Jahrhunderts. Literaturhistorisch und kritisch dargestellt. Breslau 1855 [quoted in Gundolf 1920]. Grabes 1993 Grabes, Herbert: “Narrative Sequentiality and Semantic Transformation.” In: Foltinek, Herbert/Riehle, Wolfgang/Zacharasiewicz, Waldemar (eds.): Tales and ‘their telling difference’. Zur Theorie und Geschichte der Narrativik. Festschrift für Franz K. Stanzel (Anglistische Forschungen 221). Heidelberg 1993:25–41. Grant 1991 Grant, Michael: A Short History of Classical Civilization. London 1991. Graves/Frederiksen 1991 Graves, Barbara/Frederiksen, Carl H.: “Literary Expertise in the Description of a Fictional Narrative.” In: Poetics 20, 1991:1–26. Greimas 1966 Greimas, Algirdas Julien: Sémantique structurale. Recherche de méthode. Paris 1966. Greimas 1970 Greimas, Algirdas Julien: Du Sens. Paris 1970. Greimas 1971 Greimas, Algirdas Julien: Strukturale Semantik. Methodologische Untersuchungen. Autorisierte Übersetzung aus dem Französischen von Jens Ihwe. Braunschweig 1971. Greimas 1972 Greimas, Algirdas Julien: “Elemente einer narrativen Grammatik.” In: Blumensath, Heinz (ed.): Strukturalismus in der Literaturwissenschaft. Köln 1972:47–67. Greimas 1973 Greimas, Algirdas Julien: “Sur l’Histoire événementielle et l’histoire fondamentale.” In: Koselleck, Reinhart/Stempel, Wolf-Dieter (eds.): Geschichte – Ereignis und Erzählung (Poetik und Hermeneutik 5). München 1973: 139–153. Greimas 1987 Greimas, Algirdas Julien: On Meaning. Selected Writings in Semiotic Theory. Translation by Paul J. Perron and Frank H. Collins. Foreword by Fredric Jameson. Introduction by Paul J. Perron. Minneapolis 1987. Greimas 1989 Greimas, Algirdas Julien: “On Meaning.” In: New Literary History 20/3, 1989:539–550. Greimas/Ricœur 1989 Greimas, Algirdas Julien/Ricœur, Paul: “On Narrativity.” In: New Literary History 20/3, 1989:551–562. Grimm 1984 Grimm, Jacob/Grimm, Wilhelm: Deutsches Wörterbuch. München 1984. Gundolf 1920 Gundolf, Friedrich: Goethe. Berlin 1920.

314

References

Haidu 1983 Haidu, Peter: “The Episode as Semiotic Module in Twelfth-Century Romance.” In: Poetics Today 4/4, 1983:655–681. Hallstein 1997 Hallstein, Patricia: Die Zeitstruktur in narrativen Texten. Am Beispiel von E.T.A. Hoffmanns “Das Majorat” und Achim von Arnims “Die Majoratsherren”. München 1997. Harker 1989 Harker, John W.: “Information Processing and the Reading of Literary Texts.” In: New Literary History 20, 1989:465–481. Hasselberg-Weyandt 1985 Hasselberg-Weyandt, Wilfried: Das Behalten von Geschichten. Struktur und Inhalt (Psychologie 7). Münster 1985. Hauptmeier/Schmidt 1985 Hauptmeier, Helmut/Schmidt, Siegfried J.: Einführung in die empirische Literaturwissenschaft. Wiesbaden 1985. Hegel 1954 Hegel, Georg Wilhelm Friedrich: Sämtliche Werke. Jubiläumsausgabe in zwanzig Bänden. Hg. von Hermann Glockner et al. Stuttgart 1954. Hegel 1969 Hegel, Georg Wilhelm Friedrich: Ästhetik. Hg. von Friedrich Bassenge. Frankfurt/Main 1969. Hegel 1975 Hegel, G[eorg] W[ilhelm] F[riedrich]: Aesthetics. Lectures on Fine Art. Translated by T.M. Knox. 2 Vol.s, Oxford 1975. Hegel 1977 Hegel, Georg Wilhelm Friedrich: Phänomenologie des Geistes. Frankfurt/ Main 1977. Hegel 1977a Hegel, G[eorg] W[ilhelm] F[riedrich]: Phenomenology of Spirit. Translated by A.V. Miller with Analysis of the Text and Foreword by J.N. Findlay. Oxford et al. 1977. Hegel 1991 Hegel, G[eorg] W[ilhelm] F[riedrich]: Elements of the Philosophy of Right. Ed. by Allen W. Wood. Translated by H.B. Nisbet (Cambridge Texts in the History of Political Thought). Cambridge et al. 1991. Henel 1985 Henel, Heinrich: “Anfänge der deutschen Novelle.” In: Monatshefte 77/4, 1985:433–448. Herman 1997 Herman, David: “Scripts, Sequences, and Stories. Elements of a Postclassical Narratology.” In: PMLA 112/5, 1997:1046–1059. Herman 2000 Herman, David: “Narratology as a Cognitive Science.” In: Image[&]Narrative 1, 2000. (seen 19.04.2003). Herman 2001 Herman, David: “Narrative Theory and the Cognitive Sciences.” In: Narrative Inquiry 11, 2001:1–34.

References

315

Herman 2002 Herman, David: Story Logic. Problems and Possibilities of Narrative. Lincoln, Neb. 2002. Higginbotham et al. 2000 Higginbotham, James/Pianesi, Fabio/Varzi, Achille C. (eds.): Speaking of Events. Oxford 2000. Hoermann 1996 Hoermann, Roland: “Goethe’s Masked Masque in Das Märchen – Theatrical Anticipations of Romanticism’s Self-Reflexive Peril.” In: Bernd, Clifford A./Henderson, Ingeborg/McConnell, Winder (eds.): Romanticism and Beyond. A Festschrift for John F. Fetzer (California Studies in German and European Romanticism and in the Age of Goethe 2). New York et al. 1996:79–100. Hoffmann 1985 Hoffmann, Ernst Fedor: “Die Geschichte von der Sängerin Antonelli in Goethes ‘Unterhaltungen’ und ihre Quelle in der ‘Correspondance littéraire’.” In: Goethe-Jahrbuch 102, 1985:105–143. Honig 1925 Honig, Richard: Studien zur juristischen und natürlichen Handlungseinheit. Zugleich ein Beitrag zur Strafrechtsreform. Mannheim et al. 1925. Ingarden 1968 Ingarden, Roman: Vom Erkennen des literarischen Kunstwerks. Tübingen 1968. Iser 1970 Iser, Wolfgang: Die Appellstruktur der Texte. Konstanz 1970. Jacobs 1981 Jacobs, Jürgen: “Die deutsche Erzählung im Zeitalter der Aufklärung.” In: Polheim, Konrad (ed.): Handbuch der deutschen Erzählung. Düsseldorf 1981:56–71, 564–566. Jahn 1997 Jahn, Manfred: “Frames, Preferences, and the Reading of Third-Person Narratives: Towards a Cognitive Narratology.” In: Poetics Today 18/4, 1997:441–467. Jürgens 1956 Jürgens, Ilse: “Die Stufen der sittlichen Entwicklung in Goethes ‘Unterhaltungen deutscher Ausgewanderten’.” In: Wirkendes Wort 6, 1955/56: 336–340. Kafalenos 1997 Kafalenos, Emma: “Functions after Propp: Words to Talk About How We Read Narrative.” In: Poetics Today 18/4, 1997:468–494. Kalinowski 1981 Kalinowski, Georges: “Carré semiotique et carré logique.” In: Actes Sémiotiques – Le Bulletin du Groupe de Recherches sémio-linguistiques 17, 1981:5–9. Kant [1924] Kant, Immanuel: Kritik der reinen Vernunft. Wiesbaden n. y. (= “ehemalige Kehrbachsche Ausgabe i.d. Überarbeitung von Raymund Schmidt 1924, einschl. Register der 2. Auflage 1927”).

316

References

Kant 1998 Kant, Immanuel: Critique of Pure Reason. Translated and edited by Paul Guyer and Allen W. Wood. Cambridge 1998 (The Cambridge Edition of the Works of Immanuel Kant). Kauffmann 1993 Kauffmann, Clemens: Ontologie und Handlung. Untersuchungen zu Platons Handlungstheorie. Freiburg/München 1993. Kaulbach 1979 Kaulbach, Friedrich: “Kants Theorie des Handelns.” In: Lenk, Hans (ed.): Handlungstheorien interdisziplinär. Vol. II.2: Handlungserklärungen und philosophische Handlungsinterpretation (Kritische Information 63). München 1979:643–669. Käuser 1990 Käuser, Andreas: “Das Wissen der Anthropologie: Goethes Novellen.” In: Goethe-Jahrbuch 107, 1990:158–168. Kindt/Schmidt 1979 Kindt, Walther/Schmidt, Siegfried J.: “Motivationen und Aspekte einer empirischen Literaturwissenschaft.” In: Schmidt, Siegfried J. (ed.): Empirie in Literatur- und Kunstwissenschaft. München 1979:7–42. Kindt/Müller 1999 Kindt, Tom/Müller, Hans-Harald: “Der ‘implizite Autor’. Zur Explikation und Verwendung eines umstrittenen Begriffs.” In: Jannidis, Fotis/Lauer, Gerhard/Martinez, Matias/Winko, Simone (eds.): Rückkehr des Autors. Zur Erneuerung eines umstrittenen Begriffs. Tübingen 1999:273–288. Kindt/Müller 2002 Kindt, Tom/Müller, Hans-Harald: “Der ‘implizite Autor’. Ein Begriff zwischen Narratologie und Interpretationstheorie.” (seen 19.04.2003). Kleinschnieder 1971 Kleinschnieder, Manfred: Goethes Naturstudien. Wissenschaftstheoretische und -geschichtliche Untersuchungen. Bonn 1971. Koselleck 1973 Koselleck, Reinhart: “Geschichte, Geschichten und formale Zeitstrukturen.” In: Koselleck, Reinhart/Stempel, Wolf-Dieter (eds.): Geschichte – Ereignis und Erzählung (Poetik und Hermeneutik 5). München 1973:211–222. Koselleck 1973a Koselleck, Reinhart: “Ereignis und Struktur.” In: Koselleck, Reinhart/ Stempel, Wolf-Dieter (eds.): Geschichte – Ereignis und Erzählung (Poetik und Hermeneutik 5). München 1973:560–571. Koselleck 1993 Koselleck, Reinhart: “Goethes unzeitgemäße Geschichte.” In: GoetheJahrbuch 110, 1993:27–39. Kraft 1935 Kraft, Werner: “Von Bassompierre zu Hoffmannsthal. Zur Geschichte eines Novellenmotivs.” In: Revue de littérature comparée 15, 1935:481–490. Kunz 1968 Kunz, Josef (ed.): Novelle. Darmstadt 1968. Kurzweil 1999 Kurzweil, Ray: The Age of Spiritual Machines. New York 1999.

References

317

Lämmert 1955 Lämmert, Eberhard: Bauformen des Erzählens. Stuttgart 1955. Lämmert 1973 Lämmert, Eberhard: “Zum Wandel der Geschichtserfahrung im Reflex der Romantheorie.” In: Koselleck, Reinhart/Stempel, Wolf-Dieter (eds.): Geschichte – Ereignis und Erzählung (Poetik und Hermeneutik 5). München 1973:503–515. Landauer 1921 Landauer, Gustav: “Goethes Politik. Eine Ankündigung.” In: G.L.: Der werdende Mensch. Aufsätze über Leben und Schrifttum. Potsdam 1921: 138–154 [quoted in Voßkamp/Jaumann 1992]. Lausberg 1990 Lausberg, Heinrich: Elemente der literarischen Rhetorik. Eine Einführung für Studierende der klassischen, romanischen, englischen und deutschen Philologie. Ismaning 101990. Lebowitz 1985 Lebowitz, Michael: “Story-Telling as Planning and Learning.” In: Poetics 14, 1985:483–502. Lee 1992 Lee, Jie-Oun: Eine Möglichkeit generativer Narrativik. Versuch einer Anwendung des Greimasschen Narrativikmodells auf A. v. Humboldts “Kosmos”. Bochum 1992. Lehnert 1981 Lehnert, Wendy G.: “Plot Units and Narrative Summarization.” In: Cognitive Science 4, 1981:293–331. Lenk 1979 Lenk, Hans: “Handlung als Interpretationskonstrukt. Entwurf einer konstituenten- und beschreibungstheoretischen Handlungsphilosophie.” In: H.L. (ed.): Handlungstheorien interdisziplinär. Vol. II.2: Handlungserklärungen und philosophische Handlungsinterpretation (Kritische Information 63). München 1979:279–350. Lenk 1993 Lenk, Hans: Interpretationskonstrukte: zur Kritik der interpretatorischen Vernunft. Frankfurt/Main 1993. Lenk 1994 Lenk, Hans: Von Deutungen zu Wertungen. Eine Einführung in aktuelles Philosophieren. Frankfurt/Main 1994. Lenk 1995 Lenk, Hans: Schemaspiele: über Schemainterpretation und Interpretations konstrukte. Frankfurt/Main 1995. Lenk 1997 Lenk, Hans: “Interpretationen und Imprägnationen.” In: Simon, Josef (ed.): Orientierung in Zeichen. Zeichen und Interpretationen III. Frankfurt/Main 1997:19–40. Lessing 1962 Lessing, G[otthold] E[phraim]: Hamburg Dramaturgy. With a New Introduction by Victor Lange. New York 1962. Lessing 1973 Lessing, Gotthold Ephraim: Werke. In Zusammenarbeit mit Karl Eibl, Helmut Göbel, Gerd Hillen, Albert von Schirnding und Jörg Schönert herausgegeben von Gerd Göpfert. München 1973.

318

References

Lévi-Strauss 1984 Lévi-Strauss, Claude: “Structure and Form: Reflections on a Work by Vladimir Propp.” In: Propp, Vladimir: Theory and History of Folklore. Translated by Ariadna Y. Martin and Richard P. Martin and several others. Edited, with an Introduction and Notes, by Anatoly Liberman (Theory and History of Literature 5). Manchester 1984:167–188. Linell/Korolija 1997 Linell, Per/Korolija, Natascha: “Coherence in Multi-Party Conversation.” In: Givón, Talmy (ed.): Conversation. Cognitive, Communicative and Social Perspectives. Amsterdam/Philadelphia 1997:167–205. Lockemann 1956 Lockemann, Fritz: “Die Bedeutung des Rahmens in der deutschen Novellendichtung.” In: Wirkendes Wort 6, 1955/56:208–217. Löffler 1988 Löffler, Dietrich: “Über das Erkennen von Geschichten-Strukturen.” In: Hallesche Studien zur Wirkung von Sprache und Literatur 15/8, 1988: 5–25. Lotman 1977 Lotman, Jurij: The Structure of the Artistic Text. Translated from the Russian by Ronald Vroon (Michigan Slavic Contributions 7). Ann Arbor 1977. Lübbe 1977 Lübbe, Hermann: Geschichtsbegriff und Geschichtsinteresse. Analytik und Pragmatik der Historie. Basel/Stuttgart 1977. Mandler/Johnson 1978 Mandler, Jean M./Johnson, Nancy S.: “Erzählstruktur und Erinnerungsleistung. Eine Grammatik einfacher Geschichten.” In: Haubrichs, Wolfgang (ed.): Erzählforschung. Theorien, Modelle und Methoden der Narrativik. Vol. 3 (LiLi: Zeitschrift für Literaturwissenschaft und Linguistik, Beiheft 8), Göttingen 1978:337–379. Manjali 1997 Manjali, Franson D.: Dynamical Models in Semiotics/Semantics. Lecture One: From Discrete Signs To Dynamic Semantic Continuum. Cyber Semiotic Institute, CHASS/University of Toronto, 1997. (seen 19.04. 2003). Martinez/Scheffel 1999 Martinez, Matias/Scheffel, Michael: Einführung in die Erzähltheorie. München 1999. McCarty 1991 McCarty, Willard: “Finding Implicit Patterns in Ovid’s Metamorphoses with TACT.” In: Wooldridge, Russon T. (ed.): A TACT Exemplar (CCH Working Papers 1). Toronto 1991:37–95. McCarty 2002 McCarty, Willard: “New Splashings in the Old Pond: The Cohesibility of Humanities Computing.“ In: Jahrbuch für Computerphilologie 4, 2002: 9–18. Meister 1999 Meister, Jan Christoph: “Jenseits von Lagado. Literaturwissenschaftliches Programmieren am Beispiel des Kodierungsprogramms MOVE-Parser 3.1.” In: Jahrbuch für Computerphilologie 1, 1999:71–81.

References

319

Meister 2002 Meister, Jan Christoph: “’Think Big’: Disziplinarität als wissenschaftstheoretische Benchmark der Computerphilologie.“ In: Jahrbuch für Computerphilologie 4, 2002:19–50. Merkel 1983 Merkel, Bernd G.E.: Basis-Handlungen. Zur Bedeutung des Begriffs in philosophischen Handlungstheorien (Europäische Hochschulschriften 20/124). Frankfurt/Main et al.1983. Meutsch/Schmidt 1985 Meutsch, Dietrich/Schmidt, Siegfried J.: “On the Role of Conventions in Understanding Literary Texts.” In: Poetics 14, 1985: 551–574. Mommsen 1982 Mommsen, Katharina: “Bilde Künstler! Rede nicht!” In: Brinkmann, Richard/Habersetzer, Karl-Heinz/Raabe, Paul/Selig, Karl-Ludwig/Spahr, Blake Lee (eds.): Theatrum Europaeum. Festschrift für Elida Maria Szarota. München 1982:491–516. Moya 1990 Moya, Carlos J.: The Philosophy of Action. An Introduction. Cambridge 1990. Müller 1969 Müller, Joachim: “Zur Entstehung der deutschen Novelle. Die Rahmenhandlung in Goethes ‘Unterhaltungen deutscher Ausgewanderten’ und die Thematik der deutschen Revolution.” In: Kreuzer, Helmut (ed.): Gestaltungsgeschichte und Gesellschaftsgeschichte. Stuttgart 1969:152–175. Müller 2000 Müller, Hans-Harald: “‘Flucht vom Schauplatz politischer Zerrüttung’? Kulturkonflikte und Konfliktkulturen in Goethes ‘Unterhaltungen deutscher Ausgewanderten’.” In: Gutjahr, Ortrud (ed.): Westöstlicher und nordsüdlicher Divan. Goethe in interkultureller Perspektive. Paderborn et al. 2000: 165–175. Neumann 1984 Neumann, Gerhard: “Die Anfänge deutscher Novellistik.” In: Barner, Wilfried/Lämmert, Eberhard/Oellers, Norbert (eds.): Unser Commercium. Goethes und Schillers Literaturpolitik (Veröffentlichungen der deutschen Schillergesellschaft 42). Stuttgart 1984:433–460. Nickau 1966 Nickau, Klaus: “Epeisodion und Episode. Zu einem Begriff der aristotelischen Poetik.” In: Museum Helveticum. Schweizerische Zeitschrift für klassische Altertumswissenschaft 23, 1966:155–171. Niekerk 1995 Niekerk, Carl: Bildungskrisen. Die Frage nach dem Subjekt in Goethes ‘Unterhaltungen deutscher Ausgewanderten’ (Stauffenberg Colloquium 38). Tübingen 1995. Nietzsche 1968 Nietzsche, Friedrich: The Will to Power. A New Translation by Walter Kaufmann and R.J. Hollingdale. Edited, with Commentary by Walter Kaufmann. London 1968. Nietzsche 1977 Nietzsche, Friedrich: Werke in drei Bänden. Hg. von Karl Schlechta. München 1977.

320

References

Noice/Noice 1993 Noice, Helga/Noice, Tony: “The Effects of Segmentation on the Recall of Theatrical Material.” In: Poetics 22, 1993:51–67. Nünning 1990 Nünning, Ansgar: “Bausteine zu einer konstruktivistischen Erzähltheorie.” In: Delfin 13, 1990:3–17. Ohno 1995 Ohno, Christine: “Paradigmen der Bedeutungsanalyse von Aristoteles bis Greimas: Referenz, Differenz und Typisierung.” In: Zeitschrift für Semiotik 17/3–4, 1995:319–353. Orlandi 2002 Orlandi, Tito: “Is Humanities Computing a Discipline?“ In: Jahrbuch für Computerphilologie 4, 2002:51–58. Pasternack 1975 Pasternack, Gerhard: Theoriebildung in der Literaturwissenschaft. Einführung in Grundfragen des Interpretationspluralismus. München 1975. Paul 1992 Paul, Hermann: Deutsches Wörterbuch. 9., vollständig neu bearbeitete Auflage von Helmut Henne und Georg Objartel, unter Mitarbeit von Heidrun Kämper-Jensen. Tübingen 1992. Paul/Fiebich 2003 Paul, Nora/Fiebich, Christina: “action.” In: N.P./C.F.: The Elements of Digital Storytelling. University of Minnesota School of Journalism and Mass Communication’s Institute for New Media Studies and New Directions for News. (seen 19.04.2003). Pavel 1976 Pavel, Thomas G.: La Syntaxe narrative des tragédies de Corneille. Recherches et propositions. Ottawa 1976. Pavel 1985 Pavel, Thomas G.: The Poetics of Plot. The Case of English Renaissance Drama. Manchester 1985. Pavel 1988 Pavel, Thomas G.: “Formalism in Narrative Semiotics.” In: Poetics Today 9/3, 1988:593–606. Pavel 1990 Pavel, Thomas G.: “Narrative Tectonics.” In: Poetics Today 11/2, 1990: 349–364. Pavel 1995 Pavel, Thomas G.: “Racinian Spaces.” In: Bremond, Claude/Landy, Joshua/Pavel, Thomas (eds.): Thematics. New Approaches. New York 1995: 113–126. Perron 1996 Perron, Paul: Semiotics and the Modern Quebec Novel: A Greimassian Analysis of Thériaults Agaguk. Toronto et al. 1996. Peter of Spain 1972 Peter of Spain (Petrus Hispanus Portugalensis): Tractatus. Called Afterwards Summule Logicales. First Critical Edition from the Manuscripts. Ed. by L.M. de Rijk (Philosophical Texts and Studies 22). Assen 1972.

References

321

Pettit 1982 Pettit, Philip: “Die Theorie des rationalen Menschen.” In: Hookway, Christopher/Pettit, Philip (eds.): Handlung und Interpretation: Studien zur Philosophie der Sozialwissenschaften. Berlin 1982:58–84. Pickerodt 1994 Pickerodt, Gerhart: “Seriale Momente in Goethes ‘Unterhaltungen deutscher Ausgewanderten’.” In: Giesenfeld, Günter (ed.): Endlose Geschichten. Serialität in den Medien. Ein Sammelband. Hildesheim et al. 1994: 25–32. Piltz 1982 Piltz, Anders: Die gelehrte Welt des Mittelalters. Aus dem Schwedischen von Sybille Didon. Köln/Wien 1982. Plato 1961 Plato: “The Sophist.” In: Plato: Theaetus. Sophist. With an English Translation. Cambridge, Mass. 1961:259–459. Prince 1973 Prince, Gerald: A Grammar of Stories. An Introduction. The Hague/Paris 1973. Prince 1982 Prince, Gerald: Narratology. The Form and Functioning of Narrative (Janua Linguarum, Series Major 108). Berlin et al. 1982. Prince 1988 Prince, Gerald: “The Disnarrated.” In: Style 22, 1988:1–8. Prince 1998 Prince, Gerald: “Revisiting Narrativity.” In: Grünzweig, Walter/Solbach, Andreas (eds.): Grenzüberschreitungen. Transcending Boundaries – Narratologie im Kontext. Narratology in Context. Tübingen 1998:43–51. Propp 1928 Propp, Vladimir J.: Morfologija skazki (Voprosy po tiki 12). Leningrad 1928. Rpr. St Petersburg 1992. Propp 1984 Propp, Vladimir J.: Morphology of the Folktale. Second Edition, Revised and Edited with a Preface by Louis A. Wagner. With a New Introduction by Alan Dundes (American Folklore Series Bibliographical and Special Series 9). Austin, TX 1984. Puppe 1979 Puppe, Ingeborg: Idealkonkurrenz und Einzelverbrechen. Logische Studien zum Verhältnis von Tatbestand und Handlung. Berlin 1979. Quéré 1992 Quéré, Henri: Fundamentals of Story Logic. Amsterdam/Philadelphia 1992. Raabe 1939 Raabe, August: “Der Begriff des Ungeheuren in den ‘Unterhaltungen der deutschen Ausgewanderten’.” In: Goethe. Viertelmonatsschrift der GoetheGesellschaft 4, 1939:23–39. Rapaport/Shapiro 1995 Rapaport, William/Shapiro, Stuart: “Cognition and Fiction.” In: Duchan, Judith F./Bruder, Gail A./Hewitt, Lynne E. (eds.): Deixis in Narrative. A Cognitive Science Perspective. Hillsdale, NJ 1995: 107–127.

322

References

Renner 1981 Renner, Karl Nikolaus: “‘Der Findling’. Ein Film von George Moorse nach Heinrich von Kleist.” In: Kanzog, Klaus (ed.): Erzählstrukturen – Filmstrukturen. Berlin 1981:25–58. Renner 1983 Renner, Karl Nikolaus: “Der Findling”. Eine Erzählung von Heinrich von Kleist und ein Film von George Moorse. Prinzipien einer adäquaten Wiedergabe narrativer Strukturen. München 1983. Rennie 1996 Rennie, Nicholas: “Ut Pictura Historia: Goethe’s Historical Imagination and the ‘Augenblick’.” In: Goethe Yearbook. Publications of the Goethe Society of North America 8, 1996:120–141. Rescher 1967 Rescher, Nicholas: “Aspects of Action.” In: N.R. (ed.): The Logic of Decision and Action. Pittsburgh 1967:215–219. Ricœur 1984 Ricœur, Paul: Time and Narrative. Volume 1. Translated by Kathleen McLaughlin and David Pellauer. Chicago/London 1984. Riggs 1973 Riggs, David: “‘Plot’ and ‘Episode’ in Early Neoclassical Criticism.” In: Dessen, Alan C. (ed.): Renaissance Drama. Essays on Dramatic Antecedents. Evanston 1973:149–176. Rimmon-Kenan 1983 Rimmon-Kenan, Shlomith: Narrative Fiction. Contemporary Poetics. London/New York 1983. Rindisbacher 1994 Rindisbacher, Hans J.: “Procurator or Procreator: Goethe’s Unterhaltungen as Ironic Genre Praxis.” In: Goethe Yearbook. Publications of the Goethe Society of North America 7, 1994:62–84. Ritter 1974 Ritter, Joachim (ed.): Historisches Wörterbuch der Philosophie. Darmstadt/ Basel 1974. Rockwell 2002 Rockwell, Geoffrey: “Multimedia, Is it a Discipline? The Liberal Arts and Servile Arts in Humanities Computing.“ In: Jahrbuch für Computerphilologie 4, 2002:59–70. Rohs 1986 Rohs, Peter: “Gedanken zu einer Handlungstheorie auf transzendental-philosophischer Grundlage.” In: Prauss, Gerold (ed.): Handlungstheorie und Transzendentalphilosophie. Frankfurt/Main 1986:219–245. Rumelhart 1975 Rumelhart, David E.: “Notes on a Schema for Stories.” In: Bobrow, Daniel G./Collins, Allan (eds.): Representation and Understanding: Studies in Cognitive Science. New York 1975:211–236. Rumelhart 1980 Rumelhart, David E.: “On Evaluating Story Grammars.” In: Cognitive Science 4, 1980:313–317. Ryan 1985 Ryan, Marie-Laure: “The Modal Structure of Narrative Universes.” In: Poetics Today 6/4, 1985:717–755.

References

323

Ryan 1986 Ryan, Marie-Laure: “Embedded Narratives and Tellability”. In: Style 20/3, 1986:319–340. Ryan 1991 Ryan, Marie-Laure: Possible Worlds, Artificial Intelligence and Narrative Theory. Bloomington 1991. Ryan 1992 Ryan, Marie-Laure: “The Modes of Narrativity and their Visual Metaphors.” In: Style 26/3, 1992:368–387. Sachsse 1987 Sachsse, Hans: Kausalität – Gesetzlichkeit – Wahrscheinlichkeit. Die Geschichte von Grundkategorien zur Auseinandersetzung des Menschen mit der Welt. Darmstadt 1987. Scheerer/Winkler 1976 Scheerer, Thomas M./Winkler, Markus: “Zum Versuch einer universalen Erzählgrammatik bei Claude Bremond.” In: Poetica. Zeitschrift für Sprachund Literaturwissenschaft 8, 1976:1–24. Scherner 2000 Scherner, Maximilian: “Kognitionswissenschaftliche Methoden in der Textanalyse.” In: Brinker, Klaus/Antos, Gerd/Heinemann, Wolfgang/Sager, Sven F. (eds.): Text- und Gesprächslinguistik – Linguistics of Text and Conversation. Ein internationales Handbuch zeitgenössischer Forschung – An International Handbook of Contemporary Research. Berlin/New York 2000:186–195. Schmid 1982 Schmid, Wolf: “Die narrativen Ebenen ‘Geschehen’, ‘Geschichte’, ‘Erzählung’ und ‘Präsentation der Erzählung’.” In: Wiener Slawistischer Almanach 9, 1982:83–110. Schmid 1984 Schmid, Wolf: “Der semiotische Status der narrativen Ebenen ‘Geschehen’, ‘Geschichte’, ‘Erzählung’ und ‘Präsentation der Erzählung’.” In: Oehler, Klaus (ed.): Zeichen und Realität. Akten des 3. Semiotischen Kolloquiums der Deutschen Gesellschaft für Semiotik e.V. (Probleme der Semiotik 1.II). Tübingen 1984:477–486. Schmid 1992 Schmid, Wolf: Ornamentales Erzählen in der russischen Moderne. Chechov, Babel, Zamjatin (Slavische Literaturen 2). Frankfurt/Main et al. 1992. Schmidt 1979 Schmidt, Siegfried J. (ed.): Empirie in Literatur- und Kunstwissenschaft. München 1979. Schnotz 2000 Schnotz, Wolfgang: “Das Verstehen schriftlicher Texte als Prozeß.” In: Brinker, Klaus/Antos, Gerd/Heinemann, Wolfgang/Sager, Sven F. (eds.): Text- und Gesprächslinguistik – Linguistics of Text and Conversation. Ein internationales Handbuch zeitgenössischer Forschung – An International Handbook of Contemporary Research. Berlin/New York 2000:497–506. Schütz 1974 Schütz, Alfred: Der sinnhafte Aufbau der sozialen Welt. Eine Einleitung in die verstehende Soziologie. Frankfurt/Main 1974.

324

References

Shapiro/Rapaport 1995 Shapiro, Stuart/Rapaport, William: “An Introduction to a Computational Reader of Narratives.” In: Duchan, Judith F./Bruder, Gail A./Hewitt, Lynne E. (eds.): Deixis in Narrative. A Cognitive Science Perspective. Hillsdale, NJ 1995:79–105. Shen 1985 Shen, Yeshayahu: “On Importance Hierarchy and Evaluation Devices in Narrative Texts.” In: Poetics Today 6/4, 1985:681–698. Shen 1988 Shen, Yeshayahu: “Schema Theory and the Processing of Narrative Texts: The X-Bar Story Grammar and the Notion of Discourse Topic.” In: Journal of Pragmatics 12/5–6, 1988:639–676. Shen 1990 Shen, Yeshayahu: “Centrality and Causal Relations in Narrative Texts.” In: Journal of Literary Semantics 19/1, 1990:1–29. Simon/Wallach 1999 Simon, Herbert A./Wallach, Dieter: “Editorial: Cognitive Modelling in Perspective.” In: Kognitionswissenschaft 8, 1999:1–4. Smith 1980 Smith, Barbara Herrnstein: “Narrative Versions, Narrative Theories.” In: Critical Inquiry 7/1, 1980:213–236. Söring 1981 Söring, Jürgen: “Die Verwirrung und das Wunderbare in Goethes ‘Unterhaltungen deutscher Ausgewanderten’.” In: Zeitschrift für deutsche Philologie 100, 1981:544–559. Stammen 1990 Stammen, Theo: “Johann Wolfgang von Goethe: ‘Unterhaltungen deutscher Ausgewanderten’.” In: Große Werke der Literatur: Eine Ringvorlesung an der Universität Augsburg. Tübingen 1990:123–149. Stanzel 1977 Stanzel, Franz K.: “Die Komplementärgeschichte. Entwurf einer leserorientierten Romantheorie.” In: Haubrichs, Wolfgang (ed.): Erzählforschung. Theorien, Modelle und Methoden der Narrativik. Vol. 2 (LiLi: Zeitschrift für Literaturwissenschaft und Linguistik, Beiheft 6), Göttingen 1977: 240–259. Stanzel 1995 Stanzel, Franz K.: “Historie, Historischer Roman, Historiographische Metafiktion.” In: Sprachkunst. Beiträge zur Literaturwissenschaft 26/1, 1995:113–123. Stegemann 1999 Stegemann, Bernd: Die Gemeinschaft als Drama. Eine systemtheoretische Untersuchung zu Handlung und Komunikation im Drama. Diss. Hamburg 1999. Stempel 1973 Stempel, Wolf-Dieter: “Erzählung, Beschreibung und der Historische Diskurs.” In: Koselleck, Reinhart/Stempel, Wolf-Dieter (eds.): Geschichte – Ereignis und Erzählung (Poetik und Hermeneutik 5). München 1973: 325–346.

References

325

Stierle 1975 Stierle, Karlheinz: Text als Handlung. Perspektiven einer systematischen Literaturwissenschaft. München 1975. Stierle 1981 Stierle, Karlheinz: “Text als Handlung und Text als Werk.” In: Fuhrmann, Manfred/Jauß, Hans Robert/Pannenberg, Wolfhart (eds.): Text und Applikation. Theologie, Jurisprudenz und Literaturwissenschaft im hermeneutischen Gespräch (Poetik und Hermeneutik 9). München 1981:537–545. Strohner 2000 Strohner, Hans: “Kognitive Voraussetzungen: Wissenssysteme – Wissensstrukturen – Gedächtnis.” In: Brinker, Klaus/Antos, Gerd/Heinemann, Wolfgang/ Sager, Sven F. (eds.): Text- und Gesprächslinguistik – Linguistics of Text and Conversation. Ein internationales Handbuch zeitgenössischer Forschung – An International Handbook of Contemporary Research. Berlin/New York 2000:261–274. Sugars 1996 Sugars, Cynthia: “Prochain Episode; or, Charting Canonical Time.” In: Canadian Review of Comparative Literature. Revue canadienne de littérature comparée 23/4, 1996:1125–1143. Swift 1995 Swift, Jonathan: Gulliver’s Travels. Complete, Authoritative Text with Biographical and Historical Contexts, Critical History, and Essays from Five Contemporary Critical Perspectives. Ed. by Christopher Fox (Case Studies in Contemporary Criticism). Boston/New York 1995. Taylor 1971 Taylor, S. Ortiz: “Episodic Structure and the Picaresque Novel.” In: The Journal of Narrative Technique 7–8, 1971:218–225. Tesnière 1959 Tesnière, Lucien: Elements de syntaxe structurale. Paris 1959. Titzmann 1977 Titzmann, Manfred: Strukturale Textanalyse. Theorie und Praxis der Interpretation. München 1977. Tjupa 2000 Tjupa, Valerij: “Die aktuelle Gliederung des narrativen Textes.” Abstract of a paper delivered at the University of Hamburg, Institute of Slavic Studies, June 2000. Todorov 1969 Todorov, Tzvetan: Grammaire du Décaméron (Approaches to Semiotics 3). Den Haag/Paris 1969. Todorov 1977 Todorov, Tzvetan: The Poetics of Prose. New York 1977. Tolliver 1997 Tolliver, Joyce: “From Labov and Waletzky to ‘Contextualist Narratology’: 1967–1997.” In: Journal of Narrative and Life History 7/1–4, 1997: 53–60. Tomashevskij 1967 Tomaševskij, Boris: Teorija literatury. Po tika (The Slavic Series 6). New York et al. 1967.

326

References

Tomashevsky 1965 Tomashevsky, Boris: “Thematics.” In: Russian Formalist Criticism. Four Essays. Translated with an Introduction by Lee T. Lemon and Marion J. Reis (Regents Critics Series). Lincoln 1965:59–95. Toolan 1988 Toolan, Michael J.: Narrative. A Critical Linguistic Introduction. London/ New York 1988. Topitsch 1975 Topitsch, Ernst: Die Voraussetzungen der Transzendentalphilosophie. Hamburg 1975. Tornius 1927 Tornius, Valerian (ed.): Goethes Leben. Goethe Biographie begründet von Wilhelm Bode. Fortgeführt von Valerian Tornius. 1794–1798. Der Bund mit Schiller. Berlin 1927. Unsworth 2002 Unsworth, John: “What is Humanities Computing and What is Not?“ In: Jahrbuch für Computerphilologie 4, 2002:71–83. Vaihinger 1927 Vaihinger, Hans: Die Philosophie des Als-Ob. System der theoretischen, praktischen und religiösen Fiktionen der Menschheit. Auf Grund eines idealistischen Positivismus. Mit einem Anhang über Kant und Nietzsche. Leipzig 1927. van Dijk 1976 van Dijk, Teun A.: “Philosophy of Action and Theory of Narrative.” In: Poetics 5, 1976:287–338. van Dijk 1980 van Dijk, Teun A.: Macrostructures. An Interdisciplinary Study of Global Structures in Discourse, Interaction, and Cognition. Hillsdale, NJ 1980. van Inwagen 1977 van Inwagen, Peter: “Creatures of Fiction.” In: American Philosophical Quarterly 14, 1977:299–308. Verdaasdonk/van Rees 1992 Verdaasdonk, Hugo/van Rees, Cees J.: “The Narrow Margins of Innovation in Literary Research.” In: Poetics 21, 1992:141–152. Vipond/Hunt 1987 Vipond, Douglas/Hunt, Russell A.: “Point-Driven Understanding: Pragmatic and Cognitive Dimensions of Literary Reading.” In: SPIEL 6/1, 1987: 261–277. von Graevenitz 1993 von Graevenitz, Gerhart: “Erinnerungsbild und Geschichte. Geschichtsphilosophie in Vicos ‘Neuer Wissenschaft’ und in Goethes ‘Pandora’.” In: Goethe-Jahrbuch 110, 1993:77–88. von Wilpert 1991 von Wilpert, Gero: “Die politische Sängerin. Spuk und Aktualität in Goethes ‘Sängerin Antonelli’.” In: Seminar. A Journal of German Studies 27/3, 1991:189–202.

References

327

von Wright 1979 von Wright, Georg Henrik: “Das menschliche Handeln im Lichte seiner Ursachen und Gründe.” In: Lenk, Hans (ed.): Handlungstheorien interdisziplinär. Vol. II.2: Handlungserklärungen und philosophische Handlungsinterpretation (Kritische Information 63). München 1979: 417–429. Voßkamp/Jaumann 1992 Voßkamp, Wilhelm/Jaumann, Herbert: “Kommentar zu ‘Unterhaltungen Deutscher Ausgewanderten’.” In: Jaumann, Herbert (ed.): Goethe. Sämtliche Werke. Briefe, Tagebücher und Gespräche. I. Abteilung, Bd. 9. Frankfurt/ Main 1992:1503–1611. WDG 1969 Wörterbuch der deutschen Gegenwartssprache. Berlin 1969. Weing 1998 Weing, Siegfried: “Aristotelian Foundations of German Novella Theory.” In: Seminar 34/1, 1998:45–62. Wiehl 1971 Wiehl, Reiner: “Über den Handlungsbegriff als Kategorie der Hegelschen Ästhetik.” In: Hegel-Studien 6, 1971:135–170. Wiehl 1976 Wiehl, Reiner: “Reflexionsprozesse und Handlungen.” In: Neue Hefte für Philosophie 9, 1976:17–65. Willaschek 1992 Willaschek, Marcus: Praktische Vernunft. Handlungstheorie und Moralbegründung bei Kant. Stuttgart 1992. Witte 1984 Witte, Bernd: “Das Opfer der Schlange.” In: Barner, Wilfried/Lämmert, Eberhard/Oellers, Norbert (eds.): Unser Commercium. Goethes und Schillers Literaturpolitik (Veröffentlichungen der deutschen Schillergesellschaft 42). Stuttgart 1984:461–484. Wittgenstein 1958 Wittgenstein, Ludwig: Philosophical Investigations. The English Text of the Third Edition. Translated by G.E.M. Anscombe. Englewood Cliffs, NJ 1958. Wittgenstein 1961 Wittgenstein, Ludwig: Tractatus logico-philosophicus. The German Text of Ludwig Wittgenstein’s Logisch-philosophische Abhandlung with a New Translation by D.F. Pears and B.F. McGuinness. London/New York 1961. Ziegler 1984 Ziegler, Vickie: “Goethe and the French Actress: How Clairon Became Antonelli.” In: Monatshefte 76/1, 1984:21–30. Ziolkowski 1958 Ziolkowski, Theodore: “Goethe’s ‘Unterhaltungen deutscher Ausgewanderten’: A Reappraisal.” In: Monatshefte 50, 1958:57–74.

Author Index Abel, Günter 71-73, 110 Aeschylus 38, 39 Arendt, Hanna xiii, 58, 59, 63, 74, 75, 140 Aristoteles 21, 35 Aristotle xiii, xxii, 10, 11, 21, 22, 33, 35-37, 39, 40, 48, 57, 68, 107, 110, 151, 153-158, 174, 175, 181, 193, 204, 263, 296 Bachmann-Medick, Doris 34 Bal, Mieke 112, 202 Barsch, Achim 18 Barthes, Roland 37, 38, 155, 166, 174, 201, 202 Batteux, Charles 152 Baumgartner, Alexander 260 Bauschinger, Sigrid 264, 271 Beasley, Jerry C. 156 Beddow, Michael 288 Belknap, Robert L. 20, 21 Berkemann, Jörg 43 Beyer, Wilhelm Raimund 262, 264 Bittner, Rüdiger 68 Bloomfield, Morton W. 156 Boccaccio, Giovanni 261 Bourdieu, Pierre 41, 74, 75 Boyle, Nicholas 289 Brandt, Helmut 264, 283 Bräutigam, Bernd 266, 271 Brecht, Bert 13, 262 Bremond, Claude 15, 95, 96, 100, 102, 107, 147, 162, 166, 168-170, 202, 208 Brennenstuhl, Waltraud 42 Bringsjord, Selmer ix Brooks, Peter 86, 112 Brown, Jane K. 140, 160, 267, 268, 297 Bubner, Rüdiger 57, 289 Calvino, Italo

x

Casati, Roberto 71 Caesar 146 Carlyle, Thomas 275 Caroll, Lewis 252 Chatman, Seymour 13, 16, 202 Cohn, Dorrit 16, 17, 108 Collins, Allan 166 Conrady, Karl Otto 297 Coste, Didier 16, 112 Culler, Jonathan 150 Dammann, Günter xvi, 153-155 192, 264, 268-271, 295 Danto, Arthur C. 146 Davidson, Donald 50, 71, 74 Dodd, William Nigel 156 Doležel, Lubomír 13, 20, 103 Douglas, Graham 193 Eco, Umberto 27, 88, 231, 248 Ellis, John M. 263 Else, Gerald F. 153 Engel, Johann Jakob 34 Euripides 39 Ferrucci, David A. ix Fichte, Johann Gottlieb 3-5 Fiebich, Christina 7 Fink, Gonthier-Louis 260, 264, 295 Fish, Stanley 105 Fludernik, Monika 204 Forster, E.M. 23-25, 118, 186, 187, 189, 232 Frank, Manfred 156, 166 Frederiksen, Carl H. 203 Fricke, Gerhard 264, 266, 267, 275 Fuhrmann, Manfred 35-37 Gabriel, Gottfried 7 Gaede, Friedrich 154 Gaier, Ulrich 260, 272, 293-295 Galloway, Patricia 106, 204, 205

330

Author Index

Geißler, Rolf 262, 264 Genette, Gérard 16, 23, 108, 115, 195, 202-204 Gerhardt, Volker 66-68 Goethe, Johann Wolfgang von xiii, xvi, xxii, xxiii, 32, 93, 108, 131134, 140, 183, 189, 192, 195, 219, 224, 236, 254, 257, 259-275, 283, 284, 286-305 Goldstein, Moritz 264, 265 Goodman, Nelson 16 Gottschall, Rudolf v. 260 Grabes, Herbert 10 Grant, Michael 38 Graves, Barbara 203 Greimas, Algirdas Julien xiii, 48, 95, 102, 166-180, 185, 193, 195, 268 Grimm, Jacob 29-32 Grimm, Wilhelm 29-32 Gundolf, Friedrich 261, 262 Haidu, Peter 155, 162 Hallstein, Patricia 202 Harker, John W. 203 Hasselberg-Weyandt, Wilfried 123, 260 Hauptmeier, Helmut 18 Hegel, Georg Wilhelm Friedrich 71, 80-84, 88, 156 Henel, Heinrich 263, 264 Herman, David xx, xxi, 201, 204, 254 Higginbotham, James 71 Hoermann, Roland 264, 268, 293 Hoffmann, Ernst Fedor 264, 267, 283 Homer 35, 37, 39, 153 Honig, Richard 43 Hunt, Russell A. 123 Ingarden, Roman 14, 18 Iser, Wolfgang 18 Jacobs, Jürgen 263 Jahn, Manfred 18, 204 Jaumann, Herbert 259-262, 271, 292-294 Johnson, Nancy S. 18, 161-163, 204 Jürgens, Ilse 264, 266

Kafalenos, Emma 20, 149-151 Kalinowski, Georges 169, 176, 193 Kant, Immanuel xiii, 65-72, 74-76, 88, 156, 157, 166, 272, 291, 293, 294, 298, 302 Kauffmann, Clemens 36 Kaulbach, Friedrich 67, 68 Käuser, Andreas 263, 290, 291 Kindt, Tom 8 Kindt, Walther 18 Kleinschnieder, Manfred 294 Korolija, Natascha 132, 157, 229 Koselleck, Reinhart 95, 290 Kraft, Werner 264 Kunz, Josef 263 Kurzweil, Ray ix Lacan, Jaques 112 Lämmert, Eberhard 95, 152, 153, 203 Landauer, Gustav 262 Lausberg, Heinrich 136 Lebowitz, Michael 243 Lee, Jie-Oun 202 Lehnert, Wendy G. 18, 106, 204-206, 208, 209 Lenk, Hans 68, 76-79, 86 Lessing, Gotthold Ephraim 1, 3234, 111 Lévi-Strauss, Claude 1, 19, 100 Linell, Per 132, 157, 229 Lockemann, Fritz 264-268 Löffler, Dietrich 123, 203 Lotman, Jurij 127, 145, 146, 163, 202 Lübbe, Hermann 95 Lucerna, Camilla 259 Luther 29, 31 Lüthi, Max 153 Mandler, Jean M. 161-163, 204 Manjali, Franson D. 96 Marlowe 100, 102 Martinez, Matias 23, 202 McCarty, Willard xviii, 211 Meehan, James ix, 211 Meister, Jan Christoph x, xi-xiii, xviii, 210 Mendelsohn 111 Merkel, Bernd G.E. 50

Author Index

Meutsch, Dietrich 123 Mommsen, Katharina 260, 266, 271 Moya, Carlos J. 54-57, 72, 75 Müller, Hans-Harald 8, 195, 272, 273 Müller, Joachim 263, 264, 267, 295 Neumann, Gerhard 270 Nickau, Klaus 153, 154 Niekerk, Carl 264, 270 Nietzsche, Friedrich 48, 49, 57, 63, 74, 88 Noice, Helga 160 Noice, Tony 160 Nünning, Ansgar 204 Occam xxii Ohno, Christine 167-169, 174, 175, 179 Orlandi, Tito xviii, xxii Pasternack, Gerhard 165 Paul, Hermann 29 Paul, Nora 7 Pavel, Thomas G. 13, 95, 100-104, 106, 107, 116, 149, 156, 161, 162, 165, 167, 202, 208, 210 Perron, Paul 166, 202 Peter of Spain (Petrus Hispanus Portugalensis) 174 Pettit, Philip 44 Pickerodt, Gerhart 295 Piltz, Anders 174 Plato 8, 13, 36, 37, 72, 73, 78, 82, 107, 110, 181 Prince, Gerald xi Propp, Vladimir J. xxi, 13-16, 91, 97, 99, 102, 107, 132, 145, 162, 202 Puppe, Ingeborg 43 Quéré, Henri

170, 171, 174

Raabe, August 262-266 Rapaport, William 114, 120, 123, 242, 243 Renner, Karl Nikolaus 202 Rennie, Nicholas 263, 290 Rescher, Nicholas 13, 50 Ricœur, Paul 9, 171, 174

331

Riggs, David 155, 156 Rimmon-Kenan, Shlomith 202 Rindisbacher, Hans J. 263 Ritter, Joachim 29 Rockwell, Geoffrey xviii Rohs, Peter 65, 66 Rumelhart, David E. 161, 162, 204, 205, 208 Ryan, Marie-Laure 10, 13, 15, 16, 103-106, 132, 145, 206, 209, 243 Sachsse, Hans 129, 130 Scheerer, Thomas M. 95 Scheffel, Michael 202 Scherner, Maximilian 210 Schlegel, Friedrich 259, 263 Schiller, Friedrich 260-262, 266, 270272, 292-297 Schmid, Wolf 22, 24, 46, 86, 91, 92, 114, 146, 193 Schmidt, Siegfried J. 18, 22 123 Schnotz, Wolfgang 254 Schütz, Alfred 50, 58, 60-64, 75, 77 Shapiro, Stuart 114, 120, 123, 242, 243 Shen, Yeshayahu 160-162, 204 Simon, Herbert A. 114, 123 Smith, Barbara Herrnstein 16 Söring, Jürgen 267 Stammen, Theo 264 Stanzel, Franz K. 18, 203 Stegemann, Bernd 81 Stempel, Wolf-Dieter 92, 95, 116, 146, 193 Stierle, Karlheinz 46, 146 Strohner, Hans 210 Sugars, Cynthia 156 Swift, Jonathan 199, 201 Taylor, S. Ortiz 156, 179 Tesnière, Lucien 96 Titzmann, Manfred 202 Tjupa, Valerij 163 Todorov, Tzvetan 96-98, 100,107, 113, 132, 145, 161, 202 Tolliver, Joyce 17 Tomashevsky, Boris 19-21 Toolan, Michael J. 202 Topitsch, Ernst 48 Tornius, Valerian 262

332 Unsworth, John

Author Index

xviii

Vaihinger, Hans 9 van Dijk, Teun A. 47, 112, 132, 145, 158-160, 208 van Inwagen, Peter 120 van Rees, Cees J. 18 Varzi, Achille C. 71 Verdaasdonk, Hugo 18 Vipond, Douglas 123 von Graevenitz, Gerhart 290 von Wilpert, Gero 267 von Wright, Georg Henrik 53 Voßkamp, Wilhelm 259-262, 271, 292-294

Wallach, Dieter 114, 123 Weber, Max 60 Weing, Siegfried 263 Wiehl, Reiner 68, 80-83 Wieland, Christoph Martin 30 Willaschek, Marcus 66 Wilpert, Gero von 267 Winkler, Markus 95 Witte, Bernd 271 Wittgenstein, Ludwig 51, 52, 55, 132, 144, 273, 274 Ziegler, Vickie 264, 267 Ziolkowski, Theodore 264, 266, 267

Subject Index The following subject index does not list individual occurences for high frequency words like 'action', 'episode', 'event' etc.. Relevant pages and sections for these terms should rather be identified via the table of contents. Abhandlungen 33 Ablauf 129 Absicht 32,49, 293 absolutism 78, 157 abstraction 4, 5, 16, 30, 70, 71, 96, 127, 131, 182 achèvement 96 actant 105 actant-based 171, 180 actant-focus 216, 218 action – action-based 50 – action-forming 250 – action-orientated 149, 181 – action product 251, 252, 254, 255, 274, 275, 280, 283-287 – action-related 6 actlist 250 actus 30, 32 agency 55 agent 5, 7, 8, 12, 33, 35, 40-48, 50-56, 58-65, 68-70, 73, 75, 77-79, 81, 83, 85, 96, 103, 104, 114, 115, 129, 158, 161, 184, 211, 212, 241, 242 – agential 42, 45, 46, 52, 56, 59, 60, 63, 65, 75, 162, 172 – non-agential 65, 172 agnosticism 75, 273 algorithm xiii, xix, xxii, 17, 85, 104106, 126, 206, 209-212, 223, 224, 226, 227, 229, 234-236, 242-245, 250, 253, 287 ambiguity 150, 156, 209 anachronism 115 analepsis 14, 16, 115, 152, 239 anaphora 269 – anaphoric 231 Anfangszustand 130

antagonist 102, 162, 228 anthropology 75, 290 anticlimax 284 anti-fiction 13 antimodel 13, 15 antinomy 81, 294 apperception 65-69 Auslegungssucht 259, 293 Backtracking 243 baroque 192, 269 because-relation 53 Bedeutung 7, 32, 84 Begebenheiten 69, 148, 288 Begriff 34, 76, 80 Captatio 10 catharsis 296 causality 23, 24, 26, 28, 48-50, 52, 74, 101, 114, 122, 129, 130, 150 causation 66 choreia 38, 39 chronology 97, 114, 115, 117, 138, 169, 180, 188, 227, 239 chronos 263 closure 39, 52, 57, 79, 86, 90, 154, 156, 161, 180-182, 235, 242, 295, 296, 298 codomain 119, 124 cogito 83 cognition xxii, 8, 46, 67, 69, 70, 8082, 86, 123, 143, 157, 158, 160, 180, 237, 298 cognitivism 18 coherence 15, 35, 53, 67, 69, 70, 82, 109, 112, 140, 156-158, 160-162, 181, 188, 235, 248, 251, 269, 301 cohesion 67, 70, 205, 206, 284 coincidence 40, 42, 164, 289

334

Subject Index

computer ix, x, xii, xiii, xviii, xx, xxii, 103, 105, 106, 123, 205, 206, 209, 211-213, 234, 243, 247, 250, 275 – computer-aided xviii, xix, 199, 204, 210, 274, 304 – computational 206, 209, 210, 212, 213, 242-244, 248, 255 – computerized ix, 206 Computerlinguistik xviii Computerphilologie xviii connectivity 197, 274, 280 consecutivity 92 consistency 34, 78, 212, 226 constructivism 5, 18, 79, 204 contingency 24, 49, 51, 67 contrariety 193 database ix, xi, xvi, 212, 216, 221, 222, 229, 232-234, 236, 242, 244246, 250 dechronologization 9 decomposition 210 deconstructivism 108 deduction 53, 154, 172 deictics 147, 269 denotation 7, 32, 76, 115, 145, 248 dependency 4, 8, 10, 64, 82, 96, 110, 145, 210 descriptors 216 diachrony 94, 100, 147, 149 dialogue ix, 38, 153, 262, 283 dichotomy 16, 37, 53, 60, 95, 149, 167, 178, 264, 265, 299, 300 diegesis 11, 12, 14, 41 dilemma 74, 92, 163 dirsynonym 229, 230, 236, 237 discourse event 107, 108, 113-116, 121, 123, 138-141, 143, 144, 182, 183, 214 disambiguation 209 disequilibrium 96, 97 disnarrated 14-16 dispositio 22, 135, 136 disposition 12, 44, 75, 135, 136, 199, 214, 219, 225-227, 229, 230, 233, 235, 239, 284 dithyramb 37, 39 drama 32, 33, 35, 39, 81, 82, 100, 156 Dramaturgie, Hamburgische 32, 33

ellipsis ix, 14, 16, 23, 125, 137, 212, 231, 240 elocutio 22 empiricism 208, 291, 292, 298 enactment 41 encoding 210, 214, 223, 229, 230, 238, 275, 276, 284 Endzustand 130 entelechy 37 entity 6, 19, 70, 74, 227, 242 enumeration 113, 281, 294 epeisodia 154, 155 epeisodion 153-158, 160, 163 episode – episode-forming 250 – episodic 151-153, 155-157, 160, 174, 187, 193, 230-232, 235, 251, 252, 281 episodicity 249, 251, 282 epistemology xix, 109, 264, 303, 327 epistemy 174 EpiTest xiii, xxi, 106, 210, 213, 214, 222-224, 227, 232, 234-238, 242-247, 249-254, 276, 277, 280, 284-287, 301 Ereignis 109, 130, 264 Erzählhandlung 46 Erzählung 292, 293 Erzählungen 293 escapism 261, 262 essentialism 181 etymology 14, 20, 28, 29, 31 event – discourse event 107, 108, 113-116, 121, 123, 138-141, 143, 144, 182, 183, 214 – object event 107, 108, 114-117, 123, 124, 135, 138, 140, 141, 143, 144, 150, 182-184, 214 – event-based 70, 72, 73, 110, 111, 113, 116, 117, 124, 143, 145, 148, 155, 180, 228 – event-combining 232 – event-defining 226 – event-forming 193 – eventfulness 42 – eventness xii – event-related 301 – event-sememe 222

Subject Index

EventParser xii, xxi, 106, 138, 210, 212, 214, 215, 217-226, 228-230, 234-237, 242, 245, 247, 252-254, 275, 277 eventualité 15, 147 eventus 109 exarchon 38 exegesis 255 exodos 38 experientiality 204 exposition 38, 104, 135, 136, 141, 214, 216, 218-220, 225-230, 239, 240, 266, 284, 287 fabula 15, 16, 19, 20, 22, 65, 95, 100, 107, 114, 115, 126, 138, 149, 150, 153, 154, 163, 169, 188, 194, 202, 203, 212, 253 fact-sequence 158 finalis, causa 129 finality 60, 129, 130 formalism 20, 106, 114, 169, 170 formalists xxi, 20 Fremdverstehen 50, 56, 59-62, 65, 77 game 51, 100, 104, 168, 169, 274, 293 game-plans 105 game-theory 100, 101, 103 Gegenwartssprache 29 generator 243 genre xx, 17, 20, 25, 32, 35, 39, 81, 108, 113, 119, 121, 127, 141, 155, 156, 164, 167, 187, 228, 255, 261, 263, 266, 269-271, 282, 284, 291, 295, 297, 303 Geschehen 23, 24, 49, 66, 86 Geschichte 23, 58, 86, 261 Geschmacksurteil 272 goal 33, 55, 57, 59, 62, 103, 104, 158, 202, 206 goal-orientated 31 grammaire 166, 202 grammar 16, 17, 95-104, 116, 160162, 166, 171-173, 178, 202, 205, 208 graph 280, 281, 285, 286 habitus

41, 75

335

Hand 27, 29 Handarbeit 30 Handel 30 Händel 30 Handeln 3, 6, 30, 31, 58 Handelsamt 30 Handelszweig 30 Handelunge 30 Handhabbarkeit 29 Handlung 3, 27-34, 46, 58, 66, 76, 80, 84 Handlungsplan 30 Hantalon 29 Hantalunga 30 headword 3 hermeneutics 27, 44, 247, 304 heuristics 106, 129, 171, 172, 254 histoire 16 historicism 291, 292 homogeneity 20, 119, 273 homology 176 hyperconnection 249 hyperdetermination 249 hyperfiction 206 hyperonym 138 idealism 36, 268, 291 Idealkonkurrenz 43 Ideology 129 I-function 72 immanence 28 inachèvement 96 inconsistency 14, 28, 134 indantonym 229 indeterminacy 14, 18, 215 Individualität 84 individuality 48, 83 Individuum 83, 84 Indsynonym 229-232 induction 53 inference 14, 18, 25, 51, 53, 70, 160, 208, 242 – inference-based 70 intentionalism 49 intentionality 27, 28, 31, 33, 34, 40, 42, 43, 45, 47, 50, 51, 55, 58, 76, 80, 269 intention-worlds 103 Interpretationskonstrukt 79 intradiegetic 111

336

Subject Index

isochronous 138, 226, 227, 236239, 246, 250, 276, 277, 280-282, 284-286 isochrony 227 isomorphic 20, 98, 121 isomorphism 91 juxtaposition

13, 230

Kaufhandlung 31 Kausalität 49 Khod 21 knowledge-representation Kritik 79

242

Lagado 201 Laokoon 111 linguistics xviii, 132, 157, 167, 201, 208, 212, 242 macrofact 158 macro-level xiii macrorules 160 mark-up 210, 216, 222, 223, 225, 250 matrix 128, 142, 224, 235, 242, 250, 252 meta-action 7 meta-activities 7, 10, 26, 45, 46 meta-activity 7, 8, 26 metacommentary 270 metaconstruct 182, 188, 190, 197 metadata 210 meta-event 78, 108 metaindsynonym 237 meta-interpretation 230 metalanguage 78 meta-level xiii, 10, 45, 95, 232 metaphor 10, 62, 65, 75, 163, 169, 170, 177, 195, 229, 268, 293 metaphysics 66 metapropositions 145, 242 metastructures 145 metasynonyms 230, 234, 244, 245 metaterms 234 methodology xviii, xx, xxi, 11, 63, 88, 167, 170, 255, 299 microlevel xiii, 93 microplots 227 mimesis 32, 35, 37, 39, 45, 83, 154, 181, 263

module 244 molecule 132, 190, 196 morpheme xviii, 175 Morphologie 304 morphology 19, 128, 165, 167, 178, 294, 304 motif 19-21, 93, 126, 153, 166, 192, 226, 262, 270, 272, 282 movement 1, 33, 54, 55, 73, 75, 77, 104, 202, 255 MoveParser 210 moves 82, 100-104 narratio 208 narrative-processing xi narrativité 168 narrativity xi, 168, 171 narrativization 178 narrator 8, 9, 12-15, 22-24, 27, 28, 43, 44, 53, 58, 78, 81, 105, 115, 116, 122, 125, 134, 135, 140, 141, 146, 147, 150, 192, 203, 214, 216, 265, 273, 274, 282, 288, 289, 297, 299, 301, 303 narratorial 8, 13, 14, 20, 23, 145, 146, 149, 232, 263, 271, 281, 283 Naturrecht 3 necessity 4, 12, 26, 33, 41, 54, 56, 64, 65, 70, 72, 88, 110, 122, 123, 130, 153, 159, 165, 171, 173, 253, 288, 298 negation 45, 95, 119, 121 network xi, 53, 54, 178, 234, 242, 249 networks ix, 35 nombos 39 non-actions 42 non-identity 74, 117, 119, 121, 124, 141-143, 148 non-iterativity 92 nonnarratable 15, 24 nonnarrated 13, 14 noumena 36 number-crunching 232 object – object-action 46 – object-activities 8 – object-activity 7 – object-centred 139

Subject Index

– object event 107, 108, 114-117, 123, 124, 135, 138, 140, 141, 143, 144, 150, 182-184, 214 – object-identical 227, 228, 236, 238-240 – object-level 26, 27, 78, 98 obligation-worlds 103 operator 98, 148 orator 135 output xix, 227, 247, 277, 280 over-determination 23 paralogism 69 parodos 38, 155 parser ix, 209, 210 persona 8 personality 69, 80 personification 6 Phaedrus 36 phenomenology 83, 144 phraseology 116 phrases 23, 113, 136, 215, 222 platform-dependency 210 plot 15, 16, 20-23, 26, 29, 32, 35, 86, 97, 100-104, 106, 145, 153, 155, 156, 192, 202, 204-206, 208, 263, 281-286, 288 – subplots 35 poem 34, 154, 268, 290, 296 Poesie 1, 33 poet 40, 73, 154 poeta 87 poetics xiii, xxii, 3, 21, 22, 27, 31, 33, 35-37, 39, 40, 42, 45, 86, 100, 107, 110, 146, 153-155, 166, 192, 102-204, 270, 295-298, 300, 303 poetry 1, 38, 82, 154, 199, 269 polyvalence 123, 151, 227 positivism 297 possibilités 168 post-structuralism 17, 108, 201 praxeis 21 praxis 57, 68, 72 predication 73, 79, 98, 109, 112-114, 116-120, 122-125, 128, 132-134, 136, 138, 148 prehistory 188, 226, 253 principium 66 a priori 100, 290 probability 24, 153, 267

337

prolepsis 16, 115 prolog 38, 123, 155, 210, 227, 229, 236, 242, 243, 246, 277 prophecy 115, 276 protagonist 38, 101, 102, 118, 162, 228 Protagoras 153 protension 41, 61, 129, 239, 254 prototype 295 reader-reception 22 reader-response 46 realism 13, 268 récit 16, 102, 168, 169, 202 recursion 232, 243 redundancy 232 referentiality 209, 302 referents 111, 112, 157 reification 70 Reihenfolge 32 representationalism 58 rhetoric 46, 87, 112, 135, 136, 181, 204 role-playing 4 romanticism 108, 156 romantics 156, 270 scepticism 301 schema 66, 92, 161, 172, 174, 175, 178 schemata 71, 160, 178 schematizations 18 scholasticism 174 science-fiction ix scientism 169 screenshot 218, 221, 245, 246 script 254 segment 96, 162, 163, 219, 225, 251 – segmentation 22, 156, 157, 163, 267, 276 Selbstverstehen 50, 51, 56, 59-63, 77 self-affirmation 67 self-referentiality 302 semantic network 242 semantics 7, 99, 100, 103, 180, 232, 253, 282 sémantique 169 sememe 176, 179, 180, 221, 222, 229

338

Subject Index

semiology 169 semiosis 57, 231, 234 semiotics 5, 86, 90, 105, 167, 169, 171, 173-176, 201, 202 sequentiality 9, 10, 119, 195, 226, 234 serialization 259 signification 10, 26, 55, 56, 93, 142, 170, 173-175, 178 signified 23, 26, 109 signifier 23, 146 simulation 206 Sinn 7, 30, 267 Situationshaftigkeit 95 sociology 41, 53, 60, 165 software xi, xii, xx, 106, 206, 210, 222, 227, 230, 242, 243, 253, 275-277 stasima 38 stereotype 125, 126 stimulus 52, 136 story – story-generation ix, xi, 106 – storyline 29 strategy 43, 57 stringency 61, 211 structuralism xvi, 100, 166, 174, 201-203 stylistics xviii, stylistic caesura 266 stylistic function 156 subcontrariety 193 subdivision 104, 261 subfeatures 133 subfields 59 subject-centric 210 subject-orientated 30 subroutines xi subsection 85 subtext 190, 191 syllogism 48, 68, 154, 160, 171, 174 synchrony 147, 149 synonym xii, xiii, 230, 231, 233, 235-237 synonymy 230, 231, 233 syntagm 94, 108, 116, 141, 155 syntax 134, 135, 171, 178, 210, 227, 243, 250 syuzhet 15, 16, 19, 20, 22, 91, 93, 95, 100, 107, 114, 115, 126, 127, 138, 149-151, 163, 188, 194, 202, 276

– sjuzhet

150, 203

tagger 212 tags 212 Talespin 106 Tamburlaine 102 Tasso 296 tautology 39 taxonomies 10 taxonomy 178 technology xvii, xxi, 105, 201, 211, 250 teleology xi, 26, 39, 162 tellability 15 temporality 24, 147, 188 tense 199, 239 termination 205 test-bed xxii, 213 text – text-based 207 – text-centred 103 – text-internal xii – text-orientated 145 – text-processing 138, 215, 217 – textualization 155 texture xi theme 19, 116, 212, 250, 255, 261, 262, 266, 269, 271-273, 283, 288, 290, 295, 299, 303 theology 199 theorems 92, 167 Thespis 37 transcendence 28, 55 Trauerspiele 34 Trieb 290 typology 43, 102, 156, 205, 254, 264, 265, 288, 289 underspecification 226 unerhörte Begebenheit 93 unerhörtes Ereignis 264 unnarratable 14, 150 unnarrated 15, 150 Vernunft 79 Verstand 275 virtuality 13, 96 Vorstellungsart 294, 298 Wahlverwandschaften

288

339

Subject Index

Weltanschauung 129 Weltbegebenheiten 288 Weltenlenker 28 Weltgeschehen 291 Weltgeschichte 289 Wendepunkt 263

wish-worlds yield

103

105, 282

z-axis 195 zeroing 97, 99